Public Inquiry Report On Review of Access Pricing
Public Inquiry Report On Review of Access Pricing
Public Inquiry Report On Review of Access Pricing
20 December 2017
This Public Inquiry Report was prepared in fulfilment of sections 55, 56, 61, 65, 104 and
106 of the Communications and Multimedia Act 1998
1. Introduction 6
1.1. Public Inquiry Process 6
1.2. MCMC’s legislative obligations 7
1.3. Consultation process 7
1.4. Submissions received 8
1.5. Structure of this PI Report 9
2. Principles in setting access pricing 10
2.1. Overview 10
2.2. Summary of submissions received 10
2.3. MCMC’s final view 19
3. Cost Modelling General Issues 20
3.1. Overview 20
3.2. Summary of submissions received 20
3.3. MCMC’s final view 42
4. Mobile Virtual Network Operators (MVNOs) 43
4.1. Overview 43
4.2. Summary of submissions received 43
4.3. MCMC’s final view 44
5. Weighted Average Cost of Capital (WACC) 45
5.1. Overview 45
5.2. Summary of submissions received 45
5.3. MCMC’s final view 60
6. Fixed Services 61
6.1. Fixed Services cost model 61
6.2. Summary of submissions received 61
7. Mobile Services 96
7.1. Overview 96
7.2. Summary of submissions received 96
7.3. MCMC’s final view 118
8. Infrastructure Sharing 119
8.1. Overview 119
8.2. Summary of submissions received 119
9. Digital Terrestrial Television (DTT) Broadcasting service 134
9.1. Overview 134
9.2. Summary of submissions received 134
Page 1
9.3. MCMC’s final view 137
10. Next Steps 138
Annexure – Abbreviations 139
Page 2
List of Figures
List of Tables
Page 3
SUMMARY OF MCMC’s FINAL VIEWS ON ACCESS PRICES
In this Public Inquiry the MCMC has undertaken the development of cost models in order
to calculate the costs for the facilities and services in the Access List. The costing
methodologies used to calculate the costs of the various services and the proposed prices
were fully described in the PI Paper (issued on 6 October 2017).
The PI Paper set out the MCMC’s preliminary views on the regulation of access pricing and
invited comments in response to 32 general and specific questions. After consideration of
the submissions received in response to the PI Paper as presented below in this PI Report,
the following table summarises the MCMC’s final views on regulatory pricing of the services
in the Access List.
Page 4
Service MCMC’s final view
Page 5
1. Introduction
In its Public Inquiry Paper on the Review of Access Pricing (PI Paper) issued on 6 October
2017, the MCMC detailed the approach and methodology it proposed to adopt within this
Public Inquiry:
(a) to determine which facilities and services in the Access List should be subject to
ex-ante price regulation;
(b) to determine cost-based prices for each facility or service in the Access List; and
(c) to set cost-based prices for some facilities and services in the Access List.
The MCMC noted that under section 55(1) of the Communications and Multimedia Act 1998
(CMA), the MCMC may, from time to time, make a determination on any matter specified
in the CMA. This Public Inquiry relates to access to services under Part VI, Chapter 3 of
the CMA.
Given the long-term consequences of access regulation, for this Public Inquiry the MCMC
adopted a consultative approach as provided for under the legislation in order to obtain
maximum input from both industry and the public. This approach was also designed to
promote transparency in the exercise of the MCMC’s powers.
The PI Paper set out the MCMC’s preliminary views as to which facilities and services
should be subject to price regulation and, where relevant, the proposed regulatory prices
for the period 2018 to 2020. The PI Paper invited comments on the appropriateness of
setting the proposed prices and on the methodology used to calculate the prices. The PI
Paper specifically sought comments through 32 questions.
Page 6
1.2. MCMC’s legislative obligations
Part VI of the CMA contains provisions on economic regulation including access to services.
Section 149 within Chapter 3, Part VI requires access providers to provide access to
facilities and services on reasonable terms and conditions, which, in the MCMC’s view,
include the prices that an access provider sets.
In addition to Part VI, section 198 under Chapter 4, Part VIII of the CMA contains
provisions on consumer protection including the following principles on rate setting:
(a) rates must be fair and, for similarly situated persons, not unreasonably
discriminatory;
(b) rates should be oriented toward costs and, in general, cross-subsidies should be
eliminated;
(c) rates should not contain discounts that unreasonably prejudice the competitive
opportunities of other providers;
(d) rates should be structured and levels set to attract investment into the
communications and multimedia industry; and
(e) rates should take account of the regulations and recommendations of the
international organisations of which Malaysia is a member.
As explained in the PI Paper, the MCMC views that it is required to undertake a Public
Inquiry under section 55 of the CMA in order to set prices for facilities or services in the
Access List because determination of these prices is very likely to be of significant interest
to all sectors of the economy, including providers and potential providers of these services
as well as end users of communications services.
The MCMC is now required to make any determinations arising out of this Public Inquiry
no later than 4 January 2018, which is 45 days after the close of public comments on the
PI Paper. The MCMC proposes to issue a new Commission Determination that will reflect
the MCMC’s final views as expressed in this PI Report in respect of the pricing of some of
the facilities and services in the Access List for the period 2018 to 2020.
The MCMC has consulted widely and openly with all interested stakeholders during this
Public Inquiry, including:
(a) informal consultation with a broad range of licensees prior to the release of the PI
Paper, as set out in Section 1 of the PI Paper;
Page 7
(b) presentation of the economic cost models to licensees during June 2017 and
consideration of comments received;
(c) publication of the PI Paper on 6 October 2017 and a request for comment, including
publicity on the MCMC website;
(d) making available on request public versions of the economic cost models used in
determining the proposed prices in the PI Paper; and
(e) clarifications in response to stakeholders in relation to specific items raised in the
PI Paper during the consultation period.
At the close of the Public Inquiry period at 12.00 noon on 20 November 2017, the MCMC
had received written submissions from the following parties.
Page 8
No. Submitting party Documents
Having thoroughly reviewed and assessed the submissions received on the PI Paper
against its own preliminary views, the MCMC now presents this PI Report within the 30-
day requirement of the closing date of submissions, as stipulated under section 65 of the
CMA.
The remainder of this PI Report is structured broadly to follow the PI Paper to provide a
consistent context for the MCMC’s specific questions for comment. The 32 questions in
the PI Paper are duplicated in each section with a summary of the comments received (in
alphabetical order of the submitting parties). The MCMC then sets out the rationale of its
final views on each issue:
Page 9
2. Principles in Setting Access Pricing
2.1. Overview
Part B of the PI Paper was concerned with the general principles relevant to regulatory
pricing. After some brief background on the legislative objectives in Section 5.1, Section
5.2 of the PI Paper sets out the general guidelines to be used by the MCMC in determining
which services should be subject to ex-ante price regulation and the criteria by which
regulated prices should be set. These included the recovery of appropriate costs and the
promotion of economic efficiency. In addition, in Section 5.2.3 of the PI Paper, a time
horizon of 3 years was proposed for the regulated prices.
Question 1:
Do you think that the criteria for ex-ante determination of access prices presented
remain appropriate?
Submissions received
Altel, APCC, Celcom, Digi, Maxis, Sacofa, TM, U Mobile, webe and YTL believe that the
criteria for ex-ante determination of access prices remain appropriate, while PPIT
disagreed with the MCMC’s reasoning for establishing ex-ante regulation.
APCC pointed out that the criteria test used by the MCMC for determination of ex-ante
prices is also the test applied by the European Commission and Ofcom. They believe that
the inclusion of non-transitory and high barriers to entry is necessary because the facilities
and services in the Access List are generally services which have high prices due to
inherent monopolies. This is particularly so for the provision of core network facilities
where a state sanctioned monopoly has been appointed. In such instances, regulatory
intervention is crucial. APCC also agreed to the inclusion of “where there is no trend
towards effective competition” as a criterion for ex-ante regulation. This is especially true
in Malaysia which imposes high foreign equity restrictions as a condition for acquiring
licences.
Celcom supports the position in the PI Paper that emphasises preference for commercial
negotiations over regulatory intervention. Only in cases where commercial negotiations
result in undesirable outcomes, maximum price regulation is needed. Celcom pointed out
Page 10
that there have been various successful commercial negotiations for MVNO Access,
Infrastructure Sharing, Duct Sharing and RAN Sharing on a reciprocal basis. In the mobile
market, Celcom recommends regulation only for mobile termination. However, in the fixed
market, obtaining access to facilities such as duct and manhole and high speed broadband
service have been challenging and warrant ex-ante price regulation.
Digi is of the view that setting maximum prices for essential facilities in the Access List
would help provide commercial certainty in the market and aid commercial negotiations.
Regulatory intervention in the form of ex-ante regulation is significant in the case of
essential facilities or services with high barriers to entry, when market structures do not
tend towards effective competition or where ex-post regulation is insufficient to adequately
address the identified market failure. This is particularly important for wholesale services
such as termination, transmission-related services, fixed access services, line sharing
services, HSBB services and bitstream services.
Maxis highlighted that in the fixed broadband market, the incumbent operator holds a
strong market share and has full control of all the essential facilities and/or services.
Therefore, there are key risks that the Access Provider may deliberately prolong
commercial negotiations in order to protect its market share and to gain unfair first mover
advantage. Such a situation makes it difficult for the Access Seeker to become competitive
and ex-post intervention may not be sufficient. However, in competitive markets like
mobile services, commercial negotiation should continue and ex-ante regulation for
services such as MVNO access is not appropriate.
MYTV understands the objective of price regulation and the MCMC’s role to safeguard
public interest and promote competition. However, MYTV views that the DTT services are
to be provided by MYTV to broadcasters or CASPs only and does not affect the general
public. MYTV opines that the MCMC’s intervention to regulate prices to ensure terms and
conditions are reasonable will create a stifling business environment to both MYTV and
CASPs. Instead, MYTV and CASPs should be allowed to engage in commercial negotiations
as MYTV and CASPs may have different commercial interests from one another. MYTV
highlighted that DTT services are their only source of income, while CASPs have multiple
sources such as advertisements, sponsorships, pay TV subscriptions and content royalty
fees, which they enjoy without any intervention on prices. MYTV also has to contend with
commercial negotiations with TM as the service is not subject to price regulation. On the
one hand, MYTV is hampered with high commercial costs from TM, while on the other
hand, its revenue is supressed with price regulation. MYTV pointed out that the terms and
conditions, including prices of other licensees such as Measat Broadcast Network System
and TM Net are not subject to any regulation and MYTV believes that they should be
accorded the same treatment. In conclusion, MYTV indicated its preference for commercial
negotiations.
Page 11
PPIT is of the view that the MCMC’s approach in using the market dominance study carried
out in 2014 to invoke price regulation is not just and equitable as only four out of thirteen
state backed companies (SBCs) were found to be dominant. SBCs have been promoting
sharing of infrastructure on an equal and non-discriminatory basis. Even the dominant
SBCs such as Sacofa have adopted a collaborative approach when it comes to commercial
dealings. PPIT also pointed out that the MCMC has failed to consider the countervailing
buyer power of the telcos. PPIT also highlighted that no other country in the world has
deemed it appropriate to mandate prices. PPIT cited countries such as India, New Zealand,
USA, Hong Kong and Brazil, Sweden, France, Singapore, Norway, Germany, Jordan, the
Netherlands, China and Canada as examples of countries that have not mandated prices.
TIME highlighted that while the rationale for ex-ante regulation remains, the industry
continues to be burdened as the costs do not reflect specific Access Provider’s costs,
especially WACC. Industry also submits reports on Accounting Separation that provide
the MCMC with detailed information as to whether there is price discrimination and there
are statutory provisions that prohibit tying and linking. As such, the MCMC is urged to
review its access instrument vis-à-vis the regulatory burden it imposes onto the industry
players.
TM noted that the intent of ex-ante regulation is to produce prices consistent with a
competitive market. As such, any calculations should be forward-looking rather than
backward-looking.
webe believes that it is important that the costing exercise considers all aspects of cost
and impact. The pricing should not deprive the desire for investment and should continue
to encourage service providers to invest in network expansion. At the same time, careful
measures are required to reduce investment redundancy and enhance competition to
increase network efficiency.
Discussion
The MCMC has considered carefully the comments and issues raised by the respondents
to the PI Paper. The responses have assisted in confirming and clarifying the principles
that should apply to the setting of prices for the services in the Access List.
With all ex-ante regulatory price controls, a balance needs to be struck between the needs
of end users and the needs of service providers. At this point in time, the more rapid
Page 12
increase in demand for data services with a consequential reduction in demand for voice
services needs to be recognised. The MCMC believes that it has used its best endeavours
in striking the appropriate balance between the differing needs throughout this process.
The MCMC notes the comments from respondents regarding the criteria for the imposition
of ex-ante regulation and the general agreement that the criteria for ex-ante
determination of access prices remain appropriate. The MCMC recognises the desire for
agreements to be reached through commercial negotiations where this is possible;
however, the overall objective of imposing regulatory remedies where market failure is
identified is supported by the proposed criteria:
The MCMC therefore feels that it is important to impose ex-ante price regulation on the
services in the Access List which meet these criteria.
In response to Celcom’s concerns, the MCMC notes that not all operators, especially the
smaller ones which do not own infrastructure and duct, would be in a position to enter into
reciprocal arrangements concerning Infrastructure Sharing, Duct Sharing and RAN
Sharing. In consideration of comments from TIME, TM and webe, the MCMC agrees that
the forward-looking costs of a reasonably efficient operator are the most appropriate and
this is why the LRIC+ costing methodology has been chosen and appropriate WACCs
determined.
PPIT’s concerns that the use of the market dominance study carried out in 2014 to invoke
price regulation is not just and equitable, there is a countervailing buyer power of the
telcos and that dominant SBCs have adopted a collaborative approach to commercial
arrangements are well noted by the MCMC, as well as the information about countries
where no such price regulation is imposed.
The MCMC notes MYTV’s concerns about its position in the market and its inability to secure
alternative revenue streams. However, the DTT broadcast service meets the key criteria
identified, and as such, the MCMC believes that there is a clear need for price regulation.
Question 2:
Do you think that the approach to pricing which has been adopted is appropriate? Are
there any other criteria that should be considered?
Page 13
Submissions received
All submissions, with the exception of TIME have expressed support for the approach to
pricing that has been adopted by the MCMC.
APCC agrees with the establishment of price ceilings via regulated price control and it does
not believe that ex-post remedies will be sufficient. APCC also supports the MCMC’s
approach in balancing the principles of appropriate cost recovery and the need to promote
economic efficiency in investments.
Celcom is in broad agreement with the criteria proposed by the MCMC but pointed out the
models may not have achieved these objectives.
In respect of the fixed model, despite the stated objectives, Celcom noted a number of
characteristics which run contrary to the objectives as follows:
(a) The access network has not been optimised sufficiently. The MCMC has used actual
data provided by the incumbent instead of estimating the costs of an efficient
operator;
(b) For certain services, there is a limited scope for market entry and results in
unjustifiably high prices for access seekers and end users;
(c) There is an understatement of demand growth for broadband services and this may
result in misallocation in costs and result in inefficient and distortive pricing;
(d) There is a lack of clarity on categories of cost included in determining charges and
this hinders the consultation process, which is essential to ensure a fair outcome;
and
(e) Some of the charges are higher than the current charges and no explanation has
been provided. Celcom cited interconnect links as an example.
In respect of the mobile model, Celcom is concerned with the large number of conceptual,
technical, mathematical and referencing errors in the model. There appears to be no
process for an independent check of the models.
Digi highlighted that it is exceptionally important that the information data used to derive
the wholesale prices reflect reasonable costs. This will ensure the long-term interest of
customers and send a positive encouragement to the mobile industry to continue its
interests for investment and reinvestment.
Maxis believes that the adopted principles should lead to a level of proposed prices by the
MCMC that is workable within the industry and will benefit the Access Seekers and
ultimately promote the Long-Term Benefit of End Users. Maxis welcomed the MCMC’s
Page 14
explanation that stranded and legacy assets have been excluded from the fixed model.
However, Maxis is of the view that the proposed regulated prices for services such as Fixed
Termination, End-to-End Transmission and HSBB Services are still too high. Maxis pointed
out that the proposed regulated price for Layer 3 for HSBB Network Service with Network
Service is higher than the retail price offered by TM and End-to-End Transmission prices
are higher than market prices in Malaysia. Maxis proposed that if the cost models do not
produce workable regulated prices, the MCMC should consider a retail minus approach.
While PPIT agreed with the criteria in the PI Paper, they believe that proposing regulation
solely based on a market dominance study is inappropriate. They also highlighted that on
average, passive infrastructure leasing costs only comprise about 6% of the local
operator’s operating costs and price regulation is not going to have a material impact to
consumers. PPIT also pointed out that SBCs are required to publish Reference Access
offers (RAO) and they have been negotiating commercially. To date, there has not been
any significant disputes or arbitration required to mediate any commercial misalignment.
SBCs have also maintained Access Agreements and Master Licence Agreements that are
transparent and fair to all access seekers. Since 2005, there have not been any major
price revisions in spite of rising costs of infrastructure provision and maintenance. In
addition, discounts are granted for telcos whose agreements have entered into their 8th
years and there is an ongoing negotiation for additional discounts for post 10th year’s
commercial terms. PPIT vehemently disagreed with the comments in the PI Paper that
unreasonable price terms and conditions could potentially force telcos to exercise market
power.
TIME pointed out that the approach adopted by the MCMC is not appropriate and proposed
that the MCMC takes into consideration other factors such as types of technology, size of
bandwidth, contract tenure and volume purchased.
TM requested the MCMC to be mindful of ensuring incentives are maintained for efficient
long-term investment. TM highlighted that the type of investment undertaken by fixed
operators are typically capital intensive, involving significant proportions of long-lived
assets. To embark on such projects, fixed operators must have confidence that they will
achieve a fair rate of returns over time. TM pointed out that the MCMC has altered its
approach to costing wholesale access services over the last three regulatory periods and
this led to uncertainty and has resulted in TM not recovering its cost in the medium to
long-term.
webe opines that it is important to consider the latest technology employed when deriving
dynamic efficiency to motivate service providers to adopt the latest available technology.
Page 15
YTL believes that the discrepancy between on-net and off-net pricing has a tendency of
confining users to larger networks with bigger customer base. Hence, an adjustment
would encourage users to switch to networks that currently have smaller base.
Discussion
The MCMC notes the agreement of almost all respondents to the proposals that the
regulated prices should be based on the key criteria of:
The MCMC notes Celcom’s comments regarding the fixed network model but confirms that
the data used within the model has not been taken at face value from TM and has been
subject to variations to bring it in line with the expectations for an efficient fixed network
operator. This applies to actual costs including the exclusion of legacy assets, price trends,
service volumes, service volume forecasts and international benchmarks, where
appropriate. The comments from Celcom and Maxis regarding proposed regulated prices
being higher than current prices are dealt with in connection with individual services later
in this PI Report. The MCMC notes Celcom’s comments about errors within the mobile
model and also deals with these specifically later in this PI Report.
In response to TIME’s comments that regulated prices should reflect issues such as types
of technology, size of bandwidth, contract tenure and volume purchased, the MCMC
considers that the use of the LRIC+ costing methodology adopted calculates the costs of
an efficient operator using the most appropriate forward-looking technology to calculate
the costs of services. Such costs should therefore be applied on a non-discriminatory basis
and not subject to discounts based on contract duration of volumes purchased.
The costing methodology adopted by the MCMC at each stage of implementing regulatory
price controls aims to permit appropriate cost recovery over the implementation period.
The move to the LRIC+ methodology for the wholesale Access Network Services should
therefore continue to support the necessary long-term investment decisions of TM or other
network operators.
The MCMC acknowledges the comments from PPIT regarding the successful use of
commercial negotiations and agreements, long-term price stability as well as the lack of
disputes or need for arbitration.
The impact of on-net and off-net tariffs in mobile networks is recognised as a reason for
customers to be attracted to larger networks as noted by YTL. The MCMC believes that
Page 16
adoption of LRIC+ termination rates will help to bring all services’ prices in line with costs
and help to reduce such discrepancies.
Question 3:
Do you have any comments on the appropriateness of setting regulated prices for the
period up to and including 2020?
Submissions received
The views about setting regulated prices for three years’ time horizon were divided. On
the one hand, some operators agreed with the three-year time horizon and cited the fast
evolving technology and business and regulatory certainty as reasons for maintaining the
three-year period. On the other hand, TIME and YTL proposed a 5-year time horizon,
while one operator proposed a six-year time horizon.
APCC submitted that the three-year period is in line with best practices and suggested
that the MCMC should expressly maintain the option to intervene before the end of the
three-year period if the situation calls for it.
Celcom said that it is in broad agreement with the proposed time horizon if the MCMC
would consider a shallow glide path of the rates as this would soften the impact on
industry. Celcom expressed support for the MCMC’s decision to change the depreciation
method from economic depreciation to tilted annuity. Celcom also stated that setting
prices up to 2020 requires reasonable forecast and expressed dissatisfaction over the use
of dummy data in the fixed model.
Maxis agreed with the three-year time horizon but proposed another alternative of a six-
year time horizon as it will provide longer certainty for network operators and services
providers. Maxis also proposed to include a caveat in MSAP that the regulated prices will
continue to apply in the event that MSAP expires before a new determination is issued by
the MCMC.
PPIT believes that the time horizon proposed by the MCMC is short and unreasonable and
requested a longer time horizon. A refresh of price estimates over a short period would
compel SBCs and other tower operators to adjust their prices depending on current cost
structures. The increased investment towards more innovative and environmentally
friendly passive infrastructure such as bamboo towers and carbon fibres could well reduce
the current costs of SBCs in the future, resulting in under-recovery of committed historical
costs.
Page 17
PPIT also pointed out that a CAPEX intensive business model takes a longer payback period
of around 8 to 10 years. Therefore, the rate of return for telecommunications towers is
rarely calculated at a short period of three years. SBCs have generally been using seven
years as its basis for rate of return. The towers are predominantly passive infrastructure
and their cost elements will not have meaningful variations, except in instances involving
the land the tower is built on. The common business model for telecommunications towers
is 10 years or more. The licence agreements between SBCs and mobile operators have 8
to 10 years’ tenure.
TIME proposed a validity period of at least five years as it has been the MCMC’s practice
in the past to extend the timeline of MSAP. TIME also believes that the three-year time
horizon could affect the company’s overall business and affect investors’ view of the
telecommunications industry.
TM agrees with the three-year period and highlighted that some regulators reset prices
annually. Since regulatory prices are based on demand forecasts, major discrepancies
between actual and forecast can lead to inefficient pricing that will undermine the
regulatory objectives. TM believes that the MCMC should progressively move towards
non-price regulation for competitive services such as competitive transmission routes.
YTL believes that a five-year period may be more appropriate as the revisions of RAO and
execution of access agreements take time. For services that are subject to fast evolving
technology, a glide path or scale rates would be appropriate.
U Mobile suggested that the proposed mobile termination rates for 2018 be applied
retrospectively.
Discussion
The MCMC recognises the differing views from the respondents but needs to balance the
impact of the more rapidly changing technologies with the need for long-term stability
within the sector. At this point in time, the MCMC believes that the potential impact of
changes in customer behaviour and demand for services means that a three-year
regulatory period is more realistic and will mitigate concerns expressed about the accuracy
of forecasts in the trends for voice and data services in both the fixed and mobile networks
in Malaysia.
The comments from PPIT regarding the longer term investment programmes associated
with the provision of tower services are noted; but the MCMC believes that such long-term
requirements can be accommodated, even if the current price regulation only covers a
three-year period, primarily to cater for the demand driven services.
Page 18
The proposal from Celcom for a shallower glide path would lead to ongoing over-recovery
of costs by the mobile operators and therefore not an arrangement which the MCMC would
support. Similarly, the proposal by U Mobile for the rates to be applied retrospectively
does not follow international practice and presents unforeseen financial obligations on all
operators across the industry.
On balance, the MCMC believes that the three-year time horizon presents the most
appropriate time horizon given the uncertainties highlighted by respondents.
The MCMC is convinced that the criteria that was used for access pricing continue to be
relevant. In addition, the MCMC continues to view that it is appropriate to set regulated
prices until 2020
Page 19
3. Cost Modelling General Issues
3.1. Overview
Sections 6 and 7 of the PI Paper described the methodologies the MCMC was proposing to
adopt in order to determine cost-based prices. The PI Paper sought comments on the
structure and use of each methodology.
Section 6 of the PI Paper was concerned with the available costing methodologies including
FAC, LRIC, LRIC+, pure LRIC, SAC and Step-by-Step. It proposed the continued use of
LRIC with common cost mark-up (LRIC+) as the basis for setting prices for fixed core and
mobile services. It also proposed to use the LRIC+ approach, with asset price adjustments
to reflect the presence of fully-depreciated assets for the fixed access services. It proposed
a bottom-up model based on current asset costs for Digital Terrestrial Broadcasting
Multiplexing Service and Infrastructure Sharing.
Section 7 of the PI Paper proposed the use of bottom-up models incorporating titled
annuities as the appropriate depreciation method for all services, the allocation of common
costs using Equal Proportionate Mark Up (EPMU), the inclusion of licence and spectrum
fees, the definition of relevant increments, the use of the scorched node approach, model
calibration and reconciliation, the use of glide paths and the relevance of the cost models
for arbitration of disputes. Section 7 also discussed exceptions and adjustments to LRIC-
based prices relating to co-location, duct, manhole and infrastructure sharing, HSBB and
USP subsidies and installation charges.
Question 4:
Submissions received
Celcom proposed that the MCMC use a step-by-step approach that was previously adopted
for the fixed access network. The bottom-up approach is appropriate only for contestable
elements of networks where there is a prospect of entry from an actual or potential
competitor as it sends the right build or buy signal. For non-contestable network elements,
a top-down approach should be used with prices based on historical cost. The bottom-up
approach adopted by the MCMC should adjust for any windfall gains which arise as a result
of moving from a top-down historical cost model to a bottom-up CCA model using tilted
annuities. Since a large proportion of the incumbent’s assets are old, it means that moving
Page 20
from an accounting approach to tilted annuities will allow the incumbent to over-recover
its costs.
Celcom noted that the MCMC had excluded the fully-depreciated assets. However, apart
from that, no other adjustments have been made. Celcom views that adjusting only for
fully-depreciated assets is insufficient.
Celcom pointed out some inaccuracies in the PI Paper. The MCMC had stated that Ofcom
had used Stand Alone Costs in the UK, when in actual fact Ofcom had used Distributed
Stand Alone Costs. The PI Paper also stated that regulators do not use top-down modelling
of regulated firms. While it is true that regulators do not build top-down models, some
regulators have used, with modifications, the output from top-down models to set charges
in the access network.
Digi is of the view that BULRIC+ is the appropriate costing methodology for fixed and
mobile interconnection services. There is a range of benefits of LRIC+, including allowing
efficient access providers to fully recover the costs of providing services and promoting
legitimate business interests of the providers. The outcome from this approach will enable
the mobile operators to invest and reinvest to expand coverage and ensure service
reliability and data quality.
Digi considers BULRIC+ to be more appropriate for the fixed access network as compared
to the Regulatory Asset Base approach. However, since the fixed model was populated
with dummy data, Digi found it challenging to analyse and provide valuable feedback as it
was unable to verify model calculations and comment on the reasonableness of forecast
assumptions.
On Infrastructure Sharing, Digi agrees with the MCMC’s views that the different models in
which the infrastructure may be shared requires different treatment. Digi also supported
the inclusion of “infrastructure to be shared by a stand-alone business” into the Access
Pricing as it will ensure cost-based wholesale prices and avoid excessive prices that will be
detrimental to the long-term interest of customers.
Maxis fully supports the use of the LRIC+ methodology for mobile services, as it provides
a reasonable balance between ensuring that the Access Seekers pay only for the costs
incurred by a reasonably efficient operator. LRIC+ is suitable in the Malaysian context as
it allows for the mark-up of fixed and common cost. European Regulators prefer pure
LRIC as historically, mobile termination prices have been significantly higher than the rates
in Malaysia.
Page 21
Maxis also supported the use of LRIC+ for fixed services. However, concern was raised
about the output of the Fixed Core model which is still high compared to other countries,
particularly the European countries such as UK, the Netherlands, Belgium, Germany and
Spain. Maxis also benchmarked fixed termination rates against mobile termination rates
and pointed out that in other countries such as Switzerland, Oman, Saudi Arabia, New
Zealand, Indonesia, Australia, Germany, Spain, Belgium, the Netherlands and the UK,
fixed termination prices are usually significantly lower than mobile termination rates. This
contrasts starkly with the situation in Malaysia.
Maxis highlighted that the proposed regulated prices for Layer 3 HSBB Network Service
with Network Service are substantially high and are close to TM’s retail prices. For the 10
Mbps service, it is higher than TM’s retail prices. In the absence of data in the Fixed Cost
Model, Maxis noted that it is unable to verify the data input, assumptions and network
dimensioning used by the MCMC.
For Instructure Sharing, Maxis is agreeable to using a bottom-up model. However, Maxis
disagrees with the MCMC’s approach of using purely current asset costs. Instead, the use
of combined/blended historical asset costs and current asset costs were proposed. This is
important to reflect the actual cost for most of the existing towers that have been built
many years ago. Maxis also questioned the scale of the Tower Operators used by the
MCMC and proposed to test sensitivity on the impact of volume as tower companies vary
significantly in size and scale.
Maxis stated that the output of the Tower Model is not much different from the existing
prices. Maxis highlighted that the SBCs commonly give certain amount of discounts to
Access Seekers after several years and proposed that the same approach is adopted by
the MCMC. Maxis also proposed that the MCMC review and regulate prices of other types
of towers such as monopole, monopole tree etc. The Tower Operators are also building
BTS Hotel Solutions using lamp poles and the MCMC is urged to regulate prices of these
types of towers. Further, there are fewer lattice tower sites built by SBCs nowadays.
Finally, Maxis also suggested the regulation of prices for in-building coverage antennae
(IBCA) system in order to control exorbitant prices imposed by IBCA providers such as
KLIA2.
TIME believes that access prices should only be applicable to dominant players or players
with sufficient market power.
TM recommended the use of pure LRIC for mobile services. There should be no common
costs mark-up as these can be recovered from retail services. European countries such
as Austria, Belgium, the Czech Republic, Denmark, France, Germany, Greece, Italy,
Portugal, Spain, Sweden and the UK were cited as examples. TM is facing a different
Page 22
market situation as fixed voice revenues are declining due to increased use of OTT
applications. The fixed operator in Malaysia is also constrained in its ability to recoup
common costs at the retail level. TM pointed out that the MCMC has also incorporated an
inefficient and outdated technology in its mobile LRIC model (namely 2G), thereby
artificially increasing the price of mobile services. For the fixed services, the MCMC has
arbitrarily written down some of TM’s “old” fixed assets while there are no such markdowns
for old 2G equipment in the mobile model.
The MCMC’s proposed approach to use LRIC+ for fixed access services with some asset
prices at current costs, while others are subject to price adjustments to reflect fully-
depreciated assets imply that TM can afford to price wholesale products at relatively low
levels, thereby creating barriers to new infrastructure market entry. However, a key
purpose of ex-ante regulation is to address entry barriers. TM demonstrated that without
markdowns, fixed termination rates only increased by less than 10%, while there was
significant increase of more than 40% in Layer 3 HSBB with network service. The key
rationale for use of LRIC is to determine efficient prices for uncompetitive or bottleneck
service but the MCMC’s proposed approach encompassing written down values of “old”
assets cannot deliver prices that objectively reflect efficient cost. In practice, the MCMC’s
decision to depart from the traditional LRIC approach has led to significant write down of
many network and non-network assets that make up a substantial portion of total cost
(30-35%). The MCMC’s approach of arbitrarily selecting assets and writing them down by
as much as 90% in order to match 2016 estimated depreciation with actual depreciation,
does not provide any allowance for work-in-progress in 2016 for future investment during
the regulatory period. Such an approach will have a chilling effect on investment and is
completely contrary to the Long-Term Benefit of End Users (LTBE).
TM pointed out that it is not uncommon to mark-down selected asset values to reconcile
LRIC results with historic book value depreciation. For example, the European Commission
had adopted a modified approach to asset valuation but this approach was rejected by the
regulator in New Zealand on the grounds that it was irrelevant for the New Zealand
context. TM pointed out that similarly, it is irrelevant for Malaysia as it is in a similar
situation to New Zealand where the Government has committed to a Public Private
Partnership (PPP) arrangement with fixed network providers to deploy high speed
broadband infrastructure. The MCMC has followed the European approach but the
European Commission’s recommendation is only for civil assets.
webe proposed that the technology in forward-looking mobile model should be consistent
with the fixed model where legacy costs were excluded. Other inconsistencies webe noted
are as follows:
Page 23
For fixed termination, voice traffic is not split by the use of submarine cable as the
cost of submarine component is not significant;
For Interconnect Link Service, the costs are increasing due to increasing costs of
cables and related civil infrastructure;
Trunk Transmission Service – because of significant cost of submarine cable, the
Trunk Transmission Service is separated between Peninsular Malaysia and Sabah
and Sarawak. The proposed pricing is decreasing because the decreasing cost of
transmission equipment and increasing traffic in core network. The pricing for
installation is increasing because of increasing labour cost.
End-to-End Transmission Service – because of the significant cost of submarine
cable, the transmission price is split between within Peninsular Malaysia and Sabah
and Sarawak.
YTL agrees with the methodology. However, it appears from the manual that 4G services
using the 2.3GHz and 2.6GHz TDD LTE spectrums have not been included in the costing.
Discussion
Respondents have presented a range of views about the proposed costing methodologies
based on bottom-up and LRIC+ methodologies. In determining the appropriate costing
methodology, the MCMC has sought to promote efficiency and competition whilst
representing the best interests of both the end users as well as the network operators.
The adoption of a consistent approach to modelling both fixed and mobile technologies is
important and it should be recognised that where pure LRIC has been adopted, this has
been applied to both mobile and fixed technologies.
Celcom disagrees with using a bottom-up approach for the fixed access network and
proposed two alternative solutions to modelling the fixed access network. One involved
using a top-down model of TM’s fixed access network with historical values and the second
using a step-by-step approach.
A top-down model of TM’s access network would also be based on real data from TM’s
access network. The MCMC therefore does not perceive that this approach would be
methodologically better than the approach used by the MCMC. The MCMC assumes that
the only reason why Celcom would prefer a top-down model with historical values is that
it would lead to lower costs of access network services. On the other hand, Celcom agrees
to use a bottom-up model with current values for the core network (which also leads to
lower costs of services). This appears to be cherry picking. The MCMC is of the view that
the methodology should not be selected based on the desired level of resulting costs and
that it is fair to use the same methodology consistently for modelling both the fixed access
and core networks.
Page 24
The MCMC has also decided not to use the step-by-step approach, as though it is
theoretically correct, its practical implementation requires arbitrary assumptions to be
made. Firstly, under this approach, the costs of the access network are equal to the
maintenance and replacement expenses for the access network. However, these expenses
fluctuate significantly between years and not every access network element is maintained
or replaced every year. So, average values over a long enough period, which is longer
than the modelling period, would be necessary. Long-term estimates of future
maintenance and replacements in the fixed access network are highly arbitrary and cannot
be properly verified by the MCMC. Therefore, the MCMC prefers to base its model on the
number of physical units of network equipment (for example, kilometres of cables) which
are obtained from the network inventory systems and can be verified.
The second problem with the step-by-step approach is that it does not calculate costs
separately for individual network elements and therefore it is difficult to derive distinct
costs for different services on the basis of cost causation and resource usage. Celcom
argues that the difficulty in allocating assets to elements of the network or to services,
which allegedly make it difficult to derive service-specific costs, can be addressed on the
basis of usage or other causal factors. The MCMC agrees that allocating assets to services
can be done on the basis of usage. However, the problem with the step-by-step
methodology is that the costs of different assets are mixed together into one overall figure.
A single usage factor for a mix of assets does not allow precise allocation to services, which
differ in their usage of the individual assets.
The MCMC accepts Celcom’s argument that due to the manner in which straight line
depreciation and tilted annuity work, it results in the following. In the earlier years of an
asset’s lifetime, the annuity charge is lower than the straight line depreciation plus cost of
capital; and it is reversed in the later years. Hence, by adopting straight line depreciation
plus cost of capital in the early years of an asset’s lifetime and the annuity in the later
years means that the higher cost options are chosen. Celcom adds that, because the fixed
access network assets are more than halfway through their lifetime, moving from the top-
down approach (which uses straight line depreciation plus cost of capital) to the bottom-
up approach (which uses annuity) will allow operators to over-recover their costs. The
MCMC agrees with Celcom but stresses that though the fixed access network prices were
calculated based on a step-by-step approach with straight line depreciation, the prices
were finally not regulated.
The MCMC notes Celcom’s revisions to reflect the minor inaccuracies in the PI Paper.
The MCMC notes the comments from Maxis regarding the relationships between
benchmark fixed and mobile rates in different parts of the world. These have resulted
from the use of different costing methodologies as well as levels of fixed network
Page 25
deployment and utilisation over time. However, there is a clear trend towards reducing
prices for voice termination in all networks with the potential for a move to bill and keep
at some point in the future.
The MCMC notes the statement from Maxis that the prices for HSBB services are high.
However, the MCMC does not agree that HSBB service costs are higher than the retail
prices offered by TM to its end users. Maxis reached this conclusion by adding the HSBB
installation costs to the HSBB monthly rentals and splitting the installation costs over only
24 months. However, many customers keep their broadband line for a longer period of
time so this calculation is not fully representative.
The MCMC notes the comments from TIME that such regulated rates should only apply to
dominant operators, as in the case in many other jurisdictions. However, the MCMC would
like to clarify that the legal framework under the CMA does not permit asymmetric
regulation.
The MCMC recognises TM’s situation, as is the case with both incumbent and mobile
operators around the world, where volumes of network-provided voice services are
reducing and replaced by OTT applications, but with a counterbalancing increase in
demand for data services over such networks. The LRIC models developed by the MCMC
take account of this change and apportion the appropriate amount of network
infrastructure and cost to the respective services.
The reconciliation process adopted by the MCMC aligns the asset costs predicted by the
fixed access network model and the financial records of TM to ensure that there is no over-
recovery of costs. As the model is forward-looking, costs which TM currently recognises
as work in progress and future investments are taken into account as necessary in future
years and are therefore not excluded from the model and do not therefore have “a chilling
effect on investment”. Future investments in HSBB through the PPP arrangement are also
dealt with through this process. A similar asset-based calibration was also carried out on
the mobile model. However, the MCMC did not identify the need to exclude or significantly
reduce value of certain asset types.
The MCMC notes that TM disagrees with the reconciliation process that was used which
resulted in a mark-down of TM’s assets. There was some misunderstanding and the MCMC
clarifies as follows. The mark-down does not try to bring the depreciation calculated in
the model in line with TM’s actual depreciation charge. The mark-down tries to bring the
Gross Book Value (GBV) reported by TM with the actual depreciation charge reported by
TM. The calculation of the mark-down was done in the following manner.
Page 26
TM provided the GBV of different asset groups and also the lifetimes of these assets in the
data collection. The MCMC then calculated straight line depreciation from these GBVs
(dividing the GBV by the asset lifetime, based on values provided by TM) and compared
the total depreciation calculated in this manner with the total depreciation charge reported
by TM. Consequently, it was observed that TM’s reported depreciation charge was only
half of the result obtained through the calculation. The only explanation of this situation
is that there are some fully-depreciated assets, which are still in use. Such assets are
included in the GBV but there is no depreciation charge calculated for them. The MCMC
then calculated the percentage of these fully-depreciated assets by lowering the GBV by
the mark-down so that the GBV divided by the asset lifetime is equal to the actual
depreciation charge reported by TM. The percentage of mark-down was not set equally
across all asset groups, but the MCMC selected some asset groups where it is more likely
that there will be a high level of fully-depreciated assets and adjusted the GBV by mark-
down only in these asset groups. If the MCMC applied the mark-down equally across all
asset groups, the percentage of mark-down would have been 50% because the total
depreciation charge reported by TM was only half of the total calculated depreciation.
Furthermore, the MCMC would like to stress that all inputs into this calculation are values
provided by TM and therefore the percentage of mark-down is not dependent in any way
on the model calculations. Even if the model calculated a completely different number of
units of fixed asset elements and different annual costs of these assets than in the current
model calculation, the mark-down would still be the same.
The MCMC does not agree with webe’s comment that legacy assets were excluded in the
fixed model. The MCMC clarifies that fully-depreciated assets were excluded, but not all
legacy assets, in the fixed model.
On webe’s query over the difference of treatment of the submarine cable to fixed voice
services and mobile voice services, this will be discussed in greater detail in Questions 19
and 24 respectively.
webe also pointed out the high costs of Interconnect Link Service and the change from the
previous price. This was caused by the fact that the Interconnect Link Service was
calculated as an independent, dedicated fibre cable using completely new routes. Based
on the comments from the respondents, the MCMC realized that this is not realistic and
Page 27
has revised the costing approach for the Interconnect Link Service. The discussion on this
will be provided in Question 19.
The MCMC notes the comments from YTL about the use of 4G spectrum which are dealt
with in Section 7.2 (under Question 21) of this PI Report.
Question 5:
Do you have any further comments on the elements of cost modelling which the MCMC
proposes to adopt?
Submissions received
Celcom, Maxis, PPIT and TM agrees with the MCMC’s decision to use bottom-up modelling.
Celcom believes that bottom-up models are more suitable for contestable markets as
opposed to non-contestable markets such as fixed access. The fixed access model as
constructed by the MCMC does not provide a reasonable basis for setting wholesale
services in Malaysia for HSBB or other access products. Celcom said that in Europe and
elsewhere, regulators are moving away from using bottom-up models to set wholesale
prices.
Maxis is agreeable to the MCMC’s decision to adopt a bottom-up approach but pointed out
that the key to accurate bottom-up model is accurate cost inputs and believable
assumptions and network design and dimensioning. Maxis requested the MCMC to revisit
and re-verify data inputs, cost inputs, assumptions and network dimensioning used in the
fixed model as the model produces high prices. Maxis also requested the MCMC to
reconcile or benchmark operators’ financial figures to avoid inaccurate estimates.
PPIT agreed with the use of a bottom-up approach and proposed to include some degree
of top-down reconciliation to better reflect the costs of an efficient operator.
Discussion
The MCMC notes that respondents generally support the use of bottom-up models. The
MCMC notes the comments from Celcom and Maxis regarding the suitability of bottom-up
models and data verification but believes that the cost reconciliation and corrections which
it has carried out to the final versions of the fixed access model provide consistent and
reliable service costings.
Page 28
Celcom disagrees with a bottom-up approach for the fixed access network and proposed
two alternative solutions to modelling the fixed access network. The MCMC has responded
to these comments in the previous Question 4.
Depreciation method
Submissions received
Celcom, Maxis, PPIT and TM support the MCMC’s decision to use tilted annuity
depreciation.
Celcom however, reserves its rights to comment on the actual implementation of the tilted
annuity formula.
Maxis agrees with the consistent approach across all models. For mobile models, tilted
annuities would represent more appropriate reflection of cost recovery since there is a
significant level of uncertainty in the future trends of voice and data services. However,
for the fixed network, where service demand hasn’t reached mature levels, tiled annuities
may artificially inflate unit costs. Maxis suggested that the MCMC undertake sensitivity
analysis using economic depreciation for fixed access and fixed core.
PPIT prefers tilted annuity method as a proxy for economic depreciation as this approach
takes account of the change of price trends.
Discussion
The MCMC notes the general support for the common use of tilted annuity depreciation in
all the cost models. The MCMC notes the comments from Celcom about the
implementation of the formula and has verified the method of calculation.
Maxis commented that for the fixed network, where service demand hasn’t reached mature
levels, tilted annuities may artificially inflate unit costs. The MCMC does not agree that
the demand in the fixed network has not reached mature levels. In fact, it is the converse.
Fixed network services are very mature and fixed voice services are already showing a
declining trend, which is common for the end of the service lifecycle. Maxis is probably
referring to the demand for FTTH services. However, from the perspective of demand, the
FTTH service is a replacement for the already mature ADSL service. This aspect is handled
in the fixed model, where it calculates the costs of these services based on migration from
ADSL services to FTTH services. This means that in areas where FTTH is implemented,
the customers are migrated to a fibre access network and the copper access network is
no longer used. The full and mature demand from ADSL services is transferred to the
Page 29
FTTH services in these areas, and hence, the FTTH network elements are highly utilised
from the start.
Submissions received
Celcom and Maxis support the MCMC’s decision to include appropriate common cost in the
cost model using the EPMU approach, while TM disagrees.
Maxis is of the view that it is important to recover common costs as these are integral to
business. Maxis also agreed with the MCMC that Ramsey price mark-ups are theoretically
ideal but difficult to implement.
TM believes that mobile operators can recover these at retail level, while TM is constrained
in its ability to do so. This is due to declining fixed revenues as consumers increase usage
of mobile and OTT.
Discussion
The MCMC notes the general support for the inclusion of common costs through EPMU
rather than some form of Ramsey pricing. The MCMC also notes TM’s view that common
costs should be excluded from mobile wholesale costs but is of the opinion that it should
adopt a technology neutral approach to regulation rather than discriminate arbitrarily
between network technologies. It is likely that voice revenues will decline in all networks
and all operators will be required to find alternative revenue sources.
Submissions received
Celcom and Maxis agree with the use of routing factors to allocate costs.
Celcom cautioned however, that the routing factors should be determined correctly. They
pointed out that the fixed model does not show the actual routing factors, preventing
operators from evaluating that these are applied correctly and are plausible.
Maxis believes that routing factors would reflect the use made of each network component.
Maxis also agree with the principle of cost causation adopted by the MCMC.
Page 30
Discussion
The MCMC notes the agreement to use cost causation through appropriate routing factors
within the models.
Submissions received
Celcom, Digi, Maxis and PPIT agree with the MCMC’s approach to include the cost of licence
and spectrum fees in the model.
Celcom pointed out that the model only includes spectrum costs currently incurred by
operators and there is no mention of spectrum fee for 700MHz. In addition, the amount
of spectrum fee paid by mobile operators is far in excess of the RM180 million used in the
model.
TM disagrees with the inclusion of licence and spectrum fees in the mobile model and
suggested that the MCMC removes allowance for such fees.
Discussion
The MCMC notes the general agreement to include the cost of licence and spectrum fees
in the models. The comments from Celcom regarding the inclusion of new spectrum bands
and the overall cost of spectrum are discussed in Section 7.2 (in Question 21) of this PI
Report. It is however the view of the MCMC that licences and radio spectrum (including
mobile services and microwave) represent a fundamental base upon which operators
provide both wholesale and retail services. As such, using the principle of cost causation,
licence and spectrum fees should be included in all relevant cost models.
Submissions received
Maxis and Celcom agreed with the MCMC’s decision to adopt a broad increment for the
cost models, expressed on an average basis and incorporating all the network services.
They are of the view that using marginal cost is inappropriate as it would not reflect costs
like fixed and common costs.
Page 31
Discussion
The MCMC notes that Maxis and Celcom are in agreement with the decision to adopt a
broad increment, expressed on an average basis and incorporating all network services.
They both agree that the use of marginal costing is not appropriate as this would exclude
costs such as fixed and common.
Submissions received
Celcom noted that the discussion in the PI Paper suggests that the scorched node
assumption has been used in the fixed network model for nodes containing MSAN and
other equipment higher up the network. It also states that equipment at any given node
can be optimised. Celcom highlighted two concerns with these assumptions. Firstly, it
appears that the lower level nodes, namely primary and secondary distribution points are
not optimised. Secondly, while the discussion recognises optimisation of MSANs and other
nodes, there is no indication of this having been done in the model.
There is no conclusion to the discussion on scorched node and scorched earth in the mobile
network. Celcom is of the view that scorched node should be applied and the number of
base station sites in the model should closely match the actual.
Digi submitted that modified scorched node should be adopted for the mobile cost model.
Digi also believes that clusters of population should be identified for the total Malaysian
territory using the most granular population figures from the Department of Statistics. It
is also important for the MCMC to ensure that the geo-types are associated with different
cell coverage radii based on spectrum allocation of the modelled operator. Digi agrees
with the MCMC’s approach to include licence and spectrum fees.
Maxis also supports the use of the scorched node approach as operators in Malaysia are
competitive and adopt best practice engineering in the design of network topology.
Scorched node allows for recouping of costs from appropriate existing nodes.
TM disagrees with the modified scorched node approach adopted by the MCMC for access
nodes in the fixed network where the number of aggregation nodes and rings decreases
with time. With this approach, the number of nodes accounts for coverage and not
demand. The assumption that the MSAN cabinets will be decommissioned due to fibre
expansion is unrealistic, at least within the modelling period. MSAN cabinets will still be
required to provide copper-based services despite the declining demand. TM used the
fixed model to calculate the change in service costs if a fixed number of access nodes is
Page 32
assumed and submitted that the impact on prices is significant. Therefore, TM suggested
that the MCMC keep the number of access and aggregation nodes fixed during the
modelling period.
For the mobile network, TM recommends the scorched earth pure LRIC approach, with no
common cost. When there are four operators, it is challenging to make assumptions on
where existing efficient sites should be located. As such, a modified scorched node or
scorched earth may be more practical.
Discussion
The MCMC does not agree with the statement from Celcom that in the fixed network model
the MCMC used the scorched node approach without optimisation of the nodes. All nodes
were adjusted to correspond to the number of access lines.
This is confirmed also by the comment from TM which says that the assumption that the
MSAN cabinets will be decommissioned due to fibre expansion is unrealistic, at least within
the modelling period. MSAN cabinets will still be required to provide copper-based services
despite the declining demand.
The MCMC remains of the opinion that MSANs should be decommissioned in FTTH areas.
It would be inefficient to have two parallel access networks in the same area – one copper
using MSANs and one fibre using OLT. The fixed model calculates the costs of these
services based on migration from ADSL services to FTTH services.
On the other hand, the MCMC accepts the point made by TM that in line with the scorched
node approach the number of higher level nodes (aggregation node, core node) should
not decrease with decreasing demand and the existing nodes should be modelled.
As discussed in the PI Paper, the mobile model is based on a notional operator with a 25%
market share rather than any particular existing operator in Malaysia. The reconciliation
has therefore been carried out based on the data which has been made available. As
noted by Digi, the assumption about cell radii reflect the need for both coverage and
capacity in the mobile networks and reflect the availability of radio spectrum in the various
bands.
Question 6:
Page 33
Submissions received
Celcom pointed out that the PI Paper leaves out a number of key issues. There is limited
optimisation in the fixed model and little if any in the access network where trench, duct
and cable lengths are taken from the incumbent operator. Similarly, in the mobile model,
there appears to have been little calibration with the networks of actual mobile operators.
Further, the equipment prices in the model are significantly lower than those Celcom pays.
There is also limited discussion in the PI Paper on reconciliation. For example, no mention
was made of annualisation rate. Finally, Celcom highlighted that the operating costs in
the fixed model are those of TM’s whereas it is more normal to base bottom-up model
operating costs on a range of sources.
Digi agrees with the MCMC that a bottom-up model does not take into account all the costs
necessarily incurred in providing services and submitted that it is important that common
costs are recovered with an acceptable margin in the provision of termination service.
Maxis is generally supportive of the LRIC+ methodology for mobile, fixed access and fixed
core. However, there are reservations on the implementation. Maxis expresses concern
about the asymmetry of information between the mobile model and the fixed model. For
the mobile model, the MCMC was in a position to contrast and compare at least three
datasets, while only TM’s input was used for the fixed model. There was also no evidence
of cost reconciliation, nor an audit of the input data provided by TM. Maxis expresses
concern that the results of calibration and reconciliation were not released on grounds of
confidentiality.
TM submitted that calibration and reconciliation can be difficult when multiple operators
offer the same services. Averaging the data across operators may not always result in
feasible outcomes. TM noted that the data traffic assumption used in the mobile model
does not reconcile with the operational performance described in the recent financial
report.
YTL agrees with the methodology adopted but noted that the factor of scale and capacity
of the new operator may not be the same as another operator. If standard calibration
factors are used, the small operator may not be able to recover some of the sunk and
common costs.
Discussion
The MCMC recognises the concerns of respondents about the levels of reconciliation which
have been possible between the mobile and fixed models. As the mobile sector is much
more competitive than the fixed then it has been possible to collect more comprehensive
Page 34
data from a range of mobile operators. However, additional data has been collected from
competing operators for key activities such as trenching and duct installation in order to
confirm that the costs put forward by TM are realistic. In many cases, TM’s costs were
shown to be lower than those for the competing operators.
The MCMC also confirms Celcom’s view that no efficiency adjustments in the usage of
access network equipment were performed in the model calculations. However, as
mentioned above, in the process of data collection, the MCMC verified whether the data
submitted by TM was reasonable and corresponded to values provided by TM to the MCMC
in other submissions unrelated to cost modelling. In several iterations, some of the input
values were adjusted in cooperation with TM so that the data could be accepted as being
reasonable to the satisfaction of the MCMC.
In response to Celcom’s concerns, there are limits to creating theoretical models. The
MCMC views that the creation of theoretical nets of cables (for example, concentric circles
or squares) with some theoretical density covering the area around local exchange
buildings would not deliver any realistic results.
In that regard, the only reasonable alternative would be to estimate cable routes using
GIS data and locations of local exchange buildings and subscribers. Nonetheless, the
MCMC is of the view that GIS modelling leads to a level of efficiency which cannot be
achieved in reality. GIS modelling would estimate optimal routes constructed under the
current situation. However, the fixed access network was constructed over a long period
of time. During this time, as the towns and villages developed, the locations of subscribers
also changed. It is not efficient to always dig new routes for cables in order to have the
shortest connection between local exchange buildings and the subscriber locations. In
reality, it is efficient to use the existing cable routes to the maximum extent possible and
attach new locations to the existing cable routes, even if they are not direct connections
and the lengths are not minimal. This is consistent with the scorched node approach,
while the GIS modelling actually resembles the scorched earth approach. In addition, it
is not always possible to construct the optimal route as found on the map due to digging
permits and obstacles in the path, for example.
The MCMC understands the general view that incumbent operators are generally
considered inefficient, and therefore, the idea of using real data from TM’s network is less
acceptable to the other operators. In balancing between the different perspectives, the
MCMC does not believe that even an operator in a monopolistic position without any
competitive pressure would use longer cable routes than necessary.
Based on the above reasons, the MCMC considers that using real data from the access
network of TM is a reasonable approach.
Page 35
Celcom highlighted that the operating costs in the fixed model are those of TM’s whereas
it is more normal to base bottom-up model operating costs on a range of sources. The
MCMC confirms that the operating costs used in the model were based on data from TM.
The MCMC considers this is the correct approach given the fact that the fixed model is a
model of TM’s fixed network. Unlike equipment prices, which are more or less comparable
between countries, operating costs are country-specific. Using international benchmarks
would not be appropriate in this case.
It has been necessary for the MCMC to respect the confidentiality of the information
received from individual network operators. In the case of the mobile model, these have
been anonymised as the model represents a notional operator with equipment prices that
reflect an efficient operator rather than an individual operator such as Celcom. In the case
of TM though, it has not been possible to share such data with the industry. The MCMC
notes the comments from TM about the data volumes modelled within the mobile networks
and addresses this issue further in Question 20 in the PI Report.
The MCMC notes the comments from YTL that the mobile model may not fully reflect the
costs of a smaller operator but also recognises that such an operator is likely to have much
lower levels of voice traffic and the use of the glide path will further help to mitigate the
impact.
Question 7:
Do you have any comments on the appropriateness of using glide paths and the method
by which the glide paths have been calculated?
Submissions received
Altel, Celcom, Digi and TM agree with the use of glide paths.
Celcom noted that the discussion on use of glide path is not very clear. While the MCMC
had stated that glide paths have been used in instance where the prices are sufficiently
different from regulated prices, it has used glide paths for fixed and mobile termination
but not for Interconnect Link Service, although the prices are about 40% higher. Celcom
also stated that the mobile termination rates are too low and the gradient of glide path
proposed by the MCMC is excessively steep. Such a steep gradient would result in serious
disruption in the market place and may result in the waterbed effect. Celcom believes
that a shallower gradient may be more appropriate.
Maxis firmly supports the minimum two years glide path approach used by the MCMC for
mobile termination. Since the mobile origination and termination prices in Malaysia are
Page 36
already among the lowest in the world, glide paths will soften the impact of sharp decrease
in termination revenue. Maxis also requested the MCMC to perform a comparison between
a linear interpolations method used by the MCMC with the geometric method used by some
regulators.
Maxis believes that glide paths for fixed origination and termination services are
inappropriate in light of near monopoly position held by TM in the fixed market and
considering that the fixed origination and termination rates in Malaysia are high among
the countries that they had benchmarked against Malaysia. Maxis also urged the MCMC
not to consider symmetric termination prices between fixed and mobile.
TM agreed with the MCMC on the use of glide paths as it will not cause significant disruption
to the finances of existing market players.
U Mobile highlighted that from the perspective of new operators, a glide path will mean
that they will continue to pay a higher price when costs have declined. This is exacerbated
by delays in the review process. U Mobile requested the MCMC to consider applying 2018
prices retrospectively.
One operator highlighted that glide paths should allow time for operators and customers
to adjust to new price levels and structures and should minimise disruptions. As such, the
MCMC was requested to consider a moderate reduction with a longer time horizon. They
believe a six-year time horizon would be appropriate. ACCC was cited as an example of a
regulator who had once considered a longer regulatory period to promote efficient use of,
and investment in, the infrastructure used.
Discussion
The MCMC notes the range of comments received to the proposal to use glide paths to
bring regulated prices into line with costs over time rather than by implementing them
immediately. In adopting glide paths for termination services, the MCMC seeks to move
towards the costs for each service. As noted by Maxis, the MCMC has no objective to bring
the rates for fixed and mobile termination in line. The MCMC is of the view that glide paths
are not necessary for other services.
The MCMC does not accept the argument that, because TM has a monopoly position, no
glide path should be applied to fixed services. Although the issue of avoiding disruption
to inward investment in the industry is more pressing in the case of sectors with greater
prospect of competitive entry, it is still appropriate to afford TM and its investors both
some protection from price shock and also, perhaps more importantly, to retain the
incentives to innovate provided by the glide path mechanism.
Page 37
The MCMC however recognises the issues as raised by U Mobile that throughout a glide
path period, the rates being paid are above the costs and are so providing a subsidy to all
terminating operators. This issue will be addressed under Question 26.
The MCMC takes note of the proposal made by an operator to use a six-year glide path.
In the three year’s glide path as proposed by the MCMC, the operators are already
recovering above cost, and as pointed out by U Mobile above, operators will continue to
pay higher price for termination. Therefore, the MCMC is of the view that a six-year time
horizon from 2018 is not appropriate.
Question 8:
Do you have any comments on the appropriateness of using the cost model results in
arbitrating disputes over access pricing?
Submissions received
Celcom pointed out a number of reasons why it may not be appropriate to use the results
of the fixed model in arbitration of disputes as follows:
(a) The stakeholders were not in a position to comment due to the use of dummy data;
(b) The absence of a manual explaining the approach adopted and assumptions used;
(c) The model appears to suffer a significant number of shortcomings which will tend
to bias the results upwards; and
(d) The charges produced by the model are insufficiently granular.
Celcom surmised that the fixed model used in its current state put TM’s customers and
competitors at an unfair disadvantage.
Celcom believes that there should be no role for the MCMC to arbitrate price disputes for
MVNO or other forms of mobile access since these are commercial services offered in a
competitive market. Intervention should be limited to markets where bottlenecks exist
and where operators have dominant position.
Digi is of the view that cost-based regulated prices for facilities and services can be a
useful quantitative source to be used as a reference point during arbitration proceedings.
However, the MCMC should consider the nature and context of dispute.
In principle, Maxis does not object to using the cost models to resolve disputes. However,
the service should be modelled accurately. Maxis has concerns about the manner in which
the fixed model has been developed and urged the MCMC to consider the nature of the
Page 38
dispute and where relevant, review and adopt the model to evaluate the impact on the
proposed regulated prices. Maxis is concerned that some fixed prices are higher than the
retail prices and market reality. As the basis of dispute is often motivated by Access
Seekers’ inability to compete at retail level, it is imperative that the MCMC closely monitor
the Access Providers’ retail prices to ensure that the retail offerings do not foreclose others
from competing effectively. Maxis proposed that the MCMC use the margin squeeze test
in conjunction with cost models.
PPIT does not agree that the models be used to arbitrate pricing disputes. The proposed
models are developed on an average basis and are unable to cater to individual
circumstances. They believe that the arbitration process should be examined on a case-
by-case basis and the resolution would depend on examination of individual factors that
drive operating and capital costs away from the average. PPIT pointed out that there are
many variations such as tower heights, geography, discounts, tenure, mark-ups and types
of service included that affect OPEX and CAPEX.
In addition, despite towers being in similar localities, the landlords may charge different
rents for the land sites. The soil conditions in Malaysia also vary significantly which affects
costs. PPIT pointed out that for USP projects, the MCMC had accepted the range of
RM745,545.45 to RM864,609.70 for constructing three-legged 76m light duty
telecommunications towers. However, in the PI Paper, the average cost of 75m towers is
only RM463,750.00.
PPIT also stated that the data gathering exercise appear to be selective as telcos were not
required to provide data for the model. The model also does not account for leasing with
additional sharing parties nor tenure of the lease. The MCMC also excluded some major
costs for telecommunications towers which brought the price down significantly.
TM’s main concern with the use of the cost model’s results in arbitrating disputes is the
use of marked down asset values that has led to an underestimate of TM’s costs. As time
elapses TM will undertake more capital investment, and therefore the underestimate
becomes even more significant and may lead to the use of inappropriate benchmarks or
pricing signals. If disputes arise after a considerable time from the time the models were
developed, it may be necessary to make amendments to the models as key parameters
or assumptions may have changed. TM does not favour the use of the cost model in the
existing form for dispute resolution. TM also stressed that pricing decision should not
solely rely on average costing model results but must include other dimensions such as
volume, distance, location etc.
U Mobile thinks it is reasonable for the MCMC to use the cost model results in arbitrating
disputes. However, depending on the circumstances, the MCMC may need to seek further
Page 39
information pertaining to the specifics of the disputes to ensure the appropriate cost is
determined.
webe is agreeable that the regulated prices should be the benchmark in arbitration of
disputes.
YTL pointed out that the models appear not to have included the costing for 4G services
using the 2.3GHz and 2.6GHz TDD LTE.
Discussion
The MCMC notes the concerns from respondents as to the potential limitations which the
outputs of the models may present especially with disputes which are specific to individual
locations and where it is not appropriate to use average costs or where projects may have
been the subject of USP funding. The MCMC however is of the view that the cost models
will continue to provide a useful input to such arbitration decisions where these are
required.
Although Celcom believes that there should be no role for the MCMC in the arbitration of
price disputes for MVNO or other forms of mobile access, it must be recognised that the
MCMC has an overall responsibility for the regulation of the telecommunications sector and
the imposition of good regulatory practice across the sector which includes the relationship
between MVNOs and MNOs. In addition, MVNO Access is included in the Access List and
is therefore subject to the oversight of the MCMC.
The MCMC accepts the comments by Maxis that a margin squeeze test could be used in
addition to the cost model results in arbitrating disputes over prices.
The MCMC agrees in principle with the point made by TM that if disputes arise after a
considerable period of time has elapsed from the time when the models were developed,
it may be necessary to make amendments to the models as key parameters or
assumptions may have changed.
Question 9:
Do you have any comments on the approach to setting prices for installation charges?
Submissions received
Altel believes that the proposed installation charges for Interconnect Link, Trunk and End-
to-End Transmission services are excessively high compared to the current prices. Since
Page 40
installation cost is not an immediate expense, it cannot be set to recover capital cost for
providing the service. In addition, since installation charges can vary depending on the
nature of work required, it should be determined by making reference to the actual cost
incurred.
Celcom in principle agrees with allowing efficiently incurred operating costs. However, the
way in which the cost has been estimated in the fixed model may result in over-recovery.
The installation cost has been estimated by multiplying the labour time involved by the
cost but since the model was populated with dummy data, Celcom was unable to verify
whether the time or cost is reasonable. TM has incentives to provide inflated figures and
it is unclear whether cross checks have been made. In addition, the hourly rates provided
by TM may include overheads, which would lead to double counting as mark-ups are
already included in the model.
Digi agrees with the MCMC’s approach to setting installation charges and matching it to
direct operational costs incurred. This approach provides transparency on recovery of full
costs of providing the service.
Maxis sought clarification on the direct operational costs efficiently incurred in putting the
service into operation that is being used by the MCMC in the Fixed Cost Model. To Maxis’s
understanding the operational cost for installation only includes the cost of man power to
install and turn on the service. Therefore, Maxis is surprised that the installation charges
are much higher than previous charges. For UniFi services, the package offered by TM
includes HyppTV everywhere, Installation and Activation, Wireless Modem and Cordless
Phone. It implies that TM does not impose installation charges for its UniFi Service.
TM agrees with the overall principle of determining installation charges based on direct
operational cost. In addition to the man hour cost, the direct operational cost should
account for other installation-related costs such as transport for deploying technicians.
webe is of the view that it is acceptable that the price of installation is regulated but it
should always be cost-oriented without inclusion of capital cost.
U Mobile sought clarification on how efficient direct operational cost is established for the
proposed installation cost in the PI Paper.
Discussion
The MCMC notes that there is general agreement from respondents that installation
charges should be charged separately based on the efficiently incurred non-capital costs
by the operator which is providing the service. The MCMC notes the concerns of Altel,
Page 41
Celcom and Maxis as to the level of the proposed installation charges. The costs were
calculated based on the number of working hours per installation and average hourly rate
of a technician performing the line installation. Both values were submitted by TM. Based
on the feedback from other operators, the MCMC has decided to revisit these inputs and
reduced the number of hours needed for the installation.
The MCMC notes Maxis’s comment about the TM UniFI service which is a retail service and
is therefore offered to customers as a package. As a wholesale service, the installation
and ongoing costs would need to be presented and recovered separately.
The MCMC notes the general agreement on the proposed methodologies to derive cost-
based prices: LRIC+ for the fixed and mobile networks, and a bottom-up model based on
current costs for Digital Terrestrial Multiplexing Broadcasting Service and Infrastructure
Sharing. There was also general agreement over the use of tilted annuity and broad
increment, scorched node, allocation of cost through routing factor, allocation of common
cost through EPMU and the inclusion of licence and spectrum fee in relevant cost models.
The MCMC also continues to view that a glide path over a 3-year period for the fixed and
mobile rates balances between the need to protect against price shock and to prevent
over-recovery of costs by operators. Finally, the MCMC also views that the cost models
will continue to be a useful input to arbitrate disputes, if required.
Page 42
4. Mobile Virtual Network Operators (MVNOs)
4.1. Overview
Section 8 of the PI Paper discussed the potential approaches to regulating MVNOs and
concluded that agreement through commercial negotiation was the MCMC’s preferred
approach.
Question 10:
Do you have any further comments on the proposed approach to regulating MVNO
Access?
Submissions received
Celcom, Digi, Maxis and webe are supportive of the MCMC’s position not to regulate prices
for MVNO Access. Celcom, Digi and Maxis pointed out that the mobile markets are
competitive and MVNOs have managed to establish a presence in the market.
Maxis is of the view that MVNO Access is already over regulated in Malaysia (via Access
List and MSA) as compared to many other countries in the world such as European
jurisdictions. In addition, there are also requirements for wholesale offers in the
forthcoming 700MHz spectrum tender. Maxis also carried out international benchmarking
and provided examples of the EU, Argentina, Brazil, Chile, China and Mexico where MVNO
access prices are unregulated.
Page 43
suitable remedies. Rather than setting prices for MVNO Access, it would be more important
for the MCMC to address the issue of domestic roaming. TM highlighted that many
countries have mandated domestic roaming.
webe believes that there is an overlapping meaning for domestic roaming and MVNO,
especially in their scenario where it relies 100% on a roaming partner for voice services.
While pricing for MVNO can be commercially negotiated, there must be a mechanism for
regulator to monitor and control margin squeeze.
The MCMC notes comments from respondents and the general agreement that, like in
most countries around the world, the MVNO market should be based on commercial
negotiations between mobile operators and their MVNOs. The MCMC agrees that the
principles of non-discrimination and transparency should apply within the MVNO market in
Malaysia and would expect all mobile operators to negotiate arrangements which follow
these guiding principles.
TM identified the need for regulation of national roaming. However, the MCMC believes
that such regulation could result in lower levels of investment and network roll-out at a
time when greater capacity is required in order to maintain QoS levels. National roaming
in order to reduce the number of so called “not spots” may be appropriate where there is
clear evidence that additional network roll-out is not possible, although with more
infrastructure sharing these locations should be limited.
The MCMC remains of the view that the MVNO arrangements should be made through
commercial negotiations. The MCMC would draw on information from the mobile cost
model in the event that the MCMC is required to intervene in disputes between the
operators.
Page 44
5. Weighted Average Cost of Capital (WACC)
5.1. Overview
Part C of the PI Paper presented the MCMC’s proposed approach to calculating WACCs for
the different services in the Access List. Section 9 included a discussion of the WACC
formula to be used and detailed the common parameters namely the Risk-Free Rate,
Equity Risk Premium (ERP) and Tax Rate including the proposed rates for each.
Section 10 presented the calculation of WACC values for the fixed network services,
including consideration of the impact of the HSBB funding programme, Section 11
presented the calculation of WACC values for the mobile network services, Section 12
presented the calculation of WACC values for infrastructure sharing services and Section
13 presented the calculation of WACC values for the DTT transmission services. The
proposed WACC values were then summarised in Sections 14 and 15.
Question 11:
Do you have any comments on the approach to calculating the appropriate levels of
WACC?
Submissions received
Celcom views that the proposed estimates of WACC did not consider the range of
estimation techniques which can be applied under the Capital Asset Pricing Model (CAPM)
approach. Based on the choice of parameters used, the MCMC relied on spot observations.
For example, beta estimates and gearing are based on current spot rates without due
consideration of how these estimates would change if a longer period of time was taken
into account. This is important as there could be short-term fluctuations in financial
indicators over the next 3 years.
Further, Celcom opines that the use of estimates provided by licensees were unjustified.
Firstly, no information was provided on how these parameters were derived or estimated.
Secondly, the fact that an estimate provided by an operator is similar to an average of
benchmarks is not proof that the estimate is reasonable. For example, the benchmarks
considered for the fixed asset beta vary significantly between different countries, hence,
benchmark countries must be considered carefully to ensure comparability. Thirdly, the
benchmarks considered by the MCMC are flawed, as they do not represent the type of
operator that is relevant for determining the WACC.
Page 45
Digi agrees with the proposal to use separate WACCs for fixed, mobile and DTT services.
edotco supports the computation and process to derive the WACC that is consistent with
a standard WACC calculation.
Maxis is principally in agreement with the MCMC on the approach used to calculate
separate WACCs for fixed network service, mobile network service, tower services and
DTT transmission services. It would have been ideal to have separate WACCs for the fixed
access and fixed core networks, but benchmarks and comparisons are rare, and the
incumbent itself is seen as a combined entity. Maxis is also agreeable to the WACC values
proposed for mobile network service, tower services and DTT transmission services, but
not for the fixed network services.
PPIT agrees with the proposed approach in adopting the CAPM, which is widely accepted.
Further, PPIT also agrees that fixed, telcos, tower companies and DTT transmission
operators have different risk profiles and require distinct WACC estimates.
TM agrees with the use of CAPM to estimate the WACC, and that the WACC is expected to
vary across different types of investment projects. TM also agrees that typically fibre
investment involves a higher risk profile, and highlights that in the minority of cases where
regulation is applied in Europe, a risk premium is recommended. Nevertheless, for
pragmatic reasons, TM agrees that separate WACCs should not be applied for separate
lines of business. However, the MCMC should allow for the significant risk involved in
HSBB investment in its WACC estimation. Finally, TM agrees that the estimated WACC
should not be based solely on one operator’s financial circumstances but reflect
appropriate local benchmark data with consideration of relevant overseas comparators.
Discussion
The MCMC notes that, with the exception of Celcom, there is a general consensus amongst
the respondents to both the approach and the overall values of WACC proposed for mobile
networks, infrastructure sharing and DTT services which reflect the different levels of risk
experienced within the individual markets. Celcom comments that spot rates have been
used which do not take account of fluctuations which may occur in the financial markets
in the coming years. As with all forecasting, it is impossible to be fully accurate and so
the MCMC does not think it is appropriate for it to attempt to second guess such changes.
In setting the levels of WACC for the Malaysian market, the MCMC feels that it is
appropriate to take account of the rates of local operators which reflect their individual
financial situations and which are publicly declared to financial markets. The MCMC does
Page 46
not believe that it is appropriate to use rates that are entirely based on conditions in other
markets. Whilst recognising that the average of benchmarks does not prove that rates
are correct such a comparison ensures that rates are not out of line with broad
expectations. A similar point may be made with respect to the contributions of industry
participants, whose views the MCMC seeks to take into account. Selecting appropriate
sources for benchmarking rarely results in unanimity but the operators selected have only
been used for comparison purposes rather than for the absolute setting of WACC rates.
Maxis and TM acknowledge that, whereas it could be desirable to set different rates of
WACC for fixed access and core networks as the deployment of access network
infrastructure for HSBB services potentially has a different risk profile. However, they
recognise that TM operates as an integrated telco and as such the separation of such
activities through different levels of WACC would be difficult at this point in time. Should
functional or structural separation of TM occur in the future then such an approach may
be more appropriate, though the difficulty in identifying suitable comparators is likely to
remain.
Question 12:
Do you have any comments on the proposed common parameters to be included in the
WACC calculations?
Submissions received
Digi, edotco, Maxis and PPIT support the common parameters used, while Celcom submits
that both the risk-free rate and the ERP are overstated. TM comments that the common
parameters should be based on local data. YTL believes that a single ERP should not be
used.
On the risk-free rate, Celcom opines that there was no consideration of whether the
current rate of the government bond yield is representative of the risk-free rate on a
forward-looking basis. The risk-free rate is often based on an average measured over a
longer period of time to avoid using current measurements which can be affected by
transitory effects that are not representative over the period of time of the WACC estimate.
In fact, it is also not standard regulatory practice to use a 10-year government bond as a
basis for setting the risk-free rate.
In that regard, Celcom views that short to medium term historic evidence should be used
as a basis for setting the risk-free rate, for example, based on an average of the 1-year
and 2-year historic averages. Based on the 10-year government bond yield, the averages
over the last 1, 2 and 4-years are broadly similar to the proposed MCMC rate, but would
Page 47
be slightly lower than 4% if averages of 2 or 4 years of historic data are used instead.
However, if a shorter 5-year government bond, which is considered in other jurisdictions,
is taken into account, then there will be even lower average yields for 1, 2 and 4-years.
Given the time value of money, the choice of maturity is an important consideration. The
maturity should reflect the remaining useful life of assets and the length of the regulatory
period. On the one hand, it should reflect the duration of the charge control. On the other
hand, as a result of investments that may be made by operators and where the economic
lifetimes are in excess of the charge control period, then a longer-term maturity may be
appropriate. As such, a mixed approach of using long 10-year maturity and short 5-year
maturity bonds have been used by Ofcom in the UK since 2005 until it has been suspended
due to the impact of quantitative easing in the UK and Europe on the yields of short-term
maturity bonds. As such, Celcom proposes the mix of yields of 5-year and 10-year
maturity bonds for the purpose of estimating the WACC.
On the ERP, Celcom has identified two flaws in the MCMC’s estimation: the use of the ERPs
submitted by the operators and on the use of the Professor Damodaran’s estimate.
It was unclear to Celcom on why the MCMC has removed the lowest estimate in the range
of 2.89% to 10.94% provided by the operators, rather than adjusting the rate to be based
on the long-term average. Secondly, the extent that the actual rate would be affected by
the inclusion of this lower estimate is unclear, as the entire range of ERPs proposed by the
operators was not specified.
Further, Celcom submits that the MCMC’s consideration of Professor Damodaran’s ERP
estimate effectively double counts the country risk. When the MCMC considers a Malaysian
government bond yield as the basis for the risk-free rate, the risk of investing in Malaysia,
over and above a “pure” risk-free investment, has been taken into account. Based on its
review of Professor Damodaran’s ERP, Celcom submits that his calculation has already
added the country risk premium. Hence, applying Professor Damodaran’s ERP would be
correct if the MCMC’s WACC calculation was based on a global risk-free rate, such as the
US 10-year government bond yield.
As the MCMC has not provided the details on how the average ERP submitted by the
operators was determined, Celcom proposes that the ERP as determined by Professor
Damodaran, without the addition of a specific country risk premium, should be used as
the basis for determining the cost of equity. The latest published rate is 5.13%. The rate
is consistent with the range of ERPs used by regulators in other jurisdictions, whilst the
MCMC proposed ERP falls outside the range.
Page 48
TM submits that the risk-free rate, tax rate and ERP should be set based on local Malaysian
data, and TM notes that reliance has been made primarily on Malaysian data. In addition,
for the LRIC model which is forward-looking, the risk-free rate should be set based on the
local bond market from the time period closest to the date of the Determination.
YTL submits that the use of a single ERP masks the differences in the level of returns
expected of service providers. Further, YTL believes that the WACC for Mobile Network
Services should be higher for new operators due to the risks and state of competition in
the market.
Discussion
The MCMC notes that there is agreement amongst the majority of the respondents to the
proposed common parameters within the WACC calculations. Further, it is also noted that
TM recognised that the rates were drawn from the Malaysian market.
In response to the comments raised by Celcom on the risk-free rate, the MCMC does not
agree that the use of a 10-year rate is unusual practice and it was chosen as being a
suitable compromise between the respective time horizons of the charge period and the
assets under investment by operators. Equally, the MCMC is not convinced that
consideration of the various averages put forward by Celcom leads to any clear justification
for revising the estimate.
Celcom also raises an issue with the exclusion of outlier estimates of the ERP. In point of
fact, both the highest and lowest estimates (2.89% and 10.94%) were excluded in
determining a consensus value. Removing these two values reduces the average slightly,
from 6.15% to 5.89%. The MCMC notes Celcom’s point about the inclusion of an element
of country risk in the value quoted from Professor Damodaran and accepts that the value
of 5.13% is correct.
The MCMC notes that YTL is concerned about a single rate for all mobile operators
irrespective of size. The mobile model developed by the MCMC is intended to identify the
costs of a reasonably efficient notional operator and is intended to assist with setting only
the voice origination and termination services rather than the whole spectrum of retail
services as offered by mobile operators.
Page 49
Question 13:
Do you have any comments on the WACC values proposed for the Fixed Network
Services?
Submissions received
APCC supports the use of the WACC as a method of factoring the cost of raising capital
into the equation when calculating the regulated prices of facilities and services. The use
of WACC has been widely accepted by many national authorities globally.
APCC submits that the appropriate WACC value to be applied should be 8.08%, which is
the method of calculation which considers the Government funding as a pure subsidy with
zero return. APCC supports this with its understanding that there is no specified rate of
return in TM’s PPP, after the first few years. The WACC value of 8.08% is also consistent
with the WACC values applied by national regulators in Italy, the UK, Spain, Germany and
France. If the WACC of 8.08% is not accepted, APCC proposes that a just and equitable
route is to require the disclosure of the rate of return and to have the rate of return
factored into the WACC calculation for fixed services.
Celcom submits that the gearing and equity beta assumptions proposed for the Fixed
Network Services are inappropriate.
Celcom notes that the gearing is based on TM’s submission, though the MCMC did not
justify why the rate can be considered the rate of an efficient fixed network operator. In
reviewing the data on TM’s gearing from Bloomberg, Celcom observes that over time from
2014 to 2017, there is a consistent upward trend. This suggests that in setting the WACC
over the regulatory period, a higher gearing can be justified. This is also consistent with
international benchmarks.
On the international benchmarks selected by the MCMC, Celcom highlights that Spark and
Singtel are not appropriate comparators for fixed telecommunication operations in
Malaysia. They are not the owners of the underlying network infrastructure as the assets
and operations have been transferred to separate infrastructure companies, and are
therefore, primarily retail operations. Both Spark’s and Singtel’s low levels of debt are
likely to be the result of these companies owning fewer assets against which debt can be
secured. The remaining companies from the MCMC’s benchmarks suggest a gearing of
approximately 30%. Further, based on Celcom’s review of international regulators’
decisions on WACC, the gearings range between 30% to 40%. Hence, Celcom proposes
that a gearing of 30% to 35% is more reasonable as a basis for setting the WACC of Fixed
Network Services in Malaysia.
Page 50
On the fixed network beta, Celcom comments that the equity beta submitted by TM was
used based on a comparison against a number of comparator companies. However,
Celcom views that comparison against an average is not an appropriate way to justify TM’s
submission.
Celcom also reiterates that the benchmarks used are not suitable due to the nature of
their businesses. As explained above, Spark and Singtel are different from TM; they do
not own infrastructure and focus on retail operations. Most of the operators in the
benchmark also operate mobile business, such as PLDT and Telstra, which makes them
unsuitable for a fixed-only network operator WACC.
Based on Celcom’s review of TM’s current and historic average betas provided by
Bloomberg, it notes that they are significantly different from the beta submitted by TM.
For the current, 1-year, 2-year and 4-year averages, they range between 0.61 and 0.64,
with the current on the lower end of the range. This is based on estimating the equity
beta against the main Malaysian stock market index. Another method used by regulators
in open economies, like Malaysia, for estimating equity beta is to measure the volatility of
the stock against a global index, such as the MSCI World Index. Based on this, the equity
beta derived for TM is significantly lower, ranging from 0.11 to 0.05.
Given the lack of explanation in the PI Paper, Celcom was unable to establish the marked
difference between its own finding and the MCMC’s proposed rate for asset beta. As such,
Celcom also reviewed the asset beta of regulatory decisions in other jurisdictions. It notes
that the average of those asset betas is similar to TM’s asset beta (derived from its current
equity beta measured against the local Malaysian stock market index). At the same time,
the average is very different from the MCMC’s proposed rate and is higher than Celcom’s
estimate of TM’s equity beta against the MSCI World Index.
In view of the differences between Celcom’s findings and TM’s submission, the lack of
information about TM’s own estimate and the unsuitable set of benchmarks used, Celcom
urges the MCMC to revise its estimate of the beta and to bring it in line with the actual
beta of TM.
Based on its submissions on risk-free rate, ERP, gearing and equity beta above, Celcom
calculated two fixed WACC figures and proposes an average WACC of 7.42% as a better
reflection of the actual financial data of TM, international regulatory precedent and best
practice in terms of deriving WACC.
On the impact of HSBB funding, Celcom notes that by taking into account the value of the
assets acquired through the funding without specifying the rate of return actually payable
to the funding agency, the MCMC is allowing TM to obtain super-normal profits to the
Page 51
disadvantage of consumers and Access Seekers. In that regard, rather than to account
for the funding through the WACC, Celcom proposes that the MCMC should deduct the
amount of funding from the asset values taken into account when estimating the cost of
services. Celcom disagrees that it would be difficult to identify the assets that benefited
from the funding, and the payments to the funding agency should be considered as
expenses incurred. However, if the MCMC maintains that the funding should not be
deducted from the corresponding asset values, then the WACC should be adjusted in such
a way that only the risk-free rate (which is the rate at which the Government is able to
finance the funding) is considered for the Government funded portion of TM’s assets.
Digi submits that the WACC proposed for the Fixed Network Services is overestimated,
and views that a WACC of 8.08% which is used for sensitivity analysis is a closer reflection
of WACC for a notional fixed operator. Furthermore, Digi opines that networks that have
received direct or indirect Government funding or are backed by local or state government
bodies will have lower levels of risk, and Digi notes that this is not reflected in the WACC
values.
On gearing, Digi notes that it is based on TM’s submission, which Digi views is inclusive of
the Government’s investment in HSBB. Therefore, it is likely understating the efficient
long-run gearing level in TM’s capital structure. Hence, Digi recommends to use
benchmarks, such as from S&P Global Market Intelligence that cites TM’s actual gearing
to be in excess of 51%.
Finally, Digi notes TM’s WACC that is declared in its quarterly reports, is 6.94% as of 30
June 2017, whilst the current pre-tax WACC of TM based on Bloomberg is at 6.4%, both
of which are lower than the proposed WACC.
Maxis proposes an amendment to the estimation of the asset beta by excluding PLDT and
Singtel from the list of benchmarks, thereby reducing the average levered beta to 0.65.
With that adjustment, the cost of equity is amended to 8.95%.
In addition, Maxis proposes to exclude the amount funded by the Government from TM’s
asset base in order to ensure an economically efficient outcome. This is not achieved if
only a small discount of 5% is applied to the implied cost of capital as proposed by the
MCMC. The inaccurate treatment of the amount funded by the Government will lead to
over-recovery of costs by TM at the expense of the Access Seekers.
With regard to the WACC for the Fixed Network Services, Maxis proposes to use TM’s last
reported pre-tax WACC value of 7.2%. As almost all of the data input, cost input,
assumptions and network dimensioning are based on TM’s network, the WACC should also
Page 52
be based on TM’s WACC. This value is in line with fixed services benchmarks, such as
IDA’s 7% for Netlink Trust Singapore and Australia’s NBN Co at 7.2%.
TM submits that the asset beta is a key parameter of the WACC, and the best practice is
to rely on the median value of the betas of relevant comparator companies. The use of a
large sample of firms will reduce statistical error compared to the use of a single beta of
one firm. Median values are typically preferred to averages which may be skewed by
outlier values. Hence, TM recommends that the MCMC use the median value of the
sample, which is 0.905, which is more consistent with the notional efficient operator
concept. This will revise the WACC for Fixed Network Services to 11.93%.
TM views that the discount of 5% to the implied capital costs leads to double counting of
the impact of the Government’s contribution to HSBB. The asset beta in the WACC is a
measure of systematic risk. In addition, TM highlights that in New Zealand, the regulator
has considered that Government contribution is irrelevant to the pricing of wholesale
services. TM also notes that in New Zealand there is no revenue sharing arrangement.
On the other hand, TM highlights that the European Commission endorses the addition of
risk premium to the WACC ranging from 2.5% to 4.8% as it considers that the risk involved
in such ventures are not reflected in the asset beta. The median of the risk premium is
2.85% and should be considered as a benchmark by the MCMC.
U Mobile submits that the WACC of 8.08% should be used for Fixed Network Services. As
neither TM nor MCMC has been able to ascertain the level of returns for the Government
under the PPP or isolate the services, it raises doubt as to whether there is a return.
Without evidence to indicate that the Government’s contribution is based on a commercial
basis, there can be an assumption that the RM12.2 billion is a subsidy given to TM with
zero return. U Mobile concurs that TM would not have accepted the contribution and the
Government would not have provided the funds if they could have been reasonably
secured via commercial arrangements from the capital market.
Discussion
The MCMC notes the various comments received from respondents to the proposed level
of WACC for Fixed Network Services which, with the exception of TM, indicate an over
statement of the WACC figure. Furthermore, the figures published by TM to the market
also appear to be lower than the initial calculated value. Nevertheless, the MCMC is not
persuaded that relying on TM’s own figures would lead to a better estimate of the efficient
WACC.
On the specific points raised, the MCMC accepts the argument that benchmark
comparators need to be selected and interpreted with care. However, it is not always
Page 53
possible to find exact replicas of the regulated firm. In this case, one issue is that TM is
unusual amongst fixed network operators in not having ownership and control of a fully
mobile arm. It is impractical, therefore to exclude such companies from the sample and
to do so would simply have the effect of cutting off this source of information. On the
other hand, the MCMC is persuaded that operators who have separated from their retail
arm, particularly Singtel and Spark, may unduly bias the sample.
TM also make the point that the median may be a preferable measure to the arithmetic
mean, because of the latter’s potential for being affected by outliers. An alternative would
be to exclude obvious outliers.
Excluding the two benchmark operators mentioned above results in arithmetic mean of
0.77 for the beta and 30% for the gearing and median of 0.73 and 30% for beta and
gearing respectively.
As a result of the above considerations, it is the view of the MCMC that the following
parameters will be used in the final calculation of the WACC for the fixed network services:
Parameter Value
Risk-free rate 4%
Equity Risk Premium 5.13%
Tax rate 24%
Gearing 30%
Beta 0.77
As a result of these revisions to the parameters following inputs from the various
respondents, the WACC for the Fixed Network Services will be 8.77% before the
Government of Malaysia’s investment in HSBB is taken into account. However, taking the
cost of capital for the Government’s investment at the risk-free rate of 4%, this reduces
to 8.27% as a weighted average.
Page 54
Question 14:
Do you have any comments on the WACC values proposed for the Mobile Network
Services?
Submissions received
Celcom is unable to comment on the individual parameters of the WACC values as the
rationale behind the choice of the parameters was not made clear. Nevertheless, Celcom
proposes amendments to the debt premium and gearing.
In relation to the debt premium, Celcom notes that the lower end of the range, i.e. 0.23%
is inconceivably low for a mobile operator, as compared to the debt premium for the fixed
network, estimates from Europe and those derived by Professor Damodaran. Therefore,
Celcom proposes that the lower end of the range for estimating the debt premium for the
mobile network WACC should be based on the fixed network debt premium while the upper
end of the range should be maintained at the level proposed by the MCMC.
Celcom views that the gearing calculation used by the MCMC to determine the WACC for
mobile services is flawed. Rather than use the gearing directly (financial debt/equity ratio)
when calculating the WACC, the ratios of equity and debt as a share of the total value of
the company must be used.
Hence, based on Celcom’s submissions on risk-free rate, ERP, debt risk premium and
gearing, Celcom provided two estimates of WACC for the Mobile Network Services. Taking
an average of those two estimates, Celcom proposes a WACC of 9.97% for Mobile Network
Services.
Digi and Maxis agree with the proposed WACC for Mobile Network Services.
TM submits that the MCMC’s results where the WACC for Mobile Network Services is higher
than for Fixed Network Services is inconsistent with current best practice that recognises
that a mobile business is not inherently riskier than a traditional fixed business. The recent
empirical study in 2017 on behalf of Ofcom found no evidence of differences in systematic
risk between the two types of business and that the asset beta associated with mobile
operations is likely to be similar to fixed operations. TM further explains that historically,
WACCs for mobile services were higher than those of fixed services as, at that time, mobile
businesses were relatively new compared to fixed services. Currently, mass market mobile
revenue streams are more certain than those of fixed services threatened by OTT services.
Hence, there is a trend in Europe of equal WACCs for fixed and mobile or where the WACCs
for mobile are lower than those of fixed.
Page 55
Based on TM’s examination of the EBITDA margins of the three largest Malaysian mobile
operators and regional comparators, two of the Malaysian mobile operators are amongst
the four most profitable in the regional sample. This result is consistent with a lower risk
profile and implies that the asset beta for the WACC for Mobile Services has been
overstated.
TM further comments that the median beta from the MCMC’s regional comparison for
mobile services of 0.74 should be used, consistent with best practice. This will revise the
WACC for Mobile Network Services to 11.41%. The difference between TM’s proposed
WACC for Fixed Network Services and Mobile Network Services is -0.52% which is
consistent with current regulatory and benchmark trends.
YTL submits that the use of a single ERP masks the differences in the level of returns
expected of service providers. Further, for new operators such as YTL that have not
generated net profits due to initial start-up costs will have a higher beta causing its WACC
to be higher than that of the industry. The use of an average industry WACC will lead to
under-recovery, and proposes that the WACC for Mobile Network Services should be higher
for new operators.
Discussion
The MCMC notes the various comments received from respondents to the proposed level
of WACC for Mobile Network Services and notes that Digi and Maxis support the proposed
figure whereas YTL wishes to see a high figure for new entrant operators such as itself.
Celcom has raised questions over the calculation of the gearing ratios and the debt
premium. In relation to the gearing ratio, the MCMC would concur that a market-based
calculation is preferable and this is what has been used in the calculation. In relation to
the debt premium, the MCMC is unsure why Celcom would assert that some of the figures
put forward are “inconceivably low”, given that they reflect the experience of actual
operators in Malaysia.
The MCMC notes TM’s observation that there has been a trend towards convergence in
fixed and mobile WACCs in Europe. However, the MCMC does not accept that this is, in
itself, an argument for changing any of the parameters. The approach the MCMC has
followed, in common with many other regulators, is to calculate the WACC on the basis of
its constituent parameters, rather than from an overall observation of trends elsewhere.
The MCMC notes the point made by YTL that small and new entrant operators are likely to
face higher capital costs than the larger operators, but is not persuaded that the value for
Page 56
the notional operator should be set accordingly, as this would be likely to lead to over-
recovery of costs by operators with market power, to the detriment of consumers.
After careful consideration of the comments from the respondents and taking into account
the change to the ERP, the WACC for the Mobile Network Services will be 10.00%.
Question 15:
Do you have any comments on the WACC values proposed for the Towers Services?
Submissions received
Celcom finds it difficult to comment on infrastructure and DTT WACC estimates due to a
lack of detail of how the estimates submitted have been derived. However, Celcom notes
a few inconsistencies as follows.
Celcom notes that a wide range of gearing between 12.9% to 65.6% has been provided
by the operators to the MCMC, and the clear outliers have not been removed. On the cost
of debt, Celcom highlights that the upper range, i.e. 2.5% is rather high and may be the
result of suboptimal financial structure. Hence, Celcom proposes that this value should
not be considered in setting a WACC to determine efficient access prices.
Further, Celcom views that the operations of tower companies are similar to that of
infrastructure companies and their WACC should therefore be closer to the WACC of a
fixed infrastructure company than an integrated fixed (retail and infrastructure) or a
mobile business. For example, the WACC of Openreach is set 15% lower than the rest of
BT’s business WACC.
Apart from that, Celcom comments that its submission on risk-free rate and ERP are
equally valid in deriving the WACC for Towers Services. Based on corrections of these
common parameters, Celcom proposes that the maximum WACC for Tower Services is
10.04%.
Digi views that the WACC for Tower Services has increased from the initial 10% to 12.11%.
Digi submits that due to tower companies having direct solid support from state
government bodies and lack of competition, the risk for them is mitigated. As a reference,
based on Bloomberg, the current pre-tax WACC of a public listed tower company is 6.9%,
which is much lower than the proposed WACC value. Further, Digi also notes a high debt
premium for tower companies, and expresses its concern that the sample used does not
reflect the presence of state-backing which mitigates the risk for tower companies.
Page 57
edotco views that the WACC for Tower Services should be higher, and should be materially
higher than that of the mobile operators, to reflect the higher risk of tower companies in
the market. In its view, tower companies have a weaker commercial bargaining position,
and to enable a fair return on investment, the WACC should be higher.
Maxis is agreeable to the proposed WACC value for Tower Services as appropriate for the
efficient notional tower operator in Malaysia.
PPIT and Sacofa view that it is more appropriate to select benchmark companies from
emerging markets as opposed to American and European tower companies. However,
they note that the proposed leveraged beta used is coincidentally consistent with their
suggestion. Further, PPIT and Sacofa are agreeable to the point value for WACC for Tower
Services as it falls within their proposed range. Finally, PPIT and Sacofa suggest that there
might be an error in the WACC computation and request the MCMC to review the WACC
for Tower Services.
TM highlights that the comparator data for the beta and gearing of tower companies is
extremely limited and might have limited meaning for the local environment. TM suggests
that either a larger set of benchmark data be used or alternatively, to place sole reliance
on local information for the estimation of the WACC for Tower Services.
U Mobile submits that the gearing of some of the operators are as high as 66%. In such
instances, inefficient tower operators could set unreasonably high prices and over-recover
costs to the detriment of Access Seekers.
Discussion
The MCMC notes the various comments from the respondents to the proposed level of
WACC for Tower Services.
In particular, concerns were raised about the selection of comparators and the treatment
of outliers. Although comparator companies were quoted for reference, the MCMC does
not feel that there was undue reliance on this source.
Although outliers were excluded in the calculation of the debt premium, the raw average
was mistakenly used for the gearing. When the two extreme values of zero and 65.62%
are removed, the average of the remaining figures is 23.49%.
After careful consideration of the comments from the respondents, with the changes to
the ERP and the gearing at 23.49%, the WACC for the Tower Services will be 10.08%.
Page 58
Question 16:
Do you have any comments on the WACC values proposed for the DTT Transmission
Services?
Submissions received
Celcom finds it difficult to comment on infrastructure and DTT WACC estimates due to a
lack of detail of how the estimates submitted have been derived.
Celcom nevertheless suggests that its submission on risk-free rate and ERP are equally
valid in deriving the WACC for DTT Transmission Services. Based on corrections of these
common parameters, Celcom proposes that the maximum WACC for DTT Transmission
Services is at 9.60%.
TM makes a general comment that the MCMC has partly relied on a 10-year old study for
benchmark information for parameters of the WACC for DTT Transmission Services. It is
highly unlikely that the data would be relevant and expect that the proposed WACC value
will have a very high margin of error.
Discussion
After careful consideration of the comments from the respondents and taking into account
the change to the ERP, the WACC for the DTT Transmission Services will be 9.51%.
Question 17:
Submissions received
Celcom refers to its responses in the earlier questions. In addition, Celcom notes that the
impact of applying the Monte Carlo simulation on the proposed WACC estimates is unclear.
The MCMC stated that its point estimates are the preferred values for setting the WACC
however, for the fixed services, a lower end of the Monte Carlo simulation range was
considered for sensitivity analysis. In that regard, Celcom sought clarification as to the
circumstances where the MCMC would choose to deviate from the point estimates.
Digi agrees with the approach to use selected point values. However, the WACC for Fixed
Services and Tower Services should be reviewed.
Page 59
edotco views that the WACC for Tower Services should be materially higher than that of
the Mobile Network Services. In addition, given the growth of mobile operators as against
fixed operators, edotco expects that the WACC for Mobile Network Services and Fixed
Network Services to be similar.
Maxis is largely agreeable to the proposed WACC values for mobile network services, tower
services and DTT transmission services. However, Maxis proposes to use a lower WACC
for fixed services, i.e. based on TM’s last reported pre-tax WACC value of 7.2% (derived
from Bloomberg). This is also in line with benchmarks such as IDA’s 7% for Netlink Trust
Singapore and Australia’s NBN Co at 7.2%.
TM does not agree with the use of the WACC value of 8.08% for sensitivity testing for
fixed services. This value was calculated based on the assumption that the Government
contribution to HSBB is a pure subsidy with zero return. As this is a revenue sharing
arrangement, the ‘pure subsidy’ scenario is inappropriate.
U Mobile is agreeable to the proposed WACCs, apart for the WACC for Fixed Network
Services, which should be lower.
Discussion
The MCMC notes the comments made about WACC ranges. The purpose of quoting them
was to show the effects of the individual parameter ranges under discussion of the overall
values of the WACCs. Whilst this does not provide statistical confidence bounds, it does
provide useful upper and lower limits within which the point estimates may be considered.
Page 60
6. Fixed Services
The MCMC developed a single fixed network cost model using the LRIC+ methodology to
assess the costs of providing the fixed services in the Access List. The model was based
on a network operator with similar scope and scale to that of TM. A form of the cost model
with all commercially confidential data removed was made available on request to
interested licensees during the Public Inquiry period.
Part D of the PI Paper dealt with the fixed services. Section 16 provided a list of the
services from the Access List which were costed in the model. It sets out the steps which
the modelling process followed dealing with service demand and traffic, network
dimensioning, network costing, service costing and model reconciliation. A summary of
how the initial comments from operators were taken into account following the initial
model viewing was also presented. Section 17 presented the MCMC’s proposed prices for
the fixed services.
Question 18:
Do you have any comments on the approach adopted for the fixed model?
Methodology
Submissions received
APCC supports the adoption of a bottom-up LRIC+ methodology for fixed services on the
condition that the costs of the network are adjusted by the value of the fully-depreciated
assets. This is in line with the trend in a number of countries in the EU.
Celcom does not consider that there are sufficient reasons in the PI Paper for the MCMC
to disregard the ‘first-best’ approach, which is the step-by-step approach. Firstly, the
difficulty in allocating assets to networks or services can be addressed on the basis of
usage or other causal factors. Secondly, detailed historical cost data should be readily
available. Thirdly, the cost of high data requirements on TM cannot be compared with the
higher costs borne by Access Seekers and end users as a result of implementing a sub-
optimal approach. Celcom also comments that regulators in other jurisdictions have
followed such an approach, and Celcom does not agree that the reasons given by the
MCMC are legitimate barriers to implementing the ‘first-best’ approach.
Page 61
Celcom finds that the MCMC’s justification in the PI Paper to change to a bottom-up LRIC
model based on the EC Recommendation as unconvincing. The recommendation refers to
NGA investment whereas TM’s network is largely copper-based, and to apply bottom-up
costing principles to copper investments that are heavily depreciated is different from
applying them to NGA investment. In addition, the recommendation refers to a
hypothetically efficient operator, however, Celcom notes that TM’s data is used almost
without reservation. Finally, Celcom underscores that when regulation moves from a top-
down or modified top-down approach to bottom-up approach, it is important that this does
not result in windfall gains to the incumbent through over-recovery of costs.
Celcom’s concern, as expressed in the PI Paper, is not that the bottom-up LRIC model
leads to over-recovery of investment cost. In fact, Celcom’s concern is that a network in
which its assets are more than halfway through their asset lifetime (referring to TM’s
access network), a move from top-down approach to bottom-up approach will allow
operators to over-recover their costs. This is the case even where fully-depreciated assets
are excluded, as recognised by Ofcom in its 2005 examination of BT’s copper local loop.
Likewise, Celcom asserts that its concern is not with tilted annuities methodology itself,
but with the combination of using accounting methodologies in the early years of an asset’s
lifetime with tilted annuities in later years, for assets which are rising over time. In such
circumstances, it is important to make an adjustment to eliminate any windfall gains that
this combination method grants to the incumbent. Celcom also submits that there is little
merit in showing the sensitivity using simple annuities as this is inappropriate for asset
prices that change over time.
Celcom notes that dummy data has been used instead of real data, for confidential
reasons. In response, Celcom reiterates that there are numerous publicly available
bottom-up models published with real data. Most of the data used in the fixed model such
as length of the access network, routing factors and times taken to install equipment, is
not confidential; and confidential data such as equipment costs can be disguised.
On a related point, the MCMC’s suggestion for operators to use their own data to populate
the fixed model is unrealistic. Firstly, it is a major exercise for an alternative operator to
estimate the length of TM’s trenches and cables particularly for the access network.
Secondly, whilst Celcom may be in a position to estimate data such as trench and
equipment costs, equipment costs vary significantly between countries and between
operators in the same country. Thirdly, crucial data such as the levels of calls and
broadband is unknown. Fourthly, populating the model with dummy data does not allow
third parties to verify that calculations within the model have been made correctly.
Page 62
Maxis agrees with the approach to base the fixed network model in Malaysia on TM’s
network. However, Maxis expresses concern with the lack of data in the fixed cost model
that inhibits proper assessment to be made. In that regard, Maxis suggests that the
incumbent’s costs be benchmarked against international operators’ data. Further, Maxis
submits that the problem of commercially sensitive and confidential data has been
addressed in models of other countries, such as the UK, France, Spain, Sweden, Denmark,
the Netherlands, Cyprus and Norway, through the use of disguised, rather than dummy
data. This is critical for the fixed cost model as the incumbent will have a significant and
non-replicable advantage in reviewing the model compared to other stakeholders. For
other operators, with only dummy data, it is very difficult to verify model calculations and
the reasonableness of forecast assumptions which increases the risk of model errors.
Maxis supports the use of LRIC+ costing methodology for fixed services, however, it
expresses its concerns on the outputs of the model. For example, the fixed network
termination prices are among the highest in its benchmark of countries, transmission-
related prices are much higher than current market prices and HSBB Network Services
prices are higher than the retail prices offered by TM to its end users. This seems to
indicate that input costs, network dimensioning and modelling assumptions require further
review by the MCMC.
Maxis firmly supports the approach to not include stranded assets in the fixed assets and
to model a fully NGN fixed network without any legacy assets. It reiterates the approach
of the ACCC to remove stranded assets when the NBN migrated to new fibre-based
services. In Malaysia, Maxis submits that TM will no longer offer ADSL services for
premises that have subscribed to HSBB, and suggests that a similar treatment as Australia
be followed for the copper assets. Maxis further supports the reuse of 100% of civil
infrastructure in the model, however, it suggests that the appropriate asset life be 35
years, which is the midpoint of BEREC’s Regulatory Accounting Review’s benchmark of 30
– 40 years. ACCC also uses the depreciation lifespan of 35 years.
TIME agrees to the use of the bottom-up model, however, disagrees to the use of LRIC+.
It views that LRIC+ heavily discounts on the investments made in the earlier years on
fibre and civil works. Further, the cost of fiberisation will not reduce at the same rate as
equipment and may increase due to permits and labour costs. For the fixed network, TIME
recommends the use of a FAC model for fibre investment.
TM, in its submission to Question 4, highlighted that the MCMC’s approach in writing down
many network and non-network assets departs from LRIC. In addition, this is done by
arbitrarily selecting assets and writing down as much as 90% in order to match 2016
estimated depreciation with actual depreciation. TM submits that there is no rationale for
markdowns of selected assets, particularly for the local fibre rings by 90%. These are the
Page 63
fibre rings that connect outdoor MSAN cabinets to exchanges and should be treated as E/S
fibre rings.
Discussion
The MCMC agrees with the APCC’s view that bottom-up LRIC+ is the correct methodology
for fixed services under the condition that the costs of the network are adjusted by the
value of the fully-depreciated assets. This adjustment was done in the fixed network model
which was used for the calculation of the costs of fixed services.
Celcom suggests that historical cost data should be available and high data requirements
on TM should not prohibit the usage of the step-by-step approach. The MCMC confirms
that these two reasons definitely did not prohibit the usage of the step-by-step approach
because it actually does not use any historical cost data and does not result in high data
requirements on TM. Discussion of Celcom’s comments about the use of a top-down model
or the step-by-step approach is provided in Question 4.
The MCMC notes the concerns expressed by Celcom on the effect of a change of
methodology from a top-down approach with straight-line depreciation to a bottom-up
approach with tilted annuities on the incumbent’s assets, which are in the later years of
their lifetimes. This discussion was covered in Question 4. The MCMC also agrees with
Celcom’s comment that showing sensitivity using simple annuity has little merit, because
the assets in the model change prices, and tilted annuity is therefore the preferred method.
Many submissions highlighted that the fixed model populated with only dummy input data
limits the evaluation or verification by stakeholders. The MCMC would like to stress that
this is not true. Though it is not possible to evaluate the actual input values, it is possible
to evaluate the model itself and all the calculations performed in the model. All calculation
formulas are included in the model and third parties can follow all calculation steps from
the input cells to the costs of the services. In addition, the model was accompanied by a
model manual which describes all calculations in detail. On a related point, to address
comments by Maxis, the MCMC confirms that it has reviewed and benchmarked the model
inputs after the data was received from TM. Furthermore, the MCMC has also reviewed the
model calculations and the routing factors.
The MCMC understands the desire of other stakeholders to see the real input values but it
is unfortunately not possible due to confidentiality restrictions. Celcom argues that many
of the input data should not be confidential. However, the MCMC must respect the
confidentiality placed on data by its owners.
Page 64
There was a suggestion that the MCMC should populate the model with some realistic input
values (by using benchmarks or disguised values), however, this suggestion provides no
further information to stakeholders. As mentioned above, the evaluation of the model
calculations is possible and evaluation of real input values would not be possible, even
where benchmarks or disguised values were entered into the model.
The MCMC confirms that the fixed model does not include costs of any stranded assets.
As the number of assets is driven by the service volumes, assets which are not necessary
to provide services are not included in the model. The model assumes replacement of
copper cables in the access network and MSANs with fibre cables and OLT respectively due
to the migration of customers from copper-based broadband services to fibre-based
broadband services. Stranded assets are thus removed from the model. This corresponds
to the comment from Maxis that TM will no longer offer ADSL services for premises that
have subscribed to HSBB.
The MCMC agrees with Maxis to use 35 years (which is the midpoint of BEREC’s Regulatory
Accounting Review’s benchmark of 30 – 40 years) as the lifetime of the civil infrastructure
in the model.
The MCMC does not share the opinion of TIME that LRIC+ heavily discounts the previous
fibre and civil investments made, which were expensive in the earlier rollout years, and
that a FAC model should be used for fibre investment. The change in the value of assets
is caused by the use of current cost in the model. However, current cost can also be used
in a FAC model, therefore using a FAC model would not resolve this issue. In addition,
the MCMC is of the view that it is correct to use service costs calculated on the basis of
current cost of assets for price regulation.
The issues raised by TM regarding the mark-down of fixed assets are dealt with in Question
4 on costing methodologies. TM specifically disagrees with the mark-down of local fibre
rings as they connect outdoor MSAN cabinets to exchanges and should be treated as E/S
fibre rings. The MCMC agrees with TM and has changed the calculation of the local rings
using E/S cable (access network fibre cable) instead of the transport network cable.
Service volumes
Submissions received
Celcom notes that data traffic is expected to increase by 14% which may be reasonable,
in principle. However, Celcom comments that it is unclear whether the implied busy hour
Kbps per subscriber matches TM’s actual Kbps per subscriber. In addition, it is not possible
Page 65
to observe how the forecast traffic growth rate of 14% per annum has been derived or
how it has been implemented in the model.
On the service demand and traffic, Maxis agrees on the usage of TM’s past, current and
future volumes, however, suggests that the MCMC should clarify on how it verifies and
confirms the accuracy of the data. Further, clarification should also be provided on how
the input data is benchmarked with other jurisdictions. Maxis also provided its estimates
on data traffic trends, which grows at 38%. Finally, Maxis expresses its concern that a
service volume of 1 unit is used for those regulated services where there is no take-up,
and suggests that, at a minimum, to use some initial service volume estimates and to
perform sensitivity analyses to verify the reasonableness of the estimates and its outputs.
Discussion
As requested by Celcom, the MCMC provides further explanation regarding the data traffic
in the fixed model. The expected increase of data traffic in the fixed model was estimated
by TM as part of the data collection. The future traffic was estimated individually for each
service and in total it gives the 14% increase. Therefore, the value of 14% as such was
not estimated and also was not used in the calculation in the model. The MCMC is of the
same opinion as Celcom that this increase seems to be reasonable, and accepted the
expected traffic values submitted by TM.
Maxis provided an estimate of future data traffic growth in the fixed network of 38%. As
Maxis did not provide any clear reason why the 38% growth should be better than the
14% growth expected by TM, the MCMC decided not to make any changes in this regard
to the model.
The MCMC notes Maxis’s concern regarding the use of service volume of 1 unit for services
without take-up. The MCMC would like to assure Maxis that using the volume of 1 unit
does not have any effect on the unit costs of the services. There is no network element,
which would be used only by these services which have no take-up. Therefore, there is
sufficient traffic or number of lines going through all network elements from services which
have some real take-up and the costs of network elements per unit of service are based
on reasonably utilized network elements. Adding an additional unit of the service which
has no take-up to the existing traffic or number of lines enables the model to calculate the
costs of each network element per unit of service under the existing total service demand.
Conversely, by adding higher volumes to services without take-up would artificially
increase the total traffic or number of lines in the model above the real service demand in
the market.
Page 66
Data inputs
Submissions received
Celcom notes that it is apparent that in almost all instances, the fixed model uses data
from TM without any attempt to assess whether its network is efficient. Celcom provides
a few examples: the trench length of TM’s network may be greater than that of an efficient
network; TM’s copper network may be too long; and the mix of cable types in TM’s network
may be inappropriate. The objective of the bottom-up model is to model an efficient
operator with the same coverage and demand as the incumbent. Using the incumbent’s
actual data, such as route lengths and operating costs, goes against this objective and
means that there is insufficient optimisation in the model and so it falls short of best
practice.
Celcom adds that routing for leased lines may be different from those for other traffic. A
satisfactory model would need to capture actual leased line routings or to use end-to-end
data on leased lines in order to work out optimal routings, and it is unclear to Celcom
which approach has been adopted.
Maxis provides further inputs on the values that the MCMC has used in the model
reconciliation section, where the average SMS size should be 160 characters, lifetime of
drop cables and drop points is 20 years, lifetime of BTU at 7 years and the unit price for
FTTH cabinets at RM12,000 (outdoor) or RM13,000 (indoor).
Discussion
The MCMC confirms the understanding of Celcom that routing factors for leased lines are
different from those for other types of traffic. The other issues highlighted by Celcom
regarding network optimisation and the use of TM’s data are covered in Question 6 on
costing methodology adopted.
The MCMC does not agree with Maxis that the average SMS length should be 160
characters, because 160 characters is the maximum length of an SMS and not the average
length. Therefore, the MCMC decided not to change this value in the model.
The MCMC also does not agree with Maxis that the lifetime of a BTU should be 7 years.
The MCMC is of the opinion that a BTU will be replaced more frequently due to increases
in broadband speeds. Therefore, the MCMC decided not to change this value in the model.
The MCMC decided to change the unit price of fibre cabinets to RM12,500 as submitted by
Maxis.
Page 67
Further, the MCMC decided to increase the lifetime of transport network cables, drop points
and drop cables to 25 years to be in line with the access network cables.
Network dimensioning
Submissions received
With respect to the discussion of network dimensioning in the PI Paper, Celcom says that
it was limited and inaccurate. Celcom cites that the PI Paper mentioned the dimensioning
of the access network; however, in reality there appears to be no dimensioning of the
access network. It only shows duct, trench, copper and fibre cable lengths with some
limited breakdown by part of the network for cable lengths. Celcom adds that as far as
can be detected, there appears to be no attempt to verify whether the lengths are those
of an efficient operator. The copper and fibre cable lengths are not even differentiated by
the number of pairs.
Celcom views that the discussion on contention ratios appears to have confused
dimensioning with allocation. Celcom clarifies that its concern is in the dimensioning of
the broadband network. In this regard, the average busy hour usage or Mbytes per month
is a reliable method. On the other hand, it is not obvious that using contention ratios will
either estimate demand correctly or produce correct forecasts.
Celcom further opines that the MCMC’s response on MSANs is unconvincing. The purpose
of a bottom-up model is to model an efficient network, and the fact that the model
produces the same results as in TM’s actual network does not provide confidence that the
approach is correct. In addition, it is implausible that data on the number of subscribers
at individual MSAN sites and the cost of different sizes of MSANs cannot be obtained.
Celcom comments on the statement by the MCMC that DWDM is required to send the
signal from the MSAN to the local ring fibre. The role of DWDM is not to transmit data,
but to expand the capacity that can be transmitted on a given number of fibres. In a
bottom-up model, it is argued that any form of WDM is used where it is cheaper to combine
fewer fibres with such equipment than to install additional fibres. Hence, in the Swedish
and Danish models, little WDM is used, and none (to Celcom’s knowledge) use DWDM
between the MSAN and aggregation switch, though some use Coarse Wavelength Division
Multiplexing (CWDM). Celcom also provides further anecdotes in the UK which suggests
that CWDM, which is cheaper than DWDM, is used at around 20% of MSAN sites while
DWDM is used higher up in the network. Celcom concludes that the optimal amount of
WDM and DWDM in a bottom-up model is likely to be less than in an operator’s actual
network.
Page 68
Finally, on the allocation of chassis that is based on space occupied rather than on the cost
of cards, Celcom comments an allocation on the basis of cost is more consistent with the
established EPMU approach.
Maxis is unable to comment on the network dimensioning and suggests the MCMC to
benchmark the scorched node network dimensioning used in the model with those in other
countries, such as BT in the UK and Telstra in Australia.
Discussion
The issues highlighted by Celcom regarding the dimensioning of the access network,
network optimisation and the use of TM data are covered in Question 6 on costing
methodology adopted. In addition, the MCMC confirms Celcom’s understanding that the
fixed access network dimensioning is based on existing data from TM’s access network.
The number of equipment in the access network was divided by the corresponding number
of access lines (number of access lines using the particular type of access network
equipment) and this average use of equipment was multiplied by the expected number of
access lines in each of the modelled years. The fixed model contains the formulas
performing these calculations, and accompanying manual also describes the calculations
in detail.
Celcom further mentions that in the dimensioning of the access network, the cables are
not differentiated by number of pairs. While this is true, broad categories of cables (drop
cable, drop side cable, exchange side cable) and their average prices, covering the
different number of pairs in each of the categories, were used. The MCMC agrees that
using average values instead of individual cable types differentiated by number of pairs is
less precise, but more detailed data was not available. Further, as the largest contribution
to the cost of the cable route is the trench and duct, the influence of the number of pairs
(especially in the case of fibre optic cables) is not as significant. The MCMC would like to
add that under the step-by-step approach, which is preferred by Celcom, even these broad
categories could not be distinguished.
Based on Celcom’s comments, it appears to the MCMC that there is still a lack of clarity in
the usage of contention ratios in dimensioning as opposed to allocation. The contention
ratio is the relationship between the nominal speed of a broadband line and the minimum
guaranteed speed. This means that in the busy hour the customer must be able to use at
least the nominal speed of his line multiplied by the contention ratio. The network should
be dimensioned so that it can provide at least this minimum guaranteed speed to all
customers. Therefore, this amount of traffic from broadband lines was used for network
dimensioning. The same traffic was also used for allocation of costs.
Page 69
On the MSAN sizes, the MCMC agrees with Celcom that using different MSAN sizes would
be preferable over using an average MSAN size. However, it was not possible to obtain
data about the different MSAN sizes from TM, and therefore, the MCMC used the average
MSAN size. The MCMC has verified the result calculated using the average MSAN size
against the total GBV of MSANs from TM to ensure that there was no significant effect.
Therefore, the MCMC is of the opinion that it can use the average MSAN size in the model
calculations even if it is not the most precise method of calculating the MSAN costs.
The MCMC agrees with Celcom’s submission that CWDMs would be sufficient in the local
rings. This change has now been incorporated into the final model calculation.
The MCMC does not agree with Celcom’s submission that chassis should be allocated
between individual cards based on their cost. While it is true that allocation based on cost
is consistent with EPMU, this should only be done where there is no direct causal
relationship. There is a strong causal relationship between chassis and cards because
each card occupies a slot in the chassis. Therefore, this causal relationship should be used
as a basis of the allocation and not the EPMU approach.
Submissions received
On network costing and service costing, Celcom highlights that there was no information
provided in the model on asset lifetimes and price rate changes, therefore it is difficult to
assess whether the assumptions used are reasonable. In addition, Celcom is unclear on
why operating cost information is also treated as confidential. Celcom further infers that
TM’s data and information may have been used without any cross-checks being carried
out or benchmarking of data with other operators in Malaysia or other jurisdictions.
Maxis is unable to provide further input for the network cost including the equipment cost
and economic lives, and requests for benchmarking with other jurisdictions’ data.
Likewise, for service costing, though Maxis agrees with the approach taken, with the lack
of visibility of the data, Maxis urges comparison with the common network routing factors
used in other countries.
Discussion
The MCMC notes the comments from Celcom and Maxis, and has already addressed the
issues on confidentiality of data and the verification and benchmarking of the data.
Page 70
Treatment of HSBB
Submissions received
Digi seeks clarification on how the 5% discount amount is assumed. In addition, Digi
views that the operator receiving the funds from the Government should provide more
clarity to the MCMC that they are indeed repaying the funding agency with full commercial
returns on the projects undertaken. Digi also submits that the cost of capital should
technically be lower due to the PPP.
On the treatment of HSBB subsidies, Maxis disagrees with the view of the MCMC to not
exclude the RM2.7 billion contribution from the Government from the total calculated cost
of access to Fixed Services. Maxis continues to view that a reduction in the asset base is
an economically efficient approach; rather than to apply a small discount of 5% to the
implied capital costs, as proposed by the MCMC. An ideal way is to exclude the
Government contribution and to provide additional sensitivity analysis on the impact to
the final prices. Maxis submits that the 5% discount measure, being such a significant
approach, is not sufficiently justified in the PI Paper and requests further clarification for
arriving at this quantum. Maxis notes that even when there is no return required, the
WACC is at 8.08%. This is high in comparison with the WACC applied by IDA on Netlink
Trust of 7%, where there is Government funding for fibre rollout, and that for Australia’s
NBN Co at 7.2%. In that regard, Maxis proposes that MCMC reconsider its approach on
treatment of HSBB subsidies by firstly, excluding the amount funded by the Government
from the total calculated costs for Fixed Services, particularly for Layer 2 and 3 HSBB
Network Services; and secondly, using TM’s last pre-tax WACC of 7.2%.
Discussion
Many submissions argue that applying the 5% discount on cost of capital is not enough to
deal with the Government subsidy. The MCMC has reconsidered the situation around the
Government subsidy and a final WACC value and, as discussed in the WACC section of this
PI Report, have set the cost of capital for the Government’s share of the investment in
HSBB at the risk-free rate, i.e. 4%, resulting in a WACC of 8.27% overall.
The MCMC continues to view that it is appropriate to use LRIC+ to model the fixed access
and core network. The MCMC also affirms that it is appropriate to adjust the fully-
depreciated assets in the fixed network. In addition, the fixed model accounts for the
migration effects from copper-based ADSL services to fibre-based FTTH services and
therefore does not include stranded assets. The MCMC has taken note of the submissions
Page 71
and has made the amendments to the model, where it is appropriate. Finally, the MCMC
has decided that in taking account of the Government’s contribution to the HSBB, the
WACC for the fixed network services is set at 8.27%.
Question 19:
Do you have any comments on the proposed prices for the fixed services in the Access
List?
Submissions received
Celcom supports the proposed charge level and structure for Fixed Network Origination
and Termination Services, subject to the amendments made to the fixed model, as
proposed by Celcom.
Digi notes that the absence of data in the fixed model has hampered the ability of Digi and
other Access Seekers to fully evaluate the costs and allocation of fixed assets.
Nevertheless, Digi submits that the cost of Fixed Network Termination Service should be
consistently lower than the Mobile Network Termination Service as set by other regulators.
Hence, Digi suggests that a benchmarking of TM’s costs with other regulated fixed
operators be carried out.
Maxis agrees with the approach taken to have a single Fixed Network Termination rate
and a single Fixed Network Origination rate, without distinction between PSTN and IP calls
or between local, single tandem, double tandem and submarine cable. However, Maxis
comments that prices for both Fixed Network Origination Service and Fixed Network
Termination Service are too high. According to Maxis’s benchmarked countries, the
proposed fixed termination rates are among the highest; and the fixed termination rates
are generally lower than mobile termination rates. This is not the case in Malaysia.
webe submits that it is confusing that the cost of submarine cable is not a significant
component for fixed voice traffic but it is significant for mobile voice traffic. It is only
logical that with the declining trend of volume of fixed minutes, with the same operational
cost, the cost per minute should increase. However, this is not reflected in the proposed
methodology and in the prices.
Page 72
Discussion
The MCMC confirms its decision to set only one fixed termination rate and one fixed
origination rate without distinction between PSTN and IP calls or between local, single
tandem, double tandem and submarine cable. This decision was welcomed by Celcom and
Maxis.
The MCMC does not agree with the comments made by Digi and Maxis that the cost of the
fixed termination service must be lower than the cost of the mobile termination service.
It is true, that it usually is the case, but it does not mean that the cost models must be
modified in order to achieve such a result. The MCMC is of the opinion that the costs in
both cases should be calculated objectively reflecting the reality even if the difference
between the costs of fixed and mobile termination service are not as expected.
On webe’s query on why the submarine cable cost is not significant for fixed voice services
(and therefore the fixed voice services are not split into services using and not using the
submarine cable) while it is significant for mobile voice services (and therefore the mobile
voice services are split into services using and not using the submarine cable). The reason
is that the costs of the submarine cable between Peninsular Malaysia and Sabah and
Sarawak in the fixed network are shared with data transmission services. As voice services
consume a very low capacity compared to the data transmission services, the share of
submarine cable costs allocated to fixed voice services is very low. Therefore, the cost of
the submarine cable is not significant in relation to the other costs of the fixed voice
services. Consequently, the total costs of fixed voice services within Peninsular Malaysia
or Sabah and Sarawak and between Peninsular Malaysia and Sabah and Sarawak are
almost the same. The MCMC has received further feedback during the Public Inquiry, and
will discuss the submarine cable with respect to mobile services in Question 24.
webe also comments that with the declining trend of volume of fixed minutes and the
same operational cost, the cost per minute should increase. The MCMC would like to
explain that the voice traffic represents only a small part of the total traffic in the
submarine cable and the decrease in voice minutes therefore causes only a small increase
in per unit costs. If the voice traffic would represent for example 10% of the total traffic
in the submarine cable and the amount of voice minutes would decrease by 50%, it would
mean only a 5% decrease in total traffic volumes and a 5.2% increase in per unit costs
under the assumption that data traffic would remain unchanged. However, data traffic is
actually growing so in reality even if voice traffic is decreasing the total traffic in the
submarine cable increases and therefore the per unit costs decrease.
Page 73
MCMC’s Final Views
The MCMC confirms its preliminary view and sets the final prices for the Fixed Network
Origination Service and Fixed Network Termination Service as follows.
Submissions received
Altel, in its submission to Question 9, also highlighted that the proposed installation
charges for Interconnect Link Service is excessively high compared to the current prices.
APCC agrees with the decision to regulate the prices of Interconnect Link Service as such
transmission services face high barriers to entry and remains a key input to all
telecommunications services. However, the proposed regulated prices are almost a 10-
fold increase and is highly disproportionate to the 2015 price in the Mandatory Standard
on Access Pricing in 2012. Even taking into account the migration to bottom-up LRIC+
costing method, the increasing cost of cables and related civil infrastructure, APCC views
that on balance this does not address such a substantial increase in prices. Hence, APCC
proposes to maintain the previous method used which is based on the cost of trunk fibre.
Celcom views that the price for Interconnect Link Service is not cost reflective as it is 40
times greater than the current regulated charges, and should be reduced. It is plausible
that the trench, duct and cable used have not been shared with other users. The
installation charge also appears unrealistically high. One plausible reason is that the man
hours provided by TM includes overheads which are separately applied in the fixed model.
Celcom’s own data shows that the average cost of installing an Interconnect Link is below
the RM20,000 used in the model.
Maxis agrees to regulate the prices of Interconnect Link Service. However, Maxis
highlights that within the 5 years’ span, the cost of providing Interconnect Link Service
Page 74
has increased by more than 43 times. Hence, Maxis recommends reviewing the input
data, assumptions and network dimensioning.
TM also submits that there is a more than 4,000% price jump for Interconnect Link Service
as compared to the most recent regulated price in the Mandatory Standard on Access
Pricing. It is illogical that there is such a price increase as the network has been
established since the emergence of mobile operators, has remain unchanged and is almost
fully-depreciated. This price increase would also hinder operators to move into IP
interconnection as they would enjoy higher revenue over legacy assets. Hence, TM
proposes a consistent approach where the IP environment should be applicable and the
price should be distance independent, as adopted for transmission services.
U Mobile submits that the proposed price of Interconnect Link Service has increased by
43-fold from the price in 2015, whilst the service has not changed. U Mobile sought
clarification on the vast discrepancy. Further, in the last Access Pricing review in 2012,
there was a downward trend of prices. Finally, the installation cost for Interconnect Link
Service seems excessive and unjustifiable.
webe submits that there are inconsistencies of the cost elements in the calculation of
different services. webe notes that there is a significant price jump of more than 4000%
for Interconnect Link Service as compared to the regulated prices in 2015. The increase
is not reasonable as this is not a new service and there has not been a change to the
service. In addition, most of the equipment should be almost fully-depreciated. The price
trend would not encourage service providers to move to IP interconnection as they could
enjoy higher revenue over legacy assets. Finally, webe opines that a consistent approach
should be adopted, i.e. that the proposed price should be distance independent as adopted
for transmission service.
Discussion
APCC, Celcom, Maxis, TM, U Mobile and webe commented on the high costs of the
Interconnect Link Service and the change from the previous price. This was caused by
the fact that the Interconnect Link Service was calculated as an independent, dedicated
fibre cable using a completely new route. Based on the comments from the respondents,
the MCMC realized that this is not realistic and has now recalculated the cost of the
Interconnect Link Service as a fibre link.
Altel, Celcom and U Mobile mentioned very high costs of installation of the Interconnect
Link Service. The costs were calculated based on the number of working hours per
installation and the average hourly rate of a technician performing the line installation.
Page 75
Both values were submitted by TM. Based on this feedback, the MCMC has decided to
revisit these inputs and reduced the number of hours needed for the installation.
TM and webe propose for the price of the Interconnect Link Service to be distance
independent in line with the pricing structure of the transmission services. The MCMC
does not agree with this proposal because the Interconnect Link Service does not represent
transmission in the core network of the access provider but it represents a dedicated link
between the core network of the access provider and the location of the access seeker.
The costs of the Interconnect Link Service are therefore distance dependent.
The MCMC confirms its preliminary view and sets the final prices for Interconnect Link
Service as shown in the table below.
Wholesale Local Leased Circuit Service, Trunk Transmission Service and End-to-End
Transmission Service
Submissions received
Altel, in its submission to Question 9, also highlighted that the proposed installation
charges for Trunk Transmission Service and End-to-End Transmission Service are
excessively high compared to the current prices.
APCC agrees with the decision to regulate the price of Wholesale Local Leased Circuit
Services. APCC notes that there was no distinction between the prices in Peninsular
Malaysia from those in Sabah and Sarawak, as in the Mandatory Standard on Access
Pricing in 2012. There was no reason provided and APCC is concerned that such an
approach means that the geographically averaged prices used would result in higher prices
overall. In addition, based on global benchmarks, installation costs should not exceed
USD 200, and the proposed installation prices are excessive and should be revised in line
with international benchmarks. Finally, APCC submits that the current categories of prices
are too wide and proposes the following segmentation: up to 1 Mbps, 1 Mbps – 10 Mbps,
10 Mbps – 100 Mbps, 100 Mbps – 500 Mbps, 500 Mbps – 1 Gbps and 1 Gbps – 10 Gbps.
Page 76
Celcom submits that the price of Wholesale Local Leased Circuit Service is above efficient
cost levels and should be reduced. In addition, the proposed scope of the regulated service
should be clarified. The charges of the intermediate services between 1 Mbps and 1 Gbps
should also be included. Similarly, Celcom views that the price for Trunk Transmission
Service and End-to-End Transmission Service are above efficient cost levels and should be
reduced. There should also be charges for intermediate services between 1 Mbps and 1
Gbps.
Digi supports the removal of distance and circuit-based pricing to capacity-based charging
for Trunk Transmission Services. However, Digi proposes the prevalent bandwidth range,
i.e. 5 Mbps, 10 Mbps, 20 Mbps, 100 Mbps, 200 Mbps, as there is a huge variance between
1 Mbps and 1 Gbps that are proposed by the MCMC. In addition, Digi submits that the
proposed prices for 1 Gbps appears to be overstated and is higher than the current market
rates.
Fibrecomm submits that the prices for Wholesale Local Leased Circuit Service, Trunk
Transmission Service and End-to-End Transmission Service are comparatively low when
compared to the current market price. Fibrecomm suggests that the prices should increase
taking into consideration the cost of maintenance of equipment, cables and labour for
installation, as maintenance and cost of labour increases each year. Finally, Fibrecomm
seeks clarification whether the price is for 1 Mbps or for each Mbps, where if 5 Mbps is
requested, the applicable price is the price of each Mbps multiplied by 5.
Maxis agrees with price regulation of Wholesale Local Leased Circuit Service, Trunk
Transmission Service and End-to-End Transmission Service. However, Maxis proposes for
more tiers to be included, as there could be potential conflict between the Access Providers
and Access Seekers. For example, if 100 Mbps is subscribed, should the price be the price
of 1 Mbps multiplied by 100 (RM12,600 per month) or 1 Gbps divided by 10 (RM2,859 per
month). Hence, Maxis proposes the following tiers: 1 Mbps, 10 Mbps, 100 Mbps, 200
Mbps, 500 Mbps, 750 Mbps, 1 Gbps, 3 Gbps, 5 Gbps, 10 Gbps, 20 Gbps, 50 Gbps and 100
Gbps. Secondly, the proposed regulated prices are substantially high in comparison with
the current market prices offered by Access Providers in Malaysia and regulated prices in
the UK and Australia. Hence, Maxis proposes that the input data, assumptions and
network dimensioning be reviewed.
Sacofa submits that as a fibre operator operating more than 11,000 km of fibre in Sarawak,
it was not invited to submit data and comment on the cost model. It views that the
proposed cost for Wholesale Local Leased Circuit Service, Trunk Transmission Service and
End-to-End Transmission Service does not include operational and maintenance cost of
fibre. Further, due to the geographical area, the proposed prices may not be sufficient to
cover their operational and maintenance cost.
Page 77
TIME disagrees to the proposed single price for a wide bandwidth range. Prices are
dependent on the size of bandwidth, tenure of lease and location of customers. Hence,
TIME proposes the consideration of additional parameters.
As regards to Trunk Transmission Service, TM requests clarification that the price of the
service per 1 Mbps and per 1 Gbps is costed from exchange to exchange and does not
include a connection to the POI.
With regard to End-to-End Transmission Service, TM notes that the 1 Mbps has been
costed based on a copper line, and the cost of fibre access between 1 Mbps and 1 Gbps
has not been considered. This may cause confusion during negotiation of Access
Agreements, hence TM request that either the approach for End-to-End Transmission
Service is changed to encompass 1 Mbps based on fibre access using UPE equipment or
to explicitly clarify in the PI Report that the 1 Mbps is based on copper only.
TM notes that the Wholesale Local Leased Circuit Service has been modelled to include
only the cost of access network elements, with no traffic driven elements. However, TM
highlights that this service comprises of local access (equipment and fibre access), trunk
and physical interconnection. Hence, TM sought clarification on the proposed price for this
service. Further, TM notes that there is no bandwidth cap for the 1 Gbps using DWDM,
and may lead to confusion during negotiation of Access Agreements. TM also urges that
the Wholesale Local Leased Circuit Service using DWDM, which is a customised solution,
should not be regulated.
On Wholesale Local Leased Circuit Service and End-to-End Transmission Service, U Mobile
notes that the installation cost for both are exorbitant. The proposed prices for End-to-
End Transmission Service also do not reflect current price in Malaysia or internationally.
The price for 1 Mbps per month is at least 10 times higher than the commercial prices. U
Mobile agrees to the regulation of the prices of these services and urges to review the
accuracy of the cost data.
YTL submits that the prices should be clear and unambiguous. For example, for End-to-
End Transmission Service, the prices provided are for 1 Mbps and 1 Gbps. As actual access
requests may be for more or less than these quantities, YTL request for guidelines on the
computation of prices. This applies to other services, as well.
Discussion
APCC mentioned the very high costs of the installation of the Wholesale Local Leased
Circuit Service. Altel and U Mobile add that this is the case also for installations of
transmission services. The costs were calculated based on the number of working hours
Page 78
per installation and an average hourly rate of a technician performing the line installation.
Both values were submitted by TM. Based on this feedback, the MCMC has decided to
revisit these inputs and reduced the number of hours needed for the installation.
APCC noticed that the Wholesale Local Leased Circuit Services are not split between those
in Peninsular Malaysia and those in Sabah and Sarawak. The MCMC decided not to make
this distinction because the wholesale local leased circuits represent the local tail of a
leased line and are therefore similar to access lines. None of the access network services
are split into services in Peninsular Malaysia and services in Sabah and Sarawak.
Based on the request from Celcom and TM, the MCMC would like to clarify that the
Wholesale Local Leased Circuit Service represents the tail segment of a leased circuit. It
is provided via an access line and dedicated transmission equipment on this access line.
The MCMC also clarifies that Wholesale Local Leased Circuit Service does not include trunk
transmission and physical interconnection. If these two components are used, they should
be charged separately as the Trunk Transmission Service or the Interconnect Link Service,
as applicable.
APCC, Celcom, Digi, Maxis, TIME and YTL requested that charges for intermediate speeds
between 1 Mbps and 1 Gbps for Wholesale Local Leased Circuit Service, Trunk
Transmission Service and End-to-End Transmission Service should be provided. The
MCMC has decided to add more granular segmentation by speed for Trunk Transmission
Service and End-to-End Transmission Service. However, in the case of the Wholesale
Local Leased Circuit Service, more granular segmentation would not make sense, since all
the speeds within the currently defined ranges would have the same costs. The reason is
that the wholesale local leased circuit is a line in the access network, the transmission
equipment used on this line is dedicated to this line only and the capacity cannot be shared
with other subscribers. As a result, the cost is the same for all speeds which can be
handled by the same transmission equipment type. The speed ranges used by the MCMC
represent the speeds which can be handled by the same transmission equipment type.
Celcom and Maxis also stated that the costs for the Wholesale Local Leased Circuit Service,
Trunk Transmission Service and End-to-End Transmission Service are above the level of
efficient cost and should be reduced. U Mobile also submitted that End-to-End
Transmission Service prices are too high. On the other hand, Fibrecomm submitted that
the same three services are low compared to market prices. The MCMC is of the opinion
that the costs of the services should be calculated objectively reflecting the reality and the
model should not be modified just in order to achieve the desired outcome. However,
because none of the respondents provided any justification, the MCMC did not make any
specific changes to these services based on this comment.
Page 79
Fibrecomm and Sacofa commented that costs for the Wholesale Local Leased Circuit
Service, Trunk Transmission Service and End-to-End Transmission Service are low and
that they do not include operational and maintenance cost. The MCMC confirms that the
operational and maintenance costs of fibre are included in the calculation and are
increasing each year. However, the traffic increases even more which results in lower unit
costs.
Fibrecomm and Maxis sought clarification on how prices for different speeds should be
calculated. The MCMC would like to explain that for the Trunk Transmission Service, the
costs increase linearly with the capacity (for example, costs of 5 Mbps is equal to 5 times
costs of 1 Mbps). The MCMC decided to add more granular segmentation by speed for
these services which should solve this concern expressed by Fibrecomm and Maxis.
The MCMC also clarifies, as requested by TM, that the Trunk Transmission Service includes
costs only for transmission in core network. It does not include a connection to the POI.
TM proposed to change the calculation of the End-to-End Transmission Service so that the
access segments use fibre access lines for all speeds or to explicitly clarify that the speed
of 1 Mbps uses copper access lines in the access segments. The MCMC does not see any
reason why the speed of 1 Mbps should use fibre access lines if copper access lines are
sufficient and their costs are lower. Therefore, the MCMC clarifies that the End-to-End
Transmission Service uses copper access lines for speed of 1 Mbps and fibre access lines
for speeds higher than 1 Mbps.
The MCMC confirms its preliminary view to regulate the prices for Wholesale Local Leased
Circuit Service, Trunk Transmission Service and End-to-End Transmission Service. The
MCMC has decided to not regulate the prices of Trunk Transmission Service and End-to-
End Transmission Service above 5 Gbps as there appears to be competitive supply. In the
event that there are competition complaints or disputes arising, the MCMC would
intervene, as appropriate. The final prices for Wholesale Local Leased Circuit Service,
Trunk Transmission Service and End-to-End Transmission Service as follows.
Page 80
Table 6: Wholesale Local Leased Circuit Service Final Prices
2018 2019 2020
Wholesale Local Leased Circuit Service up to 1 Mbps
48 49 50
(RM/month)
Wholesale Local Leased Circuit Service from 1 Mbps up
634 612 593
to 1 Gbps (RM/month)
Wholesale Local Leased Circuit Service from 1 Gbps up
16,042 15,432 14,869
to 10 Gbps (RM/month)
Wholesale Local Leased Circuit Service from 1 Gbps using
949 917 888
DWDM (RM/month)
Page 81
2018 2019 2020
Trunk Transmission Service within Peninsular Malaysia
8,836 7,942 7,273
and within Sabah and Sarawak 1 Gbps (RM/month)
Trunk Transmission Service between Peninsular
38,007 34,855 32,245
Malaysia and Sabah and Sarawak 1 Gbps (RM/month)
Trunk Transmission Service within Peninsular Malaysia
26,508 23,825 21,820
and within Sabah and Sarawak 3 Gbps (RM/month)
Trunk Transmission Service between Peninsular
114,022 104,565 96,734
Malaysia and Sabah and Sarawak 3 Gbps (RM/month)
Trunk Transmission Service within Peninsular Malaysia
44,180 39,709 36,367
and within Sabah and Sarawak 5 Gbps (RM/month)
Trunk Transmission Service between Peninsular
190,036 174,275 161,224
Malaysia and Sabah and Sarawak 5 Gbps (RM/month)
Page 82
2018 2019 2020
End-to-End Transmission Service between Peninsular
Malaysia and Sabah and Sarawak 500 Mbps 19,825 18,243 16,930
(RM/month)
End-to-End Transmission Service within Peninsular
Malaysia and within Sabah and Sarawak 750 Mbps 7,739 7,041 6,512
(RM/month)
End-to-End Transmission Service between Peninsular
Malaysia and Sabah and Sarawak 750 Mbps 29,104 26,753 24,802
(RM/month)
End-to-End Transmission Service within Peninsular
Malaysia and within Sabah and Sarawak 1 Gbps 10,103 9,166 8,459
(RM/month)
End-to-End Transmission Service between Peninsular
39,274 36,079 33,430
Malaysia and Sabah and Sarawak 1 Gbps (RM/month)
End-to-End Transmission Service within Peninsular
Malaysia and within Sabah and Sarawak 3 Gbps 28,406 25,660 23,597
(RM/month)
End-to-End Transmission Service between Peninsular
115,919 106,399 98,511
Malaysia and Sabah and Sarawak 3 Gbps (RM/month)
End-to-End Transmission Service within Peninsular
Malaysia and within Sabah and Sarawak 5 Gbps 46,078 41,543 38,144
(RM/month)
End-to-End Transmission Service between Peninsular
191,934 176,109 163,001
Malaysia and Sabah and Sarawak 5 Gbps (RM/month)
Submissions received
APCC agrees with the decision to regulate the price of Domestic Connectivity to
International Service. APCC also submits that there is a lack of competition in submarine
cable landing station services, with only TM providing such services and there is an artificial
monopoly. Further, APCC alleges that access to the submarine cable landing station is
conditional on the bundling of backhaul service. Hence, APCC urges the MCMC to
investigate TM’s abuse of its dominance.
Page 83
notes that it is more cost effective for an Access Seeker to use its own transmission from
the cable landing station, subscribe to Network Co-Location to co-locate its transmission
equipment in the cable landing station and subscribe to Domestic Connectivity to
International Services; rather than be forced to take Transmission Services from the
Access Provider.
TIME submits that the proposed price for Domestic Connectivity to International Service
is tremendously low. It asserts that there is a huge cost for establishing a submarine
cable landing station and other related expenses including land acquisition, approval fees,
building and terminal equipment installation and maintenance, domain experts and
physical cable landing, which runs up to millions of Ringgit annually.
TM reiterates its position provided during the Public Inquiry on Mandatory Standard on
Access with respect to Domestic Connectivity to International Service (and Network Co-
Location Service). TM highlights that due to significant national security risks and to avoid
service disruptions, TM should retain control of its existing security measures. TM
proposes that operators be allowed to apply various forms of interconnection services to
balance between national interest and fair access.
Discussion
The MCMC confirms that the understanding of Maxis is correct and that the Domestic
Connectivity to International Services is the cross connection from the access seeker’s
equipment in the submarine cable landing station into the access seeker’s capacity in the
respective submarine cable system.
Regarding the comment from TIME that the proposed price for Domestic Connectivity to
International Services is tremendously low while the costs of establishing and operating a
submarine cable landing station are very high, the MCMC has to explain, that the service
called Domestic Connectivity to International Services includes only a tie cable connecting
the equipment of access seeker located in the submarine cable landing station with the
equipment of the access provider that connects to the submarine cable system. It does
not include the co-location of the access seeker’s equipment in the submarine cable
landing station and it also does not include any costs of the international traffic. The co-
location in the submarine cable landing station is charged separately (Network Co-Location
Service based on one square metre in an international submarine cable landing station).
The MCMC confirms its preliminary view and sets the final prices for Domestic Connectivity
to International Services in the following table.
Page 84
Table 12: Domestic Connectivity to International Services Final Prices
2018 2019 2020
Domestic Connectivity to International Services monthly rental
3.17 3.33 3.49
(per 5 m) (RM/month)
Domestic Connectivity to International Services installation
85 89 94
(RM/installation)
Layer 2 HSBB Network Service with QoS and Layer 3 HSBB Network Service
Submissions received
APCC agrees with the decision to regulate the price of Layer 2 HSBB Network Service with
QoS and Layer 3 HSBB Network Service, and that the prices for the services should
decrease over time due to the decreasing costs of the transmission equipment and
increasing traffic in the core network.
Celcom perceives that the prices of HSBB services are above efficient cost levels and
should be reduced. There is also a move to set prices on a per-subscriber basis rather
than setting overall capacity charges, and only a narrow choice of products is included. In
particular, Celcom submits that the current charging structure based on overall capacity
should be maintained, with per-Mbps tariff tiers introduced for 100, 300 and 500 Mbps
total capacity with no minimum commitment. This will reduce barriers to entry and
expansion for relatively small Access Seekers such as Celcom. In addition, there is no
distinction made between residential and business users, and Celcom indicates that both
types of users have different needs. Finally, Celcom proposes an ex-ante margin squeeze
to be applied to TM’s HSBB offers.
Maxis supports the price regulation for both Layer 2 HSBB Network Service with QoS and
Layer 3 HSBB Network Service, however views the prices for both as substantially high.
Based on Maxis’s analysis of the proposed price of RM133 per month (based on RM94
recurring charge and RM39 installation charge over 24 months), it is higher than TM’s
retail price of RM129 per month. Currently, there is a RM99 promotion for 10 Mbps, that
is very close to the proposed recurring charge of RM94 per month. This would not allow
the Access Seeker to be able to compete with TM. For the higher speed such as 30 Mbps,
50 Mbps and 100 Mbps, the proposed regulated prices are not above TM’s retail prices,
however, they cause a significantly negative margin to the Access Seeker of 63%, 21%
and 19% respectively. In the UK, the regulated prices by Ofcom for the high speed fixed
broadband is only between 12% to 17% of the operator’s most popular retail prices, and
this provides a margin of between 83% to 88% for the Access Seekers to recover their
costs and compete against the incumbent’s retail packages. In contrast, the proposed
Page 85
regulated prices in Malaysia is more than 80% of the retail packages offered by TM to its
end users.
Maxis submits that, in order to sustain competition, the regulated prices of HSBB Network
Services are at the minimum 40% to 50% of the incumbent’s retail prices. This is derived
based on Ofcom’s final statement of the “Fixed Access Market Reviews: Approach to the
VULA Margin” published on 15 March 2015, where Ofcom requires BT to maintain a
minimum margin of approximately £22 to £25 between the wholesale VULA prices and its
own retail packages to end users based on its consumer’s portfolio of fibre-based
packages, rather than individual products or bundle.
Further, Maxis also submits that the structure of the regulated prices should follow that of
TM’s wholesale prices, to separate into port and bandwidth. This is also implemented by
ACCC in Australia. This structure would provide flexibility to the Access Seeker to offer
different forms of speeds to their end users.
Finally, as a fall back, Maxis proposes the MCMC to consider the retail minus approach in
particular for the HSBB Network Services that are most important to the development of
the country.
TM comments that Layer 3 HSBB Network Service without network service is not in the
Access List and should not be included under the purview of this review. Further, TM has
highlighted that the proposed pricing structure reflects retail pricing and is inappropriate
at the wholesale level. Firstly, TM will not achieve cost recovery with this pricing structure
as it makes no allowance for incremental cost in preparing and conditioning the local ring,
aggregation core, BRAS and core ring to align with the structure. Secondly, the proposed
pricing structure is not ready for immediate deployment as there needs to be modification
to the back-end system. Thirdly, the structure encourages the Access Seeker to be
positioned as a Reseller/End User than as a Service Provider.
TM also clarifies that the Layer 3 HSBB Network Service with network service should only
be applicable for High Speed Internet and does not include value-added services such as
VOD and IPTV, which are charged separately. Further, the prices are for Point of
Interconnection (POI) at an Access Provider’s premises (which includes access and core
network) within Peninsular Malaysia only. For Sabah and Sarawak, the cost of submarine
cable will be included. TM also proposes the pricing structure to be consistent with the
current wholesale offering for HSBA based on port and bandwidth. To be in line with the
pricing structure, TM proposes to remove contention ratio in the core network, as the
Access Seeker will use contention ratio. Despite the above, TM maintains that the HSBB
service should not be price regulated as there is already a significant wholesale HSBB
market and is more than sufficient to provide competitive broadband service offerings for
Page 86
the long-term benefit of end users. It is also best practice not to regulate where there is
significant capital investment deployed for new network infrastructure and where service
demand is still emerging.
Discussion
Many submissions requested that the pricing structure of HSBB services be changed to
that of BTU port and bandwidth (of the Service Gateway). The MCMC therefore decided
to develop a new pricing structure for Layer 2 HSBB Network Service with Quality of
Service and Layer 3 HSBB Network Service comprising of BTU port and Service Gateway
services. The Service Gateway component is split into Layer 2 and Layer 3. Further, the
MCMC has also removed the contention ratio in the core network, as proposed by TM.
However, the MCMC did not introduce the separation of residential and business users
which was requested by Celcom. Such a separation is not necessary because the access
seeker can set the parameters of the HSBB service for its customers and there is no
difference at the wholesale level. The BTU port costs and Service Gateway costs are the
same regardless whether the line is used by residential or by business subscribers.
The MCMC does not agree with the proposal from Maxis that a retail minus approach should
be used for pricing the HSBB services. Instead, the MCMC is of the opinion that cost-
based prices compliant with margin squeeze test represent a better option and agrees with
Celcom that margin squeeze tests can be applied to the HSBB offers of TM.
The MCMC notes the statement from Maxis that the prices for HSBB services are high.
However, the MCMC does not agree that wholesale prices should be equal to 40%-50% of
the incumbent’s retail prices just because this was the result in the UK. The MCMC is of
the opinion that the costs of the services should be calculated objectively reflecting the
reality and the model should not be modified in order to achieve the desired outcome.
The MCMC confirms the statement made by TM that the HSBB services are applicable for
high speed internet only and do not include value-added services such as VOD or IPTV.
Finally, the MCMC notes the submission made by TM that the prices for HSBB services
should only be applicable to Peninsular Malaysia, and for the BTUs that are located in
Sabah and Sarawak, the cost of submarine cable should be included. The MCMC has
requested for further details from TM and has examined this aspect carefully. The
information received indicates that there is currently a low number of BTU ports in Sabah
and Sarawak and therefore, it is assumed that there is no justification to establish POI(s)
in Sabah or/and Sarawak. As such, based on the current arrangement, the traffic is
brought to the POI in Peninsular Malaysia through the submarine cable. However, it is the
Page 87
understanding of the MCMC that TM has established a POI each in Sabah and Sarawak.
In that regard, the MCMC expects that when the number of BTU ports reaches the
appropriate level, it will be more efficient to open at least a POI in Sabah or Sarawak. The
regulated price for Layer 3 Service Gateway set by MCMC reflects this efficient situation
and does not include the cost of the submarine cable. Therefore, the same Service
Gateway charges are applicable regardless of the location of the BTUs, and there should
not be a separate submarine cable charge to the Access Seeker.
The MCMC’s final view is that the prices for Layer 2 Network Service with QoS and Layer
3 HSBB Network Service will be regulated.
As Layer 3 HSBB Network Service allows an Access Seeker to select between two POI
locations, the MCMC clarifies that the following Layer 3 HSBB Network Service prices are
applicable where the POI is at the Access Provider’s premises. Where the Access Seeker
requests that the POI be at the Access Seeker’s premises, then there is a necessity to
acquire an additional transmission service.
Similarly, an Access Seeker would also need to acquire an additional transmission service
for Layer 2 Network Service with QoS as the POI is located at the Access Seeker’s
premises.
The MCMC sets the final prices for Layer 2 HSBB Network Service with QoS and Layer 3
HSBB Network Service as follows.
Table 13: Layer 2 HSBB Network Service with QoS Final Prices
Page 88
Layer 2 HSBB Service Gateway 200 Gbps (RM/month) 659,963 580,173 519,402
Layer 2 HSBB Service Gateway 500 Gbps (RM/month) 1,649,907 1,450,432 1,298,504
Table 14: Layer 2 HSBB Network Service with QoS Installation Costs
Submissions received
APCC agrees with the decision to regulate the price of Duct and Manhole Access. However,
APCC cautions that allowing exclusivity in the Mandatory Standard on Access in 2016 sets
a precedent that exclusivity is allowed within the access regime and state-backed
companies may claim such exclusivity and hence deny access. Hence, APCC urges the
MCMC to remove all recognition of exclusivity in the access regime.
Page 89
On the Duct and Manhole Access, Celcom submits that the prices are above the efficient
cost levels and should be reduced. In addition, Celcom submits that there should not be
ex-ante regulation on Duct and Manhole Access provided between mobile operators.
Digi supports the price regulation of Duct and Manhole Access, as this is an essential input
to facilities and upstream elements. Multiple operators would have the ability to serve
locations otherwise operated by a single operator; hence, reducing duplication of facilities,
minimising disruption to the general public and increasing the speed of build-up of
broadband networks including fibre-based transmission backhaul which is required for LTE
networks.
edotco is an Access Seeker for Duct and Manhole Access. Based on its understanding,
these essential facilities are already fully-depreciated and in many cases, it is questionable
as to the level of maintenance undertaken. As such, edotco queries why the prices would
increase over time.
Maxis agrees with the MCMC’s decision to regulate the prices for Duct and Manhole Access.
Maxis observes that the proposed prices for Duct and Manhole Access are much lower than
those for Interconnect Link Service. It shows inconsistency in the fixed model and raises
questions on accuracy of data input, cost input, assumptions and network dimensioning.
The cost element for Interconnect Link Service is almost similar to the cost element for
Duct and Manhole Access. Even with the fibre cable cost for Interconnect Link Service,
the variance between the two proposed prices should not be as significant.
TIME views that the proposed price for Duct and Manhole Access as extremely low and
does not reflect the actual high costs involved in commissioning ducts and manholes,
including civil works, permitting fees, land acquisition (where applicable), labour and
maintenance.
TM provides feedback that the prices for Duct and Manhole Access is under-estimated due
to the enormous mark-down of assets by 90% and omission of civil works cost and new
service provisioning costs for backend systems. The price for Duct and Manhole Access is
inappropriate as it does not reflect the majority of exclusive areas in the new HSBB areas,
where demand is more likely to emerge. The proposed pricing structure which segregates
duct and manhole access is not operationally feasible due to the lack of inventory and
billing system. TM proposes for the pricing structure to be based on distance inclusive of
duct, manhole and cost of civil works. TM only submits that there are few countries in the
world that regulates duct and manhole access, and provides a benchmark of a selection
of those European countries, where the prices range from RM225 to RM412 per month per
km. The proposed prices are lower than the benchmark prices. Finally, TM requests for
Page 90
a grace period for new services and that the price for Duct to Manhole Access should not
be regulated.
Discussion
Celcom stated that the costs for Duct and Manhole Access are above the level of efficient
cost and should be reduced. However, because Celcom did not provide any justification,
the MCMC did not make any specific change to this service based on this comment. The
MCMC disagrees with Celcom’s view and clarifies that ex-ante prices for Duct and Manhole
Access also apply to those provided by mobile operators.
edotco asks why prices for Duct and Manhole Access increase over time if ducts and
manholes are already depreciated and there is no maintenance. The MCMC would like to
explain that if all ducts and manholes were fully-depreciated and there really is no
maintenance, then the costs of Duct and Manhole Access would be zero. However, there
are still some ducts and manholes, which are not fully-depreciated and their cost has to
be taken into account. There is also some maintenance of ducts and manholes. Both the
installation of ducts and manholes and their maintenance are labour intensive which
results in price increases over time.
The MCMC does not agree with Maxis that the costs of Duct and Manhole Access should
be similar to the costs of the Interconnect Link Service. The price of Interconnect Link
Service was already discussed above.
The MCMC agrees with TM and TIME that the costs of Duct and Manhole Access were too
low because they did not include the costs of civil works. Further, the MCMC is agreeable
to TM’s proposal that the pricing structure be based on distance inclusive of duct, manholes
and cost of civil works. The MCMC has therefore changed the calculation in the model and
has now calculated the costs of duct per km including the associated manholes and the
costs of trenches in which the ducts and manholes are laid.
The MCMC confirms its preliminary view and sets the final prices for Duct and Manhole
Access in the following table.
Page 91
Full Access Service, Line Sharing Service, Sub-loop Service, Bitstream Services, Digital
Subscriber Line Resale Service and Wholesale Line Rental Service
Submissions received
APCC disagrees with the decision of the MCMC to not regulate the price for Full Access
Service, Digital Subscriber Line Resale Service and Bitstream Services. All the services
satisfy the three criteria test: high, non-transitory barriers to entry; no trend towards
effective competition; and ex-post regulatory measures would be ineffective to remedy
the injury. Furthermore, for the Full Access Service, the moratorium specified in the
Ministerial Direction on High-Speed Broadband and Access List, Direction No. 1 of 2008,
has already lapsed; and the regulation of Full Access Service prices are more crucial today
given its utility and involvement in the HSBB network today.
Maxis proposes that the MCMC continue to regulate the prices of Full Access Service, Line
Sharing Service, Sub-loop Service, Digital Subscriber Line Resale Service, Wholesale Line
Rental Service, Bitstream with Network Service and Bitstream without Network Service.
It views that these services provide important additional options for the Access Seeker to
provide alternative and competitive fixed broadband services to end users. Currently, TM
still has a significant Streamyx customer base relying on copper technology (1,352
thousands) as compared to UniFi subscribers (1,007 thousands).
Finally, TM supports the non-price regulation for Full Access Service, Sub-loop Service,
Digital Subscriber Line Resale Service, Wholesale Line Rental Service, Bitstream with and
without Network Service and Network Co-Location.
Discussion
The MCMC has taken note of the submission from APCC and Maxis. However, in
determining which services on the Access List the MCMC should consider for price
regulation, the MCMC has taken into account the current and likely future take-up and
demand for such services. Services such as Full Access Service, Line Sharing Service and
Digital Subscriber Line Resale Service are now historic copper-based services which are
being replaced in Malaysia by HSBB fibre-based services. In order to encourage the take-
up of current technologies, the MCMC has decided not to price regulate these services.
The MCMC has decided not to regulate the prices for Full Access Service, Line Sharing
Service, Sub-loop Service, Bitstream Service with Network Service, Bitstream Service
without Network Service, Digital Subscriber Line Resale Service and Wholesale Line Rental
Page 92
Service. The updated cost model provides the following indicative prices for the respective
services.
Page 93
Table 24: Wholesale Line Rental Service Indicative Prices
2018 2019 2020
Wholesale Line Rental Service monthly rental (RM/month) 23 24 25
Wholesale Line Rental Service installation (RM/installation) 237 249 262
Submissions received
APCC disagrees with the decision of the MCMC to not regulate the price for Network Co-
Location Service. It satisfies the three criteria test: high, non-transitory barriers to entry;
no trend towards effective competition; and ex-post regulatory measures would be
ineffective to remedy the injury. In addition, APCC urges the MCMC to continue to regulate
prices for Network Co-Location Service as it did in the Mandatory Standard on Access
Pricing in 2012, given that the landscape has not undergone any substantial changes since
then. In addition, it is widely accepted that Network Co-Location Service is subjected to
price regulation in many jurisdictions, such as in Singapore.
Maxis proposes for the MCMC to continue to regulate the prices for Network Co-Location
Service as it is important for the Access Seeker to subscribe to other Facilities and Services
such as Trunk Transmission Service, Domestic Connectivity to International Service, Layer
3 HSBB Network Service without Network Service. If the prices are not regulated, the
Access Provider may impose higher prices to the Access Seeker which may be intentionally
done in order to protect their retail business and to slow down competition.
U Mobile comments that there was no explanation offered as to why Network Co-Location
Service is not price regulated. In any regard, the prices for the various types of spaces
do not reflect current practice. In many of the places leased by operators, commercial
prices prevail which tend to be higher than the regulated prices in Mandatory Standard on
Access Pricing in 2012. U Mobile suggests that it would be prudent to specify the types of
spaces that are considered bottleneck in nature, and the prices should logically be lower
than the commercially set prices.
Discussion
The MCMC notes the comments from APCC, Maxis and U Mobile that it should regulate the
prices for Co-Location Services. The MCMC is concerned that a number of different
operators with different cost structures offer Network Co-Location Services and the
imposition of regulated prices would require some operators to offer services at below cost
Page 94
prices. The MCMC therefore feels that it is more appropriate to allow commercial
negotiations to be used for these services.
The MCMC has decided not to regulate the prices for Network Co-Location Service. The
updated cost model provides the following indicative prices for Network Co-Location
Service.
Page 95
7. Mobile Services
7.1. Overview
The MCMC developed a mobile cost model based on the LRIC+ methodology for assessing
the cost of providing mobile origination and termination services. A “notional mobile
operator” version of this model with 25% market share was used to calculate proposed
prices for mobile origination and termination services.
Part E of the PI Paper concerned the mobile services. Section 18 described the
assumptions about all inputs to the model, including service demands and traffic, the
mobile network model, spectrum allocations and coverage, radio network costs, USP costs
and cost mark-ups. The impact of different levels of assumed WACC values were
described. The changes made as a result of the comments received after the initial model
viewing were also outlined.
In determining suitable regulated prices for mobile origination and termination services,
the MCMC had considered a number of issues: the likely trends in demand for voice, SMS
and particularly data services, the impact of using economic depreciation rather than tilted
annuities, the use of single rates rather than the previous local and national rates and
whether the costs for use of the submarine cable should be averaged into single national
rates or charged additionally to the specific services which use the submarine cable. These
issues were described and the MCMC’s proposed prices were also presented in the PI
Paper.
Question 20:
Do you have any comments on the proposed assumptions for the Notional Operator’s
services and volumes?
Submissions received
Celcom, Digi and Maxis view that the notional operator’s share of 25% does not represent
the Malaysian mobile market. The assumed market share was too large, resulting in
understated cost and the mobile operators not being compensated sufficiently in
accordance to the actual cost incurred to provide the mobile termination services.
According to Celcom and Digi, the current approach in the model attributes no role to the
MVNOs. Celcom suggests some compromise where the notional operator’s market share
is between 20-25%. However, Digi proposes that the MCMC adopt a forward-looking
Page 96
approach and model a 20% market share, taking into account the numerous MVNOs that
have an approximately 10% market share in 2016. Maxis highlighted that the total
prepaid market subscriber assumption for the notional operator is too aggressive and does
not represent a 25% market share in the mobile model. Maxis also views that the forecast
of data traffic per subscriber per month used in the model for 2019 and 2020 is too
aggressive as compared to the operators’ inputs. Maxis suggested that the MCMC use the
operators’ inputs to represent the actual trend in the Malaysian market.
On the other hand, TM and webe are of the opinion that the average data usage per
subscriber is understated in the mobile model. webe highlighted that both Celcom and
Maxis in their published annual reports, stated usage of more than 3GB per month in
Quarter 1 of 2017, making the 1.3GB basis for 2016 too low. TM highlighted that Celcom
reported a data usage of 6.2GB per month in Quarter 2 of 2017 (an increase of 120% from
previous year), while Digi and Maxis reported 6GB (doubled from the previous year) and
5.6GB (an increase of 97% from the previous year) respectively for Quarter 3 of 2017.
The level of usage is significantly higher than the model’s assumption. The mobile
operators’ retail pricing structure also provides an indication of mobile subscriber usage.
TM is convinced that the mobile data traffic assumed in the mobile model is significantly
lower compared to international benchmarks for countries comparable to Malaysia as well
as the financial reports of the Malaysian mobile operators. Increasing the data traffic
assumption to better represent traffic demand will have a dramatic result on the model
results for mobile origination and termination rates. TM suggests that the MCMC replace
the operators’ data with values based on the operator’s published data.
YTL views that the analysis using a notional operator distorts pricing for emerging players
offering voice services. This is because the model does not take into account the 2.3GHz
and 2.6GHz TDD spectrum, hence the prices are set solely on the homogeneity assumed
in the 4 mobile operator model. This may lead to under-recovery of cost unless different
access prices are recommended for the difference in spectrum allocated. YTL added that
the 2.3GHz and 2.6GHz have higher operational cost as more base stations are required
to achieve equivalent coverage of the notional operator operating on 900MHz and
1800MHz in addition to the 3G and 4G spectrums. YTL has achieved 80% coverage but
with lower volumes of traffic. As 2.6GHz spectrum is subject to targets set by the MCMC,
the targeted coverage implies CAPEX and OPEX expenditure irrespective of the traffic
achieved. Hence, there is under-utilized capacity.
Discussion
Celcom, Digi and Maxis assert that the MCMC has taken insufficient account of the role of
MVNOs in the Malaysian market and Celcom offers the opinion that, although Celcom
agrees that a fifth national operator should not be assumed, a market share somewhere
Page 97
between 20% and 25% should be used. The MCMC does not agree with the assertion that
insufficient account was taken of MVNOs. In fact, the total market figures based on inputs
from the four MNOs were adjusted to take account of MVNOs’ share of the market. It is a
characteristic of MVNOs, however, that although they compete with MNOs at the retail
level of the market, they do not have their own networks and instead use those of the
MNOs. It is reasonable, therefore, to assume that the Notional Operator would take a
share of this business and that its network would be correspondingly loaded.
Equally, although Maxis asserts that the assumed growth of prepaid subscribers is too
aggressive and does not reflect actual conditions in the market, the assumptions were
based on the inputs provided by Maxis and the other operators, adjusted to include the
shares of smaller operators and MVNOs. The MCMC recognises that there must be some
degree of uncertainty around projections of future market conditions, but is not persuaded
that the assumed subscriber levels are over-stated.
The views expressed by operators on the subject of data usage reveal a divergence
between Maxis, on the one side and webe and TM on the other. Whilst Maxis invites the
MCMC to consider its inputs to the modelling exercise and takes a less optimistic view of
data growth, webe and TM point to data published by the operators that points rather
starkly in the opposite direction (see the graph below). The MCMC is persuaded that the
data inputs provided by the operators represent a significant under-estimate compared to
the data published by them to the markets, both in terms of the starting levels and of the
rate of growth. The MCMC is therefore minded to adopt a more aggressive forecast for
data usage in line with the average of the reported figures and its projection, as shown in
Figure 1.
Page 98
Figure 1: Average Data Usage per Subscriber
(based on model inputs and operators’ published data)
The MCMC will revise the base case inputs to the model in line with the table below.
Question 21:
Do you have any comments on the proposed approach to the radio spectrum and
coverage assumptions?
Submissions received
Celcom supports the MCMC’s position to include the cost of coverage and spectrum fees
in the model. Celcom agrees that the allocation of costs according to routing factors is a
satisfactory approach. However, Celcom is concerned that 700MHz spectrum costs have
Page 99
not been included in the model. The amount paid by mobile operators far exceeds the
RM180 million used in the model. Celcom suggests that the model is adjusted to reflect
such costs. Celcom provided some data on the one-off payment, application fee and
irrevocable bank guarantee to substantiate this claim.
According to Digi, the model assumes a very different distribution of traffic between
technologies (2G, 3G, 4G) to that experienced by Malaysian operators. Digi views that
the traffic split between technology should not be based on population coverage as it is
largely dependent on the customer’s device usage and traffic consumption. Digi suggests
that the model be updated based on actual traffic data and forecasts.
Maxis is of the view that the spectrum assignment for the notional operator represent the
best-case scenario in the Malaysian market i.e. the current maximum spectrum
assignment. This may not be accurate as it has the potential to understate the number of
sites required, particularly for GSM. Maxis added that with refarming, spectrum for the
900MHz was reduced to 2x10 MHz. It is also not necessarily true that an efficient operator
will pay the full market price for 2x10 MHz but settle for 2x5 MHz, more so when 1800MHz
and 2100MHz are fully market valued. Maxis views that the LRIC should not model a
legacy situation. Maxis proposes for the MCMC to use the current minimum spectrum
assignments for the notional operator i.e. 2x5 MHz for 900MHz, 2x15 MHz for 1800MHz
and 2100MHz and 2x10 MHz for 2600MHz.
TM views that spectrum and coverage network costs should be excluded from the mobile
model if the increment for LRIC+ is the service being costed. If the MCMC had defined
the increment to be the whole network, then spectrum cost is included. However, if the
increment is defined to be the whole network, there is little to distinguish the LRIC+ model
from the FAC model.
YTL views that the coverage assumptions are not appropriate for voice and SMS services
provided using only higher frequencies via 4G. The analysis only takes into account the
service providers that have a combination of 900MHz, 1800MHz, 2.3GHz and 2.6GHz
spectrum, but omits service providers that provide voice and SMS services only on higher
spectrum namely 2.3GHz and 2.6GHz (VoLTE). The higher spectrum bands require 14
times more base stations compared to lower spectrum bands and thus have significantly
higher cost.
Discussion
The MCMC accepts that the costs of spectrum included in the model did not fully reflect
those of the operators, particularly in the area of 2100MHz spectrum. This has now been
corrected. With respect to the prospective allocation of 700MHz spectrum, some operators
Page 100
suggest that the costs of this should also be included. However, it is difficult to do this
without knowing the take-up of spectrum in this band. Furthermore, access to this
spectrum band would, other things being equal, be expected to result in substantial
reductions in the number of base stations required to achieve a given level of coverage as
against the higher-frequency bands currently available. However, the extent to which
such savings could be realised in practice will depend on the extent to which coverage has
already been achieved by the time the spectrum becomes available. The model inputs
currently assume that the notional operator achieves long-term plateau in rural coverage
levels of 40% for 2G and 3G, by 2019, but only 25% for 4G by that time. There would
appear to be significant scope for savings from switching to 700MHz from that point
onwards. The model has therefore been amended to show 4G rural coverage rising at a
somewhat faster rate from 2019 onwards, reaching 50% rural coverage by 2021, instead
of 40%.
Maxis suggested that the model should reflect a forward-looking position in terms of the
allocation of spectrum and proposed 2x5 MHz for 900MHz for 2G, instead of 2x10MHz to
reflect spectrum refarming. Maxis’s proposal appears to be inconsistent as it suggests not
to reflect this in the use of the refarmed spectrum for 4G, as is the case in the model.
Equally, the MCMC does not accept the argument that the total assignment of 1800MHz
spectrum should be reduced to 2x15 MHz and has continued to assume a total allocation
of 2x20 MHz, with 5MHz of that assigned to 2G.
The MCMC does not agree with YTL’s suggestion that the notional operator should be based
on the spectrum assignment of a 4G-only operator, as this would not be consistent with
the notional operator concept, reflecting an operator with 25% share of the market.
Furthermore, 4G was primarily introduced to enable faster mobile data speeds, with voice
as something of an afterthought. It would therefore be somewhat perverse to regulate a
multi-technology industry on this basis alone.
Digi’s argument that the distribution of traffic between 2G, 3G and 4G should not be as
currently calculated by the model, based on population coverage and other factors, but
based on inputs from operators. The problem with this is that only one operator was
initially able to provide a split of voice traffic in this way. Maxis, however, provided a
proposed split of voice traffic in response to the Public Inquiry. Maxis also argued that
VoLTE is at a very early stage in its introduction in Malaysia and the effect is not expected
to be more than negligible in the near term. The traffic proportions for voice are shown
in the three graphs below, which show the proportions in the model, together with the
proposals from an operator and Maxis.
Page 101
Figure 2: Distribution of Voice Traffic by 2G RAN
It may be seen that the operator and Maxis are in reasonable agreement that the
proportion of voice traffic carried on 2G will decline from a level that is fairly close to that
in the model now and at a faster rate than the model calculations suggest. On the other
hand, their views are far apart when it comes to the respective roles of 3G and 4G. This
appears to stem from a far higher starting proportion of 3G traffic in the operator’s network
than in Maxis’s network. A second difference is, as suggested by their respective
comments, a view on the part of Maxis that 4G voice will take off 2-3 years later than the
operator’s expectation.
30
20
10
0
2016 2017 2018 2019 2020 2021 2022
Page 102
Figure 4: Distribution of Voice Traffic by 4G RAN
30
20
10
0
2016 2017 2018 2019 2020 2021 2022
The MCMC is not persuaded that the Malaysian operators are generally likely to roll-out
4G as rapidly as the operator suggests.
However, the MCMC is persuaded by Digi’s argument that network selection by devices for
voice is likely to be influenced by preferences for faster data networks where these are
available. The MCMC has therefore adjusted the criteria used in the model for distributing
traffic by technology to account for this factor. The adjusted percentages now applied in
the model are also shown in Figures 2 to 4 above.
With regard to the proportion of data traffic on each RAN, Digi is incorrect to suggest that
the model allocates this purely with reference to population coverage. The MCMC would
like to clarify that additional factors relating to the relative speed of the respective RANs
for data and limitations in 2G network capacity are taken into account, resulting in profiles
that are reasonably similar to those proposed by Digi.
TM’s suggestion that spectrum and coverage costs should be excluded from the model is
only applicable with pure LRIC methodologies, which the MCMC has discounted as being
unsuitable for use in Malaysia.
The MCMC proposes to apply the following spectrum assignments in the model:
Page 103
Table 27: Spectrum Assignments applied in the Model
Question 22:
Do you have any comments on the design assumptions for the mobile model?
Submissions received
Celcom notes that the PI Paper did not provide full discussion of all the design assumptions.
A major area of concern for Celcom is the dimensioning and allocation of costs between
voice and data for 2G and 3G. This is because 2G and 3G voice are circuit switched while
data is packet switched. Celcom highlighted several concerns with regards to the model
as follows:
(a) The model does not allow operators to recover costs of stranded assets where, for
example, 2G traffic reduces from one year to another and so less capacity is
required.
(b) The model takes an optimistic view that data is 8.5 times more efficient to carry
than voice. Celcom takes a more conservative view that on EDGE, data is three
times more efficient. The model assumes that one-time slot can carry 34 Kbps of
data whereas in Celcom’s experience 25 Kbps is more reasonable. On a 3G carrier,
in principle, it can carry up to 128 voice calls per carrier, but the actual number of
calls which can be delivered is around 54 per carrier, depending on the number of
sectors used and the use of adjacent sectors on nearby sites. HSPA+ can carry 4
times more data than voice, and this was not taken into consideration in the model.
As a result, the 3G network is incorrectly dimensioned and the proportion of cost
attributed to voice is too low.
(c) The model does not take into account the efficiency difference between voice and
data traffic on 2G and 3G and the proportion of cost attributable to these services.
The model understates the equipment required to deliver traffic in the 3G network
Page 104
as the model is data focused. Since voice is carried less efficiently than data on
both 2G and 3G, more cost should be allocated to voice.
(d) The model calculates voice channels per carrier but does not use the calculation
elsewhere. It appears that the MCMC understands the constraints on the number
of voice calls but is unwilling to address this in the model.
(e) There is inconsistency in the way the network is dimensioned namely the calculation
on “CarrierbitRate3G” which calculates the bit rate per carrier in UMTS, however,
what is being modelled is a mix of 3G technologies, hence implying a higher bit
rate per carrier.
(f) Celcom notes that there is error in the calculation of 3G capacity as the model used
total number of 3G sites instead of number of upgraded 3G cell sites.
(g) The mobile network provides separate downlink and uplink capacity for data just
as it does for voice. Since the proportion of downstream data is higher than
upstream, the network should be dimensioned for the downstream data
requirement. This will then increase the percentage of costs which should be
allocated to voice.
(h) Incorrect routing factors as the model assumes all calls and all data use 2G, 3G
and 4G specific network components however the voice and data mix on 2G, 3G
and 4G are very different.
Digi agrees with the MCMC’s approach of averaging cost data provided by the mobile
operators to derive a common cost mark-up which is adopted in the model.
Maxis is agreeable to the percentage of traffic in the busy hour for data but proposes a
slightly higher percentage of traffic in the busy hour for voice i.e. 10%, as it reflects more
closely the pattern of subscribers’ traffic in Malaysia. Maxis also proposes a slightly higher
voice bit rate for 2G RAN to 12.2 Kbps in order to meet the requirement on Commission
Determination on the Mandatory Standards for Quality of Service (Public Cellular Service),
Determination No. 1 of 2015 with regards to call setup success rate, dropped call rate,
etc. Maxis fully supports the inclusion of spectrum costs and licence fees in the model as
they form part of the total costs of operating a network. Maxis proposed some revision to
the spectrum cost and highlighted that the 2100MHz spectrum cost recently issued by the
MCMC in October 2017 should be included in the model. In addition to the spectrum and
licence costs, Maxis views that numbering fees should also be included in the mobile cost
model for an accurate presentation of the total costs as an interconnect service cannot
exist without a licence and numbering block to deliver calls. Maxis also views that the
Page 105
EPMU mark-up of 25% is too low, and instead would like to propose the use of 36%, close
to the EU Models in which the common and fixed cost are approximately 40-60% of the
total costs.
TM is of the view that the average cost of the passive infrastructure in the mobile model
should reflect changes in geographic weighting. However, the cost in the model has been
clearly averaged with no transparency for this process. The mobile assumptions do show
a mix of site changes over the period of 2017 to 2020, with a greater proportion of rural
sites as coverage expands, therefore, the average cost of the passive infrastructure should
reflect this change in geographic weighting but this is ignored by the cost assumption in
the model.
webe expressed concern that 2G is still being costed in the model as this equipment could
potentially be fully-depreciated. Similar to fixed services, most service providers have
upgraded their network to NGN which should drive prices down further.
YTL highlighted that the design model does not take into account spectrum costs incurred
through the payment of apparatus assignment (AA) fees. For service providers with higher
frequencies, AA fees are substantial as more base stations are required for equivalent
coverage as the national operator.
Discussion
Celcom offers the opinion that where assets have been acquired, but are no longer needed,
for example because traffic has moved from 2G to 3G and 4G, operators should continue
to be allowed to recover costs for them via the termination rate. It further asserts that,
because the network dimensioning section of the model shows fewer 2G transponders are
required in later years, that the model, in effect, strands these assets. In the MCMC’s
view, it is an inevitable consequence of technological change and competition that some
assets will not be fully utilised for their full technical life and firms operating in an
effectively competitive market would not expect to be able to recover the full cost of
obsolete assets in the face of competition from firms with newer and more efficient assets.
It is partly for this reason that tilted annuities is used as the depreciation methodology, in
that it can reflect a profile where the economic value of assets reduces over time. It would
be unreasonable, therefore, for operators to expect to be able to recover the cost of
obsolete assets indefinitely from Access Pricing.
Nevertheless, the model does, in fact, allow for the notional operator to recover costs from
obsolete assets to some degree, in that the costing section of the model is configured so
that the quantity of assets of each type for costing purposes reduces only to the extent
that assets are retired at the end of their planned life, under conditions where the
Page 106
requirement for them is reducing. It might be argued that this introduces an element of
inefficiency to the notional operator that might not be sustainable under competitive
conditions, but in the MCMC’s view, it represents a reasonable compromise and is likely to
represent the actual practice of operators under conditions where technological transitions
are gradual.
webe makes the converse point that 2G assets should be excluded from the model,
because 2G is essentially an obsolete technology. The MCMC recognises that there is a
clear trend towards replacement of 2G and that this has been announced, or has already
taken place in several other countries in the region. However, the MCMC considers that
since 2G is still being used in Malaysia, it is appropriate to reflect a steady transition to
new technologies over time.
Whilst the MCMC would agree with Celcom that uplink and downlink traffic are handled
separately in the RANs to some degree, the operators, including Celcom, were unable to
provide a breakdown between the two. Furthermore, both uplink and downlink traffic
consume base station resources and so the MCMC does not agree that applying an
arbitrary scaling to the total to represent only notional downlink traffic is reasonable.
Celcom make a number of assertions and observations under this heading, in the context
of their overall view that the model allocates too much of the cost of RAN to data services
and not enough to voice services. These are examined in turn below.
This remark is somewhat puzzling in this context, given that switching does not
take place in the RAN, but in the core network, which is assumed for the notional
operator to be a modern common packet core network, using a soft switch.
However, it appears from the context that Celcom’s concern is actually that data
traffic is assumed to be handled more efficiently than voice, which Celcom agrees
to be the case, although it takes issue with the extent to which this is so. In the
case of 2G, Celcom propose an efficiency differential of 3x between data and voice
on a network equipped with EDGE technology, in preference to the differential for
8.5x, which it says applies in the model.
The problem with this is that, in order to put the revised assumption into effect, it
is necessary to assume either that EDGE networks offer a lower data rate than they
do, or that voice traffic can be carried more efficiently. It is not clear exactly which
Page 107
assumption Celcom would like to change in order to achieve this, or whether it
proposes some kind of arbitrary adjustment factor to achieve the preferred ratio.
However, one parameter Celcom mentioned is the TRX non-homogeneity factor,
which is applied to reflect the likelihood that traffic demands are unevenly
distributed, leading to a necessity to provide more capacity in total than an
averaged approach, as applied in the model, would lead one to expect. Whilst it is
difficult to gauge how large this effect might be, a figure of 0.7 may be considered
conservative. Ofcom, for example, used a figure of 0.5 in their 2011 model.
Celcom identifies what it describes as “a number of concerns with the way in which
the 3G radio network is dimensioned”.
In fact, given the predominance of data in the mix of traffic in Malaysia, even in
the early years of the modelled period, makes it unlikely that voice would be the
primary driver for 3G capacity and so this is unlikely to have any effect. For
example, in the base year, using the figure of 64 channels per carrier suggested by
Celcom and applying an adjustment based on the erlang table, the capacity of the
3G coverage network (1 sector and 1 carrier per site) is over 350 thousand erlangs,
but the demand is under 100 thousand erlangs. On the other hand, the MCMC
accepts that it is inappropriate to combine the voice and data traffic to drive
dimensioning based on data capacity and that capacity requirements based on
bandwidth should be based on data traffic only. The model has been amended to
reflect this.
Page 108
Celcom makes a further assertion that the model incorrectly applies routing factors with
the result that 2G, 3G and 4G specific equipment costs are divided by the whole traffic
and not by the specific traffic streams that use them. Celcom gives the example of the
respective cell site equipment. As far as the MCMC can see, this is simply incorrect. The
denominator used for the relevant calculations is the RAN-specific figures summarised on
the Equipment Outputs sheet.
Celcom identified calculation errors affecting busy hour calculations in the network
summary sheet. These have been corrected. In relation to the bit rate for voice on 2G,
the figure of 10 Kbps has been changed to 13.2 Kbps, which is the correct figure (not 14
Kbps, as Celcom suggests) and in preference to 12.2 Kbps, as Maxis suggests. The MCMC
has also corrected errors pointed out by Celcom affecting the weighting of voice traffic
expressed in Mbps between 2G, 3G and 4G, the calculation of voice Mbps in the core and
the routing factors for MGW and MSS for data services, which should have been set to
zero.
With regard to Maxis’s suggestion to include numbering fees, it is not usual practice to
include these in network cost models, because the cost of numbers is specific to
subscribers, rather in the same way that fixed access lines are. It is not appropriate,
therefore to recover these costs from origination and termination services.
Maxis also suggests that the EPMU mark-up values should be increased to the levels used
by regulators in the EU, where mark-ups are applied, which of course they are not when
pure LRIC, the methodology mandated for use in the EU, is used. The mark-ups in use in
the model were calculated on the basis of the top-down cost figures submitted by the
operators. Although the levels calculated for different operators varied, the MCMC does
not consider that it would be appropriate to assume that operators were less efficient than
the data they have supplied suggests, purely on the basis of a benchmark.
The MCMC is also not persuaded to change the assumption in the model about the
percentage of voice traffic occurring in the busy hour, because the existing figure is based
on inputs provided by the operators generally and not just by Maxis.
The MCMC does not accept TM’s argument that the model fails to take account of the
changing proportion of services using passive infrastructure over time. In the absence of
data relating services to urban versus suburban and rural sites directly, the model looks
at the proportions of sites that carry different combinations of 2G, 3G and 4G traffic and
weights the allocations of passive infrastructure accordingly. Since the mix of traffic varies
considerably among these technologies, this provides a reasonable proxy.
Page 109
With regard to YTL’s comments about AA fees, the MCMC has adjusted the cost
assumptions in the model to cover all spectrum fees.
The MCMC has corrected a number of calculation errors in the model that were pointed
out by the respondents. It is not, however, persuaded that the design assumptions or
cost allocation methodologies used in the model are incorrect, or that they require change.
Question 23:
Do you have any comments on the service costs calculated by the mobile model?
Submissions received
Celcom believes that the mobile industry in Malaysia is highly competitive, and hence
regulation should only be applied to mobile termination rates and not to other services
such as origination, SMS, MMS and data. Maxis has no objection to the mobile cost model
producing service cost for a range of mobile services and fully supports the MCMC’s
approach to regulate prices for Mobile Origination and Termination services only. The
other services should be left to commercial negotiations as these services are no longer
significant in view of application-based services gaining popularity.
Celcom believes that the model is not able to provide reliable results. If the model is
corrected as follows: WACC is revised to 9.97%; 4G bandwidth reduced from 62 Kbps to
13.65 Kbps; spectrum charges included for 700MHz and additional 2100MHz included; and
issues on network dimensioning, equipment numbers, routing factors, etc. are addressed;
then, this will increase the mobile termination rates from 1.58 sen per minute to more
than 4 sen per minute.
Maxis agrees to the use of tilted annuities as a means to annualise the capital costs for
voice termination as it is a mature service in Malaysia and this approach is a closer
approximation to reality for mobile services.
TM views that the estimated service cost is overstated as a result of inefficient use of 2G
technology in the model, understatement of mobile data traffic and extremely high cost of
core network buildings. Core network accounts for half of the total cost, of which 83%
are from core network buildings. Comparing the model’s input costs for core network
building against the financial reports of the operators suggest that the value used may
include buildings other than those just for the core network i.e. buildings associated with
retail and other operations. Similar to a selection of fixed assets with long lifetimes, for
Page 110
consistency the MCMC should also mark-down long lived assets in the mobile model such
as core network buildings, core and backhaul fibre, power supply, some site-related cost,
IBCA cost, RAN equipment, microwave equipment, etc. If these assets with lives of at
least 20 years were marked down, there is a significant reduction in the model results.
The 2G technology has been running for 22 years, thus the notional operator must have
been operating 2G technology for 22 years. Therefore, some long lived assets that are
still in use must have be written off, such assets could include passive infrastructure which
could even be shared with 3G and 4G technology.
webe has no doubt that the demand for data service will see continuous growth during the
modelled period but the model assumptions are unrealistically low in respect to mobile
data usage per subscriber. webe is also of the view that the voice service cost is over-
estimated.
YTL notes that the service costs in the mobile model are based on the notional operator
with a combination of both low and high frequency spectrum bands. As voice services
leverage mainly on the lower spectrum bands i.e. 900MHz, the cost of coverage is much
lower as compared to the operator that uses only higher frequencies i.e. 2.3GHz and
2.6GHz. Hence, the mobile model does not fully capture the cost of operators that use
only high frequency spectrum for voice services.
An Operator pointed out that the multi-RAN handsets have a preference for the highest
speed networks for data traffic. The technology choice for voice traffic is determined by
the network that the handset connects to for data purposes. Both are biased towards
higher speed technology. The operator also highlighted that the 2G network is still
relevant as it remains the primary fall back for voice services in 3G/4G networks. In
addition, where dual SIM handsets are used, the second SIM is normally configured for 2G
network usage. Thus, the operator’s challenge is in converting 25% of its subscriber base
to data user. The operator proposed the technology split for both voice and data from
2016 to 2020 in which by 2020 most of the traffic for voice and data will be on the 4G
network. The operator estimates that by adjusting the technology split, the mobile model
should reflect a 27% increase in unit cost.
Discussion
The MCMC has reviewed the calculation of core network costs in the model in light of TM’s
comments about the cost of core network buildings. It appears likely that operators have
given a total cost for this item, rather than a unit cost, and so, the model may have
incorrectly multiplied this by the number of core network sites required for the network.
An adjustment has been made to correct this.
Page 111
The points made by respondents about data service growth, 4G-only operators and the
split of traffic between 2G, 3G and 4G have already been addressed in the earlier
questions.
The MCMC has made an adjustment to the model to correct the over-estimation of core
network building costs.
Question 24:
Do you have any comments on the 3G/4G only operator, local/national call rates and
submarine cable issues?
Submissions received
Altel and webe are of the view that similar to the Fixed Network Origination and
Termination Service, the mobile model should not make distinction in prices by the use of
submarine cable as there is sharing of capacity with data traffic. webe agrees that it is
difficult to distinguish the cost of terminating calls to nearby versus distanced called party
and that commercial pricing is evolving towards a flat structure. With the emergence of
NGN, it has become more borderless. webe is surprised that the position taken for
submarine cable differs between fixed and mobile and opines that a similar position should
be adopted for both services as the submarine cable that carries mobile and fixed services
are the same. In the situation where mobile traffic outweighs the fixed traffic, the cost
for mobile calls using the submarine cable should not be treated differently to the cost for
fixed calls.
Maxis is agreeable to retain the current rates using the submarine cable to allow the Access
Provider to recover its cost of providing the service with submarine cable to the Access
Seeker who chooses to establish only one POI with the Access Provider. Maxis believes
that there is a significant cost difference between local and national origination/termination
with submarine cable. However, U Mobile views that MTR with submarine cable should
decrease in 2018 compared to the last regulated rate of 15.73 sen, with the launch of
SKR1M that provides for 4Tbps via 3,800 km of cable which links 6 locations in Peninsular
Malaysia and Sabah and Sarawak. This should translate to lower wholesale prices. Access
Seekers who utilize this service should not have to bear the inefficiencies in this area, due
to uncertainties or irregularities in the cost provided by operators of submarine cables.
Page 112
TIME is of the view that the MCMC needs to standardize the prices for mobile origination
and termination to be coherent with the pricing structure for fixed origination and
termination. For mobile origination and termination services, submarine cable connectivity
is included whereas it is absent for fixed origination and termination services. TIME
recommends that prices of origination and termination with submarine cable be
incorporated for the fixed network. Additionally, the capacity for submarine cable
connectivity between Peninsular Malaysia and Sabah and Sarawak is now expanded with
the deployment of SKR1M. With the incorporation of submarine cable cost in fixed
origination and termination services, healthier competition is being introduced in the
mobile and fixed markets.
Digi and Maxis are agreeable to combine local and national rates into a single rate within
Peninsular Malaysia and within Sabah/Sarawak as there is no significant cost difference
between these two categories.
Celcom views that since operators have a mix of 2G, 3G and 4G networks and customers
have access to 2G and 2G/3G handsets there is no rationale for modelling a 3G/4G only
operator.
Maxis views that 2G services will continue to be provided for the current regulatory period
and hence should not be excluded from the model. The planned switch-off date has yet
to be decided in Malaysia, and major operators have no immediate plans to switch-off their
2G networks as there is still a significant number of 2G and 3G handsets amongst end
users in rural areas, which still need services to be provided. Maxis agrees to the data
traffic by RAN, however, the voice traffic by RAN may not accurately represent the actual
roll-out of VoLTE services in the Malaysian market. VoLTE services in Malaysia is at the
very early stage and traffic is still small. According to Maxis, to date, only Digi and U
Mobile have soft-launched their VoLTE services in late 2016. Maxis provided a forecast of
the proportions of voice traffic by RAN with VoLTE-enabled in 2019 which indicates that
most of the traffic is still on the 2G and 3G network, with LTE picking up slowly and
reaching 49% by 2022.
TM stated that it could not find any reference within the PI Paper on 3G/4G only operators.
Nonetheless, the LRIC methodology requires the use of modern equivalent assets modelled
on an economically efficient operator. No new entrant would deploy 2G network as 3G
and 4G are more spectral efficient thus provide greater traffic capacity. If LRIC is the
approach, the notional modelled operator should not utilise 2G technology. The model
Page 113
suggests that 2G equipment is continuing to be installed as the model shows that there is
an increase in 2G microcells in 2017 and 2018 and decline thereafter and base station
controllers increase from 2017 to 2021 and remain constant thereafter. The model results
suggest an expansion of 2G network rather than transition to 3G and 4G networks. Thus,
the 2G network assets continue to incur replacement and operating cost for the lifetime of
the model whereas the MCMC’s statement that smartphone penetration reached 80% in
2016 suggests a deliberate routing of voice traffic over 2G by the notional operator rather
than a transition to 3G and 4G. TM views that 2G technology should be removed from the
mobile model so that both fixed and mobile models consistently use forward-looking
technology.
YTL expressed concern that the model does not capture the cost of 4G operators using
only higher frequencies to provide VoLTE.
Discussion
The MCMC has received significant additional information about the costs of submarine
cable transmission during the course of the Public Inquiry and this has now been reflected
in the model. The new data suggest a substantially lower cost than was previously
calculated, with the effect that calls using the submarine cable are just over one sen per
minute more expensive than those without it. In consequence, the MCMC is able to accept
the suggestion of Altel, webe and TIME that the treatment of the submarine cable should
be the same for mobile as for fixed and the rates averaged together.
The MCMC agrees with Maxis’s argument that 2G is likely to remain a feature of the mobile
telephony industry in Malaysia over the course of the regulatory period.
The MCMC has revised its cost estimates for calls using the submarine cable and its
proposed regulatory treatment of these services.
The MCMC continues to be of the view that 2G should be included in the model, because
it appears likely to remain a part of the mix of technologies in Malaysia over the regulatory
period.
Question 25:
Page 114
Submissions received
Maxis largely agrees with the sensitivity analysis done by the MCMC.
Celcom believes that sensitivity testing should be used to ensure that the model is working
correctly. However, sensitivity tests should not be considered in setting rates as the
model has many errors. Celcom notes that in the model, data usage per subscriber grew
from 0.65 GB per subscriber to 1.36 GB per subscriber between 2015 and 2016. Hence,
the comparison of data usage in Malaysia in 2016 with other countries in 2015 is likely to
be misleading.
TM suggests that the MCMC explore a sensitivity analysis for mobile data with a more
realistic assumption for data traffic in 2017. This will provide greatly improved insight into
the sensitivity of the mobile origination and termination results to the data traffic
assumptions. The “high growth” scenario clearly under-estimates the projected data
traffic. TM suggests that the MCMC consider to test the impact of using the same WACC
for fixed on mobile services as part of the sensitivity testing as there is empirical evidence
that the risk profile of mobile operators is similar to fixed operators. TM also suggested
that the MCMC use more comparable data to benchmark data usage from neighbouring
countries e.g. Singapore and Korea show substantial increase of 25% and 32%
respectively in 2017.
Discussion
The points about calculation errors and the usage of mobile data services have been
addressed in earlier questions. TM’s point about the WACC is relevant to the WACC section,
however, the MCMC is not convinced that sufficient evidence has been brought forward by
TM to support their contention that fixed and mobile operators face the same risks.
The MCMC does not consider it necessary to make any further changes to the sensitivity
analysis.
Question 26:
Do you have any comments on the proposed regulated prices?
Page 115
Submissions received
Altel, TM, U Mobile and webe concur with the MCMC on a single rate for local and national
calls as well as the convergence of mobile origination and termination rates over the period
to 2020. U Mobile views that the continued downward trend for voice traffic over the
regulatory period will result in lower unit costs.
Altel and webe added that similar to the Fixed Network Origination and Termination
Service, the mobile model should not separate the submarine cable cost as this capacity
is also shared with data traffic. In addition, Altel observes that the glide path approach is
not adopted for the Mobile Call Origination and Termination with submarine cable. In
consonance with the costing methodology adopted in the PI Paper, an increase in capacity
should effect a reduction of the unit costs. As such, Altel believes that over time, as the
costs of the network elements reduce, the regulated rates for Mobile Call Origination and
Termination with submarine cable should decline over the years as the network becomes
more efficient.
U Mobile proposes that the regulated prices for mobile origination and termination should
be more reflective of cost to ensure that there is no over-recovery or extension of
inefficiencies of the previous years into the regulatory period beginning 2018. The rates
in the mobile model indicate that voice termination for 2018 and 2019 should be 1.97 sen
and 1.74 sen respectively, before arriving at 1.58 sen in 2020. U Mobile proposes that if
a glide path is used, the price should be set at 2.20 sen in 2018 and 2019 and suggest
that these rates be used retrospectively for 2016 and 2017 which is a practical downward
step from 3.65 sen in 2015 to 1.58 sen in 2020.
Celcom views that the model is unable to produce reliable results. The figures shown
understate the costs of an efficient operator. Once the identified issues in the model are
corrected, the mobile termination rates should increase from 1.58 sen to more than 4 sen.
Maxis agrees with the MCMC’s views to set regulated prices for voice interconnection for
origination and termination services. Maxis supports LRIC+ costing methodology and the
glide path approach and pointed out that even with LRIC+ methodology, the mobile
termination rate in Malaysia is amongst the lowest. Hence, Maxis requests the MCMC to
be firm on the use of LRIC+ instead of pure LRIC as proposed by certain operators.
TM supports the MCMC’s proposal not to regulate SMS and MMS services as technology
will drive service innovation which is already resulting in a decline in SMS traffic. In such
a dynamic environment, regulating prices may result in market distortions. TM also notes
that the model estimated a unit cost for mobile calls with submarine cable that is
significantly higher than the previous regulated rate and suggests that the previous rate
Page 116
be used for the period of 2018 to 2020. TM believes that the previous regulated rate for
the submarine cable was derived from the fixed model used at that time. TM suggests
that rather than using outdated cost data, the MCMC could employ the same approach and
incorporate the per minute submarine cable costs from its fixed model, which as noted by
the MCMC is insignificant for fixed voice origination and termination. Therefore, the mobile
origination and termination rates should also not differentiate between calls with and
without submarine cable. The MCMC’s model appears to assume only voice traffic is
utilising the submarine cable, hence the entire cost is allocated to voice. It is unlikely that
the submarine cable does not carry any mobile data traffic. TM did a sensitivity of the
model by setting the submarine cable routing for data traffic to be the average of incoming
and outgoing mobile voice traffic i.e. to carry less than 1% of the data traffic, the outcome
was a relatively small difference between the rates with and without submarine cable and
significantly lower than 24.1 sen per minute. If the data traffic of the modelled operator
was increased, the differential results with and without submarine cable is even lower.
This will be further reduced if the submarine cable’s operating cost is reduced by 50% with
SKR1M. TM suggests that the MCMC review the approach on the submarine cable to set
a non-zero routing factor for data traffic or per minute cost derived from the fixed model
and reflect SKR1M’s discounted prices into the model.
YTL re-emphasizes that the proposed regulated prices do not capture the cost of 4G only
operator that provides mobile voice services (VoLTE) using only higher frequencies.
An operator proposes to adopt a cautious approach to the proposed glide path for mobile
interconnection rates as a reduction of 17% in 2018, 23% in 2019 and 32% in 2020 will
have a significant impact on the mobile operator’s finances. This is compounded with the
intense market pressure whereby the mobile industry has been recording negative growth
in the previous years. The operator proposes a gradual transition to provide for greater
regulatory certainty for their future plans and investments i.e. a 17% reduction in 2018,
followed by 23% in 2020 and 32% in 2022. Alternatively, the MCMC is requested to
consider deferment of the new rates to 2019 to alleviate significant financial disruptions.
Discussion
The MCMC notes that several operators make the point that the submarine cable system
is likely to be used for mobile data traffic as well as for voice. This aspect is related to the
earlier discussion on submarine costs where the MCMC has accepted the argument to treat
the submarine cable in a similar way for fixed and mobile voice calls.
The MCMC appreciates the point made by U Mobile that the continuation of the previous
rates may have led to some over-recovery of costs. However, as a matter of principle,
the MCMC does not consider it best practice to implement such changes retrospectively,
Page 117
as part of the reason for setting rates over a period of time is to provide an element of
certainty for operators and investors. Such uncertainty would be significantly undermined
by the application of changes retrospectively. Furthermore, although a part of the adverse
impact of over-recovery is felt by the shareholders of operators who are net out-payers,
the primary issue of concern is the potential for consequent detriment to consumers and
the MCMC doubts whether it would be practicable for this detriment to be set right
retrospectively for consumers.
The MCMC notes the general view in favour of glide paths. The suggestion of using
geometric, rather than linear interpolation, as proposed by one operator would tend to
lead to a faster initial rate of change, with slower change later in the period. This would
have the advantage of reducing the net over-recovery during the glide path period, but
must be set against the relatively rapid rate of change proposed.
The MCMC is not persuaded to change its view expressed in the PI Paper on the glide path,
nor to apply the mobile origination and termination rates retrospectively.
The MCMC continues to view that mobile origination and termination should be set using
the inputs for a notional operator with 25% market share. Prices at this level provide the
right incentives for market competition, investment in new technologies and service
innovation that lead to greater usage of new technologies. Likewise, the MCMC is not
persuaded that there is a need to regulate prices for SMS, MMS or video services.
The MCMC sets the final prices for Mobile Network Origination Service and Mobile Network
Termination Service as follows.
Page 118
8. Infrastructure Sharing
8.1. Overview
The MCMC developed a bottom-up infrastructure sharing model using tilted annuities as
the annualisation method to calculate the costs and prices for Infrastructure Sharing on
the Access List. The cost model with all commercially confidential data removed was made
available on request to interested licensees during the Public Inquiry period.
Part F of the PI Paper concerned Infrastructure Sharing, in particular towers and in-building
common antenna systems. Section 19 of the PI Paper described the infrastructure sharing
model, including changes made after the initial model viewing, and sought comment on
its completeness.
The MCMC proposed to set regulated prices for lattice type towers of 60m, 75m and 90m
and provided “reference” prices for In-Building Common Antenna (IBCA) systems which
would be taken into account in any requested interventions.
Question 27:
Do you have any comments on the approach to the modelling of tower costs?
Submissions received
Digi supports the proposal to regulate tower prices to avoid excessive wholesale price
offering.
Celcom expressed concern that the model does not show all types of towers used. Celcom
currently uses towers ranging from 150-400 feet. Celcom views that the model may not
have captured a full range of cost drivers as a 24m lattice costs RM82,686 whereas an
18m lattice costs RM107,395. It is difficult to justify lower prices for higher towers.
Celcom is also concerned with some underlying assumptions used as it appears unlikely
that power supply costs will increase by 10% per year.
edotco has serious concerns on the imposition of access pricing regulation to tower
companies and the MCMC should consider whether ex-ante price regulation is prudent.
According to edotco, regulators globally have refrained from regulating in this area except
on an ex-post basis given the complexity and bespoke nature of towers and infrastructure
solutions. edotco views that there is no compelling case for price regulation of
Page 119
Infrastructure Sharing and tower companies so ex-post access pricing regulation remains
the optimal approach for Malaysia.
(a) The proposed access pricing regime would actively discourage investment and
infrastructure sharing (tower companies) in Malaysia;
(b) The proposed force-fitting of single determined access prices on bespoke
infrastructure solutions is unworkable administratively, divorced from commercial
realities and undermines competition;
(c) The proposed access pricing regime will create market distortions by discriminating
against independent tower companies versus MNOs because the proposed regime
would not apply where tower access is swapped between and among the MNOs;
and
(d) Access pricing regulation of Infrastructure Sharing should only apply to dominant
licensees.
From edotco’s perspective, imposing regulated prices on towers even on a small subset of
tower types, may negatively impact Malaysia’s move to micro and picocells and the future
deployment of these next-generation 5G services.
Similarly, TM suggests that the MCMC revise its approach for sampling CAPEX costs for
the various tower configurations. Due to the large number of tower configurations and
some configurations being relatively uncommon, the averaging process used to derive the
model inputs may result in cost estimates that are unrepresentative of Malaysian towers.
Rejecting potential outliers could result in all of the company’s towers being ignored, if
that tower company only offered a single size by type combination that is not used by
other companies. TM highlighted that no information was provided on the number of
towers either included or omitted from the analysis, hence TM cannot ascertain if the data
sample would truly represent Malaysian towers as within the tower model, various cost
elements have been averaged. TM also noted some inconsistencies in regard to the
relationship between tower size and CAPEX as a 20m monopole is less than 60% of the
CAPEX for a 6m or a 12m monopole. The 6m, 12m and 30m monopoles have virtually the
same CAPEX. TM expects that the CAPEX and height to have a straight line relationship,
although there could be some economies of scale for certain monopole sizes. The average
CAPEX for monopoles suggests that the majority of monopoles are at least 36m, which
are quite tall and could be the outcome of a non-representative or poorly weighted sample.
PPIT views that Infrastructure Sharing should be left to commercial negotiation as there
has not been any market failure to constitute a requirement for mandated prices. Although
4 licensees have been declared dominant in certain states, there has not been any action
Page 120
from them that are detrimental to consumers. Tower sharing markets constitute different
variations and variables, and having reference prices will lead to detrimental effects on
revenue and the need to revise the business models and financial commitments with banks
and financial institutions.
PPIT highlighted that the basis of setting prices is similar between dominant and non-
dominant SBCs. Telcos have entered into 15-20 years’ agreements since 2005 which
suggest that the agreed basis and terms indicate that the licensees have not exercised
their market power, if any, to the detriment of the consumers and that the agreements
entered into are reasonable. If prices are mandated, not only SBCs will be impacted but
also independent tower companies. The modelling of the tower cost and the proposed
prices are averaging too many variations which include location, structure types, design
and height. The model also excluded substantial costs e.g. cost elements of the Telcos
who are also major providers; civil and preliminary works involved for telecommunications
tower sites and the variation orders that may be involved, especially in rural areas; and
other ancillary cost such as new permit renewal fees which are outside the control of tower
companies. The MCMC needs to be minded that it has accepted much higher tower costs
in awarding USP projects compared to that in the model and these are 3-legged light duty
telecommunication towers with lower specification as compared to heavy duty towers.
edotco and PPIT highlighted that the model may have either omitted certain cost or the
values used are too low. In addition, the modelling of the tower costs and the proposed
prices are averaging too many variations. Effective price setting should take these
variances into consideration. According to edotco, the proposed total annualised tower
site costs is 40% below their cost modelling for 3 or more tower tenants depending on the
location. It is also important that the MASP provide certainty post 2020 in the event that
the MCMC does not determine new prices before the MSAP expires. This is because unlike
call termination in which rates continue to fall due to lower equipment cost, tower
company’s cost which are directly related to tower material costs (CAPEX and OPEX) and
other cost such as permit fees and rentals are rising faster than the rate of inflation.
edotco and PPIT highlighted several concerns with regard to the model as follows: -
USP cost and subsidies have been unfairly treated in the model. If revenues are offset in
the operating cost, contributions to USP fund should also be accounted for in the model.
It is detrimental to reduce costs in this manner as they have a material impact. edotco
shared that its USP contributions are 10 times the quantum of the payments it received.
PPIT highlighted that not all SBCs receive the USP award. Phase 3 was only obtained by
2 SBCs and Phase 2 to less than 5 SBCs. The others are non SBCs recipients, some of
whom do not own towers. All SBCs contribute 6% of gross revenue towards USP. It is
Page 121
also unfair to assume the subsidy is applicable across all tower companies, without
consideration of contributions, in the model.
In terms of per tenant pricing, edotco pointed out that not all tenants are equal. Tenants
who require more space or use of resources as compared to the standard configuration
would be charged additional fees. In addition, increased rental is based on the number of
operators who are tenants on a particular site during the duration of tenancy, from a
minimum of RM500 per additional tenant per month to an average of RM2,000 per tenant
per month. The Landlord is aware of the number of tenants through site access requests.
Such cost is not included in the model to reflect the commercial realities.
PPIT highlighted that the apportioning of total costs per site to the sharing tenants remains
ambiguous. If the total cost is divided by the number of tenants, this will create
inefficiency in the industry as SBCs would not be incentivised to win new work and promote
Infrastructure Sharing. Additionally, there will be incremental OPEX cost from landlords,
USP contributions and the cost cannot be fully recovered under the equal sharing
mechanism. PPIT shared the current sharing mechanism fees imposed to licensees and
suggest to maintain the same sharing mechanism to promote efficiency.
Geographical location
edotco views that some form of geographic de-averaging of CAPEX and OPEX is required
with respect to East and West Malaysia as prices to design, build and commission towers
in Sabah and Sarawak are 30-40% higher. The OPEX for 45m to 76m towers vary by 15-
20% between East and West Malaysia. If data is inadequate, the MCMC should undertake
a detailed study to verify the cost and address the issue before regulating prices.
PPIT adds that the geographical location and specific configuration of towers are tailored
to the expansion needs of Telcos, however, the proposed model relies on a standard
configuration. Hence, the usage of the model for arbitration purposes without any
reflection to circumstances that may influence construction and installation costs is
inappropriate. PPIT pointed out that based on the data request template, the CAPEX cost
for Sabah and Sarawak is marked 200-300% higher than Peninsular Malaysia, whereas
some respondents in the PI Paper have indicated tower cost to vary by about 15-20%
between Peninsular Malaysia and Sabah and Sarawak.
Page 122
OPEX and CAPEX cost
(a) edotco highlighted that the property rental cost of RM18,598 for towers is too low,
as on average the property rental for towers is RM3,000 per month which translates
to RM36,000 per year. In addition, the one stop agency fee and local authority
permit to tower companies have increased almost 3 times in some states.
(b) edotco views that there are a number of missing OPEX cost which include
strengthening and replacement of batteries, utilities (water, sewerage), periodic
inspection for operations and health and safety, periodic maintenance of rusty
bolts, repainting, service for part failure and faults, spare parts, generator testing
and maintenance, general and administrative cost which include planning,
accounting, legal and human resource, insurance and one stop agency fee and local
authority permit fees.
(c) edotco also pointed out that fees related to one-stop-agency and local authority
permits in Terengganu have increased from RM4,000 to RM5,000 to RM15,000 per
year and this trend may be followed by other states as well.
(d) edotco views that discounts provided for longer term contracts which are subjected
to commercial negotiations should not be prohibited by the MSAP and subject to a
short phase-out period.
(e) PPIT highlighted that the proposed model averaged OPEX per tower to establish
common system charges. However, SBCs have different levels of OPEX and this
varies greatly between states due to the difference in tower configurations, terrain
and charges by local authorities in the individual states. Effective price setting
should take this variance into consideration. PPIT also highlighted that the average
cost of the 10 SBCs and Asiaspace were higher than that produced in the model for
a sample of 60 m lattice. They would have expected a narrower gap with the
inclusion of Stealth Solutions and edotco.
(f) PPIT highlighted that the CAPEX cost in the model is lower than the weighted
average and PPIT suspects that civil and construction cost have been wholly omitted
from the model. The basic site comprises of both these costs. The CAPEX cost will
further increase with a variation order and this has not been accounted into the
model. PPIT provided additional data for civil and construction cost for outdoor and
rooftop for 60m, 75m and 90m.
Maxis agrees to the use of a bottom-up costing methodology but is of the view that the
combined/blended approach of the historical asset costs and current asset cost should
Page 123
apply in the model. This is to reflect the actual cost and to avoid over-recovery by tower
operators as most of the existing towers have been constructed since many years ago.
Maxis views that there is a need to test sensitivity on the impact on volume as tower
companies vary significantly in size and scale, with most being state-based while some
are wider in scale across states. The proposed regulated prices are not very different from
the existing prices offered except that SBCs give discounts over a long tenure. Maxis
recommends that the MCMC adopt the approach of discounting as practised currently by
the SBC tower operators. Maxis suggests that the MCMC review the regulatory price for
other types of tower structure that are significantly used by Access Seekers namely
monopole, monopole tree, lamp pole and BTS Hotel solutions to avoid exorbitant charges.
Tower operators are now more focused on 15m, 18m, and 24m lamp poles and 30m and
45m monopoles. The MCMC should regulate prices for these tower types and not only
lattice.
Discussion
Operators’ comments span a considerable range of views, from those, such as Digi and
Maxis, who propose that price regulation is needed to overcome excessive wholesale
pricing, to PPIT, who maintain, as the MCMC understands it, that the willingness of
operators to enter into long-term contracts provides evidence that market power has not
been exercised to the detriment of consumers. Unfortunately, the MCMC is unable to
agree with this line of reasoning, given that, in the hypothetical case where the tower
company could exercise such power, its customers (operator) might have little choice but
to enter into such agreements. On the other hand, the MCMC does recognise that
operators have the option to build, or share, towers in many cases, though there are likely
to be exceptions to this in some places. Perhaps a stronger argument is that advanced by
PPIT in favour of differentiating the degree of price control between dominant and non-
dominant operators, although the MCMC does not accept that this should necessarily be
reflected in a cost model. PPIT further asserts that those operators who have been
declared dominant have not been guilty of actions that are detrimental to consumers.
Whilst the MCMC would agree that no such specific action has been brought to light in the
context of this Public Inquiry, the possibility of detriment to consumers also potentially
arises from more general exercise of market power, for example, in charging excessive
prices.
Page 124
edotco and PPIT complain of costs potentially omitted from the model, or being assessed
at too low a level, whereas Maxis observe that the costs produced by the model are broadly
in line with prices offered in the market. The MCMC points out that the costs entered into
the model are based on the inputs provided by operators, who were encouraged to provide
a full set of costs and to add any category of cost that they felt had been omitted from the
data request.
The MCMC notes the comments from edotco and PPIT regarding the complexities
associated with calculating the costs of multi-tenanted sites, the practice of landlords
charging additional rent when new tenants are added and the danger of encouraging
inefficiencies. The PI Paper did not propose an equal cost sharing mechanism, and the
MCMC continues to be of the view that a costing exercise of this kind is unlikely to shed
very much light on which schemes for sharing tower costs amongst tenants are reasonable
and which are otherwise. The model contains an assumption of three tenants on average,
based on the data provided by operators. The MCMC recognises that adding further
tenants may result in an increase in leasing costs, or that sites with fewer than three may
have lower leasing costs. The access regime allows flexibility in negotiating price
structures around this, including also discounts for longer tenure, provided that these are
equitable and non-discriminatory.
On the other hand, subsidies do affect the level of costs faced by operators in a direct way.
Although this is done on an averaged basis in the model to give a view of the level of costs
faced by operators overall, it is open to the MCMC to make adjustments in any particular
case, for example by removing this item where the operator in question is not in receipt
of any subsidies, or increasing it appropriately where it does receive them.
The MCMC notes the general comments from edotco and PPIT and their arguments that
the imposition of regulated pricing would have a detrimental impact on the development
of the towers sector in the future, would distort the market and discourage investment,
especially for the deployment of next-generation 5G services. The MCMC also notes the
Page 125
statements that commercial negotiations have been used successfully within the market
and that there is no evidence of market failure, even in states where licensees have been
declared as dominant.
Having considered carefully the variety of comments from respondents, the MCMC is not
minded to pursue a more detailed general set of costs, as some respondents proposed.
On the other hand, the MCMC would be likely to require further detailed specific costing in
the event of a dispute resolution, or other interventions.
Question 28:
Do you have any comments on the sensitivities and outputs from the towers cost model?
Submissions received
TM highlighted that it is unable to comment due to the lack of transparency in the model
approach.
Celcom views that the sensitivity relating to annuity method is of little value since there is
no economic rationale for simple annuities to be considered. Celcom believes price tilt
assumptions and justification of the choices made in the base case and asset lifetimes
would be useful to have sensitivities on. Celcom also highlighted that the lifetimes used
in the tower model appear not to be in line with best practice, for example 15 years for
outdoor structures but only 10 years for a rooftop structure. Celcom notes the Ofcom
model assumes a cell site life of 18 years.
Digi views that tower companies should yield a lower WACC as compared to other services
as they have lower debt premiums and lower risk levels due to solid support from state
government. Digi suggests the use of benchmarks from within the same region rather
than operators out of the region namely the US and the EU. This is to ensure that the
actual scenario is not distorted.
Maxis views that the sensitivity analysis on towers is sufficient but recommends that the
MCMC undertakes costing and sensitivity analysis for other types of towers such as
monopole, monopole tree, lamp pole and BTS Hotel solutions. It is important to avoid
exorbitant charges imposed by third party access provider to the access seeker.
PPIT express concern over the accuracy of the assumptions used and the application of
inputs derived purely by estimates. In calculating averages, the MCMC has deliberately
excluded inputs that run into hundreds of thousands of Ringgit due to the assumption of
Page 126
double counting. The omission of civil and construction costs from the model may result
in under-representation of costs. PPIT also highlighted that the commercial circumstances
of individual SBCs may vary significantly from the average parameters specified and result
in inequitable outcome.
Discussion
Celcom raised the additional point that the asset lifetimes used in the model differ from
and are generally shorter than those used by Ofcom in their (mobile) model. The MCMC
does not accept that this difference represents any departure from best practice, which is,
in the MCMC’s view, to take account of local conditions when selecting such parameters.
The MCMC does not believe that sensitivities on price tilt or asset lifetime would shed very
much additional light, given that price tilt, for example, has a fairly straightforward
relationship with price.
Question 29:
Submissions received
TM highlighted that it is unable to comment due to the lack of transparency in the model
approach.
Celcom agrees to price regulation for towers but believes that further analysis is required
as the costs of some structure type decreases with the increase in structure height.
Digi supports ex-ante regulation on tower pricing, however seeks clarity on costing for
more prevalent tower heights i.e. <45 m (monopole) and 45 m (lattice). Regulating prices
for these tower heights should also be considered by the MCMC. Digi comments that if
mark-ups are applicable on top of the prices indicated, it will result in much higher prices
Page 127
than the current market rates. Allowing mark-ups within a commercial regime will not be
effective in ensuring that prices provided are cost-based and competitive.
edotco questions the value of publishing a set of reference prices which apply to less than
20% of tower configurations, which then can be varied by a governed set of complex rules.
edotco is strongly of the view that ex-post regulation is optimal except perhaps in relation
to dominant licensees. The approach of setting annualised total cost per site per year for
a standard allocation covering 3 sharing operators, is incorrect and overly simplistic as the
standard allocation of tenants to towers is less than 2 on towers that are owned and
managed by edotco Malaysia. edotco highlighted the key factors in determining prices to
tenants include CAPEX and OPEX cost, fees and charges, occupied space, power, length of
lease, additional service request, etc. edotco argues that cost for towers have an
increasing price index, hence, does not agree on the reduction of the passive elements of
the cost structure to 6% per annum. edotco also does not consider that the proposed
prices are reflective of cost of providing lattice towers especially for multiple tenants nor
are they de-averaged by location and structure height. The proposed regulated prices will
discourage investment and infrastructure sharing especially since the tilted annuities
model caters for 1 to 2 tenants per site. As such, when there are 3 to 4 tenants, edotco
loses total income under the proposed MCMC model. This loss is further topped up with
the increase in additional costs of having multiple tenancies at the site (increased rental,
tower strengthening, permit fee, civil work, etc.) which are not included in the tower
costing model.
Maxis notes that the proposed regulated prices are not very different from the existing
prices offered except that SBCs adopt discount structures. Maxis recommends that the
MCMC adopt the approach of discounting as practised currently by the SBCs i.e. 25% from
the eighth year and 35% from the eleventh year. Maxis suggests that the MCMC review
the regulatory price for other types of tower structure that are significantly used by Access
Seekers namely monopole, monopole tree, lamp pole and BTS Hotel solutions to avoid
exorbitant charges. The proposed regulatory prices should not be used by the Tower
Operators to increase their prices.
PPIT highlighted that the proposed regulated prices are increasing from 2018 to 2020,
however, the prices offered by SBCs to the Telcos are reducing in line with the agreed
pricing methodology. Also, the proposed prices are only valid for lattice type with heights
60m, 75m and 90m. PPIT is concerned that the proposed prices are neither a ceiling nor
a floor price but a reference price and licensees are allowed to negotiate subject to
approval by the MCMC. Implementing the regulated prices may result in major re-
organisation of the tower industry due to the many variations in types and heights of
towers. If the prices will have to be referred to the MCMC for decision, this could delay
the roll-out and be a deterrent to building more telecommunication towers. In addition,
Page 128
mandating prices for 3 years is highly risky to investors as compared to adopting long-
term prices.
PPIT emphasized that the SBCs collectively disagree with the proposed approach to
regulation due to the lack of clarity around the proposed regulated prices and over
emphasis on cost-driven “buy and build” decision and ignores the possibility of Telcos
selectively choosing to focus on towers that are located in lower cost urban areas. The
higher CAPEX and OPEX to construct and maintain towers in non-urban areas may
potentially lead to under-expansion of service to those areas. This is against the spirit of
the national agenda to increase coverage and last mile access which is highly dependent
on infrastructure services being able to penetrate to non-urban areas. The move to
regulate price will also dis-incentivise investment in new infrastructure. Setting threshold
values (30% above the model input cost for individual cost items and 20% overall increase
in annualised cost) above the reference cost is too restrictive and stringent as any
additional cost should be recovered provided it can be justified.
Sacofa highlighted that currently it does not provide batteries and generators. OPEX for
generators (25 kVA) costs RM7-10k per month. The cost included in the MCMC tower
model i.e. RM43,635 is insufficient to cover the generator cost.
U Mobile views that due to the fragmented nature of the tower rental industry and a
multitude of factors affecting this sector, it is impossible to produce a pricing structure
that fits every situation or configuration. U Mobile suggests that the MCMC streamline the
processes and introduce a clear and transparent pricing structure for tower providers
throughout the country to induce efficiencies which will lower cost and benefit end users.
This needs to be done soonest as network operators often are obliged to seek access and
pay rentals without options from designated providers such as SBCs or designated network
facilities providers in a particular locale. Commercial arrangements in the form of
discounts are offered, but so is the pass-through of increasingly (unwarranted) cost. U
Mobile highlighted the following:
(a) Proposed prices for 2018 are high compared to current levels;
(b) The trend in current arrangements has been a gradual decline unlike the model;
(c) The data built into the model may not have been provided on the same basis;
(d) Issue of exclusivity needs to be removed if prices are to be regulated effectively;
(e) The model averages prices in various regions and hence is not suitable; and
(f) The proposed tower type and height does not cover the large proportion of the
towers leased by telcos e.g. 30m and 45m lattice towers. Moving forward, there
will be an increasing trend towards smaller structures like lamp poles. These need
to be regulated as telcos often do not have alternative providers, but at the same
time, are governed by the need to meet coverage/service obligation.
Page 129
webe notes that it is important that the principle of cost-based pricing is applied to towers
as tower cost is a significant portion of the total cost to provide end services to consumers
at large. webe expressed concern that the proposed prices are on a per site basis,
however, the cost per access seeker is not clearly mentioned in the PI Paper. webe
highlighted that the appointment of second and following tenants are not based on a
straight line basis but at a pre-determined percentage. webe supports the initiative to set
a price ceiling for tower rental but is of the view that the costing for towers should be
based on a duration of more than 7 years as rental of towers are long-term arrangements.
The longer tenure will enable Access Providers and Access Seekers to negotiate for volume
discounts.
YTL agrees that the regulated prices should clearly state the provisioning of space including
space to install base station i.e. cabin space and not merely space to install antennas on
towers.
Discussion
The MCMC notes the point made by Maxis and webe that the length of contracts in this
sector is significantly greater than the envisaged price control period and that this might
create undue risks, or distort appropriate pricing practices in the market.
The MCMC also notes the points made by Digi about the application of prices to other
structure types, such as monopoles. The cost of these has been included in the average,
but separate costs were not quoted, because of a lack of data points in the sample. The
MCMC recognises, however, that these structures may have lower costs than the average
and would not expect operators to increase prices purely on the basis that they are lower
than the average. Furthermore, mark-ups to cover fixed and common costs have not been
included in the model, because it is not necessary to share such costs with other services
in the case of tower companies, and so, all costs can be included in the calculation. No
further mark-ups to prices would therefore be expected.
The MCMC also notes that several respondents reiterate the point about price de-
averaging, including PPIT, despite its acknowledgment that the modelled costs might
represent no more than a reference. It is difficult in consequence to understand how PPIT
can sustain the argument that this would lead to a major reorganisation of the towers
industry.
The MCMC further notes the point made by PPIT that an average price might result in
cherry-picking on the part of operators in their build or buy decisions and a detrimental
effect on universal service objectives. However, the MCMC emphasises again that the
Page 130
proposal was not that it should set an average price, but rather that legitimate cost
variations of this kind might be set in the context of an averaged reference cost.
PPIT also suggests that its prices are on a downward trend, in contrast to the upward tilt
in the model. Whilst the MCMC accepts that this may be true, the assumptions input to
the model were those provided by the operators. It is also an incorrect interpretation of
what is proposed in the PI Paper that negotiation would only take place with the MCMC’s
approval. Tower operators and their customers are free to negotiate with each other and
the MCMC would only become involved in the process if this process fails to reach
agreement.
U Mobile raises a number of additional issues around exclusivity of access and availability
of options. Whilst they are clearly an important element of the pricing mix, and may
indeed be indicators of sub-optimal functioning of the market, they require further
examination and a wider evidence base and are not best suited to being addressed by this
current review.
Whilst the MCMC shares some of the concerns raised by the operators about the efficient
functioning of the market, in some instances, it is not persuaded that ex-ante regulation
of prices in this sector is likely to lead to an improved outcome, taking into account the
complexities of the services and pricing structures in operation in the market.
Nevertheless, the MCMC will continue to monitor developments in the sector, with a view
to intervening should instances of ineffective competition become apparent.
The updated cost model provides the following indicative prices for Infrastructure Sharing
(Towers).
Question 30:
Do you have any comments on the approach to modelling in-building common antenna
system?
Page 131
Submissions received
edotco pointed out that IBCA systems are not only deployed by MNOs but also by them.
Hence it is inaccurate to depend solely on the mobile model, especially if access to such
facility is done on a reciprocal basis between operators. If the prices for these services
are used as reference, this would create unreal expectations and hence, should not be
included. edotco deploys IBCA systems in hospitals, arenas, shopping centres, buildings,
office complexes, hotels, Government departments, etc. and the CAPEX ranges from
RM440,000 to RM790,000 to deploy between 150-400 antennas while the OPEX is between
RM12,000 to RM42,000 per year. Each site and project is unique and customised solutions
are needed. As a result, the reference price is too low and one annualised price does not
even address the variability of sites. edotco requested for the details of the costs included
in the model and sought the MCMC’s views on the essential facilities that are subjected to
access regulation in this area. If the focus of regulation is excessive rental charged by
building owners, then the MCMC should model the cost of building owners providing access
to licensees for IBCA systems.
According to Maxis, the current approach for IBCA systems in Malaysia is that the systems
are installed by the mobile operator who then allows access to other mobile operators on
a reciprocal basis. This approach has been proven effective and efficient in terms of cost
efficiency and quality of services. However, there are also third parties offering IBCA
facilities to mobile operators for example Skai in KLIA2, which raises concern on quality of
service and exorbitant prices. For the benefit of the industry, Maxis suggests that the
MCMC set maximum regulated prices for IBCA systems in order to control exorbitant prices
imposed by IBCA provider. Maxis shared the IBCA prices commonly used by the mobile
operators for their IBCA facilities and suggested the MCMC use those prices. Maxis also
suggests that sensitivity analysis be carried out for IBCA systems provided by third party
access providers.
TM doubts the value of the IBCA mobile model results as inputs for potential future
disputes as IBCA facilities vary in size and scale. The specific case may be different from
the modelled IBCA. The MCMC will need more detailed data supplied by the mobile
operators showing the variation in cost in order to facilitate potential future disputes
involving IBCA facilities.
U Mobile views that the cost in the model does not elaborate on the size of the building
and the type of IBCA system. Large buildings will have more antennas and cover a wider
area which requires a method of sharing that allows the operators to recover their cost
while extending access to other seekers. The price per antenna in the model far exceeds
Page 132
the current average of approximately RM50 per antenna. A typical IBCA system would
have 50 antennas; thus the proposed prices should be approximately 3 times the current
price paid by industry.
YTL agrees for the inclusion of IBCA in access pricing. Since there are many variations in
the deployment of in-building infrastructure and building vary in size and shape, a
convenient way to price is by area covered by the system and efficient use of spectrum
allowed i.e. multiple-input multiple-output, also known as MIMO, etc. Access pricing
should have an in-built mechanism for the calculation of costs for sharing and inclusion of
new parties.
Discussion
The MCMC notes the comments offered by edotco and Maxis that apart from the mobile
operators, other providers also provide IBCA services. The MCMC also accepts the point
made by TM, edotco and U Mobile that the costs available from the mobile model do not
elaborate on the variations of cost with the size and nature of installations.
The suggestions made by Maxis, U Mobile and YTL about how potential access prices might
be structured to take account of cost variations and factors such as the addition of new
tenants are also noted.
Finally, the MCMC accepts the point made by Maxis that the most common arrangement
for IBCA in Malaysia is for reciprocal access amongst mobile operators.
The MCMC is not persuaded that there is sufficient cause to set ex-ante price controls on
IBCA pricing in Malaysia. Nevertheless, the MCMC provides the following indicative prices
for Infrastructure Sharing (In-Building Common Antenna Systems).
Page 133
9. Digital Terrestrial Television (DTT) Broadcasting service
9.1. Overview
The MCMC developed a bottom-up cost model for the new DTT broadcasting service using
tilted annuities as the annualisation method. The cost model with all commercially
confidential data removed was made available on request to interested licensees during
the Public Inquiry period.
Part G of the PI Paper concerned the DTT broadcasting service. Section 20 of the PI Paper
described the DTT model, including treatment of High Definition (HD), Standard Definition
(SD) and radio channels, channel take-up and Set-Top Box (STB) costs and the changes
made after model viewing, and sought comment on its completeness.
The MCMC proposed to set regulated prices for the DTT service.
Question 31:
Do you have any comments on the approach to the modelling of the DTT multiplex
costs?
Submissions received
Media Prima comments that the costs of STBs should not be included in the cost of DTT
multiplex as they were part of the requirements of the tender document and should not
be included as part of the cost structure for a Common Integrated Infrastructure Provider
(CIIP).
MYTV submits that its commitment of 2 million STBs is an integral part of MYTV’s bid to
the tender which is linked to the channel demands, the channel rates offered and its
revenue stream. MYTV comments that the moratorium period is important as a lot of work
has gone into development of SIRIM-approved STBs, and would allow MYTV to sell these
SIRIM-approved STBs to set-off against its contribution of 2 million STBs. There is
currently an influx of IDTVs in the market that impacts on the take-up of STBs. MYTV
points out that the IDTVs will not be able to cater for pay TV channels or provide HbbTV
services. Hence, MYTV’s STBs are the best choice for broadcasters and it is gravely unfair
that MYTV is unable to recoup the costs of the STBs that have been developed based on
the broadcasters’ requirements. In that regard, MYTV considers that STBs should be
included in the cost model.
Page 134
MYTV provides additional information of a loan facility and requests that the annual
financing costs be included in the cost model. In addition, MYTV revised its forecast, by
increasing the number of SD channels and reducing the number of HD channels. MYTV
submits that there are currently only two HD channels that have been taken up and it does
not envision that there will be any further demand for HD channels. Finally, MYTV clarified
that radio service is a supplementary service provided to broadcasters who have
committed to lease the entire multiplexer. In that regard, MYTV submits that radio service
should be excluded and the cost of radio services should be reallocated to SD and HD
channels.
Question 32:
Do you have any comments on the proposed regulated prices?
Submissions received
Media Prima submits that the proposed price is high and higher than what it pays currently.
In addition, Media Prima suggests that the MCMC benchmark the costs and the WACC of
a relevant operator, such as Arqiva in the UK.
With the evolution of technology to compress data and the requirement of different types
of content, such as sports, news, drama and movies, this influences the data capacity (bit
rate) required. For example, news programmes do not require high bit rate whilst sports
programmes require higher bit rate. Further, there is also the capability to reduce the
range of guard interval, and hence, increase the number of channels in the multiplexer.
In that regard, Media Prima views that news programmes might not need 2.5 Mbps and a
HD channel might not require 7 Mbps for better quality. As such, Media Prima proposes
to the MCMC to also regulate the price per multiplexer or lump sum capacity to provide
the Access Seeker with greater flexibility to manage the quality for themselves.
Media Prima comments that the price for radio channels is unreasonable, as it pays around
that amount for four radio stations, Hot, Kool, One and Fly.
Further, Media Prima suggests to the MCMC to review the price for the Average channel
cost. For example, it highlights that 6 multiplexers and 240 transmitters were used. Media
Prima also submitted some price information on 5kW, 1.5kW, 250W and 100W
transmitters.
MYTV submits that the cost calculated per channel for SD, HD and radio is not related to
the ratio of bandwidth utilisation. By using radio as the point of reference, the ratio is as
follows: 40 (SD): 240 (HD): 1 (Radio). From this ratio, it can be seen that the cost output
Page 135
for a radio channel is incorrect as it does not vary so much from the price of an SD channel,
whereas an SD channel uses 40 times more bandwidth than a radio channel. It is therefore
expected that the price of SD and HD are higher than the proposed price. Finally, MYTV
suggests that it is assisting broadcasters to obtain additional revenue from advertisers and
their focus should be on the additional revenue rather than to obtain lower prices.
Discussion
The MCMC notes the opposing views offered by MYTV and Media Prima on the subject of
STBs. In the MCMC’s view, Media Prima is correct in pointing out that the offer of providing
STB was part of the tendering exercise and that the expectation was that bidders would
not seek to recover this in higher prices charged to customers.
The MCMC also notes the point made by MYTV that radio channels are only sold in
conjunction with SD or HD TV channels. However, it is clear that they are not bundled
with such channels at no additional charge. The MCMC is therefore not persuaded that
the costs of radio channels should be incorporated into regulated prices for HD and SD
channels.
Both Media Prima and MYTV offer comments about the relative bandwidth requirements
of different types of channels and Media Prima make the point that different types of
programming may suit different bandwidths on a more finely graduated scale than a simple
SD/HD distinction. Media Prima also makes the observation that additional bandwidth
within a multiplex might be released by adopting narrower guard intervals and suggests
that broadcasters should be permitted to pay for a whole multiplex and to have the
opportunity to configure the bandwidth it provides to suit their requirements. These
observations tend to lead to a conclusion that it would be inappropriate to set specific
prices for SD, HD and radio along the lines of the costs set out in the PI Paper. Allowing
some degree of flexibility for MYTV and broadcasters to agree on the price structures within
the constraints to prevent excessive pricing would seem to offer the best opportunity for
innovation by broadcasters in matching their bandwidth consumption to the needs of
viewers and listeners.
The MCMC notes the additional information provided by MYTV, including updated forecasts
and details of financing arrangements. The MCMC does not propose to take the financing
costs directly into account, as these are the type of costs covered by the WACC.
MYTV submitted a revised channels forecast showing a lower take-up as part of its
response to the Public Inquiry. As MYTV did not provide any justification for the revised
forecast, the MCMC decided not to make any changes. In addition, the MCMC has issued
several new licences to broadcasters in the recent past and is of the view that the revised
Page 136
forecast is unduly pessimistic. As a result, the original forecast has been retained in the
model.
Based on the comments received from the respondents, the MCMC has removed the
reference to HD and SD channels in the price structure and replaced them with a two-part
structure of a fixed channel and a variable bandwidth costs. This accommodates the needs
of newer broadcasters who may be buying by number of channels or Media Prima who
may be interested in buying a multiplex. In addition, it takes into consideration the
development of compression technologies that reduces the amount of bandwidth required
per channel.
Broadcasters may in that regard agree with MYTV on the price per channel basis based on
their bandwidth requirements. They have the flexibility to negotiate on the payable charge
for the SD or HD channel, so long as it does not exceed both the per-channel and per-
unit-bandwidth prices. Likewise, prices for a whole multiplex can also be agreed upon
provided that it does not exceed the price calculated through the channel and bandwidth
pricing mechanism.
The MCMC confirms its preliminary view and sets the final prices of Digital Terrestrial
Broadcasting Multiplexing Services, made up of per-channel and per-unit-bandwidth costs,
as shown in Table 32. The broadcasters can negotiate the SD or HD channel or the entire
multiplex based on the per-channel and per-unit-bandwidth costs.
Per-channel cost (RM per year) 6.1 million 7.4 million 6.8 million
Page 137
10. Next Steps
This section identifies the next steps in the regulatory process following on from this PI
Report and the subsequent Commission Determination on Mandatory Standard on Access
Pricing (MSAP).
The impact of the Determination on the access agreements currently in place between
operators are anticipated to be as follows:
All access agreements shall be amended as soon as practicable to comply with the
Determination and shall be submitted for registration by the MCMC as required
under section 150 of the CMA.
Parties to access agreements shall apply the access prices in the Determination
once the Determination comes into effect.
Any service provider that offers the regulated facilities or services in the Determination
must modify its Reference Access Offer no later than 30 days from the date on which the
Determination takes effect.
Any service providers that offer regulated facilities or services which are included in the
Determination but do not currently have a Reference Access Offer must produce one no
later than 6 months from the date on which the Determination takes effect.
MCMC
20 December 2017
Page 138
Annexure - Abbreviations
Page 139
MGW Media Gateway
MMS Multimedia Message Service
MNO Mobile Network Operator
MSAN Multi-Service Access Node
MSS Mobile Softswitch
MTR Mobile Termination Rate
MVNO Mobile Virtual Network Operator
NBV Net Book Value
NPO National Policy Objective
Ofcom The Office of Communications of the UK
OLT Optical Line Termination
OPEX Operating Expenditure
OTT Over The Top
PI Public Inquiry
POI Point of Interconnect
PPP Public Private Partnership
PSTN Public Switched Telecommunications Network
QoS Quality of Service
RAO Reference Access Offer
RAN Radio Access Network
RM Malaysian Ringgit
SBC State-backed Company
SD Standard Definition
SIM Subscriber Identity Module
SMS Short Message Service
STB Set-top Box
Tbps Terabytes per second
TDD Time Division Duplex
TRX Transceiver
UPE User Provider Edge
USP Universal Service Provision
VOD Video On Demand
VoLTE Voice over Long Term Evolution wireless technology
WACC Weighted Average Cost of Capital
WDM Wavelength Division Multiplexing
Page 140