Telematics and Informatics 24 (2007) 86100

www.elsevier.com/locate/tele

A study of MVNO diusion and market structure

in the EU, US, Hong Kong, and Singapore
Dong Hee Shin *, Michael Bartolacci
Penn State University, Information Science and Technology, Tulpehocken Road, PO Box 7009, Reading, PA 19610, United States
Received 22 July 2005; received in revised form 14 November 2005; accepted 25 November 2005

Abstract
Mobile virtual network operators (MVNOs) are emerging as alternative channels for network operators to increase
market share and drive trac over their networks. MVNOs have gained popularity recently in Europe, but have a dismal
track record in Asia. In comparing the two regions, this study analyzes how the MVNO market has changed and what
opportunities and/or threats network operators and potential MVNO entrants are likely to face. The essential question
in next generation MVNOs is how the roles of mobile network operators, service providers and content producers will
change current and future value chains. This study conducts an economic assessment of market structure and environment
for dierent countries MVNO diusion. The data on market structure is analyzed by means of factor analysis techniques
in order to group countries according to their market environments. Then, structural equation modeling is used to investigate the relations of MVNO diusion and independent variables. The results show that the MVNO diusion signicantly
relates to market and industry structure.
 2005 Elsevier Ltd. All rights reserved.
Keywords: MVNO; Market structure; Comparative study

1. Introduction
When SK Telecom, Koreas biggest mobile operator with over 18 million subscribers, established a joint
venture in 2004 with Earthlink (ISP provider in the US) to provide the MVNO service (broadband wireless
service) in US, several questions were raised. One might wonder why SK Telecom sought MVNO business
in the US and not in Asia. This particular question is very compelling because Korea has a well-established
infrastructure and dynamics that support ICT development. In addition Korea has a rapid diusion rate and a
high service penetration rate. The simple answer for this question relates recent failures of MVNOs in Asian
countries, such as Singapore, Hong Kong, and Malaysia. In Asia, the MVNO market development is not as
visible as in Europe. SK Telecom appears to be well aware of the failures of other network operators and
is instead seeking greener pastures in North America where many MVNOs have been successful. This
*

Corresponding author. Tel.: +1 6103966135; fax: +1 6103962932.

E-mail addresses: dxs75@psu.edu, dshin@psu.edu (D.H. Shin).

0736-5853/$ - see front matter  2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tele.2005.11.003

D.H. Shin, M. Bartolacci / Telematics and Informatics 24 (2007) 86100

87

example clearly brings forth the scope of this analysis: the determination of where emerging MVNOs will play
the biggest role and be the most protable.
In a similar vein, when Virgin mobile retreated from its Singapore operations in early 2003, the company
cited weak economic and market conditions in the country as the reason behind the failure of the business
(Communications Today, 2002). The failure left some questions as to what market conditions MVNOs would
be able to succeed. A cross-national study could shed light on the question as to whether the observable differences are the outcome of a specic set of factors. This research investigates the set of factors inuencing
MVNO diusion in order to answer following questions:
Under what market condition do MVNOs succeed or fail?
What are the factors aecting the diusion of MVNOs in the world markets?
What are the future MVNO business models in next generation networks?
In this research, we analyze four cases for the purposes of answering these questions: two cases for the West
(European Union and the US) and two for the East (Hong Kong and Singapore). After collecting data on
market and industry, cluster and factor analysis are used to summarize patterns of market structure in the
telecommunications industry for these cases. Each factor is characterized by a set of coecients (factor loadings) expressing its correlations with the observed variables. The variables are assigned to the factor in which
they are most loaded. As a result, the market structure indicators are split into disjoint sets with each indicator
associated with a single factor. The estimated factor loadings applied to the country-specic market structure
indicators make it possible to score countries according to each of the factors, so that rankings of countries
can be obtained in terms of factor-specic scores. It is standard practice to retain a number of the factors
which cumulatively explain a substantial part of the overall covariance. A multivariate regression model is
developed to evaluate the eects of market variables on MVNO diusion.
The ndings of this study reveals that the MVNO diusion signicantly relates to market maturity and
industry structure (horizontal or vertical integration). In Asia, mobile operators have predominantly maintained high vertical market integration strategies where they control or own the spectrum, base stations, connecting networks, systems infrastructure, customer service operations, pricing, packaging, billing, branding
and points of sale. In contrast, European and US, mobile service is becoming a commodity and network operators are more open to outsourcing parts of the value chain. Operators in these regions increasingly view
MVNOs as ecient indirect sales channels that can decrease the focus on price because of erce brand loyalty,
lower subscriber acquisition costs by tapping existing customer bases and increase average revenue per user via
dierentiated content and services.
This paper is organized as follows: the denition of an MVNO is rst briey introduced followed by a
description of how one evolves. The next section reviews previous research on MVNOs. MVNO and their
position in the value chain model is discussed in Section 4. Section 5 details the model and data employed
in the paper, Section 6 explains the main ndings from the regression model and the factor analyses. Vertical/horizontal market structures and implications for MVNOs are discussed in Section 7. The nal section
details prospects for MVNOs with respect to networks of the future.
2. Denition and evolution of MVNOs
Since its origination in 2000 in Europe, the concept of an MVNO has varied. Regulatory bodies around the
world have adopted various denitions and dierent forms of regulatory intervention depending on the extent
to which an MVNO relies on the facilities of a mobile network operator.
The ITU (2001) dened an MVNO as a carrier providing users with mobile services without its own government-issued license for bandwidth. UKs Ofcom (2004) dened it as an organization providing customers
with mobile phone services without owning any airtime. Specically, OVUM (2000) dened an MVNO as a
carrier with the following characteristics: provides customers with a mobile service, has an independent mobile
network code (business registration number recorded on the SIM card), issues independent SIM cards, directly
operates a mobile communications switch including a home location register and does not own its own
bandwidth.

88

D.H. Shin, M. Bartolacci / Telematics and Informatics 24 (2007) 86100

In summation, the term MVNO refers to an organization that does not have an assignment of 3G spectrum, but is capable of providing public cellular services to end users by accessing radio networks of one
or more 3G spectrum holders. This broad denition reects the dynamics of the MVNO market which provides exibility to potential MVNOs to establish business models according to their nancial capability. This
in turn increases the level of competition which results in lower prices and innovative service oerings that
benet mobile users.
MVNOs are entering into US market based on their success in the European market. While obstacles await
them, the US MVNO market seems to be exploding with network operators being inundated with calls from
interested companies. However, MVNOs should be wary to avoid the pitfalls that led to failures in other markets such as Virgin in the Asia-Pacic region. Its strategy backred when its activities in Singapore folded in
July 2002 after just nine months. While Virgin has succeeded in Australia, there are now other MVNO operations in question in New Zealand, Japan, and the Philippines. For this research, Singapore and Hong Kong
were selected as the Asian cases where MVNOs have not been successful. For the Western cases, we selected
ve EU countries and the US where MVNOs are currently operating. In the EU, there are four types of
MVNO adoption policies generally represented: full adoption (Denmark, Netherlands, Norway, Sweden,
and UK), partial adoption (Germany, Spain, Swiss), partial prohibition (Belgium, Portugal, France), and full
prohibition (Italy). We selected the ve countries of EU who have a full adoption policy in order to oer a fair
comparison with the other global areas.
3. Studies of MVNO diusion
A relatively small but growing number of theoretical and empirical studies of MVNOs is available. Ulset
(2002) develops a transaction cost analysis of conditions for achieving net benets from opening the value
chain in mobile communications by introducing MVNOs. His conclusion is that MVNOs oering complex
bundles of innovative, value-added services will not be competitively sustainable as separate rms, only as
more tightly integrated partners of network operators.
There are several studies investigating access charges as well as interconnection and pricing issues for
MVNOs. Hwang and Oh (2004) conduct a techno-economic evaluation for US MVNOs using a real-option
approach. Kim and Park (2004) also study the optimal access charge for an MVNO system. Kanervisto
(2005) conducts a pricing structure analysis for an MVNO in Finland. Several studies also examine the characteristics of MVNO diusion. Sarraf (2002) surveys MVNOs in Europe and nds two types of MVNO players in the market: operator-centric and enterprise-centric. The operator-centric players have telecom as their
core business with a corresponding telecom corporate culture, while the enterprise-centric players are coming
from disparate type of core business and therefore have very dierent corporate cultures. The former players
are lling the network with trac, but only earning money on mobile services. The latter players seek their
prot in another business and use the MVNO as an extra sales channel to promote their brands and
products.
Several studies examine vertical and horizontal market integrations with respect to mobile data services,
although MVNOs are not specically mentioned. Looking at European MVNOs, Kiiski (2003) nds that
there are two main industry structures in mobile data service industry: a horizontally integrated, market-driven structure with modular product architecture, and a vertically integrated structure with an integrated
product architecture. Kisski also shows that European markets tend to be horizontally integrated while
Asian countries are primarily vertically integrated markets. Jang (2003) also investigates vertical structure
and competition in Korean telecommunications services market. Although all of these analyses are worthwhile, they do not clearly specify the factors of vertical or horizontal market structure, especially with respect
to MVNOs.
OECD (organizations for economic co-operation and development) researchers, Boylaud and Nicoletti
(2003) show a signicant relation between regulation, market structure and telecom performance. They empirically investigate the linkages between regulatory regimes, market environments and performance in three services supplied by the telecommunications industry. These three services are domestic long distance,
international long distance and mobile telecommunications. Based on the comparative experience of a large
set of OECD countries over the 1990s, it provides empirical evidence that liberalization of entry and the devel-

D.H. Shin, M. Bartolacci / Telematics and Informatics 24 (2007) 86100

89

opment of eective competition in telecommunications services generally lead to higher productivity, lower
prices, and better quality.
Although there have been many studies conducted, few studies have attempted to look at the MVNO eects
and market structure from a comparative perspective. Most of empirical studies on MVNOs are limited to
single countrys case. The focus on a single country was partly related to the lack of internationally comparable data on market structures. A study by Boylaud and Nicoletti (2003) provides an example of how dierent market structures can be compared across countries. In addition, many previous studies (Noam, 2003;
Kiiski, 2003; Jang, 2003; Lehr, 2001) help to conceptualize notion of market structure and link it to MVNOs.
4. MVNO in the telecommunications value chain: vertical disintegration
The MVNO is a new player in the telecom value chain which used to be comprised of operators, service
providers, application providers, and content providers. With competition and deregulation, virtual network
operators squeeze between operators and customers. In the past, telecommunications network operators used
to oversee many functions themselves including strict control over physical networks, and a good deal of control over services, content, marketing and billing. This walled garden approach resulted in a limited number
of services and content oerings. The vertical integration of operators began to disintegrate because of the
telecommunication markets downturn, the burden of high infrastructure costs such as UMTS (universal
mobile telecommunications system) licenses and 3G (3rd generation wireless network) implementations resulting in a lower average revenue per unit (ARPU), and increasing churn in saturated markets. The telecommunications market has changed considerably recently with less vertical integration, a greater degree of
outsourcing, and the abolition of monopolies with easier entry of new players into the eld (this is a sign
of open access networks). Such vertical disintegration has encompassed not only new telecommunications
operators, but also content, service and application providers as well as new equipment providers from the
information technology eld (Lehr, 2001).
Traditionally, as Duchesneau (1975) nds in the energy industry, the treatment of vertical integration has
presented substantial problems at both the industrial and governmental policy-making levels. Both the energy
and telecommunications industries are key national infrastructures vital to a given country. Just as in the
energy industry, where vertical integration seems to discourage competitive markets, telecommunications follows suit (Economides, 1996). With the emergence of MVNOs, a question arises as to why certain markets
continue to integrate while others start to disintegrate.
The MVNO model is a catalyst to such vertical disintegration process in telecommunications industry. This
creates the situation where many major players separate from the mobile network operators seeking to access
the value chain at all levels. Unlike other simple resellers of telecommunications services such as long distance,
local exchange, and mobile network services, MVNOs add value such as brand appeal, distribution channels,
and other anities to the resale of mobile services. In addition, unlike simple resellers, who often have little
relationships with others, MVNOs typically have close and tight relationships with other actors in the value
chain. Eective MVNOs have sucient agreements with existing operators to provide a good service coverage
area; and some well-diversied MVNOs can oer a product mix that incumbent mobile operators cannot
match.
MVNOs create a role in emerging value chains for application aggregators who can better understand conditions, structures and relationships in the operators environment and who have close ties with the application community and content providers. Mobile virtual network enablers (MVNEs) are an example of created
role by MVNO. An MVNE does not have a relationship with end-user customers. Instead, an MVNE provides infrastructure and services to enable MVNOs to oer services and have a relationship with end-user customers. In particular, for any non-telecommunications company, MVNEs help them to become an MVNO.
For example, MTV (music television) became an MVNO with the help of an MVNE. An MVNE oers infrastructure and related services ranging from network element provisioning, administration, operation support
systems, and business support systems. MVNEs often provide the intermediation between MVNOs that do
not want to have any control over network elements and those that want complete control. MVNEs provide
the middle-ground in the sense that they can provide options to MVNOs for what they bring in-house versus
what they purchase from the host carrier. A network owner would not be interested in hosting small MVNOs

90

D.H. Shin, M. Bartolacci / Telematics and Informatics 24 (2007) 86100

as the setup of any new MVNO requires many resources and may not generate enough trac on its network to
be protable. An another example of an MVNE function could be as a specialist in the outsourcing of certain
billing functions. Thus an MVNE could specialize in reducing the costs of introducing several smaller
MVNOs to a network operator.
An MVNE handles all of the operations for the MVNO while the MVNO brands and markets the service.
Exactly how much behind-the-scenes work is ooaded to the MVNE depends upon the MVNOs business
strategy. For example, seven 11 outsources all aspects of its MVNO business to Ztar mobile. However, Virgin
mobile eschews the MVNE model and instead handles all its own back-oce, customer care and distribution
functions.
MVNOs and MVNEs together play a pivotal role in adding value by screening, ltering and testing
applications and building relationships between the operators and outsiders. This aggregator model enables
operators to increase revenue by outsourcing content provision to third parties. The content aggregator plays
a key role in organizing the relationships between the operators and various content providers. The aggregators make the operators, the aggregator and the handset manufacturers work closely together. By working
with handset manufacturers, the operator is able to inuence the development of intelligent handsets and to
help specify certain applications that reside on the handset. Ericsson has extensive relationships with operators throughout the world and sees its role here as an intermediary between the operators and developers
while at the same time fostering partnership relationships by giving advice and oering investment assistance
to a range of application developers (Moriana, 2002). This aggregator role benets traditional mobile operators cooperating with MVNOs to broaden the customer base at a zero cost of acquisition. To mobile operators, embracing the use of an MVNO results in revenue to oset the high cost of implementing 3G
networks.
The existence of MVNOs and MVNEs can be indicators of degree of vertical disintegration and open
access markets. MVNEs rarely ourish in a market where vertically integration is prominent. The following
section examines six indicators of market structure.
5. Summarizing market structure for empirical analysis
5.1. The choice of indicators
For the purpose of evaluating the correlations of market structure and MVNO diusion, it is useful to focus
on a limited set of indicators that capture the main relationship between MVNOs and market structure. In
general, we adopt the indicators of market structure identied by the OECD Annual Report (2004); the degree
of vertical integration, the degree of liberalization, the degree of concentration, the degree of dependency, and
the degree of segmentation. This study adds ARPU (average revenue per unit) as a market performance measure since the FCC commonly uses the ARPU measure when assessing mobile market performance (2005).
In addition to these indicators, of course, the features of the pricing scheme, the governance mechanisms,
and regulatory institutions can have important eects on performance. However, price regulation, governance
and institutional change are all qualitative in nature and are dicult to convert to into quantitative variables.
In addition, such data are not suciently available over long enough time periods and across countries to be
useful. The current regulatory environment associated with mobile telecommunications could be included, but
it is characterized by a relatively low cross-country similarity of both price regulation and regulatory institutions that can be used in an empirical analysis. These dimensions of regulation and governance were therefore
ignored in the analysis.
The focus thus became two main issues: market structure and its performance. FCC (2005) uses four variables when assessing mobile telecommunication markets (1) market structure; (2) carrier conduct; (3) consumer behavior; and (4) market performance. Because of diculty in collecting and turning the qualitative
data on carrier conduct and consumer behavior into useful quantitative data, these variables are excluded.
The following indicators of the market environment were chosen:
The degree of liberalization (the liberalization index of OECD), as measured by ranking the legal limitations on the number of competitors allowed in each market.

D.H. Shin, M. Bartolacci / Telematics and Informatics 24 (2007) 86100

91

The degree of competition in the mobile telecommunications industry, as represented by the share of new
entrants or the number of competitors (OECD Annual Report, 2002).1
The degree of concentration in the wireless telecommunications industry, as measured by HHI (Herndahl
Hirschman Index)
hX
i
2
2
2
2
2
HHI 10; 000
S 1 10; 000S i S 2 S 3    S n 
f = number of rms participating in an industry
= each rms market share
i = rm in a given industry.

iSi

This formula also can be displayed as follows without the scalar:

HHI

f
X

S 2i

i1

The degree of mobile market performance as measured by average ARPU. ARPU is calculated by
dividing billable services-derived revenues for a given period by the average number of billable subscribers for that period. Despite much criticism on its aws, ARPU is widely used to measure market
performance.
The degree of vertical integration in the telecommunications industry, as measured by the percentage valueadded providers in the market which is widely used to be the most suitable indicator (Haller, 1997, p. 319).
If a market were completely integrated, producing the entire product, the ratio would be one. At the other
extreme, if a market were vertically disintegrated, producing most of the input from dierent suppliers, this
ratio moves towards zero (Levinstein, 2000).
The degree of market segmentation as measured by the market segmentation index (Lang Research Inc.,
2001). The market segmentation index is computed by dividing the incidence of each market segment in
a particular segment by the incidence of each market segment in the total market. A score of 1.0 means
that market segment is the same as the total market in a market sample. Index values above 1.0 indicate
that the market segment is over-represented in a given sample of the overall market while index values
below 1.0 indicate that the market segment is under-represented relative to the overall market in that sample. An index value of 1.2 means that market segment is 20% more likely to be found within the particular
sample than in the general population.
We can derive from these variables the structural equations in a reduced form to explain MVNO diusion
as follows:
Q = f (concentration, competition, vertical integration, liberalization, performance, and segmentation)
We used a multivariate regression method to estimate model parameters. Factor analysis and cluster analysis methods were used in mapping the structure of the observations and relations between possible measures
for the independent variables. Based on the theoretical model above, we formulated the following empirical
model:
MVNOD CONS b1 COMP b2 LIB b3 ARPU b4 SEG b5 COMP
b6VER b7DPOL e8

Oce of communications UK provides eective competition review guidelines that sets out four broad groups of the indicators of
eective competition: market structure; supplier behavior; consumer behavior; and consumer outcomes.

92

D.H. Shin, M. Bartolacci / Telematics and Informatics 24 (2007) 86100

Table 1
Data collected
EU averagea
Singapore
Hong Kong
US

MVNO share (%)

HHI

VI

Lib

Segm.

ARPU (%)

17
1.0
3.5
12

3948
2186
1936
3736

0.23
0.32
0.60
0.59

0.86
0.54
0.25
0.34

0.93
0.56
0.81
0.84

49.91
42.54
33.0
28.4

Mobile market

Denmark
Finland
Sweden
UK
USA
Singapore
Hong Kong

Number of license holders

Share of largest operator

Share of second largest operator

8
4
5
5
8
3
3

39
67
43
32
27
54
34

19
30
25
20
24
19
18

Sources: OECD Communications Outlook, ITU Database, World Mobile Market 19962010, OMSYC.
a
EUs weighted average (i.e., HHI is weighted by the countries share of telecom revenue).

with
MVNOD: MVNO diusion,
CONS: intercept (constant),
CONP: concentration (HHI),
LIB: liberalization,
ARPU: average revenue per unit, market performance,
SEG: segmentation,
COMP: competition,
VER: vertical integration,
DPOL: dummy reecting policy, regulation, carrier conduct, consumer behavior and external environmental regimes,
e: random error term.
The dependent variable (MVNO) was dened as the total MVNO market share in a country as percentage
of the total wireless market.
Data was collected (Table 1) on a period between 2002 and 2003. It should be noted that the Singapore
(Virgin mobile) case and the Hong Kong case (Shell mobile) both terminated operations in 2002 after being
in business less than a year.
5.2. Cluster analysis and factor analysis
Based on the data above, a cluster analysis was used to categorize and group mobile markets according to
market structure and environment. Cluster analysis is used to investigate the similarity of the countries based
on the country scores on each of the estimated factors. Starting from the values taken by the scores in each
country, this technique progressively groups countries in clusters of increasing size based on (multi-dimensional) pair-wise comparisons. At each step in the clustering process, an index of inter-group similarity measuring the distance between the clusters being joined is calculated. It is standard practice to stop the clustering
procedure (i.e., select the relevant number of clusters) when the distance between clusters becomes sizeable.
Using cluster analysis, several groups of countries could be identied based on their experience with
MVNOs over the period analyzed, the most vertical being Hong Kong and Singapore, and the most horizontal, layered or open being the UK, Denmark, Netherlands, Norway, Sweden, and US. This nding is consistent to Kiiski (2003)s study on mobile industry structure.

D.H. Shin, M. Bartolacci / Telematics and Informatics 24 (2007) 86100

93

Market clusters
Vertical
Singapore, Hong Kong

Horizontal or unbundled
Most of the EU and US

After the groupings of countries were created through cluster analysis, a factor analysis was carried out to
evaluate the underlying commitment factors and to reduce number of variables in order to avoid multi-colinearity (liner inter-correlation among variables: Table 2). Factor analysis is a statistical technique aimed at nding the minimum number of latent variables that explain the maximum amount of the overall covariance of
the observed variables. The factors, which are linear combinations of the observed variables, can be interpreted in economic terms such as market structure, openness, and dependency. Each factor is characterized
by a set of coecients (factor loadings) expressing its correlation with the observed variables and the variables
are assigned to the factor in which they are most loaded (Table 3). As a result, the market structure indicators are split into disjoint sets, each of which is associated with one factor. The estimated factor loadings
applied to the country-specic market structure indicators make it possible to score countries according
to each of the factors, so that rankings of countries can be obtained in terms of factor-specic scores. It is
standard practice to retain a number of factors which cumulatively explain a substantial part of the overall
covariance.
6. Empirical model and analysis results
Overall, the empirical results allow the identication of important factors that inuence the observable pattern of MVNO diusion (Table 4). Four main factors are found to best describe the cross-country variance in
the set of indicators of MVNO and market structure.
Table estimation results summarize the estimated eects of market structure on MVNO diusion for the
selected model specications (i.e., those that were not rejected by the tests). Overall, the market structure

Table 2
Results of factor analysis rotated factor loadings
Variables/factors

Performance

Structure

Ancillary regulation

Competition
ARPU
Segmentation
Concentration
Vertical integration
Liberalization

0.628
0.721
0.023
0.129
0.093
0.003

0.052
0.188
0.930
0.900
0.838
0.254

.0083
0.114
0.149
0.033
0.199
0.349

Factor loadings measure the correlation between the individual indicators and the latent factors. Indicators are assigned to the factor to
which they are most correlated. The rotation of factor loadings is a transformation aimed at minimizing the number of indicators that are
highly correlated with more than one factor.

Table 3
Correlation matrix (full model)

Concentration
Competition
Segmentation
Liberalization
ARPU
Vertical integration

Concentration

Competition

Segmentation

Liberalization

1
0.3923
0.2128
0.2334
0.3069
0.5560

1
0.7822
0.5933
0.5932
0.4982

1
0.4983
0.6923
0.4013

1
0.5932
0.7322

ARPU

0.492
3

Vertical integration

94

D.H. Shin, M. Bartolacci / Telematics and Informatics 24 (2007) 86100

Table 4
Estimation results
Variables/factors
Concentration
Competition
Segmentation
Liberalization
ARPU

Model 1
0.0078
0.029
0.0823
1.349
**
0.27442
3.492
0.0302
0.624
*
0.0899
3.392

Vertical integration
Constant
Prob > F
R2
Adj. R2

5.934
1.170
***
0.0071
0.5517
0.4136

Model 2

Model 3

0.00688
0.305

***

0.2432
4.203
0.234
0.992
0.0603
1.9323
*
0.183
2.032
3.423
1.153
**
0.0162
0.5711
0.4053

0.0232
2.301
***
0.235
3.999
0.034
1.932
**
0.00907
3.2093
*
0.192
0.123
7.392
1.923
***
0.0012
0.5417
0.4793

Model 4

Model 5

0.01003
2.566
*
0.1372
1.212
**
0.1578
2.942
*
0.134
2.321

**

0.0382
2.903
0.0034
2.393

**

0.0197
2.9332
0.031
3.234
4.923
0.132
***
0.0052
0.5291
0.4293

0.323
3.293
**
0.0129
0.323
0.203

Dependent variable MVNO market share in 20022003, t-statistic is italic.

*
Signicant at 90% level.
**
Signicant at 95% level.
***
Signicant at 99% level.

indicators performed quite well, signicantly improving the t of the regressions. By looking at the indicators
most closely associated with each of the factors, a straightforward economic interpretation can be made: the
rst factor (associated with vertical integration of the mobile market) shows that the degree of vertical integration is signicantly associated with MVNO diusion; the vertically integrated mobile market reduces the
MVNO diusion. The second (associated with segmentation in telecom services) expresses the positive significant relationship with the MVNO diusion; the third factor (associated with mobile market competition)
shows the positive relation that higher competition is related to higher MVNO diusion; and the fourth (associated with liberalization) factor accounts for the positive relation with the process of the MVNO diusion
(reword this too, I am not sure what you mean by encouraging eect). The concentration factor has the negative relation with the MVNO diusion, but the relation is not statistically signicant. ARPU does not show a
signicant result.
A principal components method with a varimax rotation of the six variables reveals ve underlying factors
with eigenvalues of greater than one. These ve factors explained 75.9% of the variability in the eleven variables. Model 1 does not account for the vertical integration variable and model 2 does not include the competition variable. Model 3 has a better overall explanatory capacity, but the variables of competition and
liberalization are not signicant. Model 4 does not include the dependency variable, but is generally acceptable. Model 5 drops four variables and shows the lowest explanatory power. There are some recurrent observations from these models. Vertical integration and segmentation are most statistically signicant factors. The
eect of these two variables is also relatively robust.
Dividing the sample into the two groups (vertical and horizontal) identied by cluster analysis showed more
insights (Table 5). The statistical analysis shows that the liberalization factor present in the horizontal group
had a positive and signicant impact on MVNO diusion. This can be explained by the EU Access Directive
(2002) which helped to correct vertical integration in communications services and facilitated MVNO
launches. The EU Access Directive (2002) contains requirements that vertically integrated network operators
must impose accounting separation and initiate transfer pricing arrangements where they control access to
essential input facilities that are used by their competitors in a retail market. The directive also requires that
service providers be given access to these network elements if not doing so would be detrimental to a competitive market. In contrast, the liberalization factor for the vertical group was detrimental to MVNO diusion,
although the eect also was not statistically signicant.

D.H. Shin, M. Bartolacci / Telematics and Informatics 24 (2007) 86100

95

Table 5
Estimation results, country groups
Variables/factors

EU

Concentration

0.08752
1.979
0.19832
2.392
*
0.09832
3.232
**
0.2983
4.344
0.0362
1.212
*
0.0738
5.1323
**
0.1374
4.999
**
12.3049
4.923
***
0.0071
0.5517
0.4136

Competition
Segmentation
Liberalization
Dependency
ARPU
Vertical integration
Constant
Prob > F
R2
Adj. R2

USA
0.01003
0.9302

Hong Kong

Singapore

0.0128
0.4923
0.09333

0.0832
2.942
*
0.0903
3.9434

0.0393
3.2832
0.3943
2.123
**
0.0162
0.5711
0.4053

0.0434
1.232
**
3.432
4.292
***
0.0012
0.5417
0.4793

0.0920

1.5683
4.233

2.034
4.213
0.984
1.032
***
0.0052
0.5291
0.4293

Dependent variable MVNO diusion in 2002, t-statistic in italics.

*
Signicant at 90% level.
**
Signicant at 95% level.
***
Signicant at 99% level.

Model runs at the level of country sub-groups show that the overall pattern is particularly visible in the
horizontal group of countries. In this group, competition is correlated negatively with MVNO diusion (with
no signicance) and concentration does not exhibit a signicant eect. Within the countries in the vertical
group, competition has the opposite eect and is positively related to MVNO diusion.
Concentration (measured by HHI) shows a positive eect, but not statistically signicant on MVNO diusion. This nding contradicts the widely held belief by several consulting rms reports. The HHI of the US is
more eective than Europe. The US HHI is 1377 compared to that of Europe (UK-2510, Finland-4297, Denmark-3399, Sweden-3878, Norway-5656, and Netherlands-2700).2 Market concentration is low by international standards in the US and the MVNO diusion therefore has been largely unaected by market
concentration.
Concentration also is not a signicant factor in Asia. Hong Kongs HHI is low (1936), which is an eective
HHI gure. The HHI gures of Hong Kong and Singapore contribute to the insignicant result of competition and concentration. As the competition variable shows, Hong Kong and Singapore are two of the worlds
most competitive mobile markets.3 The implication is that Hong Kong and Singapore are both small scale
markets, and therefore the incentive to innovate may be lower than in large markets. As shown in the EU
and US mobile markets, larger rms innovate more. The largest operators in Hong Kong and Singapore have
relatively small market shares. Hong Kong and Singapore are the two of the most fragmented world markets;
and less fragmented markets, such as the EU and the US have much higher MVNO penetration. An explanation may be that fragmentized or segmented submarkets in smaller overall markets obviate inter-operator
2
US Mobile Subscribers by Operator, 2003 (Source: FCC Ninth Report). Verizon, 37,522 (24%), Cingular 24,027 (15%), AT&T, 21,980
(14%), Sprint, 15,900 (10%), T-Mobil, 13,128 (8.6%), Nextel, 12,882 (8.5%), All Other (including MVNO), 25,983 (17%)

HHI 10; 0000:242 0:152 0:142 0:12 0:0862 0:00852 0:172  1377
In the same manner, EUs HHI can be calculated: UK-2500, Germany-3400, Italy-3800, France-3900, Spain-4000, and Netherlands-2700
(Source mobile Communications 24 June 2004).
3
Hong Kong has 11 networks spread between 9 mobile operators servicing a population of 8.7 million with an existing penetration
rate of over 90%. Singapore has 9 networks spread between 11 mobile operators.

96

D.H. Shin, M. Bartolacci / Telematics and Informatics 24 (2007) 86100

collaboration on application development (Godfrey and Kam, 2004). As the ITU Annual Report (2004)
shows, Hong Kong has a more fragmented market sector than any other country and no mobile operator
there accounts for more than 30% of the market. This segmented market produces a low HHI in Hong Kong.
As to the HHI variable itself, the HHI may not be a good measure in such a small market (Farrell and Shapiro, 1990; Demsetz, 1974). A more appropriate one may be Gans new measurement of HHI when it applies
vertical structure, namely vertical HHI (Gans, 2005).
The market segmentation variable shows dierent results across groups. The segmentations in EU and US
show signicant correlation with the MVNO market share. This result reects the current EUs MVNE proliferation. As of 2005, there are more than 36 MVNEs in the world and half of them are located in European
countries. MVNEs accelerate markets to be segmentized. There are already many dierent players providing
mobile service operations in Europe: exclusive mobile retailers (EMR), independent mobile retailers (IMR),
Exclusive Service Providers (ExSP), Independent Service Providers (IndSP), Indirect access providers (IAP)
and nally MVNOs. These dierent players are indicators that overall markets are segmented and fragmentized in Europe.
Interestingly, this segmentation variable shows a signicant correlation with vertical integration variable.
This can be explained that the fact that European MVNEs provide the telecom operators with a relationship
to manage through a single, trusted interface for dierent MVNOs (Kristensson, 2001). In other words,
MVNEs provide a bundled platform (packaged platforms for MVNOs). These equate to turnkey type solutions that typically include automated interfaces to a carriers provisioning system, real-time or near-real-time
usage, gateways for SMS (short message service) and content, as well as the usual telco back-oce systems and
services. MVNEs take such forms as application service providers (ASPs), billing companies, content providers, hardware manufacturers, and Internet or media companies.
The highly segmented markets of Hong Kong and Singapore have both been not favorable to the MVNO
model. The fact that the MVNO model has had diculty establishing itself in such segmented market environment suggests that the intensity of the facilities-based competition creates limited opportunities for them
to be economically viable, even with a dierentiated strategy in the face of aggressive price competition. Hong
Kongs MVNOs lease network services from facilities-based providers, and compete on the basis of retail service dierentiation. Yet, because of the already fragmented sector in these markets, MVNOs have not been
eective in focusing on niche markets that can be served at lower cost with a focused marketing strategy. They
have also not focused on oering superior customer service or bundling the mobile services with other valueadded services and features. MVNOs in Hong Kong have therefore become vertically integrated service providers who contribute to and enhance the vertical market structure.
As pertaining to market performance, the ARPU variable shows no signicant correlation although it does
show a positive relation. The insignicance of ARPU implies that the main challenge prospective MVNOs
will face is not necessarily a low ARPU, but other ones such as a niche market. Jost (2004) predicts that
an MVNO can potentially survive on $5 ARPU, but that the MVNO model in general must nd a way to
survive from the cost side of the business. The positive relation implies that MVNOs are active in markets
where they deal with low margin subscribers. The undesirable role of MVNOs in this scenario winds up as
protecting the host carrier from absorbing undesirable nancial costs such as those related to volatile prepaid
churn, high capital investment, and consumer fraud. All of these can accompany an MVNO that targets lowvalue subscribers. The positive correlation further suggests that MVNOs must nd a niche among incumbent
carriers who are willing to lease capacity since most operators are unlikely to open their networks.
Overall, the empirical analysis does show key drivers of MVNO diusion, but it also provides implications
that several constellations of factors may exist that can be conducive to MVNO diusion. However, dierent
tools, such as panel regression, will need to be used to examine this further. In addition, since we set regulations and policy as dummy variables, these variables amy exhibit some signicance across dierent models.
Future studies may investigate regulatory issues in further detail.
7. Vertical and horizontal structure and the mobile telecoms industry
The data analyses show that market structure is a signicant factor with respect to MVNO diusion. A high
degree of vertical integration in Asia has prevented MVNOs there from successful active operation. Part of

D.H. Shin, M. Bartolacci / Telematics and Informatics 24 (2007) 86100

97

the consequence of vertical integration is the high prole of signicant market players with their dominant
multimedia services. The vertical integration of markets has been substantiated through strategic partnerships,
alliances and joint ventures, enabling early market penetration. The absence of regulation has adversely
aected MVNO emergence. In Asia, where regulation tends to be interventionist, there are no policies regarding MVNOs.
The European and US cases are going through a dierent phase of the business cycle. European MVNOs
are already launching multiple operations across the continent while only a few US MVNEs are working with
MVNOs. One common characteristic of both markets, though, is that MVNOs are in horizontal structures
which will eventually make the market mature. Both the EU and US cases have evolved away from vertical
integration driven by the need to achieve economies of scale and specialization. Such horizontal relationships
do not seem to exist in Asia at the present time, but it suggests a way forward for the evolution of the telecom
industry in Asia.
Network industries of Asia in general are typically described as vertical industries. In Asia vertical integration often results in increased the protability due to the elimination of the so-called double marginalization
eect (Hart and Tirole, 1991). In reality, the vertical structure in the Asian market has brought about economic eciency through the minimization of production and transaction costs, and therefore, potential consumer benets. Operators have been mostly involved in three levels of the industry; the base products, network
infrastructure and the interface with the end-users through the services and products (Sarraf, 2002). Companies deliver products for the infrastructure and the nal customers, the end-users obtain the services and products from companies situated downstream the in the value chain. Network operators oer services at dierent
levels of the value chain. Vertical integration creates an incentive for integrated suppliers to raise wholesale
prices to competing retailers as a means to increase prots in downstream markets. Vertical integration also
limits the supply of independent marketers available to potential arbitragers. With new technical platforms,
this basic value chain, based on the vertical structure of a network industry becomes more complex.
Recently, Asian governments attempt to restructure the vertical market by opening up for new operators.
In so doing, new operators can enter the market without owning a bandwidth license, the competition
increases and the vertical integration will be reduced. With a decrease in vertical integration, an increasing
number of MVNOs will enter the mobile telecom industry fullling one or several roles in the mobile telecommunications value chain. Despite regulatory change, however, the markets are still vertically integrated in nature. Outlooks for MVNOs are not so bright given this high degree of vertical structure. One example
supporting this statement is that in Asia there are more operator-centric MVNOs than enterprise-centric
MVNOs (Sarraf, 2002).4 This can be contrasted to the variety of types of MVNO players that emerged in
Western Europe. Such dierent types of MVNOs date back to the EUs telecommunications privatization
and the entrance of GSM in 1990s both of which changed the industry structure of many European countries.
Companies which were previously vertically integrated changed their operations to compete in a horizontallylayered, market driven business environment. Products became modular and standardized because the best
results were believed to be achieved by dening open standards and letting the markets do the rest (Sarraf,
2002).
8. Conclusion: future convergence value era
The emergence of MVNOs has revealed two main industry structures in the mobile telecommunications
industry: a horizontally integrated, market-driven structure with a modular product architecture, and a vertically integrated, walled-garden structure with an integrated product architecture. As shown previously,
MVNO services are much more successful in Europe than in Asian countries. One major reason for this difference in success is the dierences in the industry structures between the markets in both areas. Because of the
complexity of MVNO services, a horizontal-layered structure seems to be more successful. This nding
4

The operator-centric players have telecom as their core business with telecom culture. They are more oriented towards lling the
network with trac, and earning money on mobile services only. Since the Enterprise centric players are coming from another core
business, they have a dierent mindset. Normally they make their prots in another business and might use the MVNO as an extra sales
channel to promote their brands and products.

98

D.H. Shin, M. Bartolacci / Telematics and Informatics 24 (2007) 86100

provides future carriers and potential MVNOs with a framework to determine if the MVNO model is right
for their respective business. On the unprecedented success of Virgin mobile, and also the surge of multiple
European initiatives, both telecom and non-telecom companies in Asia (Korea, Japan and China) are now
expressing their interest in this new arena. Such companies may use the success factors discussed in this work
to address key considerations of MVNO adoption.
The Asian industrial framework (and regulatory environment to some extent) is optimized for traditional
mobile voice services. It does not allow MVNOs to develop their business models in ways that have turned to
be popular in Western Europe. In other words, MVNO services are more successful in markets where virtual
operators take intermediary and aggregating roles in order to oer a true end-to-end mobile service. The
incumbent Asian operators are not eager to change the market structure. It seems that MVNOs would have
nothing but negative consequences from the consumers point of view if the market structure changed.
In his study on market structure, Economides (1996) leaves open question of whether the horizontal or a
vertical industry structure is the better for MVNOs. To be more specic, the open questions would be: Is it
more valuable to have a transparent, market-driven industry structure or a vertically integrated structure
enabling provision of easy-to-use services? Would this more integrated structure provide a mobile telecommunications market with only one, monopoly-type player. This study partly answers such questions by showing
that a horizontally-layered market is better with respect to facilitating MVNO diusion. It is clear, however,
that the old vertically integrated telecom business models are no longer viable in an environment where the
demand for a whole range of value-added services is strong. Under such a horizontal value chain, vertical integration will be minimized or eventually disappear.
Under vertically integrated market, the owners of transport network have a real advantage by exerting
economies of scale and scope, and the potential to exploit bottleneck control of major parts of value chain.
The future of the value chain will see more and more of a separation between infrastructure and services.
The MVNO is an example that independent from its underlying infrastructure. A variety and innovative services of MVNO shows an example of redistribution of control and intelligence from infrastructure to service.
With such separation, the role of middleware services will be increasingly important while intermediating
infrastructure and applications. The development of the horizontal value chain will make a proliferation of
MVNO (or enhanced service providers) operations inevitable, with many major players (apart from the
mobile network operators themselves) seeking to access the value chain at all levels.
9. Limitations and future studies
The main goal of this study was to examine cross-national data in order to highlight possible factors related
to the observable patterns of MVNO penetration. For this purpose, a mobile market structure was analyzed
through the analysis of important economic factors related to the diusion of MVNOs. This study used an
aggregation model and data was collected for several countries for the limited time of 20022003. Therefore, it
would be helpful for future studies to conduct cross-national studies at a less aggregated level, such as using
cross-sectional time-series analysis methods.
One question that remains uncertain for MVNOs in Asia and Europe alike is the long-term viability of the
MVNO. This question relates to how the value chain will develop and where the bottlenecks will remain. Some
predict that future MVNOs will be dierent from the current voice-centric and operationally light discount
service providers of today. Some imply that the next-generation MVNOs will be distinct in their ability to
leverage future technologies such as mobile broadband, IP-based multimedia, business vertical and convergent
solutions such as biometric and cellular applications. In particular, IP-based access technologies will open up
new wholesale options for MVNOs. Next-generation MVNOs will be more complex in their approach to the
market than their rst-generation versions of today. They will enter the mobile market with unique operational models, clearer dierentiation, and a wider and more complex array of services; and they will leverage
more robust access platforms beyond GSM to deliver these services.
The current study intentionally simplies and limits the variables in order to grasp the overall understanding of MVNO diusion factors. The limited variables allowed us to set aside regulations, policy and carrier
conduct and behavior as dummy variables. As Economides (2002) contends, the structural dierences are
caused by the dissimilarities in the national regulations. The signicant results of the dummy variable imply

D.H. Shin, M. Bartolacci / Telematics and Informatics 24 (2007) 86100

99

that there is something more behind the MVNO diusion and market structure. For example, despite the overall similarities between US and EU, the lower MVNO in the US can be explained by the fact that the US operators are carefully watching MVNO development as they believe that the transition period between 2.5G and
3G will be the most fruitful period for MVNOs (Besen Group White Paper, 2003). This kind of carrier behavior will need to be investigated with respect to regulation and policy.
With regard to HHI, this study shows that HHI can give a misleading result when it is used in small size
markets. In the case of a highly fragmentized market, HHI produces an ideal index leading one to believe that
the market has eective competition. While the ARPU measure can be a useful indicator in comparing markets operating under similar business conditions and with similar products, it can also be signicantly misleading when applied to emerging markets. The reason for this is that historic ARPU development does not take
into account the rapid reduction in equipment and operating costs or the dramatic change in industry realities
which are clear phenomenon with respect to MVNOs. Therefore, both HHI and ARPU (as many other studies indicate) should be used to corroborate the results from other variables.
References
Bensen Group, 2004. Which MVNEs will be the most successful? Bensen Marketing Group White Paper, May 2004.
Communications Today, 2002. Singapore Does not Work For Virgin Mobile, Brief Article Communications Today, July 11, 2002.
Demsetz, H., 1974. Two systems of belief about monopoly. In: Goldschmid, H. et al. (Eds.), Industrial Concentration: The New Learning.
Little Brown, Boston, Massachusetts, pp. 164184.
EU Access Directive 2002/19/Ec of the European Parliament and of the Council of 7 March 2002 on access to, and interconnection of,
electronic communications networks and associated facilities (Access Directive), Ocial Journal of the European Communities.
Available from: http://europa.eu.int/eur-lex/pri/en/oj/dat/2002/l_108/l_10820020424en00070020.pdf.
Duchesneau, T., 1975. Competition in the US Energy Industry: A Report to the Energy Policy Project of the Ford Foundation. Ford
Foundation Press.
Economides, N., 1996. The economics of networks. Int. J. Ind. Organ. 14 (2), 675699.
Economides, N., 2002. US telecommunications today (October 2002). Stern School of Business Working Paper. Available from: http://
ssrn.com/abstract=378380 (accessed in April 2005).
Farrell, J., Shapiro, C., 1990. Horizontal mergers: an equilibrium analysis. Am. Econom. Rev. 80 (1), 107123.
Federal Communications Commission, 2005. WTB Seeks Comment On CMRS Market Competition, WT Docket No. 05-71, DA 05-487,
Released: February 24, 2005. Available from: <http://hraunfoss.fcc.gov/edocs_public/attachmatch/DA-05-487A1.doc>.
Gans, J.S., 2005. Concentration-based merger tests and vertical market structure. J. Econom. Literature 43 (2), 238256.
Godfrey, J., Kam, J., 2004. What it takes to kick start mobile data: the lessons from Japan and South Korea. ITU Asia 2004, Online News
Service.
Hart, O., Tirole, J., 1991. Vertical Integration and Market Foreclosure, NBER Reprints 1517, National Bureau of Economic Research,
Inc.
Haller, L., 1997. An econometric analysis of the demand for RTE cereal: product denition and unilateral market power eects.
Marketing Policy Center Research Report No. 35, University of Connecticut, Storrs, CT.
Hwang, J., Oh, J., 2004. Techno-Economic Evaluation and Analysis for MVNO (mobile virtual network operator) Adoption and
Spectrum Policy and Regulation Using Real-Option Theory Modeling. In: Proc. Int. Conf. Electronic Commerce Research, Dallas,
TX, USA, June 2004.
ITU, 2001. Mobile Virtual Network Operators, Geneva. International Telecommunications Union.
ITU Annual Report, 2004. Survey of Asian telecom market. Available from: http://www.itu.no/learchive/ITU_annual_report2004.pdf
(accessed in May 2005).
Jang, B., 2003. Vertical structure and network competition in telecommunications services market. Inform. Soc. Res. (Korean edition)
Fall, 81101.
Jost, J., 2004. Can MVNOs Succeed in Emerging Markets? Pyramid Research, 2004.
Kanervisto, J., 2005. MVNO Pricing Structures in Finland, Ministry of Transport and Communications. Finland Government
Publication, Helsinki. Available from: <http://www.mintc./www/sivut/dokumentit/julkaisu/julkaisusarja/2005/2105.pdf>, February
2005.
Kiiski, A, 2003. Mobile Data Service Industry Structure: Walled Garden versus Horizontal Integration. Research Report of Networking
Laboratory, Helsinki University of Technology.
Kim, B., Park, S., 2004. Determination of the optimal access charge for the mobile virtual network operator system. ETRI J. 26 (6), 239
251.
Kristensson, S., 2001. Mobile Virtual Network Operators. Assessing MVNO Business Opportunities, School of Business, University of
Stockholm, Master of Science Thesis, June 2001.
Lang Research Inc., 2001. Travel Activities and Motivation Survey, Touring Segmentation Report, October 2001, Canadian Tourism
Commission, ISBN: 0-7794-2270-8.

100

D.H. Shin, M. Bartolacci / Telematics and Informatics 24 (2007) 86100

Lehr, W.H., 2001. Vertical integration, industry structure, and internet telephony. In: McKnight, Lehr, W., Clark (Eds.), Internet
Telephony. The MIT Press.
Levinstein, M.C., 2000. Vertical Integration. In: Mokyr, J. (Ed.), Oxford Encyclopedia of Economic History, New York.
Moriana Group, 2002. The new value chain in telecommunications: a threat or opportunity for telecom operators? The OSA/Parlay
Report.
Noam, E., 2003. The Internet: Still Wide Open and Competitive? OII Internet Issue Brief No. 1, August 2003.
OECD Annual Report, 2002. Outlook on the competition. Available from: http://www.oecd.org/dataoecd/10/6/35892523.pdf (accessed in
June 2005).
OECD Annual Report, 2004. World IT and telecom market structure. Available from: http://www.oecd.org/dataoecd/10/6/15692529.pdf
(accessed in June 2005).
OECD Economics Department Working Papers, 2001. OECD Economics Department in its series by Olivier Boylaud and Giuseppe
Nicoletti. Regulation, market structure and performance in telecommunications.
Ofcom, 2004. The Communications Market 2004: Telecommunications, October 2004. Available from: <www.ofcom.org.uk/research/cm/
cmpdf/telecoms.pdf>.
OVUM, 2000. Virtual Mobile Services: Strategies for Fixed and Mobile Operators, June 30, 2000. OVUM Research and Consultancy,
Market Research Report.
Sarraf, M., 2002. MVNOs-The New Concept for Mobile Operators. Ericsson Report of Technology Trends. Available from: <www.ordreing-bey.org.lb/symposium/proceedings/11E_SARRAF.PDF>.
Ulset, S., 2002. Mobile virtual network operators: a strategic transaction cost analysis of preliminary experiences. Telecommun. Policy 26,
537549.

Further readings
Besanko, D., Perry, M.K., 1994. Exclusive dealing in a spatial model of retail competition. Int. J. Ind. Organ. 12, 297329.
Bolton, P., Whinston, M., 1993. Incomplete contracts, vertical integration, and supply assurance. Rev. Econ. Stud. 60, 121148.
Caillaud, B., Jullien, B., Picard, P., 1995. Competing vertical structures: precommitment and renegotiation. Econometrica 63 (3), 621646.
Cambini, C., 2000. Competition between vertically integrated networks. Inform. Econom. Policy 13 (2), 137165.
Grant, R., 1995. Contemporary Strategy Analysis: Concepts, Techniques, Applications, fourth ed. Basil Blackwell Publishers, Oxford,
2002.
Hart, O., Tirole, J., 1990. Vertical integration and market foreclosure. Brookings Papers on Economic Activity, 205285.
Strand, J., 2004. Will discount plays kill mobile operators? In: Proc. of ITU Telecom Asia 2004, pp. 7889, September 711, 2004, Busan,
Korea.
Telecoms Infotech Forum Brieng paper, October 1998, Hong Kongs 1998 Television policy review. Available from: <http://
www.trp.hku.hk/tif/9810brieng.pdf>.