Telecommunications in Small Economies: the Impact of

Liberalisation and Alternative Technologies on Universal Service

Pavlos C. Symeou and Michael G. Pollitt

University of Cambridge – Judge Business School

December 2006

ABSTRACT
Existing research on telecommunications has produced limited evidence regarding the
effects of competition on Universal Service: the requirement for telecommunications
firms to provide access to fixed voice at geographically uniform retail tariffs.
Moreover, recent literature on telecommunications has examined the impact of
alternative technologies (mobile telephony and the Internet) on Universal Service but
has also produced ambiguous evidence on the nature of the relationship. This
evidence is based on studies which have concentrated mainly on large industrial
economies or large developing economies. The literature on the liberalisation of
telecommunications is therefore lacking in research in the context of small economies.
This paper is motivated by the literature on small economies which suggests that
optimal national policies should differ with size. The paper discusses the effects of
competition and alternative technologies on Universal Service when smallness
becomes influential. An econometric model assesses these effects. Research findings
support that, for small economies, smallness constitutes a significant factor leveraging
the outcomes of competition and alternative technologies on the fixed network and
prices. Competition effects on Universal Service vary between the three
telecommunications markets. Alternative technologies effects on Universal Service
are stronger that liberalisation in the respective markets per se. Policy implications
suggest that small economies require endogenous policy that addresses their
distinctive characteristics. Small economies are more apt to encounter natural
monopoly in the fixed voice. Incumbent providers in small economies may find it
practical to serve demand in fixed voice in neighbouring economies. Technologies
that exhibit a relationship of substitution should be combined in one market and be
ruled by the same policy. Technological evolution calls for a redefinition of Universal
Service.

Keywords:
Small economies; Universal Service; Mobile telephony; Internet; Liberalisation;
Telecommunications

2
 TABLE OF CONTENTS

TABLE OF CONTENTS ....................................................................................................3

CHAPTER 1 – INTRODUCTION .......................................................................................4
CHAPTER 2 – LITERATURE REVIEW .............................................................................5
2.1 THE INCEPTION OF TELECOMMUNICATIONS REFORM – THE US, THE BRITISH AND JAPANESE REGIMES
..................................................................................................................................................................6

2.2 THE IMPACT OF THE US AND BRITISH MODELS ON EU’S LIBERALISATION ..........................................8

2.3 LIBERALISATION IN THE “REST” OF THE WORLD .................................................................................9

2.4 TELECOMMUNICATIONS REFORM – A UNIVERSAL MODEL? ................................................................10

2.5 UNIVERSAL SERVICE IN THE FIXED TELEPHONY ................................................................................12

2.5.1 Empirical studies on the effect of competition on USO............................................................14
2.5.2 The role of mobile telephony in USO.......................................................................................16

2.6 LIBERALISATION AND THE SMALLNESS OF ECONOMY .......................................................................19

2.6.1 Small economies – an overview ...............................................................................................19
2.6.2 Smallness and Efficiency..........................................................................................................21
2.6.3 Smallness and (Natural) Monopoly ..........................................................................................23
CHAPTER 3 – RESEARCH METHODOLOGY .................................................................24
3.1. DATA AND VARIABLE DESCRIPTION.................................................................................................27

3.2 THE EFFECT ON FIXED TELEDENSITY ................................................................................................32

3.3 THE EFFECT ON FIXED PRICES ...........................................................................................................38

CHAPTER 4 – DISCUSSION ...........................................................................................41
CHAPTER 5 – POLICY IMPLICATIONS .........................................................................44

3
CHAPTER 1 – INTRODUCTION
The effects of the liberalisation of telecommunications have been the centre of
research for a large number of studies. The evolving nature of the industry makes the
assessment of competition effects a nontrivial task. Competition effects are difficult to
segregate for they comprise a synthesis of inextricable concerns. A fundamental
concern for policy makers and researchers is the effect that competition has on
Universal Service.
The concept of Universal Service is ambiguous as it lacks a precise content. It
usually is meant to imply that access to fixed lines be generalised and at affordable
prices. It therefore implies that Universal Service policy should bear into account both
the development of the fixed voice network and the level of prices. The notion is also
complex as its foundations for access to voice communications are shaken by
evolving alternative technologies and changing consumer needs.
Existing research on telecommunications has produced limited evidence
regarding the effects of competition on Universal Service. Moreover, recent literature
on telecommunications has examined the impact of alternative technologies on
Universal Service but has also produced ambiguous evidence on the nature of the
relationship. This evidence is based on studies which have concentrated mainly on
large industrial economies or large developing economies. Whilst the literature on
small economies postulates that optimal national policy should differ with size the
telecommunications literature has placed nominal emphasis upon it.
This paper examines the effects of competition and alternative technologies on
Universal Service when smallness becomes influential. An econometric model
assesses these effects. Smallness is depicted by a composite index consisted of the
measures of population, GDP and arable area. The effects of liberalisation are
modelled by distinguishing between competition in the distinct markets of fixed
voice, mobile telephony and the Internet. Alternative technologies are represented by
development indicators in mobile telephony and the Internet.
Research findings support that, for small economies, smallness constitutes a
significant factor leveraging the outcomes of competition and alternative technologies
on the fixed network and prices. Competition effects on Universal Service vary
between the three telecommunications markets. Alternative technologies effects on
Universal Service are stronger that liberalisation in the respective markets per se.
Policy implications suggest that small economies require endogenous policy that

4
addresses their distinctive characteristics. For their characteristics, small economies
are more apt to encounter a natural monopoly in the fixed voice. Incumbent providers
in small economies may find it practical to serve demand in fixed voice in
neighbouring economies that may allow them exploit economies of scale.
Technologies that exhibit a relationship of substitution should be combined in one
market and be ruled by the same policy. Finally, technological evolution and market
convergence call for a redefinition of Universal Service.
The structure of the paper has as follows. Chapter 2 presents a literature
review on the liberalisation of telecommunications and elaborates on existing
empirical studies which examine the impact of competition and alternative
technologies on Universal Service. The concept of smallness is then introduced and its
implications on liberalisation are discussed. Chapter 3 concerns the empirical part of
the paper where the research methodology is unfolded and the econometric model
assessing the effects on Universal Service is discussed. Chapter 4 elaborates on the
econometric analysis and Chapter 5 builds on the research findings to suggest policy
implications.

CHAPTER 2 – LITERATURE REVIEW

Telecommunications policy has been within the scope of numerous studies covering a
broad array of themes. Review of the Economics literature has shown that these
studies have mainly taken place in large countries with developed and developing
economies but have overlooked small countries with small economies. Since a large
number of countries have adopted telecommunications policy congruent to the models
of large developed countries, the issue of smallness calls for examination as it might
influence the liberalisation process and outcomes.
This section acknowledges this gap in research. It first discusses the
liberalisation process in telecommunications by reviewing the hitherto practice and
referring to empirical work that examines the role of competition in the industry.
Then, it brings up the issue of the size of the economy and unfolds the generic
leverage that size may have on the liberalisation of an economy. It finally attempts to
create a linkage between the liberalisation of telecommunications and smallness by
associating them with the issue of Universal Service.

5
2.1 THE INCEPTION OF TELECOMMUNICATIONS REFORM – THE US, THE BRITISH AND

JAPANESE REGIMES
Traditionally, the telecommunications networks in many countries were operated as
Post, Telegraph, and Telephone administrations (PTTs) which formed part of the
governmental administration. They were usually integrated with postal services, many
of them administered directly by ministries and staffed by civil servants (Eliassen et
al., 1999). The PTTs held a de facto monopoly over the national telecommunications
infrastructures, and had a double role as both regulators and suppliers. In many cases,
they met the increasingly diversified public demand with difficulty.
This trend started to change with the splitting of US’s AT&T into one long-
distance company and the seven Baby Bells, which was followed by the privatisations
of British BT and Japanese NTT (Eliassen and Marit, 1999). That was when the
governments realised that the hitherto organisational models for their
telecommunications sector were badly suited to cope with rapid development. The
crucial role of the telecommunications industry as an infrastructure for other
industries made it imperative to retain competitive telecommunications operators both
nationally and internationally.
Moreover, the telecommunications companies understood that in order to
survive in the new technological environment, access to foreign markets was crucial
(Hills, 1986). This spilled-over to the policy makers which put international market
access at the focus of their attention. The new threats to the old monopoly incumbents
that arose from technological developments were so ground-breaking that policy
makers were forced to review their policies. This was made intense especially when
the erstwhile “natural monopoly” perception collapsed.
The leverage that regulatory reform in the US had on a series of subsequent
reforms in other countries is of paramount importance. Particularly, the effect on
Britain’s and Japan’s liberalisations was of critical importance as they later became
more or less examples to the EU and Asian countries. These three countries
experienced, in different phases, four major models of regulation (Hills, 1986) which,
in a great extent, consisted of the prevailing models in recent years’
telecommunications reform. Namely, these models consisted of public ownership
coupled with Ministry regulation; private ownership coupled with Ministry regulation;
private ownership coupled with regulation by a government appointed agency; and
self-regulation.

6
 The arguments of the proponents of deregulation in America – that
competition spurs the provision of new services, it creates more efficient allocation of
resources and that it reduces prices – were used by the proponents of privatisation in
both Britain and Japan. Just as the final agreement on the Bell break-up allowed
AT&T into new markets, Britain and Japan also extended the boundaries within
which BT and NTT might expand.
In the early 1980s when liberalisation in the US began, a similar situation
started in the Japanese equipment market. When after the divestiture of AT&T in
1984 the Federal Communications Commission (FCC) allowed specialised common
carriers entry into AT&T’s market with leased data circuits, the same process of
liberalisation beginning with the sharing of leased circuits also begun in Japan. In
addition, divestiture of NTT was among the initial plans of the Japanese regulator as
of 1981. However, it was not before 1996 that the Japanese regulators managed to
apply the American example of AT&T’s divestiture (Bohlin, 1997).
In parallel, the UK’s liberalisation began in 1981 with the corporatisation of
BT and the entry of Mercury; the first competitor to BT’s core business. The
liberalisation of mobile services and cable television markets followed, which also
inaugurated the privatisation of BT. Concurrently, a regulatory framework and
agency, Oftel (later renamed Ofcom), were put into place and a duopoly formed by
BT and Mercury was established. Markets were further liberalised in 1991 following a
review of the duopoly policy.
In all three countries privatisation and liberalisation resulted in
telecommunications markets in which one private entity had overwhelming
dominance and remained regulated. The major differences were in the extent of the
regulation still exerted over those entities and in the extent to which competition was
allowed against them. What we were witnessing at that time was re-regulation rather
than deregulation (Hills, 1986).
It is hardly surprising therefore that the terms “deregulation”, “liberalisation”,
and “privatisation” produced different political mechanisms of intervention in the
three countries. What is evident in all three, however, is a pattern of the state
struggling to retain control over markets, which were formally liberalised and which
were therefore more susceptible to foreign penetration. It is also evident that whereas
pre-divestiture regulation in the US was divided between the FCC on interstate
services and the state public commissioners on intrastate, the divestiture opened space

7
enough for the FCC to extend its regulatory decisions to intrastate matters. As in
Britain and Japan, deregulation provided the opportunity for the wider exercise of
power by the central bureaucracy (Hills, 1986).

2.2 THE IMPACT OF THE US AND BRITISH MODELS ON EU’S LIBERALISATION

The elements of the US, Japanese and British liberalisation models had considerable
influence on the process of regulatory reform in Europe, in terms of examples to be
followed and dangers to be avoided. Specifically, Britain has constituted an important
pressure and example for regulatory changes at the European and national levels
(Thatcher, 1999). The single-market programme in the EU has necessitated the
regimes in member-states and prospective members being developed very much in the
shadow of the EU regime, and being therefore very much shaped by it.
The European Commission’s initial strategy was not to use legislation to
regulate the sector, since it lacked both the legitimacy and the resources to do this in
the early 1980s (Eliassen et al., 1999). Instead, it chose to encourage harmonisation of
networks and equipment. The Commission also distinguished between three
components of regulation. The network, consisting of the physical cables, switches
and satellites; the services, including voice telephony, value-added network services
and data transmission; and, the manufacturing of customer personal equipment, such
as telephones, modems, fax machines and switching equipment.
The foundation of the EU telecommunications liberalisation was the 1987
Commission Green Paper on Telecommunications (COM(87)290 final 1987) on
which all subsequent reforms were built (Scherer and Bartsch, 1998). The new regime
was laid down in the Open Network Provision (ONP) Interconnection Directive in
1997. The ONP was designed to ensure that the public network operators were
inhibited from frustrating competition by limiting access to the networks, thereby
retaining their dominant market position not by efficiency and innovation but by
anticompetitive practices.
The new regime replaced the concept of the monopoly network operator with
the idea of many active network operators acting in competition. The Commission
reintroduced the idea of rights and duties for these operators, with the rights referring
to unfettered market and network access, and the duties to the obligation to guarantee
universal service. Those companies with Significant Market Power (SMP), defined as
those that hold over 25% of the market share, have a general duty to supply. Based on

8
the SMP proviso the Commission has gone about the task of ensuring that new market
players have the ability to compete. This was achieved by introducing sector-specific
legislation, such as by laying certain obligations on SMPs as regards interconnection
fees and numbering as well as Universal Service obligations.
An underlying requirement for this new regime was that the roles of market
regulation and service provision were split to ensure transparency in the sector
(Eliassen et al., 1999). The idea of separation of regulation and operation had been
forwarded in several countries and resulted in administrative changes in many
member-states around the turn of the decade, whereby operators were either defined
as separate units or later corporatised. Later, the ONP directive demanded the
establishment of a National Regulatory Authority (NRA) independent from the
industry which in most cases has resulted in an independent agency.
The Commission’s approach to the telecommunications sector reveals a
preference for leaving the enforcement of EU policy up to the member-state’s
regulatory agencies (Thatcher, 1996). The regime is therefore far from harmonised, let
alone uniform, albeit the member-states are moving in the same direction.

2.3 LIBERALISATION IN THE “REST” OF THE WORLD

In contrast to EU, telecommunications liberalisation in Asia, Latin America, Africa,
and Eastern Europe has not been orchestrated by a common, general regulatory
regime. The processes of regulatory reform in these countries are mainly based on the
World Trade Organisation (WTO) agreement and the guidelines proffered by the
International Telecommunications Union (ITU) the two main organisations for global
co-ordination.
The ITU’s main functions are to co-ordinate and regulate the use of the
frequency spectrum, and to ensure the interoperability of telecommunications
networks across national borders through standardisation of equipment and systems.
Of more immediate importance for trade agreements and mutual opening of markets
is the WTO. In 1997, the WTO agreement on telecommunications opened the way for
a multilateral framework for freer trade, market opening and competition. Sixty-nine
countries signed this agreement which constitute more than 90 percent of the global
market for telecommunications (Eliassen and Marit, 1999).
Concerns related to establishing a regulatory environment to secure market
access were at the centre of the negotiations on basic telecommunications. The parties

9
agreed on a set of principles relating to competition safeguards, interconnection
guarantees, transparent licensing processes, and the independence of regulators. Even
though the participants managed to reach agreement on many important matters, full
market liberalisation hinges on central principles being enforced by national
legislators and policy makers. By the February 1997 deadline, 57 of the 69
governments submitting schedules for the liberalisation of their telecommunication
sector, had wholly, or with only minor modifications, committed themselves to the
Reference Paper (Eliassen and Marit, 1999). Moreover, the individual countries have
considerable scope in which to interpret the framework. Thus, depending on how
countries handle regulations, they can undermine the market access and national
treatment commitments.
In the Latin American context, privatisations of state telecommunications
enterprises were completed in Chile (1987), Argentina (1990), Mexico (1990), and
Venezuela (1991) (Wellenius & Stern, 1996). In the former U.S.S.R and Eastern Bloc,
the collapse of the communist regimes brought about rapid progress initially in
outlining broad sector reform strategies to deal with the huge service and technology
gaps relative to the rest of Europe. In Asia, reforms were initially slower and of more
limited scope. However, numerous countries, among which Malaysia, Indonesia,
India, China, Philippines, Sri Lanka, Fiji, and Pakistan have initiated a wave of
liberalisations, decentralisations, and privatisations (Wellenius and Stern, 1996).
The most reluctant states to liberalisation have been in Africa. Although some
African countries have explored possible new forms, including joint ventures with
PTTs and introducing some competition, most governments still hesitate to consider
broader reforms and privatisation. In Africa, national security is still a politically
significant issue, compounded by a broader concern about foreign control of key
factors of economic production and distribution.

2.4 TELECOMMUNICATIONS REFORM – A UNIVERSAL MODEL?

Although the policy issues and options faced by governments in reforming the
telecommunications sector are fairly universal, the relative importance of the issues,
the package of sector solutions chosen, and especially the strategy to implement it, are
turning out to be highly country-specific (Bergman et al., 1998; Wellenius and Stern,
1996). Yet, all telecommunications reforms so far mainly involve some degree of
change along each of four directions. These encompass commercialising and

10
separating operation from government; increasing the participation of private
enterprise and capital; containing monopolies, diversifying supply of services, and
developing competition; and shifting government responsibility from ownership and
management to policy and regulation (Wellenius and Stern, 1996).
Furthermore, the design of more liberalised sector structures has required
policymakers to select among a complex array of interrelated parameters. Inter alia,
they have had to decide on which areas to open to competition and which to maintain
for monopoly provision; how many competitors to allow to enter into the competitive
domain; how long to maintain restriction on entry into the monopoly domain or
further entry into a partially opened competitive domain; how to guarantee fair
practices; and what sort of regulatory mechanism to select (Wellenius, 1993).
Governments have also had to weigh the benefits of a privatisation against the costs.
Sales of telecommunications enterprises have resulted in substantial one-time cash
infusions into the treasury, albeit, at the cost of a lost income stream and perhaps also
a convenient instrument for economic and social policy implementation (Vickers and
Yarrow, 1988).
Beyond policy, regulatory structures after reform have become more involved
and complex. This is particularly true when reform also includes privatisation of a
state enterprise. Thenceforth, the regulator needs to address a whole new set of issues.
These include, among others, how to prevent the incumbent operator from abusing its
dominant position; to ensure that a private monopolist does not make monopoly
profits; to ascertain adherence to specific technical standards; and to promote certain
economic and social goals, including universal service (Wellenius and Stern, 1996).
The effects of the liberalisation of telecommunications have been the centre of
interest for a large number of studies. The evolving nature of the industry makes the
assessment of competition effects a nontrivial task. Competition effects are difficult to
segregate and a specific concern would often occupy researchers and policy makers
with a synthesis of inextricable issues. This paper focuses on the effect of
liberalisation on Universal Service in the fixed voice a concern of unprecedented
importance for the development of an economy. The following section provokes a
discussion about Universal Service, its transition and also unfolds a number of
empirical studies that assess the impact of competition on Universal Service.

11
2.5 UNIVERSAL SERVICE IN THE FIXED TELEPHONY
Although the term Universal Service lacks a precise content it usually is meant to
imply that access to fixed lines be generalised and at affordable prices. Universal
Service aims to establish the right of a consumer to connect to the telecommunications
network at a price that does not exclude significant consumer groups (Bergman et al.,
1998). It is therefore a dual component notion and its fulfilment should involve the
measurement of both telephone density and the level of prices (Barros and Seabra,
1999a).
In most cases the incumbent firm is mandated to provide service to meet the
Universal Service Objective (USO). USO is pursued through a variety of approaches.
They mainly concentrate on extending the service to unserved populations in urban
and rural areas without discrimination and making telecommunications services
affordable for low-income citizens (Bergman et al., 1998). The requirement for
telecommunications firms to offer geographically uniform retail tariffs when the cost
of network provision varies in different regions, results in offering retail tariffs that
are distinct to their underlying costs (Armstrong, 2001). Therefore, if liberalisation
eliminates profits from existing profitable markets then the incumbent may be unable
to subsidise the unprofitable operations. For these problems, it is often presumed that
competition and USO do not mix. For the sufficient provision of USO it is necessary
that the incumbent firm exploit the economies of scale present in the industry and
reach the minimum efficient scale of operations (MES) that minimise its long-term
average cost of service.
Recent literature on the study of USO stresses that the notion requires a new
definition. The new definition should address the changing nature of the services
available through the telecoms networks and the ambiguous delineation of what is
considered as “basic service”. Also, it should be resilient to adjust to technological
advancements and facilitate the providing firm to achieve MES. According to Hudson
(2001) USO should have a set of moving targets. The definition of basic service needs
to take into consideration changes in technology and user needs 1 . Thus, goals should
be stated not in terms of a specific technology or service provider but in terms of
functions and capabilities, such as the ability to transmit voice and data. This view can
open to administrators new alternatives for policy design through a multi-network

1
Basic services for USO pertain to someone’s access to a mainline that can support the use of fax and
low speed data transmission.

12
perspective, one that considers the total sum of capabilities enabling a citizen’s access
to a national telecommunications network (Fuentes-Bautista, 2001).
Rapid technological change also dictates that the definitions of basic and
advanced services will change over time (Hudson, 2001). Advanced services are
currently interpreted as Internet access. In the future, it is likely that advanced services
will be redefined, perhaps to include access to new generations of services available
through the Internet or its successors. In rural areas, both terrestrial wireless and
satellites offer greater capacity without the cost of building out fibre and cable
networks. These technological trends have significant implications, since distance is
no longer a barrier to accessing Universal Service; costs of providing services are
declining; and new competitors can offer multiple technological solutions.
Moreover, technological convergence and blurring market barriers have
altered the structure of telecommunications as well as traditional definitions of USO
(Fuentes-Bautista, 2001). The number of fixed phone lines per hundred inhabitants,
the index traditionally used to weigh the concept, considered an “imperfect measure
of Universal Service” (ITU, 1998 p. 20). This index does not reflect the direction and
segmentation of a network’s expansion, nor does it account for other access methods
such as mobile telephony, satellite and digital channels, which are currently used to
access national and global networks.
Different technologies provide different methods of access to a local network.
Minges (1999) refers to “Universal Access”, an antecedent to Universal Service. The
combination of different access technologies contributes and broadens Universal
Service. He illustrates this through the example of the Swiss telephone system.
Between 1995 and 1998, the teledensity in Switzerland decreased from 60.6 to 56
mainlines per 100 people, as a result of residential and commercial customers
migrating to mobile cellular services and ISDN networks. However, the author argues
that the network as an aggregation of people connected to ISDN, the fixed and mobile
networks actually grew more than 20%.
The role of alternative access technologies to fixed voice gains rising
importance in Universal Service fulfilment. Banerjee and Ros (2004a) explicate this
importance in a discussion about the effects of mobile telephony on the fixed voice.
They stress that its effects can be viewed from either an economic or sociological
perspective. On the one hand mobile services can be viewed as essential to daily life;
mobile telephony expands communications possibilities between consumers and

13
businesses; and lowers transactions costs inter alia. On the other hand, mobile
communications play an indispensable role in facilitating connection between
different populations, especially those located in remote areas.
The sociological feature may attribute to mobile telephony a significant
contribution in the fulfilment of USO. This is more prevalent in the case of the
developing world. Some developing countries have badly lagged more developed
countries in the availability of fixed network capacity. The development of mobile
telephony has offered these countries the option to leapfrog existing fixed networks
and gain access to communications that previously could not (Banerjee and Ros,
2004a).
Discussion on a series of empirical studies that examine the impact of
competition on USO follows. The aim is to portray the hitherto attempts made in
economic literature to capture the relationship between competition and USO. A last
section discusses various studies which gauge the effect of alternative technologies
(particularly, mobile telephony) on the fixed network. The purpose is to demonstrate
that there is a pragmatic concern on how USO can be fulfilled, portray the complexity
of the notion, and advocate that USO is not a stationary notion, rather an evolving
one.

2.5.1 Empirical studies on the effect of competition on USO

There is a widely held presumption that competition has a negative effect on
telephone density (Barros and Seabra, 1999b). This presumption derives from the
abolition of cross-subsidies that is expected to come with liberalisation. Namely,
incumbent providers subsidise the network development in rural and less dense areas,
where costs of service are high, at the expense of more profitable areas and services.
Competition destroys cross-subsidies to a great extent resulting to some areas left with
very high costs of provision that culminate in socially unreasonable or unaffordable
prices (Gasmi et al., 2000).
On the other side, competition is believed to drive service prices downwards.
Competitive action drives telecommunications providers to acquire more efficient
technologies, increase capital productivity and minimise their costs of provision
(Armstrong, 2001). Yet, as Barros and Seabra (1999a) note affordability of prices as a
measure of Universal Service does not necessarily imply a lower level of prices in
general as different tariffs may apply to different groups for the fulfilment of USO.

14
Particularly, in the presence of high sensitivity of access demand to prices, the net
effect on teledensity upon cancellation of cross-subsidies may be positive. The effects
of competition on Universal Service thus may vary with the context that liberalisation
takes place. Empirical studies attempting to capture these effects show ambiguous
findings.
Mueller (1993) examined the expansion of telephone access in the United
States before and after the period of liberalisation. The author demonstrated that
during the monopoly the annual rate of growth of telephone connections was 5%,
while in the years following liberalisation this annual rate increased to 40%. Gillet
(1994) examined the impact of competition on USO applying her analysis after the
divestiture of AT&T. Her conclusions were that telephone penetration increased by
2.4% between 1984 and 1992 and growth rates were higher among the lower income
groups and in regions characterised by lower teledensity. An econometric study by
Wallsten (2001) examined the impact of competition, privatisation and regulation in
Africa and Latin America. The author used data on 30 countries from 1984-1997.
Specifically for competition, the study shows that it increases mainline penetration
and reduces prices for local calls.
Barros and Seabra (1999a) used aggregate data on a time-series cross-country
comparison to analyse the effects of liberalisation on USO. There findings were vague
in the sense that there was no definite conclusion as to whether competition is harmful
to Universal Service. In addition, the impact of competition on the average price of
telecommunications, for both residential and business consumers was also
indeterminate, as there was no clear downward pressure on prices from competition.
Another econometric study by Ros (1999) examined the effect of competition and
ownership of the incumbent firm on network expansion and efficiency. The author
used panel data for 110 countries for the period 1986-1995. As regards competition, it
was not possible to relate it with a positive or negative effect on fixed telephony
penetration.
A study by Li and Xu (2004) examined the impact of privatisation and
competition in the telecommunications sector of more than 160 countries using data
for the period 1990-2001. After controlling for privatisation, this paper showed that
competition did not have any detectable effects on fixed line penetration. A joint
effect of competition and privatisation though proved to have a positive leverage on
teledensity. Competition also appeared to increase the real cost of local phone calls,

15
albeit by an amount that was statistically insignificant. Again, complementarity
between privatisation and competition was restraining the rise in the cost of local
phone calls. These effects appeared to take place not in the event year but rather some
time later and the impact was similar in direction and magnitude across countries at
different levels of economic development.

2.5.2 The role of mobile telephony in USO

Mobile service adoption has grown at astonishing rates as the quality of the service
and the performance of mobile providers steadily improved since its inception in the
early 1980s. In 2002, the number of mobile subscribers worldwide surpassed the
number of fixed mainlines for the first time (Banerjee and Ros, 2004a). Yet, the
potential that mobile telephony can play a significant role in USO has been given
minor attention by the economics and telecommunications literature (Hamilton,
2003).
While on average almost one out of three people is a mobile subscriber, there
are major regional differences (ITU, 2006). In 2004, Europe’s mobile penetration rate
stood at 71%, almost twice the penetration rate of North America (43%), and nearly
four times the penetration rate of Asia (19%). Europe had almost eight times the
penetration rate of Africa, where less than one out of ten people subscribe to a mobile
service.
According to a recent EU-commissioned survey, about 12% of EU households
are using only mobile services and do not subscribe to fixed line services. This figure
ranges from 4% in Germany and Sweden to 16% and 29% respectively in France and
in Finland (Taubman and Vagliasindi, 2005). In addition, transition countries
witnessed important trends in the mobile segment of the market. First, mobile
penetration rates have been growing at exponential rates, whereas fixed line
penetration rates have at best stagnated. Second, in many countries, mobile
penetration rates have overturned fixed line penetration rates. This is the case not only
for most of the EU countries but also for countries of South Eastern Europe and
Russia.
It is therefore naive not to realise the significant leverage that mobile
telephony can have on USO. In developed economies, mobile telephony is believed to
function as complementary to fixed line and in developing countries as a substitute
(Hamilton, 2003a; Rodini et al., 2002; Taubman and Vagliasindi, 2005). Yet,

16
empirical research has not managed to provide concrete support on this postulation as
yet (Barros and Cadima, 2000).
Taubman and Vagliasindi (2005) examined the impact of usage of mobile
telephony by businesses on fixed lines penetration rates across a number of Eastern
European countries. Their results provide some evidence of some substitution effects
in place at the country level. However, complementary effects dominate at the
enterprise level. Rodini et al. (2002) note that technically, mobile is a substitute
because users can place and receive voice calls just as they do with fixed service. The
author also presents an alternative view which suggests that fixed and mobile services
are complementary. According to this, mobile service enables calls that were
otherwise impossible and in addition, mobile calls typically originate or terminate on
a fixed line so that mobile usage could stimulate fixed line usage.
In an empirical study in the Korean market, Sung and Lee (2002) examined
the relationship between mobile and fixed voice telephony. Their results showed that
a 1% increase in the number of mobile telephones results in a reduction of 0.10-0.18%
in new fixed connections and a 0.14-0.22% increase in fixed disconnections, which
indicates that mobile telephone services are becoming a substitute for fixed
telephones. Moreover, Hamilton (2003a) examined the kind of the relationship in a
number of countries in Africa. The results suggested that mobile telephones act as a
competitive force encouraging fixed-line providers to improve access. It was also
concluded that it is possible that mobile and main lines are sometimes substitutes and
at other times complements in consumption, even when fixed line access is low. The
impact of mobile telephony on the competitiveness of the telecommunications sector
was assessed by Jha and Majumdar (1999) in a set of OECD countries. Their results
highlighted the importance of mobile telephony on the increase of competitiveness
and on the efficiency of the sector.
Gruber and Verboven (2001) in a study based on data for 140 countries
during the period 1981-1997 estimated a diffusion model for mobile services. Their
research findings advocate the existence of a form of complementarity between fixed
and mobile network services. They defined this form of complementarity as
“technological complementarity” referring to the parallel growth of mobile and fixed
telephony for technological reasons. This is in contrast to “economic
complementarity” which is portrayed by parallel movements in demand in response to
mutual relative price changes. A general conclusion of this study is that mobile

17
services are well suited to provide telecommunications access in inefficient fixed line
markets.
The distinction between technological complementarity/substitution and
economic complementarity/substitution was also raised in a paper by Banerjee and
Ros (2004b). The authors performed a cluster analysis of a panel of 61 countries
examining the relationship between mobile and fixed telephony. They concluded that
while mobile and fixed telephony have generally developed apace in the more affluent
countries, relatively less affluent countries have favoured the leapfrogging of fixed by
mobile telephony – what they name technological substitution. As they note, this form
of substitution does not occur in response to relative prices; rather, when local
conditions favour the development of one form of telephony over the other. For
instance, bad experiences with existing fixed voice services have prompted consumers
to look increasingly in the direction of mobile telephony. Banerjee and Ros (2004b)
also stress that technological substitution has been the more powerful underlying
source of movement in mobile telephony.
An econometric study on the Portuguese telecommunications market by
Barros and Cadima (2000) examined the bidirectional relationship between fixed and
mobile telephony. The study used data for the period 1981-1999 and concluded that
only a unidirectional relationship exists between the two technologies; namely, mobile
telephony development has an impact on the development of fixed voice and not vice
versa. The relationship exhibits a negative effect of mobile telephony on the fixed
voice. However, there is still a seemingly positive correlation between the two
technologies as long as the two penetration rates are still rising. A claim favouring
fixed voice upon the introduction of mobile is that network externalities may increase
the overall number of calls and by this to benefit also the fixed voice providers
(Barros and Cadima, 2000).
Existing studies on the effects of the liberalisation of telecommunications on
Universal Service appear to disregard the underlying importance of the size of the
economy where liberalisation takes place. Instead, the major distinctions drawn are
between high and low income economies and between developed and developing
countries. These merely generic categorisations have not helped in producing concrete
conclusions in regard to the effects of liberalisation. Whilst the literature on small
economies postulates that regulatory and competition policy should differ with size
the telecommunications literature has placed nominal emphasis upon it. The following

18
section unfolds a review of the literature on small economies and stimulates research
towards a direction that encompasses the notion of smallness. There is high potential
that the consideration of size might cast light on the impact of competition on
Universal Service.

2.6 LIBERALISATION AND THE SMALLNESS OF ECONOMY

2.6.1 Small economies – an overview
The economic literature has generated various classifications for countries and
jurisdictions in order to facilitate the examination of various phenomena in specific
contexts. The utility of a classification of small economies was examined by Selwyn
(1980) who suggested that the level of the utility should be gauged with respect to the
classification’s analytical, predictive, and normative usefulness. Respectively, the
category of small economies should help in understanding specific country problems;
it should be useful in forecasting likely social or economic trends in a country; and it
should shed light on appropriate policies for particular countries.
The definition of small economy is arbitrary in the sense that there is no
specific line of distinction between a small from a large one. The literature on small
economies attempts to conceptualise smallness in terms of measurable variables;
notably, population, economic activity and geographic area, whether combined or
individually (Armstrong and Read, 1998). Population is the most commonly used size
measure mainly because it is easily accessible as well as proving a crude proxy for
two economic variables; the size of the domestic market and the local labour force
(Armstrong and Read, 1995b). Earlier studies on small economies used a threshold of
10 and 15 million. Over time this threshold started attenuating as smaller nations
emerged and in the 70’s international institutions adopted a one million threshold
(Armstrong and Read, 1998). The latter authors use an upper limit of three million for
their discussion on smallness.
Ambiguity in the definition of smallness based solely on population becomes
more obvious in the examples of countries used in different studies 2 . For instance,
Briguglio and Buttigieg (2004) discuss the case of Malta; Spiller and Cardilli (1997)
study the notion of smallness in Australia, Chile, Guatemala, and New Zealand;
Stewart (2004) uses Bahamas, Belize, Jamaica, St Lucia, St Vincent, and Trinidad and

2
It is noteworthy that a review of the literature on small economies shows that the study of size has
merely concentrated on the manufacturing sector and on tradable goods.

19
Tobago; and Gal (2003b) makes a reference to the Faroe Islands, Jersey, Malta,
Cyprus, New Zealand, Israel, and Australia. Evidently, the size of these countries
varies from a few thousands to several millions which imply that solely population is
not a representative indicator of smallness. It is therefore suggested that smallness be
measured by a collection of variables that involves a combination of measures of size
such as population dispersion, land area, GDP, and degree of openness to trade
(Briguglio and Buttigieg, 2004; Gal, 2003a; Armstrong and Read, 1998).
In terms of characteristics of a small economy there appears a general consent
among various studies (Armstrong and Read, 1998; Briguglio and Buttigieg, 2004;
Gal, 2003a; Stewart, 2004). A small economy’s capacity to support many competitors
is small and it depicts high concentration in many of its industries. Concentration of
an industry is determined by the size of firms operating in it and the main driving
factor to concentration is the size of minimum efficient scale of production.
Furthermore, small economies are characterised by high entry barriers. The main
entry barrier is created by scale economies 3 , by the need to produce at levels that cater
to a large portion of demand in order to achieve minimum costs. This is also
associated with another major issue, the problem of sub-optimal levels of operation.
Small economies are likely to have small markets which in turn limit
competition possibilities, due to the ease of market dominance by firms. Smallness
also renders the exploitation of advantages deriving from economies of scale difficult
manifested by higher prices. In the context of utilities, small economies tend to be
characterised by natural monopolies. To add, state aid is evitable in small economies
aiming to sustain comparative advantage in the respective market. Armstrong and
Read (1995a) identify a number of advantages arising from smallness which
nevertheless are expected to be outweighed by the disadvantages. These advantages
are primarily intangible and therefore impossible to quantify. For instance, the authors
assert that small economies have greater social homogeneity and cohesion, greater
social flexibility and openness to change.
These salient characteristics have important policy implications as they
require small economies to devise appropriate endogenous policies that offset at least

3
The postulation that economies of scale consists a barrier to entry per se has accepted considerable
criticism (see Posner, R. A. (1976) Antitrust law : an economic perspective, Chicago, University of
Chicago Press. and Vernon, M. J., Viscusi, K. and Harrington, J. (2000) Economics of Regulation and
Antitrust, MIT Press. for a discussion). It is however considered an important element for the
sustainment of competition.

20
some of the adverse effects of their small size (Gal, 2003b). However, most small
economies do not scrutinise their special economic traits in designing and applying
their antitrust laws. Rather, they adopt or rely on the statutes and established case law
of large economies, mostly of the European Union. This approach has many
recognisable advantages such as a ready basis for the law, a large body of
comprehensive case law and commentary, and network externalities. Adopting the
competition laws of larger jurisdictions is also sometimes predicated in the existence
of the hegemonic power of a large jurisdiction with the ability to impress its will, and
its competition policy, on smaller and weaker jurisdictions. The main pitfall of such
an approach is that insufficient weight is given to the unique characteristics of small
economies (Kaminarides et al., 1989).

2.6.2 Smallness and Efficiency

Armstrong and Read (1995a) stress that a concrete economic definition of small size
should address the influence of sub-optimality upon the structure of domestic output.
This methodological issue involves the effect of smallness on economies of scale and
scope, efficiency, and competitiveness. The measure of sub-optimality is manifested
by the minimum efficient scale of operations which determine the level of economic
activity that can effectively be pursued within any country.
According to Gal (2003a) smallness affects the three main indicators of social
welfare which comprise of allocative, productive, and dynamic efficiency. Whereas
competition policy in small economies should be based on the support and promotion
of efficiency, policy makers, agencies and courts often overlook considerations of
economic size. This is further exacerbated by the recent trend towards the
harmonisation and convergence of competition policies on a universal scale.
A major handicap resulting from small size is the need to firms in many
markets to produce at a level that caters to a large portion of demand to achieve
economies of scale, since they are the main determinant of minimum efficient scale
(MES) of operation (Scherer et al., 1975). Scope economies may also be a target at
the firm level through vertical integration of two or more functions down the
production and distribution chain in one firm (Williamson, 1975).
While in large economies MES tend to be too small relative to most markets
to warrant high levels of concentration, in small economies the number of MES that
firms in many markets can support is much smaller, and accordingly industrial

21
concentration is likely to be much higher. This appears to be a major explanation to
the reduced number of sustainable competitors in small economies’ industries since
the cost of inadequate production urges a firm to exit a market (Jalan, 1982). Simply
put, the smaller the economy, the higher the level of concentration as most of its
markets cannot accommodate many viable competitors, ceteris paribus (Dasgupta and
Stiglitz, 1988).
Bergman et al. (1998) stress that competition policy encounters an array of
conflicting priorities, i.e. economic vs. social objectives. In the context of small
economies, Gal (2003b) suggests that social goals should be given little or no
independent weight in formulating competition policy. Small economies should strive
to achieve economic efficiency as their main goal because they cannot afford a
competition policy that is prepared to sacrifice economic efficiency for broader policy
objectives. However, should this proposition be applied indiscriminately in small
economies it would eventually lead to social implosion (Stewart, 2004).
Structural remedies, such as the dissolution of monopolistic or oligopolistic
structures by reducing concentration, may on one level help reduce the feasibility of
market power, collusion and interdependent behaviour. On another level, they usually
involve a trade-off between enhancing competition and exploiting potential cost
efficiencies that flow from MES of production, when applied in small economies
(Gal, 2003b). Posner (1976) examined the practice of breaking up large firms as a
panacea to reducing concentration levels in any industry. The author’s critique of the
hitherto competition policy is that divestiture decisions are incorrectly justified as a
means for alleviating the problem of economies of scale considered to be a barrier of
entry. Economies of scale do not create a barrier to entry (Vernon et al., 2000). They
merely dictate the level of output that the new entrant must achieve in order to
minimise costs. Moreover, the costs of a de-concentration policy probably outweigh
the benefits which are conjectural (Posner, 1976).
Moreover, Pitelis (2003) stresses that if increasing firm size, and thus
concentration is the result of transaction cost reduction such efficiency gains should
be taken into account by regulatory bodies. For instance, a vertical integration that is
motivated by efficiency, not market power motives. Such considerations suggest a
more lenient attitude towards large firms.
The implication that industrial concentration has on competition policy is that
small economies cannot afford to transplant simplistic competition policies applied in

22
large economies, for which concentration is rarely supported. Yet, concentration
measures alone are not a good guide for competition policy for small economies.
Rather, measures of levels of concentration should be balanced with productive
efficiency considerations dictated by market size.

2.6.3 Smallness and (Natural) Monopoly

A contrasting recommendation to legacy competition policy is that attainment or
maintenance of a monopoly position per se should not be prohibited (Lien and Peng,
2001). Accordingly, the regulator should distinguish between monopolies based on
superior efficiency and those that are not. This contention pro the incumbent
monopolist extends further to the issue of structural separation of the dominant firm
(Gal, 2003b). If the dominant firm has to be divided into two or more viable firms,
this should be without jeopardising scale or scope economies. When a small economy
can support only one firm, divestiture may create a concentrated market structure –
duopoly or oligopoly – that is also prone to monopolistic exploitation.
Posner (1976) raised the question of whether it is socially desirable to promote
small business which might turn out to be pursued at the expense of the consumer.
Competition policy is an inappropriate method of trying to promote the interests of
small business as a whole. Instead, a passive role for competition policy is preferred.
Monopoly, by driving a wedge between the prices and the costs of the larger firms in
the market enables smaller firms to survive, even if their costs are higher than those of
larger firms (Posner, 1976).
According to Vernon et al. (2000) a small economy should adopt a test, based
on economic analysis, which ensures that the monopolist’s conduct actually reduces
welfare before it is prohibited. Given the difficulties involved in proving in each case
that a certain type of conduct has amounted to abuse dominance, several small
economies have included in their competition laws nonexclusive lists of practices
from large economies that are presumed to constitute abuses of power if engaged in
by a dominant firm.
In specific industries there are situations where a single firm may have much
lower costs of provision of a product than more than one firm providing the same
product. The costs of the single firm may be that low that the profit-maximising
monopoly price is actually below the competitive price. In such market conditions,
Posner (1976) suggests that society’s economic welfare would be greater if the

23
monopoly were permitted. The author explains that since in an economic analysis
competition is justified for its capacity to promote efficiency it would seem that
whenever monopoly would increase efficiency it should be encouraged. Yet, it is
extremely difficult to measure efficiency consequences of a challenged practice. This
becomes more vital in the case of small economies where the impact of natural
monopoly might be much stronger than in large economies especially in leveraging
other vertically interconnected markets (Gal, 2003a).
Condemnation of natural monopolies per se is problematic given the nature of
available remedies based on Gal (2003b). The basic remedy to restore the market to
its position absent the violation is break-up. But break-up is not an efficient solution
in the case of natural monopolies, since it would reduce productive efficiency and
needlessly sacrifice economies of scale. Instead, the joint ownership of natural
monopoly’s facilities by all competing downstream or upstream firms and the sharing
of dividends in accordance with use could be an alternative. Under this structural
solution, production or service is unified and centralised to achieve scale economies,
but ownership is decentralised among multiple owners who compete with one
another.
Discussion on the effects of liberalisation and the role of alternative
technologies on Universal Service evinces the volume of theoretical and empirical
research dedicated to the study of the notion. It also demonstrates that research has not
been exhaustive as there are unfilled gaps and emergent questions. Whilst research
has taken place primarily in large countries, whether developed or developing, this
paper raises the issue of smallness as an important factor that leverages the effects of
liberalisation and alternative technologies on Universal Service. It urges for research
beyond the conventional context of studies in large countries, hoping to enlighten and
contribute to answering pertinent questions. This is attempted through an econometric
approach which aims to capture the various relationships with Universal Service in
the context of small economies. The research methodology and econometric analysis
are unfolded in the following sections.

CHAPTER 3 – RESEARCH METHODOLOGY

The econometric model aims to assess the impact of competition and alternative
technologies on Universal Service in the context of small economies. It is a multiple
regression model that employs panel data estimation techniques and a Fixed Effects

24
transformation. Following the duality in the definition of Universal Service the model
tests for the effects both on network expansion and price levels.
As regards the network variable Gutierrez and Berg (2000) used the number of
main lines per 100 inhabitants whereas Jha and Majumdar (1999)) used the actual
number of mainlines. In addition, Wallsten (2001) used the number of payphones and
connection capacity of the network. A further suggestion by Hamilton (2003a) is that
the total number of installed mainlines plus the number of mainlines remaining on the
waiting list be used instead. This indicator is meant to capture the demand side of
fixed voice rather than merely the supply side.
Longitudinal price data are characterised by Wallsten (2001) as the most
problematic to analyse. Notwithstanding their accuracy, they may be incomparable
across countries and they often encompass extensive subsidies. Nonetheless, Wallsten
(2001) used the price of 3 minute local calls; Ros (1999) and Ros and Banerjee (2000)
used data for the cost of initial connection to the network and monthly subscriptions
for both household and business customers in addition to the price of a 3 minute local
call; and Barros and Seabra (1999a) used business and residential tariffs in the form of
call baskets.
The set of independent variables of the model encompasses notions for the
measurement of mobile telephony development; Internet development; the
liberalisation status of an economy; and the size of an economy. Mobile telephony
development is usually measured by the number of cellular subscribers in a country or
per capita rate (Gutierrez and Berg, 2000; Hamilton, 2003b). Pertaining to the
development of the Internet the ITU uses the number of Internet subscribers per 100
inhabitants, international bandwidth in Mbits per capita, ISDN and DSL subscribers
as a percentage of total internet subscribers, inter alia (ITU, 2006).
The extent of telecommunications liberalisation has been viewed in various
perspectives. For example, Jha and Majumdar (1999) employed an index taken from
the 1993 OECD Telecommunications Outlook which ranks countries according to the
degree of infrastructure liberalisation. Wallsten (2001) used the number of mobile
operators as a proxy for competition in the market. Instead, Barros and Seabra (1999a)
employed a dummy variable which drew a distinction between monopoly and non-
monopoly markets.
The last dependent variable is the size of an economy. Measures of the size of
economies have provided unclear distinction between large and small economies.

25
Selwyn (1980) noted that if we are concerned with constraints resulting from narrow
range of resources, we may identify size with physical area. If we are concerned with
manpower limitations of the small clientele for public and other services, we will
measure size in terms of population. The rationale is that different studies employ
different measures for smallness.
For this study an index for smallness suggested by Jalan (1982) is adopted
which combines population, income and geographical measures. The index is
calculated using the following formula:

I = 100/3 (P/Pmax +A/Amax + Y/Ymax),

where:
I, is the country size index for an individual country;
P, A and Y are population, arable area and GDP of each country respectively;
Pmax, Amax and Ymax represent the highest values of population, arable area and
GDP respectively.
Applying this formula on 214 countries and jurisdictions from data from the
United Nations’ Human Development Report of 2005 produced the country size
index. For many countries, the index value is very small. Particularly, the mean and
the median values of the index for the whole sample are only 2,81 and 0,53 (on a scale
from 0 to 100). There is also a discontinuity at the end of the series of country size
indices and there are three countries (USA, China and India) whose index values are
exceptionally high. In view of the skew distribution and the predominance of small
values, a convenient way of dividing countries into categories of small and not so
small may be to use the median value of the index as a cut-off point (Jalan, 1982). A
set of 112 countries and jurisdictions fall within the range of the cut-off point which
have less than 9.2 million population (Guinea), less than USD$86.17 4 billion GDP
(Singapore), and less than 19.3 thousand km² of arable land (Central African
Republic). An illustrative list of small economies is given in Appendix I.

4
GDP is based on constant prices of 2000.

26
3.1. DATA AND VARIABLE DESCRIPTION
Telecommunications related data were primarily obtained from the ITU’s
telecommunications indicators database. These data have been collected by the ITU
via questionnaires completed by national incumbents and regulatory authorities. They
consist of time-series, cross-sectional data for 214 5 countries and jurisdictions for a
number of telecommunications and other indicators. 112 of these countries fall into
the category of small economies and data for the period 1980-2004 were considered
“appropriate”, in terms of completeness for the purpose of this study. Economic,
geographical, and demographic data were obtained from the United Nations’ Human
Development Report of 2005 and the World Bank’s World Development Indicators.
As regards the econometric model, the two dependent variables depict the
fixed network expansion and prices. The former is teledensity, expressed as the
number of fixed lines per 100 inhabitants. Whereas the assessment of the demand side
of fixed voice – the joint indicator of both teledensity and unmet mainlines per 100
inhabitants - would be more suitable, lack of data limit the analysis to the met
demand. Yet, teledensity is considered a sufficient indicator for the measurement of
the penetration level of fixed voice.
The number of international outgoing calls is used as a proxy for the
measurement of the level of prices. Limitations in using price data identified by
previous studies (i.e. Banerjee and Ros, 2004a; Wallsten, 2001) called for the search
for an alternative variable. The rationale behind the selection of this variable is that a
decrease on the level of fixed voice prices would be expected to cause a decrease in
the volume of international calls. In addition, since consumers make international
calls through a fixed line, changes on this variable tend to be associated with price
changes for the fixed voice.
The independent variables used consist of variables depicting the status of
competition, the development of mobile telephony, the development of the Internet,
and the size of the economy. In order to assess the distinct impact of competition in all
three telecommunications technologies three dummy variables are employed which
demonstrate whether an economy has liberalised any of its fixed voice, mobile
telephony and the Internet market. Information on the opening of the respective
markets was obtained from the ITU’s World Telecommunication Regulatory

5
There were countries which lacked a considerable amount of data for different variables used in the
models. Therefore, different models involve a different number of countries.

27
Database. These variables allow distinguishing between the effects of competition in
different technologies, a factor that has not been taken into consideration in previous
empirical studies.
Development of mobile telephony is expressed by the number of mobile
subscribers per 100 inhabitants. Internet development is gauged by the number of
ISDN lines per capita. The latter variable was preferred to the number of Internet
subscribers per capita because of data availability and the fact that it is a better proxy
to Internet technology development. 6 Lagged variables are also used in the models in
order to address the question whether the impact of alternative technologies represents
a permanent change in teledensity or prices in the event year.
Regarding the variable for smallness, the methodology employed above
produces a composite index which draws a distinction between small and large
economies. This methodology also specifies the maximum population, GDP, and
arable area that describe a small economy. From the composite index, a dummy
variable is created which distinguishes between small economies and large
economies. Three additional variables are developed which assess the individual
effects of population, GDP and arable area. Yet, instead of plainly using actual
population, GDP and arable area values these three variables are the result of the
interaction between six other variables. Examination of the individual effects of
population, GDP and arable area is essential. However, what is of greater interest is
the effect that these variables have as their values fluctuate between the two edges that
small economies can take.
To be explicit, a first set of three dummy variables is created that
distinguishes between economies that have population less than 9.2 million; GDP less
than 86.17 billion USD; and arable area less than 19.3 thousand km². These variables
are named Pop1it, GDP1it and Arable1it. A second set of three variables is created
which takes values that derive from the following equations:

Pop2it = (population_max – populationit) / populationit

Arable2it = (arable_max - arableit) / arableit
GDP2it = (GDP_max - GDPit) / GDPit

6
A possible limitation of this variable is that it may tend to have a positive relationship with teledensity
since a subscription to the ISDN Internet network requires the installation of an additional line to the
fixed network.

28
Where:
population_max, arable_max, GDP_max, are the maximum values that small
economies are identified with;
populationit, arableit, and GDPit are the respective values for country i at time t.
i: 1,2,3,…,214
t: 1980, 1981,…,2004

The variables included in the model consist of the interaction variables

between the two sets of variables; namely, Pop1it x Pop2it; GDP1it x GDP2it; Arable1it
x Arable2it. The logic behind the derivation of these variables is twofold. First, the
variables concentrate on the distinct effect of each of the three size parameters and
consider only the group of small economies; that is, for large economies the variables
take a zero value. Second, the higher the value of the original variable the lower is the
value for the interaction variable. In this manner, the interaction variable assesses the
advantage or disadvantage created if a small economy were larger as long as it still is
within the bounds of small economies. Respective variables are created for large
economies.
A number of control variables are included in the model. Namely, these are
GDP per capita; the percentage of population living in urban areas; investment in
telecommunications as a percentage of GDP; and the logarithm of the consumer price
index. Economic variables are converted to real values of 2000. A time trend variable
is included as an indicator of temporal effects. Li and Xu (2004) in an econometric
study use such a variable to capture effects of technological changes. In support of
capturing time effects, yearly dummies are also added in order to allow modelling
possible idiosyncratic temporal effects much more intelligently and control for
unobserved time-varying effects. Table 1 describes all variables used in the regression
models.

29
Table 1: Description of variables used in the regression models.
Variable Description
Dependent Variables
teledensity Teledensity measured as the number of fixed mainlines per 100 inhabitants
logcalls The logarithm of the volume of outgoing international calls

Independent Variables
lag_real_gdp_cap Real GDP per capita in USD of 2000
t Time trend variable
urban_cap Percentage of urban population
inv_tel_per_gdp Investment in telecommunications as a percentage of GDP in USD in real prices of 2000
mobile Mobidensity measured as the number of mobile telephony subscribers per 100 inhabitants
lagmobile Mobile subscribers per 100 inhabitants lagged one year
isdn_cap ISDN channels per capita
lagisdn_cap ISDN channels per capita lagged one year
logcpi2000 Logarithm of Consumer Price Index - 2000 is the base year
pop_small Population interaction variable for small economies
arable_small Arable area interaction variable for small economies
gdp_small GDP interaction variable for small economies
gdp_large Population interaction variable for large economies
pop_large Arable area interaction variable for large economies
arable_large GDP interaction variable for large economies
smallness Dummy variable that equals 1 if the economy is small and 0 otherwise
lib_fixed Dummy variable that equals 1 if the fixed voice market has more than one provider
lib_mobile Dummy variable that equals 1 if the mobile telephony market has more than one provider
lib_internet Dummy variable that equals 1 if the Internet market has more than one provider
lib_fixed_small Interaction variable between smallness and lib_fixed
lib_mobile_small Interaction variable between smallness and lib_mobile
lib_internet_small Interaction variable between smallness and lib_internet
smallness_year(t) A vector of 24 dummy variables consisting of the interaction between yearly dummies and
smallness dummy. t represents 1981-2004. t for 1980 is omitted from the model as it is the base
year

Summary statistics are exhibited in Table 2. The mean values of the indicators
illustrate that small economies compared to large economies have more arable area;
have lower consumer prices; invest more in telecommunications as a percentage to
GDP; their population is somewhat more concentrated in cities; and they are less
affluent. They also encounter higher volumes of outgoing international calls and thus
lower prices in fixed telephony than large economies; yet, they face lower levels of
penetration in both fixed and mobile telephony and have fewer ISDN channels per
capita. Whereas these statistics do not represent causal relationships or yearly trends
with size they give a crude approximation of the general situation between small and
large economies.

30
Table 2: Summary statistics for basic indicators for the year of 2004

Mean Std. Dev. Min Max

Variable/Economies Small Large Small Large Small Large Small Large
Arable to total area 10.86 16.70 11.88 13.17 0.00* 1.03 54.45 55.69
CPI(2000)† 114.91 123.0308 16.89 35.93275 93.94 80.46 196.89 294.9
Investment in telecoms
as percentage to GDP 0.00000152 0.00000101 0.00000156 0.00000180 0.00000001 0.00000001 0.00000744 0.00001660
ISDN channels per
capita 4.59 5.82 9.74 9.07 0.00*** 0.00**** 41.82 39.53
Mobidensity 30.40 33.37 29.91 35.20 0.00** 0.12 119.38 114.14
Outgoing international
calls per capita 0.0001432 0.0000475 0.0003258 0.0000919 0.0000005 0.0000002 0.0022311 0.0006392
Population 2338722 62700000 2492606 172000000 10396 3443341 9201759 1290000000
Real GDP per capita 6478.34 7144.43 11011.05 10686.01 94.92 97.08 72936.65 38131.58
Teledensity 20.68 21.11 20.78 21.83 0.20 0.15 99.37 76.57
Urban pop. to total 21.17 21.04 10.68 9.41 3.07 0.55 60.41 39.55
Notes:
* Gibraltar Greenland, Jersey, Macau, Nauru, Tuvalu have 0 arable area
** Eritrea and Tuvalu have 0% in Mobidensity
***Belize, Eritrea, Gabon, Georgia, Micronesia (Fed. States of) have 0 ISDN channels per capita
****Benin, Cambodia, Cuba have 0 ISDN channels per capita
†The CPI(2000)’s statistics are calculated after excluding Angola and the Democratic Republic of Congo which exhibit extremely high values.

Table 3 presents a timeline of liberalisation in the three distinct markets of

fixed and mobile telephony and the Internet by drawing a distinction between small
and large economies. Generally, liberalisation in large economies has taken place
earlier than small economies and large economies appear to be more apt to liberalise
their telecommunications industry. Until 2004, almost 70% of large economies have
introduced competition in their fixed voice, 92% in mobile and 92% in the Internet,
whereas, small economies are more reluctant by accumulating 41% in the fixed voice,
64% in the mobile, and 72% in the Internet. For both categories of economies, the
fixed telephony market has the lowest percentage of liberalisation whereas the
Internet has the highest.

31
Table 3: Chronological progress in the liberalisation of telecommunications in the
Fixed and Mobile telephony and the Internet
Liberalised Large Economies In: Liberalised Small Economies In:
Total # Total #
Large Fixed Mobile Internet Small Fixed Mobile Internet
Year Economies # % # % # % Economies # % # % # %
2004 98 68 69.39 90 91.84 90 91.84 94 38 40.43 60 63.83 67 71.28
2003 98 66 67.35 90 91.84 90 91.84 94 38 40.43 60 63.83 67 71.28
2002 98 64 65.31 90 91.84 90 91.84 94 28 29.79 56 59.57 63 67.02
2001 98 62 63.27 87 88.78 87 88.78 94 24 25.53 51 54.26 60 63.83
2000 98 58 59.18 84 85.71 86 87.76 94 20 21.28 48 51.06 59 62.77
1999 98 54 55.10 75 76.53 77 78.57 94 15 15.96 38 40.43 47 50.00
1998 98 35 35.71 67 68.37 73 74.49 95 7 7.37 30 31.58 39 41.05
1997 98 21 21.43 54 55.10 54 55.10 95 7 7.37 25 26.32 25 26.32
1996 98 19 19.39 40 40.82 40 40.82 95 6 6.32 18 18.95 17 17.89
1995 98 14 14.29 34 34.69 36 36.73 95 5 5.26 11 11.58 11 11.58
1994 98 14 14.29 6 6.12 7 7.14 95 5 5.26 1 1.05 1 1.05
1993 98 14 14.29 5 5.10 6 6.12 95 5 5.26 1 1.05 1 1.05
1992 98 14 14.29 5 5.10 6 6.12 95 5 5.26 1 1.05 1 1.05
1991 98 14 14.29 5 5.10 6 6.12 95 5 5.26 1 1.05 1 1.05
1990 98 14 14.29 5 5.10 6 6.12 95 5 5.26 1 1.05 1 1.05
Source: Data obtained from ITU’s World Telecommunication Regulatory Database

3.2 THE EFFECT ON FIXED TELEDENSITY

For the assessment of the effects on teledensity two regression models employing a
Fixed Effect transformation are assessed. The first one concentrates on the effect that
competition has on teledensity in small economies. The second model also
incorporates the effect of alternative technologies on teledensity. Beginning with the
first model, it has the following form:
yit = ai + βxit + γdit +ui + εit,
where i( = 1,2,…,141) is the subscript for the cross-sectional dimension (country) and
t( = 1,2,…,25) is the subscript for the time-series dimension (year). yit is a T x 1
vector representing mainlines per 100 inhabitants; ai is a 1 x 1 scalar constant; β is a
series of coefficients corresponding to xit series of exogenous variables, γ is a series of
coefficients corresponding to dit dummy variables; ui is a T x 1 vector of the effects of
omitted time-invariant country-specific variables; and εit is a random disturbance
variable assumed to be distributed identically and independently with zero mean and
finite, constant variance. Treating the country-specific effects as fixed parameters
leads to the Fixed Effects form of panel data model estimation, while treating them as
a random variable with known distribution leads to the Random Effects form of panel
data model. In order to identify the most appropriate form of model as regards the

32
error structure, the two models were subjected to the Hausman test, the best-known
test for discriminating between the two competing models (Baltagi, 2005). The test
produced a chi² statistic of 161.96, suggesting the use of a Fixed Effects model.
A limitation of the Fixed Effects transformation is that it inherently suppresses
time-invariant effects. Therefore, it would be impossible to include the smallness
variable per se. Instead, smallness is included in the model by interacting it with year
dummies. The combined variable captures the time trend in fixed teledensity by also
considering the effect of smallness. T-statistics are estimated using robust standard
errors that account for time-series autocorrelation within each country. Furthermore,
according to Baltagi (2005) robust estimation of the standard errors often results in the
greatest precision resulting in the smallest standard errors. Table 4 illustrates the
estimates for the generic model as they are produced by STATA 9 7 . Although the full
panel comprises 5350 observations (= 214 countries × 25 years), only a fraction is
ultimately usable. That is a result of STATA omitting incomplete data rows. This
results in the creation of an unbalanced panel which nevertheless is handled by the
software’s estimation techniques.
The within-R² signifies a generally good fit of the specified model to the data
explaining 67% of the variance of fixed teledensity. The F-statistic accounts for the
joint significance of the coefficients and proves high enough to reject the null
hypothesis of insignificance in all coefficients. The value of 0.2734 at the lowest
section of the Competition model represents the correlation between the idiosyncratic
error (eit) and the regressors. The low value gives support to the assumption of strict
exogeneity of the regressors and also advocates the use of the Fixed Effects model.
Under the assumption of strict exogeneity the Fixed Effects estimator is unbiased
(Wooldridge, 2002).
Regarding the regressors’ coefficients, they are all statistically significant at
conventional levels apart from investment in telecommunications as a percentage of
GDP and eight out of twenty four smallness_year(t) variables. Lagged real GDP per
capita, urban population as a percentage of total population and the logarithm of the
Consumer Price Index are positively related with teledensity. The former result

7
Stata reports an “intercept” which can cause confusion in light of the earlier claim that the fixed effect
transformation eliminates all time-constant variables, including an overall intercept. Reporting an
overall intercept in FE estimation arises from viewing the ai as parameter to estimate. Typically, the
intercept reported is the average across i of the estimate of ai. An interpretation thus of the intercept
would be unnecessary.

33
suggests that an increase in the economy’s income results to an increase in the supply
of mainlines, however a 1% increase in teledensity would require almost an 835USD
increase in real GDP per capita. This is an extraordinary amount considering that
there are developing economies with much lower GDP per capita than 835USD.
Density of urban population is associated with an increase in teledensity. This is an
expected outcome as population concentration in the cities facilitates the reduction of
average cost of installation for additional lines, favouring network expansion. The
positive effect portrayed in the logcpi2000’s coefficient suggests a response on the
supply side of the fixed voice. Namely, an increase in the general level of prices
would induce an increase in the supply side of products and services; hence, the
installation of further landlines.
As regards the effect of smallness on teledensity, this is captured by
gdp_small and pop_small interaction variables and smallness_year(t) dummies 8 .
gdp_small has a positive effect on teledensity. That is, a unit increase in gdp_small
would result in a 0.03% increase in teledensity. According to the definition of
gdp_small when actual real GDP of a country increases gdp_small’s values attenuate.
This suggests that for small economies the higher the real GDP the lower teledensity
becomes. This contradicts with earlier finding that generic real GDP per capita is
associated with an increase in teledensity but probably this is an effect of the
standardisation of dividing GDP with population. A graphical demonstration however
of actual real GDP and gdp_small illustrates that the effect that the former has on
teledensity is minimal. Even with enormous increases in GDP the effect appears to be
negligible (see Appendix 2).
pop_small has a negative relationship with teledensity. Similarly, according to
the definition of pop_small when actual population of a country increases the
variable’s value attenuates. This suggests that small economies with population
approaching the lower bounds of large economies are expected to have higher
teledensity than economies with less population. Population as evidence of the size of
market demand illustrates the potential for minimising the average cost of service.
The smallness_year(t) dummy variables exhibit positive relationship with
telendensity. Since the variable constitutes an interaction of time trend with smallness
it suggests that in general, small economies have had higher levels of teledensity than

8
arable_small and arable_large are dropped from the model because of multicollinearity

34
large economies. The coefficients of smallness_year(t) remain stable across time
averaging at around 3.5. This finding advocates the existence of different dynamics in
the industry of telecommunications between small and large economies.
The respective size variables for large economies are pop_large and
gdp_large. pop_large has a negative effect on teledensity. Similar to pop_small
suggests that large economies with increasing population are expected to have higher
teledensity. Yet, the coefficient for large economies (-0.01) is much smaller than the
one for small economies (-0.46) implying that the underlying effect of population size
on teledensity is more critical for small economies than large economies. On the other
hand gdp_large affects positively teledensity implying a negative effect of the actual
GDP in large economies. Analogous to gdp_small, gdp_large does not depict a strong
leverage on teledensity in the context of large economies and this is illustrated by the
similarity in the respective coefficients (0.03 and 0.036).
Pertaining to the effects of competition in small economies, it appears that
liberalisation in different markets is associated with a distinct effect on teledensity.
lib_fixed_small 9 the variable that assesses the liberalisation in small economies in the
fixed market has a strong negative effect on the mainlines network, since on average,
small economies which have opened their fixed markets encounter a decline of 2.85%
in teledensity. Notwithstanding the increase in competitiveness in the market that
would be expected to increase mainlines; abolition of cross-subsidies, also brought
about with competition, appears to have an overwhelmingly negative impact on
teledensity. The effects of competition in the mobile and the Internet markets are
similarly positive. It appears that competition in these markets increases the
competitiveness in the overall telecommunications industry. The effect on the fixed
market is exhibited with a joint effect of lib_mobile_small and lib_internet_small of
2.45% which is nonetheless not enough to outweigh the negative effect of
liberalisation in the fixed market. Last, the yearly trend variable, t, suggests a general
yearly increase in teledensity of 0.38%.

9
Simultaneous inclusion of competition variables (i.e. lib_fixed) and their interaction variables with
smallness (i.e. lib_fixed_small) in the model failed as the variables exhibit high levels of collinearity
that lead STATA to drop one of the groups. Instead, the interaction variables are finally maintained in
the model as they capture the underlying effects of competition in small economies.

35
Table 4: Regression results for the Competition model and Augmented model
assessing the impact on teledensity

Dependent variable: teledensity Competition Augmented

Model model

Robust Stand. Robust Stand.

Coefficient Error t Coefficient Error t

lag_real_gdp_cap 0.0012607* 0.000072 17.52 0.0011558* 0.0001004 11.52

t 0.3761474* 0.0247733 15.18 0.3494497* 0.0274169 12.75
urban_cap 0.0264613* 0.0085085 3.11 0.0609337* 0.0179365 3.4
inv_tel_per_gdp -484.4132 559.3374 -0.87 920.2715** 555.2994 1.66
Mobile - - - 1.711169* 0.5371329 3.19
lagmobile - - - -1.650504* 0.5369087 -3.07
isdn_cap - - - 0.6468407* 0.2041921 3.17
lagisdn_cap - - - -0.9244944* 0.2142901 -4.31
logcpi2000 0.5834578* 0.0930046 6.27 0.4550983* 0.0943642 4.82
pop_small -0.4660529* 0.1113437 -4.19 -0.3664121* 0.125409 -2.92
gdp_small 0.0301723* 0.0085795 3.52 0.0161906** 0.0087158 1.86
gdp_large 0.0366413* 0.0065694 5.58 0.0272377* 0.0071797 3.79
pop_large -0.0127998* 0.0025433 -5.03 -0.0114394* 0.0023624 -4.84
lib_fixed_small -2.849138* 0.5569155 -5.12 -1.889607* 0.6488387 -2.91
lib_mobile_small 1.024339** 0.5858355 1.75 1.124483** 0.619208 1.82
lib_internet_small 1.02165** 0.6105907 1.67 0.2592679 0.6259207 0.41
Smallness_year(t) all positive values† all negative values†

Number of obs = 2274 Number of obs = 1959

Total number of countries = 141 Total number of countries = 140
Small economies = 56 Small economies = 55
R-sq: within = 0.6633 R-sq: within = 0.6965
F(35,2098) = 81.67 F(39,1780) = 85.31
corr(u_i, Xb) = 0.2734 corr(u_i, Xb) = -0.0169
Notes:
* Statistically significant at the 0.01 level
** Statistically significant at the 0.1 level
† smallness_year(t): interaction variable between the smallness dummy variable and yearly dummies. For the competition model
sixteen out of twenty four coefficients are statistically significant at the 0.05 and 0.1 levels, all with positive values. For the
augmented model the coefficients are all statistically significant at the 0.01 level, all taking negative values.
arable_small and arable_large are dropped from both models because of multicollinearity.

In a subsequent step in the econometric analysis the basic Competition Model

was augmented with variables which assess the impact of the development of
alternative technologies on teledensity. These are represented in Table 2 under the
Augmented Model heading by mobile, isdn_cap and their respective lagged variables.
Inclusion of these variables increases the model’s goodness of fit and reduces the
correlation between the idiosyncratic error and the independent variables; evidence
that in the Competition Model the effects of alternative technologies are captured by
the disturbance term (eit). In addition, there is a loss of 315 observations and one

36
country from the estimator’s calculation as a result of incomplete data, especially in
the earlier years of introduction of alternative technologies. Overall, the Augmented
Model illustrates the same relationships of regressors with the dependent variable,
though the general size of the coefficients is lower. This is expected as in the
Competition Model the impact of alternative technologies was captured by the
existing variables.
Three major changes emerge upon the inclusion of the new variables. First,
the coefficient of inv_tel_per_gdp shows significance now and a positive relationship
with teledensity. Investment in telecommunications as a percentage of GDP appears
to have a strong effect on the network development. Second, liberalisation of the
Internet market in small economies maintains a positive relationship with the
independent, yet the coefficient is statistically insignificant. It may be that
liberalisation in the Internet market per se does not have a substantial effect on
teledensity in the presence of a variable controlling for the development of the actual
Internet market. Last, the coefficients of smallness_year(t) dummies become all
negative with an average value of (-13) all statistically significant at the 0.01 level. A
possible explanation is that smallness_year(t) captures the effects of technological
change over time on teledensity. Technological change seems to be more critical
regarding its effect on teledensity in small economies than large economies. In
addition, regardless the positive effect that competition in alternative technologies has
on teledensity, which is a result of increase in competitiveness, the development of the
markets of alternative technologies overpowers this with a negative effect. This effect
illustrates the strength of economic and technological impact that alternative
technologies have on teledensity.
The effects of the development in mobile voice and the Internet are akin. The
coefficient of mobile is positive whereas the coefficient of its lagged variable is
negative. These results portray a relationship of complementarity at the event year as
both technologies show a growing trend. This reinforces the claim that growth of
mobile telephony has been mainly a demand-creating effect for general
telecommunications services (i.e. Barros and Cadima, 2000). A relationship of
substitution though is observed between mobile and fixed telephony where the
development of the former in the current year affects negatively the development of
the latter in the following year. The joint effect of the two coefficients turns out to
favouring complementarity to substitution. Similar relationships are developed

37
between lagisdn_cap and its lagged variable with teledensity. Positive effects on the
event year depict complementarity whereas the negative relationship with teledensity
on the following year suggests substitution effects. The difference with the impact of
mobile telephony is that the joint effect of isdn_cap and lagisdn_cap favours
substitution to complementarity. From the model, it is not possible to conclude
whether these effects have either an economic or a technological explanation.

3.3 THE EFFECT ON FIXED PRICES

Two semi-logarithmic regression models employing a Fixed Effect transformation are
assessed. The first one concentrates on the effect that competition has on fixed voice
prices in small economies. The second model incorporates also the effect of
alternative technologies on prices. Beginning with the first model, it has the exact
form as the teledensity model; albeit, the dependent variable is log_calls, the proxy
for the level of prices of fixed voice. Caution should be placed on the interpretation of
relationships with the dependent. On the one level, positive relationships with the
dependent would suggest a negative relationship with prices and vice versa. On
another level, the actual percentage effect of dummy variables on the dependent in a
γ
semi-logarithmic expression is e – 1, where γ is the coefficient of the respective
dummy variable (Halvorsen and Palmquist, 1980) produced by STATA and given in
Table 5. Interpretation should be analogous.
Tests akin to the teledensity model were employed to support the selection of
a Fixed Effects transformation in the model. The t-statistics were estimated using
robust standard errors to account for time-series autocorrelation within each country,
increase precision in the coefficients and generate an unbiased Fixed Effects
estimator. The within-R² signifies a very good fit of the specified model to the data
explaining 80% of the variance of prices. The F-statistic is high enough to allow the
rejection of the null hypothesis for joint insignificance in all coefficients. The value of
-0.1141 representing the correlation between the idiosyncratic error (eit) and the
regressors supports the assumption of strict exogeneity of the regressors and also
advocates the use of the Fixed Effects model.
Regarding the regressors’ coefficients, they are all statistically significant at
conventional levels apart from inv_tel_per_gdp, pop_small, and thirteen out of twenty
four smallness_year(t) coefficients. lag_real_gdp_cap has a positive effect on calls

38
volume and therefore a negative effect on prices which implies that more affluent
economies tend to have lower prices. urban_cap has an equivalent effect on prices
which suggests that concentration of population in the urban areas has a downward
effect on prices. The negative effect that logcpi2000 has on prices is unusual as it
would be expected that the level of telecommunications prices rise with general
increases in consumer prices.
GDP has a negative impact on prices in small economies, depicted by the
negative coefficient of gdp_small. This suggests that for the more-affluent small
economies, prices tend to be lower 10 . In contrast, GDP in large economies appears to
have a converse effect on prices, illustrated by the positive coefficient of gdp_large.
Namely, in more-affluent large economies fixed voice prices appear to be higher. As
regards the impact of actual population in large economies, it has a negative effect on
prices (pop_small has an inverse effect but is insignificant). Furthermore, the
smallness_year(t) variables show a positive relationship with logcalls, therefore an
inverse relationship with prices. The coefficients across time do not exhibit any
substantial fluctuation averaging at 0.9. The coefficients imply that in general, small
economies have lower prices than large economies.
Pertaining to the variables assessing the effects of competition there is
statistically significant evidence that liberalisation in small economies affects price.
Yet, a decline in prices is observed only in small economies that have liberalised their
fixed and mobile markets whereas liberalisation in the Internet exhibits an increase in
prices. The effect of liberalisation of the fixed market on prices can be related to
increased competitiveness which drives prices downwards. The liberalisation in the
mobile market also elevates overall competitiveness in the industry. By promoting an
alternative means of voice communication is expected to culminate in a pressure on
prices. Increase in prices caused by competition in the Internet market may reflect the
additional costs for newer technology, upgrades and maintenance of the existing
physical network that incumbents are incurred in order to increase bandwidth, and
provide better interconnection with local and international networks.

10
Again the sequence of steps for interpretation of this coefficient is as follows: the values of
gdp_small variable increase when actual GDP decreases; since an increase in gdp_small has a negative
effect on log_calls it implies that a decrease in actual GDP has a negative effect on log_calls and the
reverse. That is an increase in actual GDP is associated with an increase in log_calls and therefore a
decrease in prices.

39
Table 5: Regression results for the Competition model and Augmented model
assessing the impact on fixed voice prices

Dependent variable: logcalls Competition Augmented

Model model

Robust Stand. Robust Stand.

Coefficient Error t Coefficient Error T

lag_real_gdp_cap .0000197* 6.75e-06 2.92 .0000644* .0000103 6.25

t .1174295* .0028586 41.08 .1229873* .0034322 35.83
urban_cap .0061707* .0007682 8.03 .0054988* .0012953 4.25
inv_tel_per_gdp -18.68793 82.81806 -0.23 -118.354 96.60096 -1.23
mobile - - - .0243152 .0364937 0.67
- - - -.0272709 .0365576 -0.75
lagmobile
- - - -.0310343* .0084259 -3.68
isdn_cap
- - - .0125676 .0085432 1.47
Lagisdn_cap
logcpi2000 .0364545** .0142969 2.55 .0226253 .0156859 1.44
pop_small .009335 .0132162 0.71 .0397711* .0150994 2.63
gdp_small -.0032443* .0008738 -3.71 -.0032174* .000968 -3.32
gdp_large .0035567* .0005722 6.22 .003044* .0005603 5.43
pop_large -.0012789* .0004825 -2.65 -.001448* .0004356 -3.32
lib_fixed_small .1318702* .0501728 2.63 .1929631* .0691193 2.79
lib_mobile_small .1316105* .0448929 2.93 .0708524 .052265 1.36
lib_internet_small -.1879944* .0434943 -4.32 -.1592018* .0511278 -3.11
smallness_year(t) all positive values† all positive values†

Number of obs = 2216 Number of obs = 1875

Total number of countries = 140 Total number of countries = 139
Small economies: 56 Small economies: 55
R-sq: within = 0.8051 R-sq: within = 0.8116
F(35,2041) = 186.86 F(39,1697) = 216.21
corr(u_i, Xb) = -0.1141 corr(u_i, Xb) = -0.1386
Prob > F = 0.0000 Prob > F = 0.0000
Notes:
* Statistically significant at the 0.01 level
** Statistically significant at the 0.05 level
† smallness_year(t): interaction variable between the smallness dummy variable and yearly dummies. For the competition model
eleven out of twenty four coefficients are statistically significant at the 0.1 level, all with positive values. For the augmented
model eighteen out of twenty four coefficients are statistically significant at the 0.1 and 0.05 levels, all taking positive values.
arable_small and arable_large are dropped from both models because of multicollinearity.

The respective Augmented Model for prices incorporates the effects of

alternative technologies in the former model. Inclusion of these variables increases the
model’s goodness of fit and maintains the correlation between the idiosyncratic error
and the independent variables at low levels. There is a loss of 341 observations and
one country from the estimator’s calculation. Overall, the Augmented Model
illustrates the same relationships of regressors with the dependent variable, though the
general size of the coefficients is lower. The coefficient of inv_tel_per_gdp continues
to be insignificant; the coefficients of logcpi2000 and lib_mobile_small lose their

40
statistical significance; and the coefficient for pop_small exhibits a statistically
significant negative effect on prices. The latter suggests that the larger the population
in small economies the lower the prices of the fixed voice.
All smallness related variables; namely, pop_small, gdp_small, gdp_large,
pop_large, and smallness_year(t) maintain equivalent results consistent with the
Competition Model. This also applies to the effects of the competition variables. Yet,
among the newly included variables of alternative technologies, only development in
the Internet market appears to affect prices. Its positive effect is consistent with the
positive effect of liberalisation in the Internet market. Analysis also shows that the
initial impact of the introduction of competition in fixed telephony in a small
economy has a negative effect on prices. Nevertheless, subsequent development in the
market does not have an effect on prices. On the one hand, liberalisation might cause
prices to attenuate in the form of price rebalancing or downward pressure on high
monopoly prices. On the other hand, the small size of the market and large economies
of scale in small economies inhibit the achievement of minimum efficient scale of
operation that would further decrease consumer prices. Paradoxically, small
economies in average manifest lower prices than large economies. This suggests the
existence of other factors that keep prices low and outweigh the small size of the
market.

CHAPTER 4 – DISCUSSION
Existing studies on the effects of the liberalisation of telecommunications and
alternative technologies on Universal Service appear to disregard the underlying
importance of the size of the economy. Whilst the literature on small economies
postulates that the optimum policy should differ with size, the telecommunications
literature has placed nominal emphasis upon it. The paper elaborates on the
distinctiveness of smallness and stimulates research towards a direction that
encompasses the size of economy.
An econometric approach is pursued in the search for empirical evidence of
the effects of competition and alternative technologies on Universal Service in the
context of small economies. A Fixed Effects econometric model assesses the effects
of these factors on both fixed teledensity and prices. The model uses data for more
than 140 economies for the period 1980-2004. Research findings provide significant
evidence that smallness is a fundamental concept for policy and economic research.

41
 Teledensity in small economies appears to be lower than in large economies.
Lower levels of teledensity do not necessarily reflect poorer provision of service. On
the one hand, the teledensity indicator simply captures access to fixed voice and not
the actual level of access to voice telephony which is complemented by alternative
technologies. Particularly for small economies, alternative technologies are more
likely to have a considerable contribution to voice telephony relative to fixed voice, as
they feature lower costs of supply and easier network expansion. On the other
hand, competition in fixed voice appears to deter network expansion in small
economies. This negative effect outweighs the positive effect that alternative
technologies have on fixed teledensity that is a result of increasing competitiveness in
the sector. As it is most likely, smallness restrains fixed voice firms from operating at
MES and therefore they concentrate on the most profitable areas. This strategy
decreases the profit margins in these areas for incumbents from which they normally
subsidise network expansion in less profitable areas.
As regards the level of prices, small economies appear to have lower prices
than large economies. This is mainly a result of competition in the fixed voice.
Competition in the mobile market and respective development in mobile telephony do
not affect fixed voice prices. On the other hand, competition in the Internet market
and development of the Internet market increase fixed prices. Regardless, their
combined positive effect is overpowered by the stronger negative effect of
competition in fixed telephony.
The behaviour of smallness’s size components advocates that smallness has a
significant leverage on the effects of competition and alternative technologies on
Universal Service. Namely, GDP and population 11 demonstrate varying relationships
with or magnitude of effect on Universal Service, between small and large economies.
To illustrate, GDP shows negligible negative relationship with teledensity for both
groups of economies. However, GDP in small economies has a negative impact on
prices in contrast to large economies where GDP has a positive impact. The different
impact of GDP on prices might manifest different relationships between prices and
purchasing power of consumer for small and large economies. GDP’s minor effect on
teledensity might suggest that network expansion is primarily determined by the

11
No useful results are obtained for arable area as the variable is rejected from the models

42
capacity of fixed voice providers to supply and not the purchasing power of
consumers.
On another level, population has a positive effect on teledensity for both
groups of economies but this effect is substantially larger for small economies. As
regards its relationship with price, population is associated with decreases in prices for
both economies. Population denotes the size of the market. As long as the market size
increases, prices attenuate and teledensity expands. This is the underlying effect of
existence of strong economies of scale in telecommunications. Their exploitation
allows firms to achieve efficient levels of operation that enable them to generalise
service and reduce prices. Stronger effects in small economies imply that the role of
market size in small economies is critical for the provision of Universal Service.
Regarding the effects of liberalisation, competition in the three markets of
fixed voice, mobile telephony, and the Internet portrays varying effects on Universal
Service. Liberalisation in the fixed voice exhibits a negative effect on teledensity
probably as a result of the discontinuation of cross-subsidisations, which increases the
per capita cost of installing new lines. New entrants’ cream-skimming strategies
minimise the profit margin for incumbents which are normally responsible for the
provision of Universal Service. In addition, in small economies the fixed voice market
is possible to exhibit large economies of scale that inhibit competition. Competition in
fixed voice also has a negative effect on prices in small economies. This is anticipated
from increased competitiveness in the market.
Competition in mobile telephony in small economies has a positive effect on
teledensity. Increased competitiveness in the industry induced by competition in
mobile telephony, most likely motivates fixed voice providers to look into unserved
and potentially less profitable areas. Fixed voice providers face, not only customer
switching to competitors, but customer switching to alternatives technologies, as well.
Competitiveness induced by competition in mobile telephony also has a negative
effect on fixed voice prices.
With relation to competition in the Internet market, the latter exhibits a
positive effect on teledensity. Competition in the Internet increases the demand for
mainlines, since Internet users need to have access to the fixed network 12 .
Competition in the Internet in small economies also increases fixed voice prices. This

12
At least for dial-up, ISDN, and DSL connections

43
is probably a result of additional investments which are necessary to enhance
bandwidth capacity and connectivity. These investments increase the cost of the
physical network, which is consequently transferred onto consumers.
The effects of alternative technologies on teledensity are analogous to the
effects that competition in the respective markets has. Mobile telephony and the
Internet exhibit both relationships of substitution and complementarity with fixed
voice. That is, all three technologies show positive growth over time. However,
alternative technologies demonstrate an effect of substitution on fixed voice.
Development of mobile telephony and the Internet in the current year appears to have
a negative effect on development of fixed voice in the subsequent year. These
findings suggest that presently all three technology markets develop. Parallel
development increases network effects which benefit consumers. In addition,
utilisation of the fixed network by both fixed voice and the Internet reinforces a
relationship of complementarity between them. However, the relationship of
substitution illustrates that alternative technologies absorb users from the fixed voice
for their capacity to offer voice communication in a more technologically attractive
manner and at lower prices. Mobile telephony may be perceived as an alternative
means for access to voice and content communication, facilitating mobility and
development of added-value services. On the other hand, the Internet is a universal
platform that can support voice, data, and image transmission.
The impact that development in alternative technologies has on Universal
Service is stronger than the effects of liberalisation in the respective markets, per se.
Liberalisation alters the dynamics of the industry; yet, the most substantial impact on
Universal Service is caused by development in the actual markets. What is more,
liberalisation and development in the respective markets do not necessarily exhibit the
same relationships with teledensity and prices. This illustrates the magnitude of the
impact brought about by the evolving nature of technologies.

CHAPTER 5 – POLICY IMPLICATIONS

Research findings in this paper shed light on an array of pertinent issues in the
telecommunications sector. These findings are of great importance to international
policy organisations and local governments of small economies which strive to
achieve the optimum policy for the sector.

44
 A central conclusion from the paper is that distinguishing between small and
large economies has an analytical, predictive, and normative usefulness. Smallness
sheds light on particular phenomena encountered in a number of economies as a result
of size. These characteristics can be utilised to understand social or economic
concerns and obtain knowledge that allows predicting their future influence on these
concerns. Last, this knowledge can be embedded in regulatory policy for the
promotion of social welfare and pertinent regulatory tasks.
The optimum policy in small economies seems to favour the preservation of a
single provider in the fixed voice where economies of scale are large and the MES is
rather difficult to achieve by more than one operator. The effects of smallness on
competition in the fixed voice are manifested through the impact that the latter has on
Universal Service. The act of liberalisation in the market directs competing firms
towards the most profitable (crudely, the most populated) areas. Moreover, the
general profitability of the market is weakened by competition in alternative
technology markets. Consumers switch to competitors and alternative technologies.
Eventually, network expansion becomes a secondary task, subsidies are harder to
produce and incumbent providers are less apt to operate efficiently. It is not therefore
surprising that liberalised small economies may become monopolies in the less
efficient markets.
Existence of natural monopoly in the fixed voice in small economies should
not be condemned by policy makers. Liberalisation of the market can only harm
economic and social welfare and constitutes an inappropriate policy for the promotion
of Universal Service. Instead, policy for the fixed voice should facilitate the efficient
operation of the incumbent. Liberalisation in the market may take place only in the
event that the minimum efficient scale of operation for the incumbent is small relative
to the total market size so that it allows the efficient operation of additional providers.
On the part of the incumbent provider, it should aim to exploit the economies
of scale necessary to minimise its long-term average costs. This implies that,
conditional the available technology, if the total market size is not sufficient to meet
MES, the firm should seek for additional demand in neighbouring economies.
Collaboration with neighbouring providers can facilitate the minimisation of average
cost, which consequently can be expressed in reductions in prices. Serving a fraction
of a neighbouring economy’s fixed telephony may enable the host provider to meet
Universal Service requirements and the “parasite provider” to achieve MES.

45
Apparently, such parasitic models of fixed telephony may not be applicable for water-
surrounded economies. Yet, economies in the regions of the former Soviet Union and
the Gulf may exhibit suitable conditions for shared provision of service.
An important implication that derives from the theoretical and empirical
analysis in this paper concerns the relationships that evolve between
telecommunications technologies that lead to the blurring of telecommunications
markets. Existing policies which base their provisions on the nature of technology
become obsolete upon the convergence of technologies. Particularly, in small
economies, policy should be oriented on consumer needs met by technology and not
on the technology as such. A postulate by George Stingler (1975) that an industry
should encompass products and services which in the long term tend to manifest a
relationship of substitution, should also pervade regulation. If consumers of a product
A can easily switch to a product B or vice versa, then these products should be
combined in the same industry and therefore be ruled by the same competition and
regulatory policy. This may be applicable in the distinct markets of fixed and mobile
telephony, which exhibit an imminent relationship of substitution.
Emerging relationships between fixed telephony and alternative technologies
call for a redefinition for Universal Service. Access to voice telephony is a more
appropriate indicator than access to fixed voice that is currently used to gauge
Universal Service. This will allow alternative technologies to contribute to Universal
Service. Respectively, Universal Service prices should refer to the cost of a basket of
minutes of voice telephony that represents all technologies which facilitate voice
telephony. In areas where Universal Service is more efficiently implemented through
fixed telephony this should be signified in the number of minutes for fixed voice
which minimise the cost of the basket. Currently, only fixed voice providers
contribute to the Universal Service fund. This financial burden, in addition to
decreasing profit margins in the fixed voice, renders fixed voice firms incapable to
minimise their costs. Hence, a redefinition of the Universal Service will allocate this
burden to alternative providers, as well.
To conclude, economies of scale are fundamental for the efficient operation of
telecommunications firms. Structural competition policies, such as divestiture of the
incumbent operator and condemnation of merger agreements on the grounds of their
subsequent increase in market concentration should be avoided as they may sacrifice
efficiency. Currently, international telecommunications regulation is following the US

46
and the EU models, which are strict with regard to concentration issues. This
regulation is becoming less suitable for small economies. It is therefore necessary for
small economies to devise endogenous policy which would address their distinctive
characteristics.

47
 References

Armstrong, H. and Read, R. (1995a) Western European micro-states and EU

autonomous regions: the advantages of size and sovereignty. World
Development, 23, pp 1229-1245.

Armstrong, M. (2001) Access Pricing, Bypass and Universal Service. The American
Economic Review, 91, pp 297-301.

Armstrong, W. H. and Read, R. (1995b) Western European micro-states and EU

autonomous regions: the advantages of size and sovereignty. World
Development, 23, pp 1229-1245.

Armstrong, W. H. and Read, R. (1998) Trade and growth in small states: the impact
of global trade liberalisation. World Development, 21, pp 563-585.

Baltagi, H. B. (2005) Econometric analysis of panel data, West Sussex, John Wiley &
Sons, Ltd.

Banerjee, A. and Ros, J. A. (2004a) Drivers of demand growth for mobile

telecommunications services: Evidence from international panel data, Global
Economy and Digital Society, NERA.

Banerjee, A. and Ros, J. A. (2004b) Patterns in global fixed and mobile

telecommunications development: a cluster analysis. Telecommunications
Policy, 28, pp 107-132.

Barros, P. L. P. and Cadima, N. (2000) The impact of mobile phone diffusion on the
fixed-link network, London, Centre for Economic Policy Research.

Barros, P. L. P. and Seabra, M. C. (1999a) Universal service: does competition help or

hurt? Information Economics and Policy, 11, pp 45-60.

Barros, P. P. and Seabra, M. C. (1999b) Universal service: does competition help or

hurt? Information Economics and Policy, 11, pp 45-60.

Bergman, L., Doyle, C., Gual, J., Hultkrantz, L., Neven, D., Röller, L.-H. and
Waverman, L. (1998) Europe's network industries: conflicting priorities:
telecommunications, London, Centre for Economic Policy Research.

Bohlin, E. (1997) Editorial: Restructuring Japan's Telecommunications.

Telecommunications Policy, 21, pp 79-84.

Briguglio, L. and Buttigieg, E. (2004) Competition constraints in small jurisdictions.

Bank of Valetta Review, 30, pp 1-13.

Dasgupta, P. and Stiglitz, J. (1988) Learning-by-doing, Learning Structure and

Industrial and Trade Policies. Oxford Economic Papers, 40, pp 246-268.

Eliassen, A. K. and Marit, S. (1999) European telecommunications liberalisation,

London, Routledge.

48
Eliassen, A. K., Mason, T. and Sjøvaag, M. (1999) European telecommunications
policies - deregulation, re-regulation or real liberalisation?, in: E. A. Kjell and
M. Sjøvaag (eds) European Telecommunications Liberalisation, London,
Routledge.

Fuentes-Bautista, M. (2001) Universal Service in times of Reform: Affordability and

Accessibility of Telecommunication Services in Latin America, 29th Research
Conference on Communication, Information and Internet Policy, Texas.

Gal, M. S. (2003a) Competition policy for small market economies, Cambridge, MA,
Harvard University Press.

Gal, M. S. (2003b) Size matters for competition policy: prepared remarks for the
OECD session on small economies, Centre for Co-operation with Non-
members: Directorate for Financial, Fiscal and Enterprise Affairs. OECD
Global Forum on Competition., OECD.

Gasmi, F., Laffont, J. J. and Sharkey, W. W. (2000) Competition, Universal Service

and Telecommunications Policy in Developing Countries. Information
Economics and Policy, 12, pp 221-248.

Gillet, S. (1994) Technological change, market structure and universal service, 22nd
Annual Telecommunications Policy Research Conference.

Gruber, H. and Verboven, F. (2001) The evolution of markets under entry and
standards regulation - The case of global mobile telecommunications, Centre
for Economic Policy.

Gutierrez, H. L. and Berg, S. (2000) Telecommunications liberalization and

regulatory governance: lessons from Latin America. Telecommunications
Policy, 24, pp 865-884.

Halvorsen, R. and Palmquist, R. (1980) The interpretation of dummy variables in

semilogarithmic equations. American Economic Review, 70, pp 474-475.

Hamilton, J. (2003a) Are Main Lines and Mobile Phones Substitues or Complements?
Evidence from Africa. Telecommunications Policy, 27, pp 109-133.

Hamilton, J. (2003b) Are main lines and mobile phones substitutes or complements?
Evidence from Africa. Telecommunications Policy, 27, pp 109-133.

Hills, J. (1986) Deregulating Telecoms: Competition and Control in the United States,
Japan and Britain, London, Frances Pinter.

Hudson, E. H. (2001) Extending Access to the Digital Economy to Rural and

Developing Regions, Cambridge, MA, MIT Press.

ITU (1998) World telecommunication development report: Universal access and

world telecommunication indicators, Geneva, International
Telecommunication Union.

49
ITU (2006) World Telecommunication/ICT Development Report 2006: Measuring
ICT for Social and Economic Development, Geneva, International
Telecommunication Union.

Jalan, B. (1982) Problems and policies in small economies, New York, St. Martin's.

Jha, R. and Majumdar, K. S. (1999) A matter of connections: OECD

telecommunications sector productivity and the role of cellular technology
diffusion. Information Economics and Policy, 11, pp 243-269.

Kaminarides, J., Briguglio, L. and Hoogendonk, N. H. (1989) The Economic

Development of Small Countries: Problems, Strategies and Policies, Delft,
Eburon.

Li, W. and Xu, C., Lixin (2004) The impact of privatization and competition in the
telecommunications sector around the world. Journal of Law and Economics,
XLVII, pp 395-430.

Lien, D. and Peng, Y. (2001) Competition and production efficiency

Telecommunications in OECD countries. Information Economics and Policy,
13, pp 51-76.

Minges, M. (1999) Developing Policies for Universal Service, ITU/BDT Workshop on

Telecommunication Reform, Botswana.

Mueller, M. (1993) Universal service in telephone history: a reconstruction.

Telecommunications Policy, 17, pp 352-369.

Pitelis, C. (2003) Privatisation, regulation and domestic competition policy, in: G.

Wignaraja (ed) Competitiveness strategy and industrial performance in
developing countries: a manual for policy analysis, London, Taylor and
Francis.

Posner, R. A. (1976) Antitrust law : an economic perspective, Chicago, University of

Chicago Press.

Rodini, M., Ward, R. M. and Woroch, A. G. (2002) Going Mobile: Substitutability

Between Fixed and Mobile Access, Competition in Wireless: Spectrum,
Service and Technology Wars, The Public Utility Research Center, University
of Florida, Gainesville.

Ros, A. J. (1999) Does Ownership ro Competition Matter? The Effects of

Telecommunications Reform on Network Expansion and Efficiency. Journal
of Regulatory Economics, 15, pp 65-92.

Ros, J. A. and Banerjee, A. (2000) Telecommunications Privatization and Tariff

Rebalancing: Evidence from Latin America. Telecommunications Policy, 24,
pp 233-252.

Scherer, F. M., Beckenstein, A., Kaufer, E. and Murphy, R. D. (1975) The Economics
of Multi-Plant Operation: An International Comparisons Study, Cambridge,
Massachusetts, Harvard University Press.

50
Scherer, J. and Bartsch, T. (1998) Telecommunications Law and Policy in the
European Union, London, Butterworths.

Selwyn, P. (1980) Smallness and islandness. World Development, 8, pp 945-951.

Spiller, T. P. and Cardilli, G. C. (1997) The Frontier of Telecommunications

Deregulation: Small Countries Leading the Pack. The Journal of Economic
Perspectives, 11, pp 127-138.

Stewart, T. (2004) Is Flexibility Needed When Designing Competition Law for Small
Open Economies? A View from the Caribbean. Journal of World Trade, 38,
pp 725-750.

Stigler, G. J. (1975) The citizen and the State: essays on regulation, Chicago,
University of Chicago Press.

Sung, N. and Lee, Y.-H. (2002) Substitution Between Mobile and Fixed Telephones
in Korea. Review of Industrial Organization, 20, pp 367-374.

Taubman, C. and Vagliasindi, M. (2005) Fixed and Mobile Competition in Transition

Economies, European Bank for Reconstruction and Development and World
Bank.

Thatcher, M. (1996) High Technology, in: H. Kassim and A. Menon (eds) European
Union and National Industrial Policy, London, Routledge.

Thatcher, M. (1999) Liberalisation in Britain. From monopoly to regulation of

competition, in: E. A. Kjell and M. Sjøvaag (eds) European
Telecommunications Liberalisation, London, Routledge.

Vernon, M. J., Viscusi, K. and Harrington, J. (2000) Economics of Regulation and

Antitrust, MIT Press.

Vickers, J. and Yarrow, G. (1988) Privatization: An Economic Analysis, Cambridge,

MA, MIT Press.

Wallsten, J. S. (2001) An econometric analysis of telecom competition, privatization,

and regulation in Africa and Latin America. The Journal of Industrial
Economics, XLIX, pp 1-19.

Wellenius, B. (1993) Telecommunications: World Bank Experience and Strategy,

World Bank Publications.

Wellenius, B. and Stern, A. P. (1996) Implementing Reforms in the

Telecommunications Sector: Lessons from Experience, Brookfield USA,
Avebury.

Williamson, E. O. (1975) Markets and hierarchies: analysis and antitrust

implications: a study in the economics of internal organization, New York,
The Free Press.

51
Wooldridge, M. J. (2002) Econometric Analysis of Cross Section and Panel Data,
Cambridge MA, The MIT Press.

52
Appendix I: An illustration of economy categorisation based on GDP, arable area and
population combined index.

Country GDP (USD billions) Arable area (sq. km) Population Index values

Ten highest
United States 8785.00 1760180 295409638 74.33
index values
from total China 1419.00 1426210 1285227228 65.73
number of India 698.50 1617150 1087123789 61.47
countries
Russia 485.00 1234650 143899225 28.95
Brazil 611.70 589800 183912538 18.26
Japan 3630.00 44180 127923488 17.93
Canada 856.60 457440 31957882 12.74
Germany 2164.00 117910 82645291 12.59
Australia 496.70 483000 19942410 11.55
France 1605.00 184490 62052222 11.19
Ten highest
Nicaragua 5.57 19250 5376140 0.53
index values
from small Dominican Rep. 18.81 10960 8767870 0.51
economies Slovak Republic 21.38 14330 5401480 0.49
Croatia 25.25 14620 4539732 0.49
Turkmenistan 2.74 18500 4766009 0.48
Central African Rep. 1.44 19300 3985971 0.47
Moldova 1.81 18430 4217911 0.47
Rwanda 3.23 11160 8882365 0.45
Singapore 86.17 10 4272572 0.44
Latvia 7.55 18320 2318469 0.44
Ten lowest index
Micronesia (Fed. States of) 0.17 40 109691 <0.01
values from
small economies Seychelles 0.51 10 79910 <0.01
Grenada 0.31 20 102254 <0.01
Dominica 0.18 50 78534 <0.01
Saint Kitts and Nevis 0.27 70 42189 <0.01
Kiribati 0.05 20 97409 <0.01
Marshall Islands 0.06 30 59721 <0.01
Palau 0.10 40 19853 <0.01
Nauru 0.03 0 13386 <0.01
Tuvalu 0.02 0 10396 <0.01

Source: Data were obtained from the United Nations’ Human Development Report of 2005
Index value = 100/3 (P/Pmax +A/Amax + Y/Ymax),
P, A and Y are population, arable area and GDP of each country, respectively;
Pmax, Amax and Ymax represent the highest values of population, arable area and GDP respectively.

53
Appendix II: Relationship between actual GDP and gdp_small

100000000000 6000
90000000000
5000
80000000000
70000000000
4000
60000000000
GDP
USD 50000000000 3000
gdp_small
40000000000
2000
30000000000
20000000000
1000
10000000000
0 0

54