Broadband in Latin America

Broadband in
Latin America
Beyond Connectivity
Valeria Jordán
Hernán Galperin
Wilson Peres
Editors
@LIS2 - Alliance for the Information Society
in Latin America and the Caribbean, phase 2
Programme financied by the European Union
inclusion innovation development
Broadband in
Latin America
Beyond Connectivity
Valeria Jordán
Hernán Galperin
Wilson Peres
Editors
Economic Commission for Latin America and the Caribbean (ECLAC)
Santiago, Chile, December 2013
Libros de la CEPAL
Alicia Bárcena
Executive Secretary
Antonio Prado
Deputy Executive Secretary
Mario Cimoli
Chief, Division of Production, Productivity and Management
Ricardo Pérez
Chief, Publications and Web Services Division
120
The preparation of this document was coordinated by Valeria Jordán and Wilson Peres,
of the Economic Commission for Latin America and the Caribbean (ECLAC), and Hernán
Galperin, from the Regional Dialogue on the Information Society (DIRSI), in the framework
of the project “Inclusive political dialogue and exchange of experiences”, under the Alliance
for the Information Society programme phase 2 (@LIS2), which is jointly fnanced by ECLAC
and the European Union and implemented by the Division of Production, Productivity and
Management of ECLAC.
The opinions expressed in this document, which has not undergone editorial review, are the
sole responsibility of the authors and may not coincide with the views of the United Nations.
This document was produced with the fnancial assistance of the European Union. The
opinions expressed herein do not necessarily refect the offcial opinion of the European Union.
This work was conducted with the help of a grant from the International Development
Research Centre, Ottawa, Canada.
This document may be downloaded at www.cepal.org/Socinfo.
United Nations publication
ISBN: 978-92-1-121836-7 • 978-92-1-056012-2
LC/G.2583-P
Sales No. E.14.II.G.7
Copyright © United Nations, December 2013. All rights reserved
Printed at United Nations, Santiago, Chile • 2013-644
Applications for the right to reproduce this work are welcomed and should be sent to the
Secretary of the Publications Board, United Nations Headquarters, New York, N.Y. 10017,
United States. Member States and the governmental institutions may reproduce this work
without prior authorization, but are requested to mention the source and inform the United
Nations of such reproduction.
Contents
Foreword ............................................................................................................ 13
First part
Diagnosis ............................................................................................................. 15
Chapter I
The shifting digital paradigm in Latin America ............................................ 17
A. The era of cloud computing .......................................................... 18
B. Broadband in the region ................................................................ 24
C. Lessons for a new era ..................................................................... 33
Bibliography ....................................................................................................... 37
Chapter II
The demand gap: drivers and public policies ................................................ 39
A. Measuring the digital divide from the demand side ................ 40
B. Understanding the demand gap .................................................. 44
C. The situation in Latin America ..................................................... 48
1. Broadband coverage ................................................................. 50
2. The demand gap ....................................................................... 51
3. Explaining the demand gap .................................................... 52
D. Public policies to close the demand gap ..................................... 61
1. Policies for addressing the affordability barrier .................. 61
2. Policies for addressing the skills barrier ............................... 64
3. Policies for addressing the lack of interest/
relevance barrier ....................................................................... 66
4. Programmes targeting specifc population segments ......... 67
E. Conclusions ..................................................................................... 68
ECLAC 4
Bibliography ....................................................................................................... 70
Annex ................................................................................................................... 72
Chapter III
Regional and international connectivity ......................................................... 75
A. Introduction .................................................................................... 75
B. Internet connectivity ...................................................................... 76
1. Factors that affect connectivity ............................................... 76
2. Quality parameters ................................................................... 77
3. Reference prices ........................................................................ 79
4. Aggregation factor .................................................................... 80
C. South America’s dependence on international links ................ 81
1. The international bottleneck and its
strategic importance ................................................................. 82
2. Regional interconnection ......................................................... 82
3. Variables affecting costs ........................................................... 86
4. Global trends and prices .......................................................... 88
D. Conclusions ..................................................................................... 90
Bibliography ....................................................................................................... 92
Annexes ............................................................................................................... 93
Best operating practices and key performance indicators ................ 94
Second part
Economic impact .............................................................................................. 109
Chapter IV
Broadband, digitization and development ...................................................111
A. Broadband and economic growth ...............................................111
1. GDP growth ............................................................................. 113
2. Job creation .............................................................................. 118
3. Growth of household income ............................................... 121
B. Digitization and development ................................................... 124
C. Policy implications ....................................................................... 130
Bibliography ..................................................................................................... 133
Chapter V
Mobile broadband: the urgent need for speedier roll-out .......................... 135
A. Introduction .................................................................................. 135
B. Status of mobile telephony in Latin America ........................... 137
C. Mobile data networks in Latin America ................................... 140
1. Relationship between penetration and wealth creation .. 140
2. Introduction of mobile broadband networks
in Latin America ..................................................................... 141
D. A supporting theoretical model and its implications ............. 147
1. The adjustment model ........................................................... 147
Broadband in Latin America: Beyond Connectivity 5
2. Implications of the model ...................................................... 149
E. Conclusions ................................................................................... 151
Bibliography ..................................................................................................... 153
Chapter VI
Cloud computing, structural change and job creation in SMEs ................ 155
A. Introduction .................................................................................. 155
B. What is cloud computing, and how can it affect
the global economy? .................................................................... 158
C. The theoretical model ................................................................. 161
1. Labour market and job matching ......................................... 162
2. Households .............................................................................. 163
3. Technology ............................................................................... 164
4. Entry ......................................................................................... 167
5. Imperfect competition and job creation .............................. 167
6. Bargaining over wages and hours ........................................ 169
7. Business creation, hiring and IT policies ............................. 170
D. The introduction of cloud computing ...................................... 171
1. Aggregation and market clearing ........................................ 173
2. The equilibrium in the cloud economy ............................... 175
E. Calibration .................................................................................... 176
F. Transition to the cloud economy ................................................ 178
G. Conclusions ................................................................................... 182
Bibliography ..................................................................................................... 185
Third part
Public policies ................................................................................................... 187
Chapter VII
National broadband plans .............................................................................. 189
A. Introduction .................................................................................. 189
B. The end of a cycle: the changing role of the State
in telecommunications ................................................................ 191
C. Overview of national broadband plans .................................... 194
1. Argentina: Plan Nacional de Telecomunicaciones
Argentina Conectada ............................................................ 194
2. Brazil: Plano Nacional de Banda Larga ............................... 196
3. Chile: Plan Todo Chile Comunicado .................................... 197
4. Colombia: Plan Vive Digital .................................................. 198
5. México: Agenda Digital.mx ................................................... 199
D. National broadband plans: different strategies, same goal .... 201
1. Analysis and objectives.......................................................... 201
2. Investment and funding ........................................................ 202
3. Deployment models ............................................................... 202
4. Regulation and coordination with the private sector ........ 204
ECLAC 6
E. Conclusions ................................................................................... 206
Bibliography ..................................................................................................... 209
Annexes ............................................................................................................. 212
Chapter VIII
Broadband and industrial policy: the Korean experience .......................... 215
A. Broadband industrial policy: defnition and scope ................. 215
1. Sectoral industrial policy in the period before the
WTO (pre-1995: 1G) ................................................................ 220
2. Horizontal industrial policy in the WTO
(1995-2005: 2G) ........................................................................ 220
3. Targeted cutting-edge policy in the WTO
(2005-2010: 3G) ........................................................................ 220
4. Generative convergence policy in the WTO
(2010-2020: 4G-5G) .................................................................. 221
B. Broadband structure and dynamics .......................................... 223
1. Expansion and mass adoption .............................................. 223
2. Drivers of mass penetration .................................................. 226
3. Market structure ..................................................................... 229
4. ICT goods production and foreign trade ............................ 233
C. Policy convergence ....................................................................... 235
1. Broadband development policy ........................................... 237
2. Competition and regulation .................................................. 238
3. Broadband industrial policy ................................................. 240
4. Technological development policy ...................................... 242
D. The Giga Korea Plan 2020 ........................................................... 242
E. Conclusions ................................................................................... 245
Bibliography ..................................................................................................... 250
Chapter IX
Net neutrality: debate and policies ................................................................ 253
A. Introduction .................................................................................. 253
B. What is net neutrality? ................................................................ 254
1. The non-discrimination principle ........................................ 256
2. The growth of traffc and net neutrality .............................. 257
C. The situation in Europe, the United States and Asia-Pacifc 529
1. Europe and the United States ............................................... 259
2. Asia-Pacifc .............................................................................. 264
D. Situation and outlook in Latin America .................................... 266
1. Chile .......................................................................................... 267
2. Brazil ......................................................................................... 267
E. Criteria for developing a national policy .................................. 268
Bibliography ..................................................................................................... 270
Broadband in Latin America: Beyond Connectivity 7
Fourth part
The future of the ecosystem ............................................................................ 271
Chapter X
The advance ofcloud computing ................................................................... 273
A. Introduction .................................................................................. 273
B. What is cloud computing? .......................................................... 274
1. Essential characteristics ......................................................... 275
2. Service models ........................................................................ 277
3. Enabling technologies ............................................................ 280
4. Deployment models ............................................................... 282
5. Service layers ........................................................................... 283
C. Problems and challenges ............................................................. 284
1. Security and privacy .............................................................. 284
2. Reliable network access ......................................................... 286
3. Legal and regulatory aspects ................................................ 286
4. Resistance to change............................................................... 287
D. Migration to cloud computing ................................................... 288
1. Reasons for migrating ............................................................ 289
2. Assessing the benefts of cloud computing ........................ 290
3. Considerations for successful migration ............................. 292
E. The situation in Latin America ................................................... 295
1. Legal and regulatory environment ...................................... 295
2. Degree of adoption in the region .......................................... 297
F. Mobile cloud computing ............................................................. 300
1. The structure of mobile cloud computing .......................... 300
2. Challenges ............................................................................... 301
3. Mobile cloud computing ...................................................... 305
G. “Everything” in the cloud: reality or utopia? ........................... 306
1. Back to the past ....................................................................... 307
2. Technical considerations ........................................................ 308
3. Legal and procedural aspects ............................................... 310
H. Best international practices ......................................................... 311
1. Explore cloud benefts ............................................................ 312
2. Understand and manage cloud-related risks .................... 313
3. Promote service transparency ............................................... 313
4. Clarify and enhance accountability ..................................... 314
5. Ensure data portability .......................................................... 314
6. Facilitate interoperability ...................................................... 314
7. Adapt and harmonize regulatory frameworks .................. 315
8. Provide suffcient connectivity ............................................. 315
Bibliography ..................................................................................................... 316
ECLAC 8
Chapter XI
The challenge of over-the-top content and services .................................... 319
A. Introduction .................................................................................. 319
B. Over-the-top services, applications and content ..................... 321
C. Main fndings in the literature.................................................... 324
1. Changes in the value chain and the
online-industry threat ............................................................ 324
2. The network neutrality debate ............................................. 325
D. The over-the-top market in Latin America ............................... 327
1. Incumbent operators, OTT strategies and supply ........... 327
2. Netfix, Skype and WhatsApp .............................................. 332
E. The strategy and policy debate .................................................. 333
F. Conclusions ................................................................................... 336
Bibliography ..................................................................................................... 338
ECLAC Publications ........................................................................................ 341
Tables
II.1 Developed countries: size of the fxed broadband
demand gap, 2011 ................................................................................... 41
II.2 Developed countries: mobile broadband demand gap, 2011 ........... 24
II.3 United States: reasons for not adopting broadband
in the home, 2009 .................................................................................... 46
II.4 United States: reasons for not adopting broadband based
on the availability of a computer in the home, 2011 .......................... 47
II.5 Spain: reasons for not adopting broadband,
by income level, 2011 .............................................................................. 47
II.6 Internet users and broadband subscribers
in Latin America, 2011 ............................................................................ 49
II.7 Broadband coverage in Latin America, 2011-2012 ............................. 51
II.8 The fxed broadband demand gap in Latin America, 2011 ............... 51
II.9 The mobile broadband demand gap in Latin America, 2011 ........... 52
II.10 Latin America: factors explaining the demand gap ........................... 54
III.1 Internet speeds in the region according to Ookla,
1 December 2012 ..................................................................................... 77
III.2 Internet speeds in the region according to Akamai ........................... 78
III.3 Reference prices for fxed access per Mbps for download
speeds of approximately 2 Mpbs, November 2012 ............................ 80
III.A.1 Defnition of broadband ........................................................................ 96
III.A.2 Key IXP performance indicators ......................................................... 105
III.A.3 Key country performance indicators ................................................ 107
IV.1 Latin America: contribution of broadband to GDP growth ........... 113
IV.2 Colombia: contribution of broadband to GDP growth ................... 114
IV.3 Panama: contribution of broadband to GDP growth ...................... 116
IV.4 Chile: impact of broadband on job creation ...................................... 119
Broadband in Latin America: Beyond Connectivity 9
IV.5 Colombia: impact of broadband penetration growth
on employment growth ....................................................................... 120
IV.6 Dominican Republic: impact of increased broadband
penetration on employment growth .................................................. 121
IV.7 Costa Rica: impact of broadband on growth
in real household income ..................................................................... 122
IV.8 Colombia: impact of broadband penetration growth
on real household income growth ...................................................... 123
IV.9 Latin America: estimated economic impact of digitization ............ 129
IV.10 Latin America: broadband supply and demand gap, 2011 ............. 130
V.1 Model parameters ................................................................................. 149
V.2 Regression and variance analysis statistics ....................................... 149
VI.1 Change in the number of employed persons due to the
introduction of cloud computing in Argentina ................................ 181
VI.2 Business creation due to the introduction of cloud
computing in Argentina ....................................................................... 181
VI.3 Change in number of employed persons due to the introduction
of cloud computing in Brazil............................................................... 181
VI.4 Business creation due to the introduction of cloud
computing in Brazil .............................................................................. 182
VI.5 Job creation and business creation in the United States and
in the 27 European Union member countries ................................... 182
VII.A.1 ................................................................................................................ 112
VIII.1 Drivers of broadband development in the OECD
and the Republic of Korea ................................................................... 227
VIII.2 ICT sector trade balance, 2010-2011 ................................................... 234
VIII.3 Republic of Korea: master plans and strategic
frameworks for broadband development ......................................... 237
VIII.4 Broadband regulatory policies ............................................................ 239
VIII.5 The Giga Korea Plan 2020 .................................................................... 244
VIII.6 Industrial and technological development model in
Latin America and the Republic of Korea, 1960-2020 ..................... 247
IX.1 Internet services according to delay sensitivity, bandwidth
consumption and economic value...................................................... 255
IX.2 Global lP traffc projections, 2010-2015 .............................................. 257
IX.3 Global consumer Internet traffc, 2010-2015 ..................................... 258
X.1 Characteristics of cloud computing ................................................... 278
X.2 Comparison between conventional models and
cloud computing ................................................................................... 288
X.3 Chilean regulation of cloud computing ............................................ 297
X.4 Mobile cloud computing challenges and solutions ......................... 302
XI.1 Description of over-the-top-services .................................................. 321
XI.2 Service bundles including TV and frst-generation strategies ....... 328
XI.3 Second-generation strategies .............................................................. 330
ECLAC 10
Figures
I.1 Internet user penetration in 2011 .......................................................... 32
I.2 Fixed and mobile broadband penetration in 2011 ............................. 33
I.3 Fixed and mobile broadband penetrationin Latin America
and the Caribbean and in OECD, 2006-2011 ....................................... 34
I.4 2G and 3G coverage of the population in 2010 .................................. 35
I.5 Households with Internet access in urban areas,
rural areas and nationally ..................................................................... 35
I.6 Households with Internet access, by income quintile ....................... 36
I.7 Households with Internet access, by gender
of the head of household ....................................................................... 37
I.8 1Mbps fxed broadband tariffs with relation
to per capita GDP in 2012 ...................................................................... 38
I.9 Mobile broadband tariffs in relation to per capita GDP in 2012 ...... 38
I.10 Broadband connection speeds in 2012 ................................................. 39
I.11 Percentage of broadband connections, by speed range .................... 40
I.12 Bandwidth connected to the United States ......................................... 41
II.1 United States: states with the lowest broadband
penetration rates, 2010 ........................................................................... 40
II.2 Quantifcation of the fxed and mobile demand gap......................... 36
II.3 Latent fxed and mobile broadband demand
in Latin America, 2011 ............................................................................ 42
II.4 Brazil: reasons for not subscribing to Internet
in the home, 2007-2011 ........................................................................... 45
II.5 Brazil: home Internet adoption by socioeconomic
level (SEL), 2011 ...................................................................................... 46
II.6 Mexico: home Internet adoption by income decile, 2008-2010 ........ 47
II.7 Home Internet adoption gap by income quintile (Q5/Q1) .............. 48
II.8 Costa Rica: residential Internet and computer use by educational
level of the household head, 2010 ........................................................ 49
II.9 Chile: residential Internet and computer use by educational
level of the household head, 2009 ........................................................ 49
II.10 Costa Rica: residential broadband penetration by education
level and income decile, 2010 ................................................................ 50
II.11 Costa Rica: home Internet and computer use,
by age group, 2010 .................................................................................. 51
II.12 Latin America: concentration of fxed broadband supply
and average download price per Mbps ............................................... 54
III.1 Median monthly prices on North Pacifc routes for 10 Gbps,
second quarter 2010 to second quarter 2011 ....................................... 80
III.2 Trends in median prices in major cities of Latin America ................ 81
III.3 Trends in median prices in major world cities ................................... 82
IV.1 Comparative contribution of broadband to economic growth ...... 109
IV.2 Typology of countries by stage of digitization, 2011 ...................... 117
IV.3 Digitization and economic growth ..................................................... 118
Broadband in Latin America: Beyond Connectivity 11
IV.4 Returns to scale from digitization ...................................................... 119
IV.5 Digitization and employment ............................................................. 120
IV.6 Digitization and innovation in a sample of 125 countries .............. 121
V.1 Mobile telephony penetration rate in the United States and
the European Union compared with Latin America ...................... 129
V.2 Statistical signifcance of the parameters, 1995-2011 ....................... 133
V.3 Number of 3G networks in operation in Latin America ................. 134
V.4 Comparison between network launches ........................................... 135
V.5 Growth in the number of 3G users ..................................................... 136
V.6 Penetration of 3G services as of June 2012 ........................................ 136
V.7 Correlation of penetration with time in service and
GDP per capita ...................................................................................... 137
V.8 Relationship between GDP per capita and time
in service as of June 2012 ..................................................................... 138
V.9 Average quarterly increase in penetration, by network age .......... 139
V.10 Cost in economic well-being of a one-year delay in the launch
of high-speed networks ....................................................................... 142
V.11 Cost in economic well-being of a one-quarter delay
in the launch of high-speed networks ............................................... 143
VI.1 Transition of the main macroeconomic variables from the steady
state of the pre-cloud economy to that of the cloud economy ....... 170
VIII.1 OECD: fxed-line broadband subscriptions (wired)per 100
inhabitants, by technology, December 2011 ...................................... 215
VIII.2 Selected OECD countries: fxed broadband penetration ................ 216
VIII.3 OECD: ratio between broadband penetration per 100 inhabitants
and GDP per capita in PPP dollars, 2011 ........................................... 217
VIII.4 Broadband rates per Mbps of advertised speed,
September 2011 ...................................................................................... 217
VIII.5 Business adoption of broadband, 2010 or latest available year ..... 218
VIII.6 Fixed-line telephone subscribers by service provider ..................... 221
VIII.7 Number of subscribers by service and mobile technology ............ 222
VIII.8 Mobile service providers: number of subscribers
and market share .................................................................................. 223
VIII.9 High-speed broadband providers: number of
subscriptions and market share .......................................................... 224
VIII.10 High-speed broadband coverage subscribers by technology ....... 225
IX.1 Fixed broadband access, by technology, 2011 ................................... 251
X.1 Why cloud computing? ........................................................................ 281
X.2 Benefts of cloud computing in fve European countries ................ 283
X.3 Savings from cloud computing ........................................................... 284
X.4 Cloud computing regulation preparedness scorecard .................... 288
X.5 Adoption of cloud computing around the world ............................ 290
X.6 Growth of cloud computing in Brazil by level of processing ......... 290
X.7 Cloud services used in Colombia ....................................................... 291
X.8 Outlook for data centre growth in Latin America ........................... 291
ECLAC 12
X.9 Access speed by type of connection in the region ........................... 295
X.10 Mobile cloud computing revenue by region .................................... 298
Diagrams
II.1 Fixed and mobile broadband substitution
and complementarity ............................................................................. 35
II.2 Impact of fscal policy on broadband penetration ............................. 55
IV.1 Economic impact of broadband .......................................................... 104
IV.2 How broadband contributes to job creation ..................................... 110
IV.3 Components of the digitization index .............................................. 117
VIII.1 Analytical framework for broadband industrial policy .................. 211
VIII.2 Main areas of broadband industrial policy ....................................... 233
X.1 Cloud computing model ...................................................................... 266
X.2 Flexible platforms for adjusting data and applications
in the cloud ............................................................................................ 268
X.3 Development of cloud computing and software
platform infrastructure (SPI) ............................................................... 270
X.4 Service layers defnition ....................................................................... 275
X.5 The extended data centre value chain ............................................... 286
X.6 Architectures of mobile cloud computing ........................................ 292
X.7 Architecture of a CloneCloud system ................................................ 297
X.8 Evolution of computing technologies and networks ...................... 299
X.9 Action areas for promoting cloud computing .................................. 304
XI.1 Map of agents ........................................................................................ 315
Boxes
V.1 Service penetration and unique subscribers ..................................... 130
VIII.1 Broadband industrial policy approaches .......................................... 228
Maps
III.1 Capacity of principal international routes in Latin America ............ 76
X.1 Average Internet connection speed around the world .................... 273
Foreword
In November 2010, the editors of this book published Fast-tracking the digital
revolution: Broadband for Latin America and the Caribbean, which put forth
that broadband is the cornerstone of a system for economic, organizational
and social innovation that, in conjunction with complementary assets
(infrastructure, skills, production structure), was driving a positive
dynamic across all economic and social sectors.
They noted that achieving that synergy called for a new policy
approach with a comprehensive, fexible, long-term view combining
the objectives of greater productivity, innovation, social inclusion and
sustainability. This approach needed for the State to play an active role,
building capacities that would enable it to design instruments and
coordinate actions for addressing the challenges of the information society.
The editors saw that technological change was gathering
momentum, particularly in information and communication technologies
(ICTs). Their judgment was borne out by the growing penetration of 3G
telephony among the population, expanding access to broadband and
the spread of new technologies, such as cloud computing and big data
analytics. There was a clear need to step up efforts in order to meet the
demands of a hyperconnected world.
The countries of Latin America and the Caribbean have responded
to this call. The region’s public and private sectors have boosted investment
in the infrastructure needed to deploy new networks and in programmes
which encourage broadband use by individuals and businesses.
ECLAC 14
Despite this progress, the efforts have not been enough. Although
access to fxed and mobile broadband has expanded, the digital divide
with the advanced countries is still far from closed. And broadband use is
concentrated in applications for personal use and basic ICTs for business.
The impact on productivity is, therefore, low.
There are still signifcant economic, territorial and gender
disparities in terms of access, and bridging these gaps is going too slowly.
Digital policies currently in force do not attach due importance to making
infrastructure building and mass uptake of broadband part of national
industrial policy strategies.
This is crucial, because the structural change advocated by the
Economic Commission for Latin America and the Caribbean (ECLAC)
depends on a close link between digital strategies, broadband development
and sectoral industrial policy design. Only then will it be possible to make
signifcant strides towards a new, more knowledge-intensive production
structure that can generate the high-quality jobs needed for making steady
progress towards greater equality and more sustainable economic and
social development.
This publication, the result of a joint effort by the Regional Dialogue
on the Information Society (DIRSI) and the Division of Production,
Productivity and Management of ECLAC, was undertaken as part of
a research and policy proposal programme on issues relating to the
information society. ECLAC has been working on this effort since 2009,
with fnancial support from the European Union through the Inclusive
Political Dialogue and Exchange of Experiences Project of the @LIS2
Programme (Alliance for the Information Society phase 2).
DIRSI and ECLAC have made this publication available to
governments and citizens of the countries of the region in an effort to
provide an overview of the latest broadband developments in Latin
America and of the new economic and social challenges and opportunities
coming from ever faster technological change.
Alicia Bárcena
Executive Secretary
Economic Commission for Latin
America and the Caribbean (ECLAC)
First part
Diagnosis
“Do. Or do not.
There is no try.”
Yoda D´Kana
Chapter I
The shifting digital paradigm in Latin America
Roxana Barrantes, Valeria Jordán and Fernando Rojas
1
Since the release of Fast-tracking the digital revolution: Broadband for Latin
America and the Caribbean, governments in the region have accelerated
policy design and implementation to expand broadband access for the
public and for businesses (Jordán, Galperín and Peres, 2010). Two years
is a relatively short time in which to evaluate progress on expanding the
infrastructure that is so important for broadband development. And the
growing array of promising services based on applications requiring
broadband increasingly calls for urgent policy measures. However,
because of advances in Internet-based applications and services, policies
alone will not expand access. While important, they must be viewed in
a broader context within a vision of future that considers the impact
of the hyperconnected world on the global economy (World Economic
Forum, 2012).
Because of its permeable and convergent effects on the whole of the
economy and society, broadband infrastructure must be viewed not only
as a connectivity tool but also as a means of pursuing policies designed
to enhance social inclusion and economic competitiveness. The timing of
1
Roxana Barrantes is the chief researcher at the Instituto de Estudios Peruanos in Lima;
Valeria Jordán and Fernando Rojas are staff members of the Economic Commission for
Latin America and the Caribbean (ECLAC), in Santiago. Fernando Rojas is coordinator of
the Regional Broadband Observatory (ORBA).
ECLAC 18
policymaking is crucial in view of the shifting digital paradigm,
2
which
has transformed models of production, organization and social interaction
faster than any previous technological paradigm.
The commercialization of the Internet in the mid-1990s and its
expansion thanks to narrowband technology brought about radical changes,
particularly in the feld of communication, with e-mail applications and
the proliferation of online sites and information in hypertext format based
on the World Wide Web. A frst generation of policies emerged, focused
primarily on expanding Internet and computer access in the countries of
Latin America, which in the early 2000s lagged far behind in adoption of
these technologies. This gap came to be called the “digital divide”.
Between 2005 and 2010, major advances in access technologies led
to signifcant increases in data transmission speeds. Broadband emerged,
bringing with it network, device and content convergence and enabling
the development of cloud applications. These are more interactive and
audio- and video-intensive and can deliver all types of services online,
ranging from gaming and entertainment to business and social interest.
At the global level, this ushered in a second generation of access policies:
those relating to broadband.
Today, technology paths are consolidating as high-speed access
networks continue to develop and devices for accessing network services
become more widely available. The combination of cloud services and
total mobility is the backdrop for this book and this introduction. The
frst section describes the main ongoing trends, focusing on the explosive
quantitative and qualitative proliferation of structured and unstructured
data that is a virtually automatic by-product of daily activities (digital
exhaust or digital footprint). The second section describes in detail the
situation of broadband services and infrastructure in the region. And the
third section summarizes the content of the subsequent chapters and the
main conclusions of the book.
A. The era of cloud computing
As Jordán, Galperín and Peres (2010) suggested, the digital paradigm
must be examined as a set of general-purpose technologies evolving in
an asynchronous manner, with constant feedback innovations creating a
virtuous circle which generates a very dynamic technology system. Data
digitalization enables for four basic activities: generation and acquisition
of data in various formats (text, audio and video), transmission, computing
and storage. These functions are closely linked and interdependent and
make up the information and communication technologies (ICT) system.
2
See chapter V by Ernesto Flores-Roux, on mobile broadband.
Broadband in Latin America: Beyond Connectivity 19
The increase in information generation and collection, especially in
audio and video format, is exerting pressure on the three other activities,
which have different response times. These path asymmetries are at the
same time obstacles to the full use of each component in isolation and
incentives for the lagging paths to catch up with the most advanced one.
For example, throughout most of the 1990s computing capacity was far
ahead of transmission and storage capacity. The situation started to reverse
in the early 2000s, when broadband drove system development and
storage bottlenecks appeared. These are now easing as cloud computing
progresses. The end result of this process was an increase in the operating
capacity of devices and a reduction in costs, which are high considering
the cost per unit of service provided.
3
The amount of data which individuals and organizations generate,
process, consume and store has grown exponentially in recent years. Two
forces are driving this trend. The frst force is direct information generation
by users, facilitated by the proliferation of devices and convergent systems,
such as smartphones, tablets and phablets, netbooks, smart TV, social
networks and audio and video applications. The second is the information
emerging from interactions between individuals and objects, between
objects (M2M) and between Internet-connected devices with sensors (the
Internet of Things). This information can be mined to customize and target
activities such as marketing, advertising and other services connected to
social media and social graphing, which compile and use this information
on individual behaviour as a basic tool for advanced manufacturing and
new business processes such as crowdsourcing.
The amount of digital information accumulated doubles every 20
months, echoing Moore’s Law, which states that microprocessor computing
capacity doubles every 18 months. According to Gantz and Reinsel (2011),
over 1.8 zettabytes (1.8 trillion gigabytes) of information were created
and replicated in 2011, having grown more than nine-fold in fve years.
Each second of high-defnition video generates 2,000 times more bytes
than are needed to store one page of text. Businesses generate trillions of
bytes of transactional information from their clients, suppliers and other
operations. For example, according to McKinsey (2011), in 2011Walmart’s
data warehouses had a capacity of over 2.5 petabytes (about half the letters
delivered by the United States Postal Service in 2010).
This mass of information is not only coming directly from users
but also from computers and objects, which in 2011 generated more data
3
For example, in 1997 one gigaflop of processing power cost US $42,000; by 2011 the cost
had fallen to $ 1.80. In 1980, one gigabyte of storage cost about US$ 200,000; by 2011 the
cost of one terabyte (1,024 gigabytes) had fallen to US$ 100. Thus two basic conditions
were met for the artifacts spearheading a new technological and productive paradigm:
ongoing technical progress combined with plummeting costs.
ECLAC 20
than people. According to Cisco (2012), that year almost one quarter (22%)
of IP traffc originated from non-PC digital devices; the fgure is expected
to increase to 31% by 2016. It is estimated that 95% of the data explosion
consists of unstructured information, such as voice and video messages,
rather than structured information consisting of words and numbers
(The Economist, 2010).
In addition to being one of the driving forces of the digital system,
this trend has become especially important in the last fve years as it takes
the form of big data, which must be the subject of new analytics going
beyond the traditional management of relational databases (SQL). Here
too, there have been huge cost reductions. In 2011, new technologies such
as data deduplication, compression and analysis tools had driven the
cost of creating, capturing, managing and storing information down to
one sixth of what it was in 2005.
4
In addition, the increase in the amount
of information was not accompanied by an equivalent increase in the
capacity to process and store it, creating what is known as “data deluge”.
This data explosion is signifcant in the context of three major
trends: the spread of broadband and emergence of high-speed broadband,
cloud computing, and big data management and analysis.
The spread of broadband (the technology for dedicated Internet access
allowing high-speed data transfer)
5
is snowballing. It is the outcome of
new technologies feeding back (supply factor) to increase transmission
speed over a short period of time, and of burgeoning demand fuelled by
more advanced applications made possible by higher-speed, lower-cost
devices and programmes. The number of Internet users increased from
nearly 360 million in 2000 to 2.7 billion 13 years later. In 2013, broadband
subscriptions (an option which did not exist in 2000) numbered 2.8 billion.
In the early days of commercial Internet (the 1980s and 1990s), the
only access mode was dial-up modem over conventional telephone lines at
a maximum speed of 56 kbps. With this type of connection, it would have
taken about 12 minutes to download one song (5 MB) and over 28 hours
to download a low-resolution movie (700 MB). In this context, Internet use
was basically limited to exchanging e-mails and text fles, as well as web
surfng, following the emergence of the World Wide Web in mid-decade.
4
See [online] http://www.emc.com/leadership/programs/digital-universe.htm.
5
A broadband connection is generally defined as one with a transmission speed of at
least 256 kbps. However, in 2011, the Broadband Commission of the International
Telecommunication Union (ITU) fine-tuned some of the central concepts of broadband, as
a service that is always on (a user does not need to reconnect to a server each time in order
to go online) and has a large capacity, able to carry a large amount of data per second,
instead of adopting a set data transmission speed.

http://www.broadbandcommission.org/Reports/Report_2.pdf. See also the second
annex to chapter III, by Omar de León.
Broadband in Latin America: Beyond Connectivity 21
In the 2000s, the development of cable modem allowed broadband
connectivity at theoretical speeds ranging from 256 kbps to 20 Mbps. But
high tariffs limited its expansion despite the improved user functionalities
associated with better connectivity. It was not until 2005-2007 that
broadband came into widespread use, with fxed (generally, wired) and
then mobile technologies allowing theoretical transmission speeds of over
100 Mbps.
It was against this backdrop that the concept of cloud computing
emerged over the last fve years as a way to combine distributed computing
with access to data centres over a network. This made it possible to expand
utility computing and to access advanced converged applications and
services delivered in real time by streaming (ITU, 2009).
Cloud computing is a way of providing convenient, demand-based
access over a network to a shared set of confgurable computing resources
(for example, networks, servers, storage, applications and services) which
can be provided and released rapidly with minimal management effort or
interaction with a service provider. As noted by René Bustillo in chapter X
hereof, this model has fve main features: on-demand self-service, rapid
network access, resource pooling, elasticity and measured service. It
provides three types of service: Software as a Service (SaaS), Platform as a
Service (PaaS) and Infrastructure as a Service (IaaS). These may be private,
community, public or hybrid.
6
Cloud computing is forecast to grow by 18% a year between 2010
and 2016, while the information technology market is expected to grow no
more than 4%. However, by the end of this period, public cloud services
will account for only 5% of all services connected with these technologies
(Gartner, 2012). By 2014, it is expected that cloud services in the United
States will be dominated by SaaS (62% of total revenue of almost 12 billion
dollars), followed by IaaS (34%) and PaaS (only 4%) according to the March
2011 issue of Telecom Intelligence Series. The importance of SaaS led the
technical report Above the Clouds: A Berkeley View of Cloud Computing to say
that “cloud computing” refers both to applications delivered as services
over the Internet and to hardware and software systems at data centres
providing those services (Armbrust and others, 2009).
Migration to cloud computing occurs for many reasons
—usually because it costs less than operating a proprietary computing
infrastructure. In addition, it is easier to meet demand peaks because of
the greater fexibility allowed by the conversion of fxed costs (equipment)
into variable costs (lease of service based on use), economies of scale and
6
The essential facilitating technologies are rapid wide area networks (WANs), low-cost
servers and virtualization for high-performance commodity hardware.
ECLAC 22
of use, and lower costs of entering new markets with consequent positive
effects on the creation of businesses, particularly small ones, job creation
and competency building.
7

Evaluating the benefts of cloud computing is a complex proposition
because it can impact different sectors of the economy in different ways.
The main benefciaries of the growing adoption of cloud computing are,
in addition to the businesses implementing it, the providers of storage,
virtualization, networked services and security.
Like any paradigm shift, cloud computing poses problems and
challenges when frst implemented in ICT environments which were used
to keeping “everything under control”. Many potential users report that
concerns about portability (avoiding vendor lock-in) and data security
and privacy make them reluctant to use cloud services for sensitive data.
There is widespread frustration among market participants regarding the
regulatory framework, especially in respect to data privacy and support
system reliability, such as data centre power supply (Telecom, 2012).
Big data analytics for decision-making concerns data sets which
exceed the capacity of typical database software tools to capture, store,
manage and analyze information.
8
It is a response to the exploding
quantity (speed and frequency) and diversity of digital data generated
in real time because of the growing importance of technology in daily
activities. Often these data are semi-structured or unstructured, such as
data obtained from voice communications or images.
9
In this universe,
continuous data stream mining, using tools to extract data and process
them in real time (in other words, in time to change decisions before they
become irreversible), plays a crucial role, as does the combination of rapid
data streaming with slower access to historical data warehouses.
Of course, the main use of the analytics is to generate information
and knowledge based on complete information in real time. Dealing with
this new era and its abundance of data provides an opportunity to create
7
See chapter VI, by Andrea Colciago and Federico Etro.
8
Technically, big data analytics refers to tools and methodologies for converting huge
volumes of raw data into “data about data” for analytical purposes. It combines algorithms
(to detect patterns, trends and correlations with different time horizons) with advanced
visualization techniques. It often requires high-performance computing systems —in
other words, systems operating at more than one teraflop (10
12
), with parallel processing
in order to perform advanced applications (particularly applications for transaction
processing and data warehousing) efficiently, reliably and rapidly.
9
The data in question come, for example, from purchases and transactions (including
credit card information), corporate operations, information retrieval (querying, tracking,
history), social networks (identity, friendships), personal interests (tastes, favourites,
links), location, physical sensors (GPS, traffic patterns, Internet of Things) and content
(SMS, voice calls, e-mails) —in other words, information generated by traditional sources,
particularly businesses and individuals, in their daily activities.
Broadband in Latin America: Beyond Connectivity 23
value through innovation and increased effciency and competitiveness,
to anticipate and learn about behaviour and to increase the consumer
surplus and individual and collective well-being. For example, estimates
put the revenue generated by business data analysis in 2010 upwards of
US$ 100 billion, growing at an annual rate of 10%.
Big data analytics creates value mainly by enabling greater market
and population segmentation in order to target supply and products. It
also allows greater innovation in business models, products and services,
which improves existing goods, facilitates development of new products
(combining mass production with customization) and new business
and government service models. In general, in addition to enhancing
transparency and effciency because data are shared, it yields a better and
more timely analysis of the performance of organizations of all kinds and
makes it possible to adjust structures and behaviours in real time.
In the area of economic and social development, as suggested by the
United Nations Global Pulse initiative, big data analytics turns imperfect,
unstructured and complex data on the well-being of people into actionable
information which narrows time and information gaps for public policy
decision-making, providing a timely response to specifc situations and for
rapid feedback on the effectiveness of policy actions (Global Pulse, 2012).
Latin America is only just beginning to understand and disseminate
this technology path, although it is often used by major fnancial and
retail sales organizations, as well as by public sector agencies dealing with
taxation and security. According to McKinsey (2011), the geographical
distribution of new data stored in 2010 places the region on the sidelines,
with just over 50 petabytes out of a total of about 7,000 petabytes throughout
the world, of which 5,500 are stored in the United States and Europe.
Big data analytics is not without problems, particularly because of
differences in data retrieval and mining capacity and the lack of incentives
for businesses to open their data on a scale similar to what has been done
by governments (open government data) and social networks. As in the
case of cloud computing, there is also serious concern about privacy and
the extent to which data are anonymized.
10
All these paths (data explosion, cloud computing, big data analytics)
are based on and dependent on the deployment and use of high-speed
network infrastructure.
10
There are two additional, technical, issues: much of the data from sources like social
networks reflect perceptions, intentions or desires and not facts; and the increase in type
I statistical error and the resulting apophenia (seeing patterns where there are none)
because in huge volumes of data, connections open up in all directions.
ECLAC 24
B. Broadband in the region
If the opportunities described in the preceding section are to become a
reality for the region, policies must be based on the state of broadband.
This section identifes and explains the challenges using a set of indicators
shown below.
On average, only 40% of the total population of Latin America
uses the Internet, compared with almost 80% in the OECD countries as
a whole.
11
However, the averages mask substantial inequalities, as shown
by the data. Countries such as Mexico and Peru, which are important
economically, are below average. And no countries in the region even
reach the levels of Spain, which is one of the OECD countries lagging
farthest behind. Chile and Uruguay, with the highest percentage, barely
exceed 50%. In countries such as Nicaragua and Guatemala only 10% of
the population uses the Internet.
Figure I.1
INTERNET USERS PENETRATION IN 2011
(Percentages)
0
10
20
30
40
50
60
70
80
90
100
N
i
c
a
r
a
g
u
a
G
u
a
t
e
m
a
l
a
H
o
n
d
u
r
a
s
E
l

S
a
l
v
a
d
o
r
P
a
r
a
g
u
a
y
B
o
l
i
v
i
a

(
P
l
u
r
.

S
t
a
t
e

o
f
)
E
c
u
a
d
o
r
D
o
m
i
n
i
c
a
n

R
e
p
.
M
e
x
i
c
o
P
e
r
u
L
a
t
i
n

A
m
e
r
i
c
a

a
n
d
t
h
e

C
a
r
i
b
b
e
a
n
V
e
n
e
z
u
e
l
a

(
B
o
l
.

R
e
p
.

o
f
)
C
o
l
o
m
b
i
a
P
a
n
a
m
a
B
r
a
z
i
l
A
r
g
e
n
t
i
n
a
U
r
u
g
u
a
y
C
h
i
l
e
S
p
a
i
n
O
E
C
D
U
n
i
t
e
d

S
t
a
t
e
s
J
a
p
a
n
F
r
a
n
c
e
U
n
i
t
e
d

K
i
n
g
d
o
m
G
e
r
m
a
n
y
R
e
p
.

o
f

K
o
r
e
a
F
i
n
l
a
n
d
Source: Economic Commission for Latin America and the Caribbean (ECLAC), on the basis of International
Telecommunication Union (ITU), World Telecommunications Indicators Database, 2012.
Note: The data for OECD do not include Chile or Mexico.
The number of users is clearly related to broadband connectivity,
and this may, in turn, depend on whether the connection is fxed or mobile.
11
In order to avoid duplication, in this chapter the data for OECD do not include Chile
or Mexico.
Broadband in Latin America: Beyond Connectivity 25
Figure I.2 shows the importance of mobile connections with respect to
fxed connections. In OECD countries there are, on average, almost twice
as many mobile connections as fxed ones. But in Latin America and the
Caribbean there are only slightly more mobile ones, although a number
of countries still have mostly fxed connections. There could be an inverse
correlation between Internet usage penetration and mobile broadband
dissemination. In six of the seven countries in the region with lower
penetration rates (Ecuador, El Salvador, Guatemala, Honduras, Nicaragua,
Paraguay and the Plurinational State of Bolivia) mobile broadband is more
widespread than fxed broadband. This could be due to the insuffcient
deployment of wired networks or to the cost of the various types of
connection, as will be seen later when examining service affordability.
Figure I.2
FIXED AND MOBILE BROADBAND PENETRATION IN 2011
(Percentages)
20 40 60 80 100
Nicaragua
Trinidad and Tobago
Peru
Jamaica
Bolivia (Plur. State of)
Costa Rica
El Salvador
Honduras
Colombia
Guatemala
Venezuela (Bol. Rep. of)
Paraguay
Mexico
Dominican Rep.
Uruguay
China
Ecuador
Latin America and the Caribbean
Argentina
Panama
Chile
Brazil
Portugal
Italy
Germany
Spain
Australia
New Zealand
OECD
United Kingdom
United States
Denmark
Finland
Japan
Rep. of Korea
Mobile broadband Fixed broadband
0
Source: Economic Commission for Latin America and the Caribbean (ECLAC), on the basis of International
Telecommunication Union (ITU), World Telecommunications Indicators Database, 2012.
Note: The data for OECD do not include Chile or Mexico.
ECLAC 26
The fxed and mobile broadband penetration rates for 2011 were
reached from different starting points (2006) and at different speeds in
OECD countries and in the region, as shown in fgure I.3. During the
fve-year period from 2006, mobile broadband penetration in OECD grew
exponentially and caught up with fxed broadband penetration in the space
of two years. In Latin America and the Caribbean the process was slower;
mobile connections took four years to catch up with fxed connections.
Figure I.3
FIXED AND MOBILE BROADBAND PENETRATION IN LATIN AMERICA
AND THE CARIBBEAN AND IN OECD, 2006-2011
(Percentages)
Latin America and the Caribbean fixed broadband OECD fixed broadband
Latin America and the Caribbean mobile broadband OECD mobile broadband
0
10
20
30
40
50
60
2006 2007 2008 2009 2010 2011
Source: Economic Commission for Latin America and the Caribbean (ECLAC), on the basis of International
Telecommunication Union (UIT), World Telecommunications Indicators Database, 2012.
Note: The data for OECD do not include Chile or Mexico.
In addition to considering these connectivity indicators, it is
important to identify the type of network available —second generation
(2G), with limited access to data, or third generation (3G), allowing more
high speeds— and coverage, expressed as a percentage of the population
with potential access to a mobile network. Figure I.4 shows the major
differences in 3G coverage in fve selected countries in the region
(Argentina, Brazil, Colombia, Mexico and Peru) in 2010. While almost all
the population in those countries has 2G network coverage, Mexico and
Colombia lag signifcantly behind in investing in 3G networks, with only
39% and 52%, respectively, of the population covered by this technology.
National data mask substantial territorial (between the urban and
rural population), socioeconomic (between income quintiles) and gender-
based inequalities. The fgures below clearly show these inequalities
within countries, using household connection rates as the unit of analysis.
Broadband in Latin America: Beyond Connectivity 27
Figure I.4
2G AND 3G COVERAGE OF THE POPULATION IN 2010
(Percentages)
99
97
94
93
95
75 75
52
39
65
Argentina Brazil Colombia Mexico Peru
2G 3G
Source: Latin American Mobile Observatory (GSMA), 2011.
At the territorial level, where disaggregated data are available,
households in rural areas of the countries of the region clearly lag behind
with regard to Internet access. The exception are Uruguay and Costa Rica,
where over 10% of rural households have access, compared with an urban
access rate of 31% and 35.6%, respectively. In none of the other countries for
which data are available does rural access even reach 7% (see fgure I.5).
Figure I.5
HOUSEHOLDS WITH INTERNET ACCESS IN URBAN AREAS,
RURAL AREAS AND NATIONALLY
(Percentages)
38.0
34.7
28.6
24.0
19.3
13.8
12.0
8.2
8.0
10.0
15.4
6.8
10.9
1.6
2.4
0.3
1.1
0.9
43.0
35.6
31.9
31.7
24.4
21.3
17.9
15.5
11.8
0
5
10
15
20
25
30
35
40
45
50
Brazil
2011
Uruguay
2010
Chile
2009
Costa Rica
2010
Colombia
2010
Paraguay
2010
Peru
2010
Honduras
2010
El Salvador
2010
Total Rural Urban
Source: Economic Commission for Latin America and the Caribbean (ECLAC), Observatory for the
Information Society in Latin America and the Caribbean (OSILAC), on the basis of information from
household surveys conducted by the national statistical institutes. Most recent available year.
ECLAC 28
As regards the socioeconomic dimension, dividing the population
into income quintiles reveals even more signifcant inequalities in access
(see fgure I.6). In Chile, the penetration rate for quintile V is almost 70%;
for quintile I it is below 10%. In the country with the lowest penetration
rate (El Salvador), quintile V is below 30%, quintile IV is below 10%, and
there is almost no penetration in quintile I. These inequalities refect the
fact that, in several countries of the region (for example, Ecuador and
Peru), the main point of access are telecenters, where connection speeds
tend to be slower.
Figure I.6
HOUSEHOLDS WITH INTERNET ACCESS, BY INCOME QUINTILE
(Percentages)
0 20 40 60 80
Venezuela (Bol. Rep. of) 2010
Brazil 2009
Chile 2009
Costa Rica 2010
Colombia 2010
Mexico 2010
Honduras 2010
Paraguay 2010
Peru 2010
El Salvador 2010
Uruguay 2010
Venezuela (Bol. Rep. of) 2010
Brazil 2009
Chile 2009
Costa Rica 2010
Colombia 2010
Mexico 2010
Honduras 2010
Paraguay 2010
Peru 2010
El Salvador 2010
Uruguay 2010
Venezuela (Bol. Rep. of) 2010
Brazil 2009
Chile 2009
Costa Rica 2010
Colombia 2010
Mexico 2010
Honduras 2010
Paraguay 2010
Peru 2010
El Salvador 2010
Uruguay 2010
Venezuela (Bol. Rep. of) 2010
Brazil 2009
Chile 2009
Costa Rica 2010
Colombia 2010
Mexico 2010
Honduras 2010
Paraguay 2010
Peru 2010
El Salvador 2010
Uruguay 2010
Venezuela (Bol. Rep. of) 2010
Brazil 2009
Chile 2009
Costa Rica 2010
Colombia 2010
Mexico 2010
Honduras 2010
Paraguay 2010
Peru 2010
El Salvador 2010
Uruguay 2010
Q
u
i
n
t
i
l
e

I
Q
u
i
n
t
i
l
e

I
I
Q
u
i
n
t
i
l
e

I
I
I
Q
u
i
n
t
i
l
e

I
V
Q
u
i
n
t
i
l
e

V
Source: Economic Commission for Latin America and the Caribbean (ECLAC), Observatory for the
Information Society in Latin America and the Caribbean (OSILAC), on the basis of information from
household surveys conducted by the national statistical institutes. Most recent year available.
Note: Households with Internet access as a percentage of total households in each quintile.
Broadband in Latin America: Beyond Connectivity 29
Lastly, data on household Internet access according to whether
the head of household is a man or a woman reveal clear inequalities
(see fgure I.7). Chile has the widest gender gap: households headed
by men have a penetration rate of over 30%; the rate for those headed
by women is below 25%. At the other extreme, the gap is very small or
non-existent in countries such as El Salvador and Honduras, where total
penetration is extremely low.
Figure I.7
HOUSEHOLDS WITH INTERNET ACCESS, BY GENDER
OF THE HEAD OF HOUSEHOLD
(Percentages)
0
5
10
15
20
25
30
35
Chile
2009
Brazil
2009
Costa Rica
2010
Mexico
2010
Colombia
2010
Peru
2010
El Salvador
2010
Honduras
2010
Men Women
Source: Economic Commission for Latin America and the Caribbean (ECLAC), Observatory for the
Information Society in Latin America and the Caribbean (OSILAC), on the basis of information from
household surveys conducted by the national statistical institutes. Most recent year available.
Note: Households with Internet access as a percentage of total households, by gender of the head of
household.
These gaps, particularly the socioeconomic gap, refect an
affordability problem. Internet and broadband connections are very
expensive with relation to household income levels, or they require
consumption of additional goods such as computers or cell phones,
which are also costly. Figure I.8 shows an indicator expressing tariffs for
1 Mbps fxed broadband as a percentage of per capita GDP, which is a
measurement of personal income. In the more advanced countries, such as
the United Kingdom, Japan, France, Spain, Italy and Portugal, this fgure
is lower than 0.20%. In the highest-ranked countries of the region, such
as Uruguay, Panama, Chile, Argentina, Mexico, Brazil and Costa Rica,
the fgure is between 1% and 2%; for the lowest-ranked ones it is over
10% (11% in Guatemala and 16% in Nicaragua) and as high as 31% in the
Plurinational State of Bolivia.
ECLAC 30
Figure I.8
1MBPS FIXED BROADBAND TARIFFS WITH RELATION
TO PER CAPITA GDP IN 2012
(Percentages)
0.03 0.06 0.06 0.13 0.15 0.19
1.00 1.04 1.26 1.45 1.54 1.69
1.93
3.22
3.63
3.88
4.43
6.50 6.59
9.12
11.61
16.65
31.42
U
n
i
t
e
d

K
i
n
g
d
o
m
J
a
p
a
n
F
r
a
n
c
e
S
p
a
i
n
I
t
a
l
y
P
o
r
t
u
g
a
l
U
r
u
g
u
a
y
P
a
n
a
m
a
C
h
i
l
e
A
r
g
e
n
t
i
n
a
M
e
x
i
c
o
B
r
a
z
i
l
C
o
s
t
a

R
i
c
a
C
o
l
o
m
b
i
a
V
e
n
e
z
u
e
l
a

(
B
o
l
.

R
e
p
.

o
f
)
E
c
u
a
d
o
r
E
l

S
a
l
v
a
d
o
r
P
e
r
u
P
a
r
a
g
u
a
y
H
o
n
d
u
r
a
s
G
u
a
t
e
m
a
l
a
N
i
c
a
r
a
g
u
a
B
o
l
i
v
i
a
(
P
l
u
r
.

S
t
a
t
e

o
f
)
Source: Economic Commission for Latin America and the Caribbean (ECLAC), Regional Broadband
Observatory (ORBA), on the basis of tariffs published by operators as at September 2012.
But the large gaps between developed countries and the countries of
the region narrow somewhat when the indicator is for mobile broadband.
Although no country is below 0.5%, none are above 25% (see fgure I.9). In
the region, countries can be divided into those where the cost is equivalent
to more than 5% of per capita GDP (El Salvador, Peru, Paraguay, Ecuador,
Guatemala, Honduras, the Plurinational State of Bolivia, Nicaragua) and
those where it is more affordable (Uruguay, Panama, the Bolivarian Republic
of Venezuela, Chile, Mexico, Argentina, Colombia, Costa Rica and Brazil).
Figure I.9
MOBILE BROADBAND TARIFFS IN RELATION TO PER CAPITA GDP IN 2012
(Percentages)
0.62
0.94 1.04 1.04
1.25
1.49 1.50
2.06
2.73 2.85 2.99
3.19
3.40
3.75
4.16
5.18
6.12
6.48
7.55
7.99
9.75
11.28
21.65
U
n
i
t
e
d
K
i
n
g
d
o
m
I
t
a
l
y
J
a
p
a
n
P
o
r
t
u
g
a
l
S
p
a
i
n
U
r
u
g
u
a
y
F
r
a
n
c
e
P
a
n
a
m
a
V
e
n
e
z
u
e
l
a
(
B
o
l
.

R
e
p
.

o
f
)
C
h
i
l
e
M
e
x
i
c
o
A
r
g
e
n
t
i
n
a
C
o
l
o
m
b
i
a
C
o
s
t
a

R
i
c
a
B
r
a
z
i
l
E
l

S
a
l
v
a
d
o
r
P
e
r
u
P
a
r
a
g
u
a
y
E
c
u
a
d
o
r
G
u
a
t
e
m
a
l
a
H
o
n
d
u
r
a
s
B
o
l
i
v
i
a
(
P
l
u
r
.

S
t
a
t
e

o
f
)
N
i
c
a
r
a
g
u
a
Source: Economic Commission for Latin America and the Caribbean (ECLAC), Regional Broadband
Observatory (ORBA), on the basis of tariffs published by operators as at September 2012.
Broadband in Latin America: Beyond Connectivity 31
Estimates of benchmark rates for 1 Mbps connections as a
percentage of per capita GDP use a standardized speed offered and
mask differences in download or upload speeds actually observed. The
data in fgure I.10 yield two fndings. Firstly, regardless of country and
speed, there is a signifcant difference between download and upload
speed. Secondly, speeds in the countries of the region are very slow.
Chile ranks best, with average download speeds of 8 Mbps and upload
speeds of just over 2 Mbps. Ecuador is an exception in that the two
speeds are similar.
Figure I.10
BROADBAND CONNECTION SPEEDS IN 2012
0
2
4
6
8
10
12
14
F
r
a
n
c
e
S
p
a
i
n
C
h
i
l
e
B
r
a
z
i
l
M
e
x
i
c
o
I
t
a
l
y
U
r
u
g
u
a
y
A
r
g
e
n
t
i
n
a
P
a
n
a
m
a
C
o
l
o
m
b
i
a
E
c
u
a
d
o
r
N
i
c
a
r
a
g
u
a
P
e
r
u
G
u
a
t
e
m
a
l
a
H
o
n
d
u
r
a
s
P
a
r
a
g
u
a
y
E
l

S
a
l
v
a
d
o
r
C
o
s
t
a

R
i
c
a
V
e
n
e
z
u
e
l
a

(
B
o
l
.

R
e
p
.

o
f
)
B
o
l
i
v
i
a

(
P
l
u
r
.

S
t
a
t
e

o
f
)
Download speed in Mbps Upload speed in Mbps
Source: Economic Commission for Latin America and the Caribbean (ECLAC), Regional Broadband
Observatory (ORBA), on the basis of Ookla data as at 1 September 2012.
Again, national averages mask differences between countries as
regards availability. Figure I.11 shows the distribution of connections
by benchmark or offered speed: over 10 Mbps, between 4 Mbps and
10 Mbps and under 4 Mbps. In several countries of the region, more than
half the connections exceed 4 Mbps. With the exception of Guatemala,
these are all middle-income countries. However, not all middle-income
countries are in the group with connections mainly over 4 Mbps. In
Costa Rica and Peru, only 30% of connections are, and it is surprising
to see that neither Argentina nor Brazil top 40%. Indeed, connections
delivering over 10 Mbps account for no more than 10% of the total in any
country in the region.
ECLAC 32
Figure I.11
PERCENTAGE OF BROADBAND CONNECTIONS, BY SPEED RANGE
0
10
20
30
40
50
60
70
80
90
100
U
n
i
t
e
d

K
i
n
g
d
o
m
F
r
a
n
c
e
I
t
a
l
y
P
o
r
t
u
g
a
l
J
a
p
a
n
S
p
a
i
n
C
h
i
l
e
P
a
n
a
m
a
M
e
x
i
c
o
C
o
l
o
m
b
i
a
G
u
a
t
e
m
a
l
a
A
r
g
e
n
t
i
n
a
B
r
a
z
i
l
E
c
u
a
d
o
r
H
o
n
d
u
r
a
s
C
o
s
t
a

R
i
c
a
E
l

S
a
l
v
a
d
o
r
P
e
r
u
N
i
c
a
r
a
g
u
a
P
a
r
a
g
u
a
y
U
r
u
g
u
a
y
V
e
n
e
z
u
e
l
a

(
B
o
l
.

R
e
p
.

o
f
)
B
o
l
i
v
i
a

(
P
l
u
r
.

S
t
a
t
e

o
f
)
Adoption of high-speed broadband (> 10 Mbps)
Adoption of broadband (>4 Mbps)
Adoption at less than 256 Kbps
Source: Economic Commission for Latin America and the Caribbean (ECLAC), Regional Broadband
Observatory (ORBA), on the basis of Akamai data, September 2012.
Note: The three ranges in the fgure should not add up to 100% because the percentage for connection
speeds over 4 Mbps includes all speeds above that level.
These country indicators already give an idea of the magnitude
of the effort required to move to the next phase of economic and social
development that comes with broadband. As was seen in the frst section,
high-speed Internet access and the services it can provide, such as cloud
computing and big data analytics, currently have substantial connection
requirements. The countries of the region should broaden their approach
to include all the variables linked to national and regional connectivity,
especially the concentration of international bandwidth in the United States
(see fgure I.12),
12
where over 40% of the cloud servers are located. This is
in contrast to the situation in Asia and Africa, which have diversifed their
international broadband connections (Kende, 2011).
The policy conclusion to be drawn from the fgures given in this
section is clear: the countries of Latin America must commit to policies,
preferably public policies, designed to close internal and external gaps
and to own the development opportunities created by the ongoing
technological revolution.
12
This concentration affects connection costs, as shown in chapter III by Omar de León.
Broadband in Latin America: Beyond Connectivity 33
Figure I.12
BANDWIDTH CONNECTED TO THE UNITED STATES
(Percentages)
0
20
40
60
80
100
2002 2003 2004 2005 2006 2007 2008 2009 2010
Africa Asia Europe Latin America
Source: Telegeography, Global Internet Bandwidth, 2011.
C. Lessons for a new era
This publication contributes to an ongoing discussion, providing evidence
and recommendations on a set of aspects relevant to more effcient
policymaking. Public policy must refect the supply and demand scenario
described in the frst part hereof.
Chapter II, written by Raúl Katz and Hernán Galperín, shows that
the demand gap (households with an available connection which do
not sign up for the service) is wide in the region, but it is in developed
countries such as Germany, the United States and the United Kingdom, as
well. This is true not only for fxed connections but also for mobile ones,
which are usually the vehicle for universal access to telecommunications in
general and to broadband in particular. Considering the high proportion
of shared access in the region, the research shows a demand gap of 63%
for mobile broadband and 50% for fxed broadband. Why this very high
percentage? Explanations can be found in aspects linked to socioeconomic
status, mainly education level and age. This again raises the issue of
affordability: people still fnd the service expensive for their income level
and do not sign up although supply exists. The public policy implications
of these fndings are clear. In the frst place, there is a need for targeted
programmes aimed at closing the interest and skills gaps among specifc
segments (for example, women heads of household). In the second place,
regulatory policies to close the affordability gap are needed.
If policies designed to close the demand gap were to succeed in the
very short term (in other words, if all those with available broadband were to
ECLAC 34
sign up), the result would be very congested networks. Chapter III, by Omar
de León, shows that the supply of high-capacity networks in the region lags
behind what is available elsewhere, particularly as regards connections to
the main information and knowledge production hubs. It also explains that
economies of scale justifying major investments have not yet been achieved,
that the costs of international access are signifcant and high in relation to
capacity to pay, that regional connectivity is imperfect, and that content is
usually hosted remotely. All of this adversely affects average broadband
quality in the region and exacerbates the major differences between and
within countries. With poor average quality that varies considerably from
country to country, potential applications are very limited and cloud
computing is diffcult. The challenge for supply expansion policies is to
address the factors directly impacting costs by expanding traffc interchange
points, promoting web caching, and hosting content on servers located in
countries of the region.
What use will all this be? The second part of the book deals with
economic impact. As shown in chapter IV, written by Raúl Katz, in addition
to the direct effect of investment in infrastructure when networks are
built, the economic impacts of broadband expansion are based on business
productivity increases that create a producer surplus. But communication
and consumer access to information and services expand as well and
create a consumer surplus. Econometric studies for Latin America show
that a 10-percentage-point rise in broadband penetration increases GDP
by 0.158%. Using a digitization index as an independent variable instead of
just measuring penetration, a 10-point rise in the indicator increases GDP
by 0.81% and decreases unemployment by 0.82%. This index is interesting
in that it combines some dimensions studied in connection with the
demand gap while including broadband quality and the human capital
needed for ICT use, among other variables.
These effects do not distinguish between the type of broadband
system (wired or wireless). If the focus is only on how much is lost owing to
lagging deployment of wireless networks for data or broadband in general,
Ernesto Flores-Roux estimates in chapter V that there would be a loss of
approximately 1% of GDP in the region (not including Costa Rica). He also
calculates that each inhabitant of the region would lose US$ 27 (2012 PPP)
for every three months’ lag in adopting more powerful networks.
Broadband as a platform for cloud services creates enterprise
opportunities by reducing the cost of fxed capital investment in
information technologies for small and medium-sized start-ups. In the
case studies for Argentina and Brazil (chapter VI), Andrea Colciago and
Federico Etro see promise in the adoption of cloud services because, in
addition to reducing start-up costs, they make the cost function fexible
Broadband in Latin America: Beyond Connectivity 35
(converting fxed costs into variable costs), bring more new players into the
market and increase competition. As a result, new businesses are opened
and jobs are created. The estimates in the model used by the authors show
that about 900,000 jobs could be created in Brazil and 100,000 in Argentina
if cloud services were widely adopted by the business sector.
Once the impact is known and verifed, the next step is to look at the
public policy scenario. Because of the large-scale investment required and
in the absence of installed and ubiquitous high-speed networks, the public
sector is becoming increasingly involved as a driver of investment, as
shown by the review of current policies in the region set out in chapter VII
written by Hernán Galperín, Judith Mariscal and María Fernanda Viecens.
The State could resume providing telecommunication infrastructure,
either in response to declining confdence in the full privatization model
(which did not succeed in providing universal services) or because the
recent economic upswing, mainly in South America, has made sizeable
resources available. An analysis of the broadband development policies of
six countries (Argentina, Brazil, Chile, Colombia, Mexico and Peru) shows
strong government commitment refected in an increase in resources
devoted to the deployment of backbone networks. Thus the pendulum is
swinging back and creating an important challenge for the private sector,
which has so far not fully met the need for investments to expand service
so that it can operate in information societies.
As far as broadband development policies pursued by developed
countries are concerned, the experience of the Republic of Korea
analyzed by Daewon Choi in chapter VIII is noteworthy because of its
comprehensive nature and because of the joint implementation effort by
various areas of government with the private sector, which would explain
its success. Any attempt to emulate the policies pursued by the Republic
of Korea as regards scope, targeting, sustainability and commitment of
public resources must acknowledge that they were part of an industrial
policy effort and therefore directly involved large industrial frms and not
only human capacity-building and infrastructure expansion programmes.
The experience of the Republic of Korea clearly shows the paradigm shift
in broadband development policies: conceived as a pillar of national
production development, they moved higher up the priority agenda and
became more holistic in scope.
Networks, which are the basis of the transmission infrastructure,
must be neutral in order for users, whether consumers or producers, to
effectively beneft from their full potential and take advantage of the
paradigm shift towards cloud computing services, as shown in chapter IX
by René Bustillo. Among the possible ways to understand network
neutrality, the two most important ideas are that operators should not
ECLAC 36
select the content to which users can have access or block users from
visiting any Internet site. Since the region has not made much progress in
regulating neutrality, policymakers need to put the item on their agendas
and follow best practices. This is particularly important at a time when
cloud computing is expanding, as the same author shows in chapter X.
Policies are already in place, but the changing situation is outpacing
policymaking. Cloud-based services are advancing, particularly those
provided by the online content industry, known as over-the-top (OTT)
applications, services and content, such as Skype, WhatsApp and Netfix.
Chapter XI, by Juan José Ganuza and María Fernanda Viecens, shows that
operators in the region providing the infrastructure for such content are
not profting from this growing demand. On the contrary, they are facing
major investment needs to expand bandwidth with no expectation of
private proftability, precisely because the revenue is going to the players
in the content industry. As a result, operators are devising strategies
to keep themselves in the market. Among them are bundling (which
could be challenged by regulators) and development of proprietary OTT
services, requiring an important shift in operators’ strategic planning. The
big question mark is still the speeds offered by operators and the public
and regulatory policy options for increasing them.
The lessons from the different parts of this book (diagnosis,
economic impact, policies and paradigm trends) confrm that the region
must redouble its efforts to insert itself competitively in a world where
technology is changing at a dizzying pace. Simply maintaining current
investment and expansion commitments provides no assurance that the
region will have the basic infrastructure and services needed to operate
in the information society. Once again, fast-tracking the digital revolution
was and is the way to narrow gaps and take advantage of the ongoing
technological revolution. As the title of this book says: DO. OR DO NOT.
THERE IS NO TRY.
Broadband in Latin America: Beyond Connectivity 37
Bibliography
Armbrust, M. and others (2009), Above the Clouds: A Berkeley View of Cloud Computing,
Electrical Engineering and Computer Sciences, University of California at
Berkeley, 10 February.
Cisco (2012), Cisco Visual Networking Index: Forecast and Methodology, 2011-2016, May.
Gantz, J. and D. Reinsel (2011), “Extracting Value from Chaos”, IDC IVIEW, June.
Gartner (2012), IT Spending Forecast 2Q12 Update y Public Cloud Services Forecast
2Q12 Update.
Global Pulse (2012), “Big Data for Development: Challenges and Opportunities”,
New York, United Nations, May.
ITU (International Telecommunication Union) (2009), “Distributed Computing:
Utilities, Grids & Clouds”, ITU-T Technology Watch Report, No. 9.
Jordán, V., H. Galperín and W. Peres (coordinators) (2010), Fast-tracking the Digital
Revolution: Broadband for Latin America and the Caribbean (LC/R.2167), Santiago,
Chile, Economic Commission for Latin America and the Caribbean (ECLAC)/
Regional Dialogue on the Information Society (DIRSI).
Kende, M. (2011), Overview of recent changes in the IP interconnection ecosystem,
Analysis Mason, 23 January.
McKinsey (2011), Big data: the next frontier for innovation, competition, and productivity,
McKinsey Global Institute, June.
ORBA (Regional Broadband Observatory) (2012), Status of broadband in Latin America
and the Caribbean, 2012, Report of the Regional Broadband Observatory (ORBA),
Santiago, Chile, Economic Commission for Latin America and the Caribbean
(ECLAC).
Telecom (2012), Cloud computing in Latin America – Towards a virtualized reality,
Telecom, Intelligence Series, September.
The Economist (2010), “Data, data everywhere”, 23 August.
World Economic Forum (2012), The Global Information Technology Report 212. Living
in a Hyperconnected World, Geneva, INSEAD.
Chapter II
The demand gap: drivers and public policies
Raúl L. Katz and Hernán Galperin
1
The debate about the digital divide in Internet use and broadband has
largely revolved around the statistics on households that own a computer
and have adopted broadband (service penetration, in other words).
Thus, political discussion and public opinion have turned on the need
to increase take-up by expanding telecommunication network coverage.
The underlying premise is that the digital divide would be narrowed if
the issues holding up infrastructure investment were resolved. Without
denying that there is some causal relationship between investment and
the divide, it is important to stress that one of the fundamental variables
accounting for the digital divide lies on the demand side rather than the
supply side. The purpose of this chapter is to analyse the divide from this
perspective, both in industrialized countries and in Latin America.
First, quantitative information is presented to show that there is a
demand gap, even in industrialized countries. On this basis, the chapter
reviews the research carried out in the developed world, using statistics
from different countries to identify common causal variables. It then
examines the situation in Latin America, concentrating frst on quantifying
the demand gap for countries on which information is available. Following
1
Raúl L. Katz is an Adjunct Professor with the Division of Finance and Economics at
Columbia Business School and Director of Business Strategy Research at the Columbia
Institute for Tele-Information. Hernán Galperín is an Associate Professor at the University
of San Andrés in Buenos Aires.
ECLAC 40
the same process as for developed countries, it presents the results of
research done in Latin America with the aim of explaining the nature
of the demand gap. This diagnosis provides the basis for public policy
recommendations to address some of the barriers to adoption.
A. Measuring the digital divide from the demand side
The digital demand gap is defned as the number or percentage of
households or individuals who could subscribe to broadband but do
not. Such statistics are not easy to calculate because technology coverage
(i.e., households and individuals with access to broadband) is not usually
measured by public-sector entities or regulatory agencies. A large number
of national broadband strategies have been developed in recent years,
however, and in-depth analysis of the size of the service coverage shortfall
has been required for diagnostic purposes.
In the United States, according to the Federal Communications
Commission (FCC), 96% of households had access to broadband via a cable
modem in early 2008 and 82% via DSL. Current penetration statistics
indicate that 64% of United States households have purchased the service.
Thus, 32% of households could access broadband but do not. This gap
naturally varies by state (see fgure II.1).
Figure II.1
UNITED STATES: STATES WITH THE LOWEST BROADBAND
PENETRATION RATES, 2010
(Percentages)
0
10
20
30
40
50
60
70
80
M
i
s
s
i
s
s
i
p
p
i
W
e
s
t

V
i
r
g
i
n
i
a
A
l
a
b
a
m
a
N
e
w

M
e
x
i
c
o
A
r
k
a
n
s
a
s
M
a
i
n
e
S
o
u
t
h

C
a
r
o
l
i
n
a
I
n
d
i
a
n
a
M
o
n
t
a
n
a
I
o
w
a
K
e
n
t
u
c
k
y
N
o
r
t
h

D
a
k
o
t
a
T
o
t
a
l
O
k
l
a
h
o
m
a
N
e
b
r
a
s
k
a
P
e
n
n
s
y
l
v
a
n
i
a
K
a
n
s
a
s
G
e
o
r
g
i
a
Supply Demand
S
o
u
t
h

D
a
k
o
t
a
Source: Prepared by the authors on the basis of data from the Federal Communications Commission
(FCC), HSPD1207, table 14, and the United States Census Bureau.
Broadband in Latin America: Beyond Connectivity 41
As can be seen, the demand gap in a state like Mississippi is 60%
while the supply gap (households not covered by the service) is just 9%. In
Georgia, where the service adoption rate is higher, the supply gap is 8% of
households while the demand gap is 34%.
In Germany, the National Broadband Strategy published in February
2009 reported that 98% of households (39.7 million) were in a position to
access broadband. Of these, 36.7 million were covered by DSL platforms,
22 million had cable television (meaning they could access broadband
via cable modem) and 730,000 could access the Internet using wireless
platforms such as satellite. Despite this coverage, just 58% had taken up
the service. Information for other developed countries confrms that there
is a gap of this kind everywhere (see table II.1).
Table II.1
DEVELOPED COUNTRIES: SIZE OF THE FIXED BROADBAND DEMAND GAP, 2011
(Percentages)
Country Households
covered
Households
connected
Demand gap
Germany 98 58 40
Australia 89 69 20
Republic of Korea 100 93 7
Denmark 96 76 20
Spain 93 61 32
United States 96 61 35
France 100 77 23
Israel 100 83 17
Italy 95 55 40
United Kingdom 100 68 32
Sweden 100 89 11
Source: Prepared by the authors on the basis of data from the International Telecommunication Union (ITU),
the European Union (EU), the Federal Communications Commission (FCC), the German Federal Ministry of
Economy and Technology (BMWi), the Organisation for Economic Cooperation and Development (OECD).
In some developed countries (especially Germany, Italy, Spain,
the United Kingdom and the United States), a large proportion of people
who do not access the Internet via fxed broadband in the home are not
prevented from doing so by lack of service availability but refrain for other
reasons. So what are the factors accounting for this?
The demand gap issue becomes more complex when mobile
broadband is viewed as a platform capable of providing Internet access.
ECLAC 42
First of all, what is mobile broadband? The term can be applied to
service subscriptions that include the acquisition of a modem enabling a
computer to connect to the Internet (these are called USB modems, dongles
or aircards). It also applies to technologies that allow Internet access via
mobile phones. In this case, smartphones are suitable terminals because
their screen formats and interface systems provide an adequate platform
for browsing the web, answering e-mails and accessing platforms such
as Facebook, Google and YouTube. From the point of view of network
technology, third- and fourth-generation (3G and 4G) network coverage
provides the speed necessary for effcient access.
The second issue where mobile broadband is concerned is how to
measure the demand gap. Since in this case the connection is provided
to individual users (laptop or smartphone owners), the parameters used
to measure the demand gap should be different to those used for fxed
broadband: the population coverage of 3G and 4G networks and the
percentage of subscribers who have a smartphone or mobile modem.
This assumes that most smartphones operate on 3G or 4G networks,
which is not necessarily the case, although the number of subscribers
using terminals of this type on 2.5G networks is falling rapidly. Table II.2
presents estimates of the mobile broadband demand gap for some
developed countries.
Table II.2
DEVELOPED COUNTRIES: MOBILE BROADBAND DEMAND GAP, 2011
Country Population
covered (3G)
Mobile broadband
penetration
Mobile
demand gap
Germany 86.0 34.7 51.2
Australia 97.0 89.1 7.9
Republic of Korea 99.0 97.1 1.9
Denmark 97.0 57.5 39.5
Spain 90.6 36.7 53.9
United States 98.5 71.9 26.6
France 98.2 32.9 65.3
Israel 99.0 54.4 44.6
Italy 91.9 48.2 43.7
United Kingdom 95.0 42.6 52.4
Sweden 99.0 85.1 13.9
Source: Prepared by the authors on the basis of data from Wireless Intelligence and the International
Telecommunication Union (ITU).
Note: Population coverage is based on 3G networks on the assumption that Long Term Evolution (LTE)
networks will be rolled out in the same areas, at least initially.
Broadband in Latin America: Beyond Connectivity 43
Another dimension to be considered in measuring the demand gap
is the degree of substitution or complementarity between fxed and mobile
broadband. For example, mobile broadband subscribers often subscribe to
fxed broadband too, so that the two technologies complement each other
by providing Internet users with an environment of total connectivity. In
other cases, and especially in emerging countries, mobile broadband may
be a substitute for fxed broadband in three types of situations: (i) no fxed
service is available in the user’s area, (ii) the quality of the fxed service
puts it at a disadvantage to the mobile service (low speeds, for example),
or (iii) the user opts to consolidate by acquiring only a mobile service that
provides connectivity and mobility. Where substitution has taken place, it
is important to include mobile broadband-only subscribers in the total for
broadband users before looking at the demand gap. Diagram II.1 illustrates
the two types of demand gap.
Diagram II.1
FIXED AND MOBILE BROADBAND SUBSTITUTION AND COMPLEMENTARITY
Fixed broadband
coverage (ADSL,
cable modem, FTTH)
Mobile broadband
coverage (3G, 4G)
Supply
Fixed and mobile
broadband subscribers
Mobile
broadband
subscribers
Demand
Fixed
broadband
subscribers
Complementarity
Demand
gap
Demand
gap
Substitution Substitution
Source: Prepared by the authors.
There are two types of demand gap: users who only have access
to a fxed broadband service and do not acquire it (this is fairly unusual,
since fxed and mobile broadband tend to be rolled out in the same areas)
and those for whom the only option would be mobile broadband but
they do not acquire it. Mobile broadband users should therefore not be
included in the population that is considered to be part of the demand gap
because they are already purchasing an Internet access service that either
complements or substitutes fxed broadband (see fgure II.2).
ECLAC 44
Figure II.2
QUANTIFICATION OF THE FIXED AND MOBILE DEMAND GAP
(Percentages)
0
10
20
30
40
50
60
70
80
90
100
0 20 40 60 80 100
Mobile demand gap
Fixed + mobile
demand gap
Fixed
demand gap
Coverage
gap
Mobile
coverage
Fixed coverage
Fixed broadband
M
o
b
i
l
e

b
r
o
a
d
b
a
n
d
Mobile broadband-only subscribers (substitution)
Fixed- broadband-only subscribers (legacy)
Fixed + mobile broadband subscribers (complementarity)
Mobile subscribers
F
i
x
e
d

s
u
b
s
c
r
i
b
e
r
s
Source: Prepared by the authors.
Thus, the demand gap should be quantifed using the following
formulas:
Demand gap = broadband coverage (C) – broadband subscriptions (S)
where
C = population covered by fxed and mobile broadband + population
covered by fxed broadband only + population covered by mobile broadband only
S = fxed and mobile broadband subscribers (complementarity) + fxed
broadband subscribers (legacy) + mobile broadband subscribers (substitution)
Using this formula to calculate the demand gap requires a frm
understanding of parameters such as technological complementarity. The
current lack of statistics of this kind makes it necessary to analyse the
demand gap by technology type.
B. Understanding the demand gap
There is an extensive literature on the drivers of Internet adoption,
particularly for broadband. These studies concur in identifying income,
educational attainment and household age structure as the main
predictors for adoption of these services (Hauge and Prieger, 2010). A
number of studies suggest that other factors are also involved, many of
them specifc to particular countries or regions. For example, Navarro
Broadband in Latin America: Beyond Connectivity 45
and Sánchez (2011) have shown that, all other things being equal, the
likelihood of Internet use drops by 6% in Latin America if the potential
user is a woman. In the United States, studies have shown the importance
of factors such as ethnicity and profciency in English (Ono and Zavodny,
2008; NTIA, 2011). Other factors such as geographical location (rural versus
urban), the presence of school-age children and the penetration rate in
the individual or household’s geographical area (essentially, the network
effect) have also been identifed in the academic literature as drivers of
Internet adoption (Chaudhuri and Flamm, 2005; Vicente and López, 2006;
Grazzi and Vergara, 2011).
Identifying the socioeconomic factors behind Internet adoption
provides a frst approach to the demand gap issue. Nonetheless, an
analysis based on econometric studies does not allow a distinction to be
drawn between non-adoption due to supply constraints (in rural or low-
income areas, for example) and demand factors. Nor do these studies have
much to say about the reasons for non-adoption when there is an adequate
supply of services.
Studies based on surveys of people who are not Internet users do
permit some progress in this direction. This section reviews the fndings
of studies in the most developed countries, where there is a substantial
literature on the subject. As will be seen, fndings regarding the factors that
explain non-adoption in different countries are surprisingly consistent.
The evidence for Latin America will be analysed in the following section.
To start with the United States, data from the most recent surveys
show that 78% of adults use the Internet “at least occasionally” (Pew Center,
2012). Of the 22% of the population who are not users, most are individuals
over 65 years of age, adults who have not completed secondary education,
members of households earning less than US$ 30,000 a year and those with
limited English profciency, corroborating the fndings of the studies cited
above. What are the reasons for non-adoption? The responses of non-users
reveal that the main factor is lack of interest or relevance (42%), followed
by factors related to the affordability of devices and services (22%) and
lack of usage skills (21%).
Focusing on the adoption of broadband in the home, the most recent
data show that 62% of adults in the United States live in households with
broadband service.
2
However, this percentage drops to 22% for adults
who have not completed secondary education, 30% among those aged 65
and over and 41% among those with incomes of less than US$ 30,000 per
year, replicating the patterns of adoption cited above (Pew Center, 2012).
2
This figure comes from the Pew Internet Project survey and is for August 2011. Subscription
data reported by operators yield essentially similar results: according to the FCC, 64% of
households that have access to broadband subscribe to it (FCC, 2012).
ECLAC 46
As table II.3 shows, the main reasons cited by those who do not have
broadband at home fall into a pattern similar to those reported by non-
users of the Internet, most notably a lack of interest or relevance (50%) and
affordability constraints (19%).
Table II.3
UNITED STATES: REASONS FOR NOT ADOPTING BROADBAND
IN THE HOME, 2009
(Percentages and number of observations)
Reason
Have narrowband
service in the
home
No
Internet in
the home
Weighted
total
Percentage
of adult
population
Lack of relevance/interest 32 45 50 13
Cost (computer or
connectivity service)
35 15 19 5
Service unavailable 17 16 17 4
Diffculty of use 16 22 13 3
Number of observations 92 566 658
Source: J. Horrigan, Home Broadband Adoption 2009, Washington, D.C., Pew Internet and American Life
Project, 2009.
A government report on the population that does not have
broadband in the home corroborates these fndings (NTIA, 2011). The
main reason cited by individuals living in households without broadband
is a lack of interest in the service (47%), followed by reasons related to
affordability (24%) and lack of proper equipment (15%). If households
that have a computer but do not subscribe to the service and households
without either a computer or broadband are considered separately, service
affordability reasons are more prominent in the frst group, while the lack
of relevance or interest predominates in the second (see table II.4).
In Spain, the most recent data show that 61% of households have a
broadband connection (ONTSI, 2012). Among unconnected households, the
main reasons cited for not accessing the service are lack of interest (66%), the
cost of equipment or connectivity (42%) and lack of the skills or knowledge
required to use the service (29%).
3
Interestingly, as table II.5 shows, the lack of
relevance/interest is still a signifcant determinant even at the lowest income
levels, where affordability and usage skills are, as would be expected, the
most important factors. These fndings confrm the importance of digital
literacy initiatives aimed at lower-income sectors of the population.
3
Survey on the Equipment and Use of Information and Communication Technologies in
Households 2011 (INE). Percentages exceed 100% because respondents could select more
than one reason.
Broadband in Latin America: Beyond Connectivity 47
Table II.4
UNITED STATES: REASONS FOR NOT ADOPTING BROADBAND BASED
ON THE AVAILABILITY OF A COMPUTER IN THE HOME, 2011
(Percentages and number of observations)
Reason
Households with
a computer
Households without
a computer
Weighted total
Lack of relevance/interest 28 52 47
Cost (computer or
connectivity service)
37 21 24
Lack of suitable equipment
(computer)
8 17 15
Other 27 10 14
Number of households
(millions)
6.8 27.8 34.6
Source: National Telecommunications and Information Administration (NTIA), Exploring the Digital Nation,
Washington, D.C., 2011.
Table II.5
SPAIN: REASONS FOR NOT ADOPTING BROADBAND, BY INCOME LEVEL, 2011
(Percentages and number of observations)
Reason/monthly
household income
<1 100
euros
1 100 to
1 800
euros
1 800 to
2 700
euros
>2 700
euros
Total
Lack of relevance/
interest
67 65 48 42 66
Cost (computer or
connectivity service)
52 39 42 16 42
Lack of usage skills 35 27 18 12 29
Number of households
(millions)
2.5 1.2 0.3 0.1 5.6
a
Source: National Institute of Statistics (INE), Survey on the Equipment and Use of Information and
Communication Technologies in Households, 2011.
a
The difference is due to the remaining respondents (1.4 million) not reporting their income.
In the United Kingdom, the most recent fgures show that 80% of
households have an Internet connection, overwhelmingly broadband
(76% of all households) (OFCOM, 2012a). In line with the fndings of
other studies, those living in households without a connection tend to be
older (aged over 65) and on low incomes. The vast majority expressed no
intention of subscribing to the service in the next 12 months, suggesting
that the demand gap is persistent. Again, the main reason cited is lack
of relevance (66%), well above cost-related factors (16%) and lack of usage
skills (4%).
ECLAC 48
In 2010, cost-related factors were cited by 23% as the main reason for
not subscribing to the service, while in 2011 the fgure dropped to 16%. This
suggests that as access and equipment costs decline, a gap increasingly
associated with cultural or educational factors persists. Another signifcant
fnding is that 23% of non-users report that they have asked someone else
to perform a task on the Internet (such as sending e-mail or searching for
information) on their behalf. This indicates that, regardless of relevance,
important barriers associated with usage skills persist (OFCOM, 2012b).
The review of studies on non-adoption of Internet and broadband in
the most developed countries reveals results that are essentially consistent
from one country to another and yield both a sociodemographic profle
of unconnected households and the main reasons for non-adoption of the
service. In sociodemographic terms, the results of survey-based studies
bear out the evidence from econometric studies: unconnected households
tend to be composed of older persons (over 65), low-income individuals
and those who have not completed secondary education. In the United
States, ethnicity and English profciency factors (among the population of
Spanish-speaking recent immigrants) also come in.
The fndings of the different studies also concur on the reasons
given for not adopting broadband in the home. Lack of relevance or interest
consistently appears as the primary factor in non-adoption. As suggested
by OFCOM (2010a), this response may mask reasons related to costs or
a lack of usage skills, factors that consistently rank second and third in
order of importance. However, the trend seems to indicate a decline in
the importance of factors related to the affordability of equipment and
connectivity services. Thus, the persistence of a core of between 20% and
40% of households not connected to broadband in developed countries
suggests the need for long-term digital literacy policies aimed at promoting
demand for services among the households described above.
C. The situation in Latin America
In the case of Latin America, it is important to frst consider the role of shared
Internet access at locations such as the workplace, school and public access
facilities, both free (typically called telecentros in Spanish) and commercial
(public kiosks and Internet cafés). While this type of access is marginal in
more developed countries, the most recent fgures for Latin America show
that shared Internet access continues to be very signifcant, despite the
sharp increase in the number of individual broadband subscriptions. As an
example, the most recent fgures in Peru show that 65% of Internet users use
it at work or at public access facilities (INEI, 2012). In comparison, the most
recent fgure for Spain is a bare 17% of users (ONTSI, 2012).
Broadband in Latin America: Beyond Connectivity 49
The infuence of shared access in the region has resulted in a
signifcant gap between the number of Internet users and the number
of broadband subscriptions, as shown in table II.6. This gap can be
interpreted as latent broadband demand: demand for Internet access that
does not turn into service subscriptions.
Table II.6
INTERNET USERS AND BROADBAND SUBSCRIBERS IN LATIN AMERICA, 2011
(Percentages)
Country Internet users Fixed subscribers Mobile subscribers
Argentina 47.7 10.5 11.7
Bolivia (Plurinational State of) 30.0 0.7 1.9
Brazil 45.0 8.6 20.9
Chile 53.9 11.7 17.1
Colombia 40.4 6.9 3.7
Costa Rica 42.1 8.7 2.0
Dominican Republic 35.5 4.0 7.7
Ecuador 31.4 4.2 10.3
El Salvador 17.7 3.3 3.6
Guatemala 11.7 1.8 4.1
Honduras 15.9 2.7 3.7
Jamaica 31.5 3.9 1.5
Mexico 36.2 10.6 4.6
Nicaragua 10.6 1.8 1.0
Panama 42.7 7.9 14.5
Paraguay 23.9 0.9 4.5
Peru 36.5 3.5 1.4
Uruguay 51.4 13.5 9.0
Venezuela (Bolivarian
Republic of)
40.2 6.1 4.2
Source: International Telecommunication Union (ITU), Telecommunications Database 2012.
As fgure II.3 shows, latent demand measured by the difference
between users and subscribers (per 100 inhabitants) is higher in some of
the region’s more mature markets. This suggests that despite a network
effect which stimulates Internet adoption in countries with higher
penetration, the market equilibrium point keeps this latent demand from
turning into subscriptions.
ECLAC 50
Figure II.3
LATENT FIXED AND MOBILE BROADBAND DEMAND IN LATIN AMERICA, 2011
(Percentage points)
0
5
10
15
20
25
30
35
40
45
C
h
i
l
e
U
r
u
g
u
a
y
A
r
g
e
n
t
i
n
a
B
r
a
z
i
l
P
a
n
a
m
a
V
e
n
e
z
u
e
l
a
(
B
o
l
.

R
e
p
.

o
f
)
C
o
l
o
m
b
i
a
C
o
s
t
a

R
i
c
a
P
e
r
u
D
o
m
i
n
i
c
a
n

R
e
p
.
B
o
l
i
v
i
a
(
P
l
u
r
.

S
t
a
t
e

o
f
)
J
a
m
a
i
c
a
E
c
u
a
d
o
r
M
e
x
i
c
o
P
a
r
a
g
u
a
y
E
l

S
a
l
v
a
d
o
r
H
o
n
d
u
r
a
s
G
u
a
t
e
m
a
l
a
N
i
c
a
r
a
g
u
a
Fixed broadband Mobile broadband
Source: International Telecommunication Union (ITU), Telecommunications Database 2012.
The barriers preventing this latent demand from being converted
into subscriptions are associated with several factors. First, determining
whether there are coverage gaps that explain the magnitude of the latent
demand observed in the region will then make it possible to estimate the
effective demand gap (i.e., discounting coverage shortfalls) for several
countries in the region. Lastly, the factors accounting for this demand gap
are analysed on the basis of surveys conducted in various countries.
1. Broadband coverage
Broadband coverage in Latin America is relatively extensive. The
following analysis of population coverage was carried out by extrapolating
information provided by operators and regulators (see annex II.1).
As table II.7 shows, other than in some Andean countries, the
broadband supply gap in Latin America is not large. The population
coverage of fxed broadband ranges from 98% in Uruguay to 40% in the
Plurinational State of Bolivia, with an average of 79% for all the countries
analysed. This is because ADSL services over the copper network have
benefted from the historical deployment of telecommunication networks.
As might be expected, cable television network coverage is concentrated in
areas of higher population density and so overlaps with ADSL access.
For mobile broadband, this analysis is based on the deployment
of 3G networks (based on EVDO and HSPA standards), which are by
defnition more suitable for Internet access. In this case, population
coverage ranges from 96% in Colombia to 29% in the Plurinational State of
Bolivia (with an average of 76% for all the countries surveyed).
Broadband in Latin America: Beyond Connectivity 51
Table II.7
BROADBAND COVERAGE IN LATIN AMERICA, 2011-2012
(Percentage of population)
Country Fixed broadband Mobile broadband
Argentina 96 92
Bolivia (Plurinational State of) 40 29
Brazil 94 84
Chile 78 82
Colombia 81 96
Costa Rica 95 93
Ecuador 87 66
Mexico 62 77
Peru 59 63
Dominican Republic n/a 70
Uruguay 98 n/a
Source: Prepared by the authors on the basis of the methodology detailed in annex II.1.
2. The demand gap
Comparing broadband penetration and service coverage makes it
possible to estimate the size of the demand gap (table II.8). In the fxed
broadband segment, the average demand gap for the countries analysed
is 50 percentage points: less than half the households covered by a fxed
broadband service choose to subscribe to it.
Table II.8
THE FIXED BROADBAND DEMAND GAP IN LATIN AMERICA, 2011
(Percentages)
Country Coverage
Household
penetration
Demand gap
Argentina 96 39 57
Bolivia (Plurinational State of) 40 3 37
Brazil 94 29 65
Chile 78 44 34
Colombia 81 27 54
Costa Rica 95 32 63
Ecuador 87 20 67
Mexico 62 47 15
Peru 59 16 43
Uruguay 98 34 43
Average 79 29 50
Source: International Telecommunication Union (ITU), Telecommunications Database 2012.
ECLAC 52
The demand gap is even larger in the mobile segment, averaging
63 percentage points in the countries examined (see table II.9). In
accordance with the theoretical framework explained in the frst section,
a signifcant portion of mobile broadband users are also fxed broadband
users owing to the complementarity effect. As a result, the contribution of
mobile broadband to reducing the demand gap has, so far, been smaller.
Although exact fgures cannot yet be calculated because the numbers
using each technology are unknown, the rapid rate of deployment of
mobile broadband suggests that the substitution effect will increase in
importance. A progressive reduction in the overall demand gap can
therefore be anticipated for the coming years.
Table II.9
THE MOBILE BROADBAND DEMAND GAP IN LATIN AMERICA, 2011
(Percentages)
Country Coverage Subscriber penetration Demand gap
Argentina 92 19 73
Bolivia (Plurinational State of) 29 3 26
Brazil 84 21 63
Chile 82 17 65
Colombia 96 9 87
Costa Rica 93 11 82
Ecuador 66 11 55
Mexico 77 14 63
Peru 63 9 54
Dominican Republic 70 5 65
Average 75 12 63
Source: Prepared by the authors on the basis of table II.7 data for coverage and the total number of HSPA,
LTE and EVDO connections divided by the population, as reported by Wireless Intelligence.
In summary, the demand gap in the region remains large; a detailed
analysis of the underlying factors is required so that policies can be
developed to address it.
3. Explaining the demand gap
Studies conducted in different countries of the region have yielded
relatively consistent results regarding the determinants of the demand
gap. Figure II.4 presents the fndings of the survey conducted by the
Internet Management Committee in Brazil (CGI.br) to analyse why
households with computers do not subscribe to a broadband service. As
can be seen, the main reason is affordability, followed by (perceived) lack
Broadband in Latin America: Beyond Connectivity 53
of availability, lack of interest and lack of skills. There was a 10 percentage-
point decrease in reasons associated with service cost between 2007
and 2011, suggesting rising incomes in the country, falling prices and
greater segmentation of supply. Meanwhile, lack of interest and lack of
usage skills are factors that have remained relatively stable over time,
suggesting the impact of structural factors linked to defciencies in the
education system.
Figure II.4
BRAZIL: REASONS FOR NOT SUBSCRIBING TO INTERNET
IN THE HOME, 2007-2011
(Percentages of households with a computer)
0
10
20
30
40
50
60
70
2007 2008 2009 2010 2011
Cost Lack access
Lack interest Lack skills
Other
Source: CGI, Home ICT Survey.
Research results in Mexico reveal a similar pattern: among
households with a computer but no Internet connection, the main reason
cited is the cost of connectivity (60%), followed by lack of interest (19%).
In Chile, reasons associated with connectivity costs are less to the fore,
being cited by 37% of households with a computer, followed by lack of
interest (24%) and lack of usage skills (8%). In Costa Rica, by contrast, the
cost factor is more prominent at 60%, followed by lack of skills (12%) and
interest (7%). Table II.10 summarizes these results. As can be seen, with the
possible exception of Chile, service costs are the main factor explaining
the demand gap in Latin America, confrming the results obtained by
Galperín and Ruzzier (2010). However, longitudinal analysis in the case
of Brazil shows that as access costs fall, structural factors associated with
human capital gain in prominence.
The results of these studies make it possible to identify dimensions
of the demand gap associated with sociodemographic factors that are
discussed separately below.
ECLAC 54
Table II.10
LATIN AMERICA: FACTORS EXPLAINING THE DEMAND GAP
(Percentages)
Reasons cited for not having an Internet
connection in the home
(only homes with a computer)
Chile
(2009)
Brazil
(2011)
Costa
Rica
(2011)
Mexico
(2010)
Price of the service 37 48 60 60
Lack of interest 24 14 12 19
Lack of usage skills 8 10 7 n/a
Other reasons (lack of availability, use in
other locations, etc.)
31 28 21 21
Source: Chile: Survey on Internet Broadband Access, Use and Users in Chile, Universidad Alberto Hurtado/
SUBTEL, June 2009. Costa Rica: Second Assessment of the Digital Divide in the Use of Telecommunications
Services in Costa Rica, Telecommunications Directorate, February 2011. Mexico: Survey on the Availability
and Use of Information and Communication Technologies in Households, INEGI (2010). Brazil: Survey on
the Use of Information and Communication Technologies in Brazil, CGI.br, November 2011.
(a) The socioeconomic dimension of the demand gap
As the results discussed have shown, analysis of the demand gap by
income level corroborates the importance of affordability as a determinant
of Internet adoption within households. Beginning with Brazil, fgure II.5
shows the disparity between the highest-income group (socioeconomic
level A), in which household Internet adoption is almost universal, and
the lowest-income group (socioeconomic level D/E), in which household
access is marginal. The trend among households of the so-called new
middle class (socioeconomic level C) is striking, with home access more
than doubling between 2008 and 2011.
Figure II.5
BRAZIL: HOME INTERNET ADOPTION BY SOCIOECONOMIC LEVEL (SEL), 2011
(Percentages)
91 90 90
96
58
64
65
76
16
21
24
35
1
3 3
5
0
20
40
60
80
100
120
2008 2009 2010 2011
SEL A SEL B SEL C SEL D/E
Source: Survey on the Use of Information and Communication Technologies in Brazil, CGI.br.
Broadband in Latin America: Beyond Connectivity 55
For Mexico, fgure II.6 shows that the home Internet penetration
growth rate between 2008 and 2010 was greater in higher-income
deciles, suggesting a widening of the adoption gap between
socioeconomic groups.
Figure II.6
MEXICO: HOME INTERNET ADOPTION BY INCOME DECILE, 2008-2010
(Percentages)
0
10
20
30
40
50
60
70
80
X IX VIII VIII VIII VIII IV II II I
2008 2010
Source: National Institute of Statistics and Geography (INEGI), Estadísticas sobre disponibilidad y uso de
tecnologías de la información y comunicación en los hogares, Mexico City 2010.
The fndings confrm the importance of economic variables
in explaining the demand gap in the countries of the region. The
conclusions regarding trends are not decisive, however: whereas in Brazil
the increase in adoption among lower middle-class households suggests a
progressive narrowing of the penetration gap by socioeconomic level, the
trend in Mexico reveals a widening of the gaps between income deciles.
In general, as fgure II.7 shows, there has been a gradual convergence
in access levels by socioeconomic level, particularly in medium-high
income countries such as Brazil, Costa Rica, Chile and Uruguay. In any
event, these fndings merit more comprehensive longitudinal studies,
which would also make it possible to identify the factors accounting for
differences in trends between countries, which could be associated with
changes in household income, developments in service supply or policies
aimed at universalizing access.
ECLAC 56
Figure II.7
HOME INTERNET ADOPTION GAP BY INCOME QUINTILE (Q5/Q1)
0
10
20
30
40
50
60
70
80
90
100
2005 2009 2006 2010 2006 2010 2003 2009 2005 2010 2005 2010 2003 2010 2005 2010
Brazil Paraguay Costa Rica Chile Uruguay Mexico Colombia Venezuela
(Bol. Rep. of)
Source: Observatory for the Information Society in Latin America and the Caribbean (OSILAC), on the
basis of household surveys conducted in the respective countries.
(b) The education dimension
Survey fndings suggest that, after affordability, non-adoption of
home Internet services is mainly due to interest and usage skill factors that
are closely connected with the stock of human capital. This points to the
education dimension of the demand gap, which operates as a proxy for
determinants related to interest and usage skill.
In all the countries analysed, there are access gaps between
individuals with different levels of education. As an example, fgure II.8
shows that the level of home Internet adoption in Costa Rica is more than
twice as high among those who have completed secondary education as
among those who have not.
The Chilean case supports this fnding, as fgure II.9 shows. The
home Internet adoption rate in Chile is less than 25% among those who
have not completed secondary education but as high as 42% among those
who have. This confrms that completing secondary education is an
important educational threshold that fosters interest and promotes the
usage skills needed to take advantage of Internet access in the home.
Broadband in Latin America: Beyond Connectivity 57
Figure II.8
COSTA RICA: RESIDENTIAL INTERNET AND COMPUTER USE BY EDUCATIONAL
LEVEL OF THE HOUSEHOLD HEAD, 2010
0
10
20
30
40
50
60
70
80
90
100
None Incomplete
primary
Complete
primary
Incomplete
secondary
Complete
secondary
Incomplete
tertiary
Complete
tertiary
Postgraduate
Home fixed Internet Home computer
Source: Costa Rica, Telecommunications Directorate, 2011.
Figure II.9
CHILE: RESIDENTIAL INTERNET AND COMPUTER USE BY EDUCATIONAL
LEVEL OF THE HOUSEHOLD HEAD, 2009
0
20
40
60
80
100
120
Incomplete
primary
Complete
primary
Incomplete
secondary
Complete
secondary
Incomplete
tertiary
Complete
tertiary
Home fixed Internet Home computer
Source: Universidad Alberto Hurtado, Encuesta sobre acceso, uso y usuarios de Internet banda ancha en
Chile, Santiago, Chile, 2009.
ECLAC 58
The problem with these results is the strong correlation between
education attainment and socioeconomic level, making it diffcult to
isolate the effect of education on adoption by discounting the known effect
of socioeconomic level. While this type of analysis is beyond the scope of
the present paper, fgure II.10 shows that the effect of education persists
even after income deciles are controlled for.
Figure II.10
COSTA RICA: RESIDENTIAL BROADBAND PENETRATION BY EDUCATION
LEVEL AND INCOME DECILE, 2010
0
10
20
30
40
50
60
I II III IV V VI VII VIII IX X
Range Decile
Less than 37 888 I
37 888 to 54 124 II
54 125 to 69 933 III
69 934 to 88 499 IV
88 500 to 109 889 V
Primary or less Secondary Tertiary
109 890 to 138 845 VI
138 846 to 183 332 VII
183 333 to 253 406 VIII
253 407 to 416 537 IX
416 538 and more X
Range Decile
Source: Prepared by the authors on the basis of National Institute of Statistics and Census (INEC),
Household Survey, July 2010.
This analysis yields two conclusions. First, it highlights the direct
relationship between income level and access to ICTs in the home. The cut-
off would appear to be around an average income for the seventh decile.
From this decile up, the higher the level of education, the greater the
Internet and broadband adoption rate. In households with incomes below
the sixth decile, a higher level of education does not result in a signifcant
increase in service penetration, except in decile II.
Second, the educational level of the household head is an important
variable for the level of broadband adoption in Costa Rica; the level of
service take-up is more than 30% in decile II households with university-
level education. This would indicate that the infuence of education in
decile II is signifcant. One hypothesis could be that this decile includes
recent university graduates whose incomes have not increased signifcantly.
Broadband in Latin America: Beyond Connectivity 59
Meanwhile, groups with higher levels of secondary and tertiary education
and incomes above decile VIII levels have a utilization rate of around 50%.
The interaction between the two variables (socioeconomic status
and education) thus reveals a complex relationship: while income is
crucial, education acts as an adoption incentive, especially in middle- and
upper-class households.
(c) The generational dimension
As discussed above, all studies show a strong generational
component to Internet adoption. While young people tend to use new
technologies intensively regardless of their level of education or income,
adoption is signifcantly lower among older persons. Internet usage data
suggest that there is a threshold at around the age of 30, after which the
adoption rate for this technology falls off signifcantly both in the home
and elsewhere. The case of Chile is illustrative: above the age of 30, the
percentage of non-Internet users doubles (Universidad Alberto Hurtado,
2009). In Peru, the percentage of Internet users drops from 61% among
young persons aged 19 to 24 to 37% among adults aged 25 to 40 (INEI,
2012). In Brazil, 81% of young people aged up to 24 use the Internet, while
less than half (48%) of adults aged 35 to 44 do.
Consequently, the likelihood of a household subscribing to an
Internet service is associated with two factors related to household age
composition. First, the age of the household head is a determinant of the
penetration level. As fgure II.11 shows, the older the household head, the
lower the level of service penetration. What is at work in this case is a
direct effect linked to the adoption drivers discussed earlier.
Figure II.11
COSTA RICA: HOME INTERNET AND COMPUTER USE, BY AGE GROUP, 2010
0
10
20
30
40
50
60
70
80
90
15 to 24 25 to 34 35 to 44 45 to 54 55 and over
Home fixed Internet Home computer
Source: Costa Rica, Telecommunications Directorate, 2011.
ECLAC 60
On the other hand, studies suggest an indirect effect linked to the
presence of school-age children in the home. Data for 2009 confrm this effect
in Chile, as the percentage of households with Internet access rises from
39% to 43% among those with school-age children (Universidad Alberto
Hurtado, 2009). In Peru, the effect is even more signifcant: the proportion
of Internet-connected households is 57% among those with school-age
children, as against 43% among households without school-age children
(INEI, 2012). This effect is replicated in Costa Rica, although when income
and education are controlled for it is not found consistently but is associated
primarily with households whose heads have a lower level of education.
In summary, the evidence bears out fndings on what are known as
“digital natives”, showing that there is a strong age effect on the likelihood
of Internet adoption. Despite the differences in the age groups used by
different countries, the results suggest that the threshold separating the
digital natives group (among whom Internet usage is widespread) from the
so-called digital immigrants is located at around the age of 30. Regarding
the household access demand gap, the evidence suggests that the presence
of school-age children in the home has a positive effect on the expected
adoption rate, but the magnitude of this effect is small compared to that
of the factors discussed earlier. These results highlight the importance
of implementing digital literacy programmes to offset any age-related
disadvantages and thereby reduce the demand gap.
(d) Other dimensions of the demand gap
The analysis reveals that the demand gap is associated, in the
following order, with factors of income, education and household
age composition. While the evidence indicates that these factors are
common to all countries in the region, some studies also point to other
factors operating more specifcally in particular contexts or countries.
This suggests a need to tailor demand-boosting initiatives to the unique
characteristics of the demand gap in the different countries of the region.
In some countries, evidence suggests that there is still a gender
gap in Internet adoption. This is the case in Chile, where the data for
2009 indicate that the likelihood of a household having Internet access
is reduced by 7 percentage points if the household head is female
(Universidad Alberto Hurtado, 2009). The magnitude of the gender gap
is similar in Peru, where Internet usage is also 7 percentage points lower
(31% as against 38%) among women (INEI, 2012). The result in Costa Rica
is consistent with this: while 63% of men use the Internet in the home, only
54% of women do (Telecommunications Directorate, 2011).
Brazil and Mexico, by contrast, do not present any signifcant
gender differences in penetration levels, suggesting that the gap tends to
disappear in countries with higher penetration. This is consistent with the
Broadband in Latin America: Beyond Connectivity 61
fndings of Hilbert (2011), who ascertained in a recent paper reviewing
evidence from 25 countries in Latin America and Africa that differences in
ICT adoption by gender tended to disappear when controlled for education
and income.
In countries where a signifcant portion of the population is
indigenous, data reveal the existence of a usage gap associated with
Spanish-language profciency, replicating the fndings in the United States
regarding English-language profciency among immigrant populations. In
Peru, for example, just 8% of the population whose childhood language
is other than Spanish uses the Internet, compared with 40% of those for
whom it is Spanish. These results need to be analysed in greater depth,
however, given that ethnic factors are strongly associated with economic
and educational factors and therefore do not provide a prima facie basis for
determining the marginal effect of language or ethnicity on the demand
gap. Nonetheless, the results are indicative of a need for development
initiatives to target demand in these groups, as they labour under a
number of disadvantages with regard to Internet use.
D. Public policies to close the demand gap
The analysis presented above provides a framework for designing public
policies to stimulate broadband adoption. The following recommendations
are organized in accordance with the different dimensions of the demand
gap identifed in the previous section.
1. Policies for addressing the affordability barrier
The evidence shows affordability to be one of the main determinants of
the demand gap in the region. In other words, for a signifcant portion of
households in Latin America (even those that already have a computer),
the supply of connectivity is not aligned with income. Considering this,
three types of tools should be considered to better match supply to income
levels and characteristics in this segment of the population.
A frst set of policy tools aims to encourage competition in order
to bring down access prices. The tools available to achieve this goal can
be divided, generally speaking, into two major groups: those designed
to increase competition between platforms, and those that encourage
competition between services on a single platform (Cambini and Jiang,
2009). While the frst model is associated with policies in the United States,
the second is particularly applicable in the European Union.
As Katz (2009a) has shown, for various reasons the competition-
between-platforms model dominates in Latin America. The frst question
to ask, therefore, is how likely it is that competition between vertically
ECLAC 62
integrated operators might bring about a signifcant reduction in prices.
The second is what should be done if competition between operators
does not bring prices down far enough. In this case, what kind of public
initiatives can be deployed to encourage the introduction of price plans
designed to make broadband more affordable?
The Latin American experience shows that healthy competition
between broadband platforms does lead to a race to introduce better
services (competition for speed) and lower prices. The downward price
trend resulting from competition can be observed in many countries of
Latin America: the lower the level of industry concentration (as measured
by the Herfndahl-Hirschman index), the lower the average price of fxed
broadband service (see fgure II.12).
4
Figure II.12
LATIN AMERICA: CONCENTRATION OF FIXED BROADBAND SUPPLY
AND AVERAGE DOWNLOAD PRICE PER MBPS
(Herfndahl-Hirschman index and PPP dollars)
Argentina
Colombia
Brazil
Costa Rica
Venezuela
(Bol. Rep. of)
Chile
El Salvador
Ecuador
Mexico
Paraguay
Dominican Rep.
Peru
Panama
Nicaragua
0
10
20
30
40
50
60
70
80
0 2 000 4 000 6 000 8 000 10 000 12 000
Source: Prepared by the authors on the basis of data from H. Galperín, “Precios y calidad de la banda
ancha en América Latina: Benchmarking y tendencias”, Documento de Trabajo, No. 12, Buenos Aires,
Centro de Tecnología y Sociedad/ Universidad de San Andrés, 2012; and R. Katz, 2010-2012: Avances
importantes en el desarrollo del sector de telecomunicaciones en América Latina, Caracas, Andean
Development Corporation (ADC), 2012.
There is evidence that this process is under way in the region. As
Galperín (2012) shows, the implicit download price per Mbps of fxed
broadband plans in Brazil dropped by 40% between 2010 and 2012. This
trend, spurred by competition between cable operators such as Net
Serviços and telephone operators such as Telefônica and Oi, is the result of
4
The chart excludes the countries with prices at either extreme (the Plurinational State of
Bolivia and Uruguay).
Broadband in Latin America: Beyond Connectivity 63
strategies designed to capture what is the largest market in terms of size
and growth, the so-called C-segment of the population. However, the lower
prices resulting from competition may not be suffcient to ensure that the
most disadvantaged can access the service. This requires consideration of
a second set of more closely targeted tools designed to stimulate demand
and reduce fnancial barriers to access among these sectors. The main
element in this toolkit needs to be the reduction or elimination of taxes
on basic connectivity plans, either across the board or for low-income
individuals and households.
Studies have shown that the tax burden associated with broadband
services is particularly high, discouraging consumption in both the fxed
and mobile segments.
5
The tax impact on broadband is negative across
several dimensions (see diagram II.2). According to this analysis, which
is borne out by the tax reduction policy and tax incentives introduced
in the United States to promote broadband adoption, a high tax burden
raises barriers to adoption by impacting affordability. Broadband network
externalities further constrain adoption. Then, because of the high fxed
costs involved in delivering broadband, a low adoption rate raises the
average cost, reduces margins and prevents operators from cutting prices,
generating a third limitation on adoption.
Diagram II.2
IMPACT OF FISCAL POLICY ON BROADBAND PENETRATION
Broadband
use taxes
Reduced lower
broadband
take-up adoption
Reduced
network
externalities
Higher barriers
to take-up
adoption
Lower
returns to
scale
Lower
margins
Higher
prices
Reduced
lower broadband
take-up adoption
Source: R. Katz, “La brecha digital: oferta o demanda?”, Notas Enter, No. 135, Madrid, 7 July, 2009.
Considering the signifcant price elasticity of broadband services in
the region, as revealed by studies such as those of Galperín and Ruzzier
(2010) and Macedo and Carvalho (2011), reducing the tax burden could
5
See Galperín and Ruzzier (2010) for an analysis of the fixed segment and Katz and others
(2011) for the mobile segment.
ECLAC 64
have an immediate impact on the level of service adoption. Several
countries in the region have implemented such initiatives, notably Brazil,
where basic access plans in several states have been exempted from the
ICMS, a state goods and services tax with rates of up to 35%, as part of
the National Broadband Plan, and Colombia, where broadband services
for lower income strata have been exempted from value added tax (VAT).
A third set of tools involves a more proactive role for government
in the broadband supply structure. In countries where the incumbent
telephone operator is government-controlled, supply initiatives involving
very low-cost connectivity plans have been implemented. While offering
a limited service in terms of speed (between 256 kbps and 512 kbps) and
data download capacity, they do provide basic access options that work
as a frst connectivity step for low-income households. Uruguay (the
Universal Hogares plans offered by ANTEL) and the Bolivarian Republic
of Venezuela (the ABA plans of CANTV) furnish notable examples. It is
noteworthy that these plans involve a mixed payment system (a fxed
monthly payment with the option to purchase incremental data download
capacity on top of the monthly allowance), a model that has been crucial in
achieving mass adoption of mobile telephony services.
In other cases, governments have opted for direct investment in
backbone infrastructure initiatives, making it possible to structure a supply
of low-cost connectivity plans in which the private operator commits to
providing the service in accordance with government-mandated quality
and price parameters in exchange for access to the network backbone.
Examples are the National Broadband Plan in Brazil and the Argentina
Conectada plan in Argentina. Lastly, there are the more traditional supply
subsidies in areas of low private-sector return, where governments set
service features in advance and tender out the service to a private operator
which then receives the subsidy. Todo Chile Comunicado in Chile and
Vive Digital in Colombia operate in this way (see the chapter by Galperín,
Mariscal and Viecens in this book).
In conclusion, there is a wide range of public policy tools that can
encourage lower prices and the segmentation of supply into products that
are attractive to households with limited budgets. Experience suggests
that the combination of healthy competition between access platforms,
a reduced tax burden and strategic State investment in non-competitive
segments of the network can signifcantly lower barriers to broadband
affordability in the region.
2. Policies for addressing the skills barrier
The usage skills dimension of the demand gap presents enormous
challenges for the countries in the region, as it is associated with structural
Broadband in Latin America: Beyond Connectivity 65
defciencies in education systems. Without attempting to cover the broad
debate on the issue, it can be said that there are two types of initiatives
to meet these challenges. First, there are instruments designed to address
specifc gaps in computer and Internet usage skills. These initiatives
are usually implemented outside the formal education system and are
often associated with vocational training plans or other types of social
integration initiatives.
There are several examples of such initiatives, both in the region and
in more developed countries. They typically involve provision of shared
access by the State, combining connectivity with digital training at access
points. Vive Digital in Colombia, Pontos de Cultura in Brazil and Puntos
de Acceso Digital in Argentina are just some examples. Implementation
often targets specifc population segments, examples being programmes
that seek to integrate young persons into labour markets by providing ICT
training (Mariscal, Gutiérrez and Botelho, 2009).
Despite the benefts yielded by these initiatives, there are no
rigorous studies of the outcomes of such programmes in the region. This
type of analysis is important because these programmes compete in
several respects with the wide range of shared access and training options
that exist in the private sector. Moreover, as Garrido and others (2012) have
argued in their review of the literature on ICT skills and applicability, the
success of such programmes largely depends on a good match between the
type of training received and the characteristics of the target population,
as well as demand in the local job market.
In the second place, there are longer-term initiatives tied to the
national education system that seek to equip people with the long-term
ability to use and adopt new technologies. The most ambitious initiatives
involve efforts to generate ICT skills within the education system by seeing
to it that all State schools have equipment and broadband connectivity.
Programmes of this type have been widely adopted in the region,
including in Uruguay (the Ceibal plan), Argentina (the Conectar Igualdad
plan) and Chile (the Enlaces programme).
These plans have been the subject of a great deal of debate
that is beyond the scope of this chapter. In particular, the evidence on
the educational results of these initiatives has been much discussed.
6

However, there is consensus on the need to adapt the education system
to the new technology skill requirements created by Latin America’s
ongoing integration into a world that is increasingly interconnected
from both an economic and a cultural point of view. Conceptually, this
change should take place at all the different levels of education (primary,
6
See IDB (2011), among others.
ECLAC 66
secondary and tertiary) and be supplemented by training initiatives for
segments of the adult population that are at a disadvantage in terms of
their ICT training opportunities.
In this regard, initiatives undertaken by some countries that have
been leaders in broadband adoption deserve a close look. These include
digital literacy initiatives implemented in the Republic of Korea, which
has launched an online education programme for 10 million people (about
20% of the population) focused on promoting skills among older persons,
housewives, members of the armed forces, prisoners, persons with
disabilities and other groups that have fallen behind in Internet adoption.
3. Policies for addressing the lack of interest/
relevance barrier
Lack of interest or relevance consistently comes up as one of the reasons
cited by non-Internet users, regardless of their usage skills or income.
This factor does not represent a barrier as such, and it is associated
with preferences and incentives that vary from person to person. This
represents a challenge for the development of public initiatives to address
this dimension of the demand gap. However, Internet adoption studies
have revealed a variety of mechanisms that come into play in the adoption
decisions made by potential users, suggesting several possible public
policy tools.
First, Internet access is of little worth in itself in the absence of
so-called complementary goods that confer value on such access. They
include applications and content that users value and should therefore
be attractive enough to encourage people to purchase the service. With
applications whose main function is communication between users
(e-mail, social networking and fle sharing programs), usage value
increases with the number of users (the classic network effect), suggesting
that the very dynamics of adoption will lead to a gradual closing of
the demand gap in the medium term. However, the evidence indicates
that this may not be enough of an incentive for certain groups of users
to become broadband subscribers, particularly when many of these
occasional users prefer to avail themselves of the numerous public access
points that exist in the region.
This being so, public initiatives should aim at providing high value-
added applications and demonstrating tangible benefts to potential users
in the form of time or money savings or skills acquisition. This is the case
with e-government applications that are designed to optimize citizen
interaction with government and offer benefts in terms of user access
to government services. The linkage between broadband connectivity
Broadband in Latin America: Beyond Connectivity 67
plans and the development of e-government platforms is a cornerstone of
national broadband plan design. There are numerous ongoing experiences
of this type in the region that should be monitored so that best practices
can be shared with the rest of the countries.
A similar dynamic arises with content. A decade ago, the lack
of content in Spanish and Portuguese was a barrier to the widespread
adoption of Internet services. Today, the content available in both is very
extensive, if less so than in other languages, and government initiatives
should therefore be aimed at high-impact social content (like that
designed to complement ICT initiatives in schools) or content aimed at
specifc segments of the population (in the region’s indigenous languages,
for example).
Public initiatives play an important role as catalysts for new private-
sector ventures to provide local digital content and applications. Where
applications are concerned, a regulatory environment conducive to the
development of e-commerce is vital to provide legal security for providers
and customers alike. For content, the region has a substantial installed
capacity for audiovisual production that has, in recent years, expanded to
multimedia content along with information technology services, another
industry that has grown signifcantly over the last decade. Programmes
to foster closer ties between university-based research and the production
sector are also needed for the development of larger private ventures in the
local content creation sector. Promotion of these high innovation content
activities and the skilled jobs they provide is essential and forms part of
several national broadband development plans in the region.
Lastly, bandwidth consumption is akin to what is known in the
literature as an “experience good” (Shapiro and Varian, 1999). This means
that consumers lack information about the quality and value associated
with a good, the only known information being its price. It is in the act
of consumption that the value of the good is revealed and demand for it
generated. In policy terms, this highlights the importance of public access
initiatives that can successfully bring technology to segments of the
population with limited opportunities to experience the service in other
areas such as the workplace or school.
4. Programmes targeting specific population segments
The evidence set out above suggests that the demand gap has specifc
sociodemographic components calling for demand promotion policies
designed to target different population groups. The data reviewed here
have revealed three groups on which such policies should operate. First
comes the strong generation component of the demand gap, manifested
ECLAC 68
in a progressive decline in the likelihood of adoption from the age of 30,
with a sharp drop in the adoption rate after age 55. Clearly there is a need
for digital literacy plans designed for adults and older persons, like those
implemented in the Republic of Korea. In several cases, these programmes
have proved more effective when they have combined training with
subsidies for the purchase of computers and the provision of distance
learning programmes.
Secondly, there is some evidence of a persistent gender gap, albeit
one that is signifcantly smaller than in the past. The gap is particularly
noticeable among households headed by women compared with male-
headed households, as the data for Chile show. Digital literacy and
equipment programmes for this segment should consider the particular
characteristics of such households, taking advantage of the possible
presence of children there, which evidence shows to be a factor that
boosts the incentives for adoption, albeit marginally. Again, there have
been success stories with vocational training for women linked to ICT
training that has addressed the gender gap as part of a broader issue of
employment opportunities.
E. Conclusions
While recognizing that investment in broadband infrastructure plays an
important role in reducing the digital divide, this chapter has studied
a topic that is less often mentioned: the demand gap. The statistical
information that is beginning to be generated by national studies has
made it possible to reach an incipient understanding of the importance
of this gap. Of the variables accounting for the demand gap, some are
structural (e.g., household age composition and education level) while
others refect the intensity of competition in the broadband market and
economic performance in general (e.g., affordability).
As regards the most important barrier identifed in national studies
(the issue of affordability), two types of initiatives can address this. On
the one hand, competition between platforms over the long run is the
right model to trigger price reductions among the operators serving the
market. This will involve not only the traditional telecommunications
operators and cable television service providers, but also, in the coming
years, operators providing mobile broadband. While this was originally
conceived as a technology that would be acquired by the same population
sector that subscribed to fxed broadband (complementary), various
indications suggest that this product will become a substitute for fxed
broadband and bring prices down over the medium term.
Broadband in Latin America: Beyond Connectivity 69
The second initiative to address the affordability barrier is
government intervention in broadband provision, either directly (as in
Uruguay and the Bolivarian Republic of Venezuela), through backbone
infrastructure investments and agreements with private operators (Brazil
and Argentina), or by means of traditional systems of supply subsidies
(Chile and Colombia). These initiatives aim to provide basic access plans at
a very low cost, allowing low-income households to gain an appreciation
of the service and turning many users who now access the Internet at
shared access points into subscribers. By reducing the demand gap, these
initiatives can generate the economies of scale and network externalities
necessary for the healthy development of the broadband ecosystem in the
region’s markets.
The skills barrier, for its part, can be infuenced by public policies—
not just traditional ones involving the formal education system, for
instance, but continuing education initiatives as well. Some governments
have made signifcant progress in this area as they have pursued universal
broadband policies. Others, facing a need to create jobs in the short term,
have preferred to allocate resources to infrastructure deployment to cover
households that are served poorly or not at all by existing networks.
This chapter has highlighted the need to support such initiatives with
others that target the factors inhibiting increased penetration among the
population already covered by broadband. Ultimately, an understanding
of both aspects of the digital divide is required for public policymaking
based on the goals to be maximized.
Because broadband is a complex general-purpose technology, its
adoption involves the management of a larger number of variables than
is the case with mobile telephony, whose exponential growth in the region
has made up for the shortcomings of fxed telephony service coverage.
Closing the demand gap seen not only in Latin America but in many
countries of the industrialized world will require more active participation
by governments and the private sector in order to address and remove the
barriers to adoption.
ECLAC 70
Bibliography
Cambini, C. and Y. Jiang (2009), “Broadband investment and regulation: a literature
review”, Telecommunications Policy, vol. 33.
Chaudhuri, A. and K. Flamm (2005), “An analysis of the determinants of Internet
access”, Telecommunications Policy, vol. 29.
FCC (Federal Communications Commission) (2012), Eight Broadband Progress Report,
Washington, D.C.
Galperín, H. (2012), “Precios y calidad de la banda ancha en América Latina:
benchmarking y tendencias”, Documento de Trabajo, Nº 12, Buenos Aires, Centro
de Tecnología y Sociedad/Universidad de San Andrés.
Galperín, H. and C. Ruzzier (2010), “Las tarifas de banda ancha: benchmarking
y análisis”, Acelerando la revolución digital: banda ancha para América Latina y el
Caribe, LC/R.2167), Valeria Jordán, Hernán Galperin and Wilson Peres (coords.),
Santiago, Chile, Economic Commission for Latin America and the Caribbean
(ECLAC)/Regional Dialogue on the Information Societyn (DIRSI), November
[online] http://www.eclac.cl/publicaciones/xml/7/41727/LCR.2167.pdf.  
Garrido, M., J. Sullivan and A. Gordon (2012), “Understanding the links between
ICT skills training and employability: an analytical framework”, Information
Technologies and International Development, vol. 8, No. 2.
Grazzi, M. and S. Vergara (2011), “Determinants of ICT Access”, ICT in Latin America:
A Microdata Analysis (LC/R.2172), M. Balboni, S. Rovira and S. Vergara (eds.),
Santiago, Chile, Economic Commission for Latin America and the Caribbean
(ECLAC).
Hauge, J. and J. Prieger (2010), “Demand-side programs to stimulate adoption of
broadband: what works?”, Review of Network Economics, vol. 9, Nº 3.
Hilbert, M. (2011), “Digital gender divide or technologically empowered women
in developing countries? A typical case of lies, damned lies, and statistics”,
Women’s Studies International Forum, vol. 34, No. 6 .
Horrigan, J. (2009), Home Broadband Adoption 2009, Washington, D.C., Pew Internet
and American Life Project.
INEGI (National Institute of Statistics and Geography) (2010), Estadísticas sobre
disponibilidad y uso de tecnologías de la información y comunicación en los hogares,
Mexico City.
INEI (Instituto Nacional de Estadística e Informática) (2012), Las tecnologías de
información y comunicación en los hogares, Lima.
Katz, R. (2012), 2010-2012: Avances importantes en el desarrollo del sector de
telecomunicaciones en América Latina, Caracas, Development Bank of Latin America.
___ (2009), “La brecha digital: oferta o demanda?”, Notas Enter , No. 135, Madrid,
7 July.
___ (2008), La competencia entre plataformas: teoría y resultados, Madrid, Enter.
Katz, R. L., E. Flores and J. Mariscal (2011), The Impact of Taxation on the Development
of the Mobile Broadband Sector, Londres, GSMA.
Macedo, H., and A. Carvalho (2011), Broadband Economic Impact in Brazil: A
Simultaneous Equations Analysis, paper presented at the Acorn-Redecom
Conference, Lima, November.
Broadband in Latin America: Beyond Connectivity 71
Mariscal, J., L. Gutiérrez and A. Botelho (2009), “Employment and youth
inclusion into the labor force via training in information and communication
technologies: the cases of brazil, colombia, and mexico”, Information Technologies
and International Development, vol. 5, No. 2.
Navarro, L. and M. Sánchez (2011), “Gender differences in Internet use”, ICT in
Latin America: A Microdata Analysis (LC/R.2172), M. Balboni, S. Rovira and
S. Vergara (eds.), Santiago, Chile, Economic Commission for Latin America and
the Caribbean (ECLAC).
NTIA (National Telecommunications and Information Administration) (2011),
Exploring the Digital Nation, Washington, D.C.
OFCOM (2012a), Adults Media Use and Attitudes Report, London.
___ (2012b), The Communications Market, London.
Ono, H. and M. Zavodny (2008), “Immigrants, English ability, and information
technology use”, Social Forces, vol. 86, No. 4.
ONTSI (Observatorio Nacional de las Telecomunicaciones y de la Sociedad de la
información) (2012), La Sociedad en Red. Informe Anual 2011, Madrid.
Pew Center (2012), Digital Differences, Washington, D.C., Pew Internet and American
Life Project.
Shapiro, C. and H. Varian (1999), Information rules: A Strategic Guide to the Network
Economy, Cambridge, Massachusetts, Harvard Business School Press.
Telecommunications Directorate, Government of Costa Rica (2011), II Evaluación de
la Brecha Digital en el Uso de Servicios de Telecomunicaciones en Costa Rica, San Jose.
Universidad Alberto Hurtado/SUBTEL (Department of Telecommunications)
(2009), Encuesta sobre acceso, uso y usuarios de internet banda ancha en Chile,
Santiago, Chile.
Vicente, M. and A. López (2006), “Patterns of ICT diffusion across the European
Union”, Economic Letters, vol. 93.
ECLAC 72
Annex
Annex II.1
Methodology and sources used to calculate
broadband coverage
The methodology and sources used to estimate broadband coverage are
listed below. Estimates were calculated at the lowest possible level of
disaggregation considering the availability of data in each case. They are
based on the total population of each administrative unit and therefore
tend to overstate coverage, since in many cases the population residing
in a given administrative unit (usually municipal or departmental) is not
fully served.
Fixed broadband coverage:
• Argentina: authors’ estimate based on 2012 coverage data
published by the Ministry of Planning for towns, except in the
provinces of Córdoba and La Rioja, where the estimate was
carried out at the departmental level.
• Bolivia (Plurinational State of): authors’ estimate based on 2012
coverage data from the operator Entel. The estimate was carried
out at the whole city level, so coverage is overstated.
• Brazil: 2010 information published by ANATEL.
• Chile: authors’ estimate based on SUBTEL data for 2011.
• Colombia: authors’ estimate based on municipalities with at
least 50 fxed broadband connections as of the second quarter
of 2011, according to Ministry of Information Technology and
Communications data (based on SIUST).
• Costa Rica: authors’ estimate based on coverage data from
providers per district in “Estrategia Nacional de Banda Ancha”,
book 2, Diagnóstico.
• Ecuador: authors’ estimate based on the cantons where providers
reported having at least one customer in service for 2011.
• Mexico: Ministry of Communications and Transport estimate
for 2011.
• Peru: calculated from the number of residents in districts where
there was at least one client with ADSL as of December 2010.
• Uruguay: information provided by ANTEL, November 2012.
Broadband in Latin America: Beyond Connectivity 73
Mobile broadband coverage:
• Argentina: authors’ estimate based on 2012 coverage data from
operators published by the Ministry of Planning for towns,
except in the provinces of Córdoba and La Rioja, where the
estimate was carried out at the departmental level.
• Bolivia (Plurinational State of): authors’ estimate based on 2011
coverage data from Entel.
• Brazil: 2011 information published by ANATEL.
• Chile: authors’ estimate based on SUBTEL data for the
population without access to 3G coverage in 2011.
• Colombia: authors’ estimate based on 2012 municipal coverage
data from Movistar.
• Costa Rica: authors’ estimate based on SUTEL 3G coverage
testing of ICE.
• Dominican Republic: authors’ estimate based on Orange 2011
coverage data.
• Ecuador: based on Movistar coverage; a canton is considered to
have 3G service if the provider covers the most populated part of it.
• Mexico: based on COFETEL information on the population with
3G service coverage for 2011.
• Peru: calculated from the number of residents in districts where
there is a 3G connection (348 districts covered by 3G out of a total
of 1,833) as of December 2010 for mobile telecommunications.
The estimate matches the one reported by the ITU.
Chapter III
Regional and international connectivity
Omar de León
A. Introduction
This chapter examines Internet connectivity in Latin America and the
Caribbean, looking at IP capacity and transit between points in the region
and in comparison with the Northern Hemisphere, as well as wholesale
and end-user pricing. It identifes the main quality parameters for
broadband optimization, their impact on applications and content, and the
technical factors, such as aggregation, that infuence prices.
1
Present-day international connectivity and the bottleneck in the
Northern Hemisphere are analysed, along with the price levels revealed by
ad hoc studies. Physical routes between countries in this region are scarce,
and many —if not most— logical routes between countries, even neighboring
ones, are through connection points in the Northern Hemisphere. Even
Internet service providers (ISPs) in the same country often connect to one
another through other countries, which makes for a very ineffcient use
of resources. This situation is the result of a shortage of Internet Exchange
Points (IXPs) in many countries, and in Latin America as a whole.
2
1
For a detailed analysis of the concept of broadband and a proposal for Latin America, see
annex III.1, in which broadband is defined in relation to its use.
2
Annex III.2, describes the main characteristics of IXPs and criteria for evaluating them
based on international best practices.
ECLAC 76
This chapter identifes the main factors infuencing quality and
prices, including IXPs, diffculties in local hosting of national and foreign
content, the lack of effcient connectivity between countries and the lack
of competition in international connectivity, due, among other things, to
limited demand. On this basis, it suggests that policy actions should be
oriented towards improving conditions for supply and demand.
The international comparison with Europe illustrates differences
in supply and demand that, in turn, infuence wholesale and end-user
prices. Data on wholesale pricing in three regions (Europe and the United
States, Latin America and South-East Asia) show that Latin America is at a
pricing disadvantage compared with Europe and the United States, where
Internet access is highly concentrated. As for Asia, the price difference is
on the order of 2:1, which is smaller than the approximately 6:1 ratio with
the United States and Europe.
B. Internet connectivity
This section examines Internet connectivity in the countries in South
America with respect to available connection speeds, service quality
parameters and reference rates. It also evaluates the types of services that
can be provided and the quality of access offered by operators.
1. Factors that affect connectivity
The development of Internet access in South America poses a number
of problems that have a direct impact on connection speeds, quality
parameters, teledensity and rates.
Economies of scale. The economies of scale seen in the more advanced
countries are not found in most of the countries of South America, whether
due to total population size or to the distribution of income and per capita
income in relation to the intrinsic costs of broadband.
Cost of international Internet access. The longer distances involved in
connecting the countries of South America (and Latin America in general)
to the global Internet (transit on Tier 1 ISPs in Northern Hemisphere
countries) necessarily entail higher costs because of the capacity required
for such connections.
Imperfect direct regional connectivity among the countries of the
region, and within each country as well, makes it diffcult to optimize end-
user prices because of unnecessary additional transmission costs. IXPs are
one solution for interconnection within the same country, and they are the
basis for regional connectivity.
Remote content hosting. Because hosting prices are relatively high
in the countries of the region, most regional content is hosted outside
Broadband in Latin America: Beyond Connectivity 77
the region, which increases access costs. In any case, mass remote data
hosting by small-scale content providers differs from that of large-scale
content providers that use intelligence provided by content distribution or
delivery networks (CDN). These providers carry between 30% and 40% of
worldwide residential traffc, so policies favouring hosting at nodes in the
region would signifcantly improve Internet quality and pricing.
2. Quality parameters
The main quality parameters are download speed, upload speed and
network latency. Another important parameter, though it does not have
much impact on the region’s networks because of its high quality, is
the packet loss rate. Data on access speed in Latin American countries
are presented below (see table III.1).
3
The fgures are based on actual
measurements carried out by Ookla, an engine that gauges quality
parameters. The average values per country date are as of 1 December 2012.
Table III.1
INTERNET SPEEDS IN THE REGION ACCORDING TO OOKLA, 1 DECEMBER 2012
(Mbps and percentages)
Country
Download
speed
Upload
speed
Change in download
speed March-
December 2012
Change in upload
speed March-
December 2012
Argentina 4.23 1.13 12.5 19.0
Bolivia (Plurinational
State of)
1.24 0.53 96.8 76.7
Brazil 6.79 1.46 10.9 10.6
Chile 9.37 2.66 32.3 46.2
Colombia 3.92 1.55 21.4 19.2
Costa Rica 2.57 0.73 15.3 10.6
Ecuador 3.78 2.98 44.3 40.6
El Salvador 3.21 1.67 17.6 7.1
Guatemala 2.63 1.26 0.4 10.5
Honduras 2.81 1.77 7.7 14.9
Mexico 6.98 2.09 41.0 25.9
Nicaragua 3.38 2.03 6.0 0.5
Panama 3.67 1.26 24.8 -3.8
Paraguay 2.90 1.94 22.9 50.4
Peru 2.89 0.65 114.1 109.7
Uruguay 7.83 2.04 165.4 200.0
Venezuela (Bolivarian
Republic of)
1.95 0.45 27.5 15.4
Latin America average 4.13 1.54 32.1 28.6
Source: Ookla.
3
The data are published by Netindex [online] www.netindex.com.
ECLAC 78
There is a wide range of speeds, with some countries attaining only
a third of the regional average. In general, Latin American ISP download
and upload speeds are one third and one half, respectively, of those in
Europe. The rates of change are very similar for the two speeds, which
indicates that the trend is for the gap to remain the same.
Akamai’s fndings are presented as an alternative measure. They
are based on 1 trillion direct daily measurements of content requests
supported by its 100,000 servers distributed throughout the world (see
table III.2).
4
The table shows average values as well as percentages of
providers adopting speeds greater than 4 Mbps and 10 Mbps (includes
speeds greater than 4 Mbps) and those under 256 Kbps. The differences
between these and the Ookla values may be due to the fact that Akamai
directly measures the download speed of its sites, no matter where they
are located, whereas Ookla measures against servers that are usually close
to the end-user.
Table III.2
INTERNET SPEEDS IN THE REGION ACCORDING TO AKAMAI
Average
connection
speed (Kbps)
High-speed
broadband
adoption rate
(> 10 Mbps)
Broadband
adoption rate
(>4 Mbps)
Adoption rate
for speeds
below 256
Kbps
Argentina 2 244 6.35 41.80 2.67
Bolivia (Plurinational
State of)
572 0.09 1.67 19.06
Brazil 2 168 6.44 39.94 5,56
Chile 3 406 10.84 84.76 0.21
Colombia 2 653 4.69 64.47 0.43
Costa Rica 1 805 1.22 29.77 1.47
Ecuador 1 740 1.93 32.07 1.18
El Salvador 1 847 4.62 28.60 7.12
Guatemala 2 647 5.98 63.63 1.98
Honduras 1 749 2.35 30.94 1.72
Mexico 2 794 4.46 70.11 0.79
Nicaragua 1 303 0.98 12.69 2.52
Panama 2 783 3.33 74.28 0.89
Paraguay 1 238 0.32 7.89 1.15
Peru 1 644 0.42 25.64 0.87
4
See www.akamai.com and the company’s comments at https://blogs.akamai.
com/2011/11/the-future-internet.html.
Broadband in Latin America: Beyond Connectivity 79
Table III.2 (concluded)
Average
connection
speed (Kbps)
High-speed
broadband
adoption rate
(> 10 Mbps)
Broadband
adoption rate
(>4 Mbps)
Adoption rate
for speeds
below 256
Kbps
Uruguay 1 326 0.07 7.78 1.41
Venezuela (Bolivarian
Republic of)
905 0.32 4.46 6.52
Spain 4 614 27.48 87.51 0.69
England 5 576 39.42 93.10 0.35
France 4 889 31.18 90.56 0.16
Italy 4 147 17.43 89.16 0.65
Portugal 5 366 43.89 88.81 0.28
Japan 10 918 65.76 88.18 0.87
Source: Akamai and data processed by the author.
3. Reference prices
Table III.3 shows published reference prices (not those actually charged)
for fxed and mobile broadband in the countries of Latin America,
expressed in dollars per month. Prices for fxed Internet access with the
smallest number of additional services and prices for download speeds
closest to 2 Mbps (the speed defned in annex III.1 as advanced broadband)
are listed for each country and for major operators. The price per Mbps
for these services is calculated including taxes but not including initial
promotions. Given that traffc is more important than speed for mobile
broadband, within reasonable ranges, prices are for traffc close to 3
gigabytes, regardless of speed.
The wide range of prices can be attributed to factors such as
country size, the amount of competition in the access market and in
submarine cables, economies of scale and public policies. In Paraguay and
the Plurinational State of Bolivia, high prices also refect their status as
landlocked countries. The price difference for mobile broadband is smaller
than in more advanced countries because of better use of economies of
scale (high teledensity) and less international transit.
ECLAC 80
Table III.3
REFERENCE PRICES FOR FIXED ACCESS PER MBPS FOR DOWNLOAD
SPEEDS OF APPROXIMATELY 2 MPBS, NOVEMBER 2012
Country
Fixed broadband Mobile broadband
Average price
of 1 Mbps
(dollars)
Price as
percentage of
monthly per
capita GDP
Average price
(dollars)
Price as
percentage of
monthly per
capita GDP
Argentina 12.13 1.3 29.80 3.3
Bolivia (Plurinational
State of)
63.61 31.5 22.83 11.3
Brazil 19.57 1.9 34.40 3.3
Chile 13.18 1.1 33.78 2.8
Colombia 19.53 3.3 19.99 3.4
Costa Rica 13.07 1.8 27.12 3.8
Ecuador 14.77 3.9 28.75 7.6
El Salvador 13.65 4.4 16.50 5.4
Guatemala 31.20 11.8 22.21 8.4
Honduras 16.92 9.1 18.08 9.7
Mexico 9.48 1.1 25.11 3.0
Nicaragua 17.25 16.7 22.43 21.7
Panama 7.26 1.0 14.95 2.1
Paraguay 20.06 6.6 19.56 6.5
Peru 32.77 6.5 30.91 6.2
Uruguay 12.58 1.1 18.80 1.6
Venezuela (Bolivarian
Republic of)
35.01 3.9 24.63 2.7
Average 20.71 6.3 24.11 6.0
Spain 2.88 0.1 34.52 1.3
France 2.11 0.1 45.22 1.3
England 0.90 0.1 20.03 0.6
Italy 4.64 0.2 28.52 1.0
Portugal 3.50 0.2 17.19 1.4
Japan 2.15 0.1 32.48 0.9
Source: Economic Commission for Latin America and the Caribbean (ECLAC), Regional Broadband
Observatory, on the basis of survey of operators’ web pages.
4. Aggregation factor
In addition to the speed and latency parameters, there is an important
factor about which little is usually known and which is related to
service quality at peak Internet use times, such as early evening. It is
the aggregation factor, a parameter that operators deal with specifcally
Broadband in Latin America: Beyond Connectivity 81
at points in the network where bottlenecks occur. The effect of applying
inappropriately high values is packet loss and retransmissions that slow
down effective data traffc.
Aggregation is a standard part of ISP network design, especially for
international links and shared access such as wireless networks, FTTH
GPON (fber-to-the-home with Gigabit Passive Optical Networks) and HFC
networks (hybrid fbre coaxial networks for subscription cable television).
It is based on the fact that all users are not usually on the Internet at the
same time, and that even when they are online, not all of them download
information at the same time. For this reason, ISPs design bottlenecks into
their networks, including international access, for average customer use.
For international links, the correct value depends on the country, because
some countries have a lot of content hosted within their borders, or the
ISP may have large caching facilities. Generally it is on the order of 20:1;
when international linkage prices rise signifcantly, ISPs use higher values,
which reduces quality during peak hours. When a lot of content is hosted
locally, there is less usage of international channels and it is possible to use
larger aggregation factors without sacrifcing service.
C. South America’s dependence on international links
This section examines Internet connectivity, how international links
impact costs, and factors that have an indirect infuence on costs, such as
CDN storage.
There is very little competition among international cable operators
in Latin America in comparison with regions in the Northern Hemisphere,
primarily because demand has been and remains much smaller than in
the latter and because existing cables still have the capacity to expand.
However, that capacity could decline signifcantly in the near future.
Demand in Latin America (which, according to Cisco, is rising at a
compound annual rate of 51%) is likely to gradually exhaust the available
supply. This trend in demand could lead to private initiatives and inter-
government or single-government initiatives as well. The same market
arguments concerning South America apply to Central America as well;
5

prices and quality tend to vary for contracts with similar durations, and
prices are considered higher than in the Southern Cone.
5
New cables linking the region to Europe and Africa are under construction, as are cables
within the region and connecting to the United States. This could drive prices down further.
ECLAC 82
1. The international bottleneck and its
strategic importance
For historical reasons, the NAP of the Americas was completed a decade
ago. Today it is one of several major telecommunications facilities located
in the state of Florida. This initiative has generated substantial economies
of scale that result in almost all traffc between Latin America and the rest
of the world going through Florida, or, to a lesser extent, through the east
coast of the United States.
Latin America’s traffc with the rest of the world and even among
countries in the region is highly dependent on the United States, which
creates a strategic infrastructure challenge. Aside from this connection,
there is only the Atlantis 2 cable; its maximum capacity is 160 Gbps,
an insignifcant volume considering current and projected traffc.
Brazil recently announced a new cable linking it to Europe that would
substantially improve this situation, as well as a cable to Africa that may
connect with pan-African cables and provide an alternative access route
to Europe.
2. Regional interconnection
(a) Types of arrangements
There are different options for obtaining international and regional
connectivity, two of which involve peering:
(i) An ISP contracts with an international carrier for conveyance
capacity through a dedicated circuit to an Internet access point,
such as an IXP in Miami, and makes a separate purchase for access
to Internet transit at the other end. In this case, all traffc that
is exchanged uses an international circuit that always goes frst
through a remote IXP and is then routed to the end destination.
(ii) An ISP contracts with an international carrier, for direct access
to the IP network, also known as an IP transit service, for direct
access to the IP network. This gives it access to the entire Internet.
In this case, the carrier makes (technically or economically)
effcient use of its own network by exchanging the ISP’s traffc
at the point that best suits its needs, using the server with which
the ISP exchanges traffc.
(iii) Two ISPs can engage in private peering between themselves by
giving each other access to their networks. This option is chosen
Broadband in Latin America: Beyond Connectivity 83
when the ISPs have a mutual interest in interconnecting their
networks. They generally do not pay each other. This arrangement
tends to occur between major global operators, but sometimes it
involves connections between ISPs in neighboring countries.
(iv) In contrast to the private peering discussed in the previous
point, public peering is also viable for IXPs. In public peering,
an ISP interconnects with other ISPs through a single port.
The choice of one approach or another depends on the prices quoted
for capacity, transit, the interest in peering, and each ISP’s traffc profle. In
general, these four basic types of contract are available for medium-sized
and large access providers.
(b) Intra- and extra-regional capacity according
to TeleGeography
The TeleGeography study (2011) of Internet interconnection route
capacity is a partial analysis of traffc in Latin America that gives an
indication of the actual traffc situation. A link between two countries
may be carrying traffc with other destinations, and the same may be
true of links used to go from one country to a landing station located
in another country. The section below examines traffc itself based on
route studies.
The links between Latin America and the United States represent
85.5% of the capacity of all international links in the region. Intraregional
links, some of which are used to access the global Internet, represent 14.3%
of the total, whereas the capacity of links to Europe accounts for only 0.2%.
Nearly three quarters of total capacity in Latin America and the Caribbean
is in South America, a little less than one quarter is in Central America and
Mexico, and approximately 3% is in the Caribbean. Map III.1 shows the
distribution of international traffc capacity by major route.
This marked concentration of interconnection capacity with the
United States works to the detriment of intraregional connection capacity.
To date there are no reliable overall measures of the percentage of
intraregional capacity going through the United States, but it is estimated
to be about 15%. On top of that is the traffc that is exchanged with the
United States but whose true purpose is to obtain information from
regional sites (such as newspapers, TV channels and radio stations) located
in that country. It is clear that not all of that traffc needs to pass through
the United States.
ECLAC 84
Map III.1
CAPACITY OF PRINCIPAL INTERNATIONAL ROUTES IN LATIN AMERICA
United States
Dominican Rep.
Mexico
Costa Rica
Panama
Colombia
Ecuador
Peru
Chile
Argentina
Uruguay
Brazil
Venezuela
(Bol. Rep. of)
Aggregate international
Internet capacity (Gbps)
Source: TeleGeography, Global Internet Geography - Latin America, 2011.
Note: The boundaries and names shown on this map do not imply offcial endorsement or
acceptance by the United Nations.
Regional transit routes
Looking at domestic interconnection, it is diffcult to identify all of
the interconnection relationships between ISPs. This section examines two
partial alternatives. Neither one is a complete alternative —the frst one
because of practical diffculties and the second one because of problems
with the underlying information.
Methodology based on the Tracert command-line utility
Using the Tracert command-line utility,
6
a sample of routes was
traced from multiple countries to the main content sites used, which in
6
Traceroute is a diagnostic tool for tracking packets that come from a single network point
(host). An RTT or network latency statistic is obtained for these packets, which is an
Broadband in Latin America: Beyond Connectivity 85
turn are located in multiple countries, to observe the behaviour of routes
in the region. This methodology makes it possible to identify not only
national and regional interconnections, but also all of the routes used in
the domestic or regional interconnection, by identifying all intermediate
routers used. The following behaviours have been identifed:
(i) A substantial number of regional-content servers are located in
the United States and, to a lesser extent, in Europe.
(ii) CDNs are deploying more and more servers in the region,
reducing the number of network hops required to access content.
(iii) ISPs that contract for IP transit from their countries have more
direct routes to the countries of the region.
(iv) ISPs that base their connectivity on capacity going to the United
States and then transit on an IXP far from that country can access
sites located in the region through that IXP.
(v) For several countries in the region, interconnection between
ISPs in the same country takes place abroad.
Although this methodology yields clear and verifable results based
on the IP addresses of the interfaces of the routers through which the
information travels, it is not scalable for conducting a complete evaluation
of domestic and regional interconnection usage, nor does it help determine
the volumes of traffc actually exchanged on each route.
Methodology using Default-Free Zone
Another methodology for evaluating the region’s connectivity,
which includes domestic connectivity for each country, is based on the
analysis of the default-free zone (DFZ)
7
or public IPv4 routing tables for
interconnections between autonomous systems (AS)
8
in a single country
(domestic interconnection) and between different countries in a given
region (regional interconnection). This methodology does not encompass
all ISPs, so its results are not exhaustive. This limitation is insurmountable
because the underlying data are partial at their origin and do not include
all interconnections or all ISPs.
estimate of the distance between the two ends of the communication. This tool is called
traceroute on UNIX, Mac and GNU/Linux machines; in Windows it is called tracert.
7
The DFZ consists of all routers that have a global routing table and do not require a default
route to send a packet to any destination.
8
According to RFC 1930, an Internet autonomous system (AS) is a set of networks operated
by one or more network operators which has a single and clearly defined routing policy.
It has autonomy with respect to external routing. Every AS has a unique number assigned
to it under RFC 4893, called an Autonomous System Number (ASN).
ECLAC 86
The application of this methodology, presented by Patara (2010),
is intended only to determine the situation in countries that do not have
IXPs based on information as of 10 May 2010. The interconnection results
are presented in the form of a map for each country showing all ASs and
the interconnections between them. This methodology leads to the same
conclusion as the frst one, though by a different route: many countries
that do not have IXPs do not have complete domestic interconnections.
IXPs that adhere to world best practices, which will be seen later, are a
must for improving Internet quality and reducing prices (by using fewer
international links).
3. Variables affecting costs
The main factors infuencing end-user prices for Internet access are
discussed in this section. Some of them, such as IXPs, are analysed in
more detail in other sections and in annex III.2.
(a) Importance of traffic exchange points
This section describes how IXPs reduce Internet access costs. The
impact can be seen in Chile, which has 15 years of experience in national
interconnection. The high rate of connectivity in Chile makes it possible
to differentiate wholesale prices between national and international traffc
(with domestic traffc being several times cheaper than that with other
countries). The most important aspects to consider with regard to their
cost-reduction impact are:
(i) The existence of an IXP in a country can lower the costs of
connecting ISPs with each other, because it is enough for each
one to be connected to the IXP for all of them to have access to
each other’s networks through commercial agreements.
(ii) The absence of an IXP means that ISPs may have to set up a grid
structure or even connect through other countries’ IXPs.
(iii) The IXP is an effcient mechanism for routing domestic traffc,
because ISPs usually pay the IXP and interconnection costs;
when network differences keep them from peering they enter
into transit agreements.
(iv) Aggregating IXP traffc leads to lower prices for long-distance
domestic and international IP transit through ISPs, as can be
seen in Argentina and Brazil.
(v) IXPs are a natural place to host content, thereby cutting costs to
providers and improving the quality of service to the end user,
with a signifcant reduction in delays or transit times.
Broadband in Latin America: Beyond Connectivity 87
(vi) CDNs seek neutral IXPs whose management policies cannot be
manipulated to beneft any particular national or international
ISP or carrier. This is a principle that has been in place since the
early development of IXPs throughout the world.
(vii) CDNs enter into peering agreements for installation on the
following economic basis: (i) the CDN brings content to the ISP
so that the latter can operate as a “platform” in a bilateral market
between the end user and the CDN; (ii) the CDN provides value to
the ISP and thus to the end user by improving the quality of service;
(iii) the ISP provides value to the CDN by giving it direct access to
the end user; and (iv) the platform’s pricing mechanism avoids the
use of international channels, cutting costs for both sides.
(viii) The existence of multiple IXPs in a country or region facilitates
agreements between carriers so that they can spread the load
more evenly across networks and achieve route redundancy.
Web caching
Web caching, a technology that has been in use for years, is the
local storage of content that is located on distant servers and is consulted
frequently by users. This reduces the usage of international links, since the
content is accessed locally. For example, when a news site is consulted, the
system can locally download the standard images on the page that have
been stored previously so that it only needs to download from the site
the headlines and other text content that is constantly updated. In other
cases, such as YouTube, the system downloads videos that are frequently
consulted. The savings in international link usage varies, ranging from
20% to 30% depending on factors such as user behaviour, the size of the
web caching system and the quality of the technology employed.
National and regional content hosting
The countries of the region generate abundant content that is
consulted primarily by domestic users. The majority of this content tends
to be hosted at data centres located mainly in the United States, and, to a
lesser extent, in Europe. It has already been noted that a signifcant portion
of regional media sites are hosted in the United States, and it is believed
that the same is true of other content and application providers. The
reason for this tendency is that companies hosting content in advanced
countries offer lower costs and greater transmission capacity. Although it
is a widespread trend, it is useful to distinguish among the different types
of content and application providers.
(i) Major web sites of service provider companies. These include the web
sites of large production, commercial or industrial businesses,
ECLAC 88
news media, airline customer service centres, online sales sites,
e-health management centres and others. For such companies,
the quality of the hosting service and proximity to their centre
of operations are more important than price. These sites could
be hosted locally under the right conditions, mainly in terms of
quality, or they could be hosted by CDNs.
(ii) Major content and application distribution networks (CDN). These
include providers such as Google-YouTube, Akamai, Microsoft,
Limelight Networks, Level3 and others. They have developed
CDNs with nodes located in countries that offer incentives for
that purpose. Bearing in mind that Google-YouTube accounts
for about 10% of total Internet traffc and Akamai for 20%, it is
obviously important to establish favourable conditions for local
hosting of their content. The existence of an IXP is an essential
part of these companies’ decision-making process.
(iii) Small content and application providers. Since their main objective
tends to be cost reduction, aside from good quality, it is very
diffcult for them to compete with the major hosting providers
in the northern hemisphere that have economies of scale and
low costs for their main business inputs.
4. Global trends and prices
In August 2011, TeleGeography published information on changes in
pricing for transoceanic links and emphasized that competitiveness on the
North Pacifc route, where there are at least 14 cables operating and one
under construction, is similar to that of the North Atlantic route. In two
years (between the second quarter of 2009 and the second quarter of 2011),
the median monthly lease for a 10 Gbps lambda cable between Los Angeles
and Tokyo fell from US$ 98,500 to US$ 36,000 (see fgure III.1). This trend
is attributed to the deployment of three new cables in 2008-2009 and 2010,
which increased competition and available capacity. The monthly lease of
US$ 36,000 per 10 Gbps is equivalent to US$ 3.60 per Mbps, several times
lower than what is paid in South America.
Another approach is to look at the trends in the median price of
GigE ports (1 Gbps) to the last four months of 2011. In four years, the price
of 1-Gbps ports in New York and Hong Kong SAR fell at a compound
annual rate of 17%, while in London and selected cities in Latin America
(Buenos Aires, Mexico City, Rio de Janeiro, Santiago and São Paulo) they
fell approximately 20%. The relative prices of the two regions are likely to
stay on the same trajectory, unless there are signifcant changes in demand
Broadband in Latin America: Beyond Connectivity 89
for volume and available capacity in Latin America. It is believed that
prices will continue to fall in the region as new cables are installed and
demand increases.
Figure III.1
MEDIAN MONTHLY PRICES ON NORTH PACIFIC ROUTES FOR 10 GBPS,
SECOND QUARTER 2010 TO SECOND QUARTER 2011
0
20 000
40 000
60 000
80 000
100 000
120 000
Q2 2010 Q2 2011
Los Angeles - Tokyo Los Angeles - Singapore Hong Kong SAR - Los Angeles
Source: TeleGeography, Global Internet Geography Research Service.
Figure III.2 shows trends in prices in some of the major cities of
Latin America since 2008. They are converging towards a rate of US$ 40
per Mbps per month and so could stand at some US$ 30 per Mbps by the
end of 2012.
Figure III.2
TRENDS IN MEDIAN PRICES IN MAJOR CITIES OF LATIN AMERICA
0
10
20
30
40
50
60
70
80
90
2008 Q2 2008 Q4 2009 Q2 2010 Q1 2010 Q4 2011 Q2
Buenos Aires Rio de Janeiro Santiago São Paulo Mexico
Source: TeleGeography, Global Internet Geography Research Service.
ECLAC 90
Figure III.3 provides an intercontinental comparison of pricing
trends. Prices in São Paulo have fallen at the same rate as in the Northern
Hemisphere, but the ratio of 6:1 with Europe (London) and the United
States (New York) is not shrinking, nor is the 2:1 ratio with Asia (Hong
Kong SAR).
Supplementary information shows that, in keeping with the
worldwide trend, operators in the region are expanding their networks to
reach areas far away from their primary markets, so prices are expected to
fall in those areas as well.
Figure III.3
TRENDS IN MEDIAN PRICES IN MAJOR WORLD CITIES
0
10
20
30
40
50
60
70
80
90
100
2008 Q2 2008 Q4 2009 Q2 2010 Q1 2010 Q4 2011 Q2
Hong Kong SAR London New York São Paulo
Source: TeleGeography, Global Internet Geography Research Service.
D. Conclusions
The following conclusions can be drawn from the analysis presented in
this chapter and its annexes.
(i) Although progress has been made in fbre illumination in Latin
America, Internet usage is growing at a rate that is likely to lead
to saturation in the near future.
(ii) There are a number of initiatives to build international links
that could generate more competition and expand capacity in
the future.
(iii) The United States continues to predominate as an Internet
interconnection centre for Latin America, representing 85.5% of
the capacity of all of the region’s international links. Intraregional
links account for 14.3% of the total.
Broadband in Latin America: Beyond Connectivity 91
(iv) There are several different kinds of access agreements at the
wholesale level.
(v) Traffc among countries in the region travels predominantly
along routes passing through the United States.
(vi) The existence of internet exchange points or IXPs in a country
yields many advantages, including lower costs and better
quality, the possibility of moving towards lower prices for
international IP transit, incentives for local content hosting and
the installation of CDNs.
(vii) It is important to prioritize the search for greater local content
hosting capacity in two stages: (i) seek the installation of
large content and application providers as well as CDNs, and
(ii) create a favourable environment to encourage new small
content providers to begin hosting locally.
(viii) The median monthly cost per Mbps as of the end of 2011 was
US$ 31 for 10-Gbps ports and US$ 41 for 1-Gbps ports, but these
fgures are for prices at the cable landing points and do not take
local lines into account. As of year-end 2012, they may be 20%
lower. These prices are about six times higher than in Europe
and the United States, and twice the levels found in Asia.
(ix) Over the past three years, prices in the Northern Hemisphere
and in Latin America have been falling at a similar compound
annual average rate of 20%. The relative prices of the two
regions are likely to stay on the same trajectory, unless there
are signifcant changes in demand for volume and available
capacity in Latin America.
(x) Costs may come down if regional traffc can avoid more costly
routes by going through remote exchange points (Miami, for
example) rather than direct routes between IXPs in the different
countries. This situation may be improved by facilitating
regional connectivity for all operators while promoting the
installation of IXPs in all countries.
ECLAC 92
Bibliography
Patara, Ricardo (2010), NICbr. Estudio de la relación entre el AS y IXP en la región
(LAC). [Study of the relationship between AS and IXP in the region (Latin America
and the Caribbean)] Network Access Points Latin America (NAPLA).
TeleGeography (2011), Global Internet Geography Research Service.
Broadband in Latin America: Beyond Connectivity 93
Annexes
Annex III.1
Broadband definitions
ITU Recommendation I 113
This recommendation dates back to June 1997, so it refects a
completely different time and refers primarily to Integrated Services
Digital Network (ISDN) technology. It states that “broadband qualifes a
service or system requiring transmission channels capable of supporting
rates greater than the primary rate of ISDN”. The primary rate was equal
to 1.54 Mbps (1 T1) in the United States standard and 2,048 Mbps (1 E1) in
the European standard.
The ITU report of 2004
In the report titled ITU and its Activities Related to Internet Protocol (IP)
Networks published by ITU in April 2004, broadband is defned as follows:
“Although there exist various defnitions of broadband that have assigned
a minimum data rate to the term, it may be defned as transmission
capacity with suffcient bandwidth to permit combined provision of voice,
data and video, with no lower limit. ...” Broadband is defned in terms of
the services that can be provided through it, especially the combination of
services that can require speeds of 2 Mbps or higher.
Later on, in section 7.2, “What is broadband?”, the report states:
“While the term broadband is used to describe many different Internet
connection speeds, Recommendation I.113 of the ITU Standardization
Sector (ITU-T) defnes broadband as a transmission capacity that is faster
than primary rate ISDN, at 1.5 or 2.0 Mbit/s. However, this defnition is
not strictly followed. The OECD considers broadband to correspond to
transmission speeds equal to or greater than 256 kbit/s”.
“Defnitions of World Telecommunication/ICT Indicators”
Telecommunications indicators are identifed in the document
Defnitions of World Telecommunication/ICT Indicators, published by ITU in
March of 2010. Although the term broadband is not expressly defned, its
meaning can be inferred from the indicators that are defned.
Fixed broadband. It is established that the measurement of this
indicator should include connections at speeds greater than 256 Kbps. ITU
has also decided to survey indicators for fxed broadband access at speeds
of 256 Kbps to 2 Mbps, 2-10 Mbps, 10-100 Mbps, 100-1000 Mbps and more
than 1000 Mbps or 1 Gbps. The survey of subscriptions at speeds higher
than 256 Kbps has been maintained to prevent an interruption of historical
series. Thus, allowing for higher-speed indicators means that series can be
ECLAC 94
established for the present date while maintaining historical series dating
back many years.
Mobile broadband. This indicator encompasses standard mobile
subscriptions using data communications at broadband speeds, including
subscriptions with advertised speeds of 256 Kbps or higher.
“Measuring the Information Society”
The document Measuring the Information Society, published by ITU in
2011, states in box 4.1, “Defning broadband – a moving target”: “Although
many national regulators and international organizations, including ITU
and OECD, defne broadband as a connection with downstream speeds
greater than, or equal to, 256 kbit/s, there remains much debate as to
how fast a connection should be to qualify as broadband, and national
defnitions vary. The United States, for example, just recently redefned
broadband as a ‘transmission service that [currently] enables an end
user to download content from the Internet at 4 Mbps and to upload
such content at 1 Mbps over the broadband provider’s network’. Some
developing countries, for example Djibouti and Morocco, apply defnitions
of broadband that set a lower speed (128 kbit/s). However, most national
broadband defnitions are in line with the ITU/OECD defnition.
ITU News. The broadband debate: The need for speed?
The September 2011 edition of this ITU publication
9
refers to the
need to defne broadband speed. Recognizing the importance of speed, it
asks what speed is fast enough and what other factors need to be taken
into consideration to establish speed and coverage targets. In particular,
it notes that basic connectivity still makes a real difference in people’s
lives, their work and other aspects of economic and social life, without
overlooking the importance of high-speed connections.
“Broadband: A platform for progress”
10
The document Broadband: A platform for progress. A report by the
Broadband Commission for Digital Development, published by ITU and
UNESCO in June of 2011, notes in its section 2.1, “Possible defnitions”, that
several defnitions have been debated by the Commission. Three options
have emerged in this context: (i) broadband could be defned by quantitative
indicators as a function of speed and technology; (ii) broadband could be
defned by qualitative indicators, in terms of the applications it enables or
the impact it may have on economic and social factors; or (iii) a combination
of both can be used. Although it analyses the options, the document does
not reach a conclusion on a defnition for broadband.
9
See [online] http://www.itu.int/net/itunews/issues/2011/07/11.aspx.
10
See [online] www.broadbandcommission.org.
Broadband in Latin America: Beyond Connectivity 95
A proposed defnition of broadband for the countries
of the region
This proposal takes into account the documents studied and the
objective of arriving at a practical defnition that can be quantifable,
taking into account not only user needs in terms of the applications they
require, but also possible limitations specifc to each country with respect
to economies of scale, international costs and geography, among other
considerations. This defnition was developed within the framework of the
Regional Broadband Dialogue, for which ECLAC is the technical secretariat.
The objective is to provide a broadband development indicator
not intended for regulatory purposes. In addition, the goal is to translate
qualitative objectives into clear and simple quantitative indicators that
make it possible to plan, evaluate and follow up on broadband trends so
that fndings can be compared.
An initial condition for classing Internet access as broadband is
that the connection must be permanent, meaning that it is not necessary
to establish a connection every time one decides to exchange information
(“always on”). There is evidence that when using mobile broadband,
customers value “always on” more than speed.
To defne broadband, it was considered important to establish
three levels of service depending on customer requirements, taking into
account each country’s social and economic development and issues such
as e-health, e-learning and e-government.
Simple activities. These are the simplest activities carried out on the
Internet that nonetheless substantially improve quality of life as a means
of social integration. They require moderate speeds and are in line with
the ITU defnition. Simple activities consist primarily of e-mail, reading
or downloading documents and general web surfng, written online
communication (chatting) and access to government applications with low
data content.
Advanced activities. These activities require broadband to take
advantage of the main services currently offered on the Internet: social
networking, standard defnition video streaming with a certain level of
compression (for example, approximately 700 MB for a 1½ hour flm), video
conferencing, fle sharing, instant messaging, webmail and navigation at
speeds that allow for telecommuting or advanced e-health with image or
video transfer, among others.
All high-capacity activities. These include video streaming and high-
defnition (HD) IPTV, telepresence, sharing large fles, telework with easy
sharing of large volumes of data, advanced e-health with the transfer of
ECLAC 96
large volumes of image and video data, e-learning including HD videos,
among others, as well as trouble-free conduct of current activities.
Based on these demands, three levels of broadband service should
be included in the defnition: (i) basic broadband, which is provided through
wired or wireless, fxed or mobile services and enables simple activities
on the Internet; (ii) advanced broadband, which makes it possible to engage
in all of the advanced activities currently available on the Internet; and
(iii) total broadband, which enables users to comfortably engage in their
current activities and be in a position to enjoy all high-capacity activities.
In these defnitions, the main parameters are download and upload
speeds. Other parameters such as latency and jitter
11
also affect the quality
of user experience, but it is not appropriate to include them at this juncture
for two reasons. First, they do not have a great impact on quality in the
region; and second, it is diffcult to survey this information in the region.
The speed currently defned by ITU (256 Kbps, which is primarily
an international benchmark) is considered adequate for simple Internet
activities and could be set as the entry or basic level for broadband. Based
on experience, it is believed that an effective download speed of 2 Mbps
is adequate for advanced activities such as standard defnition (SD) video
streaming with a certain level of compression. This speed is chosen
because it is also the lower limit of the second level of access speed defned
by ITU in its surveys. For the third level of service, 10 Mbps is considered
suitable, as it is the next level defned by ITU, and it is also close to the
11.25 Mbps that Cisco regards as the minimum for supporting future
applications (see table III.A.1).
Table III.A.1
DEFINITION OF BROADBAND
Indicator Basic broadband
Advanced
broadband
Total broadband
Download speed 256 Kbps.
>2 Mbps. – 2048
Kbps.
>10 Mbps.
Upload speed 128 Kbps. >512 Kbps. >768 Kbps.
Availability of use Always on Always on Always on
Source: Prepared by the author.
As for upload speeds, the basic broadband service level is set at
128 Kbps, 512 Kbps for advanced broadband, and 768 Kbps for total
broadband. Because 768 Kbps does not allow for activities requiring high
11
Jitter is the term used for digital signal transit time variance. It tends to be regarded as an
undesirable signal noise.
Broadband in Latin America: Beyond Connectivity 97
upload speeds, such as SD videoconferencing, it may be considered a
little too low. However, it is well suited to the situation in Latin America,
where download speeds of 6 Mbps to 10 Mbps and upload speeds of about
800 Kbps are typically offered. In conclusion, 768 Kbps was adopted so as
not to leave out operators in the region offering more than 10 Mbps but
with upload speeds that are still low compared to what is required for
more advanced applications.
For wired connections, published speeds are used because it is
diffcult to obtain information on real speeds differentiated by bands. In
addition, since service quality control is a function that may or may not
be exercised by the regulatory authority, the survey-based defnition does
not take into account the difference between actual and published speeds.
For wireless connections, due to technical issues related to the manner in
which the resource is shared, the peak speed provided by the base station
is considered. In this case, it is impossible to defne a quality margin
because of the stochastic nature of the data fow, which depends on the
number of customers served simultaneously by the same base station and
the type of use each customer makes of the connection.
Annex III.2
Internet exchange points
Description of an IXP
Every Internet user has access to a certain network controlled by
an independent entity, usually called an Internet service provider (ISP).
Telecommunications operators in general can also be called ISPs if they
provide Internet access. The networks controlled by these ISPs are called
autonomous systems (AS).
12
An IXP is a single physical network structure to which many ISPs
connect. Any ISP connected to an IXP through a single interconnection
point may exchange traffc with any other ISP connected to the same IXP,
thereby solving the scalability problems that would arise if all ISPs had to
connect with all others on a grid. Agreements take many different forms
throughout the world. The possibility of exchanging traffc among ISPs
without going through remote points is important for the development of
regional markets, as was shown above.
12
As seen earlier, for an ISP to provide global Internet access to its customers, it must
somehow connect its AS to at least one other ISP with global Internet connectivity. The
interconnection agreement tends to be one of two types: (i) a contract for IP or Internet
transit whereby one ISP pays the other in accordance with the relative interest each has
in establishing a connection through peering, or (ii) a peer agreement whereby both ISPs
exchange traffic without any payment or other consideration.
ECLAC 98
Snapshot of IXPs in the region
The analysis begins with the list of IXPs published by the Packet
Clearing House (PCH),
13
a non-proft institution that supports operations
and analyses Internet traffc exchange areas, the economics of routing
and the development of the World Wide Web. Of the IXPs included on
the list, the most comprehensive one available, those that are no longer
interconnection points or are not currently operating were eliminated. The
others were surveyed in detail to gather data. The Member States of the
Regional Broadband Dialogue participated in this process.
14
Information surveyed in each country
The following information was gathered for analysis and
recommendations about IXPs in the region:
(i) Year of establishment and location (address, telephone number,
contact person).
(ii) Type of entity (non-proft, commercial, state-owned, among others).
(iii) List of ISPs currently connected to the IXP.
(iv) IXP traffc policies.
(v) Limitations on routes, whether the traffc exchanged is local
(one city or nearby cities), national or international.
(vi) Types of bilateral or multilateral agreements among ISPs in a
single IXP.
(vii) Payment structure (IXP entry fee, monthly or traffc-based
fee, whether a distinction is made between local, national and
international traffc, and others) and amounts.
(viii) Terms, if any, under which the ISPs in an IXP can exchange
traffc with other IXPs in the same city (peering, etc.) or different
cities (peering, transit fees, etc.).
(ix) Whether there are traffc exchanges with other IXPs in the
country, or whether the entity’s charter allows for them.
(x) Whether the IXP is at a single physical location or at various
interconnected sites.
(xi) Requirements ISPs must fulfll to participate in the IXP
and general terms for handling traffc (non-discrimination,
neutrality, etc.).
13
See [online] http://www.pch.net/home/index.php.
14
Argentina, Brazil, Chile, Colombia, Costa Rica, Ecuador, Paraguay, Peru, Plurinational
State of Bolivia and Uruguay.
Broadband in Latin America: Beyond Connectivity 99
(xii) Policies on ISPs’ physical links with the IXP: minimum capacity,
redundancy, etc.
(xiii) Whether interconnections exist or are planned with other
regional IXPs, particularly those in neighboring countries, and
the characteristics of these possible interconnections.
(xiv) Improvements in prices or quality for ISPs belonging to the IXP
in national or international IP transit agreements.
(xv) Total traffc exchanged at peak times (HP).
(xvi) Conditions required for large CDNs—Akamai, Google,
Limelight Networks, Microsoft and others—for storing content
for the national or subregional market, with respect to Internet
connections in the country, costs, special peering or similar
agreements, etc.
(xvii) Other aspects of interest to the IXP for developing its activities
appropriately.
Results of the analysis
The information was compiled primarily from the IXPs’ websites,
with support from the authorities of the Member States in the Broadband
Regional Dialogue. The countries’ situations are described below.
• Argentina. There is a single major IXP, CABASE, for which
relevant data are available. It is largely compliant with
international best practices.
• Brazil. Eight of the IXPs on the PCH list do not operate as such.
The IXPs PTT Metro and NAP do Brasil were analysed, so the
most relevant information is available for them. PTT Metro
upholds best practices.
• Chile. The seven IXPs identifed by the authorities have been
analysed. Since 1999 interconnection has been mandatory,
resulting in a market with good connectivity. Therefore, there
is no need for regulation, since the ISPs have internalized the
interconnection requirement.
• Colombia. There is little public information on IXPs.
• Costa Rica. There is no IXP, but the Interconnection Act and its
regulations provide a framework for its implementation.
• Ecuador. All relevant information has been compiled for
AEPROVI, which does adhere to best practices.
• Paraguay. There is no IXP.
ECLAC 100
• Peru. There are two main IXPs, but they do not comply entirely
with best practices.
• Plurinational State of Bolivia. There are no IXPs, although the
country’s Telecommunications Act does regulate this area.
• Uruguay. There is no IXP.
Comparative analysis of regulations
There is international evidence of the importance of installing
IXPs, and precedents have been established for setting them up and
operating them. There is also information on regulations, and documents
have been published by reputable institutions in the feld, such as the
Internet Society
15
and the ICT Regulation Toolkit,
16
emphasizing the
importance of regulation.
The following analysis reviews the concepts addressed in the
ICT Regulation Toolkit and the regulations of the three Latin American
countries (Chile, Costa Rica and the Plurinational State of Bolivia) that
have implemented regulations to encourage the development of IXPs.
ICT Regulation Toolkit
After briefy discussing the advantages of establishing IXPs, this
document explains the challenges in installing regional IXPs. It confrms
what has already been observed: that developing countries are behind in
the implementation of IXPs mainly because of regulatory, coordination
and management issues. ISPs within a single country must be convinced
that IXPs will not tip the balance towards one or more operators, and
neighboring countries’ ISPs need to understand the value of routing
traffc to the IXP rather than developing private interconnections among
themselves. The document concludes that “...it is important to have offcial
legal and regulatory support for the development of IXPs...”
Plurinational State of Bolivia
Law 164, the General Telecommunications, Information Technologies
and Communication Act, was published on 8 August 2011. Article 50
thereof (Interconnection among Internet providers) stipulates: “Internet
providers shall establish and accept interconnections among themselves
within the national territory through an Internet exchange point in order
to manage Internet traffc in accordance with the conditions established by
the regulations”.
15
See “An Introduction to Internet Interconnection Concepts and Actors”, ISOC, www.
Internetsociety.org.
16
See [online] http://www.ictregulationtoolkit.org/en/Section.2194.html.
Broadband in Latin America: Beyond Connectivity 101
This law makes it mandatory to establish and accept
interconnection among Internet service providers as a general condition.
The primary issue of Internet exchange points is established in this
manner, with details about procedures and other conditions being subject
to regulations. It represents a major step forward in a country with high
international costs for Internet access. It is expected that this regulation
will bring down costs.
Chile
The Republic of Chile, in Exempt Resolution No. 1483 of October
1999, set forth the procedure and deadlines for establishing and accepting
connections among ISPs. The concepts are addressed in the following
main articles, which are supplemented by others related to procedures
and quality control: “Article 2. In order to guarantee proper operation
and non-discrimination in the quality of Internet access services provided
to users, ISPs shall, prior to the commencement of service, establish and
accept connections among themselves to manage national Internet traffc,
without prejudice to the provisions of article 5 of this law”.
“Article 3. ... The foregoing provisions notwithstanding, the
connections established among ISPs shall ensure that the customers of the
requesting ISP are given access to content providers located both in the
requesting ISP and in the ISP receiving the request, of equivalent quality.
Similarly, the connections established among ISPs shall assure content
providers located in the requesting ISP that customers of the requesting ISP
as well as those of the ISP receiving the request have access of equivalent
quality to such providers. …”.
“Article 5. In any case, ISPs may establish other connection
topologies besides that which is specifed in the frst paragraph of
article 2 above, provided that they ensure that national Internet traffc
is exchanged through transmission channels authorized to manage
national communication. If a national Internet exchange point is
established to combine traffc from one or more ISPs, the provider of
such service shall be considered an ISP for purposes of this law”. This
article, by making interconnection through IXPs mandatory, ensures that
all national ISPs are interconnected with each other under the conditions
set forth in this resolution.
“Article 6. ISPs shall accept and put in service the aforementioned
connections under non-discriminatory conditions. Furthermore, each
ISP shall allow the customers of the ISPs that are connected pursuant
to this law to have access to all of the content it maintains, under non-
discriminatory conditions”.
ECLAC 102
As noted above, the Chilean experience is interesting in that it shows
the results of an IXP development policy that has been in place for more
than 15 years. IXPs began to emerge organically in 1995 and gained strength
after 1999 on the basis of specifc regulations. Today, there is a favourable
national connectivity environment in which IXPs, though they exist, have
a rather secondary role in the important interconnection networks that
link all ISPs. Thus, the regulation of IXPs could be unnecessary because
the market, through its own dynamic, has already established signifcant
connectivity that has even brought national wholesale IP transit prices
down to about 10% of the international transit price.
Costa Rica
The Interconnection Act and its regulations have created a
transparent environment in which mandatory interconnection does not
distinguish among technologies. In particular, there is a precise procedure
for paying for converged networks or packet-switched networks, offering
a variety of alternatives depending on, among other factors, traffc,
bandwidth and committed capacity. In any case, operators are encouraged
to negotiate with each other before resorting to the regulatory body, the
Superintendency of Telecommunications (SUTEL).
Best operating practices and key performance indicators
Current best practices according to Euro-IX
Although IXPs are sites where peering among ISPs is optimized,
they are not generally involved in the peering agreements that ISPs enter
into, nor do they dictate which ISPs each one must peer with or the terms
of their agreements. However, because they are shared infrastructures,
IXPs have requirements that ISPs must meet in order to use them properly.
There is a set of best practices based on the requirements and
characteristics of the IXPs associated with the European Internet Exchange
Association (Euro–IX).
17
These practices, widely recognized in the world
of IXPs, cover implementation and maintenance as well as rules that
ISPs are expected to follow. They are not strict rules, just a listing of the
best practices derived from the experience of the IXPs that already exist.
Based on an analysis of the importance of each of these specifcations, a
minimum number of key performance indicators (KPI) can be defned to
evaluate the quality of IXPs and that of interconnection among countries.
The rationale for the indicators lies in the benefts of effcient national
interconnection and access to international and national content. For a
more thorough analysis, KPIs are divided into two categories: those used
for analysing IXPs and those applied to the analysis of interconnection at
the country level. For each group, the benefts derived from meeting the
17
See [online] https://www.euro-ix.net/ixp-bcp.
Broadband in Latin America: Beyond Connectivity 103
KPIs are listed, which in and of itself is enough to ensure compliance. All
of these indicators are met by the most advanced IXPs. The more effcient
the interconnection, the lower the costs and the higher the quality of
Internet and content access.
The analysis does not include technical indicators relative to
equipment, routing, technologies or other requirements for establishing
an IXP node. That is a very well-known subject on which there are
internationally accepted standards and with which there is ample
experience for choosing the appropriate physical structure. In this section,
the emphasis is on aspects of IXP management per se and of commercial
relationships, which are the most lacking and which have the greatest
impact on Internet prices and quality.
KPIs for IXPs
The main operating indicators for IXPs are discussed in this
section. They are grouped into classifcations that can be used to evaluate
their impact on improved effciency in connecting national Internet traffc
(see table III.A-2).
• Allows multilateral agreements; this is a minimum requirement
for the IXP to exist and to attract the interest of ISPs.
• Allows bilateral agreements among participating ISPs. The IXP
becomes a node for ISPs to interconnect freely, depending on
each one’s assessment of its own interests. It creates more value
for ISPs, attracts their participation and improves effciency. A
bilateral agreement allows for greater fexibility, since ISPs with
a lot of traffc or with common interests can engage in high-
capacity bilateral peering in addition to the lower-capacity
multilateral arrangement. These agreements, together with
bilateral transit agreements, are the principal infrastructure for
ensuring effcient interconnection.
• Allows bilateral agreements for national transit, provided by the
IXP itself or by third parties interconnected through the IXP. The
benefts are similar to those of the previous case, in that traffc
can be aggregated and better prices and quality can be achieved
for long-distance national transit. This is an important indicator
in vast countries whose cities are Internet usage centres.
• Allows bilateral agreements for international transit, provided
by the IXP itself or by third parties interconnected through the
IXP. The rationale is similar to that of the previous point, but for
international transit.
• Is operated neutrally. It does not belong to and is not associated with
any carrier, ISP or connectivity provider. Thus, there are no conficts
of interest that would impinge on effcient operation of the IXP.
ECLAC 104
• There are no discriminatory restrictions on participation by ISPs.
This indicator ensures that any ISP wishing to exchange traffc
through the IXP may do so as long as it meets the standard
technical and commercial conditions that are applicable to all
ISPs, without discrimination.
• No fltering or discriminatory policies are applied to
participating ISPs or to the content made available by each of
them. The requesting ISP’s customers have access to equivalent-
quality content providers located both in the requesting ISP and
in the ISP receiving the request. It guarantees the quality of the
interconnection through the IXP.
• IXP usage prices are cost-oriented, or, more specifcally, they are
based on the distribution of the IXP’s investment and operating
costs, depending on how it is owned and managed. In any case,
this cost orientation makes it possible to improve Internet access
prices, because interconnection through the IXP has a lower
price, refecting opportunity costs. The value added by the IXP
is passed on entirely to the ISPs’ customers.
• Any ISP wishing to connect to an IXP may contract freely
for that access under its own responsibility. This condition
ensures that there is no exclusivity in access to the IXP, thereby
improving competitiveness. The ISP can achieve the best prices
for connecting to the IXP by building or leasing as it sees ft.
• There are no distance limitations for ISPs wishing to interconnect
through an IXP, as long as they pay for the link. In some cases,
these limitations have been established without any economic
justifcation.
• The IXP provides co-location services. This encourages ISPs
to interconnect by helping them optimize resources and
reduce costs.
• The IXP is located in places where carriers and ISPs have
points of presence nearby. This guarantees the quality of the
interconnection at the IXP, since it lowers access costs and creates
incentives for expansion. In deciding whether to use an IXP or
engage in private peering, paid or not, ISPs are heavily infuenced
by the cost of access to the IXP (physical access dependent on
distance and membership costs). Ideally, an IXP that has carrier
points of presence at its location can facilitate access and may
also make it possible to buy transit if bilateral agreements are
allowed. The State may adopt policies that provide incentives
for locating in appropriate places. For example, it may sponsor
meetings where the importance of proper location is stressed,
Broadband in Latin America: Beyond Connectivity 105
or it may facilitate the installation of IXPs near those points of
presence. This is an indicator that should be taken into account,
among others, when designing policies and actions.
• The IXP has equipment and ISP connection redundancy. This
guarantees the availability of connections at the IXP and prevents an
equipment failure at the IXP from totally disrupting interconnection.
This is a quality indicator in the operation of the IXP.
• The IXP offers robust security services, such as a continuous
power supply, access controls and fre and food protection
equipment. This is how the quality of the site can be determined
and the availability of national connections to the Internet can
be assessed in the country.
• The IXP has multiple sites. This guarantees the quality and
availability of the interconnection and also improves costs
for ISPs, which can select the nearest node, or, if they want to
connect to more than one node, they can select the best quality.
• The IXP makes available, at the ISPs’ request, information on
IXP and member traffc. This information is valuable for ISPs
when conducting cost-beneft analyses of IXP usage. The more
members there are and the more traffc is exchanged, the greater
the value to the ISP of linking to an IXP, and the more effcient
its decisions.
• The IXP has CDN nodes.
Table III.A.2
KEY IXP PERFORMANCE INDICATORS
Indicator
Interconnection functions
Allows multilateral agreements
Allows bilateral agreements among participating ISPs
Allows bilateral agreements for national transit
Allows bilateral agreements for international transit
Neutrality and non-discrimination
Is operated neutrally
There are no discriminatory restrictions on participation by ISPs
Filtering and discrimination policies are not applied
Costs
IXP usage prices are cost-oriented, or, more specifcally, they are based on the distribution
of the IXP’s investment and operating costs
Facilitation of IXP access
ISPs can contract IXP access freely
There are no limitations on IXP distance from ISPs
ECLAC 106
Table III.A.2 (concluded)
Indicator
The IXP provides co-location services
The IXP is located in places where carriers and ISPs have points of presence nearby
Redundancy and infrastructure
The IXP has equipment and ISP connection redundancy
The IXP offers robust security services
The IXP has multiple sites
Other indicators
Publishes information on IXP and member traffc
Has CDN nodes
Source: Prepared by the author.
KPIs for countries
As with the KPIs for IXPs, the main indicators for connecting
national Internet traffc originating and terminating in a country are
detailed below (see table III.A.3).
• All ISPs are interconnected so that all national communication
traffc originating and terminating in the country can be
sent to or from any ISP, either through one or more IXPs or
through private peering. This is the main indicator of Internet
connection quality, and it ensures better quality and pricing
for customers. From a country’s standpoint, it is the most
important indicator, even if it does not guarantee the effciency
of all interconnections among IXPs. Indeed, the existence
of strong IXPs meeting all of the KPIs listed above does not
guarantee compliance with this indicator.
• There are regulations providing for mandatory interconnection.
This is an informative indicator that is relevant when the
previous condition is not met.
• There is competition among IXPs according to the criteria set
forth in the country’s competitiveness regulations. Competition
is a strong incentive for linking IXP usage prices to opportunity
costs, analogous to the typical situation of non-proft IXPs that
charge in accordance with real usage costs.
• Presence of major CDNs at IXPs or in the country, with national
interconnection for all ISPs. This is an indication of how
important it is for a country to have an IXP system to improve
Internet access quality and pricing for the members of IXPs.
Hosting major providers and ensuring that the customers of
Broadband in Latin America: Beyond Connectivity 107
all ISPs have access to them yields savings of 20% to 30% on
international transit, taking into account the volume of content
on providers such as Akamai, Google-YouTube, Microsoft,
Lime Light Networks and others. As a reference, this indicator
is deemed to have been met when at least two of the CDNs
mentioned above are present.
• All IXPs allow bilateral agreements for regional and international
transit.
Table III.A.3
KEY COUNTRY PERFORMANCE INDICATORS
Indicator
All ISPs are interconnected for handling all national communication traffc originating and
terminating in the country.
There are regulations providing for mandatory interconnection. An important indicator in
cases of non-compliance with indicator 1.
There is competition among IXPs.
There are major CDNs at IXPs or in the country, with national interconnection for all ISPs.
All IXPs allow bilateral agreements for regional and international transit.
Source: Prepared by the author.
Second part
Economic impact
Chapter IV
Broadband, digitization and development
Raúl L. Katz
1
A. Broadband and economic growth
This chapter presents the fndings of a series of studies conducted
since 2009 on the contribution of broadband to development in Latin
America.
2
A frst set of results focuses on measuring the economic impact
of broadband on GDP growth, job creation and increases in average
household income. A second set evaluates the contribution of broadband
in terms of applications, services and content by means of a digitization
index that measures both technology adoption in a country and the use of
services associated with it —for example, e-government, e-commerce and
social networking. On the basis of these fndings, policy recommendations
aimed at maximizing broadband’s economic impact are proposed.
Broadband, as a general purpose technology, contributes to
economic growth through a number of effects (see diagram IV.1). First,
the deployment of telecommunications networks yields effects similar
1
Raúl L. Katz is an adjunct professor in the Finance and Economics Division at Columbia
Business School and Director of Business Strategy Research at Columbia Institute for Tele-
Information. He is also president of Telecom Advisory Services, LLC.
2
The studies were conducted within the framework of research programmes
undertaken for ECLAC, the International Telecommunication Union (ITU), the World
Economic Forum, the Governments of Colombia and Costa Rica, and associations of
telecommunications operators.
ECLAC 112
to those generated by any infrastructure project: broadband deployment
creates jobs and impacts the economy as a whole through multiplier
effects. Second, broadband use has a spillover effect on the entire
economic system, which is felt by both businesses and residential users.
On one hand, the use of broadband by the production sector results in
increased productivity, thereby helping to raise GDP. On the other, its
residential adoption increases real household income, reducing poverty
and contributing, in turn, to economic growth.
Diagram IV.1
ECONOMIC IMPACT OF BROADBAND
Broadband
deployment
Direct
benefits
Investment in
infrastructure
deployment
Residential
penetration
Consumer
surplus
Household
income
Enterprise
penetration
Total factor
productivity
Contribution
to GDP and
employment
growth
Impact of broadband
deployment on
employment
and production
Positive
externalities
Creation of
consumer
surplus
Source: Prepared by the author.
Beyond these effects, residential users who purchase broadband
beneft in terms of consumer surplus, which is the difference between what
consumers are willing to pay for the service and the market price. This
effect, though not included in the calculation of GDP, is important because
it represents an improvement in access to information, entertainment and
public services.
The following three sections describe the fndings of the author’s
studies in Latin America to date. The frst presents a model for measuring
the impact of broadband on GDP growth in the region as a whole, followed
by disaggregated results for Colombia and Panama. The second section
contains measurements of the impact of broadband on job creation in
Chile, Colombia and the Dominican Republic. The last section presents the
fndings of studies assessing the impact of broadband on the increase in
average household income in Costa Rica and Colombia.
Broadband in Latin America: Beyond Connectivity 113
1. GDP growth
The frst analysis of the economic impact of broadband in Latin America
was conducted by Katz (2011a) on the basis of a cross-sample of countries.
In the absence of time series, the analysis employed the ordinary least
squares method, using a sample of pooled data for the years 2004 and
2009. Two methodological problems arose in this analysis. First, as panel
data could not be used, it was not possible to isolate the characteristics
of each country in the model results, which could have created a
problem of “omitted variable”. This problem was mitigated, however,
by including variables such as broadband development and degree of
economic openness in the model. The second problem with this type of
model stems from endogeneity between growth of GDP per capita and
broadband penetration. Ideally, it would be possible to build a multiple-
structure model based on simultaneous equations in order to endogenize
the development of broadband as a function of per capita GDP, prices,
competition and degree of telecommunications industry regulation.
Again, lack of data did not allow the construction of such a model at the
time. The solution was to lag the variable “development of broadband” by
one year. With these caveats, the model generated the results shown in
table IV.1.
Table IV.1
LATIN AMERICA: CONTRIBUTION OF BROADBAND TO GDP GROWTH
GDP growth Coeffcient
Standard
error
t-statistic P>[t]
95% confdence
interval
Growth in broadband
penetration for 2001-
2003 and 2004-2006
0.0158715 0.0080104 1.98 0.054 -0.0002942 0.0320372
Average investment/
GDP for 2004-006
and 2007-2009
-0.0471624 0.1689699 -0.28 0.782 -0.3881575 0.2938328
Population growth
in 2004-2006 and
2007-2009
-0.4469177 1.40418 -0.32 0.752 -3.280668 2.386832
Higher education
(2002)
0.2139614 0.1108325 1.93 0.060 -0.0097076 0.4376304
Per capita GDP at
the start of 2003 and
2006
-0.0006957 0.0001806 -3.85 0.000 -0.0010602 -0.0003313
Average globalization
rate for 2001-2003
and 2004-2006
-0.0653024 0.1929498 -0.34 0.737 -0.4546908 0.324086
Constant 13.02883 12.04659 1.08 0.286 -11.28217 37.33982
ECLAC 114
Table IV.1 (concluded)
Number of observations 49
F(6,42) 7.18
Prob>F 0.0000
R
2
0.3814
Root MSE 7.024
Source: R. Katz, “The impact of broadband on economic growth”, Fast-tracking the digital revolution:
Broadband for Latin America and the Caribbean (LC/R.2167), V. Jordán, H Galperin and W Peres, Santiago,
Economic Commission for Latin America and the Caribbean (ECLAC), 2011.
The results show that, when controlling for education level and
initial per capita GDP level, a 1% increase in broadband penetration raises
GDP by 0.0158%. The broadband penetration coeffcient is positive in
sign and statistically signifcant. This fnding is consistent with the result
generated by Koutroumpis (2009) in a study of OECD countries, in which,
using a simultaneous equations model, showed that in countries where
broadband penetration averaged under 14%, a 1% increase in penetration
contributed 0.008% to GDP growth.
In 2011, increased availability of disaggregated data made it possible
to conduct studies at the national level. The frst one was carried out in
Colombia with department-level data for 2006-2010 (Katz and Callorda, 2011).
This study analyses the impact of fxed broadband on departmental GDP
growth, controlling for initial level of economic development, population
growth and human capital level (average years of education) (see table IV.2).
Table IV.2
COLOMBIA: CONTRIBUTION OF BROADBAND TO GDP GROWTH
Total Low penetration High penetration
Growth in broadband access
(percentage)
0.0036542***
(0.001282)
0.0039548***
(0.0014167)
0.0039453***
(0.0012952)
Population growth
(percentage)
0.8734808
(0.9599308)
-0.7848735
(1.019278)
4.585921**
(1.948842)
Years of education
-3.538593
(5.127222)
-1.878803
(11.28887)
3.668626
(3.831199)
GDP in 2003 (millions of
pesos)
0.0056116
(0.0284458)
-0.2697321
(0.3899207)
-0.0432453*
(0.0360005)
Adjusted R
2
0.1649 0.2088 0.2093
Prob > F 0.0103 0.0778 0.0086
Number of observations 132 64 68
Source: R. Katz and F. Callorda, Medición de Impacto del Plan Vive Digital en Colombia y de la
Masifcación de Internet en la Estrategia de Gobierno en Línea, Bogota, Centro de Investigación de la
Telecomunicaciones (CINTEL), 2011.
Note: The robustness of the model was evaluated on the basis of tests of independence between variables,
normality and covariance. In addition, the analysis was extended by performing the Doornik-Hansen
multivariate normality test. In all cases, the validity of the impact estimation models was found to be over 99%.
***, ** and * indicate signifcance levels of 1%, 10% and 15%, respectively.
Broadband in Latin America: Beyond Connectivity 115
The model shows that an increase in broadband connections has
a positive effect on GDP growth: if connections increased 10%, GDP
would increase 0.037%. This effect is less than was found in the model
for the region as a whole, mainly because the average rate of broadband
penetration in Colombia is below the average for Latin America. This
suggests the existence of a returns to scale effect —i.e. economic impact
increases with higher levels of penetration.
Growth of broadband connections is the only independent variable
that is signifcant in explaining the growth of GDP in all specifcations,
both for departments with high penetration and for departments with low
penetration (columns 3 and 4). The coeffcients for departments with high
and low penetration are similar because in no case was penetration more
than 20% (hence, by international standards, all departments have low
penetration). The model explains between 15% and 20% of the variance in
the dependent variable, indicating that there are other factors affecting the
evolution of GDP.
3
Nevertheless, the coeffcient for impact of broadband
growth is signifcant and consistent in all specifcations.
For Panama, as more information was available, Katz and
Koutroumpis (2012a) used a multiple-structure model originally developed
by Roller and Waverman (2001) for fxed-line telephony and subsequently
adapted by Koutroumpis (2009) for broadband and by Gruber and
Koutroumpis (2011) for mobile telephony. The model comprises four
equations: a production function, which models the aggregate operation of
the economy, and demand, supply and output functions, which model the
broadband market, controlling for reverse causality.
In the aggregate production function, GDP is linked to the stock of
fxed capital (excluding ICT infrastructure), the supply of skilled labour and
fxed broadband infrastructure, proxied by broadband penetration. The
demand function links broadband penetration to the price of basic service
—the number of subscribers depending on the price of broadband access—
and average individual consumption estimated on the basis of GDP per
capita. The supply function links aggregate broadband sales revenues to
broadband price levels, GDP per capita and degree of urbanization in the
country. To the extent that fxed broadband deployment is correlated with
greater urban population concentration, the supply of the service should
refect this structural variable. The output equation links the annual change
in fxed broadband penetration with broadband sales revenue, and this
change is used as an indicator of annual capital investment in broadband.
4

3
Lack of department-level data for these variables prevents their inclusion in the regression
model.
4
This assumes a stable and constant relationship between sales and investment, which
often does not occur. Information on fixed capital formation in telecommunications,
which would be a more appropriate variable, was not available.
ECLAC 116
Based on these models (see table IV.3), fxed broadband
infrastructure had a signifcant impact on growth between 2000 and 2010.
Its average annual contribution to GDP growth was estimated at 0.045%
for every 1% increase in penetration.
Table IV.3
PANAMA: CONTRIBUTION OF BROADBAND TO GDP GROWTH
Aggregate production function:
GDP
it
=a
1
K
it
+a
2
L
it
+a
3
BB_Pen
it
+Ɛ
1it
(1)
Demand function:
BB_Pen
it
=b
1
BBPr
it
+b
2
GDPC
it
+Ɛ
2it
(2)
Supply function:
BB_Rev
it
=c
1
GDPC
it
+c
2
Urb
it
+Ɛ
3it
(3)
Output function:
ΔBB_Pen
it
=d
1
BB_Rev
it
+Ɛ
4it
(4)
Variables Fixed broadband model
Growth (GDP
it
)
Labour force with secondary education (L
it
) 1.148***
Fixed capital stock (K
it
) 0.234***
Fixed broadband penetration (BB_Pen
it
) 0.045***
Constant -
Demand (BB_Pen
it
)
Fixed broadband price (BBPr
it
) -2.121***
GDP per capita (GDPC
it
) 2.443***
Constant -18.536**
Supply (BB_Rev
it
)
GDP per capita (GDPC
it
) 0.556***
Urbanization (Urb
it
) 0.374***
Constant 13.910***
Output (ΔBB_Pen
it
)
Broadband revenue (BB_Rev
it
) 4.606***
Constant -95.451***
Year effects SÍ
Observations 40
R
2
Growth 0.99
Demand 0.92
Supply 0.97
Output 0.40
Source: R. Katz and P. Koutroumpis, The Economic Impact of Broadband: Case Studies of the Philippines
and Panama, Geneva, International Telecommunication Union, 2012.
Note: The symbols ***, ** and * indicate signifcance levels of 1%, 10% and 15%, respectively.
Broadband in Latin America: Beyond Connectivity 117
Comparison of the results for Colombia and Panama confrms, frst,
the existence of returns to scale. The economic contribution of broadband
is greater in Panama, where, in 2010, fxed broadband penetration was 7.8%
versus 4.8% in Colombia. Beyond this comparative result, the multiple-
structure model for Panama yields other conclusions. In addition to the
contribution of capital, a 1% increase in skilled labour would increase GDP
by 1.15%. The model also shows the relative importance of broadband
subscription rates: a 10% decrease in broadband price would increase
broadband penetration by more than 21%.
The interpretation of the model results suggests the existence of
returns to scale from broadband. This type of effect had already been
identifed for other ICTs, such as telephony (Roller and Waverman,
2001). Comparison of the results of different estimates of the economic
contribution of broadband as a function of penetration shows evident
returns to scale (see fgure IV.1).
5
Figure IV.1
COMPARATIVE CONTRIBUTION OF BROADBAND TO ECONOMIC GROWTH
(Percentages)
Brazil
Chile
Panama
Colombia
0
5
10
15
20
25
30
35
0 5 10 15 20 25 30 35 40
Koutroumpis Katz
OECD
Low penetration
OECD
Medium penetration
OECD
High penetration
Latin America
Germany
Low penetration
Germany
High penetration
C
o
n
t
r
i
b
u
t
i
o
n

t
o

G
D
P

g
r
o
w
t
h

f
o
r

e
v
e
r
y

1
0
%
i
n
c
r
e
a
s
e

i
n

b
r
o
a
d
b
a
n
d

p
e
n
e
t
r
a
t
i
o
n
Broadband penetration
Source: R. Katz, The Impact of Broadband on the Economy: Research to Date and Policy Issues, The
Impact of Broadband on the Economy Broadband Series, Geneva, International Telecommunication Union
(ITU), 2012.
Note: The trend line does not refect observations for Latin America.
In summary, although the results are based on differently specifed
models, the greater the broadband penetration, the greater impact its
expansion will have on GDP growth. The public policy implications are
clear: maximizing the economic impact of broadband is contingent on
signifcantly increasing its penetration.
5
The significant effects in the case of Panama (not shown in this figure) are due to the
importance of broadband in a service-based economy oriented mainly towards commerce
and financial services.
ECLAC 118
2. Job creation
In addition to its impact on economic growth, broadband contributes
to job creation, although its effects in this area are more complex. First,
increased broadband penetration can increase productivity, which in
the short term may lead to net job loss (an effect that has been verifed
by the author for labour-intensive industrial sectors). Second, by
incorporating new sectors of the population into the broadband market,
the technology contributes to the creation of new businesses through an
innovation effect, which generates new jobs. Lastly, broadband can lead
to the outsourcing of certain business functions, which can create jobs
as outsourcing frms are established, but it can also lead to the loss of
opportunities if business functions in the country under consideration
are transferred to other locations. The sum of these three effects is shown
in diagram IV.2.
Diagram IV.2
HOW BROADBAND CONTRIBUTES TO JOB CREATION
Increased
broadband
Impact of
e-business
on corporate
productivity
Macroeconomic
productivity
Increased
innovation
Outsourcing
of services
Shift to services
sector
Impact on
employment
+
+
+
-
+
+
+/-
+
+
Source: Adapted by the author from a model developed by M. Fornefeld, G. Delaunay and D. Elixmann,
“The impact of broadband on growth and productivity”, study for the European Commission, Directorate
General for Information Society and Media, 2008.
It is not possible on the basis of the available information to
measure the individual impact of each of these three effects, so the
analyses had to be limited to measurement of aggregate impact. The job
creation potencial of broadband was estimated for Chile, Colombia and
the Dominican Republic.
To estimate the impact of broadband in Chile, a study was conducted
using quarterly panel data controlled for fxed effects refecting specifc
features of each region of the country that have an impact on the labour
market (industrial sectors, education levels). Information was compiled for
Broadband in Latin America: Beyond Connectivity 119
all regions (except the Santiago metropolitan region, for which quarterly
data were not available) for the period from 2001 to the fourth quarter of
2009 (see table IV.4).
Table IV.4
CHILE: IMPACT OF BROADBAND ON JOB CREATION
(12 regions, 2001-2009)
Coeffcient
Standard
error
t-statistic P>|t|
95%
confdence
interval
Economic activity
index
a
0.0003509 0.0000595 5.90 0.000 0.0002338
Change in
broadband
penetration
0.0018118 0.0004708 3.85 0.000 0.0008853
Constant 0.8682527 0.0079638 109.03 0.000 0.85258283
Number of observations 324
F(2,310) 60.89
Prob>F 0.0000
R
2
0.2820
Source: R. Katz, “The impact of broadband on economic growth”, Fast-tracking the digital revolution:
Broadband for Latin America and the Caribbean (LC/R.2167), V. Jordán, H Galperin and W Peres, Santiago,
Economic Commission for Latin America and the Caribbean (ECLAC), 2011.
a
Control variable.
The model results indicate that in Chile a 10% increase in broadband
penetration would increase the employment rate by 0.018 points. The
increase in broadband penetration appears to be statistically signifcant
in explaining the employment rate trend during the period under
consideration.
A similar analysis was performed for Colombia. In this case, a
model was constructed at the departmental level, and the relationship
between the growth of broadband connections and the rate of employment
growth was analysed, controlling for population growth and initial level
of economic development. The model was applied frst to the country as
a whole, and the information was then disaggregated by department and
level of penetration (high or low).
In these models, the effect is statistically signifcant at the national
level and for low-penetration departments; in departments with high
penetration, the coeffcient is signifcant at 24%. Population growth, on the
other hand, seems to have an effect only in high-penetration departments
(with a positive coeffcient), possibly because the innovation effect
mentioned above makes it easier to enter the market in those departments.
Initial GDP seems to have no effect on employment growth.
ECLAC 120
Table IV.5
COLOMBIA: IMPACT OF BROADBAND PENETRATION GROWTH
ON EMPLOYMENT GROWTH
Employment rate growth (%)
Dependent variable: employment rate growth between 2006 and 2010
Independent variables: growth in broadband connections, GDP 2003 and population growth
Total Low penetration High penetration
Growth in broadband
connections (percentage)
0.0003004** 0.0002951** 0.0006572
(0.0001359) (0.0001547) (0.0005495)
Population growth (percentage) 0.0159829 -0.2538734 0.5937073*
(0.5114836) (0.7899623) (0.3761862)
GDP 2003 (millions of pesos) 0.0053431 -0.1084577 0.0003309
(0.0077051) (0.1308956) (0.0090124)
Adjusted R
2
0.0110 0.0318 0.0338
Prob > F 0.0730 0.0321 0.4351
Number of observations 132 64 68
Source: R. Katz and F. Callorda, Medición de Impacto del Plan Vive Digital en Colombia y de la
Masificación de Internet en la Estrategia de Gobierno en Línea, Bogota, Centro de Investigación de la
Telecomunicaciones (CINTEL), 2011.
Note: The symbols ***, ** and * indicate signifcance at a level of 1%, 10% and 15%, respectively.
In the study of broadband impact on job creation in the Dominican
Republic, a model was constructed using panel data for 32 provinces.
Unlike the models used for Chile and Colombia, the objective was to
determine the impact of broadband in reducing unemployment.
The results show the high impact of broadband (see table IV.6), with
a 1% increase in penetration reducing unemployment by 0.29 percentage
points. Other variables affecting the unemployment rate indirectly are, as
expected, change in the number of industrial establishments between 2008
and 2009 and the importance of the construction sector in 2009. Thus, a
combination of increased broadband penetration, growth in construction
and an increase in the number of industrial establishments has a signifcant
impact in reducing unemployment.
Based on the coeffcients, the contribution of broadband relative to
the other two variables is higher than would have been expected. Part of
this effect is due to greater growth in broadband penetration in the capital,
Santo Domingo, and in a resort area, Altagracia. To determine the relative
contribution of broadband, it would be important to include in the model
a variable refecting the importance of this sector in each province. Such
a variable is not available for all provinces, however. Hence, although the
model shows that broadband plays an important role in creating jobs, it
is diffcult to measure its impact in relation to the development of key
industrial sectors in the country.
Broadband in Latin America: Beyond Connectivity 121
Table IV.6
DOMINICAN REPUBLIC: IMPACT OF INCREASED BROADBAND
PENETRATION ON EMPLOYMENT GROWTH
Growth in
unemployment
Coeffcient
Standard
error
t-statistic P>t
95% confdence
interval
Population growth 0.72442 0.24939 2.90 0.0070 0.21180 1.23704
Change in broadband
penetration
–0.29529 0.13290 –2.22 0.0350 –0.56846 –0.02211
Change in number of
establishments
–0.14959 0.04728 –3.16 0.0040 –0.24678 –0.05241
Value of construction
industry 2009
0.69456 0.14588 4.76 0.0000 0.39469 0.99443
Change in construction
2008-2009
–0.64299 0.12787 –5.03 0.0000 –0.90583 –0.38015
Constant 0.74317 0.37360 1.99 0.0570 –0.02477 1.51111
Number of observations 32
F(5,26) 12.70
Prob>F 0.0000
R
2
0.4175
Source: R. Katz, The Impact of Broadband on the Economy: Research to Date and Policy Issues, The
Impact of Broadband on the Economy Broadband Series, Geneva, International Telecommunication Union
(ITU), 2012.
3. Growth of household income
This third economic effect of broadband is important since an increase
in the growth of average household income has an impact on poverty
reduction in a country. This aspect is crucial because, although it has
been shown that broadband contributes to GDP growth, it is important
to ensure that this growth does not favour only the higher-income
population, resulting in greater social polarization (Fernandez-Ardevol
and Vázquez Grenno, 2011). Studies in Costa Rica and Colombia have
looked at this question.
In a study assessing the economic impact of Costa Rica’s national
broadband strategy, Katz (2011) conducted an analysis based on the
National Household Survey between 2005 and 2009. Panel data with
random effects were used for estimates where the results by region are
specifc to a given period (see table IV.7).
6
6
At the same time, the White method was used to correct for potential error bias and thus
to increase the statistical significance of the coefficients.
ECLAC 122
Table IV.7
COSTA RICA: IMPACT OF BROADBAND ON GROWTH
IN REAL HOUSEHOLD INCOME
Growth in
household income
Coeffcient
Standard
error
Z p>|z|
Confdence
interval 95%
Household income (-1) -0.000337 0.000033 10.08 0.0000 -0.0004 -0.0003
Broadband penetration 2.960308 0.970254 3.05 0.0020 1.0586 4.8620
No education -4.603882 0,889184 -5.18 0.0000 -6.3437 -2.8611
< 3 persons 1.923927 0,446712 4.31 0.0000 1.0484 2.7995
Manufacturing 2.526376 1.017825 2.48 0.0130 0.5315 4.5213
Agriculture 0.708006 0.195230 3.63 0.0000 0.3254 1.0907
Hotels and restaurants 2.665666 0.302174 8.82 0.0000 2.0734 3.2579
Exports (-1) 0.010438 0.001638 6.37 0.0000 0.0072 0.0136
Constant -98.568610 31.663730 -3.11 0.0020 -160.6284 -36.5088
Number of observations 24
Number of groups 6
R
2
within groups 0.8029
R
2
between groups 0.8119
R
2
total 0.7971
Source: R. Katz, “Impacto económico de la Estrategia Nacional de Banda Ancha”, Gobierno de Costa
Rica, Rectoría de Telecomunicaciones, Estrategia Nacional de Banda Ancha, San José, 2011.
According to the model results, a one-percentage-point increase
in regional broadband penetration raises average household income by
2.96%; the rise in household income is larger if the head of household
is employed in manufacturing or in the tourism sector (hotels and
restaurants). Achieving broadband penetration of 10% would increase
real average monthly household income in Costa Rica by the equivalent of
US$ 48. Thus, if broadband penetration reached 16%, average household
income would increase by US$ 141. In turn, these increases in household
income would contribute to GDP growth by boosting consumption.
For Colombia, the specifed model aimed to study the impact
of growth in broadband connections on the growth of real household
income in 2006-2010. Following the literature, controls were included for
population growth, human capital, share of output accounted for by the
mining sector and initial level of wealth (as measured by the percentage of
households with unmet basic needs (UBNs) in 2005).
Broadband in Latin America: Beyond Connectivity 123
Table IV.8
COLOMBIA: IMPACT OF BROADBAND PENETRATION GROWTH
ON REAL HOUSEHOLD INCOME GROWTH
Percentage increase in real household income, controlled for human capital
a
Total Low penetration High penetration
Growth in broadband
connections (percentage)
0.0034083 0.0035966 0.0025196
(0.0011585) (0.0013686 ) (0.0011616)
Population growth
(percentage)
-2.533624 -5.520381 1.702465
(1.245529) (1.361513) (1.19664)
Years of education 1.462938 0.4542847 0.1371095
(0.7531259) (1.273384) (0.7649286)
Mining sector output
(percentage)
7.816958 9.122359 8.837977
(4.226792) (4.701466) (8.11938)
Households with UBNs 2005
(percentage)
19.7768 31.17167 -34.74956
(9.51923) (10.61504) (28.60452)
Adjusted R
2
0.1885 0.2986 0.1435
Prob > F 0.0101 0.0006 0.0672
Number of observations 132 64 68
Source: R. Katz and F. Callorda, Medición de Impacto del Plan Vive Digital en Colombia y de la
Masificación de Internet en la Estrategia de Gobierno en Línea, Bogota, Centro de Investigación de la
Telecomunicaciones (CINTEL), 2011.
Note: the symbols ***, ** and * indicate signifcance at a level of 1%, 10% and 15%, respectively.
a
Dependent variable: growth in real household income between 2006 and 2010. Independent variables:
growth in broadband connections, population growth, years of education, mining sector output and
households with unmet basic needs in 2005.
The main fnding is that if the number of connections increases by
10% in a year, real household income would rise by 0.034% (see table IV.8).
Growth in broadband penetration consistently explains increases in
real household income in the three models (national level, departments
with low penetration and departments with high penetration). The
effect appears to be greater in departments with low penetration,
although, except in Bogota, no department had more than 9% broadband
penetration in 2010 (in other words, all departments had low penetration
by international standards). Thus, it can be concluded that none of
Colombia’s departments had reached a suffcient level of penetration in
2010 to see returns to scale of the type that exist in OECD countries.
7
7
In keeping with the human capital theory, a larger number of years of education results
in greater growth in household income. Moreover, In accordance with the Solow growth
model, there is a trend towards income convergence, as indicated by the coefficient
“households with UBNs in 2005”.
ECLAC 124
B. Digitization and development
Beyond the impact of broadband access, it is important to study the
combined impact of all services and applications enabled by broadband.
The concept of digitization —defned as the ability to use digital
technologies to generate, process and share information— was developed
for that purpose (Katz and Koutroumpis, 2012b). This indicator measures
not only the penetration of the technology, but also the use of applications
and the consumption of content at three levels: (i) individuals, businesses
and government; (ii)  goods and services production processes; and
(iii) provision of public services.
In order for digitization to reach its full potential, it must meet the
following conditions with respect to infrastructure: affordability (prices),
technological accessibility (network coverage) and technological reliability
(capacity and access speed). A composite index based on the 23 indicators
listed in diagram IV.3 was created in order to measure a country’s level of
development in terms of digitization.
The digitization index is composed not only of infrastructure
indicators but also includes information regarding the adoption of
applications and services transmitted by broadband —for example, use
of e-commerce, mobile broadband, social networks, and e-government
services. Calculations of the digitization index for a sample of 184 countries
in 2011 indicated that countries pass through four stages of development
(see fgure IV.2).
Generally speaking, industrialized countries score above 50 on the
digitization index. Countries in transition to advanced digitization have
scores ranging from 35 to 50; this group includes countries in the Middle
East, Eastern Europe and South-East Asia and some Latin American
countries (Argentina, Brazil, Chile, Colombia, Costa Rica, Mexico, Panama
and Uruguay). The majority of Latin American countries and some
African and Asian countries have scores between 20 and 35, while the
least developed countries score under 20.
Broadband in Latin America: Beyond Connectivity 125
Diagram IV.3
COMPONENTS OF THE DIGITIZATION INDEX
Digitization
index
Affordability
Price of ICT products and services for
individuals and businesses
Reliability
Level of reliability of networks that
carry digital information
Access
Adoption of terminals and networks
that enable individual users and
businesses to access services,
applications and content
Capacity
Network capacity as an indicator of
quality of service
Use
Processes, indicating assimilation of
digital technologies Adoption of
applications and changes in business
Human capital
Labour force level of training
in use of ICT
Average prices for fixed-line and mobile
telephony and broadband adjusted for GDP
per capita. Includes installation costs, fixed
per-minute costs, mobile connection costs
and broadband subscription costs
Measured as level of investment per subscriber
in fixed-line, mobile and broadband services
Average of the sum of fixed and mobile
broadband penetration, PC penetration and
mobile network coverage
Measures international access network
bandwidth and nominal download speed
offered in the market
Average penetration of e-commerce,
e-government services, mobile data services,
social network usage, percentage of individuals
using the Internet, data as a percentage of
mobile average revenue per user (ARPU)
Percentage of engineers in the labour force
and percentage of population with higher
education
Source: R. Katz, P. Koutroumpis and F. Callorda, “The Latin American path towards digitization”, Info,
vol. 15, Nº3, 2013, pp. 6-24.
Figure IV.2
TYPOLOGY OF COUNTRIES BY STAGE OF DIGITIzATION, 2011
0
10
20
30
40
50
60
70
Transitional Advanced Emerging Constrained
Region Constrained Emerging Transitional Advanced
Sub-Saharan Africa 35 6 1 0
South-East Asia 8 6 0 0
East Asia and the Pacifc 5 7 4 6
CIS and Russian Fed. 3 3 3 2
Middle East and
Northern Africa
4 9 7 2
Latin America and
the Caribbean
3 14 13 0
Eastern Europe 0 3 13 4
North America 0 0 1 2
Western Europe 0 0 3 17
Source: R. Katz, P. Koutroumpis and F. Callorda, “The Latin American path towards digitization”, Info,
vol. 15, Nº3, 2013, pp. 6-24.
ECLAC 126
Analysis of the six sub-indices in the digitization index reveals
that the biggest difference between advanced and emerging countries is
not necessarily in technological infrastructure but rather in network use
and capacity and in availability of the local human capital necessary to
develop applications and content. Although there are large differences
between developed and emerging countries in access to fxed broadband
(one of the components of the index), the recent deployment of mobile
broadband has done much to narrow the gaps. For all countries, then,
the sub-index for use of technology never reaches the same level of
development as the access sub-index, although the numerical distance
between the infrastructure and use of technology sub-indices is less in
advanced countries than in developing ones. Overcoming the lag in the
adoption of applications and content, especially in countries at a medium
stage of development, is the greatest technological challenge. In other
words, the focus of a technology strategy should be more on applications
and services than on infrastructure development. Models of the economic
impact of digitization show that it has greater effect than broadband or
mobile telephony by themselves (see fgure IV.3).
Figure IV.3
DIGITIZATION AND ECONOMIC GROWTH
(Percentages)
0.23
0.14
0.08
0.26
0.24
0.16
0.08
0.09
0.17
0.11
0.2
0.81
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
O
E
C
D

-

h
i
g
h
O
E
C
D


-

m
e
d
i
u
m
O
E
C
D

-

l
o
w
G
e
r
m
a
n
y

-

h
i
g
h
G
e
r
m
a
n
y

-

l
o
w
L
a
t
i
n

A
m
e
r
i
c
a
B
r
a
z
i
l
C
h
i
l
e
M
a
l
a
y
s
i
a
L
o
w

i
n
c
o
m
e
H
i
g
h

i
n
c
o
m
e
D
i
g
i
t
i
z
a
t
i
o
n
Fixed broadband studies Mobile telephone
studies
I
m
p
a
c
t

o
f

1
0
%

i
n
c
r
e
a
s
e

i
n

t
h
e

i
n
d
e
p
e
n
d
e
n
t

v
a
r
i
a
b
l
e

o
n

G
D
P

p
e
r

c
a
p
i
t
a
Source: R. Katz and P. Koutroumpis, “Measuring Socio-Economic Digitization: A Paradigm Shift”,
2012, unpublished.
Broadband in Latin America: Beyond Connectivity 127
A 10% rise in the digitization index results in a 0.81% increase in GDP
per capita. This result is highly signifcant as it suggests that the economic
impact of ICT is the result of cumulative adoption of all technologies, as
well as of the assimilation of content and applications. Achieving high
broadband penetration is just a telecommunications policy objective;
maximizing its economic impact requires a combination of policies in the
areas of telecommunications, information technology, and content and
applications. In addition, disaggregating the economic impact model in
order to measure the contribution of digitization by stage of digitization
again reveals the existence of returns to scale (see fgure IV.4).
Figure IV.4
RETURNS TO SCALE FROM DIGITIZATION
(Percentages)
I
m
p
a
c
t

o
f

1
0
%

i
n
c
r
e
a
s
e

i
n

d
i
g
i
t
i
z
a
t
i
o
n

i
n
d
e
x

o
n

G
D
P
0.751 0.774
0.777
0.681
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
General
model
Constrained Emerging Transitional Advanced
Stage of digitization
Source: R. Katz and P. Koutroumpis, “Measuring Socio-Economic Digitization: A Paradigm Shift”, 2012,
unpublished.
While, as shown above, the overall model indicates that a 10%
increase in digitization results in a 0.81% increase in GDP per capita,
for advanced-stage countries the fgure is 0.681% and for countries with
lower levels of digitization it is between 0.751% and 0.777%. The results
presented in fgure IV.4 confrm the hypothesis of increasing returns,
but they could also indicate that diminishing returns begin to emerge at
an advanced stage of digitization. The impact is less pronounced among
advanced-stage countries than among countries in the transitional
and emerging stages. The latter two stages are closely linked and are
in a different position in relation to countries at a constrained stage.
Obviously, there is considerable heterogeneity within these categories,
ECLAC 128
which could affect the results. However, from an aggregate perspective
it is clear that there are increasing returns to scale in the process and that
the benefts began to be seen at a digitization level of 30, with a saturation
effect appearing at around 65.
Digitization also has a greater impact on job creation than
broadband alone (see fgure IV.5). A 10% increase in the digitization
index results in an increase of 0.82% in employment. This effect can, once
again, be attributed to two trends. First, the deployment and assimilation
of ICT contributes more to job growth in technology-intensive sectors
(software development, business process outsourcing, manufacturing
of equipment and parts). Second, the assimilation of ICT has a spillover
effect on other sectors of the economy, especially commerce, fnancial
services and health services.
Figure IV.5
DIGITIZATION AND EMPLOYMENT
I
m
p
a
c
t

o
f

a

1
0
-
p
o
i
n
t

i
n
c
r
e
a
s
e

i
n

d
i
g
i
t
i
z
a
t
i
o
n

i
n
d
e
x

o
n

j
o
b

c
r
e
a
t
i
o
n
0.3
0.06
0.03
0.38
0.29
0.45
0.02
0.82
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Broadband studies
U
n
i
t
e
d

S
t
a
t
e
s

(
m
e
t
r
o
p
o
l
i
t
a
n

a
r
e
a
s
)
G
e
r
m
a
n
y

–

h
i
g
h
G
e
r
m
a
n
y

–

l
o
w
M
e
x
i
c
o
A
r
g
e
n
t
i
n
a
B
r
a
z
i
l
C
h
i
l
e
D
i
g
i
t
i
z
a
t
i
o
n
Source: R. Katz and P. Koutroumpis, “Measuring Socio-Economic Digitization: A Paradigm Shift”, 2012,
unpublished.
Lastly, a country’s level of digitization is closely correlated with
the innovation capacity of its economy (see fgure IV.6), although, as
in the models presented above, it is not possible to confrm a causality
relationship. The impact of digitization on innovation capacity would
appear to be attributable to the capacity of the digitization environment to
facilitate the creation of new products and services that add value.
Broadband in Latin America: Beyond Connectivity 129
Figure IV.6
DIGITIzATION AND INNOVATION IN A SAMPLE OF 125 COUNTRIES
0
10
20
30
40
50
60
70
0 10 20 30 40 50 60 70
I
n
n
o
v
a
t
i
o
n

i
n
d
e
x
Digitization index
Source: R. Katz and P. Koutroumpis, “Measuring Socio-Economic Digitization: A Paradigm Shift”, 2012,
unpublished.
Based on the correlation coeffcient, a 10% increase in digitization
would result in a 6.4% increase in innovation. This boost to innovation
comes from the introduction of ICT-facilitated services and applications
(including new ones like telemedicine, Internet searching, e-commerce,
distance education and social networks) and new forms of commerce and
fnancial intermediation. These estimates show that the economic impact
of digitization is signifcant (see table IV.9).
Table IV.9
LATIN AMERICA: ESTIMATED ECONOMIC IMPACT OF DIGITIZATION
Country
Indicators (2011)
Change as a result of 10% increase
in digitization index
a
Digitization
index
GDP per
capita
b

(dollars)
Innovation
index
Digitization
index
GDP per
capita
b

(dollars)
Innovation
index
Argentina 41.32 10 881 34.40 45.45 10 969 36.60
Brazil 36.61 12 594 36.60 40.27 12 696 38.94
Chile 45.33 13 738 42.70 49.86 13 849 45.43
Colombia 38.33 7 121 35.50 42.16 7 179 37.77
Costa Rica 37.33 8 644 36.30 41.06 8 714 38.62
Ecuador 32.75 4 504 28.50 36.03 4 540 30.32
El Salvador 29.56 3 602 29.50 32.52 3 631 31.39
Mexico 37.05 9 980 32.90 40.76 10 061 35.01
Panama 44.29 8 740 30.90 48.72 8 811 32.88
Paraguay 28.68 3 594 31.60 31.55 3 623 33.62
Peru 32.20 5 860 34.10 35.42 5 907 36.28
Uruguay 42.78 14 294 35.10 47.06 14 410 37.35
Source: World Economic Forum, Maximizing the Impact of Digitization, Global Information Technology
Report (GITR), Geneva, 2012; R. Katz and P. Koutroumpis, “Measuring Socio-Economic Digitization: A
Paradigm Shift”, 2012, unpublished.
a
An ICT development policy results in a 10% increase in the digitization index.
b
Constant United States dollars.
ECLAC 130
In conclusion, digitization has a positive economic impact. Every
10% rise in the digitization index increases GDP by 0.81% and decreases
the unemployment rate by 0.82%. In this process, there are increasing
returns to scale, the benefts of which are seen mainly after the index
reaches a threshold of 30 points; a saturation point is reached at around 50
points. This suggests that countries should accelerate the development of
digitization, in particular for usage, applications and content, in order to
maximize its impact on economic growth.
C. Policy implications
There is abundant empirical evidence of the economic impact of
broadband and its positive externalities in innovation, productivity and
business restructuring. Studies are beginning to show that these effects
vary with the environment in which broadband is deployed (more or
less developed regions). This highlights the need for prospective impact
studies in order to target plans and investment, as well as to coordinate
broadband deployment with regional economic development programmes
in less developed areas. From an impact research standpoint, more work is
needed on threshold and saturation levels in order to identify quantitative
targets for digitization programmes. This is because the broadband
demand gap is the main obstacle standing in the way of increasing the
digitization index in Latin America (see table IV.10).
Table IV.10
LATIN AMERICA: BROADBAND SUPPLY AND DEMAND GAP, 2011
(Percentages)
Country
Fixed broadband Mobile broadband
Supply gap
(network
coverage)
Demand
gap
a
Supply gap
(network coverage)
Demand
gap
Argentina 4 55 8 73
Bolivia (Plurinational
State of)
60 37 71 26
Brazil 6 65 16 63
Chile 22 34 28 65
Colombia 19 54 4 87
Costa Rica 5 63 7 82
Ecuador 13 67 34 55
Mexico 38 15 23 63
Peru 41 43 37 54
Source: Katz and Galperín (2013).
a
The difference between the percentage of population covered and the penetration rate.
Broadband in Latin America: Beyond Connectivity 131
As expected, the percentage of households that could obtain fxed
broadband service but do not do so is signifcant. Excluding countries
with low fxed-line telephony coverage (Peru and the Plurinational State of
Bolivia), the demand gap ranges from 67% (Ecuador) to 15% (Mexico). The
situation with respect to mobile broadband is similar: excluding countries
with low mobile coverage (Plurinational State of Bolivia), the demand gap
is signifcant, ranging from 87% (Colombia) to 54% (Peru). However, mobile
broadband is still in the embryonic stages; dissemination projections
indicate that the demand gap will narrow signifcantly in the coming years.
The demand gap is determined by generational, education and
economic factors. Studies in industrialized and developing countries show
that broadband adoption and Internet access are associated with younger
generations. In the study of Costa Rica, over 80% of individuals accessing
the Internet by means of a computer at home were between 15 and 24
years of age. The number drops substantially among age groups over 45
and is only 20% among persons aged over 55. Just as broadband adoption
is associated with younger generations, greater household Internet and
computer penetration are associated with higher levels of education. In
the Costa Rican study, households with lower levels of education showed
considerably lower adoption rates (less than 50% in terms of service use).
In contrast, more than 70% of households headed by an individual with
post-secondary education used computers and broadband. In keeping
with the affordability argument put forward by Galperin and Ruzzier
(2010), the third explanation for the demand gap is the economic factor.
In this context, public policies aimed at addressing the demand gap
should be underpinned by four fundamental principles. First, to the extent
that affordability is one of the main obstacles to broadband adoption, it
is important to highlight the benefts of increased competition among
private operators as a factor leading to lower prices. Second, beyond the
benefts of competition, the State should play a key role in promoting
programmes designed to encourage broadband adoption. Areas to be
prioritized include education and training programmes, deployment of
e-government services that enhance the appeal of broadband service and
implementation of subsidy programmes for the purchase of equipment.
Third, one of the most important factors in stimulating broadband
adoption is the development of applications that meet individual needs, in
both the social and the economic realms. The role of the applications and
equipment ecosystem in fostering demand is critical. Lastly, public policies
aimed at promoting broadband adoption should be long-range ones, since
some of their results will not materialize in the short term. Accordingly,
such initiatives should stem from State policies that go beyond political
and electoral cycles.
ECLAC 132
Bearing in mind these principles, four public policy areas for
encouraging broadband adoption and increasing digitization can be
identifed. From an economic perspective, consideration should be given
to eliminating sales tax on basic broadband service and on computers
(especially import duties) and to offering subsidies to reduce monthly
service fees for certain benefciaries. It is also important to negotiate with
broadband providers with a view to ensuring the availability of affordable
broadband services.
In the area of education, ICT should fgure more prominently in the
curriculum. Secondary schools and specialized training institutions should
be encouraged to offer short courses or university extension classes on
ICT. Governments should promote digital literacy programmes targeting
disadvantaged groups, older persons and persons with disabilities. They
should also introduce initiatives to promote broadband adoption by small
and medium-sized enterprises (SMEs). Measures to be considered include
reducing taxes on the purchase of computer equipment and on broadband
services, allowing accelerated depreciation of equipment and offering
discounts or rewards to companies using ICT and broadband for their
transactions with the government. Such economic incentives should be
accompanied by training programmes for SME workers and advisory
services to enable business owners to install and obtain maximum beneft
from ICT.
Finally, with regard to incentives for broadband adoption arising from
direct government action, it is important to develop content on portals
providing information on cultural topics, preventive health care and public
services. Mechanisms should also be put in place to promote e-government
services such as electronic tax payment systems, e-procurement systems
for use by suppliers of goods and services to the government and the
development of platforms to facilitate telework.
Broadband in Latin America: Beyond Connectivity 133
Bibliography
Fernández-Ardevol, M. and J. Vázquez Grenno (2011), “Estimación de la contribución
de la telefonía móvil al crecimiento y reducción de la pobreza”, Comunicación
Móvil y Desarrollo Económico y Social en América Latina, M. Fernández-Ardevol,
H. Galperin and M. Castells, Barcelona, Ariel.
Fornefeld, M., G. Delaunay and D. Elixmann (2008), “The impact of broadband
on growth and productivity”, study for the European Commission
(DG Information Society and Media), MICUS.
Galperin, H. and C. Ruzzier (2010), “Broadband tariffs: benchmarking and analysis”,
Fast-tracking the digital revolution: Broadband for Latin America and the Caribbean,
V. Jordán, H. Galperin and W. Peres, ECLAC-DIRSI, Santiago.
Gruber, H. and P. Koutroumpis (2011), “Mobile telecommunications and the impact
on economic development”, Economic Policy, vol. 67, July.
Jordana, J. (2001), “Desigualtats digitals i societat de la informació: un debat
pendent”, Papers de la Fundació, No. 130, Barcelona, Fundació Rafael Campalans.
Katz, R. (2009), El papel de las TIC en el desarrollo: Propuesta de América Latina a los retos
económicos actuales, Colección Fundación Telefónica, Madrid, Ariel.
___ (2011a), “The impact of broadband on economic growth”, Fast-tracking the digital
revolution: Broadband for Latin America and the Caribbean (LC/R.2167), V. Jordan,
H Galperin and W. Peres, ECLAC-DIRSI, Santiago.
___ (2011b), “Impacto económico de la Estrategia Nacional de Banda Ancha”,
Gobierno de Costa Rica, Rectoría de Telecomunicaciones, Estrategia Nacional
de Banda Ancha, San José.
___ (2012), The Impact of Broadband on the Economy: Research to Date and Policy
Issues, The Impact of Broadband on the Economy Broadband Series, Geneva,
International Telecommunication Union (ITU).
Katz, R. and F. Callorda (2011), Medición de Impacto del Plan Vive Digital en Colombia y
de la Masifcación de Internet en la Estrategia de Gobierno en Línea, Bogotá, Centro
de Investigación de la Telecomunicaciones (CINTEL).
Katz, R. and H. Galperin (2013), “The demand gap: drivers and public policies”,
H. Galperin, V. Jordán and W. Peres, Broadband in Latin America: beyond
connectivity, ECLAC-DIRSI, Santiago, Chile.
Katz, R. and P. Koutroumpis (2012), The Economic Impact of Broadband: Case Studies
of the Philippines and Panama, Geneva, International Telecommunication Union.
Katz, R., P. Koutroumpis and F. Callorda (2013), “The Latin American Path Towards
Digitalization”, Info, vol. 15, No. 3, pp. 6-24.
Koutroumpis, P. (2009), “The Economic Impact of Broadband on Growth:
A Simultaneous Approach”, Telecommunications Policy, No. 33.
Roller, L-H. and L. Waverman (2001), “Telecommunications Infrastructure and
Economic Development: A simultaneous approach”, American Economic Review,
vol. 91, No. 4.
World Economic Forum (2012), Maximizing the Impact of Digitization, Global
Information Technology Report (GITR), Geneva.
Chapter V
Mobile broadband: the urgent need
for speedier roll-out
Ernesto M. Flores-Roux
1
A. Introduction
The importance of telecommunications in development has been
recognized for decades. Much of the literature that has looked at this
phenomenon has focused on assessing its impact on the transformation
of specifc social and economic groups. A pioneering study by Jensen
(2007) to measure the impact of access to information and communications
technologies (ICT) identifed the positive effect that mobile telephony has
had on fshing communities in southern India by reducing transaction
costs. Several other studies have assessed the contribution of mobile
telephony to growth (Waverman and others, 2005; Qiang, 2010; Katz,
2010a), employment (Katz, 2010b) and productivity (Waverman, 2009;
García Zaballos, 2012, Waverman and others, 2005).
Mobile telephony is not new. The frst real attempts to establish a
mobile telecommunications network date back to the Second World War,
2

1
Ernesto Flores-Roux is a professor at the Center for Economics Research and Teaching
(CIDE) in Mexico City.
2
The first network came into operation in Saint Louis, Missouri, in 1946. The hands-off
concept (the ability to move from one cell to another without interrupting a call) was
proposed one year later by Bell Labs.
ECLAC 136
but it was not until 1978 that Bell Labs launched the frst test network
in Chicago, for which AT&T received a commercial use license in 1982.
Around that same time, other countries, most with high per capita income,
also installed mobile networks.
3
However, early telephony was very different from today’s. Due
to the cost of service for users —the price of terminal equipment as well
as user rates— the market was very restricted. In other words, it was a
luxury service. It is worth remembering that in the early 1980s, AT&T
conducted an in-depth study of this market, seeking to estimate how many
mobile telephones would be in use at the turn of the century. That study
highlighted several problems with the product: the terminals were large
and heavy, the batteries ran out quickly, network coverage was patchy and
the price was exorbitant. Accordingly, the conclusion was that the total
market would amount to approximately 900,000 terminals (The Economist,
1999). If these assumptions and the conclusion drawn from them had been
accurate, the decision made by AT&T would have been the right one: in
1984, when its local and long-distance services were split up, the company
spun off its mobile operations to what would come to be known as the Baby
Bells. Ten years later, when it became apparent that the mobile telephony
market would be larger by several orders of magnitude than originally
estimated, AT&T returned to the market by acquiring McCaw Cellular
Communications, Inc., completing the operation in 1995 and establishing
AT&T Wireless (http://www.corp.att.com/history/milestones.html).
By then, it was already clear that mobile telephony would be a
market with considerable potential, but the role that it would play in
making voice services universally available outside the developed world
was not yet understood. Over the past 20 years, virtually every country in
the world has experienced a sweeping social and economic transformation
driven by mobile telecommunications. There have been several waves,
from the early days when mobile telecommunications was a service for a
small segment of the population, followed by the widespread roll-out of
voice services, to the explosive growth of data services today.
This chapter describes some of the road traveled since mobile
services began to be used on a large scale and discusses some of the causes
of this phenomenon. Contrary to what was suggested in the literature in
the early 1990s by authors such as Mueller (1998), the penetration of ICT,
particularly mobile services, is not associated only with wealth. There
is virtually no correlation between GDP per  capita and time in service
in Latin America: in other words, the decision to launch mobile data
3
The first 1G (first generation) network was launched in the Nordic countries (Sweden,
Denmark, Finland and Norway) in 1981 by NMT (Nordic Mobile Telephone) and offered
full roaming capabilities.
Broadband in Latin America: Beyond Connectivity 137
networks does not seem to be related to income levels. This chapter argues
that the launch of networks did much to speed growth of the service. An
important factor behind the faster penetration rate is the amount of time
that networks have been in place: an earlier network launch guarantees
earlier adoption, regardless of how wealthy a country is or how developed
the preceding technology was.
B. Status of mobile telephony in Latin America
As of late 1995, Latin America
4
had only 3.6  million mobile terminals
in operation,
5
for a penetration rate of 0.77%. The United States and the
European Union had 33.8  million terminals (12.7%) and 22.1  million
terminals (4.6%), respectively, with penetration rates sixteen and six times
greater than in Latin America. Fifteen years later, the gap has closed. In
fact, there are more mobile telephones per 100 inhabitants in Latin America
than there are in the United States (see fgure V.1). As may be seen in the
fgure, the convergence in penetration rates is a phenomenon of the past
six years.
Figure V.1
MOBILE TELEPHONY PENETRATION RATE IN THE UNITED STATES AND
THE EUROPEAN UNION COMPARED WITH LATIN AMERICA
0
2
4
6
8
10
12
14
16
18
1995 1996 1997 1998 1999 2000 2001 2002
United States/Latin America European Union/Latin America
4
In this chapter, references to Latin America include Argentina, the Bolivarian Republic
of Venezuela, Brazil, Chile, Colombia, Costa Rica, the Dominican Republic, Ecuador,
El Salvador, Guatemala, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru, the
Plurinational State of Bolivia and Uruguay. Cuba has a mobile telephony penetration rate
of 14.1%, with a subscriber rate of 12.7%. That country’s data are not included in this
analysis so as to prevent findings that cannot be generalized to the region as a whole.
5
All figures on market size (penetration, access, and number of subscribers) from the year
2000 and after were taken from Wireless Intelligence. For statistics prior to that year, the
figures reported by the International Telecommunication Union (ITU) were used.
ECLAC 138
Figure V.1 (concluded)
United States/Latin America European Union/Latin America
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
Source: ITU, Wireless Intelligence.
Penetration rates above 100% are generally interpreted to mean
that mobile telephony use has become universal in a region where voice
services were a luxury enjoyed by the few just over a decade ago. However,
there is a problem with the traditional measure of penetration: this
indicator is increasingly divergent from the real indicator of how universal
access has become because it measures the total number of connections,
not the number of people who own a mobile telephone (see box V.1).
Today, slightly more than half of the population uses mobile services, a
fgure whose importance should not be played down as it indicates great
progress towards achieving universal access, but it is also an indication of
the work yet to be done.
Box V.1
SERVICE PENETRATION AND UNIQUE SUBSCRIBERS
Penetration, defned as the number of connections per 100 inhabitants, has
been the traditional measure used to quantify the use of telecommunications
services. However, it has been recognized for some time that this fgure
does not accurately refect the number of people who actually subscribe to
the service. Anecdotal evidence suggests that many people have more than
one mobile line, a number of connections reported by frms or regulators are
not active, and there are many machine-to-machine (M2M) connections, so
interpreting penetration as the number of people per 100 inhabitants who
have a mobile telephone overstates the real number of unique subscribers.
But just how infated is this fgure? In October 2012, the association that
represents the interests of mobile operators worldwide (GSMA) published a
frst report addressing the issue on a global scale, based on feld research and
statistical models (Gillet, 2012). These were the fndings: in June 2012, there
were 6.6 billion lines worldwide excluding M2M, but 10% of connections were
not active, reducing the number to 5.9 billion. With every consumer using an
average of 1.85 SIM cards, this brings the total number of unique subscribers
in the world to 3.2  billion. Accordingly, the global penetration rate was just
44%, which suggests that not only is there still considerable growth potential,
but also that service coverage is far from being truly universal
Broadband in Latin America: Beyond Connectivity 139
Box V.1 (concluded)
Global mobile telephony penetration, June 2012
(Connections per 100 inhabitants)
75.2
51.2
69.5
73.7
38.4
31.3
89.3
44.0
134.9
109.5
106.0 106.0
83.3
64.1
Europe Latin America
and the Caribbean
United States
and Canada
Oceania Asia Africa
Unique subscribers Mulltiple SIM cards and M2M Global average
According to GSMA, as of the same date, 51.2% of the population of
Latin America and the Caribbean had a mobile telephone, with the fgure
running as low as 37.6% in Mexico to as high as 66% in El Salvador.
Latin America: mobile telephony penetration, June 2012
(Connections per 100 inhabitants)
12.7
44.0
37.6
50.1
54.8
50.1
48.2
54.5
53.6
54.2
53.0
51.6
57.7
61.6
55.6
55.6
60.8
66.0
63.6
14.1
78.4
84.3
85.8
89.1
92.8
94.7
99.1
99.2
99.5
102.7
105.8
110.0
127.3
131.3
139.3
142.3
143.6
145.4
Cuba
Bolivia (Plur. State of)
Mexico
Costa Rica
Dominican Rep.
Nicaragua
Peru
Honduras
Colombia
Guatemala
Paraguay
Venezuela (Bol. Rep. of)
Ecuador
Panama
Brazil
Argentina
Uruguay
El Salvador
Chile
Unique subscribers Mulltiple SIM cards and M2M
51.3
109.5
Total average
Average unique subscribers
Source: GSMA, Wireless Intelligence (2012).
If mobile telephony has been the main vehicle for the spread of
voice services in Latin America, it can be expected that mobile broadband
will provide universal access to the global information and knowledge
network. This was the assertion made by Flores-Roux and Mariscal
(2010), who posited, based on preliminary data, the substitution and
complementarity effects of fxed and mobile broadband.
ECLAC 140
C. Mobile data networks in Latin America
The spread of mobile services is linked not only to technological progress
and steadily lower service delivery costs but also to its cost structure.
The marginal investment needed to add new consumers is not the main
cost component, unlike in the case of fxed telecommunications, which
require a single-purpose last-mile investment. Other than the sales and
distribution costs and possible subsidies for terminal equipment, the cost
attributable to each new user is very low. The cost incurred is associated
with use —which is refected in price structures and has made prepaid
or pay-per-use services possible— not with the investment in a dedicated
connection, which must be amortized.
These characteristics have enabled a steady stream of investments
to be made in setting up new networks and upgrading old ones. Although
in many cases the patterns of service introduction have correlated with
the development level of countries, this has not been the case in Latin
America. There is a strong correlation between the development of mobile
telecommunications and the wealth of the region’s countries, but launch
patterns seem to correspond not with levels of economic development but
with spectrum licensing and liberalization processes, which have stood in
the way of fully capitalizing on the potential social and economic impact
of mobile telecommunications.
This section analyses the relationship between service penetration
and wealth creation, measured by GDP per capita, as well as the launch of
mobile telecommunications networks in the region.
1. Relationship between penetration and wealth creation
One of the most common approaches in telecommunications sector
studies is to show how penetration is dependent on per  capita income,
generally defned as GDP per capita in terms of purchasing power parity
(PPP). However, the fact that penetration rates in Latin American countries
and in regions with substantially higher GDPs have converged makes this
point somewhat moot, providing as it does information that is of little
consequence and in some cases irrelevant. In fact, when analysed over
time, the correlation between penetration and GDP per  capita gradually
diminishes until virtually disappearing.
Based on the model specifed as “penetration
t
= k
0t
+ k
1t
GDP
pc,t,
”
where t represents a moment in time, and verifying the statistical
signifcance of the parameters k
0t
and k
1t
, the wealth creation factor in Latin
America was highly indicative of penetration until the middle of the last
decade. However, over the past fve years, as levels of access around the
world began to converge, this parameter lost its explanatory power, as
Broadband in Latin America: Beyond Connectivity 141
illustrated in fgure V.2. The fgure also shows how the signifcance of the
constant grew, which indicates a convergence in penetration rates.
Figure V.2
STATISTICAL SIGNIFICANCE OF THE PARAMETERS, 1995-2011
-4
-2
0
2
4
6
8
1
9
9
5
1
9
9
6
1
9
9
7
1
9
9
8
1
9
9
9
2
0
0
0
2
0
0
1
2
0
0
2
2
0
0
3
2
0
0
4
2
0
0
5
2
0
0
6
2
0
0
7
2
0
0
8
2
0
0
9
2
0
1
0
2
0
1
1
k
o
(Constant)
k
1
(GDP per capita)
Value of t where the estimated parameter is not significant
Source: Prepared by the author on the basis of data from Wireless Intelligence.
Note: Value of the statistic t for the regressions “penetration = k
0
+ k
1
GDP
pc
” from 1995 to 2011. For values
of |t|>2, the estimate of the value of the parameter is signifcant at a confdence level of 95% or greater.
Beginning in 2007, the correlation between wealth and penetration
is no longer statistically signifcant. Indeed, it falls sharply in 2004-2008,
refecting convergence towards a reasonably constant level of penetration
throughout the region, regardless of wealth.
If the hypothesis that mobile platforms will bring about universal
broadband service is accepted as true, a pattern of convergence similar to
the one seen for voice services should be expected. If so, the relationship
with income will be a time variable that will partly explain the adoption
rate but will not be a determining factor in achieving near-universal levels
of use. The following section further elaborates on these relationships.
2. Introduction of mobile broadband networks
in Latin America
Although the world’s frst mobile broadband networks (3G)
6
were installed
at the beginning of the twenty-frst century, the frst one in Latin America
6
3G or later-generation networks are those that use one of the following technologies:
EVDO, HSDPA, HSPA+ or LTE (networks that use the latter are considered to be 4G).
ECLAC 142
did not appear until the frst half of 2004, in Guatemala.
7
Not many latest-
generation networks were launched in the following three years; by the
frst half of 2007, only six Latin American countries had these services,
which were provided by a total of nine networks (see fgure V.3). However,
over the next two years, 3G networks came online in nearly every country
in the region. By mid- 2009 there were 47 networks in 17 countries (Cuba
and Costa Rica
8
did not have 3G services). Since that time, spectrum
licensing has increased and mobile broadband networks have been set
up by nearly all incumbent operators in the region, as well as some new
entrants, so that every country in the region (except Cuba) now has these
services, which are provided by 78 different networks.
9
Figure V.3
NUMBER OF 3G NETWORKS IN OPERATION IN LATIN AMERICA
1 1
2
5 5
6
9
20
30
45
47
50
51
54
56
73
78
1 1 2 4 4 5 6 10 15 16 17 17 17 17 17 18 18
1
s
t

h
a
l
f
-
2
0
0
4
2
n
d

h
a
l
f
-
2
0
0
4
1
s
t

h
a
l
f
-
2
0
0
5
2
n
d

h
a
l
f
-
2
0
0
5
1
s
t

h
a
l
f
-
2
0
0
6
2
n
d

h
a
l
f
-
2
0
0
6
1
s
t

h
a
l
f
-
2
0
0
7
2
n
d

h
a
l
f
-
2
0
0
7
1
s
t

h
a
l
f
-
2
0
0
8
2
n
d

h
a
l
f
-
2
0
0
8
1
s
t

h
a
l
f
-
2
0
0
9
2
n
d

h
a
l
f
-
2
0
0
9
1
s
t

h
a
l
f
-
2
0
1
0
2
n
d

h
a
l
f
-
2
0
1
0
1
s
t

h
a
l
f
-
2
0
1
1
2
n
d

h
a
l
f
-
2
0
1
1
1
s
t

h
a
l
f
-
2
0
1
2
Countries with
a 3G network
Source: Wireless Intelligence.
Note: EVDO, HSDPA, HSPA+ and LTE technologies.
There have been some technological discontinuities in mobile
telephony. These have been so signifcant that their names have come to form
part of everyday language: 1G, 2G, 3G and now 4G. Mobile data became a
reality with the step from 2G to 3G, a process that may be compared to the
transition from 1G to 2G. The adoption of digital technologies (commonly
7
BellSouth Guatemala began to operate a CDMA2000 1xEV-DO network in May 2004. This
company was subsequently acquired by Telefónica Móviles, S.A., which is now part of
Telefónica S.A.
8
As will be explained later, the development of Costa Rica’s telecommunications sector has
been a typical of sector development in the rest of the region.
9
Different networks are defined as the networks in different countries that are run by the
operators that provide services in the region, as well as when they have more than one
technology deployed.
Broadband in Latin America: Beyond Connectivity 143
known as 2G), which replaced analog technologies, began in Latin America
in the mid-1990s and accelerated dramatically with the arrival of GSM
technology in the region, which replaced the technologies that had initially
been deployed (TDMA and CDMA).
The time analysis shows that there do not seem to be any signifcant
differences between the advent of 3G networks in Latin America and the
transition from analog (1G) to digital (2G) telephony. Mobile broadband
was slightly slower to take off, but the pace at which the networks were
deployed subsequently accelerated (see fgure V.4).
10
Figure V.4
COMPARISON BETWEEN NETWORK LAUNCHES
(Number of networks in operation)
0
20
40
60
80
100
120
123456789
1
0
1
1
1
2
1
3
1
4
1
5
1
6
1
7
1
8
1
9
2
0
2
1
2
2
2
3
2
4
2
5
2
6
2
7
2
8
2
9
3
0
3
1
3
2
3
3
1 = Second half of 1994
Six-month periods
1 = First half of 2004
Digitalization
3G Networks
Digitalization
3G Networks
Source: Wireless Intelligence, 4G Americas, press archives.
The introduction of new networks offering Internet and other
broadband services was clearly refected in the number of users. From
just 2.4  million connections
11
in the region in the frst half of 2007, the
fgure had jumped to 117  million connections in 2012. From the second
half of 2009 (when the technology had already become nearly universally
available) to June 2012, the annual regional growth rate was an impressive
92% (see fgure V.5).
10
The figure reflects the consolidation of companies between 2000 and 2005, as well as the
migration from TDMA and CDMA to GSM. There are still some companies in the region
that keep more than one 2G or 2.5G network in operation, which explains why there are
over 100 different networks in 19 countries.
11
This includes the total number of connections to 3G networks, regardless of whether these
are users of mobile data services or just voice services delivered on this platform.
ECLAC 144
Figure V.5
GROWTH IN THE NUMBER OF 3G USERS
(Number of connections at end-quarter; in millions)
0.0 0.1 0.2 0.3 0.4 0.6 0.9 1.2 1.5 1.8 2.1 2.4 2.8
4.2
5.4
7.1
10
13
16
19
23
32
38
43
51
59
68
79
92
104
117
Q
4

2
0
0
4
Q
1

2
0
0
5
Q
1

2
0
0
6
Q
1

2
0
0
8
Q
1

2
0
0
9
Q
1

2
0
1
0
Q
3

2
0
0
8
Q
3

2
0
0
9
Q
2

2
0
0
5
Q
3

2
0
0
5
Q
4

2
0
0
5
Q
2

2
0
0
6
Q
3

2
0
0
6
Q
1

2
0
0
7
Q
4

2
0
0
6
Q
2

2
0
0
7
Q
3

2
0
0
7
Q
4

2
0
0
7
Q
2

2
0
0
8
Q
4

2
0
0
8
Q
2

2
0
0
9
Q
4

2
0
0
9
Q
2

2
0
1
0
Q
3

2
0
1
0
Q
4

2
0
1
0
Q
2

2
0
1
1
Q
3

2
0
1
1
Q
1

2
0
1
1
Q
1

2
0
1
2
Q
4

2
0
1
1
Q
2

2
0
1
2
Source: Wireless Intelligence.
There are several reasons for this surge: more appropriation, lower
service delivery costs (owing both to the maturation of the technology and
greater network use, as well as the realization of economies of scale) and a
substantial reduction in the cost of terminal equipment (which is a major
entry barrier for consumers). However, the rate of adoption between the
different countries is still very uneven (see fgure V.6), and this should
have an explanation beyond levels of wealth.
Figure V.6
PENETRATION OF 3G SERVICES AS OF JUNE 2012
(Connections per 100 inhabitants)
5.3
5.8
7.1
7.3
7.3
7.8
8.0
8.6
10.8
12.0
12.3
20.2
23.0
23.4
26.3
26.6
29.5
29.7
Paraguay
Bolivia (Plur. State of)
Nicaragua
Panama
Guatemala
Honduras
El Salvador
Dominican Rep.
Colombia
Ecuador
Peru
Mexico
Argentina
Chile
Uruguay
Costa Rica
Venezuela (Bol. Rep. of)
Brazil
Source: Wireless Intelligence.
Broadband in Latin America: Beyond Connectivity 145
The correlation between GDP per  capita and penetration of 3G
services has been relatively weak in Latin America. Figure V.7 tracks the
correlation of penetration with both GDP per  capita and the amount of
time that the service has been available (measured in quarters, which is the
minimum unit of time for which data are available for the entire region).
As shown in the fgure, it was not until late 2010 that the correlation with
GDP surpassed the correlation with time in service.
In other words, at any moment t after mobile data networks have
been installed, the rate of adoption has been tied both to wealth and to the
time in which the service has been available. Excluding Costa Rica, which
has been an atypical case,
12
the data point to a strong correlation between
penetration and the amount of time in service.
Figure V.7
CORRELATION OF PENETRATION WITH TIME IN SERVICE AND GDP PER CAPITA
(Correlation factors)
0.1
0
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Q
1

2
0
0
6
Q
2

2
0
0
6
Q
3

2
0
0
6
Q
4

2
0
0
6
Q
1

2
0
0
7
Q
2

2
0
0
7
Q
3

2
0
0
7
Q
4

2
0
0
7
Q
1

2
0
0
8
Q
2

2
0
0
8
Q
3

2
0
0
8
Q
4

2
0
0
8
Q
1

2
0
0
9
Q
2

2
0
0
9
Q
3

2
0
0
9
Q
4

2
0
0
9
Q
1

2
0
1
0
Q
2

2
0
1
0
Q
3

2
0
1
0
Q
4

2
0
1
0
Q
1

2
0
1
1
Q
2

2
0
1
1
Q
3

2
0
1
1
Q
4

2
0
1
1
Q
1

2
0
1
2
Q
2

2
0
1
2
Penetration and
time in service
Penetration and
GDP per capita
Source: Prepared by the author on the basis of data from Wireless Intelligence.
Note: For each quarter, the set of data observed at that moment for all countries in Latin America that
already had 3G networks available was used.
To conclude the correlation analysis, it is important to note that there
is virtually no correlation between GDP per capita and time in service. In
other words, the decision to launch mobile data networks does not seem to
be related to levels of income (see fgure V.8).
12
The State-owned electricity company, ICE, had a monopoly on the delivery of mobile services
—the first 3G network (WCDMA) was launched under the “Kölbi” brand in December
2009— until the arrival of Telefónica (Movistar) and América Móvil (Claro), which launched
their networks in November 2011. During the two-year period in which ICE was the only
provider of 3G services, the penetration rate climbed to just over 6%. By late 2011, it had
already increased to 11%, rising to 26.6% just six months later. By late 2012, it was over 35%.
ECLAC 146
Figure V.8
RELATIONSHIP BETWEEN GDP PER CAPITA AND TIME
IN SERVICE AS OF JUNE 2012
(Correlation factors)
0
5
10
15
20
25
30
35
5 000 10 000 15 000 20 000
Costa Rica
GDP per capita (PPP)
Q
u
a
r
t
e
r
s

i
n

s
e
r
v
i
c
e
0
Source: Prepared by the author on the basis of data from Wireless Intelligence and the World Bank.
Note: Each dot represents a country in the region as of June 2012.
Lastly, in an analysis of how penetration increases according
to network age, the longer the network has been in service the more
the adoption rate raises. The speed, or rate, of adoption (the number
of percentage points of penetration that are added to the baseline in a
given period of time) went from 0.15 percentage points of penetration per
quarter initially to 0.4 percentage points in the third year. By the ffth year,
the rate had climbed to over one percentage point (see fgure V.9). At the
end of the sixth year, the rate slows somewhat but rebounds one year later.
This is largely explained by a reduction in the number of countries in the
considered universe starting in the twenty-fourth quarter; by 2012, there
were only three countries that had at least one 3G network in operation for
over seven years.
Based on the foregoing, it can concluded that the growth of mobile
broadband has followed a pattern similar to that of the transition from
analog to digital mobile telephony. Adoption has been correlated with
GDP per capita, but another important factor has been the length of time
in which latest-generation networks have been available. Moreover, there
has not been any correlation between the installation of networks and the
wealth of individual countries in the region.
Broadband in Latin America: Beyond Connectivity 147
Figure V.9
AVERAGE QUARTERLY INCREASE IN PENETRATION, BY NETWORK AGE
(Percentage points)
0
0.5
1.0
1.5
2.0
0 5 10 15 20 25 30
Number of quarters with 3G network in operation
Source: Prepared by the author on the basis of data from Wireless Intelligence.
D. A supporting theoretical model
and its implications
To test the hypothesis of the importance of time for penetration, a simple
adjustment model was constructed for the data observed in Latin America
from the fourth quarter of 2004 to June 2012. Subsequently, the model
was used to estimate the average cost of a delayed roll-out of mobile
broadband networks.
1. The adjustment model
The model that was developed seeks to explain the penetration of mobile
data services in terms of GDP per capita (GDP
pc
, measured in PPP dollars
per quarter) and time in service of the network (t, measured in quarters)
(model 1).
13
The fact that the rate of adoption increases over time means
that the relationship between penetration and time is not linear; therefore,
a quadratic model starting at the origin was used.
14

13
For an explanation and rationale of the models used, see the chapter “Smoothing and
extrapolation of time series” in Pindyck and Rubinfeld (1998).
14
The model can be marginally improved by using an estimate of parameters for an
equation in an adoption model, which normally follows an S-curve, also described in
Pindyck and Rubinfeld (1998). Given that the beginning of the adoption curve is being
analysed and the turning point has not likely been reached yet, using a quadratic model
as a proxy for an S-curve is acceptable. The ordinate was taken at the origin (i.e., the
linear term kit was eliminated) since no shift has been observed in Latin America: in all
cases, the penetration rate begins to climb in the first quarter in which the network is
brought online. The data confirm this, inasmuch as the variable t, with the inclusion of t
2
,
is not statistically significant.
ECLAC 148
Three variants of the model were considered. The frst includes a
country indicator as a fxed effect (FE
country
) to incorporate the existence
of differences among countries (model  2). The second adds the year of
observation as a fxed effect, in order to incorporate fuctuations in the
economy of the region (model 3). Lastly, a model was studied taking as
a fxed effect the (total) mobile penetration in each country at the time of
launch (MP
t0
). Models 2 and 4 are statistically equivalent.
15
penetration = k
0
+ k
1
GDP
pc
+ k
2
t
2
(1)
penetration = k
0
+ k
1
GDPpc+ k
2
t
2
+ FE
country
(2)
penetration = k
0
+ k
1
GDP
pc
+ k
2
t
2
+ FE
country
+ FE
year
(3)
penetration = k
0
+ k
1
GDP
pc
+ k
2
t
2
+ MP
t0
(4)
The models were adjusted to the database, excluding Costa Rica
(where, as previously indicated, the pattern has differed from the rest
of the region). Cuba is not included either because it does not have a 3G
network in operation and so cannot be represented in the model, given
that t > 0 is required.
For Latin America, 376 observations were available for different
points in time from the fourth quarter of 2004 to June 2012, grouped into 17
data sets, each one representing a country. The number of observations per
country varies depending on when mobile data services were launched.
Across all of the models considered, the results are consistent and
demonstrate that time in service does in fact have a signifcant impact on
penetration growth, regardless of wealth or legacy technology penetration
rate. All estimated parameters are highly signifcant, and the critical value
of the statistic F is near 0 (see tables V.1 and V.2).
How can these results be interpreted? In the period encompassed by
the data, the unweighted linear average of quarterly GDP per capita for all
countries in Latin America was 2,160 PPP dollars, with compound annual
growth of 3.4%, according to data from the World Bank. Using models 2
and 3, it can be estimated that 68% of the growth in the penetration of
mobile data services as of the end of the frst year came from GDP growth
and the rest can reasonably be attributed to network time in service. For
the second, third and fourth years, this fgure drops to 35%, 19% and 12%,
inverting the ratio. In other words, in the frst year, the impact of GDP is
approximately double the impact of network time in service, but by the
fourth year the impact of time in service is 7.5 times greater than GDP.
15
The two models are equivalent inasmuch as MP
t0
ends up functioning as a fixed effect
because only the value at t
0
was considered.
Broadband in Latin America: Beyond Connectivity 149
Table V.1
MODEL PARAMETERS
Coeffcient Standard error T-statistic P>ItI
Model 1
Constant -0.031 119 0.004 668 -6.667 0.00
GDP
PC
0.000 017 0.000 001 8.754 0.00
t
2
0.000 202 0.000 008 23.992 0.00
Model 2
Constant -0.281 923 0.038 878 -7.252 0.00
GDP
PC
0.000 092 0.000 011 8.223 0.00
t
2
0.000 197 0.027 949 23.539 0.00
Model 3
Constant -0.260 783 0.055 592 -4.691 0.00
GDP
PC
0.000 108 0.000 012 8.803 0.00
t
2
0.000 232 0.000 020 11.646 0.00
Source: Prepared by the author.
Table V.2
REGRESSION AND VARIANCE ANALYSIS STATISTICS
Model 1 Model 2 Model 3
R
2
0.665 0.838 0.868
Adjusted R
2
0.663 0.829 0.858
F 370.01 102.35 88.40
P>F 0.00 0.00 0.00
Number of groups 17 17
Number of observations 376 376
Source: Prepared by the author.
2. Implications of the model
The analysis based on the model reveals two basic points. A signifcant
portion of growth in the penetration rate is not explained by GDP
per capita, but rather by network time in service. In addition, the portion
of penetration that is explained by time in service increases as the time in
service lengthens, which is to be expected during the period of adoption
of a technology. In other words, an earlier launch guarantees earlier and
swifter adoption, regardless of how wealthy a country is or how developed
the preceding technology was.
Under this assumption and given that telecommunications, as a
general purpose technology, has a major spillover effect on the economy,
ECLAC 150
the approximate cost in terms of well-being that countries incur by
delaying decisions to launch telecommunications networks can be
estimated. This is important because, among other reasons, Latin America
is now engaged in launching new latest-generation networks (LTE), and
the region’s governments are making additional spectrum available to
the market (primarily, the digital dividend, or frequencies in the 700 MHz
band, as well as the 2.6 GHZ bands). Furthermore, all of the countries are
transitioning to digital television and working ton the analog switch-off,
which will free up more spectrum (a potential “second digital dividend” in
the 600 MHz band). Deferring or deliberately postponing these processes
comes at a high cost, as will be discussed later.
Based on the literature, the most conservative estimate of the
impact of broadband penetration on GDP is used in this chapter. In Katz
(2010a), it was estimated that 10 points of broadband penetration adds
0.17 percentage points to GDP growth (see also the article by Raúl Katz
herein). Other authors, such as Qiang (2009) and García Zaballo (2012),
argue that the impact may be much greater. The former puts the impact in
developing countries at 1.38 percentage points, while the latter estimates
an average increase of 3.2% in GDP and an increase in productivity of 2.6
percentage points.
Using the fgure published by Katz (2010a), it can be estimated that a
region-wide delay of one year in the launch of mobile broadband networks
represents a loss in spillover benefts for the economy equivalent to
66 billion 2012 PPP dollars in the sixth year (see fgure V.10). Considering
that regional GDP in 2011 was approximately US$ 6.88 trillion, the delay in
launch is equivalent to 1% of the wealth that is generated.
Figure V.10
COST IN ECONOMIC WELL-BEING OF A ONE-YEAR DELAY IN THE LAUNCH
OF HIGH-SPEED NETWORKS
(Impact on GDP in year 6 – millions of 2012 PPP dollars)
21 690
17 560
8 294
4 228
3 561
3 420
2 580
1 012
787
520
543
517
366
330
251
225
118
66 002
Brazil
Mexico
Argentina
Colombia
Venezuela (Bol. Rep. of)
Chile
Peru
Ecuador
Dominican Rep.
Guatemala
Uruguay
Panama
El Salvador
Bolivia (Plur. State of)
Paraguay
Honduras
Nicaragua
Total Latin America
Source: Prepared by the author.
Broadband in Latin America: Beyond Connectivity 151
To make it even more dramatic, the cost of delaying launch
by just one quarter can be estimated. In the sixth year, regional GDP
would be negatively impacted by between 0.15 percentage points and
0.26 percentage points, equivalent to an average of US$ 27 per  capita in
2012 PPP terms (see fgure V.11). Regionally, the value of this unrealized
spillover approaches 15 billion PPP dollars.
Figure V.11
COST IN ECONOMIC WELL-BEING OF A ONE-QUARTER DELAY
IN THE LAUNCH OF HIGH-SPEED NETWORKS
(Impact on GDP in year 6 – 2012 PPP dollars)
0.15
0.16
0.16
0.17
0.17
0.18
0.19
0.20
0.20
0.20
0.22
0.22
0.23
0.24
0.24
0.26
0.26
Nicaragua
Honduras
Bolivia (Plur. State of)
Guatemala
Paraguay
El Salvador
Ecuador
Dominican Rep.
Peru
Colombia
Brazil
Venezuela (Bol. Rep. of)
Panama
Mexico
Uruguay
Chile
Argentina
6
8
9
10
11
17
20
22
25
25
31
34
40
42
44
54
56
Impact on GDP
(percentage)
Impact on GDP per capita
(2012 PPP dollars)
Source: Prepared by the author.
E. Conclusions
A number of conclusions with direct public policy implications can be
drawn from these analyses.
(i) In the medium term, mobile broadband penetration will likely
vary from country to country, as is presently the case with mobile
telephony. However, per capita income will be less and less of a
determining factor in penetration. Given that penetration is the
principal measure of the spread of access, its evolution should
be a frst-order concern of public policymakers.
16

16
This chapter does not address the issue of use, but it is plausible to assume that average
income is directly related to intensity of use. This variable should be the second item of
concern for the authorities, since spillover benefits are achieved primarily through actual
use of technology, not just mere access to it (see the chapter herein by Galperín and Katz
on the demand gap).
ECLAC 152
(ii) Because penetration and the rate of adoption increase as
networks mature, earlier launches make it possible to reach the
saturation point sooner.
(iii) Penetration has a direct positive impact on the well-being
of the population, so earlier launches translate into greater
wealth creation.
In general, an early launch is associated with efforts to prevent
delays, to the extent possible, in the processes that make it possible to
bring networks online. The main barriers are spectrum licensing and the
processing of concessions, licenses and operating permits. However, there
are other barriers, including those associated with obtaining construction
permits, testing and standardization, which result in delays in network
roll-out and expansion. As suggested earlier, this has a substantial
quantifable impact on economic development and well-being.
Accordingly, the region’s governments should focus a good part of
their efforts on eliminating barriers so these services, as well as new services
to be launched in the future as technology evolves, can be provided.
The possibility of applying instruments to accelerate network roll-
out should also be considered. There are several approaches to this, but
spectrum allocation contract requirements have proven effective, as have
fscal incentives tied to investment and roll-out. Lastly, it is important
to note that national broadband plans have virtually ignored mobile
broadband. Inasmuch as mobile telecommunications are among the
strongest drivers of the spread of broadband service, actions to promote its
development should be at the core of any public policy in the sector.
The amount of time that a network has been available is a basic
indicator of the total penetration achieved in a country. Any unwarranted
delay comes at a real cost to the economy and to the well-being of the
population. Since much of the decision to launch services is subject to
government actions, the authorities control a strategic variable. There are
not many variables over which the authorities do have control, but this is
one and it can be used to further social and economic development. Failing
to take action and allowing delays to mount will work against efforts to
close the digital divide in Latin America.
Broadband in Latin America: Beyond Connectivity 153
Bibliography
Flores-Roux, E. M. and J. Mariscal (2010), “Oportunidades y desafíos de la banda
ancha móvil”, Acelerando la revolución digital: banda ancha para América Latina y
el Caribe, V. Jordán, W. Peres and H. Galperin (eds.), Santiago, Chile, Economic
Commission for Latin America and the Caribbean (ECLAC).
García-Zaballos, A. and G. A. Truitt Nakata (2012), “Construyendo puentes, creando
oportunidades: La banda ancha como catalizados del desarrollo económico y
social en los países de América Latina y el Caribe. La visión de la industria”,
Washington, D.C., Inter-American Development Bank, citing A. García-Zaballos
and R. López-Rivas. “Control gubernamental sobre el impacto socio-económico
de la banda ancha en los países ALC”, working paper.
Gillet, J. (2012), “Global mobile penetration – subscribers versus connections”,
Wireless Intelligence/GSMA, October.
Jensen, R. (2007),“The Digital Provide: Information (Technology), Market
Performance, and Welfare in the South Indian Fisheries Sector”, The Quarterly
Journal of Economics, vol. 122, No. 3.
Katz, R. L. (2010a), “La contribución de la banda ancha al desarrollo económico”,
Acelerando la revolución digital: banda ancha para América Latina y el Caribe,
V. Jordán, W. Peres and H. Galperín (eds.), Santiago, Chile, Economic
Commission for Latin America and the Caribbean (ECLAC).
Katz, R. L. (2010b), “The impact of broadband on the economy; Research to date and
policy issues”, paper presented at the 10th Global Symposium for Regulators
“Enabling Tomorrow’s Digital World”, Dakar, Senegal.
Mueller, M. L. (1997), Universal Service: Interconnection, Competition, and Monopoly in
the Making of American Telecommunications, AEI Series on Telecommunications
Deregulation, MIT Press.
Pindyck, R. S. and D. K. Rubinfeld (1998), Econometric Models and Economic Forecasts,
Irwin/McGraw-Hill.
Qiang, C. Z. (2010), “Broadband infrastructure investment in stimulus packages:
Relevance for developing countries”, Info, vol. 12, No. 2.
Qiang, C. Z. and C. M. Rossotto (2009), “Economic impacts of broadband on growth:
A simultaneous approach”, Telecommunications Policy, No. 33.
The Economist (1999), “Cutting the cord”, 7 October [online] http://www.economist.
com/node/246152.
Waverman, L., M. Meschi and M. Fuss (2005), “The impact of telecoms on economic
growth in developing countries”, Vodafone Policy Paper Series, No. 2, London.
Waverman, L. and others (2009). Economic Impact of Broadband: An Empirical
Study, LECG, London.
Wireless Intelligence [online] www.wirelessintelligence.com.
World Bank [online] www.worldbank.org and data.worldbank.org.
Chapter VI
Cloud computing, structural change
and job creation in SMEs
Andrea Colciago and Federico Etro
1
A. Introduction
Structural change and technology adoption are key aspects of economic
development in emerging economies. This chapter studies them by
looking at a particular contemporary experience: the adoption of a new
IT innovation, and by focusing on the leading emerging markets of
Latin America.
Cloud computing is an Internet-based general purpose technology
that is spreading at the global level with an important economic impact
in both developed and developing countries. Through this technology,
information can be stored in servers and provided online as a service
to business custumers on a pay-as-you-go basis. The provision of this
technology is emerging in a rather competitive environment, with a
number of leading players offering cloud solutions with aggressive pricing
(Fershtman and Gandal, 2012). Its adoption allows frms to avoid large up-
front costs for hardware and software and turn part of them into variable
1
Andrea Colciago is from the Department of Economics of the University of Milano
Bicocca. Federico Etro is from the Department of Economics of the University Cà Foscari
in Venice.
ECLAC 156
costs to be optimized at the margin. As a result they will be able to adjust
their demand for IT according to their needs. The reduction in required
up-front investment will be crucial for business creation and for the
growth of young frms which typically face stringent fnancial constraints
and thus cannot engage in large initial investments. Moreover, the new
technology is likely to promote new investment in R&D for applications
developed in the cloud (Borek and others, 2012).
Emerging countries, including Latin American ones, are expected
to beneft a lot from this revolution as it could close the information and
technology gap with more advanced developed countries (De Oliveira and
OgasawaraIs, 2012).
2
This chapter evaluates the impact of the adoption of cloud
computing on output, employment and business creation, focusing on
selected countries —in particular, Latin American countries— as well as
the United States and European Union countries for reference purposes.
The authors used a macroeconomic model with Endogenous Market
Structures (EMSs) in the spirit of Etro and Colciago (2010), Colciago and
Rossi (2011) and Bilbiie and others (2012). The analysis goes beyond the
preliminary attempt by Etro (2009a) at estimating the impact of cloud
computing in the European Union because:
(i) It employs a more advanced theoretical model augmented with
a job matching model that can more realistically reproduce fows
in and out of employment and with a separate (competitive)
sector providing cloud services to the production sector;
(ii) It revises the calibration accordingly, focusing on a global
perspective and in particular not only on the European Union
but also on the United States and on selected emerging countries
from Latin America, namely Brazil and Argentina; and
(iii) It provides a different conceptual experiment focused on
extrapolating the medium- and long-run impact of the adoption
of the new technology on aggregate variables so as to emphasize
the potential net gain from the new technology.
The model is characterized by three industries: the fnal good one,
the industry producing the physical IT stock, and the industry providing
IT maintainance and development services. The fnal good industry
features many sectors, where the dynamics of the number of market
competitors is endogenous. Firms face a cost of market entry which they
will decide they can afford only if it is compensated by the expectation
2
The literature on structural change and innovation has emphasized the importance of
structural change toward technology-intensive industries for emerging countries (for
instance, see Cimoli and others, 2011, on Brazil).
Broadband in Latin America: Beyond Connectivity 157
of future profts. Firms produce fnal goods using labour and physical
capital. The stock of capital takes the form of IT hardware which the frm
has to install and maintain over time. The industry producing IT adopts
physical capital as the only input, while in the industry providing IT
maintainance and development services the input is labour. The fraction
of workers employed in the latter industry is defned as IT employment.
The labour market is characterized by frictions following the literature
on job search and matching (Mortensen and Pissarides, 1994; Merz, 1995;
Andolfatto, 1996; Pissarides, 2000). In the fnal sector, both new frms
and incumbent frms need to hire workers from the pool of unemployed
agents who are looking for a job and to set up a stock of IT material before
starting production. Similarly, the industry providing IT services faces
labour market frictions.
In line with the observations made above, the model counterpart of
the introduction of cloud computing will be a reduction in the installation
and maintainance cost of IT for the individual frm. IT services, hardware
and software can be outsourced by the frm. In particular, they can be
obtained on demand by the provider of cloud IT services. This reduces the
upfront sunk entry costs, since the frm no longer needs to build up a stock
of IT before starting production. Also, it reduces IT stock maintainance
costs. As a result of lower entry costs, there is a stronger incentive for
new frms to enter the market; this promotes investment and impacts the
demand faced by incumbent producers. In turn, the creation of new frms
and the growth of the existing ones promotes both competition and job
creation, affecting the unemployment rate. It is important to note that,
although maintainance and development costs decline for the individual
frm, higher IT usage promotes a long-run increase (after some periods of
reduction) in IT employment.
The quantitative results provided by the model are then translated
into empirical results using data on business creation and job creation
relative to Brazil, Argentina and, as a reference, the European Union
(Etro, 2009a) and the United States. The empirical literature supports the
view that new frms have a fundamental role for the creation of new jobs.
Haltiwanger and others (2010), on the basis of United States manufacturing
data between 1972 and 1986, estimate that 25% of annual gross job
creation is due to start-ups. Similarly, Jaimovich and Floetotto (2008)
focus on employment data at the establishment level. They estimate that
the average fraction of quarterly job gains (losses) that can be explained
by the opening (closing) of establishments is about 20%. Therefore, the
authors’ analysis of job creation derived from business creation appears
well-grounded in the macroempirical literature. The simulation shows
a substantial impact of the structural technological change associated
with the introduction of cloud computing on business and job creation: in
ECLAC 158
particular, the long-run estimates herein show the creation of 900,000 new
jobs in Brazil and 100,000 in Agentina, which can be compared with about
3 million new jobs in both the United States and the European Union.
This chapter is organized as follows. Section B reviews the literature
on the economics of cloud computing and introduces its aspects concerning
Latin American countries and the European Union and the United States.
Section C develops the basic DSGE model with IT investment. Section D
augments it with the introduction of cloud computing. Section E calibrates
the model on Brazil, Argentina and the European Union and the United
States. Section F analyzes the transition to the cloud economy, performs
the simulation and discusses the results. The conclusions are set out in
Section G.
B. What is cloud computing, and how can it affect
the global economy?
Cloud computing is defned by the United States National Institute of
Standards and Technology as “a model for enabling convenient, on-
demand network access to a shared pool of confgurable computing
resources (e.g., networks, servers, applications and services) that can be
rapidly provisioned and released with minimal management effort or
service provider interaction.” Through cloud computing, frms will be
able to rent computing power (both hardware and software) in their latest
versions and storage from a service provider, and to pay on demand, as
they already do for other inputs such as energy and electricity (see the
chapter on the subject in this book).
A new general purpose technology such as cloud computing can
exert a number of effects on the economy. First of all, cloud computing
can provide cost effciencies in the private sector; across all industries its
use is directly related to lower entry costs and therefore to the creation of
new businesses and the diffusion of what Lanvin and Passman (2008) call
e-business skills in the managerial environment, the ability to exploit new
opportunities provided by ICT and, again, to establish new businesses.
The introduction of cloud computing can also create multilateral network
effects between businesses and increased productivity within them, and
it can promote entry and innovation in all the sectors where ICT costs are
relevant and are drastically reduced by the adoption of this technology.
Second, it can provide huge cost savings and more effciency in large
areas of the public sector, including hospitals and health care (especially
to provide information and technologies in remote or poorer locations),
education (especially for e-learning and universities) and the activity of
government agencies with periodic spikes in usage. Last, beyond cost
Broadband in Latin America: Beyond Connectivity 159
effciencies, substantial positive externalities are expected from cloud
computing because of energy savings: the improvement of energy
effciency may contribute to the reduction of total carbon emissions in a
substantial way.
Recent research (Etro, 2009a) based on the work of Etro and
Colciago (2010) has provided the frst simulation in the literature of the
economic impact of the diffusion of cloud computing in Europe focusing
on incentives to promote business creation. The initial adoption of cloud
computing solutions started in the United States, followed by the countries
of the European Union, and now by some developing countries including
the more advanced Asian and Latin American ones. While the initial focus
in Etro (2009a) was on Europe, it is clear that similar ideas and results
apply to other countries and also to developing countries, where there is
equally wide attention being paid to the diffusion of this new technology.
For instance, see Kuyucu (2011) on Turkey or Center of Information
Development (2010) on China.
3

This chapter focuses on Latin American countries such as Brazil
and Argentina, where key players are already active. The frst mover in
the feld, Amazon, has recently started operations in two data centres
in São Paulo. The entry of Amazon Web Services in the Latin American
cloud space is of crucial importance to the region, but other players are
also building data centres or starting to plan operations in Brazil and
other Latin American countries. Private clouds are already spread, and the
challenge is now to move to public clouds. Local players, such as Vurbia
Technologies, are already active in Argentina and elsewhere. The scenario
appears quite dynamic even if it is clearly less developed compared with
European and North American realities.
The key point of this research is that, somewhat surprisingly, a big
portion of the benefts associated with the adoption of the new technology
derives from indirect mechanisms active in non-IT sectors rather than
from the direct effciencies in the IT sector. This chapter reports on some
refnements of this study which take into consideration aspects that were
neglected in earlier work, namely the decomposition of the process of job
creation across countries and macrosectors and between job creation and
job destruction, and the role of public fnances.
Starting from conservative assumptions on the cost reduction
process associated with the diffusion of cloud computing, Etro (2009a)
3
In perspective, the Chinese market, with its wide public sector, could become a major
area of development of cloud computing. According to the China Center of Information
Development (2010), Chinese local governments are investing heavily in setting up data
centres, and the central government has designated cloud computing as a strategic new
business for the next years.
ECLAC 160
suggested that the diffusion of cloud computing could provide a positive
and substantial contribution to the annual growth rate (up to a few
decimal points), helping to create several hundreds of thousands of new
jobs every year through the development of a few hundred thousand new
small and medium-sized enterprises (SMEs) throughout the 27 European
Union member countries (EU-27). The driving mechanism behind the
positive contribution works through incentives to create new frms, and
in particular SMEs. One of the main obstacles to entry in new markets is
the high up-front costs of entry, often associated with physical (and IT)
capital spending. Cloud computing allows potential entrants to save on the
fxed costs associated with hardware/software adoption and on general
IT investment, and turns part of this capital expenditure into operating
expenditure, that is, into variable costs. This reduces the constraints
on entry and promotes business creation. The importance of such a
mechanism is well known at the policy level, especially in Europe, where
SMEs play a crucial role in the production structure, but it is quite relevant
also in emerging economies such as those of Latin America.
To evaluate the impact of cloud computing, the authors have
adopted a macroeconomic approach emphasizing the effects that
this innovation has on the cost structure of frms investing in IT (and
consequently the incentives to create and expand new business), on the
market structure and on the level of competition in their sectors, and
ultimately on the induced effects for aggregate production, employment
and other macroeconomic variables. The methodology is based on a
dynamic stochastic general equilibrium calibrated model augmented with
endogenous market structures in line with recent developments in the
macroeconomic literature.
The model follows the framework introduced by Etro and Colciago
(2010) and Colciago and Etro (2010) and recently extended to include the
dynamics of the labour market by Colciago and Rossi (2011), and it has
been augmented with a public sector producing goods and services. The
analysis goes beyond the preliminary attempt by Etro (2009a) at estimating
the impact of cloud computing in the European Union because it employs
a more advanced theoretical model augmented with a job matching model
able to reproduce fows in and out of employment in a more realistic
way and with a separate (competitive) sector providing cloud services to
the production sector. It revises the calibration accordingly, focusing on
a global perspective and, in particular, not only on the European Union
but also on the United States and on selected emerging countries in
Latin America, namely Brazil and Argentina, and it involved a different
conceptual experiment focused on extrapolating the medium- and long-
run impact of the adoption of the new technology on aggregate variables,
so as to emphasize the potential net gain from the new technology.
Broadband in Latin America: Beyond Connectivity 161
Two key factors for the impact of cloud computing are, on the
one hand, the size of the cost savings in ICT spending and, on the other
hand, the reduction of the fxed costs of production. On the frst point,
the business literature emphasizes large savings. The International
Data Corporation (IDC) (2008, 2009) estimated a total cost reduction up
to 40%-50% in the private sector, but a more prudential estimate in a
negative scenario could go down to 10%. Estimates for the public sector
are more limited, ranging between a 10% reduction in total costs in a
pessimistic scenario and of 30% in an optimistic one (but West, 2010,
suggests a range between 25% and 50% in successful cases). It should
also be kept in mind that the portion of these potential benefts that
will be translated on the private sector will also depend on the level of
competition in the provision of cloud services. As noted by Fershtman and
Gandal (2012), there are reasons to believe that competition in the feld will
be strong enough to conjecture an approximately competitive provision
of cloud services and a wide translation of the cost savings. Overall, a
conservative assumption would require keeping the estimated effective
reduction in savings at around 10%.
The estimates in Etro (2009a) suggested a permanent creation of
about 400,000 new SMEs. The largest impact was expected to occur in
the aggregate sectors of wholesale and retail trade and of real estate and
other fnancial and business activities. That exercise did show a strong
impact on the creation of new SMEs, on the magnitude of a few hundred
thousand in the whole European Union. Incidentally, this is consistent
with the conclusions of studies by IDC (2009) arguing that cloud services
could add US$ 800 billion in net new business revenues between 2009 and
2013. The revised exercise herein shows now a substantial impact of the
technological change on business and job creation at the global level: the
long-run estimates will suggest a likely creation of more than 3 million
new jobs in the European Union and in the United States, and about one
million in Brazil and Argentina.
C. The theoretical model
This section describes the main model to be used for simulations.
The structure of the model is a basic DSGE model that is standard in
the macroeconomic analysis (Kydland and Prescott, 1982; Christiano,
Eichenbaum and Evans, 2005), augmented with endogenous market
structures as opposed to perfectly competitive or monopolistically
competitive markets (Colciago and Etro, 2010, and Etro and Colciago, 2010).
4

4
See Etro (2009b) for a survey of the related literature on endogenous market structures in
general equilibrium.
ECLAC 162
The economy features a continuum of atomistic sectors, or industries, on
the unit interval. Each sector is characterized by different frms producing
a good in different varieties, using labour and IT material as input. In
turn, the sectoral goods are imperfect substitutes for each other and are
aggregated into a fnal good. The IT material is produced by a perfectly
competitive frm using physical capital as the only input.
Households use the fnal good for consumption and investment
purposes. Price competition and endogenous frms’ entry is modelled at
the sectoral level, where frms also face search and matching frictions in
hiring workers, modelled in the tradition of the literature on job search
(Mortensen and Pissarides, 1994; Pissarides, 2000). In the pre-cloud
economy they also face IT material maintainance costs.
1. Labour market and job matching
The labour market is characterized by search and matching frictions,
as in Andolfatto (1996) and Merz (1995). Firms producing in need to
post vacancies in order to hire new workers. Unemployed workers and
vacancies combine according to a CRS matching function and deliver
new hires, or matches, in each period. The matching function is assumed
to be a Cobb-Douglas one:
(1)
where refects the effciency of the matching process, is the
total number of vacancies created at time and is the unemployment
rate. The probability that a frm flls a vacancy is given by ,
while the probability to fnd a job for an unemployed worker reads as
. Firms and individuals take both probabilities as given.
Matches become productive in the same period in which they are formed.
Each frm separates exogenously from a fraction of existing workers
each period, where is the probability that a worker stays with a frm until
the next period. As a result a worker may separate from a job for two reasons:
either because the frm where the job is located exits from the market or
because the match is destroyed. Given that population is normalized to
one, the number of unemployed workers and the unemployment rate are
identical. Therefore, given labour at time as , the unemployment rate is:

and represents also the fraction of agents searching for a job.
Given this functional form, the probability of flling a vacancy can
be expressed as:
(2)
Broadband in Latin America: Beyond Connectivity 163
where , the probability of fnding a job, is defned as:
(3)
and their ratio as:
2. Households
Using the family construct of Merz (1995) a representative household can
be referred to as consisting of a continuum of individuals of mass one.
Members of the household insure each other against the risk of being
unemployed. The representative family has lifetime utility:
(4)
where is the discount factor, the variable represents
individual hours worked and is the consumption of the fnal good. The
family receives real labour income , where is the real wage,
and profts from the ownership of frms. Unemployed individuals
receive a real unemployment beneft , hence the overall beneft for the
household is . This is fnanced through lump sum taxation by
the government. The households hold the stock of physical capital, ,
which evolves according to
(5)
where is investment in capital. The household chooses how
much to save in riskless bonds, physical capital and the creation of new
frms according to standard Euler and asset pricing equations. The frst
order condition (FOC) with respect to employment, , is

(6)
here is the marginal value to the household of having one
member employed rather than unemployed and is the marginal
utility of consumption. Equation (6) indicates that the household’s shadow
value of one additional employed member (the left-hand side) has four
ECLAC 164
components: (i) the increase in utility generated by having an additional
member employed, given by the real wage expressed in utils; (ii) the
decrease in utility due to more hours dedicated to work, given by the
marginal disutility of employment; (iii) the foregone utility value of the
unemployment beneft; and (iv) the continuation utility value, given by the
contribution of a current match to next period household’s employment.
3. Technology
There are four types of frms in the economy. They are (i) the producers
of intermediate goods; (ii) the fnal good producer; (iii) the producers of IT
material; and (iv) the providers of maintanance services for IT. The fnal good
is an aggregate of a continuum of mass one of sectoral goods defned as
(7)
where denotes output of sector and is the elasticity of
substitution between any two different sectoral goods. The fnal good
producers behave competitively. In each sector , there are frms
producing differentiated goods that are aggregated into a sectoral good by
a CES aggregating function defned as
(8)
where is the production of good in sector is the
elasticity of substitution between sectoral goods. As in Etro and Colciago
(2010), the authors assume a unit elasticity of substitution between goods
belonging to different sectors. This makes it possible to realistically separate
limited substitutability at the aggregated level, and high substitutability at
the disaggregated level. Each frm in sector produces a (intermediate)
differentiated good with the following production function
(9)
where represents technology which is common across sectors
and evolves exogenously over time. The variable is frm ’s time-
workforce used for the production of the fnal good, and represents
hours per employee. In the remainder, frms in the intermediate goods
sector are referred to as producers. The variable is the amount of
IT material involved in the production process. The latter is produced by
Broadband in Latin America: Beyond Connectivity 165
a perfectly competitive frm which uses physical capital as the only input.
In each period a fow of IT, defned as , is produced with technology
(10)
where is the stock of capital in the economy and is the
productivity of the IT industry. Given perfect competition the price of IT
material is the marginal cost of production. The latter can be obtained
by proft maximization of the producer of IT services as .
Period-t real profts of a producer are defned as
(11)
where is the real wage paid by frm , represents the
number of vacancies posted at time , is the output cost of keeping a
vacancy open, is period investment in IT and is the price of a
unit of IT in terms of the fnal good. Notice that is the real price of
frm i’s output.
The term represents maintainance and development
costs of the IT stock. These services are provided by a frm which operates
in perfect competition with technology , where represents
the number of workers employed in the industry. The provider of
maintainance services also faces search costs in the labour market. It hires
workers by posting vacancies at an output cost equal to , taking as given
hours and the real wage determined in the bargaining process between
workers and frms operating in the fnal good industry.
5
Its workforce
evolves according to . The problem faced by the
provider of maintainance services can thus be written as:
Proft maximization requires
this condition equates the marginal cost of hiring a worker with
the marginal beneft. The latter is given by a discounted stream of a frm’s
expected future net earnings from the marginal worker. A maintainance
technology is assumed such that the fnal good producer must acquire
units of maintainance services for each unit of IT owned. As a
5
In this case it is indifferent for a member of the household to work in the IT sector or in the
final sector.
ECLAC 166
result, the individual demand for maintanance services is
and profts of the fnal good producer can be rewritten as
The value of a fnal good producer is the expected discounted value
of its future profts
(12)
where is the households’ stochastic
discount factor which takes into account that frms’ survival probability is
. Firms which do not exit from the market have a time , individual
workforce given by
(13)
and a stock of IT equal to
(14)
where is the depreciation rate of IT material. The unit
intersectoral elasticity of substitution implies that the nominal expenditure,
, is identical across sectors. Thus, the fnal producer’s demand for
each sectoral good is
(15)
where is the price index of sector and is the price of the
fnal good at period . Denoting with the price of good in sector
the demand faced by the producer of each variant is
(16)
where is defned as
(17)
,
Broadband in Latin America: Beyond Connectivity 167
Using (16) and (15) the individual demand of good can be written
as a function of aggregate expenditure,
(18)
4. Entry
At the beginning of each period new frms enter into sector (0,1),
while at the end of the period a fraction (0; 1) of market participants
exits from the market for exogenous reasons.
6
As a result, the number of
frms in a sector , follows the equation of motion:
(19)
where is the number of new entrants in sector at time .
Following Bilbiie and others (2012) it is assumed that new entrants at time
will only start producing at time +1 and that the probability of exit from
the market, , is independent of the period of entry and identical across
sectors. The assumption of an exogenous constant exit rate is adopted for
tractability, but it also has empirical support. Using United States annual
data on manufacturing, Lee and Mukoyama (2007) fnd that, while the
entry rate is procyclical, annual exit rates are similar across booms and
recessions. The entry process and the mode of competition within each
sector are described in detail below. Prior to entry, frms face a sunk
entry cost to be paid in order to serve the market. It is made up of
two components
The frst term represents the cost associated with regulation
and barriers to entry, which is common across sectors. It is exogenous and
expressed in units of the fnal good. The second component of the entry
cost refects instead the fact that in order to start production in the next
period new frms must set up a stock of IT. This requires an amount of
investment in IT given by .
7
If the frm exits from the market its
IT stock is lost. Firms will enter into the market up to the point where their
6
As discussed in Bilbiie and others (2012), if macroeconomic shocks are small enough is
positive in every period. New entrants finance entry on the stock market.
7
The parameter Φ is an indicator variable which takes value 0 or 1 and allows to nest the
post-cloud economy into the pre-cloud economy. As will be clear below, it takes value 1
before the introduction of cloud services and value 0 after the introduction of the cloud.
ECLAC 168
value, represented by the discounted value of their future profts, equals
the sunk entry cost .
5. Imperfect competition and job creation
Let us consider competition à la Bertrand. Each frm chooses;
; to maximize , taking as
given the price of the other frms in the sector. In a symmetric equilibrium
optimal pricing implies that the relative price chosen by frms is
(20)
where the markup over the marginal cost is given by
(21)

The latter is decreasing in accordance with the number of frms in
the sector. Further, when the markup tends to , i.e. the
traditional one under monopolistic competition. The frst order condition
(FOC) with respect to vacancies reads as
(22)
Thus, the frm sets the value of the marginal worker, , equal to
the expected cost of hiring the worker, . The FOC with respect to
delivers
(23)
Combining the latter two equations delivers the following Job
Creation Condition (JCC)
where the authors used the pricing condition to substitute for
. Since the ratio increases in the number of frms,
it follows that competition leads to a rise in the marginal cost and hence in
the equilibrium marginal revenue. For this reason the marginal revenue
product of labour (MRP), given by

,
also rises with competition. Thus, stronger competition promotes the
creation of vacancies and employment due to its positive effect on the MRP
of labour. The frm will invest in IT up to the point where
Broadband in Latin America: Beyond Connectivity 169
(24)
Increasing IT by one unit today costs . The beneft associated
with the marginal unit of IT is given by the discounted marginal revenue
product of IT net of maintainance costs, the frst term at the RHS, added
to the discounted value that the additional unit of IT will have tomorrow,
. Since IT is a stock variable, the frm is forced to look
ahead when taking decisions concerning optimal investment in IT.
6. Bargaining over wages and hours
As in Trigari (2009), bargaining takes place along two dimensions: the real
wage and the hours of work. The authors assume Nash bargaining. That is,
the frm and the worker choose the wage and the hours of work to
maximize the Nash product
(25)
where is the frm value of having an additional worker, while
is the household surplus expressed in units of consumption. The
parameter refects the parties’ relative bargaining power. The FOC with
respect to the real wage is
(26)
Using the defnition of in equation (23) and that of given by
equation (6), after some manipulations, yields the wage equation
(27)

where . The wage shares costs
and benefts associated with the match according to the parameter . The
worker is rewarded for a fraction of the frm’s revenues and savings
of hiring costs and compensated for a fraction 1- of the disutility he
or she suffers from supplying labour and the foregone unemployment
benefts. A distinguishing feature of this approach is that the wage
depends on the degree of competition in the goods market. The direct
effect of competition on the real wage is captured through the term
, which represents the share of the
MRP which goes to workers. Entry leads to an increase in the ratio
and hence in the MRP which goes to the workers. Thus, everything else
being equal, stronger competition shifts the wage curve up. This result
is similar to that in the static model by Blanchard and Giavazzi (2003),
ECLAC 170
who fnd a positive effect of competition on the real wage. The FOC with
respect to yields
(28)
Because the frm and the worker bargain simultaneously about
wages and hours, the outcome is (privately) effcient and the wage does
not play an allocational role for hours. Stronger competition leads to an
increase in hours bargained between the workers and frms for the same
reasons for which competition positively affects the wage schedule.
7. Business creation, hiring and IT policies
Let be, respectively, real profts, the number
of vacancies posted by a new frm and investment in IT. Symmetrically,
defne, respectively, individual profts and vacancies
posted by an incumbent producer. New frms and incumbent frms are
characterized by the same size, . Thus, the optimal hiring policy of
new frms, which have no initial workforce, consists of posting at time t as
many vacancies as required to hire workers. As a result
Since it has to be the case that
(29)
Hence, a new frm posts more vacancies than an incumbent
producer. For this reason, and given vacancy posting is costly, the profts
of new frms are lower than those of incumbent frms, in particular
Notice also that a new entrant must set up a stock of IT before
starting production next period. Given the IT choice is symmetric across
producers they have to invest during time as much as required to reach a
stock of IT identical to that held by incumbent producers at the end of time
, that is . The sunk entry cost for a new frm can thus be
written as
In each period the level of entry is determined endogenously to
equate the value of a new entrant, , to the entry cost
(30)
.
Broadband in Latin America: Beyond Connectivity 171
Notice that prospective new entrants have lower value than
producing frms because if they do not exit from the market before starting
production they will have to set up a workforce in their frst period
of activity. The difference in the value between a frm which is already
producing and a prospective entrant is, in fact, the discounted value of the
higher vacancy posting cost that the latter will suffer, with respect to the
former, in the frst period of activity. Formally
(31)
where is the value of a producing frm (both new frms and
incumbent frms) at time .
D. The introduction of cloud computing
As described in the introduction, the creation of cloud facilities implies
that IT services, hardware and software can be outsourced by the frm.
In particular, delivery of computing and storage capacity is provided as
a service that can be obtained on demand. This technological change is
modelled as follows. It is assumed that after the introduction of cloud
computing all producers of intermediate goods will no longer own a stock
of IT but will rent it from the provider of cloud services. The existing
stock of IT is transferred to the cloud-services provider. There are two main
consequences spreading from this modelling device. The frst one is that
maintenance and depreciation costs associated with the stock of IT will be
borne by the provider of IT services. The second one is that new entrants will
no longer need to build up a stock of IT before starting production. As a result,
up-front sunk entry costs faced by potential entrants are sharply reduced.
Production of IT services is carried out with the same production
function considered earlier, namely equation (10). However, proft
maximization must now take into account that the production of new IT
services will no longer be sold, but rented to intermediate goods producers.
The authors defne as the rental rate of IT services. As a result the IT
produced at time , will contribute to the holdings of IT by the cloud
provider. Period profts of the cloud provider are
The cloud-provider solves the following problem
(32)
(33)
ECLAC 172
After defning as the Lagrange multipliers on the constraint, it
can be shown that the FOCs for this problem are
(34)
and
(35)
The frst FOC implies that the provider of cloud services will rent
capital up to the point where the rental rate of capital equals the marginal
revenue product of capital. Combing the FOCs delivers
(36)

Proft maximization requires that the marginal cost of production of
an additional unit of IT at time , must be identical to the marginal
revenue that the frms obtain from producing the additional unit. The
latter is given by the rental rate at which the additional unit will be rented
out tomorrow net of the maintainance costs, plus the continuation value,
. The marginal revenue has to be discounted since the
IT producers will be able to rent the additional unit produced in just at
period +1.
Sectoral goods producers will now simply demand IT up to the point
where the rental cost of IT equals the marginal product of IT. Formally, the
FOC with respect to is no longer equation (24), but
(37)
Notice that combining equations (37) and (36) we recover equation
(24). Time entrants will no longer need to face up-front entry costs in
terms of IT. Once they are in the market and start production they will
demand IT services up to the point where condition (37) is satisfed. As a
result, frms will enter the market up to the point where
(38)
This means that the IT policy of a new entrant does not differ
from that of an incumbent producer. As in the pre-cloud economy, goods
producing frms need to set up a workforce before starting production. For
Broadband in Latin America: Beyond Connectivity 173
this reason the difference between the hiring policy of new entrants and
that of incumbent producers is the same as that spelled out above.
1. Aggregation and market clearing
Considering that sectors are symmetric and have a unit mass, the sectoral
number of frms and new entrants also represents their aggregate
counterpart. Thus, the dynamics of the aggregate number of frms is
As aggregate expenditure and sectoral expenditure are identical, it
follows that . Considering and
the individual production function we obtain
(39)
The aggregate production function features a form of increasing
returns. In this case a productivity shock impacts directly on output, but
also through the frm creation channel. Total vacancies posted at period
are
where is the number of incumbent producers and
is the number of new frms. Aggregating the budget constraints of
households yields the aggregate resource constraint of the economy
(40)
which states that the sum of consumption and investment in new
entrants and capital must equal the sum of labour income and aggregate
profts, , distributed to households at time . Aggregate profts are
defned as
8
(41)
where


and

8
Since the producer of IT material and the provider of IT services operate in perfect
competition, they make no profits.
ECLAC 174
In this case
Since , and defning
as the number of vacancies created by the producers of the fnal
good, it follows that
Notice that

hence


Since, aggregate profts can be
expressed as


The aggregate number of workers is , which leads
fnally to
As a result, the clearing of the market for the fnal good requires

Broadband in Latin America: Beyond Connectivity 175
or
(43)
(44)
Notice that the total IT produced at time must be equal to the sum
of investment by incumbent frms and that of new entrants, that is
thus, using , it follows
and fnally
(45)
The dynamics of the aggregate stock of IT reads as follows
2. The equilibrium in the cloud economy
In the cloud economy the producers of the fnal good rent their stock of
IT from the cloud services producer. Profts of an incumbent producer are
thus given by
while those of a new frm are
Aggregating as above leads to

Given that , it follows
while the entry condition is simply . As a result, the
market clearing condition reads as
ECLAC 176
(46)
where
E. Calibration
There are relatively few analyses of labour market dynamics at a
macroeconomic level in Brazil and Argentina. In the past few years, most
studies of labour markets have focused on microeconomic aspects such as
wage differentials, convergence and demand for skilled workers, among
other themes.
Siqueira (2009) is one of the few studies that analyzes variables
such as gross job creation, job destruction and job fnding in Brazil. He
fnds that job transitions in Brazil are more frequent than in most of OECD
countries where the same statistics are available. One possible explanation
for this high mobility in the Brazilian labour force is the presence of
informal jobs and self-employed workers, classes which are widely known
to be the most fexible. As suggested by Siquieira (2009), although this
feature can account for some part of the high worker mobility, it cannot
account for all of it. All this implies that labour market fows in Brazil are
closer to those of the dynamic United States labour market than to those
of the more rigid European countries. For this reason, in the remainder of
this section the authors calibrate the model, and in particular parameters
relative to the labour market, following two alternative strategies. The
frst one is meant to mimic the United States and Latin American labour
markets, characterized by high fows. The second one is instead based on
the more rigid European market.
Calibration is conducted on a quarterly basis as in Shimer (2005)
and Blanchard and Galì (2010) among others. Initially, the authors describe
parameters which are common across calibration strategies. The discount
factor, , is set to the standard value of 0.99. Campos and Iootty (2005)
report an average yearly exit rate for Brazil of 9.8%, very close to the 10%
reported by Bilbiie and others (2012) for the United States. For this reason
the authors set = 0.025.
The baseline value for the entry cost is set such that the ratio of
investment in new frms and physical capital is close to 15%, as in Bilbiie
Broadband in Latin America: Beyond Connectivity 177
and others (2012). The implied steady state price markup is about 35%. This
value is within the range estimated by Oliveira Martins and Scarpetta
(1999) for a large number of United States manufacturing sectors. With no
loss of generality, the value of is such that steady state hours equals one.
In this case the Frisch elasticity of labour supply reduces to , to which
a low value of 0.5 is assigned in line with the evidence. The authors take
as the baseline value for the intersectoral elasticity of substitution = 6,
as estimated by Christiano, Eichenbaum and Evans (2005) using United
States quarterly data between 1965 and 1995. As standard in the literature
the steady state marginal productivity of labour, , is set to 1. The same
value is set for the marginal productivity of capital in the IT producing
sector, .
The elasticity of matches to unemployment is =1/2, within the
range of the plausible values of 0.5 to 0.7 reported by Petrongolo and
Pissarides (2001) in their survey of the literature on the estimation of the
matching function. In the baseline parameterization the authors impose
symmetry in bargaining and set =1/2, as in the bulk of the literature.
Since the authors consider a labour-leisure choice, the overall replacement
rate is given by the sum of the unemployment insurance beneft and
the disutility cost of working. The authors calibrate the latter to 0.95
consistently with Hagerdon and Manovskii (2008). The cost of posting a
vacancy is obtained by equating the steady state version of the JCC and
the steady state wage setting equation. Finally the authors set =0.025
and =0.025. These fgures imply that the maintainance cost of the IT
stock is about 10% a year, and that the IT stock of a frm fully depreciates
in about 10 years.
Next the authors turn to parameters that differ across calibration
strategies and that characterize the fexibility of the labour market. In the
fexible environment they set the separation rate to 0.1, as suggested
by estimates provided by Hall (1995) and Davis and others (1996). They
then set the effciency parameter in matching, , and the steady state
job market tightness to target an average job fnding rate, z, equal to 0.7
and a vacancy flling rate, , equal to 0.9. The authors draw the latter value
from Andolfatto (1996) and Dee Haan and others (2000), and the former
from Blanchard and Galì (2010). Notice that a job fnding rate equal to 0.7
corresponds, approximately, to a monthly rate of 0.3.
The rigid market is instead characterized by a separation rate equal
to 0.03 in line with the estimates from the Labour Force Survey in Bell
and Smith (2002) and by a job fnding rate equal to 0.25 as in Thomas and
Zanetti (2009). Finally the authors set the vacancy flling rate to 0.7 in line
with estimates reported by the ECB (2002).
ECLAC 178
F. Transition to the cloud economy
This section evaluates the effect of the introduction of cloud computing.
Following the strategy in Etro (2009a), we assume that IT is produced more
effciently. This is obtained featuring a 1% increase in in the long run.
The authors also assume that bringing IT services on the cloud leads to a
reduction in the units of maintainance services required for each single
IT unit installed. This is formalized by assuming a 5% reduction in m.
Figure VI.1 displays percentage deviations from the pre-cloud steady state
when all existing frms adopt the cloud technology. Solid lines refer to
the case of a fexible labour market, dashed lines to that of a more rigid
market. Time on the horizontal axis is in quarters.
Figure VI.1
TRANSITION OF THE MAIN MACROECONOMIC VARIABLES FROM THE STEADY
STATE OF THE PRE-CLOUD ECONOMY TO THAT OF THE CLOUD ECONOMY
(Percentage deviations from the pre-cloud steady state
on the vertical axis, quarters on the horizontal one)
0 5 10 15 20 25 30 35 40 45 50
−1
−0.5
0
0.5
Employment in the IT sector
0 5 10 20 30 40 50
0
2
4
6
Number of firms
0 5 10 20 30 40 50
−5
0
5
10
IT stock
Flexible labor market Rigid labor market
Broadband in Latin America: Beyond Connectivity 179
Figure VI.1 (concluded)
Flexible labor market Rigid labor market
0 5 10 20 30 40 50
−0.01
0
0.01
0.02
0.03
Total employment
0 5 10 20 30 40 50
−0.8
−0.6
−0.4
−0.2
0
Price markup
0 5 10 20 30 40 50
−3
−2
−1
0
Entry cost
Source: Prepared by the author.
The introduction of cloud computing lowers the up-front entry costs
in terms of IT. The implied reduction in the latter is about 3% with respect
to that in the pre-cloud economy. This stimulates entry of new frms. Given
that entry is subject to a one-period time-to-build lag, the total number of
frms, N
t
, does not change on impact but builds up gradually. New frms
post a large amount of vacancies to reach their desired size. This results
in a persistent change in aggregate employment.A larger number of frms
leads to a stronger usage of IT. Higher employment together with a higher
IT stock lead in turn to a sustained increase in aggregate output. Finally,
the creation of new frms and the growth of the existing ones promotes
competition in the fnal good sector which translates into a lower markup.
Notice that employment in the IT sector initially goes down, due to
more effcient technology of maintainance, i.e. lower m. However, as the
stock of IT rises to its new long-run level, IT employment increases above
ECLAC 180
its initial value. A crucial role for the increase in employment, both in the
aggregate and in the IT sector, is played by the creation of new frms.
While the previous description holds for both the fexible and the
rigid characterization of the labour market, the variation in employment
is, as expected, more pronounced in the market characterized by more
fexibility. It is interesting that this is also mirrored in the changes in the
number of frms and in the price markups, which are more relevant in the
market characterized by higher worker fows.
Finally, to get a sense of the implications of this analysis, the authors
translate the results obtained with the model into the potential change
in the number of employed persons and number of business since the
introduction of cloud computuing. Argentina and Brazil are analyzed at
the sectoral level, but aggregate fgures for the United States and the EU-27
are also provided.
In line with the simulation above, the authors assume that the
economy is currently in the steady state of the pre-cloud situation. The pre-
cloud number of frms and employed persons is obtained by averaging the
number of frms and the number of workers over the past fve years. Then
the authors obtain the change in the employment and number of frms by
applying to the averages just obtained the rates of change implied by the
simulation. Data for Brazil come from the Relação Anual de Informações
Sociais (RAIS) of the Brazilian Ministry of Labour, which requires by law
that all formally-registered frms report information each year on each
worker employed by the frm. Data relative to Argentina come instead
from the Observatorio de Empleo y Dinámica Empresarial. The source
of United States data is the Business Dynamics Statistics (BDS) database,
while for the eurozone infomation on the number of frms and number of
employed persons is from Eurostat. As specifed above the authors assume
that Latin America and the United States are characterized by higher
labour market fexibility, while the European Union is characterized as a
rigid labour market. Tables VI.1 to VI.5 show the change in the number of
employed persons and the number of frms created after 5 and 10 years
from the adoption of cloud computing. This makes it possible to evaluate
the effects of this new technology in both the medium and the long run.
The focus is on those sectors that could be affected by the introduction of
cloud computing.
9
9
As an example, the authors do not consider sectors as mining and agriculture which are
characterized by aspects, such as natual capacity constraints, which are not captured by
their model.
Broadband in Latin America: Beyond Connectivity 181
Table VI.1
CHANGE IN THE NUMBER OF EMPLOYED PERSONS DUE TO THE
INTRODUCTION OF CLOUD COMPUTING IN ARGENTINA
Sector 5 years 10 years
Manufacturing
Utilities
Construction
Wholesail and retail trade
Hotels and restaurants
Transport
Financial services
Housing services
Education
Social services
Other services
25 600
1 200
9 200
22 400
4 900
11 100
3 200
17 200
8 500
5 600
8 400
28 100
1 300
10 100
24 700
5 400
12 200
3 500
18 900
9 300
6 200
9 200
Total 117 300 128 900
 Source: Prepared by the author.
Table VI.2
BUSINESS CREATION DUE TO THE INTRODUCTION OF CLOUD
COMPUTING IN ARGENTINA
(Number of new frms)
Sector 5 years 10 years
Manufacturing
Utilities
Construction
Wholesail and retail trade
Hotels and restaurants
Transport
Financial services
Housing services
Education
Social services
Other services
2 900
50
1 200
7 900
1 300
2 600
250
4 400
400
1 100
2 600
3 500
70
1 500
9 600
1 600
3 200
350
5 400
480
1 400
3 200
Total 24 700 30 300
Source: Prepared by the author.
Table VI.3
CHANGE IN NUMBER OF EMPLOYED PERSONS DUE TO THE INTRODUCTION
OF CLOUD COMPUTING IN BRAZIL
Sector 5 years 10 years
Industry
Construction
Wholesail and retail trade
Services
171 000
50 700
169 300
470 000
188 000
56 000
186 000
515 000
Total 861 000 945 000
Source: Prepared by the author.
ECLAC 182
Table VI.4
BUSINESS CREATION DUE TO THE INTRODUCTION
OF CLOUD COMPUTING IN BRAZIL
(Number of new frms)
Sector 5 years 10 years
Industry
Construction
Wholesail and retail trade
Services
17 150
17 750
131 500
35 700
21 000
22 650
191 500
167 500
Total 202 100 402 650
Source: Prepared by the author.
Table VI.5
JOB CREATION AND BUSINESS CREATION IN THE UNITED STATES AND
IN THE 27 EUROPEAN UNION MEMBER COUNTRIES
Job creation 5 years 10 years
United States
EU 27
2 892 000
3 127 000
3 179 000
3 431 000
Business creation 5 years 10 years
United States
EU 27
1 346 000
369 000
1 651 000
451 000
Source: Prepared by the author.
The authors’ simulations show a substantial impact of the structural
technological change associated with the introduction of cloud computing
on business and employment. In particular, their long-run estimates
show the creation of about 900,000 new jobs in Brazil and approximately
100,000 in Agentina. They have disaggregated the results for the two
Latin American countries for the sectors on which they could have a more
detailed analysis of the market structure. For reference purposes one
can compare the numbers with about 3 million new jobs expected under
similar assumptions in both the United States and the European Union. In
light of this, the impact of the adoption of cloud computing for the Latin
American countries appears quite large, especially for Brazil.
G. Conclusions
This chapter has provided a simulation of the economic impact of the
diffusion of cloud computing in Latin America, the United States and
the European Union. The authors evaluate the impact of the adoption of
cloud computing on output, employment and business creation using a
macroeconomic model characterized by endogenous market structures
and job matching. This setting features an economy with many sectors,
Broadband in Latin America: Beyond Connectivity 183
where the dynamics of the number of market competitors is endogenous.
Firms face a cost of market entry which they will decide they can afford
only if it is compensated by the expectation of future profts. In the
authors’ model, frms produce the fnal goods using labour and physical
capital. The stock of capital takes the form of IT hardware which the frm
has to install and maintain over time. Firms employ multiple workers,
and the labour market is characterized by frictions. Both new frms and
incumbent frms need to hire workers from the pool of unemployed agents
who are looking for a job and to set up a stock of IT material before starting
production. The simulation shows a substantial impact of a technological
change such as the introduction of cloud computing in terms of business
and job creation.
Contrary to the concern of many experts in the feld, the authors’
framework does not suggest a reduction in IT employment due to the
adoption of the cloud technology, at least in the medium and long run.
Part of the positive effects of cloud computing are going to be positively
related to the speed of adoption of the new technology. Of course there
are a number of factors that may slow down this adoption, such as a lack
of understanding of the cloud on the part of frms, systemic risk, security,
privacy and interoperability issues, reliability, jurisdictional complexity,
data governance, loss of IT control and general status quo inertia. For this
reason, the authors’ research suggests that policymakers should promote
as much as possible a rapid adoption of cloud computing. Concrete
interventions include (beyond the expansion of the broadband capacity,
of course):
(i) international agreements in favour of unrestricted fow of data
across borders (since data centres are located in different countries
with different privacy laws, data portability remains a key issue
for the diffusion of cloud computing)
10
and to agree with industry
leaders a minimum set of technological standards and process
standards to be respected in the provision of cloud computing
services to guarantee data security, privacy and portability, and
promote a healthy diffusion of the new technology;
(ii) introduction of fscal incentives for the adoption of cloud
computing, targeting dynamic sectors (for instance, governments
could fnance, up to a limit, the variable costs of computing for
all the domestic and foreign frms that decide to adopt a cloud
computing solution) or sectors were environmental gains are
likely to be larger.
10
On the security risks of cloud computing, see Ahmad (2010); on privacy issues, see
Ranganathan (2010).
ECLAC 184
(iii) introduction of government support for the reallocation of
employment within the IT feld (from IT departments, especially
of small frm,s to different destinations in the IT sector).
These policies may be studied in such a way to optimize the process
of adoption of the new technology and to strengthen the propagation of its
benefts. Such a normative analysis is left for future research.
Broadband in Latin America: Beyond Connectivity 185
Bibliography
Ahmad, M. (2010), Security Risks of Cloud Computing and Its Emergence as
5th Utility Service, Information Security and Assurance, Communications in
Computer and Information Science, 76, 209-219.
Andolfatto, D. (1996), Business Cycles and Labor Market Search, American Economic
Review, 86, 112-32.
Bell B. and J. Smith (2002), On Gross Worker Flows in the United Kingdom: Evidence
from the Labour Force Survey, Bank of England WP 160, Bank of England.
Bilbiie, F., F. Ghironi and M. Melitz (2012), Endogenous Entry, Product Variety, and
Business Cycles, Journal of Political Economy, 120, 2, 304-44.
Blanchard, O. and J. Galì (2010), Labor Market Frictions and Monetary Policy:
A New Keynesian Model with Unemployment, American Economic Journal:
Macroeconomics, 2, 2, 1-30.
Blanchard, O. and F. Giavazzi (2003), Macroeconomic Effects of Regulation and
Deregulation in Goods and Labor Markets, Quarterly Journal of Economics, 118,
3, 879-907.
Borek, C. L. Christensen, P. Hess, G. Rafert and J. Lerner (2012), Lost in the Clouds:
The Impact of Copyright Scope on Investment in Cloud Computing Ventures,
mimeo, Harvard University.
Campos, N. and M. Iootty (2005), Firm Entry and Exit in Brazil: Cross-sectoral
Evidence from Manufacturing Industry, Anais do XXXIII Encontro Nacional
de Economia.
Center of Information Development (2010), Whitepaper on Cloud Industry
Development in China, CCID Group, Beijing.
Cimoli, M., W. Pereira, G. Porcile and F. Scatolin (2011), Structural Change,
Technology, and Economic Growth: Brazil and the CIBS in a Comparative
Perspective, Economic Change and Restructuring, 44, 1, 25-47.
Christiano, L., M. Eichenbaum and C. Evans (2005), Nominal Rigidities and the
Dynamic Effects of a Shock to Monetary Policy, Journal of Political Economy,
113, 1-45.
Colciago, A. and F. Etro (2010), Real Business Cycles with Cournot Competition and
Endogenous Entry, Journal of Macroeconomics, 32, 4, 1101-17.
Colciago, A. and L. Rossi (2011), Endogenous Market Structures and Labor Market
Dynamics, mimeo, University of Milano Bicocca.
Davis, S.J., J. C. Haltiwanger and S. Schuh, 1996, Job Creation and Job Destruction, The
MIT Press, Cambridge
Den Haan, W., G. Ramey, J. Watson (2000), Job Destruction and the Propagation of
Shocks, American Economic Review, 90, 482-98.
De Oliveira, D. and E. OgasawaraIs (2012), Cloud Computing the Solution for
Brazilian Researchers? , International Journal of Computer Applications, 6, 8, 19-23.
ECB (2002), Labor Market Mismatches in the Euro Area Countries, European
Central Bank.
Etro, F. (2009a), The Economic Impact of Cloud Computing on Business Creation,
Employment and Output in the E.U., Review of Business and Economics, 54, 2,
179-208.
Etro, F. (2009b), Endogenous Market Structures and the Macroeconomy, New York and
Berlin, Springer.
ECLAC 186
Etro, F. and A. Colciago (2010), Endogenous Market Structure and the Business
Cycle, The Economic Journal, 120, 549, 1201-33.
Fershtman, C. and N. Gandal (2012), Migration to the Cloud Ecosystem: Ushering
in a New Generation of Platform Competition, Communications & Strategies, 85,
1, CEPR Discussion Paper 8907.
Hagedorn, M. and Manovski, J. (20080, The Cyclical Behavior of Equilibrium
Unemployment and Vacancies Revisited, American Economic Review, 98, 4,
1692-706.
Hall, R. (1995), Lost Jobs, Brookings Papers on Economic Activity, Economic Studies
Program, The Brookings Institution, 26, 221-74.
Haltiwanger, J. C., R. S. Jarmin and J. Miranda (2010), Who Creates Jobs? Small vs.
Large vs. Young, NBER Working Paper No. 16300.
Jaimovich, N. and M. Floetotto (2008), Firm Dynamics, mark up Variations, and the
Business Cycle, Journal of Monetary Economics, 55, 7, 1238-52.
International Data Corporation (2008), IT Cloud Services Forecast – 2008-2012: A Key
Driver for Growth, mimeo, 8 October.
International Data Corporation (2009), White Paper. Aid to Recovery, 9 October.
Kydland, F. and E. Prescott (1982), Time to Build and Aggregate Fluctuations,
Econometrica, 50, 6, 1345-70.
Kuyucu, A.D.H. (2011), The playground of cloud computing in Turkey, Procedia
Computer Science, 3, 459-63.
Lanvin, B. and P. Passman (2008), Building E-skills for the Information Age, Chapter
1.6, Global Information technology Report 2007-2008, WEF.
Lee, Y. and T. Mukoyama (2008), Entry, Exit, and Plant-level Dynamics over the
Business Cycle, mimeo, Federal Reserve Bank of Cleveland.
Merz, M. (1995), Search in the Labor Market and the Real Business Cycle, Journal of
Monetary Economics, 36, 269-300.
Mortensen, D. and C. Pissarides (1994), Job Creation and Job Destruction in the
Theory of Unemployment, Review of Economic Studies, 61, 3, 397-415.
Oliveira Martins, J. and S. Scarpetta (1999), The Level and Cyclical Behavior of
Mark-ups Across Countries and Market Structures, OECD Working Paper No.
213, OECD Publishing.
Petrongolo, B. and C. Pissarides (2001), Looking into the Black Box: A Survey of the
Matching Function, Journal of Economic Literature, 39, 2, 390-431.
Pissarides, C. (2000), Equilibrium Unemployment Theory, Cambridge: MIT Press.
Ranganathan, V. (2010), Privacy Issues with Cloud Applications, IS Channel, 5,
1, 16-20.
Siqueira, F. (2009), The Ins and Outs of Cyclical Unemployment in Brazil - A First
Assessment, mimeo, Economic School of Sao Paulo - Getulio Vargas Foundation.
Shimer, R. (2005), The Cyclical Behavior of Equilibrium Unemployment and
Vacancies, American Economic Review, 95, 1, 25-49.
Thomas, C. and F. Zanetti (2009), Labor Market Reform and Price Stability: An
Application to the Euro Area, Journal of Monetary Economics, 56, 6, 885-99.
Trigari, A. (2009), Equilibrium Unemployment, Job fows and Infation Dynamics,
Journal of Money, Credit and Banking, 41, 1, 1-33.
West, D. (2010), Saving Money Through Cloud Computing, mimeo, Governance
Studies at Brookings, Washington, D.C.
Third part
Public policies
Chapter VII
National broadband plans
Hernán Galperin
Judith Mariscal
María Fernanda Viecens
1
A. Introduction
Sizeable public investment in the deployment of new network
infrastructure and ambitious government initiatives to develop
broadband services are unequivocal signs of the changing role of the
State in the telecommunications sector. Contrary to the consensus that
prevailed until a few years ago, governments are no longer content
with regulating private-sector activity and using universalization funds
to correct market failures. Government fnancing of networks and
infrastructure, State participation in operating those networks, and a
proactive telecommunications industry policy stance are all back in the
toolbox for telecommunications sector policymakers.
1
Hernán Galperin is a professor at the Universidad de San Andrés in Buenos Aires; Judith
Mariscal is a professor at the Centro de Investigación y Docencia Económicas (CIDE)
in Mexico City, and María Fernanda Viecens is a researcher at the Universidad de San
Andrés and at the Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET)
in Buenos Aires.
ECLAC 190
This change is seen in developed and developing countries alike
and has been particularly evident in the proliferation of what are called
‘national broadband policies.’ This term encompasses the diverse set
of initiatives adopted over the past fve years whose core objective is to
accelerate the roll-out and adoption of broadband services. The larger
countries of Latin America have been particularly proactive in the design
and implementation of national broadband plans, which have been
accompanied by growing efforts towards regional coordination.
This changing role of the State in the telecommunications sector
has raised a number of questions. To what extent does it represent a
return to the days before market liberalization and privatization of State-
owned operators? What safeguards should be adopted to avoid market
distortions and the crowding out of private investment? What should be
the objectives of State intervention in the market for broadband services,
and what are the best instruments for achieving them? How to regulate
the activities of operators that receive public subsidies or are directly
controlled by the State?
This chapter takes up these issues by means of a comparative
analysis of the network deployment objectives, instruments and models set
out by a number of countries of Latin America in their national broadband
plans. One of the main objectives of this chapter is to identify similarities
and differences, both among the plans put in place by countries in the
region and in comparison with initiatives taken by developed countries.
Another goal is to analyse these plans in the context of broader political
changes in the region, especially in the wake of the international economic
crisis of the late 1990s. The recommendations focus on the development of
regulatory frameworks for broadband service that ensure complementarity
between public-sector initiatives and the promotion of private-sector
competition and investment.
This chapter is divided into fve sections. The following section
identifes the factors that have led the governments of the region to launch
initiatives supporting the roll-out and adoption of broadband services.
Section C provides a snapshot of the national broadband plans of fve
countries (Argentina, Brazil, Chile, Colombia and Mexico), which are also
outlined in the table in the annex. These countries were selected because
they are useful examples and because information on implementation of
the plans in question was available. Section D identifes the similarities
and differences in the objectives, instruments and management of these
plans, which are then compared with initiatives in developed countries.
The European experience with regulatory guidelines for achieving
balance between public initiatives and private investment is of particular
interest. The fnal section discusses policy recommendations and sets out
the conclusions of the study.
Broadband in Latin America: Beyond Connectivity 191
B. The end of a cycle: the changing role
of the State in telecommunications
In Latin America, the cycle of policies geared towards deregulating the
telecommunications industry and privatizing State-owned operators
began in the late 1980s and was showing the frst signs of having run
its course by late in the frst decade of the 2000s. This waning might at
frst seem surprising because the cycle saw leaps in service coverage,
exponential growth in investment and a virtuous process of technological
innovation, new business models and new services (Estache and others,
2002; Jordán and others, 2010).
Studies of the impact of the reforms on aggregate well-being have
demonstrated that the indirect effect on employment has been positive,
and that the direct effect in terms of layoffs in privatized enterprises
has been largely offset by the growth in total employment in the sector
(McKenzie and Mookherjee, 2003). Furthermore, several studies have
documented the positive or neutral distributive impact of privatizing and
opening the telecommunications market, despite tariff adjustments linked
to reforms in the sector (particularly in local fxed services) (Navajas, 1999;
Ennis and Pinto, 2003).
So, if most of the evidence points to the success of the reform process
that began in the 1990s, why are governments interested in changing it
and heightening State intervention in the telecommunications sector? This
chapter identifes various drivers. Some have to do with the changing
economic and political context in the region; others are associated with
trends within the telecommunications sector itself.
To start with the context factors, the frst is what has been called the
political ‘left turn’ taken by the governments of Latin America in the early
2000s (Castañeda, 2006; Levitsky and Roberts, 2011). This shift is relevant
in that it brought with it a policy agenda calling for more State intervention
in the economy and a return to the developmental and industrial policy
tradition that had characterized Latin America during the greater part of
the twentieth century (Corrales, 2008). Specifcally, the economic crisis of
1998-2002 (when per capita output in the region contracted and poverty
and inequality levels increased) penalized several of the governments
associated with market reforms as public opinion shifted towards
candidates with redistributive agendas calling for more State intervention
in the economy (Murillo and others, 2011).
This shift has been especially noticeable in the sweeping public-
opinion rejection of the privatization of public service enterprises in
Latin America following the crisis. The level of support for privatization
plummeted from 46% in 1998 to 19% in 2004, after which it recovered
ECLAC 192
slightly (Latinobarómetro, 2011). Studies have identifed numerous reasons
for the scant support in the region for private management of public
services, primary among which are lack of confdence in the capacity of
governments to regulate private operators adequately (Panizza and Yañez,
2006), the continued existence of private monopolies in several sectors
(Murillo and Martinez Gallardo, 2006) and the unequal distribution of the
benefts of privatization (Shirley, 2004).
The second context factor is the signifcant improvement in the
terms of trade of many countries in the region, particularly in South
America. Several authors have pointed out that the combination of fscal
and external surpluses resulting from what they refer to as the ‘commodity
boom’ not only gave governments the resources needed to invest heavily
in infrastructure but also reduced the macroeconomic risk involved in
State operation of public service enterprises (Weyland, 2009; Murillo and
others, 2011). By enabling the rapid build-up of foreign reserves and easing
the external debt burden, the macroeconomic boom cycle that began in the
early 2000s increased the capacity of governments to carry out agendas
providing for redistribution and State intervention in strategic industries.
This context is the inverse of the situation in the late 1980s,
when the State controlled the incumbent operators but was unable
to make the investments needed to modernize and expand national
telecommunications network coverage while the private sector had both
the resources and the know-how to do so. Two decades later, in a context of
global economic uncertainty, private investors are hesitant to make large
investments in network infrastructure, particularly the riskier ones with
lower expected rates of return such as deploying the fbre backbone beyond
the confnes of large urban centres. This has encouraged the governments
of the region —benefting from favourable terms of trade— to fll the void
left by the slowdown in private investment in the sector.
Focusing on the telecommunications sector, the third factor behind
the public policy about-face is the growing body of evidence about the
contribution that telecommunications —and broadband in particular—
make to aggregate economic growth and job creation. The evidence of
the importance of telecommunications for economic growth and business
effciency is nothing new and has, for fxed telephony, been available
at least since the 1980s (Hardy, 1980; Leff, 1984). This contribution was
subsequently confrmed by studies of mobile telephony (Roller and
Waverman, 2001) and broadband services (Qiang and Rossotto, 2009;
Koutrompis, 2009).
The difference lies in the role of the State in ensuring that the supply
of services is enough to harness the potential of telecommunications as
an engine for the other sectors of the economy. In other words, while the
Broadband in Latin America: Beyond Connectivity 193
empirical evidence just shows the positive impact of network roll-out for
competitiveness and employment, in the late 2000s an emerging consensus
began to call for proactive public policies to speed deployment, particularly
in the case of next-generation networks (NGNs) (OECD, 2009; ECLAC,
2010; ITU Broadband Commission, 2011). These policies were part of the
economic stimulus packages implemented in several countries in response
to the 2008 economic crisis, once the deployment of broadband had been
linked to increases in competitiveness and employment (Qiang, 2010).
Just as the State played a key role in developing the electricity grid,
transportation systems and the telecommunications network itself before
the 1980s, the new consensus is calling for governments to take on a
similar role in modernizing the telecommunications infrastructure in the
twenty-frst century.
2
The fourth factor explaining the policy shift in the sector has to do
with the limited impact of the instruments designed during the reform
process in terms of mitigating geographical disparities in network
deployment and access to services. Basically, the evidence points to failures
in monitoring and enforcing private operators’ network deployment
obligations and to problems in designing and implementing universal
service funds (USFs). Various studies (for example, REGULATEL, 2006)
have shown that private-sector investment during the past two decades
has been concentrated in urban areas with higher per capita income,
perpetuating the gap in access to services between urban and rural areas
and by income level (Grazzi and Vergara, 2011). The limited impact of
USFs in Latin America has been widely documented as well (Stern, 2009;
Barrantes, 2011).
3
Towards the end of the frst decade of the 2000s, the governments
of Latin America reached basically the same conclusions. On the one
hand, the limited penetration of high-capacity backbone networks and
2
Obviously, the consensus is not sweeping; Kenny (2011) has reviewed the evidence and
summarized the criticisms of public investment in next-generation networks.
3
Brazil’s Universal Telecommunications Service Fund (FUST) was set up in 2000 and
collects approximately US$ 800 million annually. But the fund has never been used, due
to legal obstacles to its implementation. In Argentina, USF implementation and collection
problems date back to renegotiation of contracts with fixed telephony operators after
the 2011 economic crisis. Strictly speaking, Mexico does not have a USF but rather a
temporary fund (the Telecommunications Social Coverage Fund) set up in 2002 to bring
fixed telephony to isolated localities. Other countries, such as Colombia, Chile and Peru,
have managed to achieve better results in implementing USFs. However, as pointed out
by Barrantes (2011), even where USFs were more or less successfully implemented they
have two basic problems. The first has to do with funding constraints and, therefore, with
limited impact. The second is that by focusing on shared fixed telephony and Internet
access they have lost ground to new technologies such as broadband, which require
infrastructure deployment and service provision models.
ECLAC 194
the lack of competition for access networks outside of large urban centres
were having a negative effect on broadband service coverage, quality and
price. On the other hand, the instruments for promoting infrastructure
deployment and mitigating regional imbalances in access to services
were insuffcient for meeting the new need for connectivity in homes,
businesses and public institutions. In that context, State intervention in the
deployment of networks in areas with limited private return was deemed
to be not only a matter of equity but also of taking the fullest advantage of
the positive externalities to be derived from broadband consumption.
The fnal factor for understanding the shift in telecommunications
sector policy focus in Latin America is the spread of broadband incentive
policies implemented by the more developed countries. A few Southeast
Asian countries have taken the lead in deploying next-generation
networks, and their success can be seen in deployment and adoption
rankings.
4
The leadership of countries such as the Republic of Korea is
not associated so much with successful market reforms as with proactive
public policies combining private-sector incentives with signifcant public
investment in infrastructure, training, and research and development
(R&D) (Kim and others, 2010; Choi herein). For the developing countries,
the lesson to be learned from these success stories is, therefore, the need
to complement private enterprise with greater State involvement in
channelling investment and stimulating demand for broadband.
C. Overview of national broadband plans
1. Argentina: Plan Nacional de Telecomunicaciones
Argentina Conectada
Argentina Conectada (the National Telecommunications Plan) was
introduced in October 2010 to integrate ongoing initiatives in the sphere
of new information and communications technologies (ICT) such as the
roll-out of digital television and the introduction of notebook computers in
schools, as well as to respond to uneven access to high-speed transmission
networks in the region.
5
The goal is to widen coverage and improve
broadband Internet service access, particularly in areas that are not
proftable for commercial operators (Plan Argentina Conectada, 2011).
The plan hinges on, among other things, deployment of a federal
fbre-optic network extending some 58,000 kilometres, combining
4
Some of the most notable are the ICT Development Index (IDI) prepared by the
International Telecommunication Union (ITU), the Network Readiness Index (NRI)
designed by the World Economic Forum, and the broadband reports produced by the
Organisation for Economic Cooperation and Development (OECD).
5
Argentina Conectada was passed by Presidential Decree No.1552 in October 2010.
Broadband in Latin America: Beyond Connectivity 195
(i) the construction of a 22,000-kilometre provincial network and an
18,000-kilometre interprovincial network with (ii) the lighting of stretches
of dark fbre-optic cable by the electrical company Transener (in which
the State is a shareholder) and (iii) fbre exchange agreements with several
private operators. This backbone network is expected to cover 97% of the
population by 2015, with local operators being responsible for providing
last-mile service.
The federal fbre optic network is operated by AR-SAT, a State-
owned enterprise created in 2006 to take over the assets of Nahuelsat,
a privately-owned satellite operator that, unable to meet its liabilities,
had agreed to transfer its assets to the recently-created State enterprise.
The government plan emphasized the need to lower costs and increase
competitiveness in the wholesale access market, but it did not structurally
separate the new State operator and thus left the door open for AR-SAT to
step into the retail side.
The plan includes initiatives that complement deployment of
the fbre-optic network. For example, it promotes the creation of public
access and training centres (called knowledge access hubs and digital
access points) as well as digital literacy and ICT research initiatives.
On the regulatory front, the plan includes initiatives to promote
competition, such as a tender to award additional spectrum for mobile
broadband services and reactivating the universal service fund. The
plan also includes aid in the form of credit and technical assistance
for cooperatives and small private broadband service operators that
the plan sees as key for achieving coverage and adoption targets.
Investments under the plan are expected to total 8 billion Argentine
pesos (approximately US$ 1.8 billion) over fve years (2011-2015), of which
3.7 billion Argentine pesos (approximately US$ 840 million) represents
investment in the backbone fbre-optic network.
So far, tenders have been offered for the construction of 11 segments
(approximately 18,700 kilometres) of the federal fbre-optic network and
for 2,500 kilometres of provincial network, representing a total investment
of 2.85 billion Argentine pesos (approximately US$ 640 million). The new
national data centre out of which AR-SAT will run the new network is
also under construction. Laying of the fbre-optic cable across the Strait
of Magellan to Tierra del Fuego island has been completed. Fifty shared
access and training centres have been inaugurated in several provinces,
and cooperation agreements with neighbouring countries, particularly
Brazil and Uruguay, for the creation of Internet exchange points are
moving forward.
6
6
See Argentina Conectada, Informe de Gestión 2012 [online] http://www.argentinaconectada.
gov.ar.
ECLAC 196
2. Brazil: Plano Nacional de Banda Larga
The Plano Nacional de Banda Larga (National Broadband Plan) (PNBL)
was developed during 2010 under the administration of President Lula
and endorsed by the administration of President Dilma Rouseff. The fve
broad objectives of the plan are: make it easier for the population to access
broadband Internet services; speed social and economic development;
promote digital inclusion; reduce social and regional inequalities; and
promote job and income creation.
7
The plan especially seeks to lower the
cost of Internet access and increase broadband service coverage and quality.
It provides for action along four fronts: regulation and infrastructure
standards; tax incentives for telecommunications services; production and
technology policy; and deployment of a national fbre-optic network.
The primary focus of the national fbre-optic network is deployment
of a network connecting the 27 state capitals so as to meet government
agency connectivity requirements, and to offer connectivity to localities
where there is no private operator service or the supply is of low quality
and high cost. The goal is to connect 4,278 (76%) of the 5,564 municipalities
around the country within four years (2011-2014) at a total investment cost
of 5.7 billion reais (approximately US$ 3.3 billion). The network is planned
to extend over 30,000 kilometres and will include the laying of new fbre
cables and making use of the idle fbre-optic capacity of State-controlled
enterprises like Petrobras and Eletrobras.
Under the PNBL, Telebras will build and operate the network and
the State will inject 3.2 billion reais (approximately US$ 1.8 billion) to
capitalize the company. Telebras (the incumbent State telecommunications
company until its privatization in 1998) was reactivated by the government
in 2010 to lead infrastructure initiatives under the plan. It will supply
capacity to the wholesale market and reach the end user by means
of agreements with last-mile operators. The plan requires that these
agreements include an offer to supply the client with access at 1 Mbps for
35 reais per month (about US$ 20). As in Argentina, Brazil’s plan provides
that Telebras can, on terms set by the telecommunications sector regulator
(Anatel), operate in the retail segment in localities where there are no local
operators or where service is poor.
The entry of a State operator into the market has been questioned by
the major private operators (Jensen, 2011). In spite of this, some of them have
already signed agreements with Telebras. The smaller operators see it as an
opportunity to change the conditions of a market in which fve operators
control more than 90% of the market (PNBL, 2010). Meanwhile, Anatel,
7
The Plano Nacional de Banda Larga (PNBL) was established by Presidential Decree
No. 7175 of May 2010.
Broadband in Latin America: Beyond Connectivity 197
in an effort to promote competition, adopted a general competition plan
that empowers it to require companies with substantial market leverage to
share their infrastructure with other operators, although it declared a nine-
year regulatory holiday for the deployment of fbre-optic networks.
The PNBL also envisages fscal incentives, support for research and
development, and fnancing for equipment using domestically-produced
technology, as well as leveraging government purchasing power to
promote the production of national technology. The State expects to invest
some 2.5 billion reais (approximately US$ 1.45 billion) in these initiatives,
on top of 7.5 billion reais (US$ 4.17 billion) in National Bank for Economic
and Social Development (BNDES) credit facilities.
Telebras has already signed contracts for taking the fber-optic
backbone network to 40% of Brazil’s municipalities by the end of 2012.
And this State operator has signed data capacity contracts with private
operators who have committed to offer 1-Mbps connections at a cost of 35
reais per month with a guaranteed speed that cannot be lower than 20% of
the advertised speed. The frst of these contracts was signed in June 2011
with the company Sadnet for an area in the state of Goiás, where Telebras
offers 100 Mbps of capacity at a cost of less than 200 reais (US$ 115) per
Mb per month (Pena, 2012). In addition, several tender processes have been
initiated for the construction of a number of segments of the new network.
3. Chile: Plan Todo Chile Comunicado
The Plan Todo Chile Comunicado (All Chile Connected Plan) project
seeks to spur production development, education and the insertion of
communities in isolated rural areas of the country by supplying them
with mobile broadband coverage. Most (68%) of these communities are
small (less than 1,000 inhabitants) and are thus of little interest to private
operators. The initiative has been implemented within the framework of
the Telecommunications Development Fund (FDT), which was created to
increase telecommunications service coverage in low-income and isolated
areas and earmarks national budget funds for subsidies to encourage
telecommunications enterprises to provide services to such areas.
The Plan Todo Chile Comunicado project (2010) is a public-private
partnership where the government subsidizes operators to provide end-
user services in predetermined areas under the conditions established
in the tender specifcations for the plan. The initiative was launched in
2010 with the aim of providing mobile broadband services to more than
3 million inhabitants of 1,474 isolated rural localities upon completion of
its three phases by 2012. The tender was awarded to the private operator
Entel and involves a total investment of approximately US$ 110 million, of
ECLAC 198
which US$ 65 million will be contributed by Entel and the remainder in
equal parts by the FDT and local governments.
The plan is up and running in the more than 1,000 localities where
Entel offers mobile broadband Internet access services at a monthly rate of
14,220 Chilean pesos (approximately US$ 30) with a maximum download
speed of 1 Mbps. The operator also offers a daily access mode at a rate of
1,886 Chilean pesos (approximately US$ 4).
4. Colombia: Plan Vive Digital
The Plan Vive Digital (Live Digital Plan) was launched in 2010 to deliver
mass Internet access in Colombia. It sets three basic goals to be achieved
within fve years: triple the number of municipalities connected to the
national fbre-optic network, connect 50% of micro, small and medium-
sized enterprises and 50% of households to the network, and quadruple
the number of Internet connections around the country—from 2.2 million
connections in 2010 to 8.8 million connections in 2014. The plan seeks to
take fbre-optic connectivity to 62% of the country’s 1,120 municipalities
(representing 90% of the population) and establish shared access centres in
all localities with more than 100 inhabitants (Vive Digital, 2010).
One of the main initiatives under the plan is the National Fibre-
Optic Project for deployment of backbone fbre-optic cable to those
municipalities not connected to high-speed transmission networks under a
public-private partnership model. In July 2011, the Ministry of Information
and Communications Technologies (responsible for implementing the
plan) published the tender conditions for the National Fibre-Optic Project.
In November 2011, the project was awarded to the Colombia Provisional
Fibre-Optic Union, consisting of the companies Total Play and TV Azteca,
controlled by Mexico’s Grupo Salinas. The government of Colombia agreed
to contribute 415 billion Colombian pesos (approximately US$ 237 million),
which was estimated to be about one third of the total investment needed
for deployment of the network. According to the Ministry, out of the
four bids tendered, the project was awarded to the consortium offering
coverage to the greatest number of municipalities (1,078).
8
According to the tender specifcations, the operator must confgure
the fbre-optic transmission network, operate it and connect it to the target
group of municipalities ensuring freedom of access, transparency, non-
discriminatory treatment, free competition, effciency and user rights.
Once the infrastructure has been deployed in the municipalities, the
company shall operate and manage the network for 15 years pursuant to
8
See [online] www.mintic.gov.co/index.php/fibra-inicio/53-sitio-fibra-optica/sitio-fibra-
noticias/542-20111104licitacionfibra..
Broadband in Latin America: Beyond Connectivity 199
the tender specifcations, which include considerations such as free Internet
broadband access for the 2,000 public institutions in the municipalities
connected to the network. At the end of this period, ownership of the
infrastructure will pass to the tender awardee and come under the
standard regulations governing the rest of the private operators. There are
no structural separation requirements, so the fbre-optic network operator
may provide access services to the end user.
Other initiatives under the plan include establishing a legal and
regulatory framework for convergence, effcient infrastructure utilization
and incentives for the software and digital content industries. (In July
2011 the withholding tax on software development frms in Colombia
was reduced from 11% to 3.5%.) For e-government services, the plan calls
for 100% of national entities and 50% of territorial entities to provide
government services online by 2014. In addition, import tariffs on Internet
access terminals such as computers, tablets and smartphones were
eliminated in the second half of 2011. The plan also provides for designing
ICT capacity-building programmes for segments of the population with
lower resources and the removal of the value-added tax on broadband
services to those population segments.
5. México: Agenda Digital.mx
During the administration of President Calderon (2006-2012) measures
were taken to promote Internet deployment and use. In the frst quarter
of 2012, these ongoing initiatives were grouped into two broadband
development programmes: Acciones para el Fortalecimiento de la Banda
Ancha y las Tecnologías de la Comunicación y la Información (Actions
for Strengthening Broadband and Information and Communications
Technologies) and Agenda Digital.mx (Digital Agenda.mx). The frst
programme focuses on boosting the supply of telecommunications
services by encouraging private investment in infrastructure and
government investment. The second centres on developing demand for
services by promoting the adoption and use of ICT and developing the
market for telecommunications-based content and applications. The four
goals are: promote investment in the infrastructure needed for providing
broadband services nationwide; reduce the cost of broadband services;
increase digital literacy; and encourage uptake of Internet services.
One of the initiatives in place by 2011 was for operationalizing
the fbre-optic backbone network of the Federal Electricity Commission
(CFE), a public corporation and the main electricity provider in Mexico.
This network extends over 34,000 kilometres and has 36 fbre-optic
cables, of which CFE uses only six (Mariscal and Flores-Roux, 2009). In
2010, the Communications and Transport Secretariat (SCT) auctioned
ECLAC 200
off a couple of dark fbre cables from the CFE network on three routes
totalling 19,500 kilometres for a period of 20 years, during which time the
operator agreed to make complementary investments aimed at boosting
network coverage and capacity. A consortium made up of the Spanish
company Telefónica and Mexico’s Televisa won the tender with a bid
of approximately US$ 70 million that included a commitment to invest
another US$ 103 million in extending the network by 1,700 kilometres.
The new operator can offer intercity data-transfer services under non-
discriminatory conditions with the proviso that third parties be allowed
to use the dark fbre in those parts of the network not being used by the
consortium (SCT, 2012).
Another infrastructure development initiative is the ‘Fibre-to-
the-Node’ programme. It is being funded by the National Infrastructure
Fund (FONADIN), which can provide non-reimbursable subsidies for
private operators to deploy fbre-optic cable “in localities with insuffcient
coverage or where there is market failure” (SCT, 2012). More than 400
municipalities have been identifed as meeting these requirements, and
interested parties have been invited to bid for the frst stage in the state
of Guanajuato. Under the public-private partnership planned for this
initiative, the licensee will be required to share the infrastructure and
comply with non-discriminatory rules of access. These projects, because of
their nature, will be subject to asymmetric rate regulation.
The e-Mexico programme, renamed Coordinación de la Sociedad
de la Información y el Conocimiento (Information and Knowledge Society
Coordinating Offce) (CSIC), has relaunched its digital literacy strategy
with the Campaña Nacional por la Inclusión Digital de los Adultos, geared
towards digital literacy for persons aged 25 to 54. Partnering with the
Instituto Nacional para la Educación de los Adultos (INEA), has enabled
the campaign to train between 300,000 and 500,000 persons per year
during the past fve years. The government, in order to promote access for
low-resource segments of the population, is seeking to increase the number
of Digital Community Centres from 6,788 to 24,000 by the end of 2012. This
will be achieved mainly by increasing satellite connectivity capacity. CSIC
will also launch a social network called Digital Club to build technological
capacities and encourage technological entrepreneurship among young
people. Thirty-seven Digital Clubs are planned at Mexican Youth Institute
facilities and some secondary schools.
Broadband in Latin America: Beyond Connectivity 201
D. National broadband plans: different strategies,
same goal
The broadband plans rolled out in Latin America have much in common
with initiatives launched by the governments of many countries. This
section identifes the similarities and differences between initiatives put
in place by the countries of the region and compares them with those
undertaken by more developed countries.
1. Analysis and objectives
A comparative analysis of the plans implemented in Latin America with
those undertaken in developed countries reveals, frst of all, important
differences as to the diagnosis of the situation. In the wealthy countries,
the main issue is incomplete deployment of high-speed access networks
(especially fbre-to-the-home); in the countries of Latin America, the issue is
a lack of interurban backbone network capacity. That is why the developed
countries seek to encourage deployment of high-speed access networks
(usually in excess of 30 Mbps) while the countries of Latin America are
focusing more on correcting regional imbalances in the supply of frst-
generation broadband services with more modest target access speeds
(usually around 1 Mbps). In other words, for the more developed countries
the priority is to improve service quality, while the rationale for initiatives
in the region is to expand geographical coverage and increase the number
of connections.
Initiatives in Latin America are focused almost exclusively on
increasing coverage and encouraging competition on the wholesale-
backbone-segments of the network, particularly in areas where there are
few or no private operators. This reduces (but does not eliminate) the
problem of market distortions and displacement of private investment that
public investment in network infrastructure could generate because the
focus is on areas not covered by fbre optic networks (as in Colombia and
Mexico) or on areas with a single incumbent provider (characteristic of the
plans of Argentina, Brazil and, to a certain extent, Mexico). This strategy
stands in contrast to those being followed by some developed countries,
such as Australia, New Zealand and Singapore, whose plans have no
specifc geographical focus.
ECLAC 202
2. Investment and funding
A comparison of the estimated investment costs and sources of funding
for the national broadband plans of Latin American countries yields
several interesting results. Public investment in the deployment of network
infrastructure in the region ranges from US$ 2.60 per capita in Chile to
US$ 21 per capita in Argentina (see annex). These differences are due in
part, to differences in length and the segments of the planned network
that are already built. Compared with more ambitious initiatives, such
as in Australia and New Zealand (Given, 2010), where public investment
per capita is US$ 845 and US$ 245, respectively, the levels of investment
planned by the countries of Latin America are low. The comparison is,
however, qualifed by the fact that Australia’s and New Zealand’s plans
include fbre-optic services to households with access speeds of up to 100
Mbps, while the plans implemented in Latin America have much more
modest objectives.
Taking into consideration those objectives, the geographical
conditions, and the network architecture, it would seem more appropriate
to compare Latin American broadband plans with initiatives in the United
States of America and Canada, where per-capita public investment in
network infrastructure amounts to US$ 8 and US$ 5, respectively (Qiang,
2010).
9
This comparison shows how sizeable the investment commitment is
in the case of Argentina (US$ 21 per capita) and, to a lesser degree, Brazil
(US$ 9.2 per capita), particularly when the per capita GDP of the United
States is almost three times that of Argentina and four times that of Brazil.
The opposite applies to Chile, which, with a GDP per capita similar to that
of Argentina, plans to invest ten times less.
As for funding, the region’s national broadband plans are not well
coordinated with existing supplier subsidy schemes, such as USFs. The
fact that funding from national broadband plans is not earmarked and
is therefore subject to fscal budget variations and macroeconomic cycles
poses a signifcant challenge for these initiatives in the medium term.
This affects Argentina and Brazil, in particular, where new State-owned
operators have to fnd the delicate balance between fnancial sustainability
and achieving their objectives of serving the least proftable segments of
the market.
3. Deployment models
A comparison of the national plans implemented by the countries of the
region reveals many similarities in terms of their analysis of the situation
9
The figure for the United States includes only the US$ 2.5 billion to be spent under the
Broadband Initiatives Program (BIP) for network infrastructure deployment.
Broadband in Latin America: Beyond Connectivity 203
and their rationale and goals. But differences arise when it comes to
the policy instruments the governments have chosen, especially for
deployment of national fbre-optic backbone networks. There are two
approaches: the one taken by Argentina and Brazil, where backbone
network deployment and operation is assigned to a State-controlled
company, and the public-private partnership (PPP) model adopted in
Colombia, Mexico and Chile. Each model has different implications
regarding the role of the State in the sector.
One of the most signifcant aspects refers to the level of fnancial
commitment by the State in each model. In the State-operator model,
the government is responsible for the investment cost of laying the
entire backbone network, with the private sector covering the last-mile
investment (in both cases, there are provisions for soft loans to small local
operators). There is, therefore, an implicit connection between public and
private investment, although how they are linked is not spelled out in
the plans. By contrast, under the PPP model the split between public and
private investment is defned, thus reducing the State’s initial investment
commitment as well as its future commitments for maintaining and
operating the network. In the cases examined, public funding ranges from
38% of the estimated total investment in Colombia to 45% in Chile.
In the State-operator model the level of public investment is higher
because the government must cover 100% of the investment in new
network infrastructure (capital expenditure, or CAPEX) as well as the long-
term cost of operating the network (operating expenditure, or OPEX). But
it should be borne in mind that in such cases the investment is in assets of
which the State retains ownership whereas in Chile’s and Colombia’s PPP
models the State subsidizes deployment of the network that, after a certain
period, is handed over to the private operator.
The PPP model adopted by Chile is closely related to the market-
reform paradigm for the sector. In this model, a private operator provides
services in unproftable areas in exchange for a subsidy whose amount is
established by tender. A well-designed scheme of this kind optimizes the
use of public funds and minimizes the displacement of private investment
(Wallsten, 2009). In Colombia, the tender process has also been used to
determine the fnal amount of the State subsidy, with 15-year agreements
providing for set contributions by the State. In cases such as these, the
implementation mechanism seeks to promote market competition in areas
where there is little potential for private return. In contrast, the strategy
in Argentina and Brazil, and to a certain extent in Mexico, is to promote
market competition by creating a new backbone network operator that
would put pressure on the incumbent operators and drive down the price
of access.
ECLAC 204
The PPP scheme adopted by Mexico differs in that the State auctions
the use of existing but underutilized infrastructure (the fbre network
owned by the State-owned electrical company, the Federal Electricity
Commission, CFE). Despite some of the aforementioned weaknesses in
the design of the tender, such approaches usually allow full advantage to
be taken of the combination of State-owned assets and the fnancial and
management capacity of the private sector. Although the scheme depends
on the availability of this kind of infrastructure, there are usually many
State-owned assets that could leverage private investment in new network
infrastructure (such as dark fbre, ducts and associated rights of way, poles
and towers for wireless network equipment, and so on).
10
4. Regulation and coordination with the private sector
Another feature of broadband plans in Latin America is that they combine
public investment in backbone network infrastructure with last-mile
services provided by the private sector. Both in Argentina and in Brazil,
the plans promote small- and medium-sized local access operators by
means of soft loans, technical training and facilities for interconnection at
local Internet exchange points. In these countries, the entry of the State
operator in the retail market segment is seen as a strategy of last resort,
although the plans do not set out the criteria for such intervention.
However, unlike initiatives in some developed countries (such as Australia
and New Zealand) that include the creation of a State-controlled backbone
network operator, Argentina, Brazil and Colombia have no provisions for
the structural or functional separation of operators (AR-SAT, Telebras and
the Colombian Provisional Fibre Optic Union, respectively). However,
Mexico and Colombia have set up safeguards in the tender contracts that
bar discrimination in access by the new network operator.
In Argentina and Brazil, the regulatory framework and the
authority responsible for monitoring the conduct of the State operator
are still under discussion. Both theoretical arguments and international
experience suggest the need for State operators to be held to the same
regulatory standards as other operators in the market. If State operators
provide services and develop infrastructure in areas that are not proftable,
subsidies should be awarded transparently via mechanisms that optimize
public investment. Such safeguards turn out to be particularly important
when the plans do not provide for long-term fnancing mechanisms for the
State operator.
10
See ITU (2008) for a discussion of the subject.
Broadband in Latin America: Beyond Connectivity 205
Europe has experience in setting rules for State aid for next-
generation network projects. These rules seek to avoid displacing
private investment and to create a lasting balance between public
and private initiatives. They were created in September 2009 by the
European Commission in response to the wave of broadband initiatives
implemented by European Community member governments. Known
as the Broadband Guidelines (European Commission, 2009), these
directives defne and guide the actions of European governments in the
deployment of telecommunications infrastructure and refer specifcally
to the application of the principles on State aid for the establishment of
broadband networks.
11
This is the reason why the directives seek to
establish clear rules governing where, and how, public funds may be used
for the deployment of broadband networks.
The Guidelines are based on the distinction between competitive
areas (the ‘black areas’), in which State aid is not allowed, and areas that
are unproftable or unserved (the ‘white’ and ‘grey’ areas, respectively) in
which State aid might be justifable under certain conditions. The areas are
defned in the following manner: (i) white areas: broadband services are not
currently available, nor are they likely to be deployed by private investors
in the near future (which is understood to be within the next three
years); (ii) black areas: there are at least two broadband network providers,
and broadband services are supplied under competitive conditions
(competition understood as based on broadband infrastructure) and
(iii) grey areas: there is only one network supplier operating under
monopolistic conditions; in this situation, the Commission requires a
detailed analysis and evaluation before State aid can be authorized.
Since the directives were published, the Commission has heard
approximately 50 cases of State aid for the roll-out of next-generation access
networks.
12
Most of the projects presented by European governments have
involved broadband deployment in rural areas not covered by the private
sector (white areas) and have been authorized without objection by the
Commission.
13
A few cases have required a more detailed examination of
11
European Community competition policy prohibits unjustified State aid that could distort
competition.
12
For example, during 2010 projects amounting to some 1.8 billion euros were approved.
See “State Aid: Commission approves record amount of State aid for the deployment of
broadband networks in 2010”, Press Release (IP/11/54), 20 January 2011.
13
Some examples of projects authorized by the Commission are Broadband support
in rural areas of Germany, National broadband plan for rural areas in Italy, High-
speed construction aid in sparsely-populated areas in Finland, and RAIN (Rural Area
Information Technology Network) in Lithuania.
ECLAC 206
the non-white areas impacted by the projects.
14
In general, the directives
have helped to minimize the crowding out of private investment,
establishing conditions for complementarity between public and private
sector investment in the deployment of next-generation networks.
E. Conclusions
During the past decade, the role of the State in the telecommunications
industry in the countries of Latin America has changed signifcantly.
At frst glance, this change seems paradoxical in view of the evidence
concerning the success of market reforms in the sector. Part of this paradox
is explained by rapid technological change, and particularly by the
growing importance of broadband Internet services in the economic and
social fabric. Even though private investment has generated exponential
growth in the levels of access to basic telephony services, an examination
of the main countries in the region reveals the insuffcient development of
both fbre backbone networks and the capacity of households, frms and
the State itself to absorb the new services associated with the Internet.
However, the success that some countries leading the way in the adoption
of broadband services (the Republic of Korea, in particular) have had with
proactive public policies has drawn the attention of countries in the region.
This chapter puts forth that the national broadband plans adopted
in Latin America in recent years were shaped by factors in the political and
economic context of the early twenty-frst century and the development
of the sector itself. However, these plans do not represent a wholesale
return to the State-owned operator model of the period prior to the market
reforms in the sector. As has been seen, the new State-owned operators
created in the framework of the national broadband plans are mandated
above all to operate in the wholesale access market in areas where private
investment has been insuffcient or non-existent, coordinating with private
last-mile operators under non-discriminatory conditions. Other countries
have adopted several different public-private partnership models which
14
This has been the case with the Xarxa Oberta project in Cataluña, Spain. This project
involved the deployment of a fibre optic network that would connect all the public
facilities (schools, health centres, police stations, courts of justice, and so on) in the
region. This aspect of the project did not, in itself, trigger an examination because self-
provision by the State is not considered State aid. However, the project also allowed the
network service provider to use any surplus network to provide wholesale broadband
access to third-party operators. It was considered that this aspect of the project could
distort competition by allowing an operator receiving State funding to enter the private
broadband market as a wholesale provider. The Commission then resolved that providing
wholesale broadband access should be restricted to the white areas of Cataluña and be
barred in the big cities in the region (Ganuza and Viecens, 2011).
Broadband in Latin America: Beyond Connectivity 207
ensure complementarity between public funding and private operation of
the new network (Falch and Henten, 2010).
All in all, there have been no rigid dichotomies between State
operators and competition between private actors, but rather a range of
policies seeking complementarities to promote the development of the
sector. The new role of the State is a mixed model in which the State
and the private sector partner in the deployment of a general-purpose
technology (broadband Internet access) with many economic and social
externalities. The specifc circumstances facing the telecommunications
sector call for close cooperation between the State and private actors in
order to maximize the contribution of the sector to the economic and social
development goals of the region.
It is still too soon to assess the impact of national broadband
plans in Latin America. However, some issues are bound to arise in the
immediate future. First, as noted above, State-run telecommunications
companies bring to mind the multiple diffculties that the old State-owned
telephone companies used to face. Operating these enterprises effciently
and transparently is the core challenge of national broadband plans. To
this end, emulating the instruments of State intervention that have been
successful in other countries must be considered in the institutional
context of each country. The capacity of the State to manage a complex
infrastructure network when technology and patterns of demand are
changing rapidly must be evaluated carefully by each government.
Moreover, the international economic context is cyclical in nature, enabling
governments to take on major network infrastructure investments with
relative ease today but requiring sustainable models for funding new
networks over the medium term.
Second, complementarity between the public and private sectors will
work as long as the mechanisms safeguarding competition and promoting
private investment are not weakened. This chapter has pointed out the
many challenges that setting transparent rules for State-owned or State-
subsidized operators presents. This is a long-standing debate in countries
(such as Uruguay and Costa Rica) where the State has maintained a
signifcant holding in the incumbent operator, and it is surfacing in many
other countries in the region. It is important to take advantage of successful
mixed-management experiences in the telecommunications industry and
learn from the experiences of other regions in regulatory matters, such as
the case of the European Union, discussed earlier.
Finally, to return to the question of the role of the State in the
sector, there is consensus as to the need for policies that address market
failures in the provision of services, particularly those that have a strong
impact on social and economic well-being. This need is less apparent in
ECLAC 208
areas supplied by a single operator, usually the historical incumbent. Is
the duplication of publicly funded network segments the most effcient
tool for ensuring competitive access prices? Obviously, there are other
tools (such as regulating the cost of broadband access and unbundling
network components that cannot be replicated) that the countries of the
region should not overlook. This will require continuing to strengthen
the technical expertise of industry regulators and building up institutions
that enable these regulators to implement rules adapted to the new context
of the sector.
Broadband in Latin America: Beyond Connectivity 209
Bibliography
Barrantes, R. (2011), “Uso de los fondos de acceso universal de telecomunicaciones
en países de América Latina y el Caribe”, Project Documents, No.429
(LC/W.429) Santiago, Chile, Economic Commission for Latin America and
the Caribbean (ECLAC).
Castañeda, J. (2006), “Latin America’s left turn”, Foreign Affairs, vol. 85, No. 3.
Corrales, J. (2008), “The Backlash against Market Reforms”, Constructing Democratic
Governance in Latin America, J.I. Domínguez. y M. Shifter (eds.), Baltimore, Johns
Hopkins University Press.
ECLAC (Economic Commission for Latin America and the Caribbean) (2010), Banda
ancha: Una urgencia para América Latina y el Caribe (LC/R.2158), Santiago,
Chile.Ennis, H. and S. Pinto (2003), “Privatization and income distribution
in Argentina”, unpublished.Estache, A., M. Manacorda and T. Valletti (2002),
“Telecommunications reforms, access regulation, and Internet adoption in Latin
America”, Economica, No. 2.
European Commission (2009), “Guidelines for the application of state aid
rules in relation to the rapid deployment of broadband networks” [online]
http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:C:2013:025:000
1:0026:EN:PDF.
Falch, M. y A. Henten (2010), “Public private partnerships as a tool for stimulating
investments in broadband”, Telecommunications Policy, vol. 34.
Ganuza, J.J. y M.F. Viecens (2011), “Deployment of high-speed broadband
infrastructures during the economic crisis. The case of Xarxa Oberta ”,
Telecommunications Policy, vol. 35.
Given, J. (2010), “Take your partners: Public private interplay in Australian and
New Zealand plans for next generation broadband”, Telecommunications Policy,
vol. 34, No. 9.
Grazzi, M. and S. Vergara (2011), “Determinants of ICT Access”, ICT in Latin America:
A Microdata Analysis, S. Vergara, S. Rovira and M. Balboni (eds.), Santiago, Chile,
Economic Commission for Latin America and the Caribbean (ECLAC).
Hardy, A. (1980), “The role of the telephone in economic development”,
Telecommunications Policy, No. 4.
ITU (International Telecommunication Union) (2008), Trends in Telecommunication
Reform: Six Degrees of Sharing, Geneva.
ITU Broadband Commission (2011), Broadband: A platform for progress, Geneva, ITU/
UNESCO.
Jensen, M. (2011), Broadband in Brazil: A Multipronged Public Sector Approach to Digital
Inclusion, Washington, D.C., infoDev/World Bank.
Jordán, V., H. Galperín and W. Peres (2010), Acelerando la revolución digital: Banda
ancha para América Latina y el Caribe (LC/R.2167), Santiago, Chile, Economic
Commission for Latin America and the Caribbean (ECLAC).
Kim, Y., T. Kelly and S. Raja (2010), Building Broadband: Strategies and Policies for the
Developing World, Washington, D.C., World Bank.
Kenny, C. (2011), Overselling Broadband: a critique of the recommendations of the
Broadband Commission for Digital Development, Center for Global Development.
Koutrompis, P. (2009), “The economic impact of broadband on growth: a
simultaneous approach”, Telecommunications Policy, vol. 33.
ECLAC 210
Latinbarómetro (2011), Latinobarómetro 2011 Report, Santiago, Chile.
Leff, N. (1984), “Externalities, information costs, and social beneft-cost analysis
for economic development: An example from telecommunications”, Economic
Development and Cultural Change, vol. 32, No. 2.
Levitsky, S. and K. Roberts (2011), The Resurgence of the Latin American Left, John
Hopkins Press.
Mariscal, J. and E. Flores-Roux (2009), “Propuesta de licitación de la fbra oscura
propiedad de la CFE. Solución que genera escasez artifcial, tanto presente
como futura”, Regional Dialogue on the Information Society (DIRSI).
McKenzie, D. and D. Mookherjee (2003), “The distributive impact of privatization
in Latin America: evidence from four countries”, Economía, Latin American and
Caribbean Economic Association (LACEA).
Murillo, M. V. and C. Martínez Gallardo (2006), “Political competition and policy
adoption: market reforms in Latin American public utilities”, American Journal
of Political Science, vol. 51, No. 1.
Murillo, M. V., V. Oliveros and M. Vaishnav (2011), “Voting for the Left or governing
on the Left?”, Latin America’s Left Turn, S. Levitsky and R. Kenneth (eds.),
Cambridge University Press.
Navajas, F. (1999), Structural Reforms and the Distributional Effects of Price Changes in
Argentina, Buenos Aires, FIEL.
OECD (Organisation for Economic Cooperation and Development) (2009), The Role
of Communication Infrastructure Investment in Economic Recovery, Paris.
Panizza, U. and M. Yañez (2006), “Why are Latin Americans so unhappy about
reforms”, Working Paper, No. 567, Washington, D.C., Inter-American
Development Bank.
Pena, A. (2012), “A Banda Larga e o Cenário de Telecomunicações Brasileiro”,
Revista de Direito, Estado e Telecomunicações, vol. 3, No. 1.
Plan Argentina Conectada (2011), Plan de Acción del Plan Nacional de
Telecomunicaciones “Argentina Conectada”, Ministerio de Planifcación
Federal, Inversión Pública y Servicios [online] http://www.argentinaconectada.
gob.ar//adjuntos/139/documentos/000/025/0000025555.pdf.
PNBL (Programa Nacional de Banda Larga) (2010), “Brasil Conectado. Programa
Nacional de Banda Larga. Documento base do Programa Nacional de Banda
Larga”, Publicação da Secretaria-Executiva do Comitê Gestor do Programa de
Inclusão Digital [en línea] www.planalto.gov.br/brasilconectado.
Proyecto Todo Chile Comunicado (2010), Proyecto Bicentenario “Red de
Internet Rural: Todo Chile Comunicado”, Ministerio de Transportes y
Telecomunicaciones, Subsecretaría de Telecomunicaciones [online] http://www.
subtel.gob.cl/prontus_subtel/site/artic/20100819/asocfle/20100819103226/
ppt_bicentenario_fdt_red_internet_rural.pdf.
Qiang, C. (2010), Broadband Infrastructure in Stimulus Packages: Relevance for
Developing Countries, World Bank.
Qiang, C. and C. Rossotto (2009), “Economic impacts of broadband”, Information
and Communications for Development, World Bank.
Regulatel (Latin American Telecommunications Regulators Forum) (2006), Nuevos
modelos para el acceso universal de los servicios de telecomunicaciones en América Latina.
Roller, L. and L. Waverman (2001), “Telecommunications infrastructure and economic
development: a simultaneous approach”, American Economic Review, No. 4.
Broadband in Latin America: Beyond Connectivity 211
SCT (Communications and Transport Secretariat) (2012), Acciones para el
fortalecimiento de la banda ancha y las tecnologías de información y comunicación,
Mexico City.
Shirley, M. (2004), “Why is sector reform so unpopular in Latin America”, Roland
Coase Institute Working Paper Series, No. 4.
Stern, P. (2009), “Objetivos y obligaciones de acceso universal en el sector de las
telecomunicaciones en América Latina”, Más allá del mercado: las políticas de
servicio universal en América Latina, J. Caldaza, A. Costas and J. Jordana (eds.),
Fundación CIDOB.
Vive Digital (2011), Ministerio de Tecnologías de la Información y Comunicaciones,
República de Colombia [online] http://www.vivedigital.gov.co/fles/Vivo_
Vive_Digital.pdf.
Wallsten, S. (2009). Reverse Auctions and Universal Telecommunications Service:
Lessons from Global Experience. Federal Communications Law Journal 61(2):
373-394.
Weyland, K. (2009) The Rise of Latin America’s Two Lefts: Insights from Rentier
State Theory, Comparative Politics 41(2):145-164.
Wallsten, S. (2009), “Reverse auctions and universal telecommunications service:
lessons from global experience”, Federal Communications Law Journal, vol. 61, No. 2.
Weyland, K. (2009), “The rise of Latin America’s two Lefts: insights from Rentier
State Theory”, Comparative Politics, vol. 41, No. 2.
ECLAC 212
Annexes
Annex VII.1
Main elements of broadband plans in selected countries
Table VII.A.1
Argentina Brazil Chile Colombia Mexico
Population 40 738 000 195 498 000 17 133 000 46 299 000 110 675 000
Lan area (km
2
) 2 780 400 8 514 877 756 102 2 070 408 1 972 550
Name of the
initiative
Plan Argentina
Conectada
Plano
Nacional de
Banda Larga
(PNBL)
Plan todo Chile
Comunicado
Plan Vive Digital Agenda Digital.
mx
Implementation
period
2011-2015 2010-2014 2010-2012 2010-2014 2012-2015
Population
objective
97% of the total
population
89% of
the total
population
90% of the rural
population (3 million)
90% Not specifed
Geographical
objective
100% de of
localities
(9,400
localities).
76% of
municipalities
1 474 rural localities 62% of
municipalities
Not specifed
State and
small- and
medium-sized
enterprise
connectivity
objectives
100% of public
schools and
government
offces
connected .
100% of
schools,
health-care
centres,
libraries
and local
governments
connected
2 133 schools, 1108
day-care centres,
534 health-care
centres connected
100% of health-
care centres and
public schools
connected; 50%
of small and
medium-sized
enterprises
covered
100% of
schools, public
libraries and
health-care
centres and
offces of the
three levels of
government
Price and/
or quality
objectives
10Mbps. 1 Mbps at
US$ 20 per
month.
1 Mbps at US$ 30
per month
1 Mbps Not specifed
Fiscal
incentives
Incentives for
local Internet
service
providers as
yet unspecifed
No Universal
Service Fund
obligations
for small/
medium-
sized Internet
service
providers
Not included Reductions
in tariffs on
imported
equipment,
VAT exemption
on broadband
services for
low-income
population
segments
Not included
Total public
investment
US$ 1.8 billion US$3.25
billion
US$ 45 million US$ 2 25 billion Not specifed
Total
investment per
capita
US$ 44.2 US$ 16.6 US$ 2.6 US$ 48.6 Not specifed
Broadband in Latin America: Beyond Connectivity 213
Argentina Brazil Chile Colombia Mexico
Source of
public funding
General
national
government
funds
General
national
government
funds
50% regional
governments, 50%
Telecommunications
Development Fund
(FDT), private sector.
General national
government
funds
General
national
government
funds, private
sector
Public
investment
in network
infrastructure
US$ 840
million
US$1.8 billion US$ 45 million (Entel
to contribute about
another US$ 55
million).
US$ 230 million
(Colombia
Provisional Fibre
Optic Union to
contribute about
another US$ 370
million).
Not specifed
Per capita
public
investment
in network
infrastructure
US$ 21 US$ 9.2 US$ 2.6 US$ 5 Not specifed
Network
characteristics
58 000 km
of national/
provincial fbre
optic network.
35 000 km
of national/
provincial
fbre optic
network
12 optical nodes and
mobile broadband for
the last mile
17 000 km
of national/
provincial fbre
optic network
22 000 km
of national/
provincial fbre
optic network
Ownership and
management
of the network
AR-SAT (public
operator)
Telebras
(public
operator)
Entel (private
operator).
Colombia
Provisional
Fibre Optic
Union (private
operator).
Telefónica/
Televisa
consortium
(private
operator).
Source: Argentina Conectada (2011), Colombia Plan Vive Digital (2011), Plano Nacional de Banda Larga
(PNBL) (2010), Todo Chile Conectado (2010), Agenda Digital.mx (2012).
Table VII.A.1 (concluded)
Chapter VIII
Broadband and industrial policy:
the Korean experience
Daewon Choi
1
A. Broadband industrial policy: definition and scope
The industrial policy debate has advanced considerably in recent years.
The focus has been on the dilemma between accepting horizontal
policies based on World Trade Organization (WTO) rules and regulatory
framework, and the possibility of developing a space for sectoral policies.
This discussion is inseparable from another debate unfolding at the same
time on policy instruments and how effective they are.
Experts do not agree on the costs and benefts of industrial policy.
At the national level, some hold that the newly industrialized economies
used industrial policy effectively in the early phases of industrialization
from 1960 to 1980. Others argue that these instruments, including direct
subsidies, have undermined the general effciency of the economy.
Similarly, there are those who hold that during the liberalization and
privatization process of the 1980s and 1990s, Latin America eliminated
the policy instruments that caused distortions in industrialization
1
Consultant to the Economic Commission for Latin America and the Caribbean (ECLAC)
and Vice President of the Korean Association of Trade and Industry Studies.
ECLAC 216
through import substitution; others contend that the structural barriers to
industrial development in the region have not been lowered.
The conventional debate on industrial policy came to a head in the
mid-1990s. With the establishment of the WTO in 1995, the multilateral
trading system reached an agreement prohibiting the use of direct
subsidies as an industrial policy tool. This was a decisive moment for the
global industrial policy framework in that it created a system based on
guidelines that included a dispute settlement mechanism for regulating
the principles of most-favoured-nation and national treatment. Since then,
government intervention by means of fnancial support to promote export
competitiveness at the sectoral level has been subject to close multilateral
examination under the WTO Agreement on Subsidies and Countervailing
Measures and other standards.
While this institutionalized horizontal industrial policies, it did not
completely mark the end of sector-based industrial policy. In particular,
academic circles in favour of trade policy must still answer the following
questions. What industrial policy instruments are available following
the establishment of a system based on the WTO rules and regulatory
framework? Is a targeted policy actionable under the WTO Agreement
on Subsidies and Countervailing Measures? Is there room for industrial
policy over and above the approved instruments, such as support for
technological research and development (R&D) or regional development?
Experts are divided with regard to how much the horizontal
approach determines industrial policy space in the framework of the
system based on WTO rules. While some hold that the multilateral trade
system has created a global environment with equal conditions favouring
horizontal policies, others argue that compliance with WTO rules does not
preclude a policy space in which national economic policy can effectively
give rise to targeted industrial policy.
This context has informed most of the academic debate on
industrial policy in the trade policy arena, constraining the various
currents of thought and their contributions over the past decade. While
this trade policy context will most likely continue to frame the industrial
policy debate in the immediate future, globalization is now being
driven by revolutionary advances in information and communications
technologies (ICTs). The WTO multilateral trading system was conceived
during the Uruguay Round (from the mid-1980s to the mid-1990s), when
the Internet had not yet emerged as a trade tool.
Consequently, the WTO agreements —the General Agreement on
Tariffs and Trade (GATT), the General Agreement on Trade in Services
(GATS) and the Agreement on Trade-Related Aspects of Intellectual
Broadband in Latin America: Beyond Connectivity 217
Property Rights (TRIPS)— were drafted without taking into account the
trade environment that the Internet would later create. As a result, the
Basic Telecommunications Agreement and the Information Technology
Agreement (ITA), two important WTO guidelines on ICTs, were
addressed only as a postscript. In subsequent decades, especially in the
2000s, the impact of the Internet increased so much that devices based on
Internet protocol (IP) now outnumber the world population. In the 2010s,
IP-based ICTs have become one of the main channels of trade activity
around the world.
This has created a vacuum in the debate on industrial policy in
the digital era, which translates into the following questions. Can the
territory-based trade policy system, tariffs and technical barriers to trade
be effectively applied to the new industrial and technological policy
framework that is taking shape? Are the devices, platforms, contents and
interconnected networks blurring the harmonized classifcation systems
for goods and services? Will this new hyperconnected world ushered
in by broadband respond consistently to the conventional trade policy
regime of the past decade? Will IP-based broadband services, operating at
speeds of gigabits per second (Gbps) or terabits per second (Tbps), displace
the dispute settlement model that depends on rules of origin for digital
goods, providing faster resolution? If a set of digital goods and broadband
services (such as geolocation services) account for more than 30% of the
value of an automobile, can a code against unfair competition and the
WTO Agreement on Subsidies and Countervailing Measures effectively
safeguard equal conditions in the multilateral trade system?
This chapter makes a contribution to the debate on industrial policy
from the perspective of the development of broadband and the industrial
policy associated with that development. It moves the debate rooted in
conventional trade policy towards one on the new industrial policy taking
into account the various factors outlined below.
First, while the dichotomy between horizontal and sectoral
industrial policies is theoretically valid, there is an increasingly fuzzy
area of industrial policy that is so broad that it is hard to defne as either
horizontal or sectoral. For example, the development of broadband
networks serves as a horizontal cornerstone for the economy, but the rules,
standards, and regulations are based on a sectoral focus that cannot be
classifed as horizontal.
Second, there is an area of industrial policy for broadband
development that could be defned as a cutting-edge in terms of
institutional arrangements and instruments. Through a combination of
legal and technical barriers, among others, it is possible to promote the
development of the cutting edge of a given industry without using direct
ECLAC 218
subsidies or choosing winners. For example, an R&D subsidy to achieve
fourth-generation (4G) standards is permissible at the horizontal level,
while the creation of a specifc long-term evolution (LTE) standard for
commercialization is clearly sectoral but nevertheless compatible with the
WTO Agreement on Technical Barriers to Trade (TBT).
Third, areas with an R&D-based industrial policy show a strong
correlation between the targeted support of these activities in priority
sectors and the type of instrument used. For example, the choice of high-
priority sectors to receive government support through administrative and
regulatory incentives can make for more effcient R&D resource allocation
in the pre-commercialization phase. This was the case with code division
multiple access (CDMA) technology in the Republic of Korea, and the
process is being replicated in China, for example, with time-division long-
term evolution (TD-LTE) technology.
Fourth, there is an increasing convergence in production, services
and consumption that transcends any single sector, industry or frm,
making a defnitive classifcation irrelevant in terms of sanctions based
on price distortions in the multilateral trade system. Many products
comprising platform devices for broadband development cannot be
classifed exclusively as goods or services, because they can serve as
components but function as services. For example, customs duties on the
cross-border movement of a DVD as a physical unit can be uniform or non-
existent, due to an exemption in accordance with the WTO Information
Technology Agreement (ITA), but in cyberspace the tax rate would depend
on the value generated in the downloaded content. Another example is a
cross-subsidy for broadband infrastructure and related devices, which can
reduce the cost of using convergent services for businesses with an eye to
increasing the competitiveness of digital services exports.
Finally, because the level of national production is increasingly
tied to the global production supply chains, it can be hard to defne the
boundary between rules of origin and product value. SMS messages and
voice transmissions may have little commercial value, but when gigabytes
and terabytes of data are involved there can be diffculties in managing
large volumes of data and cloud computing as part of industrial policy. This
often happens with Internet protocol, which obstructs national jurisdiction
for maintaining a protection system. The case of Google in China, which
involves a certain degree of broadband-based industrial policy, is a good
example. At the same time, it is diffcult to confgure the IP ecosystem
within the set of products governed at the national level, given that global
interoperability is critical for hyperconnectivity in the future.
For these reasons, it is urgent to review the role of broadband-
related services and products in industrial and technological development,
Broadband in Latin America: Beyond Connectivity 219
outside conventional trade policy. This is particularly necessary at a time
when global actors in the feld of networks, devices, platforms and content
are moving to a new generation of regulatory policies and infrastructure.
In a highly connective digital world, the difference between the fow
of Kbps and gigabyte storage may be only a technical issue, but in the
physical world of convergence, the fow of gigadata and terabyte cloud
storage can give rise to a different ecosystem affecting all the modules
along the production and distribution chain.
In this chapter, broadband and hyperconnectivity are considered
essential to the new industrial policy, termed broadband industrial policy,
insofar as they can represent different facets of trade policy. The following
typology is useful for analysing broadband industrial policy in the context
of the WTO-based multilateral trade system: sectoral industrial policy
in the period before the WTO (pre-1995: frst generation-1G); horizontal
industrial policy in the WTO (1995-2005: second generation-2G); targeted
cutting-edge policy in the WTO (2005-2010: third generation-3G); and
generative convergence policy in the WTO (2010-2020: fourth and ffth
generations-4G-5G). Diagram VIII.1 illustrates the analytical framework
for broadband industrial policy.
The transition from the frst to the second generation took almost a
decade, but the transition to the third generation took just fve years. The
OECD economies are expected to move from a 3G environment to a 4G era
in just a few years, which will accelerate both hyperconnectivity and the
corresponding digital divide with developing countries.
Diagram VIII.1
ANALYTICAL FRAMEWORK FOR BROADBAND INDUSTRIAL POLICY
Broadband policy
Industrial
policy
Broadband
industrial policy
WTO
1995
2G
2000
3G
2010
4G -5G
2020
Hyperconnectivity
(Gbps)
Mobile
broadband
(Mbps)
Internet and
FTTx (Kbps)
Offline PC
From
sectoral
to horizontal
policies
Targeted
cutting-edge
policies
Generative
convergence
policies
Source: Prepared by the author.
ECLAC 220
1. Sectoral industrial policy in the period before
the WTO (pre-1995: 1G)
Before the establishment of the WTO, many countries (most notably
the newly industrialized economies) implemented sectoral industrial
policies often based on targeted subsidies. This coincided with the frst
generation of ICTs, in which most communications went through fxed-
line telephones and the commercialization of mobile technology was
still in its infancy. It was in this period that the Republic of Korea laid
the groundwork for the development of CDMA technology based on
government support for R&D. The government, research centres and
businesses worked together to commercialize this technology in the
period before the WTO.
2. Horizontal industrial policy in the WTO (1995-2005: 2G)
With the creation of the WTO in 1995, especially in the context of the
Agreement on Subsidies and Countervailing Measures, direct fnancial
subsidies to selected industries and frms were prohibited. Violation of
this rule was subject to WTO regulations in the framework of the dispute
settlement mechanism. This provision led member countries to use
subsidies less frequently and more indirectly. In this second-generation
phase, CDMA technology became commercially viable even without
government subsidies, and an alliance was forged between the R&D
centres and the private sector responsible for commercialization. Early 2G-
3G was based on the standardization of mobile CDMA technology.
3. Targeted cutting-edge policy in the WTO
(2005-2010: 3G)
Restrictions on the use of subsidies left governments with few industrial
policy instruments during this phase. Nevertheless, a few countries
somewhat selectively promoted national industrial development so as
to foster key cutting-edge technologies. Instead of subsidies, the Korean
government implemented a major policy for deployment of high-speed
landline broadband and the standardization of wireless broadband
(WiBro) for 3G-3.5G mobile technology. While the development of
high-speed broadband was undeniably successful, the promotion of
3.5G WiBro mobile technology did not produce the desired results in
the national and international markets. The government’s targeted
cutting-edge approach pushed the limits of innovation to a new level
at a time when only a few countries had reached the third-generation
technological frontier.
Broadband in Latin America: Beyond Connectivity 221
4. Generative convergence policy in the WTO
(2010-2020: 4G-5G)
During this period, government policy for industrial and technological
development began to converge towards an intensive use of digital data
and ICTs to produce a hybrid industry with embedded data capacity,
which in turn produces a new form of connection with nearby industries.
This period saw the global fnancial crisis and the stalling of the WTO
Doha Round; this created a new policy space for industrial development,
especially in the BRIC countries (Brazil, Russia, India and China). The main
features of this period would be, on the one hand, the use of government
support to protect domestic industry and, on the other, a convergent policy
approach for the main industries —including automobiles, biotechnology,
mechatronics, shipbuilding and petrochemicals— to build a new platform
for big-data industry, which in turn feeds back into these activities. Since
many countries are loath to undo the achievements of the WTO, the
new protectionist policy will increasingly be channelled through a new
generative industrial policy in order to avoid a deadlock in the multilateral
world trading system. In this sense, two of the main components of
industrial policy will be, frst, large-scale deployment of ultra-fast
broadband and giga-level mobile broadband such as LTE and, second, an
industrial policy based on renewable energies to establish an emissions
trading scheme. Cloud computing technology will take the world trading
system to a new frontier, which will begin with 4G-5G mobile technologies.
This period corresponds to broadband industrial policy.
Unlike the debate on conventional industrial policy, broadband
industrial policy moves away from the dichotomy between horizontal
policy and targeted policy and is more aligned with a cutting-edge
policy. In addition, when targeted R&D support is combined with
technical standards, regulations and systems integration for production,
distribution and consumption, it becomes a facet of industrial policy. For
example, while a cloud-computing-based service with a direct link to
smart devices can produce signifcant value, the rules of origin on value
production might not be the same as the distinction between onshore and
offshore production of physical goods.
To clarify the discussion of broadband industrial policy, the
following assumptions shall be used for analytical purposes. First, this
policy differs from conventional industrial policy in that it exists in
a connected world. Some analysts predict that by 2030 the number of
connected devices and artefacts will be three times the world population,
which underscores the impact of the connected world on the global
economy. According to this position, broadband infrastructure in not
ECLAC 222
just a functional tool for connectivity but also a means of applying a new
industrial policy.
Second, broadband industrial policy differs from conventional
industrial policy in that the generalized effect of the convergence of people-
to-people or peer-to-peer (P2P) connections, machine-to-machine (M2M)
connections and people-to-machine (P2M) connections will create new
economic opportunities in which the promotion of a unique device and an
exclusive platform may not guarantee national industry competitiveness.
Third, unlike technical trade barriers aimed at boosting national
competitiveness, the hyperconnected world requires a totally new
ecosystem in which the de facto standards coexist with de jure ones and
where the national standard in a large market can lay the groundwork for
a de facto one, as in the case of the TD-LTE standard in China, contrary to
conventional industrial policy.
Fourth, the growing importance of the electronic commerce
system will require a complementary platform for interoperability
with the multilateral trading system. For example, the WTO General
Agreement on Trade in Services seeks to refect an age of electronic data
exchange in which large volumes of data and cloud computing, among
other developments, are changing the concept of service delivery and
could require a different industrial policy instrument. Just as the spread
of mobile 2G services led to the Basic Telecommunications Agreement
and the spread of the Internet resulted in the Information Technology
Agreement within the framework of the WTO, the consolidation of the
connected world could create the need for a different global agreement to
address emerging industrial policy issues.
Finally, the digital divide will, in the future, open a broadband
industrial policy gap that will cause countries lacking such a policy to
lag behind in terms of economic development and growth opportunities.
The conventional analysis of industrial policy in Latin America and East
Asia in 1960-1980 centred on the difference between import-substitution
industrialization and export-oriented industrialization. Since 1990
the major difference has been the scope of the industrial policy space.
Similarly, the decisive difference in the coming decade will revolve around
economic development grounded in the existence —or lack thereof— of a
broadband industrial policy.
This chapter analyses broadband development in the Republic of
Korea from the perspective of broadband industrial policy. Following
this introduction, section B describes the dynamics and structure of
the broadband market. Section C tracks the industrial policy space
in the 1990s and 2000s in the context of broadband industrial policy.
Broadband in Latin America: Beyond Connectivity 223
Section D examines the challenges presented by the development of
hyperconnectivity and outlines the Giga Korea Plan 2020 for facing these
challenges by fne-tuning broadband industrial policy. The fnal section
sets out conclusions and useful lessons for Latin America.
B. Broadband structure and dynamics
1. Expansion and mass adoption
This is one of the main requirements for developing markets, attracting
investment, adopting new technologies and ensuring pluralism in
services and content. This section provides statistical data on broadband
development trends in the Republic of Korea. This country ranks ffth
among the OECD economies in fxed-line broadband subscriptions
(wired) per 100 inhabitants, with 35% of the population connected to
broadband services (see fgure VIII.1). By type of technology, fbre optic
cable and local area network (LAN) connectivity top the list, with 20%
of the population connected —fve to ten times more than the OECD
average. The fact that fbre optic connectivity in the Republic of Korea is
double that of Sweden means that the former leads the world in terms of
broadband connection speed.
Figure VIII.1
OECD: FIXED-LINE BROADBAND SUBSCRIPTIONS (WIRED) PER 100
INHABITANTS, BY TECHNOLOGY, DECEMBER 2011
(Percentages)
OECD average
0
5
10
15
20
25
30
35
40
DSL Cable Fibre/LAN Other
S
w
i
t
z
e
r
l
a
n
d
N
e
t
h
e
r
l
a
n
d
s
D
e
n
m
a
r
k
F
r
a
n
c
e
N
o
r
w
a
y
R
e
p
.

o
f

K
o
r
e
a
I
c
e
l
a
n
d
U
n
i
t
e
d

K
i
n
g
d
o
m
G
e
r
m
a
n
y
L
u
x
e
m
b
o
u
r
g
S
w
e
d
e
n
B
e
l
g
i
u
m
C
a
n
a
d
a
F
i
n
l
a
n
d
U
n
i
t
e
d

S
t
a
t
e
s
J
a
p
a
n
N
e
w

Z
e
a
l
a
n
d
A
u
s
t
r
i
a
E
s
t
o
n
i
a
I
s
r
a
e
l
A
u
s
t
r
a
l
i
a
S
p
a
i
n
S
l
o
v
e
n
i
a
I
t
a
l
y
I
r
e
l
a
n
d
G
r
e
e
c
e
P
o
r
t
u
g
a
l
H
u
n
g
a
r
y
C
z
e
c
h

R
e
p
.
P
o
l
a
n
d
S
l
o
v
a
k

R
e
p
.
C
h
i
l
e
M
e
x
i
c
o
T
u
r
k
e
y
O
E
C
D
Source: Organisation for Economic Cooperation and Development (OECD).
ECLAC 224
Another important factor is that high-speed broadband in the
Republic of Korea began in the pioneering phase of broadband deployment
in the early 2000s. Figure VIII.2 shows how far in the lead the Republic
of Korea was in disseminating broadband between 2001 and 2005. In
that period, its broadband policy differed from the OECD norm: whereas
average broadband connectivity in the OECD was largely market-driven,
in the Republic of Korea it was driven by government policy. Case studies
of the Korean experience need to take this factor into account.
Figure VIII.2
SELECTED OECD COUNTRIES: FIXED BROADBAND PENETRATION
(Percentages)
Switzerland Netherlands Denmark Norway Rep. of Korea Iceland Sweden
0
5
10
15
20
25
30
35
40
45
2
0
0
1
-
Q
2
2
0
0
1
-
Q
4
2
0
0
2
-
Q
2
2
0
0
2
-
Q
4
2
0
0
3
-
Q
2
2
0
0
3
-
Q
4
2
0
0
4
-
Q
2
2
0
0
4
-
Q
4
2
0
0
5
-
Q
2
2
0
0
5
-
Q
4
2
0
0
6
-
Q
2
2
0
0
6
-
Q
4
2
0
0
7
-
Q
2
2
0
0
7
-
Q
4
2
0
0
8
-
Q
2
2
0
0
8
-
Q
4
2
0
0
9
-
Q
2
2
0
0
9
-
0
4
2
0
1
0
-
Q
2
2
0
1
0
-
Q
4
2
0
1
1
-
Q
2
2
0
1
1
-
Q
4
Source: Organisation for Economic Cooperation and Development (OECD).
The extraordinary development of broadband in the Republic
of Korea is also refected in the ratio between the penetration rate (per
100 inhabitants) and GDP per capita (in PPP dollars). There is a marked
correlation (on the order of 65%) between these variables in the OECD
economies. As shown in fgure VIII.3, which gives the ratio between the
two variables for 2011, the Republic of Korea is clearly in the lead and only
Estonia is close behind.
With regard to relative connection cost and quality, fgure VIII.4
shows that the Republic of Korea offers the most affordable broadband
services in the world, at a minimum price of 0.21 PPP dollars per Mbps
and a maximum price of 1.93 PPP dollars per Mbps in September 2011. In
other countries such as Germany and Chile, the same service costs 2 to
20 times more. Sweden offers a more affordable service at one extreme,
but the maximum rate is at least ten times greater. The ramifcations for
industrial policy are substantial given the impact of broadband on growth,
employment and productivity (see the chapter by Raúl Katz herein).
Broadband in Latin America: Beyond Connectivity 225
Figure VIII.3
OECD: RATIO BETWEEN BROADBAND PENETRATION PER 100 INHABITANTS
AND GDP PER CAPITA IN PPP DOLLARS, 2011
L
u
x
e
m
b
o
u
r
g
I
r
e
l
a
n
d
S
l
o
v
a
k

R
e
p
.
N
o
r
w
a
y
U
n
i
t
e
d

S
t
a
t
e
s
A
u
s
t
r
a
l
i
a
C
z
e
c
h

R
e
p
.
A
u
s
t
r
i
a
T
u
r
k
e
y
C
h
i
l
e
I
t
a
l
y
M
e
x
i
c
o
P
o
l
a
n
d
S
p
a
i
n
S
w
e
d
e
n
F
i
n
l
a
n
d
C
a
n
a
d
a
S
w
i
t
z
e
r
l
a
n
d
G
r
e
e
c
e
J
a
p
a
n
P
o
r
t
u
g
a
l
B
e
l
g
i
u
m
G
e
r
m
a
n
y
S
l
o
v
e
n
i
a
N
e
w

Z
e
a
l
a
n
d
N
e
t
h
e
r
l
a
n
d
s
D
e
n
m
a
r
k
I
s
r
a
e
l
U
n
i
t
e
d

K
i
n
g
d
o
m
I
c
e
l
a
n
d
H
u
n
g
a
r
y
F
r
a
n
c
e
E
s
t
o
n
i
a
R
e
p
.

o
f

K
o
r
e
a
0.0000
0.0002
0.0004
0.0006
0.0008
0.0010
0.0012
0.0014
Source: Organisation for Economic Cooperation and Development (OECD).
Note: The simple correlation between the variables is 0.65. The penetration rate is for December 2011.
Figure VIII.4
BROADBAND RATES PER MBPS OF ADVERTISED SPEED, SEPTEMBER 2011
(PPP dollars)
4.65
1.68
1.64
1.56
1.39
1.10
1.09
0.92
0.89
0.78
0.77
0.70
0.67
0.66
0.65
0.64
0.64
0.63
0.61
0.58
0.48
0.47
0.46
0.43
0.42
0.40
0.38
0.34
0.31
0.29
0.28
0.21
0.10
41.14
55.91
40.26
193.38
28.18
71.49
71.70
20.66
84.76
11.55
40.40
74.27
6.63
52.57
8.89
81.14
15.91
7.90
4.20
22.26
3.84
182.00
172.57
21.01
30.88
72.13
46.21
5.73
35.73
18.16
16.34
1.93
19.22
33.00
0 1 10 100 1000
0.06
Mexico
Greece
Chile
New Zealand
Israel
United States
Spain
Norway
Turkey
Luxembourg
Belgium
Switzerland
Iceland
Ireland
Netherlands
Italy
Czech Rep.
Australia
United Kingdom
Austria
Denmark
Poland
Portugal
Slovak Rep.
Germany
Canada
Slovenia
France
Hungary
Finland
Estonia
Rep. of Korea
Sweden
Japan
Source: Organisation for Economic Cooperation and Development (OECD).
ECLAC 226
The Republic of Korea ranks frst in the world in broadband use by
the private sector and service companies: almost 100% use IP services for
management and transaction purposes (see fgure VIII.5). This is an advantage
that is recognized and valued by businesses and helps build support for
industrial policy in favour of broadband-based products and services.
Figure VIII.5
BUSINESS ADOPTION OF BROADBAND, 2010 OR LATEST AVAILABLE YEAR
(Percentage of frms with 10 employees or more)
R
e
p
.

o
f

K
o
r
e
a

(
2
0
0
9
)
S
w
i
t
z
e
r
l
a
n
d

(
2
0
0
8
)
A
u
s
t
r
a
l
i
a

(
2
0
0
9
)
S
p
a
i
n
I
c
e
l
a
n
d
N
e
w

Z
e
a
l
a
n
d
C
a
n
a
d
a

(
2
0
0
7
)
F
i
n
l
a
n
d

(
2
0
0
9
)
F
r
a
n
c
e
S
w
e
d
e
n
N
e
t
h
e
r
l
a
n
d
s
G
e
r
m
a
n
y
T
u
r
k
e
y
B
e
l
g
i
u
m

(
2
0
0
9
)
L
u
x
e
m
b
o
u
r
g

(
2
0
0
9
)
E
s
t
o
n
i
a
U
n
i
t
e
d

K
i
n
g
d
o
m
S
l
o
v
e
n
i
a
N
o
r
w
a
y
I
r
e
l
a
n
d
D
e
n
m
a
r
k
C
z
e
c
h

R
e
p
.
E
U
2
7
P
o
r
t
u
g
a
l
I
t
a
l
y
A
u
s
t
r
i
a
G
r
e
e
c
e
J
a
p
a
n
H
u
n
g
a
r
y
S
l
o
v
a
k

R
e
p
.
P
o
l
a
n
d
M
e
x
i
c
o

(
2
0
0
8
)
0
25
50
75
100
Source: Organisation for Economic Cooperation and Development (OECD).
Note: For Japan, frms with 100 employees or more. For Mexico, frms with 20 employees or more. For
New Zealand, frms with 6 employees or more and sales in excess of US$ 30,000. For Switzerland, frms
with 5 employees or more and a connection speed of 144 Kilobits per second or faster (mobile and fxed).
2. Drivers of mass penetration
Several studies explore the drivers of high-speed broadband infrastructure
penetration. The following analysis focuses on studies covering 20 to
25 OECD countries because these countries account for most of the
broadband deployed in the world. These studies, however, do not capture
the factors driving broadband penetration in the BRICs, which will be
equally important in the future due to the size of their digital economies.
The main factors explaining the degree of high-speed broadband
deployment include price of service, transmission speed, population
density (urban areas), GDP per capita, average user age, competition
among platforms, unbundling regulations, other government regulations
and social and cultural factors.
The factors behind successful high-speed broadband infrastructure
deployment are the same in most OECD economies. In table VIII.1, this
Broadband in Latin America: Beyond Connectivity 227
is illustrated with the blue circle around price of service, transmission
speed and GDP per capita, among other variables. These cost factors are
positively correlated (or negatively correlated in the case of price) with
broadband development.
Table VIII.1
DRIVERS OF BROADBAND DEVELOPMENT IN THE OECD
AND THE REPUBLIC OF KOREA
Case study Factor variables
Price of
service
Transmission
speed
Population
density
(urban
areas)
GDP
per
capita
Number of
users
Average
user age
Competition
among
platforms
Unbundling
regulations
Other
government
regulations
Social
and
cultural
factors
OECD
case
studies
(including
the
Republic
of Korea)

Yoon and
Byun (2008)
- + + + +
Atkinson,
Correa and
Hedlund
(2008)
- + + + (-) (+)
Wallesten
(2006)
+ + +/-
Cava-
Ferreruela
and Alabau-
Muñoz
(2006)
(-) + + + +
Berkman
(2010)
(Inter:+
Intra: -)
(+/-)
Grosso
(2006),
de Ridder
(2007)
+ +
Höffer
(2007)
+
Waverman
and others
(2007)
+
Korean
case
studies:
unique
factors
Aizu (2002), +
UIT (2003), +
Lau, Kim and
Atkin (2005),
+
Ovum
Consulting
(2009), Kim,
Kelly and
Raja (2010),
+ + +
Berkman
(2010)
+ +
Source: Prepared by the author.
ECLAC 228
The cost factor is largely related to competition policy: the degree of
competition that it spurs depends on government regulations and policy.
Therefore, the government plays a key role in establishing an appropriate
regulatory environment that leads to healthy competition. This is illustrated
by the green circle in table VIII.1. In the Republic of Korea, the government
initially used competition among platforms to speed broadband deployment
after communication services were liberalized in the mid-1990s. On
achieving this objective in the 2000s, the government shifted its position to
service-based competition in order to improve service quality. The authors
of a number of studies agree that the government’s regulatory framework
is fundamental for increasing competition and managing market effciency.
Most studies fnd that competition among platforms is positively correlated
with broadband use, while analysts such as Berkman (2010) warn of an
inverse correlation with excessive competition within platforms.
In addition to these variables, several studies explore factors that can
explain high-speed broadband penetration in the Republic of Korea but are
not present in other countries. These studies emphasize two key factors: the
government’s strategic policy with a fexible regulatory framework, and the
sociocultural factor. These are shown in the red circle in table VIII.1.
Most OECD governments have institutionalized broadband
strategies and policies, but the speed and regularity with which the
Korean government has revised its stance by designing new regulatory
frameworks to respond to technological and market innovation are fairly
unusual, if not unique.
ICT laws have enjoyed bipartisan support, and the quick succession
of laws and regulations helped the government respond quickly to
market requirements. For example, Lau, Kim and Atkin (2005) and Ovum
Consulting (2009) argue that in addition to competition among platforms
and services, the government has used population dynamics and tastes
and the construction industry (buildings, apartments and offces) as part
of the broadband infrastructure. The certifcation mechanism for buildings
equipped with high-speed Internet attracted new residents, served as a
guarantee of quality and provided an incentive for residential and business
rentals. The certifcation system worked very well; almost all buildings
now display the “broadband certifcation mark” in order to attract clients.
Atkinson, Correa and Hedlund (2008) and Kim, Kelly and Raja (2010)
agree that the Korean government’s strategic conception and regulatory
framework have been excellent. Not only has it maintained a consistent
leadership position over the years, but it has applied its policy instruments
fexibly. Since 1995, the government has systematically designed strategic
frameworks every two years, changing its policy instruments in
accordance with the prevailing demands of technological innovation and
the market environment.
Broadband in Latin America: Beyond Connectivity 229
3. Market structure
To analyse the structure of the broadband-based services market, this
section describes the government policy that led to the transition from
a monopoly to a duopoly, from a duopoly to three-way competition and
fnally to broad-based competition, including mobile virtual network
operators (MVNOs).
The government encouraged competition in fxed-line telephony
services in the early 2000s, which resulted in three-way competition
between KT, SK and LGU+. The near-monopolistic position of KT as the
incumbent supplier of fxed-line services fell from 90% in 2007 to 83%
in 2011 (see fgure VIII.6). Despite the policy for demonopolizing fxed-
line services, other private actors were not much interested in offering
this class of services because most of the value added is generated in the
mobile sector. According to 2012 data, the joint market share of the other
two suppliers of fxed-line services was less than 15%.
Figure VIII.6
FIXED-LINE TELEPHONE SUBSCRIBERS BY SERVICE PROVIDER
(Per 10,000 people and in percentages)
0
500
1 000
1 500
2 000
2 500
2007 2008 2009 2010 2011
Jun
2011
Jul
2011
Aug
2011
Sep
2011
Oct
2011
Nov
2011
Dec
KT SK LGU+
0
10
20
30
40
50
60
70
80
90
100
2007 2008 2009 2010 2011
Jun
2011
Jul
2011
Aug
2011
Sep
2011
Oct
2011
Nov
2011
Dec
KT SK LGU+
Source: Korea Information Society Development Institute (KISDI) (2012).
ECLAC 230
Figure VIII.7 reveals an uptrend in IP-based services, with an
increase in broadband that offsets the drop in fxed lines. With regard
to the number of mobile service subscribers, coverage approached 100%
in 2010 and exceeded 110% en 2012, with more than 30 million people
using smartphone services. This means that there are three mobile phone
users for every fxed-line user and that smartphones have become the
cornerstone of mobile connectivity. All mobile service users are expected
to be using smartphones by 2015. Because the mobile virtual network
operators currently offer VoIP free of charge, consumers have an incentive
to adopt smartphone services.
Figure VIII.7
NUMBER OF SUBSCRIBERS BY SERVICE AND MOBILE TECHNOLOGY
(10,000 people)
Local fixed line Mobile Broadband
0
2 000
3 000
4 000
5 000
6 000
1 000
2007 2008 2009 2010 2011
Jun
2011
Jul
2011
Aug
2011
Sep
2011
Oct
2011
Nov
2011
Dec
0
1 000
2 000
3 000
4 000
5 000
6 000
2009 2010 2010 2010 2010 2011 2011 2011 2011
3G smartphones 2G telephones
Jun Sep Dec Mar Jun Sep Dec Mar
Source: Korea Information Society Development Institute (KISDI) (2012).
Among mobile service providers, SK has the largest market share,
with around 27 million customers, followed by KT, with about 17 million,
and LGU+, with less than 10 million (see fgure VIII.8). More than half of
Broadband in Latin America: Beyond Connectivity 231
all mobile service users are with SK; the rest are divided between KT and
LGU+. LGU+ frst promoted LTE-based 4G services in early 2012; SK and
KT followed suit immediately. By late 2012, there was a massive transition
to this type of technology. Another interesting aspect of 4G services is the
upsurge in mobile virtual network operators, who lease the network from
the incumbent provider and offer proftable, value-added services, often at
a lower rate.
The hypermarket chain HomePlus, an affliate of the British
company Tesco, entered into negotiations with KT for the provision
of mobile virtual network operator services, including near-feld
communication and automatic payment services. It is expected that within
a few years, several value-added service providers will have a considerable
share of the mobile virtual network operator market, with a variety of
services targeting specifc clients that use high-speed 4G connectivity.
Figure VIII.8
MOBILE SERVICE PROVIDERS: NUMBER OF SUBSCRIBERS AND MARKET SHARE
(Per 10,000 people and in percentages)
0
500
1 000
1 500
2 000
2 500
3 000
2007 2008 2009 2010 2011
Jun
2011
Jul
2011
Aug
2011
Sep
2011
Oct
2011
Nov
2011
Dec
SK KT LGU+ MVNO
0
10
20
30
40
50
60
70
80
90
100
2007 2008 2009 2010 2011
Jun
2011
Jul
2011
Aug
2011
Sep
2011
Oct
2011
Nov
2011
Dec
SK KT LGU+ MVNO
Source: Korea Information Society Development Institute (KISDI) (2012).
ECLAC 232
In addition to mobile virtual network operators, there is a growing
share of high-speed broadband connection providers (see fgure VIII.9). The
proportion of medium-sized broadband service providers is growing; they
have overtaken SK and LGU+. In 2011, they ranked second among high-
speed broadband providers and accounted for more than 20% of the market.
Figure VIII.9
HIGH-SPEED BROADBAND PROVIDERS: NUMBER OF
SUBSCRIPTIONS AND MARKET SHARE
(Per 10,000 people and in percentages)
0
100
200
300
400
500
600
700
800
900
2007 2008 2009 2010 2011
Jun
2011
Jul
2011
Aug
2011
Sep
2011
Oct
2011
Nov
2011
Dec
SK KT LGU+ Others
0
10
20
30
40
50
60
70
80
90
100
2007 2008 2009 2010 2011
Jun
2011
Jul
2011
Aug
2011
Sep
2011
Oct
2011
Nov
2011
Dec
KT SK LGU+ Others
Source: Korea Information Society Development Institute (KISDI) (2012).
Since 2012, LAN services have become the primary broadband
service line, followed by hybrid fbre coaxial (HFC) networks and fbre-to-
the-home (FTTH). Since 2010, LAN and FTTH connections have reached
a turning point in broadband development and, together with HFC
networks, represent over 80% of the connection technologies in use. By
contrast, the share of connections via digital subscriber lines (xDSL) has
fallen to a secondary level (see fgure VIII.10). By 2015, LAN and FTTH are
expected to account for over 90% of broadband connectivity, keeping the
country at the forefront of world high-speed broadband development.
Broadband in Latin America: Beyond Connectivity 233
Figure VIII.10
HIGH-SPEED BROADBAND COVERAGE SUBSCRIBERS BY TECHNOLOGY
(Per 10,000 people and in percentages)
0
100
200
300
400
500
600
700
2007 2008 2009 2010 2011
Jun
2011
Jul
2011
Aug
2011
Sep
2011
Oct
2011
Nov
2011
Dec
xDSL LAN HFC FTTH
0
10
20
30
40
50
60
70
80
90
100
2007 2008 2009 2010 2011
Jun
2011
Jul
2011
Aug
2011
Sep
2011
Oct
2011
Nov
2011
Dec
xDSL LAN HFC FTTH
Source: Korea Information Society Development Institute (KISDI) (2012).
4. ICT goods production and foreign trade
There is a strong correlation between broadband development and the
ICT industry in the Republic of Korea. This was deliberately fostered
by government policy in order to take full advantage of the broadband
infrastructure so as to create a market for the new ICTs. In general, the
Korean ICT industry surged in 2005-2011, growing by 11.7% in 2005, 8.7%
in 2007 and 17.7% in 2011. This latter fgure shows that the industry was
not long affected by the 2008 global recession. The growth pattern for the
ICT sector also stands in contrast to other sectors, which grew at annual
rates of less than 4% in 2005-2011.
The ICT industry´s share in GDP reached 11.4% in 2011, versus 8.6%
in 2005 and 9.5% in 2007. During implementation of the e-Korea Plan (2005),
ICT production was less than US$ 33.0 billion; it surged to US$ 49.7 billion
under the Smart-Korea Plan (2011). According to a study of 500 companies
in the sector, in the period comprising the u-Korea Plan (2007) and the
ECLAC 234
Smart-Korea Plan (2011), income increased 49%, the workforce grew 18%
and R&D spending on ICT convergence rose 72%. The global market
share of smartphones produced in the Republic of Korea grew tenfold in
just a few years, from a mere 2% in 2007 to 24% in 2011. That same year,
Korea’s Samsung and LG held a 57.8% share of the smartphone market
in the United States. Worldwide, Samsung (21.6%) and LG (3.4%) led the
mobile phone market in 2012, followed by Nokia (19.9%), Apple (6.9%) and
the Chinese frms ZTE (4,3%), Huawei (2.6%) and TCL (2.2%) (Chosun Daily,
21 August 2012, B3). If this trend continues, the Republic of Korea and
China will account for half of all smartphone production and sales in the
world in 2015.
With regard to trade, the Korean ICT industry exports US$ 15.0
billion a month, on average, and imports US$ 7.0 billion, maintaining a
favourable trade balance of between US$ 70.0 billion and US$ 80.0 billion
a year. As shown in table VIII.2, the ICT sector trade surplus was over
US$ 35.0 billion in the frst half of 2010 and 2011 and stood at some
US$ 70.0 billion at the end of each year. This demonstrates the sector’s solid
export performance, which has contributed to a favourable trade balance in
the past few decades.
Table VIII.2
ICT SECTOR TRADE BALANCE, 2010-2011
(Millions of dollars and growth rates)
First half 2010 First half 2011
Exports All industries 275 340
(24.4%)
221 320
(34.3%)
ICT industry 77 370
(6.3%)
72 810
(38.4%)
Imports All industries 257 970
(26.6%)
203 790
(40.2%)
ICT industry 40 580
(15%)
35 290
(25.6%)
Trade balance All industries 17 370 17 540
ICT industry 36 800 37 520
Source: Ministry of Knowledge Economy (MKE) (2012).
Notes: Data for 2011 are estimates. Growth rates are in parentheses.
The main export products of the Korean ICT industry include
semiconductors, display panels, cellular phones (including smartphones),
television sets, computers and computer parts and home devices. In
recent years, smartphone exports have posted double-digit growth. They
were followed by computers and related devices (such as solid-state
drives), which rivalled other major traditional export products such as
Broadband in Latin America: Beyond Connectivity 235
semiconductors and computer devices. Foreign sales of memory chips
contracted 15.8% in 2011; systems-on-chip (SoC) exports increased 14.8%,
to US$ 1.52 billion.
In August 2012, executives at Samsung decided to convert their main
semiconductor factory in the United States into a dedicated production
line for systems on chips, mainly semiconductors for smartphone chips.
This refects a shift from an industrial policy based on PCs to one based
on broadband. In anticipation of IP-based digital television services, 3D
and smart television exports have increased, while television component
exports have fallen considerably in recent years. With the slowdown of the
world economy, exports of display panels also decreased by 8.2%.
In recent years, the structure of the country’s trading partners in
the ICT sector has changed substantially. The United States still represents
a large share of the market (27.4% of exports of home devices), but sales
to that country have fallen. By contrast, trade with China has expanded,
and it is now twice the level of trade with the United States and Japan
combined. Sales of home devices to other emerging economies grew 14%,
with an increase in exports to Eastern Europe (91.4%) and Latin America
(52.1%) due to the rising demand for ICT products and related components.
As part of a shared production strategy at the world level and to avoid
protectionist trade barriers such as the tax on industrialized products
in Brazil, some Korean companies are moving their production base to
Latin America. In 2011, Brazil received its frst Korean investment for the
production of semiconductors, which heralds a new era of ICT production
in that country and could be a frst step towards overcoming the past
decade’s lag in attracting foreign direct investment in the sector. More
foreign direct investment is expected to fow into Latin America to tap the
dynamic ICT market, opening space for intra-industry trade policies with
supplier countries like China, the Republic of Korea and Japan, as well as
the current partners: Europe and the United States.
C. Policy convergence
This section describes the Republic of Korea’s broadband industrial
policy. It suggests that the conventional broadband development policy
commonly recommended by the OECD and case studies of Korea based on
conventional factors such as population density, gross domestic product
(GDP) per capita and cultural trends might be insuffcient for explaining
the driving factors behind Korea’s policy on ultra-broadband development.
This study focuses instead on government measures for implementing
an industrial policy aimed at using broadband not only as a consumer
good but also as a means of achieving a higher standard of industrial and
technological development.
ECLAC 236
From a trade policy perspective, broadband development should
be a natural outcome of market-driven competition, privatization and
liberalization, which would translate into an appropriate price. This has, in
fact, been the case with value-added services in second-generation mobile
connectivity, which has provided connectivity access for more than half
the world population. From the perspective of industrial policy, however,
broadband development has been used as an intensive government policy
geared towards encouraging technological innovation in order to lead the
competition and adjust prices.
This section demonstrates the limits of conventional studies on
broadband development, explaining why the conventional wisdom behind
Korea’s “broadband magic” has not been replicated in other developing
countries with similar conditions in terms of population density (for
example, Bangladesh), GDP per capita (Portugal or Spain), local loop
unbundling (LLU) (most OECD countries), video game addiction (Brazil),
digital literacy (the Netherlands) and addiction to speed (almost all
countries), among other factors.
In terms of key variables, conventional studies on broadband
development emphasize the trade policy perspective, whereas this
study stresses the industrial policy viewpoint. This is shown in the
following equation.
Box VIII.1
BROADBAND INDUSTRIAL POLICY APPROACHES
A. Broadband development policy (BDP) approach:
BDP= f (trade policy based on connectivity, passive policy on
technological innovation, little convergence between broadband and
industry, user-based education)
B. Broadband industrial policy (BIP) approach:
BIP= f (industrial policy based on value added, active policy on
technological innovation, substantial convergence between broadband and
industry, producer-based education)
C. Synthetic proposition (SP):
SP=f (BDP+BIP)
Source: Prepared by the author.
While most of the factors considered decisive for the successful
launching of world-class broadband infrastructure are well-founded,
this chapter addresses a dimension of broadband-based economic
industrial policy that acted as both a driving force and a drag on use of
the network for deployment of new classes of industrial products and
digital services. Broadband industrial policy does not necessarily generate
Broadband in Latin America: Beyond Connectivity 237
the best infrastructure for the economy, so this causality should be closely
examined. Nonetheless, it is proposed herein that in the case of the
Republic of Korea, there is, at a minimum, a strong correlation between
broadband development policy and broadband industrial policy. To
support this argument, the next section reviews the literature on broadband
development in the OECD countries, including the Republic of Korea.
1. Broadband development policy
As seen above, broadband development in the Republic of Korea was
intentionally and aggressively driven by government policy. Among the
key elements of that policy, the government provided a vision, a strategy
and, in particular, the following master plans: the National Strategic
Plan on Broadband Infrastructure (1995); the Cyber Korea 21 Plan (2000);
the e-Korea Vision (2005); the u-Korea Plan (2006); and the Giga Korea
Plan (2011). Table VIII.3 summarizes these plans as well as the strategic
framework for promoting broadband development between 1995 and 2012.
Table VIII.3
REPUBLIC OF KOREA: MASTER PLANS AND STRATEGIC
FRAMEWORKS FOR BROADBAND DEVELOPMENT
Year Strategy and main instruments
Underlying
technologies
Broadband
speed objective
1995 National strategic plan on
broadband infrastructure
ATM, ADSL, cable
modem
1Mbps
1999 Certifcation of buildings
equipped with broadband
1999-2002 Cyber Korea 21 2G: VDSL, FTTH,
FTTB, W-CDMA
10 Mbps
2001-2005 Third revision of the strategic
plan on broadband infrastructure
2002-2006 e-Korea Vision 2006 3G: WiBro, HSDPA,
FTTH, FTTB
50 Mbps
2003-2007 Broadband IT Korea Vision 2007
2004-2010 u-Sensor Network (USN) Plan
2004-2010 Broadband Convergence
Network (BcN) Plan
2006 u-Korea Plan FTTH, FTTB, WiBro,
W-CDMA, HSDPA
100 Mbps
2008
Basic plan for national
informatization
2011-2020 Giga Korea Plan 4G: LTE (TD-LTE) 1 Gbps
Source: Prepared by the author, on the basis of data from the Korea National Computerization Agency
(master plans highlighted in colour).
ECLAC 238
The World Bank (2009) stresses that, in addition to promotion policy,
the Republic of Korea’s oversight laws and regulations (which could be
emulated in developing countries to accelerate broadband penetration)
have played an important role. These regulatory and legal frameworks
for oversight include the Telecommunications Framework Law, the
Telecommunications Trade Law, the Regulatory Law on Fair and Equitable
Trade Practices, regulation of network access, regulation of Internet service
provider (ISP) peering, local loop unbundling, regulation of signifcant
market power, number portability for voice over Internet protocol (VoIP),
digital rights management, regulation of cybersecurity and digital crimes,
and the ethical use of information.
The government was also active in terms of policies to promote
universal broadband access, including the following: master plans for
closing the digital divide; rural connectivity strategy; low-interest loans for
rural development; subsidized services for citizens living in poverty, older
adults and persons with disabilities; free Internet access centres in rural
and remote areas; and broadband access in all schools, even in rural areas.
Some authors hold that the users’ cultural milieu could have
contributed to the fast deployment of broadband, underscoring their
willingness to accept new technologies. According to the International
Telecommunication Union (2003), the youngest generations of Koreans
are very connected to the communications media culture, movies, online
games, music and digital entertainment; this makes for a favourable
environment for the fast deployment of high-speed Internet. According
to Lau, Kim and Atkin (2005), the Internet has brought the stock market
into homes and turned online home banking and home trading systems
(HTS) into an instrument for personal asset management. Since the early
2000s, 67% of stock market transactions in the country are carried out
online, which illustrates the large demand for this type of application that
is contributing to the fast growth of the broadband network.
2. Competition and regulation
One of the key factors in the development of a world-class broadband
system in the Republic of Korea is the fexible use of regulatory policies to
adapt to changing technology and market trends. After the liberalization
conducted at the urging of the WTO, the government effectively used
competition policy as part of its carefully planned and structured incentives
for broadband development. Table VIII.4 summarizes the main competition
policy instruments for broadband development, from breaking up the
telecommunications service monopoly to the policy on competition among
platforms and, subsequently, on service-based competition.
Broadband in Latin America: Beyond Connectivity 239
Table VIII.4
BROADBAND REGULATORY POLICIES
Year Policies and regulations
Infrastructure
vs. services
Before
1982
Telecommunications services provided by MPT
Telecommunications monopoly
Monopoly
1982 KT founded as a public corporation
1984 Dacom founded;
KT launches analogue mobile services
1988 KMT spun off of KT
1991 Promotion of competition in value-added services;
Dacom enters international market
1992 Competition in localization services
1994 Second cellular licence issued (Sinsegi Telecom);
Order issued on accounting separation
1995 Competition in the long-distance market (Dacom);
Launch of cable television services, with KEPCO and KT as
operators;
Establishment of the WTO;
Managed competition between infrastructures launched
Platform-based
broadband
development
policy
1996 27 new licences granted; 3 personal communications services;
6 trunking radio systems; 1 Cd-2 second-generation wireless
telephone;
2 dedicated lines; 1 localization; third international operator (Onse);
3 wireless data transmissions
1997 10 new licences granted; one local operator, one long-distance
operator, six TRS, one localization;
The WTO Basic Telecommunications Agreement enters into effect;
Introduction of resale-based competition
1999 Hanaro starts bulding telephone and local broadband Internet
services;
Universal service regulation;
KT launches broadband Internet service
2001 Launch of DBS (Skylife);
Regulation of local loop unbundling;
Rebalancing of KT’s local tariff
2003 Regulation of interconnection
2004 Revision of the regulations on competition among infrastructures
and the move to service-based competition;
Broadband Internet redefned as basic service;
Introduction of number portability
Service-based
broadband
development
policy
2005 Two-year prohibition of the mobile telephone subsidy
2006 Launch of WiBro and high-speed downlink packet access (HSDPA)
services;
Establishment of KCC (Korea Communications);
2007 Authorization of bundling services by the main operators;
Temporary regulation on the elimination of the mobile telephone
subsidy
2008 Regulation of number portability for VoIP
2010 Regulation on new lease service providers for mobile virtual
network operators (MVNO)
Convergence-
based
broadband
development
policy
2012 Launch of Giga Korea Plan
Source: Prepared by the author, on the basis of Korea Information Society Development Institute (KISDI) (2012).
ECLAC 240
Three main characteristics emerge. First, the government effectively
used competition in the private sector for granting licences, following
an approach geared towards improving services through facilities for
accessing new infrastructure. Second, the government did not rely solely on
private-sector leadership to create a world-class broadband infrastructure,
because the private sector has always tended to seek returns without
making new investments in higher-level infrastructure. The authorities
managed to compel the private sector to invest in the next generation
of broadband services. Third, the government utilized the broadband
infrastructure to boost production in the ICT industries —including IP-
based devices such as headphones, displays, digital television, radio-
frequency identifcation (RFID), near feld communication (NFC) and so
on— within a broader industrial policy scheme.
3. Broadband industrial policy
The Republic of Korea is unique in its transition from conventional
industrial policy to broadband-based industrial policy. The development
of the chemical and heavy manufacturing industries in the 1970s and
1980s is a classic example of “getting the price wrong” in line with
Amsden (1995), with a set of policies and instruments aimed at selectively
protecting domestic industry.
In the mid-1990s, the Republic of Korea joined the OECD in the area
of capital account convertibility. It later came into compliance with WTO
regulations prohibiting the use of direct subsidies in the framework of the
Agreement on Subsidies and Countervailing Measures. With the end of
the controlled fnancial system era, the infux of short-term capital along
with current account defcits triggered the fnancial crisis of 1997.
The industrial policy model has been under review since the
creation of the WTO in the mid-1990s. In this process, industrial policy for
information technologies was combined with communications policy to
take full advantage of its potential. This was achieved through measures to
increase Internet use as a fundamental tool in the promotion of four major
components of broadband-based development: communication networks,
device industry, platform technologies and content (see diagram VIII.2).
These four components have made up broadband industrial policy
since the advent of commercial Internet in the mid-1990s. This policy has
gone through two phases of development; the third phase has only just
begun. The frst began with CDMA technology in the early 1990s, just
before the WTO agreement entered into effect. The successful development
of this technology and the commercialization of the CDMA standard
contributed a good deal to the dissemination of mobile broadband and
Broadband in Latin America: Beyond Connectivity 241
related devices. The CDMA initiative was under the direct supervision
of the Electronics and Telecommunications Research Institute (ETRI), a
government research agency, with the collaboration of the private sector
and other major research centres.
Diagram VIII.2
MAIN AREAS OF BROADBAND INDUSTRIAL POLICY
Platform Content
Devices Network
infrastructure
Communications
technology
policy
Information
technology
policy
Platform
policy
Applications
policy
Source: Prepared by the author.
An important advance in the second phase was the launch of WiBro
mobile broadband technology (a counterpart of WiMAX technology)
in the 2000s. The effective use of the mobile standard with the frst
commercialization of CDMA gave the industry a lot of room for promoting
technological innovations. It also provided justifcation for the development
of broadband trunking networks such as FTTH and FTTx. From this
perspective, the concentration of world-class broadband infrastructure in
the Republic of Korea cannot simply be attributed to network development
policy. As illustrated by the case of CDMA and WiBro technologies,
broadband deployment is closely tied to industrial policy.
The transition to 4G in the early 2010s marks the beginning of the
third phase of broadband industrial policy, when both CDMA and WiBro
saw decreasing returns to scale because of the slow rate of adoption, the
lack of commercial viability and the emergence of new technical standards.
As was the case with CDMA, the advantage of being the forerunner
in technological standards was considerably smaller in the second phase,
for three reasons. First, CDMA lost technical superiority over time and
LTE became the pioneer. Second, 4G technology, such as TD-LTE, has
emerged as the de jure ITU standard and the de facto standard of the
ECLAC 242
major companies. Third, the development and adoption of TD-LTE as the
4G standard in China give China a great advantage because its internal
market for 4G is potentially larger than the world market. In the face of
these challenges, the Korean authorities launched the Giga Korea Plan in
2011 and announced the corresponding action plan in 2012.
4. Technological development policy
The Technological Development Promotion Law was enacted in April 2009
to develop the national economy by increasing international competitiveness
through the creation and dissemination of new technologies.
2
The law
targets three areas for promotion: technological development, adaptation of
foreign technology and new technologies.
Technological development aims to create new methods applicable
to materials, manufactured products, processes and equipment systems,
among others, through industrial technology research and its outcomes,
including prototypes and pilot tests. Adapting and upgrading existing
technology involves activities to create new technologies by analysing,
complementing and improving on technologies in place in other countries.
New technologies include greenfeld technological development carried
out in the country and the assimilation and updating of imported
technology, certifed by the Ministry of Knowledge Economy.
Within the framework of this law, specifc R&D projects can
be designed to develop key industrial technologies, choose research
tasks annually and assign research programmes to institutions and
organizations. The next section discusses the Giga Korea Plan as an
example of broadband industrial policy in the context of technological
development promotion policy.
D. The Giga Korea Plan 2020
As mentioned earlier, broadband industrial policy has gone through
three phases: (i) CDMA-based mobile broadband policy in the 1990s;
(ii) WiBro-based 3G mobile broadband policy in the 2000s; and (iii) the
Giga Korea Plan based on 4G broadband industrial policy in the 2010s. The
combination of broadband development policy and broadband industrial
policy has exhibited a growing synergy, with revolutionary technological
advances and development opportunities impacting society as a whole.
3
2
See the technological development promotion law of the Republic of Korea (No. 9630)
enacted in April 2009.
3
Machina Research’s Mobile Broadband Global Forecast & Analysis 2010-20 report offers a full
and detailed analysis of global opportunities for mobile broadband services. The report
includes the following key findings. First, mobile broadband connections for personal and
Broadband in Latin America: Beyond Connectivity 243
In 2011 the government established the National Committee on
the Giga Korea Plan, made up of the ministries of education and science,
knowledge economy, security and public administration, culture,
and defence, as well as the diffusion and communication commission
and the national computerization commission. The Electronics and
Telecommunications Research Institute (ETRI) provided support in terms
of internal coordination. An inter-ministerial work division was established
to avoid unnecessary overlapping of tasks in the various ministries, while
taking into account that synergies depend on working together.
The government plans to establish an ICT hub for four sectors:
networks, devices, platforms and content. The goal is for there to be 50
national software companies by 2020. In the same period, the government
expects that the development of essential mobile broadband technology
for processing large volumes of data will generate revenue the order of
US$ 100 billion and create 700,000 jobs (Korea IT Times, 2012).
One of the goals of the Giga plan is to build a Gbps broadband
infrastructure before 2020 through fxed and wireless network
convergence. With this type of network, it will be possible to download
a 3D movie in less than three minutes, versus a current download time
of three hours. The plan also calls for development of a 32Tbps fexible
fve-stage WDM optical transmission system by 2020 and the production
of holographic devices and mobile 16-core 3D devices with speeds of
2.5 Ghz, which will allow a virtual reality generation interface capable of
processing hundreds of terabytes with minimal energy consumption.
A key element of the Giga plan is the development of ultra-fast
multiple cloud computing with holographic media processing that
requires a capacity equivalent to 10 million PCs, or double-digit Tbps, for
multiple heterogeneous platform services based on simultaneous data
transmission. Table VIII.5 summarizes the Giga Korea Plan 2020.
laptop computers (through USB modems, embedded modules and so forth) will increase
from 140 million at year-end 2010 to 1.5 billion in 2020, as a result of lower prices, increased
coverage and capacity and the lack of fixed alternatives for broadband provision in many
emerging markets. Second, mobile broadband will be increasingly data-capable. The
number of 3G+ and 4G connections will grow from 1.5 billion to 7.3 billion in 2020. The
use of e-book readers and tablets with wireless wide area network (WWAN) capability
will increase steadily in the mobile broadband environment, growing from 20 million in
2010 to 230 million in 2020. By 2020 the vast majority of these will be 4G+. Third, mobile
data traffic will increase from 2.3 exabytes in 2010 to 41 exabytes in 2020 (a 17-fold jump).
Finally, total revenue from WWAN data services for mobile telephones, PCs and laptop
computers and tablets is expected to approach US$ 400 billion in 2020, versus US$ 100
billion in 2010. In contrast to the trend for traffic, and due to the fact that the prices per
Mb are generally higher, this increased income will largely come from mobile telephone
services. This will create a large number of mobile network operators in the 4G era
(http://www.machinaresearch.com/mobilebroadband2020.html).
ECLAC 244
Table VIII.5
THE GIGA KOREA PLAN 2020
Classifcation Stages
Preliminary stage
(pre-2012)
Stages I and II
(2013-2017)
Stage III
(2018-2020)
Networks General
objective
Mbps wireless
network
500 Mbps
fxed-wireless
convergence
network
1Gbps fxed-
wireless
convergence
network
Example:
download a 3D
movie (25GB)
3 hours 7 minutes 3 minutes
Main network
access node
Optical packet
switching with
hundreds of Gbps
4Tbps fexible
three-stage WDM
optical transmission
system
32Tbps fexible fve-
stage WDM optical
transmission system
Devices General
objective
2K-level mobile
devices
4K 3D non-glass
mobile devices
8K 3D holographic
mobile devices
Data
processing
capacity
[email protected]
Mobile Core
[email protected]
Mobile Core
[email protected]
Mobile Core
Interface Multi-touch inputs Biodata recognition
interface technology
Virtual reality
generation interface
Data storage Hundreds of
gigabytes with
low power
consumption
Several Tbytes
with low power
consumption
Hundreds of Tbytes
with low power
consumption
Platforms General
objective
T (1012) IT
platform
(= 1,000 PCs)
Multiple UHD cloud
computing
(= 100,000 PCs)
Holographic media
processing
(2-digit Tbps)
IT platform HD media
processing
(2-digit Gbps)
UHD media
processing
(3-digit Gbps)
Multiple ultra-fast
cloud computing
(= 10 million PCs)
Cloud
computing
service
Mono platform-
as-a-service
cloud computing
Multiple platform-
as-a-service cloud
computing
Multiple
heterogeneous
platform-as-a-
service cloud
computing
Content General
objective
3D content
processing
Biorecognition
content processing
Holographic content
processing
Display Glass-based 3D
HD
4K non-glass UHD 8K/3D digital
hologram UHD
Virtual reality CAVE-type virtual
reality
Interaction-based
virtual reality
Auto-progressive
virtual reality
Source: Prepared by the author, on the basis of data from the Ministry of Knowledge Economy (MKE) and
offcial documents.
Broadband in Latin America: Beyond Connectivity 245
E. Conclusions
This chapter has explored Korean broadband industrial policy as an
example of the creation of a national broadband infrastructure combined
with broadband-based industrial development. Previous studies by the
World Bank, the OECD and the ITU focused on broadband development
from the network perspective, without looking in depth at the development
of the ICT industry. This approach arises from seeing the ICT industry
from a horizontal perspective and considering broadband development as
a function of market-driven deregulation, privatization, competition and
network neutrality policies and policies targeting demand for applications
such as e-government and e-business.
While this approach is useful for many developing countries and
emerging economies that lack a strong ICT industry, other countries
have room for revising their broadband development policy from the
perspective of industrial policy. The case of the Republic of Korea
shows that the provision of networks is only one of the core elements of
broadband industrial policy, which also encompasses devices, platforms
and content.
Latin America missed out on the opportunity to take advantage
of the frst wave of broadband industrial policy.
4
This was mainly due
to the prevalence of integrated horizontal polices in the economic model
in the period following the debt crisis of the 1980s and the liberalization
process of the 1990s. Although the countries of Latin America are now
joining the second wave of broadband development, the link between
broadband development and industrial policy is a weak one in many
countries in the region.
This stands in sharp contrast to the Republic of Korea, where space
was made for targeted cutting-edge policies within the WTO regulations.
5

This chapter has examined the case of CDMA technology as an example
of broadband industrial policy, WiBro technology as another example of
3G mobile broadband industrial policy and the Giga Korea Plan as a fresh
initiative for 4G super-broadband industrial policy.
At the dawn of a new 4G era, action under the Giga Korea Plan has
turned to reducing reliance on CDMA and WiBro as technology continues
to advance. This third stage of broadband development, called the era
4
The governments of the countries of Latin America did not take full advantage of the
opportunities. There is some outsourcing of the assembly of PCs and components, mainly
in Mexico and Costa Rica, and some production for the domestic market in Brazil, but
their aggregate impact was modest. The lack of a broadband-based industrial policy
is a constraint, from the perspective of both industrial production and technological
development.
5
For more information on targeted cutting-edge policies, see Peres and Primi (2009).
ECLAC 246
of hyperconnectivity, offers emerging economies a new opportunity for
addressing the fundamentals of broadband industrial policy.
In the global context, China is on track to lead 4G broadband
development, on the strength of the sheer number of its mobile
smartphone users. Although it remains to be seen whether China
can lead the hyperconnected world in the next decade, its broadband
development policies will undoubtedly have repercussions far beyond
a policy based solely on network deployment. China, the Republic of
Korea and Japan are moving into a new stage of broadband industrial
development, with synergies between industrial development and
technological development. The Association of Southeast Asian Nations
(ASEAN) will also take part in the division of labour for giga broadband
industrial development. This will mirror the earlier “fying geese” model
of the Asian division of labour, which was led by Japan following the
Plaza Accord in the 1980s. This time, the group could be led by China,
the Republic of Korea and Japan, under a new model that might be called
the “smart dragon”, in which the regional economy will grow rapidly
with hyperconnected ICTs, led by China.
If Latin America participates in this process, there should be
a reasonable window of opportunity for a new wave of industrial
development over the next decade. Otherwise, the region’s economies
could lose another decade of ICT development if their policy focus is
penetration rate and network speed with no signifcant impact on industrial
development. The third stage of broadband industrial development in the
2010s will undoubtedly be different from the frst. In addition to building
on the frst stage, it will usher in unprecedented developments such as
the convergence of information, communications and environmental
technologies related to lifestyles and climate change, establishing a new
platform for technological and industrial development.
In sum, the Korean case displays a classic combination of
broadband development and the simultaneous development of networks,
devices and, to a lesser extent, platforms and content as part of broader-
based industrial development policy. Table VIII.6 compares the
trajectories of horizontal industrial policies and the cutting-edge policies
of the 1990s and 2000s and sets forth a new scope for a convergence
policy that could change the outlook for the industrial competitiveness
of Latin America in general.
Broadband in Latin America: Beyond Connectivity 247
Table VIII.6
INDUSTRIAL AND TECHNOLOGICAL DEVELOPMENT MODEL IN LATIN AMERICA
AND THE REPUBLIC OF KOREA, 1960-2020
Industrial policy
period
1960-1995 1995-2010 2010-2020
Global framework GATT WTO Post-Doha
WTO
Broadband
infrastructure
1G 2G-3G 4G-5G
Industrial policy
path in
Latin America
Sectoral policy Horizontal
policy
Convergence
policy
Policy
instruments
Import-
substitution
industrialization
Various Various
Endogenous
development
Strong Weak Medium
Global supply
chain
Weak Medium Strong
Industrial policy
path in the Republic
of Korea
Sectoral policy Targeted,
cutting-edge
policy
Generative
convergence
policy
Policy
instruments
Import-
substitution
industrialization
+ export-
oriented
industrialization
Research and
development,
standards,
broadband
Research and
development,
standards,
broadband-
based
convergence
Endogenous
development
Strong Strong Strong
Global supply
chain
Strong Strong Strong
Source: Prepared by the author.
This chapter concludes with some recommendations on broadband
development and industrial development aimed at policymakers in
Latin America.
(i) Shift the focus of broadband development policy from a
network-based approach to an industrial development
approach. The latter cannot be achieved without the former, but
the focus has to change in order to achieve the best outcomes in
the coming decade.
(ii) Establish a national giga plan for the next decade, with four
components: networks, devices, platforms and content. Each
area should take into account the ways and means of inserting
ECLAC 248
the national economy into the global economy through an
appropriate division of labour in production and distribution.
(iii) Design a demand-driven giga broadband policy, harnessing
government promotion policy and market forces in order to
develop e-government, e-business, e-learning, e-health and
other application platforms.
(iv) Draw up a universal service plan to bring mobile broadband
to rural communities and other areas with poor coverage.
Governments can promote public-private partnerships, which
assume a division of labour between the two sectors; the former
would be responsible for the national fbre optic network, while
the latter would invest in alternative mobile technologies such
as WiMAX and HSDPA where the private sector can respond
more quickly to fast-changing technologies.
(v) Closely study the development of the 4G standard in China,
such as TD-LTE, to assess the effects of a new standard on
national, regional and global mobile broadband markets.
Brazil has begun working on TD-LTE for the development of
4G broadband. Although this process involves multiple actors,
China’s move towards 4G could have substantial repercussions
on broadband industrial development in Latin America.
(vi) Gradually change the traditional curriculum approach to one
based on smart learning, given the predictions that the next
decade will open up an enormous gap in the region in terms
of human resource development in the hyperconnected world.
This requires universal literacy, to the extent possible, because
as Amsden (1995) points out, the leap forward in the Republic
of Korea was only possible after the attainment of universal
education in the 1960s.
(vii) Attract foreign direct investment for 4G development and
deployment. This requires a proactive broadband industrial
policy to create synergies between network deployment and
industrial development.
(viii) Develop intra-industry trade with the suppliers of manufactured
broadband products as part of the new trade policy. At present,
some countries in the region have a good pool of skilled labour,
which can contribute to the adaptation and development of new
technologies if the technology transfer channels are secured.
Together with foreign direct investment, intra-industry trade
can pave the way.
Broadband in Latin America: Beyond Connectivity 249
(ix) Draw up a national policy for the endogenization of foreign
and new technologies. The Giga Korea Plan could have policy
spillover effects for in Latin America. The 4G and 5G stages could
be a major turning point for endogenization because—even for
the developed economies—many parts and components are
new-generation. This could provide an untapped opportunity
to harness the potential of ICTs and smart technologies in Latin
America. The economic development policy behind the 4G-
5G stage needs to be reverse engineered so as to increasingly
orient industrial policy towards the convergence of traditional
industries with ICTs, based on the analysis of big data, cloud
computing and memory applications. The region has the
potential to catch up with the latest trends through a new
industrial policy based on the adaptation and endogenization
of technological innovations.
(x) Draft a master plan for internationalization and globalization
in the hyperconnected economy. In some cases, trade policy
needs to be redefned in accordance with the new geography of
industrial development. As mentioned above, “smart dragon”
is a new industrial and technological development model in
Asia encompassing two thirds of the world population of smart
mobile technology users. The global economic shift to the east in
the 4G and 5G era requires a broadband approach to trade and
industrial policy in Latin America (instead of the narrowband
approach to trade policy), in order to boost inter-institutional
cooperation and industrial and technological development. To
paraphrase Fernando Fajnzylber in ECLAC (1990), the main
task for the region consists of establishing a new framework for
the development of a broadband industrial policy for “smart”
production transformation with equity.
ECLAC 250
Bibliography
Aizu, I. (2002), “A comparative study of broadband in Asia”, Asia Network Research
Discussion Paper.
Amsden, A. (1995), Getting the Price Wrong: The Case of Korea, MIT.
Atkinson, R. D. (2009), “The role of competition in a national broadband policy,
Journal on Telecommunications & High Technology, vol.7.
Atkinson, R.D., D. K. Correa and J.A. Hedlund (2008), “Explaining International
Broadband Leadership”, Social Science Research Network [online]
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1128203.
Berkman Centre for Internet and Society (2010), Roadmap for Open ICT Ecosystems,
Harvard Law School.
Cava-Ferreruela, I. and A. Alabau-Muñoz (2006), “Broadband policy assessment:
A cross-national empirical analysis”, Telecommunications Policy, vol. 30, No. 8.
De Ridder, J. (2007), “Catching-up in Broadband - What Will it Take?”, Paris,
Organisation for Economic Cooperation and Development (OECD), July.
ECLAC (Economic Commission for Latin America and the Caribbean) (1990),
Changing Production Patterns with Social Equity (LC/G.1601-P), Santiago, Chile.
Government of Korea (2009), Technology Development Promotion Act of Korea
(No. 9630), April.
___ (2011), Giga Korea Plan 2020, Seoul.
Grosso, M. (2006), “Determinants of Broadband Penetration in OECD Nations”,
Working Paper, Regulatory Development Branch, Australian Competition and
Consumer Commission.
Höffer, F. (2007), “Cost and benefts from infrastructure competition: estimating
welfare effects from broadband access competition”, Telecommunications Policy,
vol. 31, No. 6-7.
ITU (International Telecommunication Union) (2003-12), ITU Annual Report
(2003-12); Measuring the Information Society 2012 [online] http://www.itu.
int/net/pressoffce/press_releases/2012/70.aspx.
Kim, Y., T. Kelly and S. Raja (2010), “Building broadband: Strategies and policies for
the developing world”, GICT Publication by KISDI.
Korea Information Society Development Institute (KISDI) (2012), 2012 ICT Inductry
Outlook of Korea, Seoul.
Korea IT Times (2012), [online] http://www.koreaittimes.com/story/21141/vision-
it-powered-korea-future, 25 April.
Lau, T.Y., S.W. Kim and D. Atkin (2005), “An examination of factors contributing
to South Korea’s global leadership in broadband adoption”, Telematics and
Informatics, No. 22.
Machina Research (2012), Mobile Broadband Global Forecast & Analysis 2011-20, Sector
Report, Machina Research.
Ministry of Knowledge Economy (2012), San Ueop Baek Seo 2011, Republic of Korea.
Ovum Consulting (2009), Broadband Policy Development in the Republic of Korea, A
Report for the Global Information and Communications Technologies Department,
World Bank.
Peres, W. and A. Primi (2009), Theory and Practice of Industrial Policy: Evidence from
the Latin American Experience, Santiago, Chile, Economic Commission for Latin
America and the Caribbean (ECLAC).
Broadband in Latin America: Beyond Connectivity 251
Wallesten, S. (2006), “Broadband and Unbundling Regulations in OECD Countries”,
AEI-Brookings Joint Centre Working Paper, No. 06-16, Technology Policy Institute.
Waverman, L., M. Meschi, B. Reillier and K. Dasgupta (2007), Access Regulation
and Infraestructure Investment in the Telecommunications Sector: An Empirical
Investigation. With the support of ETNO, London, UK.
World Bank (2009), Broadband Policy Development in the Republic of Korea. A Report for
the Global Information and Communications Technologies Department, October.
WTO (World Trade Organization) (1995), Agreement on Subsides and Countervailing
Measures, Geneva.
Chapter IX
Net neutrality: debate and policies
René Bustillo
A. Introduction
The issues surrounding net neutrality (broadly defned as the principle that
all Internet traffc should be treated without discrimination) are changing
the broadband industry all over the world through regulatory and
market forces. Many regulatory bodies are trying to adopt strict network
regulations, changing the way that telecommunications service providers
operate. From a market perspective, consumer demand for variety and
greater bandwidth for Internet-based services is constantly growing. This
can cause network congestion, which not only affects Internet use but also
the viability of other services often offered by providers, particularly voice
services. And so the way in which the broadband industry deals with
these challenges will depend largely on how regulatory bodies address
the issue.
Although discussions on net neutrality often turn into an ethical,
political and social debate, the challenge is above all a technical one.
Network congestion has become the focal point of discussions on net
neutrality. Service providers claim that because the number of broadband
subscribers and high bandwidth consumption applications has risen, their
networks will very soon be unable to offer the speed and service quality
expected by users. The key to understanding net neutrality and its effect
ECLAC 254
on telecommunications policy and industry lies in grasping the technical
issues behind Internet gridlock and the mitigation technologies and
practices used and adapted to tackle the problem since the Internet began.
B. What is net neutrality?
There are several defnitions of “net neutrality”, but they basically refer
to the same thing: allowing the unrestricted fow of all information over
the Internet. But such unrestricted freedom would have far-reaching
technological, economic, regulatory, legal, social and other consequences.
The very fact that the idea behind the Internet was non-
discrimination in carrying information means that limiting what can and
cannot fow through it would come across as artifcial or even pointless.
But structure of the networks and services that make up the Internet is
so complex that absolutes do not apply in the case of net neutrality. Some
knowledge of networks and topologies can be useful for understanding the
underlying problem of providing unrestricted access to Internet resources.
One of the most controversial statements about neutrality was
made several years ago: “In a world of dumb terminals, networks had to
be smart. But in a world of smart terminals, networks have to be dumb”
(Gilder, 1992). The idea was that telex and telephone networks should be
smart enough to route communications and compensate for the fact that
the devices themselves were not smart at all and were unable to make
decisions relating to rerouting communications or managing congestion.
Computer networks were still in the early stages of development, but it
was thought that with smart terminals at each end the network would be
nothing more than a “bit pipe”. In other words, the network would not
have to make decisions on communication traffc or destination. It just had
to send the information.
An absolutely neutral or “dumb” network is not technically
feasible, especially in such a complicated system as the Internet. The task
of ensuring data exchange on a global scale and, especially, attempts to
improve service quality have forced Internet engineers to make networks
smart by adding “smart” routers and gateways. However, as the Internet
begins to play an increasingly important role in the economy and the
technical ability to identify certain packets and treat them differently
grows, new criteria for discriminating against certain types of traffc are
put forth (Bocache, Mikheyev and Paque, 2007).
Information technology experts —even those who advocate net
neutrality— agree that, strictly speaking, the Internet was never 100%
neutral (Wu, 2005). For example, the Transmission Control Protocol/
Broadband in Latin America: Beyond Connectivity 255
Internet Protocol (TCP/IP) network distinguishes between different
classes of traffc and indirectly gives some priority over others.
The Internet was designed to work differently from traditional
telephone networks. Since it is based on the principle of packet switching
instead of circuit switching, the Internet is more fexible (when a circuit is
unavailable, the packets can take another route) and, in theory, less reliable
(transmission quality varies because packets are not routed through a
dedicated circuit). The Internet is therefore seen as smart at the ends: at the
devices connected to it.
In the early days, most Internet traffc was largely composed of data
fles and messages that were not delay-sensitive. Today, it carries all kinds
of traffc that can be classifed as delay-sensitive or non-delay-sensitive, or
high- or low-bandwidth-consumption. Table IX.1 classes Internet services
according to their delay-sensitivity, bandwidth consumption and economic
value perceived by users.
Table IX.1
INTERNET SERVICES ACCORDING TO DELAY SENSITIVITY, BANDWIDTH
CONSUMPTION AND ECONOMIC VALUE
Service Delay sensitivity Bandwidth consumption Value / willingness to pay
P2P fle sharing Low Very high (no limit) Low
YouTube Low (buffered) Medium (320 – 600 Kbps) Low
NetFlix streaming Low (buffered) High (up to 4 Mbps) Low
E-mail Low Very low Low
VoIP Medium - high Low (30 – 80 Kbps) Medium
Online gaming High Low (30 – 80 Kbps) Medium
Videoconferences High medium High
Telemedicine High High (up to 8 Mbps) High
Source: G. Pehnelt, “The economics of net neutrality revisited”, Jena Economic Research Papers, 27
October 2008.
The table shows, frst, that combining the different kinds of traffc
makes for a mix that would be diffcult for any network to manage unless
its resources were unlimited. Data networks are usually made up of access
points (physical or wireless), a core network and, lastly, connections to other
networks. Constraints at any part of the network could lead to packet loss
and lower quality that some applications would tolerate better than others.
However, since TCP/IP networks operating under normal conditions do
not distinguish between traffc according to delay-sensitive, application
packets with high bandwidth consumption and low delay-sensitive food
the network to the detriment of those carrying data applications, such as
VoIP and videoconferencing, that do have lag issues.
ECLAC 256
The conclusion is that under normal traffc conditions, delay-
sensitive applications and services cannot maintain the same standard of
quality as applications which are not affected by expected data network
delay. The situation grows more complicated when there is traffc
overload, because services that cannot tolerate lag are the frst to suffer the
consequences of equal treatment of all data packets.
1. The non-discrimination principle
Non-discrimination is a core principle of ICT regulation. Broadly, it
should be understood as barring service providers from giving different
treatment to users and other providers operating under the same
conditions. Discrimination can be found in pricing (two users who are
charged different rates for the same service) or in facilities (different access
quality or features for different users). Discrimination in net neutrality
can take many shapes: charging users differently according to the type of
traffc they use the most (like web pages with or without high multimedia
download content), blocking certain applications or lowering service
quality (as is the case of T-Mobile 3G and Skype in Germany).
According to some authors, the underlying principle of network non-
discrimination is to give users the right to use devices and applications
that are not harmful to the network while giving innovators the freedom
to provide them (Wu, 2005). However, providers tend to be against this
principle, claiming that, in the long run, unrestricted freedom impairs
network and service quality and requires investments that revenue from
services provided does not cover. The classic example is VoIP over data
networks and the Internet, which voice providers (fxed and mobile)
oppose and block when it starts to eat into their revenue.
Peha (2006) analyses the benefts and risks of mandating net
neutrality and suggests a policy designed to protect the benefcial uses of
discrimination which might allow the network operators to:
• Provide different quality of service to different classes of traffc,
using explicit prioritization or other techniques. These techniques
can be used to favor traffc with stricter quality of service
requirements, and/or traffc sent using a higher-priced service.
• Charge a different price for different classes of traffc. The higher
price would be justifed because the traffc requires superior
quality of service, consumes more of a limited resource, has a
greater adverse effect on other traffc, or is otherwise linked to
cost (or opportunity cost).
• Block traffc that poses a threat to security.
• Charge the senders of information, recipients, or both.
Broadband in Latin America: Beyond Connectivity 257
• Offer proprietary content or unique services to their customers
(without using their dominant control over the last-mile
connection to favour their own content or service).
• Block traffc originating from an attached device that might
reasonably be believed to be harmful to the network or its
users, such as one that does not follow prescribed protocols
and algorithms.
• Use any form of discrimination they wish, if the broadband
market becomes truly competitive.
2. The growth of traffic and net neutrality
One of the most important considerations for any analysis of net neutrality
is the fact that Internet traffc has been soaring. Projections are that IP
traffc in petabytes (PB) per month will increase four-fold between 2010
and 2015 (see table IX.2). The situation is particularly concerning in Latin
America since growth will mean a seven-fold increase in traffc. As a
result, service providers will have to adapt the networks to these levels of
data traffc while ensuring the viability of the business model.
Table IX.2
GLOBAL IP TRAFFIC PROJECTIONS, 2010-2015
(Petabytes per month and percentages)
2010 2011 2012 2013 2014 2015
CAGR
2010-2015
By type
Fixed Internet 14 955 20 650 27 434 35 879 46 290 59 354 32
Managed IP 4 989 6 839 9 014 11 352 13 189 14 848 24
Mobile data 237 546 1 163 2 198 3 806 6 254 92
By segment
Consumer 16 221 23 130 31 592 42 063 54 270 70 045 34
Business 3 930 4 894 6 011 7 357 8 997 10 410 22
By location
North America 6 998 9 947 12 978 16 116 18 848 22 274 26
Western Europe 4 776 6 496 8 819 11 774 15 187 18 858 32
Asia-Pacifc 5 368 7 317 9 847 13 341 18 060 24 150 35
Japan 1 414 1 923 2 540 3 283 4 019 4 762 27
Latin America 665 993 1 465 2 158 3 238 4 681 48
Central and Eastern
Europe
708 1 004 1 413 1 955 2 700 3 713 39
Middle East and Africa 253 366 550 802 1 235 2 019 52
Total IP traffc 20 151 28 023 37 603 49 420 63 267 80 456 32
Source: Cisco Systems, “Cisco Visual Networking Index”, 2011.
ECLAC 258
The service providers’ main argument against net neutrality is that
with the expected growth in traffc, being unable to select or prioritize data
packets will make it diffcult to maintain the performance level required
to carry all of the traffc. This is even more diffcult in wireless broadband
networks, since problems setting up base stations and spectrum availability
constraints prevent access networks from growing as quickly as required.
Another equally worrying aspect of IP traffc growth is global
consumer traffc, which encompasses any traffc that is not confned
to a single service provider’s network. In other words, it is any traffc
which crosses from one network to another, as is the case for most traffc
generated by web browsing. Table IX.3 shows the projected growth of this
traffc to 2015, disaggregated by network and subsegment.
Table IX.3
GLOBAL CONSUMER INTERNET TRAFFIC, 2010-2015
(Petabytes per month and percentages)
2010 2011 2012 2013 2014 2015
CAGR
2010-2015
By network
Fixed 12 355 17 467 23 618 31 318 40 842 53 282 34
Mobile 174 399 858 1 654 2 930 4 931 95
By subsegment
File sharing 4 968 6 017 7 277 8 867 11 040 13 797 23
Internet video 4 672 8 079 12 146 17 583 24 357 33 620 48
Web, e-mail and data 2 393 3 113 4 146 5 325 6 769 8 592 29
Video calling 308 442 659 905 1 251 1 736 41
Online gaming 49 68 95 133 187 290 43
Voice over IP (VoIP) 138 147 153 157 160 168 4
Others 0 1 1 3 8 11 132
By location
North America 3 301 5 000 6 579 8 306 10 012 12 537 31
Western Europe 3 147 4 360 6 075 8 224 10 841 13 896 35
Asia-Pacifc 4 403 6 006 8 142 11 129 15 249 20 758 36
Japan 638 932 1 317 1 807 2 344 2 968 36
Latin America 482 735 1 106 1 667 2 577 3 850 52
Central and Eastern Europe 454 667 971 1 381 1 963 2 805 44
Middle East and Africa 103 166 286 459 784 1 399 68
Global consumer traffc 12 528 17 866 24 476 32 973 43 771 58 214 36
Source: Cisco Systems, “Cisco Visual Networking Index”, 2011.
As can be seen, mobile Internet traffc growth is soaring (28-fold)
and far outpacing fxed networks. Internet video traffc (one of the most
popular applications at present) is growing disproportionately. Although
IP television (IPTV) traffc does not fall into this category, projected growth
will also impact service provider access networks. In view of the above,
Broadband in Latin America: Beyond Connectivity 259
providers have adopted a radical stance on net neutrality, saying that if
they cannot differentiate the traffc on their networks the business model
will become unsustainable in a few years.
C. The situation in Europe, the United States
and Asia-Pacific
Net neutrality is an issue in many places worldwide, particularly in the
European Union countries and the United States. It has been addressed
from different angles, because measures to ensure absolute neutrality
would have major implications for network operability.
To identify the best international practices, net neutrality was
examined in several regions of the world. For Latin America, there are
useful lessons to be learned from the experience of Europe and the United
States regardless of any differences in degree of development of electronic
communications. The methods used in Asia-Pacifc could also be a source
of useful models for looking at the issue from different perspectives.
1. Europe and the United States
Broadband development has taken different paths in Europe and the
United States. ADSL, which is the primary means of fxed broadband
access in Europe, is not as prominent in the United States, where cable
Internet access is the most common (see fgure IX.1). This is mainly for
historical and geographical reasons that made cable providers more likely
to provide broadband access than was the case in most of the countries
of Europe. In addition, the broadband regulatory framework is much less
interventionist in Europe than in the United States, where it is tightly
regulated at the federal and state levels.
Figure IX.1
FIXED BROADBAND ACCESS, BY TECHNOLOGY, 2011
(Percentages)
0
10
20
30
40
50
60
70
80
90
100
Canada United
States
United
Kingdom
France Germany Italy Japan Australia
DSL Cable modem Fibre/LAN Other
Source: Canadian Radio-television and Telecommunications Commission, CRTC Communications
Monitoring Report, 2011.
ECLAC 260
a) European Union
The European Union (EU) is made up of 27 States that are subject
to community law. All of the European countries examined herein (with
the exception of Norway) fall within the scope of European regulations
and must adapt their national regulations to the provisions of net
neutrality directives.
The regulatory framework for electronic communications networks
and services (the “Regulatory Framework”) is the basis for all national
telecommunications laws in the EU Member States. The Regulatory
Framework provides general and technology neutral rules applying to
all electronic communications networks and services covering fxed
and wireless telecommunications, data transmission and broadcasting
transmission. It contains provisions for the structure and functioning
of national telecommunications providers, and the framework for both
general rules applying to all providers of electronic communications
networks and services and particular rules which may only be imposed by
national regulatory authorities (NRA) on operators with signifcant market
power (SMP). The Regulatory Framework only relates to the provision of
electronic communications networks and services, and does not cover the
content of these services.
One of the main objectives of the Regulatory Framework is to align
sectoral regulation of the electronic communications market with general
competition principles. As a consequence, the Regulatory Framework
adopts the principle that ex ante regulation should only be imposed where
there is ineffective competition, for example, in markets where there are
one or more undertakings with SMP and where competition law remedies
are not suffcient to address the problem (Enaux and Escribano, 2011).
The Commission attaches high importance to preserving the open
and neutral character of the Internet, taking full account of the will of
the co-legislators now to enshrine net neutrality as a policy objective and
regulatory principle to be promoted by national regulatory authorities,
alongside the strengthening of related transparency requirements and
the creation of safeguard powers for national regulatory authorities to
prevent the degradation of services and the hindering or slowing down
of traffc over public networks. The Commission will monitor closely the
implementation of these provisions in the Member States, introducing a
particular focus on how the ‘net freedoms’ of European citizens are being
safeguarded in its annual Progress Report to the European Parliament and
the Council. In the meantime, the Commission will monitor the impact
of market and technological developments on ‘net freedoms’ reporting
to the European Parliament and Council on whether additional guidance
is required, and will invoke its existing competition law powers to deal
Broadband in Latin America: Beyond Connectivity 261
with any anti-competitive practices that may emerge (Offcial Journal of
the European Union, L 337, 18 December 2009). This declaration is the
starting point for regulating net neutrality because it refers to preserving
the open and neutral character of the Internet. However, the term “Internet
neutrality” has a different nuance from “net neutrality”. Internet neutrality
refers to all of the elements which make up the public network, which in
many cases does not include the access network or sections of the network
which belong to a particular provider. Net neutrality is much more general
and includes all of the elements and applications required for users to
connect with each other or other devices connected to the “network of
networks”. Although the terms are used without distinction, they differ
in scope.
Although the declaration states that Member States must report to
the European Parliament on measures taken to protect Internet freedoms,
there is no consensus with respect to what these freedoms encompass. Nor
is the declaration in and of itself enough to guarantee user rights against,
for example, content blocking or fltering.
The third European Union Telecom package (A6-0272/2009) refers
to the Council common position for adopting the European Parliament
and Council directive amending Directive 2002/21/EC on a common
regulatory framework for electronic communications networks and
services, Directive 2002/19/EC on access to, and interconnection of,
electronic communications networks and associated facilities, and
Directive 2002/20/EC on the authorization of electronic communications
networks and services. The Telecom package sets out three sets of
measures that pertain to neutrality. They are:
• To enshrine the principle of neutrality as a regulatory objective,
both in its economic aspect (fostering true competition between
Internet access providers and content providers for the beneft of
the consumer, “including for the transmission of content”) and
in its social aspect (favouring end-user access to information and
preserving their ability to disseminate and use the applications
of their choice).
• To impose transparency obligations on operators for managing
traffc and network access restrictions so as to ensure the
protection of the principle of neutrality via competition (new
compulsory information included in electronic communications
service agreements must appear clearly and in detail and be
easily accessible: traffc management procedures, restrictions on
access to certain services or equipment, measures to ensure the
network’s security and integrity, among others).
ECLAC 262
• To grant new powers to the regulators to prevent violations of the
principle of neutrality (power to defne minimum requirements
in terms of quality of service; dispute resolution powers extended
to disputes over the transmission of traffc between operators
and other companies, including content providers).
The third Telecom package contains specifc provisions on
actions that can be taken by the regulator to ensure that users’ rights
are not violated. The measures focus on competition and information
transparency, making it clear that the policy focus is on protecting user
rights via a healthy, undistorted market.
b) United States
In March 2010 the Federal Communications Commission (FCC),
the communications regulatory body for the United States, presented
the frst National Broadband Plan and set the country on a new
telecommunications path that is very pro-consumer, pro-privacy and pro-
competition. The plan also proposes signifcant changes to access regimes,
wholesale facilities and competition in both the telecommunications and
the broadcasting markets.
Communications policy in the United States is jointly regulated by
federal and state governments. At the federal level, the Communications
Act of 1934, amended by the Telecommunications Act of 1996 (together,
the “Act”), established a national policy geared towards widespread, rapid
and effcient communication services that are universally available at
affordable rates in a competitive market. The Act tasks the Commission
with implementing a regulation which promotes these policies in a way
that is consistent with “public interest, convenience, and necessity”.
Each State’s Public Services Commission controls the rates and sets the
obligations of fxed telephony operators. Although the United States
made the transition from a monopolistic to a competitive communications
market in 1996, the industry continues to be subject to oversight by the
FCC and the states.
In 2005, the FCC issued a unanimous policy statement spelling out
four fundamental rights of Internet users. Users are entitled to (i) access the
lawful Internet content of their choice; (ii) run applications and use services
of their choice, subject to the needs of law enforcement; (iii) connect their
choice of legal devices that do not harm the network; and (iv) competition
among network providers, application and service providers, and content
providers. After further discussion, in December 2010 the FCC decided
to impose two even tighter regulations on Internet access providers: no
Broadband in Latin America: Beyond Connectivity 263
blocking and no unreasonable discrimination in transmitting traffc
(Federal Communications Commission, 2011).
In short, the frst requirement is transparency. Fixed and mobile
broadband providers must disclose the network management practices,
performance features, and terms and conditions of their broadband
services. Secondly, there must be no blocking: fxed broadband providers
may not block lawful content, applications, services, or non-harmful
devices; mobile broadband providers may not block lawful websites
or block applications that compete with their voice or video telephony
services. Thirdly, there must be no unreasonable discrimination: fxed
broadband providers may not unreasonably discriminate in transmitting
lawful network traffc.
These rules, applied with the complementary principle of reasonable
network management, will ensure the continued freedom and openness
that have enabled the development of the Internet. This framework
provides greater clarity and certainty for consumers, innovators, investors
and broadband providers, as well as the fexibility that providers must
have to effciently manage their networks. It enables a virtuous circle of
innovation and investment in which new uses of the network–including
new content, applications, services and devices–lead to increased end-user
demand for broadband, driving network improvements that in turn lead
to more innovative uses.
In 2011 two bills were proposed, one by the Senate (Preventing
Real Online Threats to Economic Creativity and Theft of Intellectual
Property Act, or PIPA) and another by the House of Representatives
(Stop Online Piracy Act, or SOPA). Both bills are based on similar
principles and essentially propose the same thing: to combat piracy of
United States copyright-protected products, the law should require that
content providers and Internet access providers block access to websites
which infringe these provisions. These regulations have caused a stir
in the United States and other countries amid claims that they violate
the principles of freedom of expression and presumption of innocence
and that, if adopted, the effects will be similar to the “Great Firewall of
China”. The difference between PIPA and SOPA is that while the former
aims to block domains, the latter also aims to block IP addresses. These
bills are still being debated in the United States Congress. If they are
approved, websites that breach United States copyright laws would
be blocked within fve days and then the owners would be prosecuted
under criminal law (United States Senate, 2011; United States House of
Representatives, 2011).
ECLAC 264
2. Asia-Pacific
a) Republic of Korea
The Republic of Korea is one of the most advanced broadband
markets in the world and a global leader in the deployment of FTTx. This
is partly because it implemented a sound national broadband development
strategy that received wide political support. The government adopted an
initiative known as the Korea Information Infrastructure (KII) Plan, which
aimed to connect 84% of households to broadband services with speeds
of up to 1Mbps by 2005. Its next major objective was to deploy 1 Gbps
broadband services by 2012.
The government adopted two development programmes: the
Broadband Convergence Network (BcN) and IT839. Their aim was to
create a widespread network enabling customers to communicate at any
time using a number of devices, such as fxed and mobile telephones,
personal computers and home networks. In 2004, the government selected
three consortia led by Korea Telecom (KT), DACOM and South Korea
Telecom (SKT) to develop self-fnanced BcN prototypes. The objective
was to create the best BcNs in the world, which would be able to provide
up to 100 million households with fxed multimedia broadband and 100
million users with wireless services. This programme was followed by an
UltraBroadband Convergence Network (UBcN) programme running from
2009 to 2013 and focused on developing 1Gbps broadband services.
A comparative study of 16 countries carried out at the start of 2011
ranked the Republic of Korea as the most advanced country in terms of
government planning for broadband development. Surprisingly, the
government plans to spend less than 1% of its budget on the plan and aims
to fund it by promoting private investment (Point Topic, 2011).
Although the Republic of Korea is one of the leading countries in
broadband development worldwide, net neutrality policy and regulation
is still embryonic. While the Telecommunications Business Act does not
specifcally mention net neutrality and the Korean Communications
Commission (KCC) has not issued an offcial policy, there was a case in
which a broadband operator was sanctioned for blocking one provider’s
VOD service. The KCC found that this blocking was a prohibited activity
under the law. The dispute was settled by the VOD services provider
agreeing to pay the broadband company a fee for using the network. Since
Internet traffc is constantly rising, it is hoped that net neutrality will move
higher up the agenda.
Broadband in Latin America: Beyond Connectivity 265
b) Japan
Japan was relatively late in joining the broadband revolution but
started experiencing rapid growth in the service in 2001. A major reason
for the lag was the slow liberalization of the telecommunications market,
which allowed the incumbent Nippon Telegraph and Telephone (NTT)
to control the market in many ways. Despite this, liberalization started
in 1999 when NTT was divided into a holding company with fve major
subsidiaries: NTT East, NTT West (local telephone companies), NTT
Communications (long distance), NTT Docomo (mobile) and NTT Data
(information services). Following liberalization, the broadband market
grew quickly; by the end of 2009 Japan had the third largest in the world,
after the United States and China.
In September 2007, the Ministry of Internal Affairs and
Communications of Japan (MIAC) published the “Report on network
neutrality”, which identifed two basic issues: fair distribution of network
development costs and fair access to the telecommunications operators’
network, including content providers. It discusses who should bear
the development costs and whether telecommunications operators can
participate in packet shaping (or blocking traffc) to guarantee network
service quality. It focuses on whether major consumers should be
required to pay additional charges depending on their use of packets
and whether mass content distributors should pay additional charges to
Internet service providers (ISP). Since there is no specifc law prohibiting
such charges, the report concludes that these matters should be left up
to the telecommunications market (Ministry of Internal Affairs and
Communications, 2007).
As for traffc management, four telecommunications operator
consortia published a guide in May 2008 which is in line with the
discussion in the “Report on network neutrality”. It states that packet
shaping could violate the Telecommunications Business Law (TBL)
because telecommunications secrecy is protected by this law. However,
packet shaping may be permitted under exceptional circumstances, such
as when users have diffculties accessing a network because of major users’
traffc or a specifc application that overloads the network. The guide also
states that telecommunications operators must inform users of rates and
any traffc shaping, and how and when it will take place.
The Ministry of Information and Communications (MIC) set up
the Study Group on a Framework for Competition Rules to Address the
Transition to IP Based Networks in 2005. On 19 September 2007, the Group
ECLAC 266
determined that there is net neutrality when the following three conditions
are met: consumers are entitled to (i) use IP-based networks fexibly
and access the content/application layer freely; (ii) connect to IP-based
networks freely through terminals that comply with technical standards
provided by laws and regulations and these devices may connect to each
other fexibly; and (iii) use the communication layer and the platform layer
free from discrimination at a reasonable price.
In line with the MIC initiative on reasonable traffc management,
the associations of Internet providers, telecommunications operators and
cable companies created the “Guide for Traffc Management” in 2010.
According to this guide:
• The principal means of dealing with the rise in traffc include
investing in the network and increasing its capacity; traffc
management should be used only in exceptional circumstances.
• Traffc management must be designed to deal with network
congestion (there must be offcial data to prove there is
congestion). However, it is not a legitimate means for dealing
with copyright infringement or data security issues.
• In order not to endanger the secrecy of any means of
communication (article 21 of the Constitution of Japan), the
ISPs must obtain individual and explicit consent from the users
unless the act performed is in the pursuit of lawful business
(article 35 of the Criminal Code).
• To ensure fairness in use (article 6 of TBL), traffc management
shall be appropriate and non-discriminatory unless there are
reasons to justify unfair treatment. Since ISPs and others are
expanding the content business, such acts (discriminatory
treatment) would hinder efforts to ensure fair competition.
• ISP must disclose their traffc management data in advance and
in accordance with the revised telecommunications guidelines
of the Consumer Protection Act. Since the confguration packet
of an ISP may affect the broadband ecosystem, all parties
concerned, including Internet providers and mobile virtual
network operators (MVNO) must be informed.
D. Situation and outlook in Latin America
Few countries of the region have legislation or regulations on net neutrality.
For most of them this is a new issue that has not been fully addressed,
even though national regulations governing the sector spell out principles
Broadband in Latin America: Beyond Connectivity 267
which enshrine users’ right to free access. All of the countries, with the
exception of Chile and, to a certain extent, Brazil, have only recently started
to discuss net neutrality; any experience they have in implementing the
corresponding legal and regulatory framework is minimal.
1. Chile
In August 2010, Chile became the frst country in the world to adopt a net
neutrality law. It bars telecommunications operators and Internet service
providers from arbitrarily blocking, interfering with, discriminating,
hindering or restricting an Internet user’s right to use, send, receive or
offer any lawful content, application, service or any other type of lawful
activity or use through the Internet.
This law focuses on how access providers should calculate and
publish Internet access service quality indicators, and it identifes some
of the statistical measures that should be calculated and reported. It
also states that access providers may only block services, content and
applications at a user’s request. In terms of the use of devices, it states that
users may connect to the network using any lawful device which does not
harm the security or quality of a service provided to others.
The regulatory decree of 18 March 2011 covers and elaborates
on all of the aspects set out in the law, focusing on user rights and how
users may exercise them when faced with an unfavourable situation. The
decree establishes up front the information that Internet access providers
should provide, such as the characteristics of the services offered to the
users, quality indicators and traffc management measures. It also sets
a time limit for access providers to provide this information upon user
request. Under the provisions of this regulation, any action preventing or
hampering a user from exercising the right to access information on the
features of the access services offered is regarded as a practice restricting
freedom of content, applications and services.
2. Brazil
Although there is no specifc legislation on neutrality in Brazil, there
are some references to it in the General Telecommunications Law.
Article 3 grants all users of telecommunication services the right to non-
discrimination in conditions for access and use. On these legal grounds,
the regulatory body (ANATEL) seeks to regulate net neutrality by drawing
up specifc rules for the Multimedia Communications Services Regulation,
which is currently being updated.
Despite regulatory advances, net neutrality is a more general concept
in Brazil and is not limited to Internet services. The rules to be implemented
ECLAC 268
by ANATEL will bar all network and service providers, without exception,
from blocking or discriminating against any type of traffc, regardless of
content (voice, data, video) or technology used on the network.
Exceptions may be made to these provisions if they are required to
ensure service and network stability, provided that the privacy of users
and communications is respected and competition is not affected.
E. Criteria for developing a national policy
In view of the above, a number of considerations should be taken into
account with respect to net neutrality.
Free choice. When developing a national policy on net neutrality,
priority should be given to the user’s right to freely use, send, receive and
offer any lawful content, application or service through the Internet, unless
they are prohibited under statutory provisions. The user must also be able
to freely use any lawful instrument, device or apparatus on the Internet
which does not harm service security or quality. No characteristic of the
network should prevent or hamper a user from exercising the right to
free choice. The only condition should be lawful use of the Internet, since
combating cybercrime takes priority over individual rights.
Non-discrimination. Network and telecommunications services
providers which provide Internet access must always treat content,
applications and services equally, without discriminating against source
or ownership. Network and telecommunications services providers which
offer Internet access can make offers based on market segment needs or
use and consumption by users; this should not be taken as discrimination.
Transparency. Network and telecommunications services providers
offering Internet access must disclose their network management policies
to users and other providers that have access to their network, such as
content or applications providers. Transparency is essential for users
to know the conditions of the services they are offered. Net neutrality
policymakers must bear in mind that users have the right to be informed
about the conditions under which they will be provided Internet access
and whether traffc management practices will be used which may affect
their privacy and the quality of the service they will receive.
Right to information. Network and telecommunications services
providers offering Internet access should provide users with complete
information on service terms and conditions, including speed, quality
Broadband in Latin America: Beyond Connectivity 269
and traffc management practices for each plan offered by the provider
or hired by the user. This principle is closely related to and complements
information transparency in that information targets each user as an
individual client.
The Internet as basic infrastructure. The Internet is essential
infrastructure, both for the fow of economic activity and as an input for
many other services of great social value, such as access to educational
information and the dissemination and discussion of political views. The
economic value of reliable Internet access should not be left out of the debate
on net neutrality which will inevitably take place between access providers
and governments. After all, the access providers’ business model is based
on the idea not only that their investments will be recouped from users but,
in addition, service providers and applications must contribute through
what are known as bilateral markets. In any event, those agreements must
be left to free competition and negotiation between parties.
“Neutral” Internet. One of the most hotly debated issues is whether
net neutrality and Internet neutrality are synonymous. Net neutrality is
the principle that all Internet traffc generated within the network should
be treated equally regardless of factors such as content, access type or
user characteristics. Internet neutrality, however, is a much more complex
concept which deals with issues such as whether the information within
the network of networks is being intentionally altered or blocked because
the availability of network resources is limited at the source. Obviously,
things like cybercrime often justify not only blocking the website involved
in criminal activities but also prosecuting those responsible. In any event,
it should be understood that the Internet will never be absolutely neutral
and that a poorly implemented net neutrality principle is no reason for
allowing websites, for example, to publish child pornography or copyright-
infringed material or, quite simply, threaten the future of the web.
The Chilean law on net neutrality, which is the most advanced in
the region, explicitly excludes unlawful content, applications and services.
Therefore, there is no reason to bar (without a court order) fltering of
unlawful content, applications or services provided that the flter does not
affect lawful content that may be hosted on the same website or operate
with the same IP address as unlawful content. It does refer to protection
against malicious acts by blocking (without waiting for a court order)
outgoing and incoming traffc from anyone who has been identifed as
a hacker on the grounds that they are attacking the provider’s or third
party’s equipment through the Internet.
ECLAC 270
Bibliography
Bocache, Romina, Andrei Mikheyev and Virginia Paque (2007), The Net Neutrality -
Debate and Development, March.
Cisco Systems Inc. (2011), Cisco Visual Networking Index: Forecast and
Methodology, 2010–2015, 1 June.
CRTC (Canadian Radio-television and Telecommunications Commission)
(2011), CRTC Communications Monitoring Report 2011 [online] http://www.crtc.
gc.ca/eng/publications/reports/PolicyMonitoring/2011/cmr6.htm.
Enaux, Christoph and Blanca Escribano (2011), “An Overview of the EU Regulatory
Framework”, The International Comparative Legal Guide to: Telecommunication
Laws and Regulations, Olswang LLP, 2012.
FCC (Federal Communications Commission) (2011), “Preserving the Open
Internet”; [GN Docket No. 09–191; WC Docket No. 07–52; FCC 10–201]; Federal
Register / Vol. 76, No. 185 / Friday, 23 September / Rules and Regulations.
Gilder, George (1992), “La llegada de la fbroesfera”, Forbes ASAP, 7 de diciembre.
MIAC (Ministry of Internal Affairs and Communications of Japan) (2007), Report on
Net Neutrality. Working Group on Net Neutrality, September.
Peha, Jon M. (2006), “The Benefts and Risks of Mandating Net neutrality, and the
Quest for a Balanced Policy”, paper presented at the 34th Telecommunications
Policy Research Conference, Carnegie Mellon University, September.
Pehnelt, Gernot (2008), “The economics of net neutrality revisited”, Jena Economic
Research Papers, 27 October.
Point Topic (2011), “South Korea Broadband Overview, 16 September [online]
http://point-topic.com/content/operatorSource/profiles2/south-korea-
broadband-overview.htm.
United States House of Representatives (2011) “Stop Online Piracy Act”, 112th
Congress, 1st Session; House Judiciary Committee; 26 October [online]
http://judiciary.house.gov/hearings/pdf/112%20HR%203261.pdf.
United States Senate (2011), Federal Communications Commission (FCC), “Preserving
the Open Internet” [GN Docket No. 09–191; WC Docket No. 07–52; FCC 10–201];
Federal Register, vol. 76, No. 185, 23 September/ Rules and Regulations [online]
http://judiciary.house.gov/hearings/pdf/112%20HR%203261.pdf.
Wu, Tim (2005), “Net neutrality, broadband discrimination”, Journal of
Telecommunications and High Technology Law.
Fourth part
The future of the ecosystem
Chapter X
The advance of cloud computing
René Bustillo
A. Introduction
A topic of growing importance in the information technologies (IT) feld is
cloud computing also known as cloud services, “the cloud”, computational
cloud or cloud concepts. Like all new trends, cloud computing has sparked
different opinions. There are those who see in it many benefts and others
who claim it is nothing new or who view it as a passing fad. What is clear is
that cloud computing in Latin America, both in government offces and in
private frms, continues to grow rapidly. The adoption of cloud computing
services is expanding primarily because their architecture offers the
benefts of shared services over isolated products. Shared services help
an organization focus on its core business and enable its departments to
reduce the gap between available computing capacity (which is frequently
oversized) and systems demand (generally low-volume, with occasional
peaks). This results in a much more effcient usage-based model.
Cloud computing is not one single technology but rather a
combination of many technologies. Its elements may look like those
from the earlier days of computing, but advances in virtualization,
storage, connectivity and processing capacity have, together, created a
new technical ecosystem. The result is a fundamentally different and
attractive phenomenon.
ECLAC 274
B. What is cloud computing?
Cloud computing is a model for enabling convenient, on-demand
network access to a shared pool of confgurable computing resources (for
example networks, servers, storage, applications and services) that can
be rapidly provisioned and released with minimal management effort or
service provider interaction. This model fosters availability and has fve
essential characteristics (on-demand self-service, broad network access,
resource pooling, rapid elasticity, and measured service), three service
models (Software as a Service, or SaaS; Platform as a Service, or PaaS;
and Infrastructure as a Service, or IaaS), and four deployment models
(private cloud, community cloud, public cloud and hybrid cloud). The key
enabling technologies are fast wide-area networks (WAN); powerful and
inexpensive server computers; and virtualization for high-performance
commodity hardware. Diagram X.1 shows how these elements are
interrelated and the degree of development of each one.
Diagram X.1
CLOUD COMPUTING MODEL
Public Private Hybrid Community
Software
as a service (SaaS)
Platform as a
service (PaaS)
Infrastructure as
a service (IaaS)
Broad
network access
Rapid elasticity
Measured
service
On-demand
self-service
Resource pooling
Essential
characteristics
Service
models
Deployment
models
Source: National Institute of Standards and Technology (NIST).
The cloud computing model offers the promise of massive cost
savings combined with greater fexibility. In developed countries, it is
increasingly seen as crucial that government and industry speed uptake
of this technology in response to economic constraints. But there is
opposition to this technology because it challenges traditional approaches
to datacentre and business applications design and management and
poses unresolved issues of security, interoperability and portability.
Broadband in Latin America: Beyond Connectivity 275
1. Essential characteristics
The defnition provided by the United States National Institute of
Standards and Technology (NIST) includes the fve essential characteristics
described below.
a) On-demand self-service
This allows for use of cloud computing resources as needed, without
human interaction between the user and the service provider. With on-
demand self-service, a consumer can programme the use of services in
the cloud, such as computing and storage, as needed, in addition to the
management and implementation of these services. In order to be effective
and acceptable to the consumer, the self-service interface must be easy to
use and must provide effcient means for managing the services offered.
Ease-of-use and elimination of human interaction boost effciency and cost
savings for the user and the service provider.
b) Broad network access
If this technology is to be an effective alternative to in-house
datacentres, large bandwith communication links need to be available
for connecting to cloud services. One of the main economic justifcations
for cloud computing is that the reduced cost of communication with the
broadband network in the cloud provides access to a larger amounts IT
resources, which in turn can lead to a high degree of utilization. Many
organizations are using a three-tier architecture to connect a wide
variety of computer platforms, such as laptop computers, printers, mobile
telephones and PDA, to the wide area network. This three-tier architecture
includes access switches that connect desktop devices to aggregation
switches, which in turn control the fow, and routers and switches that
provide connection to the WAN and traffc management.
1
c) Resource pooling
The cloud must have a large and fexible pool of resources to satisfy
consumer needs, generate economies of scale, and meet level-of-service
requirements. Applications need resources for running them, and these
resources must be effciently allocated for optimal performance. Resources
may physically reside in many different geographic locations and be
1
The result of this approach is latency times of 50 microseconds or more, which causes
lags in cloud computing. To ensure good performance, the switching environment must
have a latency time of 10 microseconds or less. A two-tiered approach that eliminates the
aggregation layer can fulfill this requirement, using 10 GB (10 Gigabits/second) Ethernet
switches and the next 100 GB Ethernet switches.
ECLAC 276
allocated as virtual computing components as needed. As the NIST puts it,
“[t]here is a sense of location independence in that the customer generally
has no control or knowledge over the exact location of the provided
resources but may be able to specify location at a higher level of abstraction
(e.g., country, state, or datacenter).”
d) Rapid elasticity
Rapid elasticity refers to the cloud’s capacity to expand or reduce
the allocated resources promptly and effciently in order to meet the
requirements of the self-service characteristic. This allocation can be
done automatically, and the consumer will have the impression of an
unlimited quantity of dynamic resources that can be purchased as and
when needed. One of the characteristics of elasticity is that it allows rapid
development and implementation of loosely coupled services that can
be adjusted independently of other services and do not depend on their
elasticity. Diagram X.2 shows how IT platforms have evolved over time,
beginning with local memory caches in the server, moving on through
distributed memory caches and elastic caching platforms and, fnally,
elastic application platforms in the era of cloud computing,. This is a
key consideration with respect to the ease and attractiveness of cloud
computing systems, because the possibility of expanding computing
capacity according to need can only be achieved in traditional servers at
the cost of steep investments in hardware that would stand idle most of
the time.
Diagram X.2
FLEXIBLE PLATFORMS FOR ADJUSTING DATA AND APPLICATIONS IN THE CLOUD
Local cache
(server)
Distributed
cache platform
(server)
Elastic
cache platform
(cluster)
Elastic
application
platform
(cloud)
Source: Prepared by the author.
Broadband in Latin America: Beyond Connectivity 277
e) Measured service
Given the characteristics of service-oriented cloud computing, the
quantity of cloud resources that a consumer uses can be dynamically
and automatically assigned and monitored. Customers can then be billed
only for their actual use of the resources assigned to them for a particular
session. The NIST describes measured service as follows: “[c]loud systems
automatically control and optimize resource use by leveraging a metering
capability at some level of abstraction appropriate to the type of service
(e.g., storage, processing, bandwidth, and active user accounts). Resource
usage can be monitored, controlled, and reported, providing transparency
for both the provider and consumer of the utilized service” (Mell and
Grance, 2011). The concept of measured service is closely related to that of
on-demand self-service. However, the innovation is that the user is billed
only for what is used and does not have to purchase costly hardware and
software which, in some cases, will represent a very high fxed cost and
poses a barrier to market entry. The other advantage is that if at some point
the user decides to change to a different commercial activity, it does not
have to modify or dispose of the IT platforms, which generally are not its
property, since it merely has a service contract with the service provider.
2. Service models
The generally accepted classifcation scheme for cloud computing involves
the software platform infrastructure (SPI) model, which includes three
principal services provided via the cloud (SaaS, PaaS and IaaS). Although
there are other concepts that suggest variations of this scheme, the SPI
framework is currently the most widely accepted classifcation of cloud
computing. The NIST uses this framework, and most cloud service
providers support the concept.
Diagram X.3 shows the development path of cloud computing and
the SPI model, and how service providers have adopted the various service
models that the cloud offers. Cloud computing is a new concept, but the
technologies that came before it and have contributed to the development
of the cloud services that are commercially available today have been
around for decades. Supercomputing and cluster computing existed 20
years ago, and grid computing was based on these platforms.
Although there are many similarities between utility computing
and cloud computing, the former includes information processing as yet
another service, like water or electricity. The similarity breaks down at this
point, as cloud computing includes more elements within the service than
utility computing does. Table X.1 shows the features of cloud computing
ECLAC 278
in relation to other technologies. Because it developed from other IT
architectures, cloud computing is much more comprehensive than its
predecessors and encompasses their features.
Diagram X.3
DEVELOPMENT OF CLOUD COMPUTING AND SOFTWARE
PLATFORM INFRASTRUCTURE (SPI)
Cluster computing
Grid computing
Utility computing
Cloud computing
PaaS IaaS SaaS
Google App Engine
salesforce.com
Microsoft Azure
Zoho Suite
MobileMe
Google Docs
Amazon ECZ y S3
Sun Microsystems
Terremark
DropBox
Source: D. Baran, “Cloud Computing Basics”, WEBGUILD, July 2008 [online] http://www.webguild.
org/20080729/cloud-computing-basics.
Table X.1
CHARACTERISTICS OF CLOUD COMPUTING
IT architectures Metered use Web-based costumer Distributed processing
Cloud computing X X X
Utility computing X
Grid computing X
Virtualization
technology
X
Source: F. Belfort, “Panorama do mercado brasileiro de Cloud Computing”, presentation at the Cloud
Conference 2012, Frost & Sullivan, August 2012.
a) Software as a cloud service
The NIST defnes SaaS as follows: “The capability provided to
the consumer is to use the provider’s applications running on a cloud
infrastructure. The applications are accessible from various client devices
through either a thin client interface, such as a web browser (e.g., web-
based email). The consumer does not manage or control the underlying
cloud infrastructure including network, servers, operating systems,
storage, or even individual application capabilities, with the possible
Broadband in Latin America: Beyond Connectivity 279
exception of limited user-specifc application confguration settings” (Mell
and Grance, 2011).
At a high level, SaaS offers various benefts throughout the
organizational structure. First, it enables an organization to outsource
application hosting to an independent software vendor or other software
service provider. This nearly always reduces the cost of licenses, hardware,
management and other resources needed for internally hosting the
application. SaaS also benefts the application provider by increasing its
control over the use of the software, limiting the distribution of unlicensed
copies and allowing the software vendor to exert greater upgrade and
patch management control. SaaS also enables the provider to create and
control multiple revenue streams with a one-to-many model, reducing
duplication of software packages and overhead. Moreover, end users in
remote or branch offces can access the application more readily through a
browser, and start-up is enormously simplifed. Apart from modifcations
to peripheral devices (such as frewalls) to allow specialized ports
references, the end-user’s hardware requirements are also minimal.
b) Platform as a cloud service
The NIST defnes PaaS as follows: “The capability provided to the
consumer is to deploy onto the cloud infrastructure consumer-created or
acquired applications created using programming languages, libraries,
services, and tools supported by the provider. The consumer does not
manage or control the underlying cloud infrastructure including network,
servers, operating systems, or storage, but has control over the deployed
applications and possibly confguration settings for the application-
hosting environment.”
The PaaS model offers lower entry costs for application designers and
distributors, supporting the complete development of the web application
software lifecycle and eliminating the need to acquire hardware and
software resources. A PaaS solution can include a complete end-to-end
application solution for the development, testing and deployment of an
application or it can be smaller and more specialized, focusing on a specifc
area such as content management. If a software development platform is
to be considered as a PaaS solution it must have the following elements:
(i) baseline supervision of applications must be used in order to improve
the processing platform; (ii) the solution must offer perfect integration
with other cloud resources, such as databases and other web-based
infrastructure components and services; (iii) dynamic multi-tenancy
must be achievable, and collaboration via the cloud between developers,
customers and users over the entire lifecycle of the software must be readily
achievable; (iv) security, privacy and confdentiality must be maintained as
a basic service; and (v) the development platform must be browser-based.
ECLAC 280
c) Infrastructure as a cloud service
The IaaS model is the one that most clearly demonstrates the
difference between traditional IT infrastructure and cloud-based
infrastructure services. The NIST defnition for IaaS is as follows: “The
capability provided to the consumer is to provision processing, storage,
networks, and other fundamental computing resources where the
consumer is able to deploy and run arbitrary software, which can include
operating systems and applications. The consumer does not manage or
control the underlying cloud infrastructure but has control over operating
systems, storage, and deployed applications; and possibly limited control
of select networking components (e.g., host frewalls).”
The benefts of IaaS are similar to those of other *aaS models. Smaller
frms have access to a much greater range of IT solutions and technology,
and dynamic infrastructure scalability allows IaaS consumers to adapt
their needs at a more detailed level. The costs of organizing computing
systems infrastructure have traditionally made up a large portion of
business expenses. Leasing or buying dedicated hardware and software
and engaging in-house experts or consultants consume a signifcant
portion of any frm’s resources. Using the IaaS model (often in combination
with the SaaS or PaaS model) offers a level of scalability that can swiftly
respond to demand in a way that conventional acquisition, deployment
and maintenance of IT infrastructure cannot.
3. Enabling technologies
The essential enabling technologies for cloud computing include fast wide-
area networks, powerful and inexpensive computer servers, and virtualization
for high-performance commodity hardware, as described below.
a) Fast wide-area networks
An important requirement for cloud computing is the existence of
high-speed networks, more specifcally wide-area networks (WAN) that
allow access to remote servers at high speeds in the order of gigabits per
second (or even faster in the future). The evolution of WANs has allowed
the development of very high-speed connections worldwide, an aspect
that facilitates the transfer of enormous volumes of information in very
short times. Map X.1 shows average Internet access speed worldwide;
many regions, particularly North America, Asia-Pacifc and Europe, have
high averages.
Much of Latin America still has relatively low average Internet access
speeds, refecting the fact that the region lacks widespread deployment of
Broadband in Latin America: Beyond Connectivity 281
physical networks and that deployment of fbre-optic backbone networks
is still limited. The challenge facing cloud service providers in Latin
America is to ensure that their customers have access to suffcient speed
and, even more important, that the WANs have enough capacity and speed
to provide reliable and secure services.
Map X.1
AVERAGE INTERNET CONNECTION SPEED AROUND THE WORLD
Source: Akamai, State of the Internet Report, second quarter 2012.
Note: The boundaries and names shown on this map do not imply offcial endorsement or acceptance
by the United Nations.
b) Powerful, low-cost computer servers
Another important enabling technology is very high-speed, low-
cost servers capable of handling clients’ processes in the cloud in cost-
effcient way. Moore’s law, which holds that the number of transistors in an
integrated circuit will double approximately every 18 months, allows these
objectives to be achieved. The direct consequence of Moore’s Law is that
prices fall as performance rises: a computer that is today worth US $3,000
will cost half that amount next year, and will be obsolete in two years.
In 26 years the number of transistors in an integrated circuit will have
increased 3,200 times. This empirical law is one of the main reasons why
server equipment can be expected to fall in price and increase in speed; in
other words, servers will be even more powerful and less expensive.
ECLAC 282
c) Virtualization for high-performance
commodity hardware
An indirect consequence of Moore’s Law is that high-performance
hardware can be found on the market at prices that render them an
undifferentiated product (a commodity). Such equipment can be offered by
cloud service providers in an almost generic manner, so that it makes sense
to offer them in virtual form. The need for specialized equipment is receding
constantly, and frms’ IT requirements can be covered by generic systems.
4. Deployment models
In each of the three service models described there are multiple deployment
models. For example, an SaaS delivery model can be presented to users
in one of many deployment models, such as a private or a public cloud.
These deployment models are not technically or functionally related to
any delivery model. Any one of them can exist in any of the deployment
scenarios, although a given service/deployment pairing model may be
more common than others (for example, SaaS/public cloud).
Depending on an organization’s use of the cloud and its relationship
with the enterprise as a whole, these deployment models may make use
of either external or internal clouds. Each of these models must share
the following basic principles: (i) each deployment model uses devices
connected to the Internet; (ii) each model offers dynamic scaling of virtual
resources; and (iii) the users of each model normally have no control over
the technology that is used.
The cloud computing deployment models are:
• Private cloud. The cloud infrastructure is operated for exclusive
use by a single organization. It may be managed by the
organization or a third party, and it may exist on or off premises.
• Community cloud. The cloud infrastructure is shared by various
organizations and is compatible with a specifc community that
has shared concerns (e.g., mission, security requirements, policy,
and compliance considerations). It may be managed by the
organization or a third party, and it may exist on or off premises.
• Public cloud. The cloud infrastructure is provisioned for open
use by the general public or a large number of enterprises and is
owned by a service vendor.
• Hybrid cloud. The cloud infrastructure is a composition of two
or more distinct cloud infrastructures (private, community, or
public) that remain unique entities but are bound together by
standardized or proprietary technology that enables data and
application portability (“cloud bursting” is an example).
Broadband in Latin America: Beyond Connectivity 283
An organization may deploy one or several models, depending on
which offers the best solution. For example, a critical application that has
compliance or some other type of security specifcations may need a hybrid
or private cloud model. On the other hand, a general application needed for
a temporary project could better lend itself to a public cloud. In none of these
four models is the physical location of the infrastructure or the application
specifed. A co-location facility may host both public and private clouds.
5. Service layers
The provisioning of cloud services can be more readily understood when
analysed from the viewpoint of the architecture’s component layers.
Craig-Wood (2010) offers an analysis of the layers that make up the typical
structure of networked computing and provides a visual diagram with
examples for each layer. Diagram X.4 shows the service layers and the
levels at which the different services are located within this structure.
Diagram X.4
SERVICE LAYERS DEFINITION
Service stack
components
as a Service
S
o
f
t
w
a
r
e
/
A
p
l
i
c
a
t
i
o
n
P
l
a
t
f
o
r
m
I
n
f
r
a
s
t
r
u
c
t
u
r
e
I
n
f
r
a
s
t
r
u
c
t
u
r
e
-
a
a
S
P
l
a
t
f
o
r
m
-
a
a
S
a
S
o
f
t
w
a
r
e
-
a
a
S
*
People
Client device
Hosted
aplication
software
Infrastructure
software
Operating
system
Virtualisation
layer
Physical
servers
Networking
and firewalling
Data centre
mechanical
and electrical
Your Company Ltd
Source: Craig-Wood (2010).
Note: Brand names for illustrative/example purposes, only and examples are not exhaustive.
a
Assumed to incorporate subordinate layers.
ECLAC 284
In the frst place, this structure of layers bears a great similarity to
the seven-layer ISO/OSI (International Standards Organization’s Open
System Interconnect) structure. The four bottom layers include most of the
hardware used for data communication and processing. These four layers
are the ones that an IaaS provider would offer to its clients, as indicated in
the right-hand portion of the diagram. Platform refers to the next two higher
layers (operating system and infrastructure software), which are offered by
PaaS providers in addition to the underlying infrastructure. It is at this level
that applications are developed and where cloud computing has a specifc
market of application developers. Finally, the software and the applications
are located in the two top layers, with the next-to-top coinciding with the
platform. This is where SaaS, the most common cloud service, is offered.
C. Problems and challenges
Cloud computing, like any paradigm shift, presents problems and
challenges when it is used in IT environments that usually kept everything
“under control”. The typical IT department today consists of personnel
devoted exclusively to developing computer networks and equipment and
keeping them running. These tasks take a great deal of time and much of
a frm’s resources. While there are a number of challenges to be addressed
in adopting cloud computing, the main ones are described below.
1. Security and privacy
Many potential users say that concerns about data privacy and
confdentiality make them less willing to use cloud computing services
for sensitive data. In the cloud, data is stored on remote machines that
are shared with other users. This makes many users concerned about the
potential for business competitors or government authorities to access
their data in the cloud without their knowledge or consent.
Governments would like to mandate and apply national legal
requirements for data stored in the cloud, and many have done so. Given
the cross-border nature of the cloud, though, national measures to protect
data privacy and confdentiality have only limited capacity to reassure
users. There is a desire for greater global consistency in data privacy
requirements applying to the cloud, but government actors point out
that fundamental differences in their approaches make comprehensive
international agreements on this issue unlikely. For example, the United
States has a stricter regulatory regime for specifc sectors, such as health
care, where privacy and confdentiality issues are especially sensitive. The
European Union has blanket data privacy laws.
Given these regulatory challenges, users concerned about data
privacy and confdentiality will ultimately have to rely on market
Broadband in Latin America: Beyond Connectivity 285
mechanisms to assess provider reliability. Nonetheless, there is no
guarantee that appropriate market mechanisms will emerge in a timely
fashion (World Economic Forum, 2011).
Users are concerned that data in the cloud are more susceptible to
cyberattacks, because aggregating multiple users’ data and services on a
single platform makes them a more attractive target. Providers point out
that no security mechanisms are infallible and all come with trade-offs.
Encryption can be expensive, and using hypervisors to virtually isolate a
user’s applications and data can still leave vulnerabilities. More broadly,
industry and governments alike express concern that technical security
mechanisms such as encryption could give users a false sense of security.
Encryption is only as effective as the user’s control of who has the key,
and it does not solve the problem of a malicious insider or of users being
manipulated into giving access. These concerns are bound up with wider
questions of how to manage and verify identities.
When users store their data on their premises, it is clear who is
accountable if the data is corrupted, lost or temporarily inaccessible. This
is not necessarily the case when the data is stored in the cloud. When it
is unclear whether a problem lies with the cloud provider or with the
networks the user is using to access the cloud, users are concerned that
they will be unable to establish who is accountable and to seek redress.
Many users’ data may be shared on one machine, so users are
concerned about the potential for problems with one user’s services
affecting another’s. Government actors express concern about the resilience
of cloud providers to distributed denial-of-service (DDoS) attacks and note
there is a disincentive for providers to report breaches and problems. Some
industry stakeholders, however, believe they are already being transparent
enough, especially given that most client agreements require the service
provider to notify the client of any data leakage or loss.
With all its benefts, cloud computing also brings with it concerns
about information security and privacy, as a result of its size, structure
and geographic dispersal. Krutz and Dean Vines (2010) identify the
following problems:
• Leakage and unauthorized data access among virtual machines
running on the same server.
• Failure of a provider to handle and protect sensitive information
properly.
• Release of critical or sensitive data to law enforcement or
government agencies without client´s knowledge and approval.
• Inability to follow compliance and regulatory requirements.
• System crashes or failures that make service unavailable for
extended periods of time.
ECLAC 286
• Hacker invasions to obtain and distribute sensitive information
from client applications.
• Provider’s lack of robust security protection.
• Low interoperability of the kind that that would allow a client to
easily move applications among different cloud providers and
avoid lock-in.
Cloud users also worry whether their data will be continuously
available over long periods of time, and whether the cloud provider might
surreptitiously exploit sensitive data for its own purposes.
2. Reliable network access
Reliable network access includes many aspects, some of which have been
covered above in relation to data protection measures. However, reliable
network access includes a technical component that, while implicit in the
issues discussed, is not very carefully analysed. That component is reliable
access to the network.
Reliable network access implies a level of quality that allows data
to be transferred as required, with confdence that communications
will always be available when needed. Although this aspect does not
create major problems in countries with highly developed networks, in
many Latin American countries there are still problems with the lack of
availability of networks and their reliability.
In the case of cloud computing, communication services are
generally outsourced to specialized frms (“telcos”). Proper provision of
services usually requires an above-average level of service and availability,
and this must be spelled out in the service agreements. If the networks
are proprietary, there must be assurance that individual failures will not
compromise the functioning of the communications systems as a whole.
In other words, there must be enough alternative communication routes to
cope with interruptions (planned or not) and to foresee situations such as a
mass cut-off of communications because of a submarine cable break. This
is particularly important when it comes to storing data in various locales
in different countries or regions.
3. Legal and regulatory aspects
It is not surprising that the nascent cloud service industry fnds government
regulations to be complex and contradictory. There is little agreement
on the kind of regulation necessary in general or internationally. For
example, one European Union regulator does not believe that data privacy
Broadband in Latin America: Beyond Connectivity 287
regulations should be updated for cloud computing, a viewpoint quite
different from that of the industry (World Economic Forum, 2010).
The World Economic Forum suggests that governments adapt
and harmonize regulations for the cloud in order to improve their
enforceability and reduce divergence across jurisdictions, while
considering the maturity of the industry. There is widespread frustration
among market stakeholders with the regulatory framework for cloud
computing, especially in the areas of data privacy and security. Regulations
are often inconsistent, conficting and diffcult to comply with for users
and providers operating globally. This holds back users from moving to
the cloud, as they fear regulatory provisions are insuffcient to protect
their data from being unduly accessed by law enforcement or retained by
providers. And when regulations effectively force data to remain within
national borders (either directly by restricting data transfers outside the
jurisdiction, or indirectly through a lack of cross-jurisdictional alignment)
they hold back cloud providers from making improvements that would
create economies of scale by using multiple storage locations.
As a long-term goal, governments may wish to explore a macro-
regulatory framework that will be more fexible and keep pace with rapid
technological change. Options include co-regulation, whereby industry
takes the lead in identifying necessary provisions and governments take a
policy and oversight role. This would mean a harmonized approach to the
underlying principles that guide regulation, which currently differ across
jurisdictions: the sectoral approach to data privacy regulation taken by the
United States stands in particular contrast to the European Union’s more
universal one. Minimum regulatory standards are not a solution; they are
often not enough to reduce complexity because they do not stop countries
from introducing additional provisions.
As a step in this direction, governments should continue the
dialogue with providers to better understand the impact of regulations.
Data protection authorities can play an important role in interpreting and
harmonizing legal frameworks to more effectively meet user and provider
needs. They can also provide understandable and authoritative guidance
about respective responsibilities and protections, and the recourse
available in the event of breaches.
4. Resistance to change
One fnal factor, which is not related to technology but rather to human
behaviour, is resistance to change. This is a very natural human trait that
comes to the fore when traditional ways of doing things are confronted
with new ones. Table X.2 shows the differences between the conventional
IT model and the models associated with cloud computing.
ECLAC 288
Table X.2
COMPARISON BETWEEN CONVENTIONAL MODELS AND CLOUD COMPUTING
Conventional model Cloud computing
Purchase model Buy assets and build a
technological architecture
Lease services
Business model Pay for fxed and administrative
assets
Monthly payment
Access model From the internal network to the
corporate desktop
Via Internet, to any device
(telework)
Technical model Individual leasing, not shared, static Scalable, fexible, dynamic,
multiuser
Marketing model Face-to-face sales Online sales
Service model Call centre - service points Contact centre - Anywhere
Source: Chamber of Commerce of Medellín, Computación en la Nube. 2011.
For one, a frm’s IT personnel, after years of buying assets and
building systems, are now faced with the challenge of leasing those same
services. This can generate resistance among IT department staff who will
feel that their main responsibilities are being outsourced. The business
model changes, and instead of paying for hardware and software at
unplanned intervals the frm now makes monthly service payments to the
cloud provider.
The virtualization of activities is another aspect that can readily
create resistance, as now, instead of single-point access, systems can be
used from anywhere. Concerns associated with lack of control over data
in the system are frequently reported as a potential problem by IT staff
and managers. Yet perhaps the most alarming aspect for those responsible
for systems is that the servers that used to be next to their offces will
disappear and be replaced by servers in locations over which they have
no control. These aspects relating to the new cloud working environment,
together with the uncomfortable feeling that what were considered routine
activities (like server maintenance and software updates) are no longer
required and their time must be devoted to development work, tend to
spark unease and resistance among IT personnel.
D. Migration to cloud computing
Migration to cloud computing occurs for many reasons, the most frequent
being to reduce the costs of operating in-house computer infrastructure.
But there are also other reasons why a user may decide to migrate, and
when: these will depend on various aspects relating to how a business is
structured. Those in the best position to analyse migration possibilities are
likely to be users who have not yet begun operations, as they do not need
Broadband in Latin America: Beyond Connectivity 289
to consider what they will do with an existing computer system. Those
who have the greatest diffculties are likely to be users who have invested
heavily in equipment and software.
1. Reasons for migrating
An analysis of the reasons that an enterprise may have for migrating its
IT operations to cloud computing shows that all kinds of considerations
come into play. Lockheed Martin Corporation (2011) conducted a survey
of frms that adopted cloud computing for their IT activities; responses to
the question as to why they chose this technology are set out in fgure X.1.
Figure X.1
WHY CLOUD COMPUTING?
(Percentages)
31
32
33
37
50
51
52
53
57
59
60
61
Reduce IT capital costs
Reduce IT operating costs
Ability to obtain new resources quickly
Simplify IT infrastructure and management
Increase overall IT flexibility and agility
Ability to replace existing solutions
Ability to scale up and down quickly
Improve service availability and uptime
Don’t have to spend funds to expand
Energy efficiency, lower carbon footprint
Reduce IT staff
Reduce IT staff
Source: Lockheed Martin Corporation, When the cloud makes sense, The Download - Cloud Computing
Research Study, 2011.
The frst reason is savings, both on licenses and on service
management and equipment. If an infrastructure is 100% cloud-based
there is no need to install any kind of hardware beyond the terminals. It is
this simplicity for the user and the fact that a much smaller investment is
required to begin work that is the real strength of this technology.
There are also savings on hardware: there is no need to choose
between a laptop or a desktop computer, or to determine which is cheaper
and, often, faster. In the world of cloud computing, the user can buy an
inexpensive thin client laptop that can be connected to a screen and a
keyboard. Then, all that is needed is to connect to a provider and have all
the performance and memory desired. Next, when the user would normally
have to replace an obsolete laptop, it can still use its thin client, because the
provider, and not the equipment itself, supplies the performance.
ECLAC 290
Fast, low-risk deployment is another beneft. With cloud computing
infrastructure, work can start very quickly. There is no need to spend a lot
of time and money up front before launching a session in a new solution.
Applications based on this technology will be available in a matter of a
few weeks, and they will be highly customized.
Automatic upgrades are still another incentive, as the cloud
computing model has no adverse impact on IT resources. When upgrading
to the latest version of the application, the new technology does not
force the user to decide between upgrading or keeping its work, because
customizing and add-ons are automatically saved during the upgrade.
Information portability is a further attraction. Initially, most
providers geared their services to corporate users, but with the passage of
time private users have begun to make massive use of this concept, almost
unknowingly, through the use of mobile telephone services, in particular
smartphones and tablet PCs.
The cloud model is also more environmentally friendly. Desktop
computers use about 150 watts each; with cloud computing there is no need
for machines that consume more than 10 watts. Of course, to this must be
added the power consumed by data centres. But these serve a great number
of terminals, so the cloud offers substantial overall savings in energy.
Another beneft has to do with the content industry. In the past,
various methods were used to prevent illegal copying of music and
flms, but each method had its problems. Some CD players couldn’t
read protected copies of CDs. DRM7 technology is yet another effort
by some companies to protect their contents, sparking debate about its
implementation and the design of methods for evading it. However, the
cloud will provide another digital rights management (DRM) beneft for
content producers offering flms, games or music directly to the consumer.
Content will be designed to be run on a cloud computing system and it
will take more time and effort to make illegal copies of flms and music
distributed by this medium. The beneft also extends to consumers: the
cost of videos and games will fall because users will pay only for what
they use during the time requested.
2. Assessing the benefits of cloud computing
Assessing the benefts of cloud computing is complicated because it can
impact different sectors of the economy in different ways. The United
Kingdom’s Centre for Economics and Business Research evaluated the
impact of adopting this technology in France, Germany, Italy, Spain and
the United Kingdom: the fndings are shown in fgure X.2 (Centre for
Economics and Business Research, 2010).
Broadband in Latin America: Beyond Connectivity 291
Figure X.2
BENEFITS OF CLOUD COMPUTING IN FIVE EUROPEAN COUNTRIES
(Percentages)
0
10
20
30
40
50
60
70
80
90
100
France Germany Italy Spain United
Kingdom
EMEA
Business development opportunities Business creation
Net cost savings Multiplier effects
Source: Centre for Economics and Business Research, 2010.
Quantifying the economic benefts of cloud computing involved
aggregating the individual benefts identifed and quantifed, taking into
account the macroeconomic outlook for each of the fve countries, and
assumed rates of adoption for each aggregate industry sector in each
country. Business development opportunities are cited in the report
as one of the economic benefts obtained, but in most of the countries
analysed (except for the United Kingdom) they do not exceed 15% of the
total. A much greater economic impact comes from business creation,
which in France exceeds 30% of the total. Net IT operating cost savings
(the difference between IT expenses before and after introduction of the
technology) are shown in red. The economic benefts on this score vary
between 15 and 20% of the total.
The last item analysed, accounting for 37% of the economic benefts,
is the multiplier effects on the economy. This item is generally complicated
to evaluate, as it requires a detailed analysis of the collateral benefts of
using the technologies and the positive impact on other related segments
of the economy. What is undeniable is how vast the multiplier effects of
cloud computing can be in an industrialized society.
Cloud computing technologies result in business IT budget cost
savings which, in turn, boost proftability. There are three ways for frms
to cut costs. These are:
• Capital expenditure: by eliminating server and storage costs
and replacing these with pay-as-you go cloud computing
capabilities, frms can reduce their net IT capital cost.
ECLAC 292
• Labour costs: by outsourcing IT services, frms can reduce their IT
headcount and/or redeploy staff into more productive areas of
IT departments such as application development.
• Power and cooling costs: by eliminating the need to power and
cool server and data centres, frms can save a substantial amount
on energy bills.
But adopting cloud computing technology will involve new
expenses, particularly when using hybrid and public cloud computing
models. The cost reduction from use of this technology in the fve countries
analysed is shown in fgure X.3.
Figure X.3
SAVINGS FROM CLOUD COMPUTING
(Percentages of spending on information technologies)
0
10
20
30
40
50
60
70
80
90
Capital expenditure Staffing cost Power and cooling costs
Private Hybrid Public
Source: Centre for Economics and Business Research (2010).
3. Considerations for successful migration
There are many advantages to adopting cloud computing. As indicated
above, many benefts can be reaped from its implementation (IT Industry
Innovation Council, 2011). They are:
(i) Simplicity: The cloud reduces to almost negligible the technical
knowledge a business owner needs. The technical side of setting
up, operating and maintaining any part of the ICT process is
taken care of by the cloud provider, enabling frms to focus on
their business.
Broadband in Latin America: Beyond Connectivity 293
(ii) Accessibility: Cloud accessibility makes information available
to small businesses irrespective of where it is stored, through a
multitude of devices and limited only by Internet access.
(iii) Flexibility: In many respects, cloud technology offers a greater
value proposition to small businesses than to larger enterprises
when it comes to fexibility. Given the speed with which the
business and technology landscape changes, small businesses
need to be responsive, nimble and equipped to adapt their
operations quickly. With cloud-based services priced per user
or per subscription, small businesses can grow their technology
capability in parallel with their business requirements and
growth. Rather than depending on potentially risky forecasts
and predictions, with cloud-based solutions they can adapt and
expand IT capability on an ad hoc, as-needed basis. And given the
cyclical nature of business, they can increase and reduce additional
resource capability in line with actual business fuctuations. They
also have fexibility in terms of how much cloud resources they
use. While on the one hand they can base their entire IT systems
in the cloud with negligible IT hardware requirements onsite, on
the other they can select only those components that suit them
and their business (e-mail hosting, databases or storage).
(iv) Affordability: Enterprise business applications such as customer
relationship management (CRM) programmes and enterprise
resource programmes (ERP) are typically expensive to acquire,
install and maintain. In a cloud computing model, these sorts of
applications become much more affordable and thus accessible
to SMEs. Capital investment in infrastructure, including servers,
storage and software, is avoided. Hardware and software
upgrades and software version control become unnecessary,
with responsibility falling instead to the cloud service provider.
(v) Improved productivity: With routine IT and network management
tasks performed by the cloud service provider, SMEs can avoid
the need to dedicate or redirect resources to maintain the
systems they rely on to run their business. This directly improves
productivity and enables SMEs to focus on issues relevant to
growing and improving the business and staying competitive.
One of the frst considerations when migrating to cloud computing
is to decide which services to outsource. This depends on such aspects as
the desired level of security, the need to develop proprietary applications
and legal or contractual requirements with respect to maintaining control
over information.
ECLAC 294
In the conventional service provision model, the frm or organization
hosts both the infrastructure and the platform and applications on its
premises. A migration to IaaS means leasing all the infrastructure from
the service provider, so that the application development platforms and
the applications themselves, as well as the software, remain under the
control of the organization. A more advanced level involves migrating
the layers corresponding to the applications development platform to the
cloud, outsourcing the development platform services under the PaaS
scheme while retaining the applications and the software in the frm.
Finally, when all the layers are placed in the cloud, SaaS is contracted from
the provider and very little remains housed within the organization.
If there is a dedicated IT department structure, phasing in is
probably the least traumatic approach for gradual migration. Nevertheless,
many frms opt to migrate all the services and infrastructure directly and
to contract SaaS from the provider at the outset, especially in the case of
SMEs with little in-house infrastructure and variable IT needs.
When it comes to data centres, it is useful to examine their value
chain. Diagram X.5 illustrates this chain and the service providers
involved at each step.
Diagram X.5
THE EXTENDED DATA CENTRE VALUE CHAIN
Transport/
infrastructure
Value-added
network
Hosting
Security Service
development
Creation of
content and
applications
Systems and
applications
integration
Client
interface
Network service providers IT integrators
Telecom operators Software suppliers
Pure data centres
ASP/SaaS/Cloud computing
ISP/Web hosting
Full service provider (partnerships)
Source: CloudConf LATAM, 2012.
The frst two elements of the chain comprise the services of network
providers, which include transport and infrastructure as well as value-
added networks. In addition to networks, many telecommunications
operators offer hosting, security and service development, tasks typical of
a conventional pure data centre. The creation of content and applications is
normally in the hands of the software suppliers, but various other service
providers such as IT integrators and ASP, SaaS and ISP providers that host
webpages can also offer them. Systems or applications integration and
Broadband in Latin America: Beyond Connectivity 295
client interface (through CRMs, for example) can be handled by the SaaS
provider, but there are also other possibilities.
The choice of outsourcing arrangements will depend to a large
extent on the availability of services in the country or region. Ideally, it
is best to maintain a contractual relationship with a single provider.
However, it is rare to fnd providers that are involved in the entire value
chain and can offer all the services.
E. The situation in Latin America
The current situation in Latin America with respect to cloud computing is
diffcult to assess because such services have been available in the region
for only a short time. “Cloud adoption is increasing but is still not at US
levels... lots of evaluations and planning but not too many implementations.
Current implementations are mostly private and in multinationals
following headquarters guidelines. The main benefciaries of the increase
in cloud adoption (besides the companies that actually implement it) are
storage, virtualization, networking and security vendors” (Grava, 2012).
One of the main concerns in Latin America is about moving
information outside of national boundaries, which makes it diffcult to
adopt shared environments in the United States or Europe.
1. Legal and regulatory environment
Latin America is making progress with legislation for the comprehensive
governance of business relations in cloud computing. There is much
debate in Colombia over the controversial Lleras Law that will regulate a
series of aspects relating to information and could grant Internet providers
tight control over data, something that could affect cloud computing either
positively or negatively, depending on how it is applied. In Chile, work
has started on a draft data protection law. Argentina has some regulations
in this area, such as Law 25.326 on the protection of personal data, but
stakeholders agree that it needs to be updated promptly because it has
shortcomings similar to those found in the rest of the region.
Business Software Alliance (2011) examined preparedness for cloud
computing in 24 countries (see fgure X.4), including 3 in Latin America
(Mexico, Argentina and Brazil).
Mexico and Argentina earn a total score of more than 50%, while
Brazil ranks last in the list of countries selected. The fndings show
convincingly that, in a region with an enormous potential for the growth
of cloud computing, the regulatory and legal framework is in its infancy
and, in the few cases where it exists, inadequate.
ECLAC 296
Figure X.4
CLOUD COMPUTING REGULATION PREPAREDNESS SCORECARD
(Percentages)
0
10
20
30
40
50
60
70
80
90
Japan United
States
Rep. of
Korea
Spain Malaysia Mexico Argentina India Brazil
ICT readiness, broadband deployment
Promotion of free trade
Support for industry-led standards, international harmonization of rules
Intellectual property
Cybercrime
Security
Privacy
Source: Business Software Alliance, BSA Global Cloud Computing Scorecard - A Blueprint for economic
opportunity, 2011.
The only Latin American country certifed for international data
transfer is Argentina, with Uruguay and Mexico on its heels (Martinez
Fazzalari, 2011). The Argentine law follows the European approach, in the
sense that it prohibits the transfer of personal data to any country that does
not offer adequate protection. Mexico was the frst country in the region
to adopt a regulation on cloud computing, as part of its federal law on
protection of personal data held by private parties. “What is noteworthy is
that Mexico has and enforces explicit conditions for the contracts signed,
for example, by a frm with a third-party cloud service provider, as well
as technical requirements and information requirements for guaranteeing
that when the information is sent to the cloud there is no risk to the rights
of personal data owners” (Commission on Transparency and Access
to Information of the State of Nuevo León, 2012). Chile has developed a
number of legal instruments in this area; they are listed in table X.3.
In conclusion, legal and regulatory provisions in Latin American
countries with the greatest degree of cloud computing use are still in their
infancy and, in many respects, inadequate to the task. More coordination
at the regional level is needed to establish a uniform legal and regulatory
framework for the free transfer of data between countries and to ensure
security for users of this technology.
Broadband in Latin America: Beyond Connectivity 297
Table X.3
CHILEAN REGULATION OF CLOUD COMPUTING
Law/
regulation
Date Description
Law 19,799 2002 Electronic documents and electronic signature, including certifcation
(Ministry of Economy, Development and Reconstruction)
DS/181 2002 Regulations to Law 19.799 on electronic documents and electronic
signature, including certifcation (Ministry of Economy, Development
and Reconstruction)
DS/81 2004 Technical standards for central government entities on interoperability
of electronic documents (General Secretariat of the Offce of the
President)
DS/83 2004 Minimum security standards for the use, storage, access and
distribution of electronic documents. Also regulates the electronic
relationship between the Government and the citizens (General
Secretariat of the Offce of the President)
DS/93 2006 Standards for proper handling of electronic messages, in particular
spam (General Secretariat of the Offce of the President)
DS/100 2006 Minimum standards for government websites (General Secretariat of
the Offce
of the President)
DS/158 2007 Amends DS/81 of 2004. Approves technical standards for central
government entities on interoperability of electronic documents
(General Secretariat of the Offce of
the President)
DS/271 2008 Regulates the registration of document systems in the archives of the
Administrator of Systems and Metadata for central government entities
(Ministry of Economy, Development and Reconstruction)
Guía Web
V2.0
2008 Guidelines for design and development of government entities’ web
platforms (Ministry of Economy, Development and Reconstruction)
Source: Prepared by the author on the basis of offcial documents.
2. Degree of adoption in the region
One of the frst problems in addressing this issue is the lack of information
and data, as there are not many specialized frms producing studies of this
kind in Latin America. There are recent data available for Brazil, which
hosted CloudConf LATAM 2012; for other countries, a few private or
academic entities have produced studies on the adoption of technology,
based primarily on surveys.
IDC Worldwide IT Public Cloud Services Forecast (2010) estimates
that cloud-based services will increase by 61% in Latin America between
2010 and 2014, with especially strong growth in the areas of storage,
servers and applications. This probably refects the fact that Latin America
is at an early stage of development in the use of cloud computing and that,
for this reason, fast growth rates are due to initial mass adoption of the
technology. Figure X.5 shows the degree of adoption of this technology
ECLAC 298
around the world, highlighting the position of Brazil over the coming two
years. The growth of non-critical applications is relatively high, refecting
perhaps the proliferation of applications such as social networks. Figure
X.6 shows growth by level of processing in Brazil. Although it has no
proper legislation for the development of cloud computing, Brazil is
setting the pace for the region.
Figure X.5
ADOPTION OF CLOUD COMPUTING AROUND THE WORLD
(Percentages)
Now
Using for critical applications
Using for less important applications
In two years
Using for critical applications
Using for less important applications
All nations
Australia
Brazil
China
France
Germany
Japan
Singapore
United Kingdom
United States
All nations
Australia
Brazil
China
France
Germany
Japan
Singapore
United Kingdom
United States
19
25
25
29
25
30
33
28
26
23
13
17
0 20 40 60 80
13
35
37
33
8 3
37
25 25
28
30
31
28
33
29
30
22
29
29
29
27
34
35
31
13 6
43
29
0 20 40 60 80
Source: Accenture Institute for High Performance.
Figure X.6
GROWTH OF CLOUD COMPUTING IN BRAZIL BY LEVEL OF PROCESSING
(Percentages)
19
41
86
39
42
88
Platform
Infrastructure
Application
2012 2010
Source: F. Belfort, “Panorama do mercado brasileiro de Cloud Computing”, presentation at the Cloud
Conference 2012, Frost & Sullivan, August 2012.
Colombia is another country that is promoting the use of cloud
technologies: an independent evaluation found that nearly 50% of frms
adopting cloud services do so through SaaS (see fgure X.7).
Broadband in Latin America: Beyond Connectivity 299
Data centres are an important aspect of cloud computing, as the
provision of mass storage and offshore processing requires relatively
large spaces in different countries. Figure X.8 is a projection by McKinsey
of the data centre landscape in fve countries of Latin America (Moreira,
2012). It shows that by 2014 Brazil will have the most room for the growth
of cloud computing data centres. Mexico and Argentina will also be
developing greater capacities for offering storage and data processing
technologies.
Figure X.7
CLOUD SERVICES USED IN COLOMBIA
(Percentages)
0
10
20
30
40
50
60
Software as a service
(SaaS)
Platform as a service
(PaaS)
Infrastructure as a service
(IaaS)
Source: F. Rueda, El uso de la computación en la nube (Cloud Computing), Departamento de Ingeniería de
Sistemas y Computación, Universidad de Los Andes, Colombia, 2012.
Figure X.8
OUTLOOK FOR DATA CENTRE GROWTH IN LATIN AMERICA
(Thousands of square meters)
0
50
100
150
200
250
300
2010 2011 2012 2013 2014
Colombia Chile Argentina Mexico Brazil
Source: R. Moreira, “IT and Cloud Computing in Brazil: Public Policies”, CloudConf LATAM 2012, August,
São Paulo, Brazil, 2012, based on McKinsey projections.
ECLAC 300
F. Mobile cloud computing
Uptake of cloud computing has been expanding worldwide since 2007, as
has the use of smartphones that connect to the Internet thanks to the rapid
growth of wireless network technology. Mobility and ubiquity are both
important features of the new-generation networks that offer a range of
customized services through various kinds of terminals and access modes.
The combination of a ubiquitous mobile network and cloud computing has
generated a new modality: mobile cloud computing (MCC).
The Mobile Cloud Computing Forum defnes MCC as follows:
“Mobile Cloud Computing at its simplest refers to an infrastructure where
both the data storage and the data processing happen outside of the mobile
device. Mobile cloud applications move the computing power and data
storage away from mobile phones and into the cloud, bringing applications
and mobile computing to not just smartphone users but a much broader
range of mobile subscribers” (Aepona, 2010).
1. The structure of mobile cloud computing
The basic structure of mobile cloud computing relies on the fusion of two
technologies: mobile data access and cloud computing. It leverages both
technologies to provide the user with tools for on-demand access to powerful
computing capacity, free of the constraints of a physical connection. Diagram
X.6 shows the typical architecture of a mobile cloud computing system.
Diagram X.6
ARCHITECTURES OF MOBILE CLOUD COMPUTING
Source: Dinh, Lee, Niyato and Wang (2012).
Broadband in Latin America: Beyond Connectivity 301
In MCC, the mobile devices are connected to the mobile networks
via base transceiver stations (BTS), access points or satellites that establish
and control the connections (wireless links) and functional interfaces
between the networks and mobile devices. Mobile users’ requests and
information (e.g., ID and location) are transmitted to the central processors
that are connected to servers providing mobile network services. Here,
mobile network operators can provide mobile users with services such as
AAA (authentication, authorization, and accounting) based on the home
agent (HA) and subscribers’ data stored in databases. After that, the
subscribers’ requests are delivered to a cloud through the Internet. In the
cloud, cloud controllers process the requests to provide mobile users with
the corresponding cloud services.
The great advantage of this architecture is that the user can use a
mobile device to draw on powerful computing facilities that overcome
intrinsic limitations (storage, computing capacity) and work from any
location where an untethered connection is available. In theory, a mobile
device with fairly simple computing capacities and memory could provide
a user with the same facilities as a sophisticated computer, provided
there is reliable, high-speed communication and bandwidth available.
In a sense, this is a return to the 1980s concept of powerful mainframe
computers and a set of access devices known as “dumb terminals”. In this
case the mainframe computer is replaced by the cloud, the wiring and
communications by the Internet, and the terminals by mobile devices. The
advantages for the mobile user are:
• Extended battery life. This is one of the main concerns of a mobile
user, and offoading to the cloud the running of applications that
consume CPU time can provide energy savings of up to 45%.
• Improved data storage capacity and processing power. Memory
capacity is a constraint with mobile devices, and the ability
to store information in the cloud allows signifcant savings of
this resource. Similarly, applications that demand substantial
processing power can be offoaded to the cloud with considerably
improved processing times.
• Reliability. Storing data or running applications in clouds is an
effcient way to boost reliability, as various computers in the
network keep backup copies or interim results.
Nevertheless, there are many challenges to be overcome in
deploying MCC, as examined in the following subsection.
2. Challenges
Mobile devices (like smartphones and tablet PCs) are increasingly
becoming an essential part of life as the most effective, convenient and,
ECLAC 302
above all, ubiquitous communication tools. Mobile users have at their
disposal a wide array of services ranging from mobile applications (such as
iPhone apps and Google Apps) that run on the devices or on remote servers
via wireless networks. The rapid progress of mobile computing is driving
the development of IT technology as well as the felds of commerce and
industry. However, the mobile devices face many challenges in terms of
resources (e.g., battery life, storage, and bandwidth) and communications
(like mobility and security). These resource constraints are holding back
the improvement of service quality. Table X.4 lists the challenges that MCC
must address, and some possible solutions.
Table X.4
MOBILE CLOUD COMPUTING CHALLENGES AND SOLUTIONS
Challenges Solutions
Limitations of mobile devices Virtualization and image management
Task migration
Quality of communications Bandwidth upgrading
Reducing data delivery time
Division of applications services Elastic application division mechanism
Source: H. Qi and A. Gani, Research on Mobile Cloud Computing: Review, Trend and Perspectives, Faculty
of Computer Science and Information Technology, Kuala Lumpur, University of Malaya, 2011.
a) Limitations of mobile devices
In examining the use of mobile devices in the cloud the frst thing
that stands out is resource constraint. Although smartphones have been
much improved in aspects such as CPU and memory capacity, storage,
screen size, wireless communication, geolocation technology and
operating systems, they still have serious limitations such as insuffcient
computing capability and power for running complicated applications. In
comparison with PCs and laptops in a given situation, smartphones have
3 times less processing capacity, 8 times less memory, 5 to 10 times less
storage capacity and 10 times less network bandwidth (Qi and Gani, 2011).
Normally, a smartphone needs to be charged every day as battery
power is depleted by dialling calls, sending messages, surfng the Internet,
accessing social networks, and other Internet applications. According
to current trends, increased mobile computing capacity and rapid
development of screen technology will lead to ever more complicated
applications running on smartphones. If battery technology cannot be
improved apace, then saving battery power will become an increasingly
important issue.
Broadband in Latin America: Beyond Connectivity 303
The processing capacity, storage, battery life, and communication
capabilities of smartphones will improve as mobile computing develops.
However, their limitations will remain one of major challenges in mobile
cloud computing. The challenges and solutions for tablet PCs are similar.
b) Communication quality
Unlike fxed networks (copper or fbre) where the physical
connection ensures bandwidth consistency, the data transfer rate in a
mobile cloud computing environment is constantly changing and the
connection is discontinuous due to unequal network coverage and hand-
off. Furthermore, data centres in large companies and service provider
Internet resources are usually far from end-users, especially mobile device
users. In a wireless network, network latency delay may be as much as
200 ms in the last mile while it is only 50 ms in a wired network. Other
variables such as changes in application throughput, user mobility, and
even the weather can lead to changes in bandwidth and network coverage.
Therefore, handover delay in a mobile network is greater than in a wired
network. Figure X.9 shows average access speed as a function of the ratio
of mobile to fxed broadband connections in Latin America.
Figure X.9
ACCESS SPEED BY TYPE OF CONNECTION IN THE REGION
Argentina
Bolivia (Plur. State of)
Brazil
Chile
Colombia
Costa Rica
Ecuador
El Salvador
Guatemala
Mexico
Nicaragua
Panama
Paraguay
Peru
Dominican Rep.
Uruguay
Venezuela (Bol. Rep. of)
-
500
1 000
1 500
2 000
2 500
3 000
3 500
4 000
- 1 2 3 4 5 6
A
v
e
r
a
g
e

I
n
t
e
r
n
e
t

a
c
c
e
s
s

s
p
e
e
d

i
n

K
b
p
s
Ratio of mobile to fixed broadband subscribers
Source: Prepared by the author on the basis of data from ITU and Akamai
Note: The size of the circle is proportionate to the number of Internet users.
As the ratio of mobile to fxed broadband connections rises, average
access speed declines. Nevertheless, many countries with a high mobile-
to-fxed broadband ratio do not have a good correlation with the dotted
line, which is an adjustment curve for data on Latin America. In other
ECLAC 304
words, countries below the dotted line have a below-average connection
speed and are likely to have mobile bandwidth performance that is more
marginal the further they are from the curve.
c) Partitioning application services
In a mobile cloud computing environment, resource constraints
mean that some computing-intensive and data-intensive applications
cannot be run on mobile devices because they can be a substantial power
drain. Therefore, applications and use of cloud computing capacity must
be split to achieve those purposes. The core computing tasks are processed
in the cloud, and the mobile devices are responsible only for some simple
tasks. In this process, the major issues affecting the performance of mobile
cloud computing are data processing in data centres and the mobile device,
network handover delay, and data delivery time.
For a given standard, providing a guaranteed quality of cloud
service means considering the following: (i) optimal application split
between cloud and mobile device; (ii) interaction between low latency
and code offoad; (iii) suffcient bandwidth capacity between cloud and
mobile device for high-speed data transmission; (iv) user-oriented cloud
application performance; (v) mechanism for self-adaptation of mobile
cloud computing; and (vi) optimal levels of consumption and excess
capacity of mobile devices and cloud servers. The following strategies can
be used to respond to these challenges:
• Upgrade bandwidth for wireless connection, making the web
content more suitable for the mobile network using regional
data centres.
• Deploy the request processing node at the “edge” of the cloud in
order to reduce data delivery time.
• Duplicate mobile devices in the cloud, using virtualization and
image technologies, to handle data-intensive computing and
power-intensive computing processes, such as virus scanning
in mobile devices.
• Dynamically optimize applications for insertion in the cloud
and the split with mobile terminals.
As an example of the third strategy, the concept of CloneCloud was
introduced by Chun in 2011. The method uses virtual machine migration
technology to offoad execution blocks of applications from mobile devices
to the CloneCloud, seamlessly and partially, in order to extend (either
fully- or semi-automatically) or modify smartphone-based execution to a
distributed environment (smartphone computing plus cloud computing). In
a CloneCloud system the clone is a mirror image of a smartphone running
Broadband in Latin America: Beyond Connectivity 305
on a virtual machine. Unlike smartphones, a clone has more hardware,
software, network and power resources in a virtual machine, which
provides an environment more suitable to processing tasks. Diagram X.7
shows the execution of tasks within a CloneCloud architecture.
Diagram X.7
ARCHITECTURE OF A CLONECLOUD SYSTEM
Execution blocks
Operating system
Hardware
Smartphone image
Smartphone
Execution blocks
Operating system
Hardware
Execution blocks
Operating system
Virtual hardware
Virtual machines
Hardware
Distributed computing
Single-machine computing
Source: Dinh and others, A Survey of Mobile Cloud Computing: Architecture, Applications, and
Approaches, Wireless Communications and Mobile Computing, Wiley, 2012.
As the diagram shows, a smartphone task is divided into fve
different execution blocks, and the smartphone is cloned (virtualized) as
an image in a distributed computing environment. Next, the image passes
some computing- or power-intensive blocks (the green blocks) to the cloud
for processing. Once those execution blocks have been completed, the
output will be passed from the CloneCloud to the smartphone. A major
advantage of the CloneCloud is enhanced smartphone performance.
Another advantage is reduced battery consumption, as the telephone
does not use its CPU as frequently. A disadvantage is handover delay
due to bandwidth limitations. As the speed of data transmission between
smartphones and the base station is not consistent, the CloneCloud will be
unavailable if mobile users enter a signal blind spot.
3. Mobile cloud computing
The forecast that mobile cloud computing would generate US$ 29 billion in
revenues in 2014 was made in the wake of an estimate by ABI Research that
the number of mobile subscribers in the cloud would reach nearly a billion
that year, representing about 19% of all mobile subscribers (Bahl, 2011). This
represents a huge increase over the 42.8 million mobile cloud computing
subscribers in 2008. The forecasts, by region, are shown in fgure X.10.
ECLAC 306
Figure X.10
MOBILE CLOUD COMPUTING REVENUE BY REGION
(Millions of dollars)
2009 2010 2011 2012 2013 2014 2015
Rest of world 736 946 1 051 1 156 1 471 1 576 1 891
Asia-Pacific 6 620 8 511 9 141 10 507 11 873 12 609 13 659
Europe 2 312 2 732 3 047 3 572 3 783 3 993 4 413
North America 4 623 5 569 6 830 8 091 9 457 10 822 11 558
0
5 000
10 000
15 000
20 000
25 000
30 000
35 000
M
i
l
l
i
o
n
s

o
f

d
o
l
l
a
r
s
Source: V. Bahl, “Cloud in the Palm of your Hands”, 8 July 2011, on the basis of data and projections from
ABI Research.
Although predictions indicate strong growth in cloud use for
mobile applications in Latin America, its share is still small. Projections
for the “rest of world” (which includes Africa and the region) are for
only US $1.9 billion by 2015. There is not much additional information on
current or future use of mobile cloud computing in the region. In any case,
if the trend for the use of cloud facilities through smart mobile devices
(telephones or tablets) points to 20% by 2014, this would mean that a ffth of
mobile broadband customers in Latin America would be using the cloud.
G. “Everything” in the cloud: reality or utopia?
The main incentive for using cloud computing is and will continue to
be the ability to use, as if they were owned, computer resources (such as
data storage, processing, programmes and databases) that are beyond the
capacity or reach of the individual or business user. The premise is that if
enough data transmission capacity is available, and if it is reliable and of
good quality, many of the tasks that normally require large processing or
local storage capacity can be offoaded to the cloud. But can “everything”
really be sent to the cloud without problems? This section examines the
possibility that processing will really allow computing resources to be
distributed ubiquitously and that consumers can make use of them just as
they use electricity or drinking water.
Broadband in Latin America: Beyond Connectivity 307
1. Back to the past
Computing systems have evolved ever more swiftly in recent decades, and
their dissemination has also accelerated thanks to the omnipresent Internet.
The “network of networks” has not only contributed to the dissemination of
information and access to knowledge but has also created another concept
of society, thanks to popularization of its use in activities of all kinds. But
what if all these advances are closing a cycle that would mean going back to
what things were like several decades ago? Diagram X.8 shows how mass
use of cloud computing could close the cycle.
Diagram X.8
EVOLUTION OF COMPUTING TECHNOLOGIES AND NETWORKS
1980s
Mainframes and
dumb terminals
2020s
Smart cloud and dumb devices?
1990s
Networked smart PCs
and laptops
2000s
Notebooks and netbooks
with Internet Access
Networked servers
2010 to present
WANs with ultrafast connections to the
cloud, tablets and smartphones with
mobile broadband and cloud computing
Source: Prepared by the author.
In the 1980s, the typical structure for information systems in
use since the 1970s involved mainframe computers, where intelligence
resided outside the access terminals (which were therefore called
“dumb”). Although personal computers became popular in the mid-
1980s, they did not have the capacity to perform very complicated
tasks. It was only in the 1990s that smart desktops and laptops became
widespread and local area networks (LAN) came into general use. In any
case, the intelligence was no longer located in centralized equipment, and
individual device capacities improved steadily. The turn of the century
saw the spread of portable devices with Internet access, and intelligence
shifted to network servers. In the early 2010s, networks started to rely
increasingly on connectivity and the Internet, and devices with wireless
broadband connections became popular. This was also the era of cloud
ECLAC 308
computing, using both fxed and mobile connections, based on the idea
that intelligence would migrate to the cloud and allow access devices
with limited capacities to draw on powerful remote computing systems.
What will the coming decade bring? While some believe that
evolution is not cyclical but tends instead to gradual improvement,
reality increasingly points to a system where intelligence resides not
in devices or networks but outside them. Dependence on connectivity
is increasing; not being connected to the Internet is equivalent to being
isolated, especially when it comes to mobile devices. As the tendency is
for each person to have one or more connectivity devices, expectations
are that terminals will go back to being, essentially, dumb and that
everyone will access a smart cloud to store data, run processes, manage
operations and perform any kind of computing.
2. Technical considerations
The notion that everything can be stored or run in the cloud poses
some technical challenges that the architects of these structures are
seeking to resolve by various means. The main challenges for ensuring
that most processing and storage can migrate to the cloud are (i) data
synchronization and sequencing; (ii) transport of information; (iii) delay
management and (iv) growth management.
a) Data synchronization and sequencing
Information management and processing requires that
processes be transferred and run remotely in the cloud. Running
requires that processes be synchronized and that the data processed
be delivered in sequence. Cloud computing faces many of the same
technological challenges as parallel computing, with the additional
complication that some of the processes may be running on the
user’s own device. The management of information and processes in
sequence also places burdens on cloud computing, because resources
such as cached memory and temporary storage must be duplicated
to maintain the order of intermediate data and results. Structured
information management is a technique that can soften the impact of
managing process synchronization, since the processing of fat fles
and unstructured information imposes very heavy requirements on
cloud processing and creates synchronization problems. Another way
of ensuring synchronization and sequencing is to use ultrafast servers
which allow for shorter processing times than if the user were using
its own computing resources. To put it another way, the slowest cloud
processing must be faster than the fastest processing by customers
using their own resources.
Broadband in Latin America: Beyond Connectivity 309
b) Transport of information
Management of information and processes in the network requires
massive transport of information to and from the cloud by users.
Communication platforms vary tremendously not only in their technology
but also in their quality of performance in different geographical locations.
Because cloud computing servers can be located in different countries,
each with its own standards not only for contracted speed but also for
transmission quality, the transport of huge volumes of data can encounter
diffculties of all kinds. The most complicated problem, of course, is the
inability to transmit or receive information on time, thereby exceeding
user´s waiting capacity. This is particularly true when it comes to real-time
processing of information or high-performance online applications. The
issue of broadband connection to the cloud and its components is crucial,
particularly for applications that handle enormous quantities of data.
c) Delay management
The issue of managing delays is closely related to management of
distributed information in the network. Diverse structures and network
computing architectures have been designed to provide resilience to
unpredictable delays, including that of retaining execution of delay-
sensitive applications in user devices. However, this is not always possible,
since one of the main reasons for moving everything to the cloud is
precisely that the user cannot perform the processes on its own system
within a reasonable time. The problem of delay management is closely
related to that of synchronization and sequencing, which is alleviated
through the use of ever faster links and the splitting of processes into
subprocesses that can be run in parallel on different servers. Delays are
particularly complicated in the management of mobile cloud computing,
given the variable conditions of wireless communication.
d) Growth management
As cloud computing spreads, storage and processing capacity needs
are growing almost exponentially. That capacity, distributed among users
around the world, is now migrating to a limited number of processing
centres that, for reasons of security, must have duplicate resources (such as
memory, servers and optical disks) to avoid delays or loss of data.
Gmail and Facebook offer two examples of the problems of
managing accelerated growth. They both require very high processing
and distributed storage capacities, and those needs are increasing daily
with the growing ranks of increasingly demanding users. Users do not
in fact need to store the information in their accounts (except perhaps
temporarily), as it is available somewhere in the cloud. What is required,
ECLAC 310
then, is a great increase in resources to meet users’ needs to have
“everything” in the cloud. This can be achieved technologically by adding
to and expanding computer resources or processing centres as migration
continues. Relieving the user of the need to have signifcant computing or
storage capacity in its own equipment, then, has a heavy impact in terms
of network overload, but the benefts outweigh the drawbacks.
3. Legal and procedural aspects
As indicated, the technical aspects associated with transferring virtually
all or at least most processes and applications to the cloud can be resolved.
The technology is such that in the future there will be a cloud with enough
capacity to handle processes that are now performed outside it. The
chief diffculties in making the cloud ubiquitous are not necessarily of a
technical nature, however: legal and procedural aspects may in fact be more
important. The factors with the greatest impact on full transfer to a global
cloud environment are (i) the diversity of the applicable legal systems;
(ii) responsibility for data integrity; and (iii) security and confdentiality.
a) Diversity of applicable legal systems
Cloud computing is not confned to any country or region, as it can
readily operate in a variety of geographic locations at any point in time.
It is here that differences among countries in the legal treatment of a
specifc situation become relevant. For example, storage of a certain type of
information in one country may constitute a crime in another country. The
storage and distribution of users’ personal information such as medical
records or images can be subject to treatment that differs among countries.
One very contentious case involves the United States Patriot Act (Library
of Congress, 2001), which allows intelligence agencies to review and
intercept any information that could be related to terrorist activities. This
law is confned in its application to the territory of the United States; it
cannot be applied legally to information in the cloud that is hosted outside
that country, even though the user might be physically located there. Even
more complex is the case of a cloud computing user in a third country
who may have no idea that the servers are located in the United States and
may not suspect that its information is subject to scrutiny by that country’s
intelligence agencies. Uncertainty about which legislation they will be
subject to is a compelling reason for the reluctance of potential cloud
computing customers to use the service.
b) Responsibility for data integrity
More than one service provider is involved in cloud computing, so
responsibility for the integrity of the data delivered and processed in the
cloud may easily fall on more than one party. The cloud service provider
Broadband in Latin America: Beyond Connectivity 311
contracts the services of an Internet access provider and may also use the
services of still other providers. Responsibility for data integrity is then
shared, and the problem of clearly identifying the accountable party when
information is lost or corrupted represents an obstacle to full migration to
the cloud.
c) Security and confidentiality
A fnal crucial aspect of migrating data and processes to the
cloud is the question of how secure and confdential that information
will stay. The problems that large corporations and cloud service
providers have faced with leakage of their own or clients’ confdential
information have been a warning signal that user privacy can
be compromised. While it is true that such information could be
intercepted and extracted by hackers even if users keep it in their own
computer systems, the users are responsible for ensuring that this does
not happen. But who is accountable if that information is somewhere in
the cloud and even the provider does not know its exact location? This
is certainly a tremendous obstacle to generalized adoption of cloud
services and is frequently cited as the key consideration in making the
decision to migrate.
H. Best international practices
Because cloud computing is relatively new, best practices have been
emerging only recently and in countries (especially the United States and
in Europe, as well as some countries in Asia-Pacifc) that were pioneers in
its deployment. In Latin America there are incipient efforts to introduce
this technology and lay the groundwork for widespread uptake.
The World Economic Forum partnered with the consulting frm
Accenture to prepared a compendium on the promotion of cloud
computing and guidelines on priorities for industry and government
(World Economic Forum, 2011). These action areas are set out in
diagram X.9 and are the result of careful analysis of the problems and
benefts of cloud computing. This section transcribes the best practices
listed in the study.
While the underlying issues are complex and contentious, eight
critical action areas were selected by government representatives and
companies —including many of the largest cloud providers and regulators
from Europe and North America— and then confrmed in the private
session about cloud computing held during the 2011World Economic
Forum Annual Meeting. These action areas are put forward as a charter for
further engagement among key stakeholders. They are intended to form a
ECLAC 312
cohesive agenda, bringing together several areas in which there are existing
but disparate initiatives. This step should lead to industry and government
collaboration to further defne and implement the necessary actions to
move the agenda forward and accelerate the uptake of cloud technologies.
Diagram X.9
ACTION AREAS FOR PROMOTING CLOUD COMPUTING
Benefits
• Accelerate innovation
• Better serve customers
• Lower organizational expenses
• Improve IT efficiency and flexibility
• Bring socioeconomic improvements
• Level the playing field
Action areas
1. Explore cloud benefits
2. Understand and manage cloud risks

3. Promote service transparency
4. Clarify and enhance accountability
5. Ensure data portability
6. Facilitate interoperability
7. Adapt and harmonize regulation
8. Provide sufficient connectivity
Issue areas
Data governance issues
Data location and jurisdiction; privacy and
confidentiality; data ownership
Security issues
Interoperability and portability; reliability;
service-level commitment; ecosystem maturity
Business environment issues
Authorized access; integrity and
availability; data loss; data destruction
Source: Prepared by the author, on the basis of World Economic Forum, Advancing cloud computing:
What to do now? Priorities for Industry and Governments, 2011.
1. Explore cloud benefits
Cloud ecosystem participants should dedicate additional resources to
understanding the benefts of cloud and accelerating the adoption of
innovative applications of cloud technology. Topics include product and
process innovation and job creation, collaboration, broad delivery of IP,
government effectiveness and effciency, and other economic benefts.
Underlying many of the issues discussed previously is a sense that
the benefts of cloud computing —beyond those related to IT effciencies—
are not well understood. This manifests itself as a problem in two main
ways. First, users may be held back from moving to the cloud if they
perceive the risks more clearly than the benefts. Second, regulators fnd
it hard to make balanced decisions that are in line with the legal principle
of proportionality if they lack a clear sense of how their decisions could
potentially impact the macroeconomic and societal benefts of the cloud
as well as the risks. The principle of proportionality argues, among other
provisions, that regulation should detract as little as possible from the
benefts of what is being regulated.
Broadband in Latin America: Beyond Connectivity 313
2. Understand and manage cloud-related risks
Relevant stakeholders (providers and government) should encourage
research into the unique risk drivers in cloud computing and identify
potential solutions. The fipside of clearly understanding the potential
benefts of cloud is ensuring that perceptions of risk are also grounded
in reality. It is arguable that several of the stakeholder concerns apply just
as much to the public Internet as to the cloud, where data centres may be
protected by security mechanisms that are so sophisticated they actually
reduce risk rather than exacerbate it. If concerns are indeed overstated, the
development of the cloud would be needlessly held back.
Risk mitigation strategies need to address the different risk
profles of different types of data, such as personal data and trade
secrets. Innovative approaches to managing risk could include industry
players developing codes of conduct and mutual assistance schemes
whereby providers agree to assume responsibility for each other’s service
commitments in the event of outages or breaches. A better understanding
of risks would also facilitate the development of nascent cloud insurance
models to offer compensation to customers in the event of losses caused by
the cloud.
3. Promote service transparency
Providers of cloud services should make available to customers information
about how their services are provided and how they perform. This includes
letting customers know how data is secured, where data is stored and/or
what jurisdictional provisions apply, how and by whom it can be accessed,
and how it can be deleted. Greater transparency (i.e. public disclosure)
about cloud computing would go a long way towards addressing many of
the stakeholder issues detailed above —notably privacy and confdentiality,
data ownership, security, liability and reliability. Clearer and more easily
accessible information about cloud service delivery models and offers
would accelerate the development of the market by improving levels of
user trust and facilitating the creation of aggregated services provided by
multiple providers.
4. Clarify and enhance accountability
Industry, regulatory bodies and third parties should collaborate to
create and implement more consistent and comprehensive approaches to
accountability for how cloud services are provided. Complementing greater
transparency, greater clarity about accountability would accelerate uptake
of cloud computing among potential users, who are currently reluctant to
ECLAC 314
entrust mission-critical services to the cloud. Users want to know who is
accountable if service levels are unsatisfactory, if they are unable to access
data they put in the cloud or if it is accessed by unauthorized persons or
government agencies. In particular, users want clarity about accountability
for service delivery in situations where providers leverage sub-contractors,
get acquired or go out of business.
5. Ensure data portability
Cloud service providers should provide ways for users to easily retrieve
data they have input to clouds, without an onerous fee and in a timely
manner. The fear of vendor lock-in holds back many potential users
of cloud, while many government stakeholders are concerned about
maintaining competitiveness in the cloud market. These concerns are
lessened if it becomes quicker, easier and cheaper for users to move data,
and perhaps applications, between different cloud providers and between
user premises and the cloud. Users should be aware, however, that due to
economies of scale in the cloud and particular cloud architectures, it may
be economically unfeasible to roll back from the cloud to an on-premises
solution. Work on facilitating data portability also needs to be aligned
with work on common approaches to data ownership and protection,
law enforcement access and liability. Providing meta-data and context
information, in addition to the actual data entered, can signifcantly
increase the options available to customers.
6. Facilitate interoperability
Industry players should pursue the evolution of cloud offerings with the
goal of facilitating interoperability among multiple (private and public)
clouds. This will accelerate the growth of the overall cloud ecosystem.
There has been notable progress recently in developing offerings that
allow users to customize their own solutions by simultaneously using
services from multiple cloud providers. As with data portability, every step
towards greater interoperability helps to address stakeholder concerns
about competitiveness and lock-in. It may also accelerate innovation and
help address challenges related to data privacy and security. Fostering
cloud interoperability will also likely extend to a broad range of ecosystem
players, including providers of connectivity and application developers,
who will need to adopt relevant architectures and provide enabling
services such as highly reliable cross-cloud connectivity.
7. Adapt and harmonize regulatory frameworks
Governments should adapt and harmonize regulations relevant to cloud
with the aim of improving their applicability and reducing divergence
Broadband in Latin America: Beyond Connectivity 315
across jurisdictions, while considering the maturity of the overall industry.
There is widespread frustration among stakeholders about the regulatory
environment for cloud computing, especially in the areas of data
privacy and security. Regulations are often inconsistent, conficting and
diffcult to apply for users and providers operating globally. This holds
back users from moving to the cloud, as they fear regulatory provisions
are insuffcient to protect their data from being unduly accessed by law
enforcement or retained by providers. And when regulations effectively
force data to remain within national borders —either directly by imposing
restrictions on data transfers outside the jurisdiction, or indirectly through
a lack of cross-jurisdictional alignment— they hold back cloud providers
from realizing improvements that come from achieving scale through
multiple locations.
8. Provide sufficient connectivity
Industry, government and relevant agencies should identify connectivity
requirements for cloud services (wired and wireless) and promote
the commensurate deployment of networks. To be able to use cloud
computing with confdence, users need easy access to the cloud. They need
guarantees about the speed, reliability and robustness of networks, both
fxed and mobile. In particular, in an environment where market needs
may be moving faster than the technology, users need to be confdent that
current telecom investments will be suffcient to support future services.
Given that cloud computing uses data centres that need to be able to
handle massive amounts of traffc, the planning of networks and data
centres needs to be coordinated. Developing a framework describing what
services can be provided with various levels of connectivity could also
help national governments promote and prioritize investments that will
sustain future growth opportunities.
ECLAC 316
Bibliography
Aepona Ltd. (2010), Mobile Cloud Computing - Solution Brief, November.
Akamai (2012), “State of the Internet”, Q2 Report [online] http://spanish.akamai.
com/enes/stateoftheinternet/.
Bahl, Victor (2011), Cloud in the Palm of your Hands 8 July.
Baran, Daya (2008), Cloud Computing Basics, WEBGUILD, July [online]
http://www.webguild.org/20080729/cloud-computing-basics.
Belfort, Fernando (2012), “Panorama do mercado brasileiro de Cloud Computing”,
presentation at the Cloud Conference 2012, Frost & Sullivan, August.
Business Software Alliance (2011), BSA Global Cloud Computing Scorecard - A Blueprint
for economic opportunity.
Centre for Economics and Business Research (2010), The Cloud Dividend: Part One -
The economic benefts of cloud computing to business and the wider EMEA economy
France, Germany, Italy, Spain and the UK, Report for EMC, December.
Chamber of Commerce of Medellín for Antioquia (2011), Computación en la Nube.
Commission on Transparency and Access to Information of the State of Nuevo Leon
(2012), Cloud Computing y Protección de Datos en América Latina, 27 May [online]
www.qumulos.com/tweets/mexico-pionero-en-el-cloud-computing/.
Craig-Wood, Kate (2010), “IaaS vs. PaaS vs. SaaS Defnition”, 18 May [online]
http://www.katescomment.com/iaas-paas-saas-defnition/.
Dinh, Hoang T., Chonho Lee, Dusit Niyato and Ping Wang (2012), A Survey of
Mobile Cloud Computing: Architecture, Applications, and Approaches, Wireless
Communications and Mobile Computing–Wiley.
Grava, Wilson (2012), Paths to the Cloud: Cloud Computing in Latin America and the
New Channel Engagement Model, 26-27 April [online] www.hawkeyechannel.com.
IDC Worldwide IT Public Cloud Services Forecast (2010)
IT Industry Innovation Council, Australia (2011), Cloud Computing – Opportunities
and Challenges, 11 October.
Krutz, Ronald L. and Russell Dean Vines (2010), Cloud Security - A Comprehensive
Guide to Secure Cloud Computing, Wiley Publishing.
Library of Congress (2001), “Uniting and Strengthening America by Providing
Appropriate Tools Required to Intercept and Obstruct Terrorism (USA Patriot
Act)”, http://thomas.loc.gov/cgi-bin/bdquery/z?d107:H.R.3162.
Lockheed Martin Corporation (2011), When the cloud makes sense, The Download -
Cloud Computing Research Study, 1105 Government Information Group Custom
Report.
Martínez Fazzalari, Raúl (2011), Aspectos Regulatorios - Los retos de Cloud Computing,
Logicalis Now, July.
Mell, Peter and Timothy Grance (2011), The NIST Defnition of Cloud Computing,
Recommendations of the National Institute of Standards and Technology, September.
Moreira, Rafael H. R. (2012), “IT and Cloud Computing in Brazil: Public Policies”,
CloudConf LATAM 2012, August, São Paulo, Brazil.
NIST (National Institute of Standards and Technology) (2011a), Cloud Computing
Synopsis and Recommendations, Computer Security Division Information,
Technology Laboratory, May.
___ (2011b), NIST US Government Cloud Computing Technology Roadmap - Technical
Considerations for USG Cloud Computing Deployment Decisions, November.
Broadband in Latin America: Beyond Connectivity 317
Qi, Han, and Abdullah Gani (2011), Research on Mobile Cloud Computing: Review,
Trend and Perspectives, Faculty of Computer Science and Information Technology,
University of Malaya.
Rueda, Francisco (2012), El uso de la computación en la nube (Cloud Computing),
Departamento de Ingeniería de Sistemas y Computación, Universidad de Los
Andes, Colombia.
World Economic Forum/Accenture (2011), Advancing cloud computing: What to do
now? Priorities for Industry and Governments.
___ (2010), Exploring the Future of Cloud Computing: Riding the Next Wave of
Technology-Driven Transformation.
Chapter XI
The challenge of over-the-top
content and services
Juan José Ganuza and María Fernanda Viecens
1
A. Introduction
Incumbent telecommunications operators providing services such as fxed
and mobile telephony, broadband and pay-TV are being invaded by the
online content industry and so-called over-the-top (OTT) applications,
services and content. The best-known examples of these include Skype,
WhatsApp, video games and flm streaming (Netfix, Pandora). A key
feature of this industry is that the Internet service supplier (ISP) is not
involved in the distribution of OTT applications, services or content and
does not earn any revenue from them. Yet OTT providers need Internet
providers’ infrastructure to reach their users, and they deliver products
that generally compete with the ISPs’ proprietary ones, such as voice,
instant messaging and online TV.
The advent of smartphones with ever-lower prices, together with
the upgrading of access networks, has helped fuel the growth of OTTs in
the communications market. And technological change has transformed
1
Juan José Ganuza is a professor at Universitat Pompeu Fabra, Barcelona; and María
Fernanda Viecens is a research fellow at Universidad de San Andrés and the National
Scientific and Technical Research Council (CONICET), Buenos Aires.
ECLAC 320
the creative industries and impacted their production and supply
cost structure. For example, digitalization has reduced conservation,
reproduction and distribution costs (Weeds, 2012), and this has fuelled the
exploding supply of online content.
Paradoxically, broadband offered by incumbent operators provides
the platform for developing the new businesses that now threaten them.
Meanwhile, OTTs fuel burgeoning traffc fow and demand for greater
bandwidth, which creates the need for network investment.
Despite the OTT boom, in the literature on the reasons for incumbent
operators to invest in infrastructure provision, and particularly next-
generation access networks (NGAN, fbre optic networks) there is little
discussion of the impact of OTTs on investment incentives for incumbent
operators.
2
As for infrastructure investment needs, in developed countries the
focus is on how to bring fbre to homes (or, similarly, to guarantee high
access speeds),
3
while the strategies and initiatives deployed in Latin
America prioritize correcting regional imbalances in the supply of frst-
generation broadband services, with much more modest access-speed
targets. In other words, while developed countries focus on upgrading
service quality, plans in the region aim to expand geographic coverage
and increase the number of connections (for an analysis of government
initiatives in the region see the chapter by Galperín, Mariscal and Viecens
herein). Slow access speeds, or limited coverage for high-speed services,
could restrict the use of OTTs among the population of Latin America or
confne access to groups with high access speeds (certain neighbourhoods
and large cities).
This chapter examines the level of development and uptake of
the over-the-top market in Latin America. In particular, it discusses the
strategies deployed by incumbent operators in Latin America in response
to the threat posed by OTTs. The chapter is organized as follows: the next
section describes OTT content, applications and services and identifes the
dilemma or challenge that each one represents for incumbent network
operators. Section C reviews the literature on the subject, while section D
analyses the OTT services and content market in Latin America, together
2
As noted below, Ganuza and Viecens (2012) aim to fill part of that gap, using an industrial-
economy theoretical model to analyse the interaction between the development of the
NGN industry and the content market. Many studies analyse the incentives to invest
and the effect of wholesale-access price regulation (see Hoernig and others, 2012 for a
review of this literature). Others focus on the high fixed-variable cost ratio of deploying
an industry based on fibre optics (Noam, 2010).
3
For example, the Digital Agenda for Europe calls for 100% of Europeans to have access at
speeds of 30 Mbps and for 50% to be subscribed to 100 Mbps connections by 2020.
Broadband in Latin America: Beyond Connectivity 321
with supply and the strategies deployed by incumbent operators. Section E
discusses the current situation and unresolved issues, while section F sets
forth conclusions.
B. Over-the-top services, applications and content
This section provides a description of OTT services with a view to
organizing available information on the subject that has yet to be set out
systematically. It also takes an initial look at the challenges that OTTs
pose for incumbent network and service operators. The approach is not
restricted to OTT content (which was the frst to use the expression OTT).
Table XI.1 outlines the different types of OTTs, providing examples, and it
identifes the main ways in which the incumbent operators’ main sources
of revenue are being cannibalized.
Table XI.1
DESCRIPTION OF OVER-THE-TOP-SERVICES
OTT
Minimum speed
requirements
for good-quality
access
Challenge for
the incumbent
operator
Implications for
the incumbent
operator
Communication
services
VoIP: Skype, chat
with and without
video, Gmail,
WhatsApp
1-2MBps Substitute for
SMS.
More competition
Loss of value
of proprietary
service
Applications Social networks:
Facebook, LinkedIn,
Twitter
1 MBps Substitute for
telephony and
SMS
More competition
Loss of value
of proprietary
service
Content OTT-TV, OTT-Video
streaming and video
on demand (VoD):
Netfix
Netmovies, Hulu,
UltraViolet,
Cuevana,
YouTube
6–10 MBps Substitute for
TV
Content supplier
disintermediation
Less potential for
differentiation
Loss of
advertising
audience
Online videogames 1–4 MBps
Online music 1–3 MBps
Source: Prepared by the authors.
The table lists three general types of OTT: voice and instant-
messaging services, applications basically related to social networks, and
video and audio content.
ECLAC 322
Bijl and Peitz (2010), discuss the profound impact that voice-over-
Internet-protocol (VoIP) telephony is having on the telecoms industry in
general. They analyse the competition between an incumbent operator
who, when an operator offering VoIP telephony enters the network, is
forced to offer this service as well, in addition to that provided over the
switched public telephony network. In this context, the authors study the
effect that regulated access to the switched network has on competition
and on incentives for investment in VoIP. Nonetheless, they restrict their
analysis to what they call “managed” VoIP and disregard “unmanaged”
VoIP such as Skype. In contrast, this chapter’s examination of OTT includes
“unmanaged” VoIP services, whose business models differ from those of
incumbent telephone service operators.
There is some disagreement as to whether instant messaging is a
substitute for SMS
4
as indicated in the table. While not a perfect substitute,
the capacity of instant messaging to eat into SMS revenue is undeniable.
According to Nikou and others (2012), European operators are experiencing
a drop in revenues as a result of OTT services (such as WhatsApp) and
social networks (like Facebook and Twitter). They note that young people
in particular are massively switching from SMS to these free-of-charge
Internet-based services.
5
Access to online or OTT content enables disintermediation: as
Internet access and use becomes more widespread and is available at high
speeds (made possible by NGNs), the owners and managers of certain
kinds of content no longer need to negotiate with telecom operators (or
television providers) to reach their customers and can do so directly
through a website.
6
The table highlights the ways most OTT services cannibalize the
revenues of incumbent operators, but it fails to note that OTTs also mean
higher management costs, congestion and infrastructure needs because
of greater use of capacity and demand for speed. For example, while
4
See, for example [online] http://www.pyramidresearch.com/points/item/120810.htm.
5
OTT TV/Video means the distribution of video or television directly over the Internet
to users connected to an electronic device. This differs from Internet Protocol Television
(IPTV), which is also TV by IP but of the “managed” type. IPTV requires a private cable
line and uses the Internet protocol (it uses the same IP to provide Internet), so a certain
minimum level of bandwidth is needed (at least 3-6 MBps download). In contrast, Web TV
and OTT TV/Video require even better access conditions (at least 6-10 MBps download).
High-definition (HD) requirements are more demanding still: 10-25 MBps for IPTV HD
and 25-50 MBps for video HD.
6
A very clear example of this is provided by the recently announced UltraViolet Internet
platform, www.uvvo.com, created by the major Hollywood film studios (Paramount
Pictures, Sony Pictures Entertainment, Twentieth Century Fox, Universal Pictures and
Warner Bros., among others) to offer consumers a wide selection and freedom to choose
digital films, television programmes, and other streaming entertainment. There are also
platforms that stream sporting events.
Broadband in Latin America: Beyond Connectivity 323
online video games and music streaming may not represent challenges
for the operators’ own products, they do drive network traffc up.
7
To provide a more complete picture of OTTs, this section concludes
with a map of the main agents involved (see diagram XI.1).
Diagram XI.1
MAP OF AGENTS
Providers of OTT communication services, applications and content
Device manufacturers
• Smart TV
• Video consoles
• Set-top boxes
• Smartphones
• Tablets
• Computers and PCs
Network, telecom services
and Internet operators
Users
Source: Prepared by the authors.
To be able to receive OTT services, users need a device and
Internet access, which means that OTTs, devices and Internet access
are complementary products for the user. At the same time, as shown
in table XI.1, OTTs deliver products that are close substitutes for those
supplied by telecom and Internet service providers.
8
The links between
the different types of agents defne a complex strategic relationship that
determines each party’s prices and incentives to invest. This chapter
focuses on the strategic relationship between OTTs and Internet
providers. The range of devices that can be connected to the Internet
is broad and varied; their manufacturers compete to attract users.
Nonetheless, although technological advances in these devices are
7
In response to the Telesemana.com (2012) survey question “Where do you think OTTs
have the greatest negative impact?”, 38% replied management costs; 49% mentioned
revenue cannibalization; and 13%, referred to relations with users.
8
Chen and Nalebuff (2006) interpret the emergence of services such as Skype as a problem
of competition between complementary services. The Skype voice service competes with
the telecom operator; but it also complements with the network operator from the user
perspective, because the user cannot use Skype without an Internet connection. This type
of rivalry raises concerns that the broadband supplier may have initiatives to degrade the
quality of Skype to make its own telephony service more attractive.
ECLAC 324
playing a fundamental role in the development of OTTs, only passing
reference will be made to them in this chapter.
9
C. Main findings in the literature
The development of OTTs is linked to two ongoing debates in the telecoms
literature: changes in the value chain caused by the booming online
industry, and whether maintaining network neutrality is a good or bad
idea. Changes in the value chain are a result of technological progress and
vigorous innovation in the sector. In the face of these changes, the debate
on network neutrality is being driven by incumbent network operators
who see value chain revenue shifts in which they are likely to be the
losers and are being forced to seek alternatives to reverse the process.
The following paragraphs discuss the main fndings and unanswered
questions in the literature.
1. Changes in the value chain and the
online-industry threat
The emergence and boom in Internet-based industries is posing major
challenges for some traditional sectors of the economy, such as printing,
music, advertising and communications media.
10
Incumbent telecom
operators are not ignoring the new markets and services that are emerging
on the back of Internet-access potential, particularly OTT services,
applications and content.
According to Grove and Baumann (2012), although incumbent
integrated operators (providing both infrastructure and proprietary
services) have tried to develop VoIP products and services, along with
video portals and social communities, and have also taken advantage
of the possibilities offered by IPTV, they have been unable to achieve
growth or penetration rates comparable to those achieved by services such
as YouTube and Netfix. The question is why these integrated operators
are consistently left behind by suppliers of services based purely on the
Internet (OTT services). They argue that, over the long haul, an integrated
operator could achieve better results by controlling both elements
9
Content Delivery or Distribution Networks (CDNs) have started to play an important
role in this group, by providing infrastructure solutions for more effective delivery to
customers. Some are owned by network operators and others by companies such as
Akamai and Microsoft. For a more detailed discussion of them and their impact in Latin
America, see León (2012).
10
For example, Seamans and Zhu (2010) conduct an empirical analysis of the effect on the
local newspaper market of a website starting to offer classified advertisement services;
Athey and others (2012) focus on the fact that Internet enables the consumer to switch
easily between different vendors.
Broadband in Latin America: Beyond Connectivity 325
(infrastructure and services); but, in the short run, a pure service provider
can improve performance more quickly because its product’s spatial
confguration is smaller.
Dedrick and others (2011) break down the mobile phone distribution
value chain, but a major shortcoming of their study is that they do not
consider the role of applications (including OTTs). Shin (2012) analyses
the development of mobile VoIP (mVoIP); following what Nikou and
others (2012) describe for European operators, Shin reports how mobile
phone operators in the Republic of Korea have faced a drop in voice calls
owing to Internet call services and applications based on text messaging
for smartphones. Feijoo and others (2012) track the development of mobile
gaming applications, noting that the arrival of the frst wave of smartphones
and fat-rate broadband connections in 2006 heralded a shift in market
power from network operators to the providers of applications and devices.
Nonetheless, they also note that it was the arrival of the iPhone in late 2007
which drastically rearranged the mobile gaming market.
11
Ganuza and Viecens (2012) analyse the options open to providers of
content (such as sports events and Hollywood movies) for offering their
products through online portals (OTT content) —in other words, ways to
reach consumers without having to negotiate with intermediate operators
(disintermediation). Their article shows that NGNs and the resulting high
access speeds will reallocate rents among agents in the value chain and
will, in particular, transfer them from network and service operators to
content providers. Disintermediation by content providers will cause
telecom service operators to lose an important source of differentiation
and income, forcing them to seek new resources to compete.
2. The network neutrality debate
Network neutrality means that all content, of whatever type or origin,
is treated equally by the infrastructure provider; discussion on this
subject is complex and controversial (see the chapter by René Bustillo
herein). Some infrastructure operators complain that content suppliers
are free-riding on their infrastructure and recognize that they are being
threatened by the new possibilities and initiatives provided by OTTs
(Huertas Sánchez and others, 2011).
12
Major Internet providers have
proposed that application and content providers should pay a surcharge
11
In this market, the standard consumption model involves downloading games to a
smartphone from the application store (Feijoo and others, 2012).
12
In relation to the use of AT&T infrastructure by Google, MSN, Vonage and others, Ed
Whitacre, former CEO of AT&T, famously remarked “Now what they would like to do is
use my pipes free, but I ain’t going to let them do that because we have spent this capital
and we have to have a return on it.” Business Week, 7 November, 2005.
ECLAC 326
for access to the ISPs’ residential customers, as well as differential rates
for priority content. For example, Shin (2012) reports that mVoIP triggered
a debate on network neutrality in the Republic of Korea, where telecom
operators are mobilizing to restrict smartphone applications from using
their networks for free Internet phone services.
The network neutrality debate has also been interpreted as a
confrontation between the United States (where the leading content
providers are located) and Europe (which has strong incumbent network
operators). European frms are trying to put pressure on companies with
network-intensive applications, particularly Apple and Google, to force
them to share costs (Dedrick and others, 2011). On the other hand, content
providers in the United States play an important role in the content
market worldwide; they have the advantage of being the frst movers,
and they are have close links with the leading software and hardware
suppliers (Noam, 2008).
Economides (2011a) draws attention to the potential negative effect
on society of letting network operators assess a surcharge on content
providers. In his opinion, given the strong network externalities of the
Internet, the value of the network depends essentially on two elements:
the number of users (adoption) and the number of content creators. At the
same time, the value of the network to users depends on content quantity
and quality; the value of the content grows with the number of users. As
Internet providers do not internalize these network effects, they do not
consider the total effect of content providers on society when setting prices.
This would have a bearing on the value that users attribute to the network
and would erode the virtuous circle that has characterized Internet
development and innovation. The introduction of this type of pricing,
with its concomitant transaction costs, would be particularly damaging for
agents that have contributed to the explosion of content and innovation
over the past few years (small businesses, start-ups and individual content
providers). Economides (2011b) also argues that the combination of an ISP
with high market power and the user cost of switching from one operator
to another would make it easier for ISPs to engage in uncompetitive
practices in the absence of neutrality regulation. For example Chen and
Nalebuff (2006) point out that a broadband provider could potentially have
an incentive to degrade the quality of an OTT such as Skype, with a view
to making its own telephony service more attractive.
13
13
Nonetheless, their model finds that an Internet provider does not have incentives to
degrade the quality of a provider such as Skype because such services enhance the value
of its network.
Broadband in Latin America: Beyond Connectivity 327
Some broadband providers have claimed that regulating neutrality
would discourage investment in networks. Consequently, one of the
arguments for allowing network operators to charge content suppliers
for use of the network has been to provide incentives for the former to
invest (Krämer and Wiewiorra, 2009). But Economides (2011a) refutes this
argument, holding that allowing a charge for priority access would mean
that the higher the network congestion the higher the potential charge, so
there would be no incentive to invest to decongest the network. Similarly,
the model developed by Cheng, Bandyopadhyay and Guo (2010) predicts
that incentives for the broadband provider to expand capacity are greater
under neutrality. More capacity leads to less congestion and more highly-
valued Internet services, resulting in benefts for the operators. Lastly,
Choi and Kim (2010) provide a full and formal analysis of the relationship
between neutrality and investment incentives, concluding that it is
impossible to reach unambiguous general conclusions on that relationship.
D. The over-the-top market in Latin America
1. Incumbent operators, OTT strategies and supply
The challenges and changes in the industry are unfolding alongside
development of the content market. Service bundling, particularly bundles
including TV, can be seen as frst-generation strategies deployed by
operators to compete and to maintain market share in the new industry.
For example, IPTV is an alternative for competing with online content
offered by OTTs such as Netfix, which would give an incumbent ADLS
operator an income-producing system equivalent to cable television.
Incumbent operators are also looking at second-generation strategies
involving initiatives with OTT features or innovations in activities with
which those operators are familiar. This section analyses the frst- and
second-generation strategies of operators in Latin America.
In the frst-generation strategy, where proprietary services are
bundled with TV, what is available in Latin America displays a wide
variety of combinations of technologies and agreements. These include
cable operators (pioneers in double-and triple-play bundles) and operators
which, when developing a strategy for entering the content market, have
used a different technology than that used for communication services.
And some ADSL providers have started to offer IPTV, while others are
entering into agreements with satellite TV suppliers (DTH).
Table XI.2 lists operators offering bundles with pay-TV, by country
and type of technology. In 18 of the 24 countries considered there is a least
one bundle that includes pay-TV.
ECLAC 328
Table XI.2
SERVICE BUNDLES INCLUDING TV AND FIRST-GENERATION STRATEGIES
Country
Triple-Play +
cable TV
Double Play
(Internet –TV) +
cable
Double
Play
(Internet
+ cable
TV
with
ADSL+
IPTV)
Triple-
Play with
ADSL+
IPTV
Double play
and triple-
play with
technology
mix
Double play
and triple
play with
agreements
between
companies
Argentina Telecentro Telefónica
+ DIRECTV
Telecom +
DIRECTV
Fibertel+
Cablevisión
Barbados
Belice
Bolivia
(Plurinational
State of)
Corporación
COMTECO

Brazil NET NET Telefónica GVT
(DTH+IPTV)
CTBC (DTH
and cable) Oi
(DTH)a

Chile VTR VTR Telsur Telefónica
Telsur
Telefónica
(DTH)
Claro (DTH and
cable) TV

Colombia Claro (cable
TV) Telefonica
(DTH)
ETB (ADSL)
+DIRECTV
Costa Rica RACSA
(cable)
+Cablevisión
Cuba
Dominican
Republic
TRICOM Aster Claro Claro (DTH)
WIND Telecom
(MMDS)

Ecuador Portal Grupo
Cable TV
Claro (TV
cable)

El Salvador Claro (cable TV
and DTH)
Tigo (cable TV)

Guatemala TVO Telecom Claro (cable TV
and DTH)

Honduras Claro (cable TV
and DTH)

Jamaica Flow Jamaica
Mexico Cablevisión
Cablemas
Cablecom
Megacable
Cablevisión
Cablemas
Cablecom
Megacable
Maxcom Maxcom Telmex+Dish
Broadband in Latin America: Beyond Connectivity 329
Country
Triple-Play +
cable TV
Double Play
(Internet –TV) +
cable
Double
Play
(Internet
+ cable
TV
with
ADSL+
IPTV)
Triple-
Play with
ADSL+
IPTV
Double play
and triple-
play with
technology
mix
Double play
and triple
play with
agreements
between
companies
Nicaragua Claro (cable TV
and DTH)

Panama Cable Onda
Paraguay
Peru StarGlobalCom Telefónica
(DTH)

Suriname
Trinidad and
Tobago
Flow Flow
Uruguay
Venezuela
(Bolivarian
Republic of)
Intercable Intercable Telefónica
(DTH)

Source: Prepared by the authors on the basis of data from CETyS-UdeSA, “Relevamiento planes de banda
ancha fja segmento residencial de principales operadores de cada país en América Latina”, 2012; and
surveys undertaken by the authors.
* Oi announced its plans to launch IPTV services over FTTH in high density areas by the end of 2012.
The table shows that IPTV is very incipient in the region, unlike in
more developed countries where this type of service has been available
for several years. For example, Telefónica was already offering IPTV
(Imagenio) throughout Spain in 2005, although the frst pilot tests were
conducted in 2000. France Telecom has offered IPTV in France since 2003,
and Deustche Telecom in Germany since 2004. In Latin America, IPTV was
frst launched by Maxcom in Mexico in 2007. Telefónica offers IPTV under
the Imagenio brand in a number of Chilean and Brazilian cities. IPTV is
also offered in some cities in Chile by Telefónica del Sur (Telsur) and in
some cities in the Dominican Republic by Claro (América Móvil-Telmex).
In brief, IPTV is available in just 4 of the 24 countries surveyed.
In Argentina and Mexico, two of the largest markets in the region,
the incumbent frms offer triple-play packages under agreements with
satellite TV companies. This could be explained by the fact that the
regulatory frameworks (particularly concession contracts) do not allow
incumbent network operators to offer converged IPTV services.
14
14
There is a debate in Mexico over the need to change the incumbent Telmex’s concession
contract for it to be able to offer IPTV. In Argentina, local cooperatives are allowed to
provide IPTV, and there are several cases of this happening, albeit on a local basis.
Table XI.2 (concluded)
ECLAC 330
In addition to the cable companies in each country, two incumbent
company brands have a strong presence in the supply of TV in the region:
Telefónica’s Movistar TV and Claro TV from América Móvil-Telmex.
Movistar TV provides satellite TV in Chile, Colombia, Peru and the
Bolivarian Republic of Venezuela (in the latter country Telefónica offers
all services, but none in a bundle). Claro TV provides cable or satellite
television in eight countries, particularly in Central America. The two
brands overlap only in Chile and Colombia.
Table XI.3 provides a non-exhaustive list of the second-generation
strategies developed by incumbent operators in Latin America and
launched in 2010-2012.
Table XI.3
SECOND-GENERATION STRATEGIES
Operator
Name of the
initiative
Strategy
Countries of the region
where it is active
Telefónica TU Me Free calls and instant
messaging between users.
Worldwide (iPhone and
Android devices)
América Móvil-
Telmex
Claro Messenger Instant messaging service for
Claro clients.
Argentina, Peru,
Panama, Dominican
Republic, Guatemala
and Honduras
Telefónica Shopping de
Movistar
Movistar clients can buy
music, games, devices, etc.
on the company’s commercial
website.
Argentina
Ecuador
El Salvador
Telefónica Movistar video Entertainment web portal
offering access to exclusive
contents. Free for Movistar
clients.
Peru
Telefónica On Video For Telefónica clients. Argentina
Terra
(Telefónica
group)
Agreements with
Samsung, LG
and Philips
Application of live channels for
smart TVs.
América Móvil-
Telmex
Claro Ideas
Entertainment
Online contents for clients. Argentina, Peru,
Uruguay and to be
launched in Chile
Telmex Claro Video Online videos for clients. Colombia
América Móvil-
Telmex
Portal UnoTV Online and streaming content. Mexico
Totalplay Total Movie Online movies. Mexico
Oi Video distribution of and
implementation of VOD.
To be launched in Brazil
Sky Brasil
(Direc TV
group)
Sky Online Online video club with movies,
television series and music that
Sky subscribers can access.
A monthly subscription is
available through Club Sky
Online www.skyonline.com.br.
Brazil
Broadband in Latin America: Beyond Connectivity 331
Operator
Name of the
initiative
Strategy
Countries of the region
where it is active
Cablevisión,
Megacable,
VTR,
CableOnda,
MovistarTV,
Wind Telecom
The operator
includes
Moviecity Play in
its range.
VOD service that makes it
possible to watch over the
Internet almost 1,200 titles of
this premium channel. The
subscriber can access the
service from anywhere in Latin
America by entering www.
moviecityplay.com with a
username and password.
Cablevisión in
Argentina, Megacable
in Mexico, VTR in
Chile, CableOnda in
Panamá, MovistarTV in
Peru, Wind Telecom in
Dominican Republic
Source: Prepared by the authors on the basis of the websites of the respective operators.
Two types of initiatives stand out in the table: instant messaging tied
to the operator’s brand, and online content supply. On the frst of these,
Nikou and others (2012) show that European mobile phone operators base
their defensive strategy on providing more secure and reliable services than
those offered by Skype and WhatsApp, for example. This is consistent with
the fndings of a survey for Latin America presented in Telesemana.com
(2012), which show that one of the advantages that telecom operators have
over OTT providers is linked to the guarantees they can offer in terms of
billing and service security levels— features that are particularly important
for the high-end segments. At the same time, it is striking that applications
such as TU Me and Claro Messenger are successful among users who
demand less in terms of level and security. The instant-messaging market has
strong network externalities, so free applications that do not discriminate by
type of network or device and do not require any type of membership (such
as WhatsApp) would be expected to be market leaders. In terms of online
content supply, video sales (VOD) of movies and television series, games,
and devices lead the feld. In all cases, only the operator’s clients have access
to web-based portals with content, sometimes on an exclusive basis.
Apart from those listed in the table, other Telefónica initiatives
include BlueVía, an open platform where external developers can interact
to create their own applications; and participation in Wayra, an ICT global
digital business incubator in place in Europe and Latin America sponsored
by frms such as Microsoft and Nokia along with several Spanish partners.
Some device manufacturers are also venturing into new markets.
Examples are LG with its NetCast platform, and Sony, which has launched
a free online content platform, Crackle, in 17 Latin American countries. A
key feature of this platform is that the content is free for users because the
business model is based on advertising.
15
15
New players less closely tied to the sector are also venturing into the OTT world. For
example, Walmart launched the Vudu service in Mexico.
Table XI.3 (concluded)
ECLAC 332
Noam (2008) identifes three generations of TV technology:
(i) broadcast TV; (ii) cable TV, satellite TV and home video; and (iii) Internet
TV and mobile TV. The information in the foregoing tables shows that
second-generation TV technology has enabled frst-generation strategies
deployed by incumbent telecom operators to enter the content market.
Third-generation TV fosters and is represented by OTT content, one of the
main threats facing incumbent operators. Table XI.3 shows that incumbent
operators (both ADSL technology providers and in those using cable and
satellite technology) are reacting to this by offering online content on their
own channels.
2. Netflix, Skype and WhatsApp
The following paragraphs summarize the as-yet scant information that
exists for the region on the three leading OTTs, each in its own segment.
Skype has over 250 million users worldwide (and is growing by
54 million per month), although less than 10% of them pay to use the
service.
16
Roughly 16% of these users are in the United States, 13% in the
rest of America (so the percentage for Latin America is less than that
fgure) and 57% in Europe. In the United States, over 9% of Internet users
also use Skype, compared with 7% in America and 16% in Europe.
17
In
September 2011, Skype was acquired by Microsoft for US$8.5 billion.
WhatsApp was created in August 2009. Although it does not
publish data on its number of users, it is known to have handled about 2
billion messages per day in February 2012.
18
It remains among the most
downloaded applications in 40 European Union countries, Asia, the Near
East and Latin America. In Germany, it led the downloads ranking in 2011;
in Spain it has over 6 million users.
19
Netfix did not venture into Latin America until 2011, so it is
a new product for Latin American users although it has been in the
United States market for several years. The company has more than
27 million customers in the United States, Canada, Latin America, the
United Kingdom and Ireland, of whom just over 1 million are in Latin
America (ir.netfix.com). Analysts view Netfix earnings in Latin America
as disappointing, because 10 months after its launch it had only reached
0.75% of households. This stands in clear contrast to the 6% of households
16
Interview with Alejandro Arnaiz, the company’s market development manager for Latin
America [online] http://mundocontact.com/skype-va-por-mil-millones-de-suscriptores/.
17
See [online] http://skypenumerology.blogspot.com.ar/.
18
See [online] http://www.cnnexpansion.com/tecnologia/2012/04/04/whatsapp-tumba-
a-los-mensajes-de-texto.
19
See [online] http://tecnologia.elpais.com/tecnologia/2012/07/03/actualidad/134134
0111_145629.html.
Broadband in Latin America: Beyond Connectivity 333
achieved in Canada nine months after its launch, and 4% in United
Kingdom/Ireland after six months.
20
E. The strategy and policy debate
The invasion of the sector by the online content industry and unmanaged
VOIP communication services (OTT) will be boosted further when the
population has better-quality access in terms of download speed, thereby
allowing for a better OTT experience.
This chapter has shown that the strategies that incumbent
operators have followed and continue to develop in order to compete and
maintain market share boil down to two: service bundling (fxed and
mobile telephony, Internet and TV) and the development of proprietary
applications that compete directly with OTTs.
The bundling strategy tends to be effective for shifting market power
from a given product to others. But it is questionable as a competition
policy, and some national regulators might decide to control it. In addition,
the only segments of demand that can afford to purchase bundles tend to
be those with high purchasing power, which are very sensitive to service
quality, so all of the bundled products need to be very quality-competitive.
The improving audiovisual-content quality of OTTs such as Netfix is a
threat to bundling that includes pay-TV.
The second strategy seems highly risky, because incumbent
operators in principle have no competitive advantage in creating OTTs.
The OTT market is highly competitive in terms of innovation, it is global,
and it has “winner-takes-all” characteristics. These conditions mean
that the leading OTT suppliers are few and highly specialized, and they
beneft from signifcant economies of scale (fxed costs of innovation
and acquisition of content, etc.). In contrast, incumbent operators do not
specialize in a specifc service or content, and they face constraints in
leveraging economies of scale because they are not as well positioned to
enter markets dominated by the leading providers.
Table IX.2 shows that the main incumbent operators in Latin
America —Telefónica (Movistar) and América Móvil-Telmex (Claro)—
have been and remain highly active in developing bundling strategies.
Both have positioned themselves in the pay-TV market, using various
technologies, developing IPTV (albeit still on an incipient basis), or
entering into agreements with satellite TV companies. Telefónica has been
launching new applications or second-generation strategies, which is not
20
See [online] http://www.pyramidresearch.com/points/item/120730.htm.
ECLAC 334
surprising given the greater challenges it faces in its developed-country
markets, where connection speeds are faster than in Latin America.
Fifty-one percent of the Latin American operators surveyed in
Telesemana.com (2012) reported that deploying a strategy to face OTTs is
not a priority for them. This could be due (at least in part) to the region’s
immature smartphone market and the fact that network coverage remains
a constraint despite growing penetration in countries such as Argentina
and Chile. This is consistent with the fndings of Sabbag and others
(2012), showing that digitalization maturity levels in the region’s countries
are well below those in countries where OTTs are booming. Similarly,
the slow average connection speed suggests that the region is behind
in terms of OTT development. On this point, bandwidth and service
quality constraints suggest that it is not ready to provide mass access to
OTT videos. At the same time, the low penetration rates in Latin America
achieved by OTTs such as Netfix could be explained at least partly by the
different regulations and piracy controls between the region and countries
where the penetration of these OTTs is considerably higher. On this point,
users in Latin America tend to think that once the Internet connection has
been paid for everything that circulates on the network is free and there is
no reason to pay for content that can be found on the network at zero cost
and with little effort.
21
Some authors argue that, in certain circumstances, later adoption
can bring advantages linked to lower equipment costs or learning from
the experience of other countries, to avoid becoming locked-in to obsolete
standards (Dedrick and others, 2011). For example, Latin America’s lagging
behind might ultimately mean that IPTV does not develop in the region to
the extent seen in more advanced countries. In fact, Huertas Sánchez and
others (2011) note that some European operators entering the market have
decided to abandon IPTV because they cannot compete with OTT content.
Nonetheless, underlying this entire discussion is a long-standing
core issue that the region’s operators will need to address sooner or later.
Over-the-top services offer users the possibility of obtaining, through a
simple high-speed Internet connection, the same services (in some cases
of better quality) as those obtained from incumbent operators, and at a
very low price or even for free. At the same time, the high-speed Internet
market is very competitive. Firstly, it hardly allows any differentiation
between companies. In other words, it is not clear where incumbent
operators’ profts will come from unless they fnd a way to differentiate
their service. For example, with investment in broadband, European IPTV
21
See, for example, the post “10 razones para no contratar a Netflix” [10 reasons not to sign
up for Netflix] on Taringa, the largest social network originated in Latin America. It is one
of the 10 most visited sites in Argentina and has nearly 20 million members in the region.
Broadband in Latin America: Beyond Connectivity 335
operators entered the pay-television market offering exclusive premium
content to attract subscribers (Weeds, 2012). In that case, exclusive content
was the differentiator. With high-speed access networks allowing non-
exclusive and widespread OTT content, telecom operators will be forced
to innovate their business strategies (Ganuza and Viecens, 2012). Secondly,
Internet-provider market entry barriers might not be high, depending on
the regulations, which increases the threat of entry by new market players.
For example, the fact that network deployment entails, essentially, a major
civil work project makes infrastructure construction companies natural
candidates to become new market participants in the medium or long
term. All of these factors raise questions about the operators’ traditional
business model and their return on investments in new networks.
Consequently, from the social standpoint, OTTs can reduce the incentives
for incumbent telecom operators to invest in deploying networks that
would provide access at the speeds needed for a good OTT experience.
The above is consistent with the active government interventions
currently occurring internationally. Nonetheless, as noted in the
introduction, interventions in the region have prioritized expanding
geographic coverage and increasing the number of accesses while giving
less importance to quality and speed (see the chapter by Galperín, Mariscal
and Viecens herein). Consequently, over the short and medium term, the
deployment of high-speed access networks in Latin America (and the
OTT services that need them) will probably be confned to countries and
regions that have demand segments with high purchasing power. But even
in these segments, incumbent operators may not have suffcient incentives
for deployment. Nonetheless, and for redistributive reasons, government
intervention to subsidize the selective deployment of networks to the
home would be debatable (national plans in Latin America only involve
laying backbone fbre networks). In contrast, investments for the provision
of good quality public services (schools, hospitals, public administration)
could have a very high social return in the medium and long terms. For
example, Ganuza and Viecens (2011) analyse the infrastructure deployment
project planned by the Generalitat of Catalonia (Spain). It is a government
investment to bring fbre to all of the autonomous community’s public
administration sites in order to ensure that citizens have access to high-
speed services with a high social return, such as education, health and
security. In contrast, the use of public funds to bring high speeds to homes
would be hard to justify from a social point of view, since household
use of high-quality connections is almost always only for entertainment
purposes (videos and high-defnition movies).
Most OTT services only require speeds of up to 2Mbps, so
guaranteeing access to them is a more feasible goal for the countries of the
region to pursue, at least in the medium term. This could provide a chance
ECLAC 336
to reduce telecom service costs, guaranteeing access to communication
services and OTT applications. At the same time, from the regulatory
standpoint, providing services through OTTs makes it simpler to compare
price vectors, because the important thing is the basic data transmission
cost. This simplifcation could help countries with weak regulatory
authorities to design effective policies.
A feasible medium-term objective with a high social return for the
countries of the region would therefore consist of ensuring mass access
to OTT communication services and applications (ideally with 2 Mbps
speed), while letting the private sector provide access at higher speeds to
ensure enjoyment of OTT content (less clear social return). Lastly, certain
public services, such as education and health care, could justify the use
of public funds for the deployment of high-speed access networks, to
guarantee better quality in these services.
F. Conclusions
The traditional revenue model based on subscriptions and metered
services is showing signs of coming to its end. At the same time,
growing communications services such as Skype and WhatsApp do not
generate revenue for the access providers, but they do take up network
space and compete with proprietary services. In addition, frms that
are emerging with the new industry (such as Google and Yahoo) are
exploiting new potential revenue sources based on the two-sided nature
of their markets, in many cases sourcing their revenue mainly from
advertising.
22
Incumbent operators do not have experience in using this
model and fnd it hard to adapt (Huigen and others, 2008). At the same
time, their market power is being eroded by competition and is shifting
towards content providers. If the current trend persists, the time could
come when the user just wants a household Internet connection with
speeds that enable access to quality services. How will telecom operators
position themselves in this new industry? At one extreme, some authors
argue that infrastructure operators should just focus on their role as “bit-
pipes” instead of continuing to operate as integrated frms providing both
services and infrastructure (Grove and Baumann, 2012). The debate is
ongoing, particularly in countries where infrastructure makes it possible
to offer consumers coverage and relatively high access speeds. One of
the objectives of this chapter has been to examine how the scenario is
22
Two-sided markets or platforms exist when companies create value by facilitating
interaction between two different consumer groups. The success of the platform hinges
on the pricing structure for each side of the market, based on the network externalities
generated by each one. For example, for the Google platform the two sides of the market
are users and advertisers.
Broadband in Latin America: Beyond Connectivity 337
unfolding in Latin America and to discuss how the region’s challenges and
potential differ from those seen in more developed regions and countries.
Part of this discussion may seem somewhat irrelevant for the
region insofar as it lacks the infrastructure and technology needed to
ensure quality OTT services and content. Current broadband and service
conditions are far from being able to provide access to quality video
streaming, for example. Consequently, there is some way to go before
Latin American operators feel genuinely threatened by OTTs such as Web
TV/OTT TV.
The case of unmanaged VoIP services (such as Skype and Gmail
video chat) could be different because they can be used at slower access
speeds, although a minimum speed is still needed for reasonable service.
WhatsApp does pose a clear challenge to text messages and even voice
messaging as its smartphone penetration rate increases. Nonetheless,
at the present time there is no widespread concern among the region’s
operators in relation to these OTT.
23
What economic policy implications or recommendations can be
drawn in this context? The previous section suggested that the medium-
term objective should be to ensure mass access to OTT communication
services and applications (ideally with speeds of 2Mbps) and let the private
sector supply access at higher speeds for enjoying OTT content (which has
a lower social return). On the other hand, it should be borne in mind that
high Internet access speeds, along with their potential for developing new
business models such as OTTs and the implications for telecom operators,
are a technological revolution. Every technological revolution throughout
history has produced winners and losers, and what ultimately needs
to be taken into account is the effect on society’s aggregate welfare.
24

Consequently, governments should facilitate this process and refrain
from introducing measures that could obstruct it. Decisions such as those
relating to network neutrality should be taken with a long-term view and
not in response to pressure from vested interests or short-term objectives.
Internet, and the new services and business models that accompany it,
have greatly enhanced consumer welfare. Product and service variety,
competition leading to lower prices, the shortening of distances and
consequent reduction in transport costs for Internet users, have highly
positive repercussions on their welfare. A good economic and regulatory
policy clearly should ensure and empower these possibilities in society.
23
For example, in response to the Telesemana.com (2012) question “Do you believe over-
the-top (OTT) services such as Google, Twitter and WhatsApp pose a threat to the mobile
broadband business?”, 42% of respondents said no; 8% said they were not sure; and 50%
said yes.
24
The positive effect of broadband on growth is well documented. See Czernich and others
(2011) and Koutroumpis (2009), and also the chapter by Raúl Katz herein.
ECLAC 338
Bibliography
Athey, S., E. Calvano and J. S. Gans (2012), “The Impact of the Internet on Advertising
Markets for News Media”, unpublished.
CETyS-UdeSA (2012), “Relevamiento planes de banda ancha fja segmento
residencial de principales operadores de cada país en América Latina”.
Chen, M. K. and B. Nalebuff (2006), “One-Way Essential Complements”, Working
Papers on Economic Applications and Policy, No. 22, Yale School of Management,
Research Paper Series.
Cheng, H. K., S. Bandyopadhyay and H. Guo (2010), “The Debate on Net Neutrality:
A Policy Perspective”, Information Systems Research, No. 22.
Choi, J. P. and B.-C. Kim (2010), “Net neutrality and investment incentives”, The
RAND Journal of Economics, No. 41. 
Czernich, N., O. Falck, T. Kretschmer and L. Wößmann (2011), “Broadband
Infrastructure and Economic Growth”, The Economic Journal, No. 121.
De Bijl, P.W.J. and M. Peitz (2010), “Regulatory legacy, VoIP adoption, and
investment incentives”, Telecommunications Policy, No. 34.
De León, O. (2012), “Desarrollo de la conectividad nacional y regional en América
Latina”, Project Document, No. 502 (LC/W.502), Santiago, Chile, Economic
Commission for Latin America and the Caribbean (ECLAC).
Dedrick, J., K.L. Kraemer y G. Linden (2011), “The distribution of value in the
mobile phone supply chain”, Telecommunications Policy, No. 35.
Economides, N. (2011a), “Why Imposing New Tolls on Third-Party Content
and Applications Threatens Innovation and Will Not Improve Broadband
Providers’ Investment”, Net Neutrality: Contributions to the Debate J. P. Martínez
(ed.), Fundación Telefónica.
___ (2011b), “Broadband Openness Rules Are Fully Justifed by Economic Research”,
Communications and Strategies, No. 84.
Feijoo, C., J. L. Gómez-Barroso, J. M. Aguado and S. Ramos (2012), “Mobile gaming:
Industry challenges and policy implications”, Telecommunications Policy, vol. 36,
No. 3.
Ganuza, J.J. and M.F. Viecens (2012), “Exclusive content and the Next Generations
Networks”, Information Economics and Policy, forthcoming.
___ (2011), “Deployment of high-speed broadband infrastructures during the
economic crisis. The case of Xarxa Oberta”, Telecommunications Policy, vol. 35,
No. 9-10.
Grove, N. and O. Baumann (2012), “Complexity in the telecommunications industry:
When integrating infrastructure and services backfres”, Telecommunications
Policy, No. 36.
Hoernig, S., S. Jay, K.H. Neumann, M. Peitz, T. Plueckebaum e I. Vogelsang (2012),
“The impact of different fbre access network technologies on cost, competition
and welfare”, Telecommunications Policy, No. 36.
Huertas Sánchez, J.C. Domínguez, J. Lacasa and V. Sanz Fernández (2011),
“Modelos over the top (OTT): regulación y competencia en los nuevos mercados
de Internet”, Revista GEER, No. 6, Telefónica.
Huigen, J. and M. Cave (2008), “Regulation and the promotion of investment in
next generation networks—A European dilemma”, Telecommunications Policy,
No. 32 .
Broadband in Latin America: Beyond Connectivity 339
Koutroumpis, P. (2009), “The economic impact of broadband on growth: A
simultaneous approach”, Telecommunications Policy, No. 33.
Krämer, J. and L. Wiewiorra (2009), “Innovation through Discrimination!? A
Formal Analysis of the Net Neutrality Debate”[online] http://ssrn.com/
abstract=1444423.
Noam, E. (2010), “Regulation 3.0 for Telecom 3.0”, Telecommunications Policy, No. 34.
___ (2008), “If fber is the medium, what is the message? Next-Generation Content
for Next-Generation Networks”, Communications and Strategies, Special Issue,
November.
Nikou, S., H. Bouwmany and M. de Reuver (2012), “The potential of converged
mobile telecommunications services: A conjoint analysis”, info, vol. 14, No. 5.
Sabbag, K., R. Friedrich, B. El-Darwiche, M. Singh, S. Ganediwalla and R. Katz
(2012), “Maximizing the Impact of Digitization”, The Global Information
Technology Report, S. Dutta and B. Bilbao-Osorio, Geneva, World Economic
Forum and Insead.
Seamans, R. and F. Zhu (2010), “Technology Shocks in Multi-Sided Markets: The
Impact of Craigslist on Local Newspapers”, NET Institute Working Paper,
No. 10-11.
Shin, D-H. (2012), “What makes consumers use VoIP over mobile phones? Free
riding or consumerization of new service”, Telecommunications Policy, No. 36.
Telesemana.com (2012), “Encuesta OTT 2012, Relación estratégica que mantendrán
los operadores con los proveedores OTT”.
Weeds, H. (2012), “Superstars and the Long Tail: The Impact of Technology on Market
Structure in Media Industries”, Information Economics and Policy, vol. 24, No. 1.









Publicaciones de la CEPAL
ECLAC publications
Comisión Económica para América Latina y el Caribe
Economic Commission for Latin America and the Caribbean
Casilla 179-D, Santiago de Chile.

Véalas en: www.cepal.org/publicaciones
Publications may be accessed at: www.eclac.org

Contacto / Contact: [email protected]


Revista CEPAL / CEPAL Review
La Revista se inició en 1976 como parte del Programa de Publicaciones de la Comisión Económica para
América Latina y el Caribe, con el propósito de contribuir al examen de los problemas del desarrollo
socioeconómico de la región. Las opiniones expresadas en los artículos firmados, incluidas las
colaboraciones de los funcionarios de la Secretaría, son las de los autores y, por lo tanto, no reflejan
necesariamente los puntos de vista de la Organización.
La Revista CEPAL se publica en español e inglés tres veces por año.

CEPAL Review first appeared in 1976 as part of the Publications Programme of the Economic
Commission for Latin America and the Caribbean, its aim being to make a contribution to the study of the
economic and social development problems of the region. The views expressed in signed articles,
including those by Secretariat staff members, are those of the authors and therefore do not necessarily
reflect the point of view of the Organization.
CEPAL Review is published in Spanish and English versions three times a year.

Informes periódicos institucionales / Annual reports
Todos disponibles para años anteriores / Issues for previous years also available
• Balance Actualizado de América Latina y el Caribe 2012 - abril de 2013, 24 p.
• Updated Economic Overview of Latin America and the Caribbean 2012 - April 2012, 24 p.
• Balance Preliminar de las Economías de América Latina y el Caribe 2012, 84 p.
Preliminary Overview of the Economies of Latin America and the Caribbean 2012, 82 p.
• Estudio Económico de América Latina y el Caribe 2012, 162 p.
Economic Survey of Latin America and the Caribbean 2012, 154 p.
• Panorama de la Inserción Internacional de América Latina y el Caribe 2011-2012, 126 p.
Latin America and the Caribbean in the World Economy 2011-2012, 116 p.
• Panorama Social de América Latina, 2012, 252 p.
Social Panorama of Latin America, 2012, 238 p.
• La Inversión Extranjera Directa en América Latina y el Caribe 2011, Documento informativo,
152 p.
Foreign Direct Investment in Latin America and the Caribbean 2011, Briefing paper, 142 p.
• Anuario Estadístico de América Latina y el Caribe 2012 / Statistical Yearbook for Latin America
and the Caribbean 2012, 224 p.


Libros de la CEPAL
118 Sistemas de innovación en Centroamérica. Fortalecimiento a través de la integración regional,
Ramón Padilla Pérez (ed.), 2013, 222 p.
117 Envejecimiento, solidaridad y protección social en América Latina y el Caribe. La hora de avanzar
hacia la igualdad, Sandra Huenchuan, 2013. 190 p.
117 Ageing, solidarity and social protection in Latin America and the Caribbean Time for
progress towards equality, Sandra Huenchuan, 2013, 176 p.
116 Los fundamentos de la planificación del desarrollo en América Latina y el Caribe. Textos
seleccionados del ILPES (1962-1972), Ricardo Martner y Jorge Máttar (comps.), 2012, 196 p.
115 The changing nature of Asian-Latin American economic relations, German King, José
Carlos Mattos, Nanno Mulder and Osvaldo Rosales (eds.), 2012, 196 p.
114 China y América Latina y el Caribe. Hacia una relación económica y comercial estratégica,
Osvaldo Rosales y Mikio Kuwayama, 2012, 258 p.
114 China and Latin America and the Caribbean Building a strategic economic and trade
relationship, Osvaldo Rosales y Mikio Kuwayama, 2012, 244 p.
113 Competitividad, sostenibilidad e inclusión social en la agricultura: Nuevas direcciones en el diseño
de políticas en América Latina
y el Caribe, Octavio Sotomayor, Adrián Rodríguez y Mônica Rodrigues, 2012, 352 p.
112 El desarrollo inclusivo en América Latina y el Caribe. Ensayos sobre políticas de convergencia
productiva para la igualdad, Ricardo Infante (ed.), 2011, 384 p.
111 Protección social inclusiva en América Latina. Una mirada integral, un enfoque de derechos,
Simone Cecchini y Rodrigo Martínez, 2011, 284 p.
110 Envejecimiento en América Latina. Sistema de pensiones y protección social integral, Antonio
Prado y Ana Sojo (eds.), 2010, 304 p.
109 Modeling Public Policies in Latin America and the Caribbean, Carlos de Miguel, José Durán
Lima, Paolo Giordiano, Julio Guzmán, Andrés Schuschny and Masazaku Watanuki (eds.), 2011,
322 p.
108 Alianzas público-privadas. Para una nueva visión estratégica del desarrollo, Robert Devlin y
Graciela Moguillansky, 2010, 196 p.
107 Políticas de apoyo a las pymes en América Latina. Entre avances innovadores y desafíos
institucionales, Carlos Ferraro y Giovanni Stumpo, 2010, 392 p.
106 Temas controversiales en negociaciones comerciales Norte-Sur, Osvaldo Rosales V. y Sebastián
Sáez C. (comps.), 2011, 322 p.

Copublicaciones recientes / Recent co-publications
Decentralization and reform in Latin America. Improving Intergovernmental Relations, Giorgio Brosio
and Juan Pablo Jiménez (eds.), ECLAC / Edward Elgar Publishing, United Kingdom, 2012.
Sentido de pertenencia en sociedades fragmentadas. América Latina desde una perspectiva global, Martín
Hopenhayn y Ana Sojo (comps.), CEPAL / Siglo Veintiuno, Argentina, 2011.
Las clases medias en América Latina. Retrospectiva y nuevas tendencias, Rolando Franco, Martín
Hopenhayn y Arturo León (eds.), CEPAL / Siglo XXI, México, 2010.
Innovation and Economic Development. The Impact of Information and Communication
Technologies in Latin America, Mario Cimoli, André Hofman and Nanno Mulder,
ECLAC / Edward Elgar Publishing, United Kingdom, 2010.
Las clases medias en América Latina. Retrospectiva y nuevas tendencias, Rolando Franco, Martín
Hopenhayn y Arturo León (eds.), CEPAL / Siglo Veintiuno, México, 2010.
Sesenta años de la CEPAL. Textos seleccionados del decenio 1998-2008, Ricardo Bielschowsky
(comp.), CEPAL / Siglo Veintiuno, Argentina, 2010.
El nuevo escenario laboral latinoamericano. Regulación, protección y políticas activas en los
mercados de trabajo, Jürgen Weller (ed.), CEPAL / Siglo Veintiuno, Argentina, 2010.
Internacionalización y expansión de las empresas eléctricas españolas en América Latina, Patricio
Rozas Balbontín, CEPAL / Lom, Chile, 2009.


Coediciones recientes / Recent co-editions
Juventud y bono demográfico en Iberoamérica, Paulo Saad, Tim Miller, Ciro Martínez y Mauricio Holz,
CEPAL/OIJ/UNFPA, Chile, 2012.
Perspectivas económicas de América Latina 2013. Políticas de Pymes para el Cambio Estructural,
OCDE / CEPAL, Chile, 2012.
Latin American Economic Outlook 2013. SME Policies For Structural Change,
OECD / ECLAC, Chile, 2012.
Perspectivas de la agricultura y del desarrollo rural en las Américas: una mirada hacia América Latina y el
Caribe 2013, CEPAL / FAO / IICA, Chile, Octubre, 2012.
Reforma fiscal en América Latina. ¿Qué fiscalidad para qué desarrollo?, Alicia Bárcena y Narcís Serra
(editores), CEPAL/SEGIB / CIDOB, Chile, 2012.
La sostenibilidad del desarrollo a 20 años de la Cumbre para la Tierra. Avances, brechas y
lineamientos estratégicos para América Latina y el Caribe, CEPAL / Naciones Unidas, 2012.
Sustainable development 20 years on from the Earth Summit. Progress, gaps and strategic
guidelines for Latin America and the Caribbean, ECLAC / United Nations, 2012.
Perspectivas económicas de América Latina 2012.Transformación del Estado para el desarrollo,
CEPAL / OCDE, 2011.
Latin America Outlook 2012. Transforming the State for Development, ECLAC/OECD, 2011.
Perspectives économiques de l’Amérique latine 2012. Transformation de l’État et
Développement, CEPALC / OCDE, 2012.
Breeding Latin American Tigers. Operational principles for rehabilitating industrial policies,
Robert Devlin and Graciela Moguillansky, ECLAC / World Bank, 2011.
Espacios iberoamericanos: Hacia una nueva arquitectura del Estado para el desarrollo,
CEPAL / SEGIB, 2011.
Espaços ibero-americanos: A uma nova arquitetura do Estado para o desenvolvimento. CEPAL
/ SEGIB, 2011.
Cuadernos de la CEPAL
100 Construyendo autonomía. Compromiso e indicadores de género, Karina Batthyáni Dighiero, 2012,
338 p.
99 Si no se cuenta, no cuenta, Diane Alméras y Coral Calderón Magaña (coordinadoras),
2012, 394 p.
98 Macroeconomic cooperation for uncertain times: The REDIMA experience, Rodrigo
Cárcamo-Díaz, 2012,164 p.
97 El financiamiento de la infraestructura: Propuestas para el desarrollo sostenible de una política
sectorial, Patricio Rozas Balbontín, José Luis Bonifaz y Gustavo Guerra-García, 2012, 414 p.
96 Una mirada a la crisis desde los márgenes, Sonia Montaño (coordinadora), 2011, 102 p.
95 Programas de transferencias condicionadas. Balance de la experiencia reciente en América
Latina y el Caribe, Simone Cecchini y Aldo Madariaga, 2011, 226 p.
95 Conditional cash transfer programmes. The recent experience in Latin America and the
Caribbean, Simone Cecchini and Aldo Madariaga, 2011, 220 p.
94 El cuidado en acción. Entre el derecho y el trabajo, Sonia Montaño Virreira y Coral Calderón
Magaña (coords.), 2010, 236 p.
93 Privilegiadas y discriminadas. Las trabajadoras del sector financiero, Flavia Marco Navarro y
María Nieves Rico Ibáñez (eds.), 2009, 300 p.

Cuadernos estadísticos de la CEPAL
40 América Latina y el Caribe: Índices de precios al consumidor. Serie enero de 1995 a junio de
2012. Solo disponible en CD, 2012.
39 América Latina y el Caribe: indicadores macroeconómicos del turismo. Solo disponible en CD, 2010.
38 Indicadores ambientales de América Latina y el Caribe, 2009. Solo disponible en CD, 2010.
37 América Latina y el Caribe: Series históricas de estadísticas económicas 1950-2008. Solo
disponible en CD, 2009.

Observatorio demográfico / Demographic Observatory
Edición bilingüe (español e inglés) que proporciona información estadística actualizada, referente a
estimaciones y proyecciones de población de los países de América Latina y el Caribe. Incluye
también indicadores demográficos de interés, tales como tasas de natalidad, mortalidad, esperanza
de vida al nacer, distribución de la población, etc.
Desde 2013 el Observatorio aparece una vez al año.
Bilingual publication (Spanish and English) proving up-to-date estimates and projections of the
populations of the Latin American and Caribbean countries. Also includes various demographic
indicators of interest such as fertility and mortality rates, life expectancy, measures of population
distribution, etc.
Since 2013, the Observatory appears once a year.



Notas de población
Revista especializada que publica artículos e informes acerca de las investigaciones más recientes
sobre la dinámica demográfica en la región, en español, con resúmenes en español e inglés. También
incluye información sobre actividades científicas y profesionales en el campo de población.
La revista se publica desde 1973 y aparece dos veces al año, en junio y diciembre.

Specialized journal which publishes articles and reports on recent studies of demographic dynamics in
the region, in Spanish with abstracts in Spanish and English. Also includes information on scientific
and professional activities in the field of population.
Published since 1973, the journal appears twice a year in June and December.

Series de la CEPAL
Comercio Internacional / Desarrollo Productivo / Desarrollo Territorial / Estudios Estadísticos y
Prospectivos / Estudios y Perspectivas (Bogotá, Brasilia, Buenos Aires, México, Montevideo) /
Studies and Perspectives (The Caribbean, Washington) / Financiamiento del Desarrollo / Gestión
Pública / Informes y Estudios Especiales / Macroeconomía del Desarrollo / Manuales / Medio
Ambiente y Desarrollo / Asuntos de Género (ex Mujer y Desarrollo) / Población y Desarrollo / Políticas
Fiscales / Políticas Sociales / Recursos Naturales e Infraestructura / Reformas Económicas /
Seminarios y Conferencias.
Véase el listado completo en: www.cepal.org/publicaciones /
A complete listing is available at: www.eclac.org/publications


Las publicaciones de las Naciones Unidas y de la
Comisión Económica para América Latina y el Caribe (CEPAL)
se pueden adquirir a través de:
Publicaciones de las Naciones Unidas
National Book Network
15200 NBN Way
Blue Ridge Summit, PA 17214
Estados Unidos

Tel. (1 888)254-4286
Fax (1-800)338-4550
Contacto: [email protected]
Pedidos: [email protected]


Publications of the United Nations and the Economic
Commission for Latin America and the Caribbean (ECLAC)
can be ordered through:
United Nations Publications
National Book Network
15200 NBN Way
Blue Ridge Summit, PA 17214
USA

Tel. (1 888)254-4286
Fax (1-800)338-4550
Contact: [email protected]
Orders: [email protected]


www.un.org/publications

Economic Commission for Latin America and the Caribbean (ECLAC)
Comisión Económica para América Latina y el Caribe (CEPAL)
www.eclac.org
Broadband in Latin America
Beyond Connectivity
Valeria Jordán
Hernán Galperin
Wilson Peres
Editors
First edition
Printed at United Nations, Santiago, Chile • E1300644 • December 2013
ISBN 978-92-1-121836-7 • E-ISBN: 978-92-1-056012-2 • Sales No. E.14.II.G.7
Copyright © United Nations 2013 • 2013-644
Based on a model of the broadband ecosystem developed by the
authors in their 2010 book, Fast-tracking the digital revolution:
Broadband for Latin America and the Caribbean, they now analyse
the evolution of this technology in Latin America, assess the economic
impacts of its diffusion, propose public policies for its expansion and
debate the future of the ecosystem.
The policy analyses and proposals presented in the book focus on national
programmes to foster universal broadband access and the debate on
Internet neutrality. The study of the current trends highlights the progress
of cloud computing and the new developments induced by the entrance
of over-the-top operators in the region. This book underscores the need
to expand regional and national Internet traffic exchange points (IXPs)
and the relevance of the increasing demand gap, which poses new
challenges beyond those related to access and connectivity.
This book is the result of the collaboration between the United
Nations Economic Commission for Latin America and the Caribbean
(ECLAC), within the framework of the Inclusive Political Dialogue and
Exchange of Experiences Project of the @LIS2 Programme (Alliance
for the Information Society phase 2) of the European Commission,
and the Regional Dialogue on the Information Society (DIRSI).