It consists of actions aimed at helping SMES to gain benefits from the digital economy. Innovation Union:
The Digital Agenda, one of the seven flagship initiatives of the Europe 2020 Strategy, has the goal of creating a flourishing digital economy by 2020.
based on FÁS Regional Labour market Bulletin 2012 & CSO Figures 22 The available data indicates that
Subdivision of data into North & South Tipperary areas not available. 28 Case study: Benefi ts of participation in EU Projects Tyndall National Institute was established in 2004 under a formal agreement between University college Cork and the Minister for Enterprise, Trade and Innovation.
There are currently 15 industry-led research centres working in fields such as IT Innovation, Biorefining & Bioenergy, Data Analytics and Manufacturing Research.
c) Data pertaining to employment, turnover and exports in the Region from the Central Statistics Offi ce and;
A national clustering policy is essential to provide support and structure to cluster development. Additionally, there is a requirement for policies to ensure the creation of the type of environment that companies need
and public awareness in the areas of cluster structure, cluster development and the regional benefits of clustering;
Baseline Data: The initial sections of this Strategy form a baseline assessment of the current research and innovation activities and strengths in the Region.
Priority Area A-Future Networks & Communications Priority Area B-Data Analytics, Management, Security & Privacy Priority Area C-Digital Platforms, Content & Applications
In this fast-changing digital era, one of the key challenges in measuring the information society is the lack of up-to-date data, in particular in developing countries.
ITU is joining the international statistical community in looking into ways of using new and emerging data sources such as those associated with big data to better provide timely and relevant evidence for policy-making.
Calls for a data revolution are prominent in the international debates around the post-2015 development agenda
store and analyse huge amounts of data, as well as being a major source of big data in their own right.
Big data from mobile operators, for example, are real-time and low-cost and have one of the greatest development potentials in view of the widespread use and availability of mobile networks and services.
This report provides the reader with a comprehensive and critical overview of the role of big data from the telecommunication sector,
for use in social and economic development policy and for monitoring the future information society.
I trust that the data and analysis contained in this report will be of great value to the ITU membership,
Acknowledgements The 2014 edition of the Measuring the Information Society Report was prepared by the ICT Data and Statistics Division within the Telecommunication Development Bureau of ITU.
and Michael Minges to the compilation of data on international bandwidth, revenue and investment. Helpful inputs and suggestions were received from Joan Calzada Aymerich from the University of Barcelona (Chapter 4), Jake Kendall from the Gates Foundation, Anoush Tatevossian and Alex Rutherford from UN Global Pulse,
The report includes data from Eurostat, OECD, IMF, Informa, the UNESCO Institute for Statistics, the United nations Population Division and the World bank,
ITU also appreciates the cooperation of countries that have provided data included in this report. The report was edited by Anthony Pitt and Bruce Granger, ITU English Translation Section.
152 Chapter 5. The role of big data for ICT monitoring and for development...173 5. 1 Introduction...
173 5. 2 Big data sources, trends and analytics...175 5. 3 Telecommunication data and their potential for big data analytics...
181 5. 4 Big data from mobile telecommunications for development and for better monitoring...185 5. 5 Challenges and the way forward...
195 Chapter 5 Annex...207 List of references...213 Annex 1. CT Development Index (IDI) methodology...
221 Annex 2. ICT price data methodology...231 Annex 3. Statistical tables of indicators used to compute the IDI...
156 5. 1 The five Vs of big data...176 5. 2 An overview of telecom network data...
182 5. 3 Customer profiling using telecom big data...184 xii List of boxes 1. 1 Final review of the WSIS targets:
Achievements, challenges and the way forward...26 1. 2 A decade of successful international cooperation on ICT measurement...
28 1. 4 What is a data revolution?..30 2. 1 ITU discussion forums on ICT statistics...
158 5. 1 How big data saves energy Vestas Wind Systems improves turbine performance...177 5. 2 How Twitter helps understand key post-2015 development concerns...
179 5. 3 How mobile operators currently use data to track service uptake, business performance and revenues...
183 5. 4 Using mobile data for development...187 5. 5 How mobile network data can track population displacements an example from the 2010 Haiti earthquake...
188 5. 6 Leveraging mobile network data for transportation and urban planning in Sri lanka...189 5. 7 Poverty mapping in Côte d'ivoire using mobile network data...
190 5. 8 Using mobile-phone data to track the creditworthiness of the unbanked...191 5. 9 Using mobile big data
and mobile networks for implementing surveys...193 List of tables 1. 1 Rural population covered by a mobile-cellular signal, 2012.4 1. 2 Total Internet domain registrations by world region, 2003,2008
and 2013.18 2. 1 IDI values and changes, 2012 and 2013.41 2. 2 ICT Development Index (IDI),
2012 and 2013.42 xiii xiv 2. 3 IDI access sub-index, 2012 and 2013.43 2. 4 IDI use sub-index, 2012 and 2013.44 2. 5 IDI skills sub-index, 2012 and 2013.45 2. 6 Most dynamic
163 4. 15 ICT Price Basket and sub-baskets, 2013.166 5. 1 Sources of big data...
The data also show a continuous increase in Internet usage, with growth in the number of Internet users in all countries and increasing availability of online content,
whereby urban citizens enjoy ubiquitous mobile network coverage, affordable high-speed Internet services and the higher levels of skills required to make effective use of online content and services,
covering the fixed and mobile (voice and data) market segments, and considering both subscriptions and household access data.
This will be followed by a presentation of the latest trends in terms of investment and revenue in the telecom sector.
in particular in the context of the post-2015 development debate and the WSIS+10 review, the demand for a data revolution,
and the role of big data for ICT monitoring. 1. 2 The voice market In line with developments in recent years,
or SIM CARD the total numbers and growth rates strongly point to market saturation. Whether this will change in the near future,
ITU World Telecommunication/ICT Indicators database. 39.2 26.3 24.9 15.8 12.7 8. 7 1. 3 05 10 15 20 25
In addition, mobile-cellular population coverage has reached 93 per cent globally: in other words, almost every person on the globe lives within reach of a mobile-cellular signal and,
Closer examination and disaggregation of the data reveal, however, that digital divides still exist and that some people are excluded still from access to communication networks.
Even though rural population coverage is very high, at 87 per cent globally, at end 2012 around 450 million people worldwide still lived out of reach of a mobile signal.
ITU World Telecommunication/ICT Indicators database. 162.7 124.7 109.9 108.5 96.4 89.2 69.3 0 20 40 60 80 100 120
For countries where data are available, the number of mobile subscriptions far exceeds the number of mobile phone users (Partnership, 2014).
which data are available, but is closing with the availability of affordable mobile-phone services in rural areas.
Further research and data would be necessary to determine people's access to, and use of, voice communications and to identify other potential barriers,
Partnership (2014) based on ITU data. Overall mobile-cellular population coverage(%)Rural population covered(%)Rural population covered (millions) Rural population not covered (millions) Africa 88 79 498
129 Americas 99 96 171 9 Asia 92 87 2 017 309 Europe 99 98 196 3 Oceania 96 81
ITU World Telecommunication/ICT Indicators database. 27.7 16.7 14.3 9. 8 7. 7 3. 1 0. 4 05 10 15
The data on fixed-and mobile-broadband uptake confirm what has been observed on the ground. In developed countries, fixedbroadband infrastructure and services were available much earlier than in most developing countries,
ITU World Telecommunication/ICT Indicators database. 83.7 32.0 21.1 6. 3 0 10 20 30 40 50 60 70 80
for example when people use multiple devices (e g. smartphone, tablet) and SIM CARDS. Looking towards the future, the growth potential for mobile broadband looks promising,
data based on ITU and Telecom Advisory Services calculations. more and more countries upgrade their mobile networks. As mentioned earlier, 2g population coverage stands at over 90 per cent worldwide.
Data on 3g population coverage are less available. According to ITU estimates, global 3g population coverage stood at around 50 per cent by end 2012,
and there were still sizeable ruralurban gaps. Rural population coverage ranged from 100 per cent in the Gulf countries of United arab emirates
and Bahrain to zero in some African countries (Chart 1. 6). These numbers are expected, however to change significantly in the near future,
as more and more countries are deploying 3g+technologies and services, and given the strong growth in mobile-broadband subscriptions.
New data collected by ITU on the deployment of fibre transmission capacity in countries shows that by end 2013,
a closer look at the data also reveals major disparities across regions: Asia and the Pacific (in particular China and India) accounts for more than 85 per cent of the total length of backbone networks (Chart 1. 7, left.
Partnership (2014) based on ITU data. Percentage of rural population covered by at least a 3g mobile network 2012
such bandwidth being a key requirement for delivering data-intensive applications and services through high-speed networks.
ITU World Telecommunication/ICT Indicators database. connectivity, because of the strong internal demand and also its location:
ITU World Telecommunication/ICT Indicators database. 221 420 1'213 702 4'384 11'572 8'074 19'037 21
the Information Society Report 2014 The latest ITU data show that by end 2014, almost 44 per cent of the world's households will have Internet access at home, up from 40 per cent one year earlier and 30 per cent four years earlier (Chart 1. 10).
ITU World Telecommunication/ICT Indicators database. 78.0 57.4 53.0 43.6 36.0 35.9 11.1 0 10 20 30 40 50 60
In countries where data are available, rural household access falls far below urban household access,
2014). 8 Available data also show that Internet access in rural households is growing slowly, 78.4 43.6 31.2 5. 0 0 10 20 30 40 50 60 70
but data are not readily available for those countries. As has been illustrated earlier, network deployment is limited still
Data on public access is collected by ITU through its household questionnaire but only few countries report data on this indicator,
let alone broken down by urban and rural populations. Public access can be provided by commercial facilities,
while with 60 per cent coverage half of all rural areas would be connected. 9 The World Report series published by the International Federation of Library Associations
ITU World Telecommunication/ICT Indicators database. 78.3 40.4 32.4 8. 0%0 10 20 30 40 50 60 70 80
The data include both global top-level domain (gtld) and country code top-level domain (cctld) registrations,
Data sourced from Facebook. 0 500 1'000 1'500 2013 2012 2011 2010 2009 2008 2007 2006 2005 2004 Millions
International data on ICT access and use by enterprises are collected annually by the United nations Conference on Trade and Development (UNCTAD),
In the developing world, data on ICTS in enterprises are scarce and only collected by few countries.
Data supplied by Zooknic, compiled from cctld and other sources. Figures exclude fifteen cctlds which act as virtual gtlds. 2003 2008 2013 Millions%Millions%Millions%World 59.7 100.0 173.4 100.0 245.2 100.0 Developed 49.6
There is little data on the use of ICTS by government organizations and those countries that do have data are usually the more advanced ones with high levels of connectivity in general.
More information is available about government services provided online, tracked by the United nations through its E-government Survey,
The latest data show that, today, governments of all countries have established central websites and that more than 50 per cent of countries provide links to local and/or regional government agencies'sites (UNDESA, 2014).
UNCTAD Information Economy Database, 2014, available at unctadstat. unctad. org. Percentage of enterprises 92 85 78 72 43 31 30 50 18 14 0 10 20 30 40 50 60 70
Data from United nations E-government Survey (2014. Chart 1. 21: E-government services provided by countries (transactional services, left,
Data from United nations E-government Survey (2014). 101 73 60 46 44 42 41 40 39 33 29 27 76 0
and helps enhance education administration through the electronic exchange of forms, data and other information.
The latest available data from the UNESCO Institute for Statistics (UIS) 22 show that, in developed countries, the vast majority of schools have Internet access,
which data are available. In developing countries, school access to Internet is lower on average, although much progress has been made in recent years.
Data on broadband in secondary schools in Bangladesh are not available. Data for Nicaragua, Philippines and Indonesia do not include upper secondary.
Data for European countries and Costa rica refer to lower secondary. Data for Guyana, Nicaragua and Indonesia refer to primary and lower secondary.
Data for Cambodia include pre-primary schools. Data for Morocco, Tunisia, Guyana, Montserrat, Dominican republic, Nicaragua, Colombia, Trinidad and tobago, Bangladesh, Philippines, Sri lanka, Azerbaijan, Bhutan, Cambodia, Kazakhstan, Malaysia, Maldives, Singapore, Belarus
and the Russian Federation refer to public schools. In Suriname, there are no private schools in upper secondary.
Data for Palestine refer to West bank schools only. Source: UIS database, Partnership on Measuring ICT for Development WSIS Targets Questionnaire, 2013.
Percentage of schools Americas 0 10 20 30 40 50 60 70 80 90 100 Internet Fixed broadband Internet Percentage of schools
0 10 20 30 40 50 60 70 80 90 100 Asia and Africa Bangladeshnepal Kyrgyzstan Cambodiaphilippinessri Lanka Azerbaijanpalestine Indonesiabhutan Qatar
Saudi Arabiajordanturkey Iran I. R. Omanmalaysia Mongoliakazakhstan Thailandjapanarmeniageorgia Maldives Bahrain Brunei Darussalamchina, Hong kong Korea (Rep.)Singapore Ethiopiamorocco Sudansenegal Lesothobotwanaegypt Algeriatunisia Mauritius
rural areas often suffer from much lower network coverage and hence ICT uptake compared with urban areas.
Data for Philippines refer to primary and lower secondary. Data for Venezuela refer to primary only.
Data for Montserrat and Saint lucia refer to secondary only. Data for Palestine refer to West bank schools only.
Data for Bahrain, Belarus, Morocco and Tunisia refer to 2008. For Morocco, ICTQUALIFIED teachers figures refer to 2008.
Data for Azerbaijan, Barbados, Jordan, Saint lucia, Singapore, Trinidad and tobago, Uruguay, Philippines and Sri lanka refer to public schools only.
Source: UIS database, Partnership Questionnaire on WSIS Indicators, 2014. Proportion of teachers trained to teach basic computer skills
(or computing)(%Proportion of teachers trained to teach subjects using ICT(%)Anguilla Argentina Azerbaijan Bahrain Barbados Belarus Cayman islands Chile Montserrat Cuba Egypt Jordan Lithuania Malaysia China
, Hong kong Morocco Nauru Oman Palestine Qatar Saint kitts and nevis Saint lucia St vincent and the Grenadines Singapore Thailand Trinidad and tobago Tunisia Turks and Caicos Isl.
Available data collected by UIS at the international level shows that education systems in countries seem to put more emphasis on training teachers to teach subjects using ICTS than on training teachers to teach basic computer skills or computing (i e.
The document also includes two targets on data and monitoring and stresses the need to take urgent 25 Measuring the Information Society Report 2014 steps to improve the quality,
coverage and availability of disaggregated data to ensure that no one is left behind. 26 The role of ICTS as a key development enabler,
and encourages Member States to collect ICT data at the national level. Monitoring progress towards achievement of the WSIS outcomes has been an integral component of the WSIS process.
it referred to the Partnership on Measuring ICT for Development and its contribution towards developing indicators and collecting data and statistics on ICT.
to collect national-level data on the indicators identified to measure the WSIS targets. The results of the survey are featured in the report
along with other data sources. The Final review of the WSIS targets: Achievements, challenges and the way forward report is available at:
Using different data sources and big data analytics An important element in the discussion related to the post-2015 development agenda and the setting of measurable goals and targets has been need the to take a fresh look at the way data
and statistics are collected, analysed and disseminated, in view of the large data gaps prevailing in many developing countries in basic statistics in the areas of the economy, health education, labour, etc.,
all of which are crucial to monitoring the MDGS, and the future SDGS. This call for a data revolution was enunciated first in the report to the UN Secretary-general of the High-level Panel of Eminent Persons on the Post-2015 Development Agenda published in 2013 (see Box 1. 4 on the data revolution.
Since then, it has received considerable attention within the statistical community, as well as in other circles concerned with the lack of official statistics and development data.
According to the discussions, a data revolution considers new data sources, in addition to existing official sources: besides governments, other stakeholders such as the private sector, civil 27 Measuring the Information Society Report 2014 Box 1. 2:
A decade of successful international cooperation on ICT measurement In 2004, ICT measurement was still in its infancy,
calls surfaced for reliable and comparable data in order to take stock of the emerging information society, identify digital divides and measure the developmental impacts of ICTS.
covering many aspects of the information society and economy, is used widely by countries in the course of their national ICT data collection.
Data availability has increased also significantly over the past decade. In particular there are more comparable data on ICT infrastructure, household access and Internet users.
For example, at the beginning of the century, only around a dozen developing countries collected data on Internet users, while today there are almost 50 developing countries collecting this indicator through official surveys (Chart Box 1. 2). Data on household access to the Internet
or a computer are now being collected by more than 100 economies worldwide, and data on Internet use in businesses by almost 70 countries,
although not on a regular basis (Partnership UNSC 2011). Similarly, whereas no data were available on ICT access
and use in schools, they have started now to be compiled in many developing countries (see section 1. 5). At the same time,
major data gaps remain, in particular in developing countries and LDCS. This concerns, notably, statistics on ICT use by individuals, businesses, governments and other public-sector organizations, ICT-related employment data,
as well as data related to online security and cybercrime, gender and youth, and cultural and environmental aspects.
The growing information society will increasingly require more and better statistics to assess the social,
economic and environmental impacts of ICTS. The Partnership, in close collaboration with national statistical systems and the international donor community, will continue its efforts to address these challenges.
Number of countries collecting Internet user data through official surveys, by level of development Note: Chart shows countries that have collected data on the number of Internet users through official national surveys.
Data are presented in three-year intervals and include countries that have collected data for at least one year within those intervals.
Source: ITU. 0 2 11 24 36 49 1 3 27 37 41 40 1 5 38 61 77 89 0
20 40 60 80 100 Developing Developed World 1997 1998-002001-03 2007-09 2004-06 2010-12 (and before) Chapter 1
New data sources could include big data (mostly provided by private-sector companies) which could help improve the timeliness and completeness of data,
without compromising the relevance, impartiality and methodological soundness of the statistics (UNSC, 2014). The topic of big data is gaining momentum in the statistical community.
Chief statisticians gathering at the UNSC meetings in 2013 recognized that big data constitute a source of information that cannot be ignored by official statisticians
and that official statisticians must organize and take urgent action to exploit the possibilities and harness the challenges effectively (UNSC, 2014). 30 In view of declining responses to national household and business surveys in a number of countries,
big data could provide important sources of more timely and relevant information, thus complementing official statistics on the economy, society and environment.
and additional targets for Goal 2 are being reviewed and adjusted, based on contributions from Member States. c Due to data limitations,
currently mobile-broadband signal coverage is considering in determining this target. d Data being compiled by the Global Cybersecurity Index (GCI).
thus potentially becoming big data sources as well. At the UNSC meeting in 2014, the commission reiterated its call for the global statistical community to take action,
and supported the proposal to create a global working group on the use of big data for official statistics.
31 To make an inventory of ongoing activities and concrete examples regarding the use of big data for official statistics at regional,
access to data and legislation related to big data To address the issue of obtaining access at no cost to big data from the private sector for official statistical purposes,
as well as the issue of access to transborder data or access to data on transboundary phenomena To develop guidelines to classify the various types of big data sources
and approaches To develop methodological guidelines related to big data, including guidelines for all the legal aspects To formulate an adequate communication strategy for data providers
and users on the issue of use of big data for official statistics To reach out to other communities, especially those more experienced in IT issues or in the use of open data platforms.
The UN Global Working group on Big data for Official Statistics was launched formally in June 2014, under the auspices of the UN Statistics Division.
The mandate of the group, of which ITU is a member, includes: provide strategic vision, direction and coordination of a global programme on big data for official statistics;
promote practical use of sources of big data for official statistics; provide solutions for methodological, legal and privacy issues;
promote capacity building; foster communication and advocacy of the use of big data for policy applications;
and build public trust in the use of private-sector big data for official statistics. ICTS are part of the debate on the data revolution, big data and, more broadly,
emerging data issues in the post-2015 development debate. First, the ICT sector in itself represents a new source of data,
provided by, for example, Internet and telecommunication companies. Second, the spread and use of ICTS allow public and private entities across all economic sectors to produce,
store and analyse huge amounts of data. At the same time, however, monitoring access to and use of ICTS by people
public entities and private enterprises will be essential in order to identify the extent to which stakeholders in the ICT sector can be used as an alternative data source.
Without ICTS, no ICT-driven data revolution will take place. In view of the link between big data and ICTS, work is under way in ITU with a view to contributing to the debate
and identifying new ways and means of exploiting the potential of big data. The focus is primarily on the telecommunication/ICT sector as a source of big data,
including players such as operators and service providers, in the fixed, mobile and Internet sectors. Delegates attending the eleventh World Telecommunication/ICT Indicators Symposium (WTIS) in Mexico city in December 2014 recommended that ITU should further examine the challenges and opportunities of big data,
in particular data coming from ICT companies; that regulatory authorities should explore the development of guidelines on how big data could be produced,
exploited and stored; and that national statistical offices, in cooperation with other relevant agencies, should look into the opportunities for big data and address current challenges in terms of big data quality,
Chapter 1. Recent information society developments 30 veracity and privacy within the framework of the fundamental principles of official statistics. 33 The big data approach taken by ITU so far focuses on the following areas and questions:
Standardization: 34 Which standards are required to facilitate interoperability and allow technology integration in the big data value chain?
Which definitions, taxonomies, secure architectures and technology roadmaps need to be developed for big data analytics and technology infrastructures?
What is the relationship between cloud computing and big data in view of security frameworks? Which techniques are needed for data anonymization for aggregated datasets such as mobile-phone records?
How is exploited big data in different industries; what are the specific challenges faced; and how can these challenges be addressed through international standards?
Regulation: 35 What are the key regulatory issues at stake and how can and should big data be regulated?
How does big data impact on the regulation of privacy, copyright and intellectual property rights (IPR), transparency and digital security issues?
Box 1. 4: What is a data revolution? The report of the High-level Panel of Eminent Persons on the Post-2015 Development Agenda to the UN Secretary-general,
which was published in May 2013, has called for, inter alia, a data revolution taking advantage of new technology and improved connectivity:
We also call for a data revolution for sustainable development, with a new international initiative to improve the quality of statistics and information available to people and governments.
This has prompted a revolution in the debates taking place in the statistical communities at both the international and national levels on
what such a data revolution could entail and how it could be implemented. While no internationally agreed concept has thus far been defined,
the following elements seem to be part of a data revolution: 32 In view of the ubiquitous availability of communication networks, the use of new information technologies (e g. mobile technologies) should be leveraged for improving the collection
and dissemination of data Data should be further disaggregated (by gender, income, age, geography, etc.)
to ensure that no one is left out; in this regard, traditional statistical processes should be made more efficient Sustained investment in national statistical capacity,
both technical and institutional, is essential and needs to receive a major push from the international donor community The focus should go beyond data dissemination
and also include investment in the development of concepts, measurement frameworks, classifications and standards New,
nontraditional data sources should be explored and leveraged to complement existing ones and satisfy the demand for data needs in new areas, such as big data,
geospatial information and geographical information systems Open data policies should be envisaged to ensure accountability and promote transparency The role of data,
statistics and monitoring for policymaking and decision-making should be increased. 31 Measuring the Information Society Report 2014
What is the link between big data and open data (crowdsourcing, cloud computing, etc.)?is there need a to regulate data management and service providers?
How can market dominance in the area of big data be prevented and the rights of the data owners protected?
ICT data collection and analysis: How can big data complement existing ICT statistics to better monitor information-society developments?
Which type of data from ICT companies are most useful and for which purposes? Which new ICT indicators could be produced from big data sources?
What are the key issues that need to be addressed, and by whom, in terms of collecting and disseminating big data in telecommunications?
What is the role of national statistical offices and how can big data complement official ICT data?
How can big data from telecommunications inform not only ICT but broader development policy in real time, leading to prompt and more effective action?
Chapter 5 of this report addresses some of these questions and provides suggestions and recommendations for the way forward.
Chapter 1. Recent information society developments 32 1 Refers to countries where fixed-telephone penetration increased by more than 1 per cent in 2014.2 See https://gsmaintelligence. com/.3 http
://www. censusindia. gov. in/2011census/hlo/Data sheet/India/Communication. pdf. 4 4g refers to fourth-generation mobile network or service.
It is a mobile-broadband standard offering both mobility and very high bandwidth, such as long-term evolution (LTE) networks (ITU Trends 2014). 5 Data collection on Europe and North america will follow in 2014.6 For a list of IXPS,
see for instance http://www. datacentermap. com/ixps. html. 7 For more details on international submarine fibre-optic links,
see Telegeography's Submarine cable Map 2014, available at: http://submarine-cable-map-2014. telegeography. com. 8 For further discussion on progress made towards connecting rural households to the Internet,
see Partnership (2014), Chapter on Target 1. 9 Universal Postal Union (forthcoming 2014). Development strategies for the postal sector:
An economic perspective. 10 See http://www. ifla. org/faife/world-report. 11 Source: IMF World Economic Outlook Database, April 2014.12 Source:
The Economist, April 12 2014, Nigeria's GDP step change. 13 Telefónica, for instance, reduced its net debt by EUR 4 819 million in 2012 after several years of sustained increases in borrowings.
/18 Eurostat news release of 16 december 2013 and http://epp. eurostat. ec. europa. eu/statistics explained/index. php/Social media -statistics on the use by enterprises. 19 Data refer mostly to the year
highlighted a number of elements that should be part of a data revolution. 33 See final report of WTIS-13, available at:
http://www. itu. int/en/ITU-D/Statistics/Pages/events/wtis2013/default. aspx. 34 For further information on the work on big data carried out by the ITU Telecommunication
see http://www. itu. int/en/ITU-T/techwatch/Pages/big data-standards. aspx. 35 A background document on big data that was prepared for GSR-14 is available at http
and as more and better data become available. For example what was considered basic infrastructure in the past such as fixed-telephone lines is fast becoming less relevant in the light of increasing fixed-mobile substitution.
Data availability and quality. Data are required for a large number of countries, as the IDI is a global index.
There is relative paucity of ICT-related data, especially on ICT usage, in the majority of developing countries.
In particular, the three indicators included in the skills sub-index should be considered as proxies until data directly relating to ICT skills become available for more countries.
The results of various statistical analyses. The statistical associations between various indicators were examined, and principal components analysis (PCA) was used to examine the underlying nature of the data
and to explore whether the different dimensions are statistically well-balanced. While the basic methodology has remained the same
Given the dynamic nature of the ICT sector and related data availability, the indicators included in the IDI
and to experts in the field of ICT statistics and data collection, work through online discussion forums and annual face-to-face meetings.
Multi-SIM ownership is driving up mobile-cellular subscription numbers, which is an issue in prepaid and, to a lesser extent, also in postpaid mobile markets.
one possibility would be to replace the subscription-based (supply-side) data with data based on national household surveys (demand-side indicators).
still only 42 countries reported these data to ITU for at least one year between 2011 and 2013.
It is therefore too early to substitute the current mobile-cellular subscription data in the IDI with mobile-phone user data.
In view of the methodological difficulties in collecting harmonized data on international Internet bandwidth a review of the definition of the indicator is currently under discussion in EGTI.
Preparation of the complete data set. This step includes filling in missing values using various statistical techniques.
Normalization of data. This is necessary in order to transform the values of the IDI indicators into the same unit of measurement.
Rescaling of data. The data were rescaled on a scale from 0 to 10 in order to compare the values of the indicators and the sub-indices.
Weighting of indicators and sub-indices. The indicator weights were chosen based on the principal components analysis (PCA) results.
This chapter presents the IDI based on data from 2013 in comparison with 2012. It should be noted that IDI 2012 values have changed from those published in the previous edition of this report as a result of:
Country data revisions. As more accurate data become available, countries provide ITU with revised statistics for previous years,
which have been taken into consideration. This also allows ITU to identify inconsistencies and revise previous estimates.
In each new edition, some countries are excluded and others added based on data availability. Overall, this version of the IDI includes 166 countries/economies as compared with 157 in last year's edition.
According to data from the European union (EU), 85 per cent of Danes have some level of computer skills (compared to the EU average of 67 per cent)
the digital economy accounted for more than 5. 8 per cent of GDP, and it continues to grow.
Denmark's national target even exceeds the EU's Digital Agenda objective of 100 per cent coverage of households with broadband speeds of 30 Mbit/s
In both indicators although Sweden has a slightly higher wireless-broadband penetration Denmark surpasses the other top five IDI countries (see Chart 2. 1). In terms of LTE population coverage
ITU World Telecommunication/ICT Indicators database. 33 35 36 38 40 110 75 87 105 107 0 50 100 150
ITU World Telecommunication/ICT Indicators database. have access to mobile broadband at speeds of at least 10 Mbit/s. 10 The Republic of korea ranks second in the IDI 2013.
Full coverage having being achieved (by April 2012, LTE was available nationwide), the wirelessbroadband market is showing signs of saturation, with little growth over the past years.
Data also show that the Republic of korea achieves the highest advertised fixed-broadband speeds, with all subscriptions providing at least 10 Mbit/s. This compares with 75 per cent of fixed-broadband subscriptions at advertised speeds of at least 10 0 20 40 60 80 100 120
There is however a sizeable domestic demand for data driven by the high volume of local content,
Sweden and Finland are the countries with the highest LTE coverage in the European region (European commission, 2014a).
Data from the EU confirm that household access is correlated highly with regular use of the Internet
and sets ambitious targets to have 50 per cent of homes subscribed to ultra-fast broadband (at least 100 Mbit/s) and coverage of all households by broadband speeds of at least 30 Mbit/s by 2020.16
The UK government aims to achieve coverage of at least 90 per cent in 2016, and has made GBP 530 million of funding available for the roll out of networks in sparsely populated underserved areas.
Data from the European commission's Digital Agenda underline the competitiveness of the European fixedbroadband market:
ITU World Telecommunication/ICT Indicators database. 0. 0 0. 1 0. 3 1. 0 2. 5 15.6 02468 10 12
The launch of mobile-broadband services by the country's only private-owned operator Tashi Cell in late 2013 has helped to increase coverage and competition in the market,
Proportion of households with a computer and proportion of households with Internet access, 2012-2013, Qatar ITU World Telecommunication/ICT Indicators database. 91.5 88.1 97.2 96.4 0
and promotions on mobile data plans. 35 During 2013, operators further extended their wireless infrastructure and services throughout Thailand,
Based on the 2013 and 2012 data presented in this chapter, the current (2013) global divide is measured
Data change very little over time and advances in skills do not show immediate effects.
Above 2. 78 LCC (2. 78 and below) Data not available 59 Measuring the Information Society Report 2014 the profitability of various kinds of economic activities (Sachs, 2012.
Apart from the potential relationship of these variables with ICT developments, they were selected for their high data availability for a large number of countries.
Data on urban population, population density, land area and GNI p. c. are sourced from the World bank.
Data on population size are sourced from UNPD.**Correlation is significant at 0. 01 level.
mobile-cellular coverage for rural populations has reached very high levels, with almost 90 per cent of the world's rural inhabitants covered by a 2g mobile-cellular signal by 2013.
On the other hand, 3g mobilecellular coverage was comparatively low for 0123456789 10 0 10000 20000 30000 40000 50000 60000 GNI p. c. USD) r=0. 775 IDI
when it comes to data on Internet access and use. Access to the Internet (be it narrowband
People living in rural areas, particularly in developing countries, are disadvantaged compared to their urban counterparts because of lower service coverage;
a correlation analysis between IDI values and MDG indicators was conducted for 2011 where data are available for both sets of indicators.
of which data pertaining to both the MDG indicators and the IDI are available for at least 16 countries. 41 The MDG indicators measuring Goal 2 (literacy rate
A comparison of IDI values of the 147 economies for which data were available for 2002 and 2013 shows that the global IDI value has doubled almost from 2. 52 in 2002 to 4. 88 in 2013
this would require different sets of data (including micro data) collected from official surveys. Therefore, the analysis should be considered as a first step in an attempt to quantify the relationship between ICT performance and MDG progress.
which sufficient data are available Almost all MDG indicators that are included under MDG 1, MDG 4, MDG 5,
and women using appropriate technologies MDG4 Reduce child mortality) Data collected through ICT applications allow health professionals to assess child health
SMS reminders for appointments or medical treatment) Improve access to medical databases and electronic health records (EHRS) Link community health workers to the national health system/specialist support MDG7 (Ensure environmental sustainability) Reduce greenhouse gas (GHG) emissions
Smart planning and reliable access to real-time data for climate monitoring, as well as the implementation of early warning systems Reduce energy and water consumption through smart transportation and logistics, dematerialization and other technologies.
Achieve, by 2015, universal access to reproductive health 5. 3 Contraceptive prevalence rate 5. 4 Adolescent birth rate 5. 5 Antenatal care coverage (at least one
Insufficient data.++The indicator or similar indicator is included in the IDI. Source: UN for the list of MDG indicators and ITU for the correlation results.
5. 5 Antenatal care coverage (at least one visit and at least four visits) IDI Goal 7 7. 2 CO2 emissions,
through the analysis of health data collected through public health applications and by serving as a platform for exchange and advocacy.
120 140 160 180-1 2 3 4 5 IDI r=0. 529 Antenatal care coverage, at least four visits, percentage 0
but no relationship exists between antiretroviral therapy coverage among people with advanced HIV and IDI (see Chart 2. 16).
The use of smartphones to capture essential data on the patients and monitor their treatment has accelerated progress.
where 2002 and 2011 data are available for both sets of indicators. The following steps were performed:
which data are available for 2002 and 2011 for both the MDG indicators and the IDI The number of countries included in the analysis ranges from 16 to 101.
and this indicator in the earlier section where 2011 data were analysed, improvements in the level of ICT access
This type of analysis will require different data sets, including micro data on ICT usage collected from official surveys.
Micro data offer analysts and researchers ample information and considerable flexibility to apply quantitative models that identify relationships
and interactions between indicators and topics covered in a survey, thereby fostering the diversity and quality of research and analysis. Chapter 2. The ICT Development Index (IDI) 80 1 This section is based on the 2013 edition of Measuring the Information Society.
as well as the 2009 edition of Measuring the Information Society (ITU, 2009), which describe the methodology in more detail. 2 Data on the indicators included in the skills sub-index are sourced from the UNESCO Institute for Statistics (UIS).
http://data. worldbank. org/about/country-classifications/country-and-lending-groups#High income. 7 Based on 2011 data:
the in-scope population for data on Internet users is aged individuals 16-74.20 Refers to the indicator active mobile-broadband subscriptions.
as well as the household access data, excludes transient labourers, which account for a significant proportion of residents in Qatar.
According to data from ICTQATAR transient labourers make up 27 per cent of the overall population. 33 http://qnbn. qa/qatar-vision-2030/34 http://www. nbtc. go. th/wps/portal/NTC/!
-data-plans--900198.36 http://www. telegeography. com/products/commsupdate/articles/2013/05/09/true-4g-launch-trumps-rivals-ais-claims-800000
World bank, see http://data. worldbank. org/indicator/NY. GNP. PCAP. CD. 39 See: Czernich, N.,Falck, O.,Kretschmer, T. and Woessmann, L. 2009), Broadband Infrastructure and Economic growth, http://papers. ssrn. com/sol3/papers. cfm?
if there is available data for at least one year between 2010 and 2012. The number of countries included in the analysis varies from 16 to 101 developing countries.
since it is the closest year to available MDG data. 42 Broadband Commission. Transformative Solutions for 2015 and Beyond a Report of the Broadband Commission Task force on Sustainable Development and http://www. broadbandcommission. org/Documents/Climate/BD-bbcomm-climate. pdf and The State
Improving coverage is particularly challenging in vast rural areas and where the reach of basic infrastructure,
ITU World Telecommunication/ICT Indicators database. 0 20 40 60 80 100 120 140 160 180 2012 2013 Per 100 inhabitants
at 43 per cent (after Botswana with 74 per cent) following an expansion of network coverage throughout the archipelago. 3 Large-scale infrastructure rollout also helped to increase uptake of wireless broadband in Nigeria4 (from 5 per cent in 2012
Data from South african operators show that not only is wirelessbroadband penetration reaching higher levels 29per cent by end 2013
but customers are consuming more data, indicating an increase in the intensity of usage. MTN reported a growth of 63 per cent in data volumes in the first half of 2013
and Vodacom reported that on average users were generating 75 per cent more data traffic per device than a year ago. 5 Wireless broadband is of particular importance in the region
because fixed-broadband infrastructure is lacking. The vast majority of African countries 32 out of 38 had fixed a-broadband penetration of less than 1 per cent by end 2013.
and high levels of multi-SIM ownership (GSMA and Deloitte, 2013). Furthermore, the very high mobile-cellular penetration rates reached in the GCC countries are driven by large transient worker and expatriate populations.
Data from household surveys show that the actual number of people using a mobile-cellular phone is much lower than the number of subscriptions.
ITU World Telecommunication/ICT Indicators database. 2012 2013 Per 100 inhabitants Bahrain United Arab Emiratesqatar 0 20 40 60 80 100
Data from the annual ICT household survey show that, since 2008, computers have replaced telephones as the most commonly available ICT device in Thai homes.
ITU World Telecommunication/ICT Indicators database. 2012 2013 Per 100 inhabitants 0 20 40 60 80 100 120 140 Macao, Chinasingapore
The mobile markets in the CIS are predominately prepaid, with typically high rates of multi-SIM ownership.
Data from household surveys collected in a number of CIS countries underline that mobilecellular penetration,
ITU World Telecommunication/ICT Indicators database. 2012 2013 Per 100 inhabitants Belarus 05 10 15 20 25 30 35 Azerbaijanrussian Federationmoldova
Data from the Eurobarometer underlines this finding: on average, 92 per cent of European union citizens (the majority of countries in the region are members of the EU) had access to a mobile phone in 2013 (European commission, 2014b.
Fast broadband coverage (defined at 30 Mbit/s) should be available throughout the entire EU
Data on Individuals using the Internet for Eurostat members are sourced from Eurostat. Eurostat collects data for Internet users aged 16-74 years old.
Source: ITU World Telecommunication/ICT Indicators database. World Developed 0 10 20 30 40 50 60 70 80 90 100%101 Measuring the Information Society Report 2014 penetration stands
at 6 per cent, all European countries exceed the global average penetration. A well-developed ICT infrastructure and the availability of high-speed broadband Internet access and relevant content are reflected in a higher proportion of Internet users in the region.
or further extended 3g coverage in 2013, spurring growth in the mobile sector. The United states has the highest wirelessbroadband penetration, at 94 per cent by end Chart 3. 13:
ITU World Telecommunication/ICT Indicators database. 0 10 20 30 40 50 60 70 80 90 2012 2013 Per 100 households
and coverage was extended massively throughout the country in 2013. The operator Verizon had achieved 97 per cent LTE population coverage,
and the majority of all data traffic is carried by the LTE network. 27 Very high increases were reported by Brazil,
where 40 million new wireless-broadband subscriptions were added within a year, resulting in a penetration of 52 per cent by end 2013.
Internet user data from Gulf countries are not comparable, as they refer to different populations. Data from Bahrain and Qatar refer to the overall population,
i e. including expatriate/transient workers. Data from United arab emirates are estimated by ITU based on base data excluding the transient worker population. 12 Reported in activated external capacity. 13 http://www. ofca. gov. hk
/en/industry focus/telecommunications/facility based/infrastructures/submarine cables/index. html. 14 http://submarinenetworks. com/systems/intra-asia/sjc/sjc-cable-system. 15 http
/26 Data reported by the country refer to 2012.27 http://www. verizonwireless. com/wcms/consumer/4g lte. html
In response to the demand for global benchmarks on ICT prices, ITU has been collecting ICT price data following a harmonized methodology since 2008.
Since 2012, the data collection has been extended to include mobilebroadband prices. These data have proved to be useful for the international comparison of ICT prices across more than 160 countries,
and for identifying those cases where prices constitute a barrier to ICT uptake. This year's analysis of ICT prices goes beyond simply measuring affordability,
They include end-2013 data for each of the three price sets contained in the IPB (fixed-telephone
The coverage gaps in the fixed-telephone network have been filled by the mobile-cellular network, which covers 93 per cent of the global population.
This compares with a global 3g coverage of around 50 per cent by end 2012
Based on 140 economies for which 2008-2013 data on fixed-telephone and mobile-cellular prices were available.
and PPP$ 28.4 (or USD 19.5) per month for a prepaid mobilebroadband service with a 500 MB monthly data allowance. 7 Despite the limitations of comparing such different services,
The fixed-telephone market is the most mature segment of those included in the ITU price data collection exercise.
78 per cent of the countries with price data had already fully or partially liberalized their fixed-telephone market in 2008,
Based on 140 economies for which 2008-2013 data on fixed-telephone and mobile-cellular prices were available.
*Data correspond to the GNI per capita (Atlas method) in 2013 or latest available year adjusted with the international inflation rates.**
because data on GNI p. c. are not available for the last five years. Source:
GNI p. c. and PPP$ values are based on World bank data. Rank Economy Fixed-telephone sub-basket GNI p. c.,USD, 2013*Rank Economy Fixed-telephone sub-basket GNI p. c
*Data correspond to the GNI per capita (Atlas method) in 2013 or latest available year adjusted with the international inflation rates.**
because data on GNI p. c. are not available for the last five years. Source:
GNI p. c. and PPP$ values are based on World bank data. Rank Economy Mobile-cellular sub-basket GNI p. c.,USD, 2013*Rank Economy Mobile-cellular sub-basket GNI p. c
fixed broadband remains the de facto option for accessing high-volume Internet applications such as file sharing (less than 1 per cent of total filesharing traffic was transmitted through mobile networks in 2013)
In addition, CISCO estimates that 45 per cent of total mobile data traffic was offloaded onto fixed networks in 2013 (CISCO,
Based on 143 economies for which 2008-2013 data on fixed-broadband prices were available. Source:
This is in line with the findings on bundle adoption from household surveys (European commission, 2014) and data on fixed (wired)- broadband subscriptions by speed
Based on 143 economies for which 2008-2013 data on fixed-broadband prices were available. Source:
Based on 165 economies for which 2013 data on fixed-broadband prices were available. Source: ITU.
GNI p. c. values are based on World bank data. 0. 0 0. 5 1. 0 1. 5 2. 0 2. 5 3
GNI p. c. values are based on World bank data. 02468 10 12 As a%of GNI p. c. Russian Federationkazakhstan Belarus Ukraineazerbaijan Armeniageorgia*Uzbekistanmoldova Kyrgyzstan
This may be explained by the limited coverage of Qualitynet, which suggests the need for more investment in broadband network equipment
GNI p. c. values are based on World bank data. for fixed-broadband uptake in the country.
GNI p. c. values are based on World bank data. 05 10 15 20 As a%of GNI p. c. 85.8 66.1 United states Trinidad & Tobago Venezuela
GNI p. c. values are based on World bank data. There are 13 countries in the Asia and the Pacific region where entry-level fixed-broadband plans cost more than 10 per cent of GNI p. c. These include some large countries, such as Pakistan and the Philippines,
the latest data on international connectivity show that this may remain an issue in Kiribati (45 Mbit/s), Marshall islands (45 Mbit/s), Micronesia (45 Mbit/s), Samoa (135
GNI p. c. values are based on World bank data. However, a comparison with other regions shows that it is feasible to improve the affordability of fixed-broadband prices in Africa significantly
*Data correspond to the GNI per capita (Atlas method) in 2013 or latest available year adjusted with the international inflation rates.**
because data on GNI p. c. are not available for the last five years. Source:
GNI p. c. and PPP$ values are based on World bank data. Rank Economy Fixed-broadband sub-basket Speed in Mbit/s Cap per month in GB GNI p. c.,USD, 2013*Rank Economy Fixed
and this figure will grow as more and more mobilebroadband networks are deployed, until eventually 3g coverage approaches mobilecellular coverage (93 per cent).
On top of the main types of mobile-broadband plans for which ITU collects data on prices (Figure 4. 1),
and Verizon in the United states, are allowing customers to pool the data consumed by different devices in a single subscription. 23 In addition,
and has unlimited access to the Internet at a given speed, with neither time nor data volume constraints.
and almost all of them include data volume caps, e g. USD 10 for 50 MB per month.
Several operators also offer advantageous plans based on a combination of time and data volume limitations, e g.
most mobile-broadband plans are priced on the basis of the data allowance (i e. the data volume in MB included in the plan) and not the speed.
A comparison of mobile-broadband prices across time would reflect the changes in pricing structures (changes in data allowances,
The global average price for a computer-based mobile-broadband service with 1 GB monthly data allowance was PPP$ 36.6 (or USD 24.4) for prepaid plans and PPP$ 30.0 (or USD 19.2
The average cost for a handset-based mobilebroadband service with 500 MB monthly data allowance was PPP$ 25.3 (or USD 16.9) for prepaid plans and PPP$ 25.7 (or USD 17.6) for postpaid
because the monthly data allowance was half as large. Nevertheless, the reduction in price was not proportional to the reduction in the data allowance,
confirming that the price per GB is lower for larger data allowances, the equivalent of a volume discount.
Unlike in the case of computer-based mobile-broadband services, the prices for postpaid and prepaid handsetbased mobile-broadband plans were similar,
Based on 119 economies for which data on mobilebroadband prices were available for the four types of plans.
Indeed, handsetbased mobile-broadband plans with a monthly data allowance of 500 MB are about eight times more affordable in developed countries than in developing countries, on average (Chart 4. 14.
The average price for a computer-based mobile-broadband service with 1 GB monthly data allowance corresponded to more than 20 per cent of GNI p. c. in Africa,
Based on 119 economies for which data on mobilebroadband prices were available for the four types of plans.
Based on 119 economies for which data on mobilebroadband prices were available for the four types of plans.
with mobile-broadband subscribers consuming much less than 500 MB of Internet data per month, supported by the fact that several African operators offer discount plans for occasional use.
For instance, Internet video cannot be consumed on the basis of such limited data allowances, and even Internet radio would need to be limited.
Based on 119 economies for which data on mobilebroadband prices were available for the four types of plans.
Country data for The americas reveal that there are a number of countries which have high prepaid computerbased prices
because minimum packages include a monthly data allowance much larger than 1 GB (Table 4. 8). This is the case of Chile (USD 73 for 14 GB), Antigua and barbuda (USD 63
Such high monthly data allowances for prepaid mobile-broadband dongles suggest that these services target high-end customers, rather than the average user.
Postpaid mobilebroadband dongles include much lower monthly data allowances in The americas (Table 3. 7), suggesting that postpaid rather than prepaid is the base offer for regular computer-based mobile-broadband customers.
Average prices for computer-based mobilebroadband plans with a monthly data allowance of 1 GB suggest that mobile broadband could be a cheaper alternative to fixed broadband in many Chapter 4. ICT prices and the role
Percentages are calculated on the basis of the total number of countries with data available in each region:
and include a monthly data allowance of at least 1 GB. Although the minimum data allowance is the same,
in practice most fixed-broadband plans allow unlimited data use (Table 4. 4), whereas most computerbased mobile-broadband plans with a minimum monthly data allowance of 1 GB really do have a cap of 1 GB (Tables 4. 7 and 4. 8). In almost half
of the African countries included in the price benchmark, mobile-broadband prices were more than USD 10 cheaper per month than fixed-broadband prices.
Taking into account the GNI p. c. levels in Africa such price savings could make the difference between a service being affordable or not.
c.,USD, 2013*Monthly data allowance (MB) as%of GNI p. c. USD PPP$ 1 Austria 0. 13 5. 31 4. 62
p. c.,USD, 2013*Monthly data allowance (MB) as%of GNI p. c. USD PPP$ 80 Maldives 2. 08 9. 7 12.6
*Data correspond to the GNI per capita (Atlas method) in 2013 or latest available year adjusted with the international inflation rates.**
because data on GNI p. c. are not available for the last five years. Bundles include:
GNI p. c. and PPP$ values are based on World bank data. Chapter 4. ICT prices and the role of competition 134 Table 4. 6:
. c.,USD, 2013*Monthly data allowance (MB) as%of GNI p. c. USD PPP$ 1 Norway 0. 1 8. 34 5 102
*Data correspond to the GNI per capita (Atlas method) in 2013 or latest available year adjusted with the international inflation rates.**
because data on GNI p. c. are not available for the last five years. Bundles include:
GNI p. c. and PPP$ values are based on World bank data. Rank Economy Mobile-broadband, prepaid handset-based (500 MB) GNI p. c.,USD, 2013*Monthly data allowance (MB) as%of GNI p
. c. USD PPP$ 80 Antigua & Barbuda 2. 58 27.78 33.8 12'910 1'024 81 India 2. 58 3. 38
. c.,USD, 2013*Monthly data allowance (MB) as%of GNI p. c. USD PPP$ 1 Austria 0. 13 5. 31 4. 62
*Data correspond to the GNI per capita (Atlas method) in 2013 or latest available year adjusted with the international inflation rates.**
because data on GNI p. c. are not available for the last five years. Source:
GNI p. c. and PPP$ values are based on World bank data. Rank Economy Mobile-broadband, postpaid computer-based (1 GB) GNI p. c.,USD, 2013*Monthly data allowance (MB) as%of GNI p
. c. USD PPP$ 76 Montenegro 2. 89 17.49 28.93 7'260 3 77 China 2. 95 16.14 26.60 6'560
. c.,USD, 2013*Monthly data allowance (MB) as%of GNI p. c. USD PPP$ 1 Austria 0. 13 5. 31 4. 62
*Data correspond to the GNI per capita (Atlas method) in 2013 or latest available year adjusted with the international inflation rates.**
because data on GNI p. c. are not available for the last five years. Source:
GNI p. c. and PPP$ values are based on World bank data. Rank Economy Mobile-broadband, prepaid computer-based (1 GB) GNI p. c.,USD, 2013*Monthly data allowance (MB) as%of GNI p
. c. USD PPP$ 76 South africa 4. 82 28.90 51.17 7'190 1 77 Suriname 4. 91 37.88 64.17 9'260
and hunger have contributed to making available more data on household and individual economic welfare, as well as its distribution.
which data on the distribution of household income or consumption expenditure are available. The objective is to explore how factors such as income inequality may affect people's access to,
as a way of classifying countries by income levels for instance by the World bank. 30 Data on household income, on the other hand, measure only people's economic welfare,
Data are collected by national statistical offices by means of household income and expenditure surveys (HIES)
In order to classify income data by deciles, individuals are placed in ascending order according to the household income attributed to them,
such as access and availability of products, may be relevant. 34 Data on household consumption expenditure are obtained also from household surveys.
when making international comparisons based on data on household economic welfare because there are several methodological issues that limit comparability, notably:
householdlevel economic data provide information (not available from macroeconomic indicators) on the actual income and expenditure capacity of households in a country and the differences across households.
these data can be used to obtain a finer-grain indication of the affordability of broadband services for households from different economic levels.
Data for the United states and Sweden are sourced from the OECD Database on Income Distribution
Data for South africa and Viet nam are sourced from the World bank's Povcalnet and refer to 2008.0 20 40 60 80 100 10 20 30 40 50 60 70 80 90 100%population Sweden United states%of total
Chart 4. 18 uses data on income inequality to reveal differences in the affordability of fixed-broadband services between these two countries,
Household disposable income based on World bank's Povcalnet data adjusted with ITU estimates on average persons per household. would need to spend 8. 5 per cent of their household disposable income,
economic data at the household level also make it possible to determine more precisely the affordability of residential fixed-broadband services.
Household consumption based on World bank's Povcalnet data adjusted with ITU estimates on average persons per household.
and applying this threshold to both household disposable income and expenditure data, the following conclusions can be drawn:
In more than 85 per cent of countries for which data are available, the richest 20 per cent of the population can afford an entry-level fixed-broadband plan.
In 40 per cent of countries for which data are available, a basic fixedbroadband subscription still represents more than 5 per cent of household income/consumption for over half of the population.
As a result, in most developing countries for which data on household income or expenditure distribution are fixed available
Data on household disposable income refer to 2011 or latest year available.*‘*‘Lowest 20%'refers to the price divided by the average income of the first and second income/consumption deciles.‘
Household disposable income for OECD countries and the Russian Federation based on data from the OECD Database on Income Distribution adjusted with ITU estimates on average persons per household.
Household disposable income and consumption expenditure for other countries based on World bank's Povcalnet data adjusted with ITU estimates on average persons per household.
Data on household consumption expenditure refer to 2011 or latest year available.*‘*‘Lowest 20%'refers to the price divided by the average expenditure of the first and second consumption expenditure deciles.‘
Household consumption expenditure based on World bank's Povcalnet data adjusted with ITU estimates on average persons per household.
In the Arab States, the comparison is limited by lack of data on the distribution of household income/expenditure.
Available data suggest that basic fixedbroadband plans are affordable for 90 per cent of the population in Jordan and Tunisia
Table 4. 11 and Table 4. 12 show the price of prepaid handset-based mobile-broadband plans (with a 500 MB monthly data allowance) as a percentage of disposable household
which ITU collects price data (Chart 4. 13), and is currently the mobile-broadband service that is available in most countries (Chart 4. 11). 147 Measuring the Information Society Report 2014 Handset-based mobile-broadband services are affordable
which data are available. This is also the situation in Armenia, Dominican republic and Egypt. In these countries, a prepaid handset-based mobile-broadband plan represents
However, subscription data show that, in developed countries, handset-based mobile-broadband subscriptions are individual,
Data on household disposable income refer to 2011 or latest year available.*‘*‘Lowest 20%'refers to the price divided by the average income of the first and second income deciles.‘
Household disposable income for OECD countries and the Russian Federation based on data from the OECD Database on Income Distribution adjusted with ITU estimates on average persons per household.
Household disposable income for other countries based on World bank's Povcalnet data adjusted with ITU estimates on average persons per household.
Data on household consumption expenditure refer to 2011 or latest year available.*‘*‘Lowest 20%'refers to the price divided by the average expenditure of the first and second expenditure deciles.‘
Household consumption expenditure based on World bank's Povcalnet data adjusted with ITU estimates on average persons per household.
Chart 4. 20 complements the previous analysis by showing the affordability of handset-based mobilebroadband plans assuming that each member of the household has his/her own SIM CARD with a mobile-broadband plan.
or consumption per person. 39 Data show that handset-based mobilebroadband prices (with a 500 MB monthly allowance) are affordable for a majority of the population in developed countries,
Data on equivalized household income and consumption expenditure per decile refer 2011 or latest available year.
Equivalized household disposable income for OECD countries and the Russian Federation based on data from the OECD Database on Income Distribution.
Equivalized household disposable income and consumption for other countries based on World bank's Povcalnet data adjusted with ITU estimates on average persons per household. high-income economies (Australia and New zealand) to unaffordable for a majority
Data for the Arab States are limited to four countries. In Tunisia and Jordan, individual handset-based mobile-broadband subscriptions are affordable for a majority of the population.
On the basis of the data presented, it can be concluded that income inequality does not only have an impact on the proportion of households within a country that have access to mobilebroadband services,
Among all ICT services mobile cellular and fixed broadband have been selected for the analysis because of the availability of comprehensive data series on the prices for these two services,
and is limited often in scope because of lack of data for developing countries. Quantitative studies are in several cases restricted to samples of EU and OECD countries
by using ITU price and regulatory data for up to 144 countries in the period 2008-2013.
and to check to what extent the quantitative results based on telecommunication data from EU, OECD and specific countries hold true in a global context.
since data availability and comparability for such a large number of countries is limited more. Therefore, more specific quantitative results,
Simple averages for 140 economies with available data on fixed-broadband prices and competition for the period 2008-2013.
Herfindahl-Hirschman Index (HHI) data sourced from Informa. Chart 4. 22: Competition in mobile markets and mobilecellular prices as a percentage of GNI p. c.,2008-2013 Note:
Simple averages for 140 economies with available data on mobile-cellular prices and competition for the period 2008-2013.
Herfindahl-Hirschman Index (HHI) data sourced from Informa. competition and the openness of the market to private and foreign investment.
the following section presents an econometric model based on panel data regression. This enables us to go beyond descriptive statistics and draw some robust conclusions on the link between competition and prices.
For the analyses in this section, data from the Regulatory Tracker have been extracted for clusters 1, 2 and 3,
Choice of the model The analysis was conducted through econometric modelling using panel regressions for up to 144 countries based on data for the five-year period from 2008 to 2013.
Panel data regression is a statistical technique which is used to assess how variations in a set of variables over a given time period relate to Figure 4. 3:
and 1 GB of data allowance included per month. However, entry-level fixed-broadband plans in several countries offer higher speeds and larger data allowances.
In order to measure how these enhanced features affect prices, a variable on fixedbroadband speed and another one on capped data allowances are included as controls in the model for fixed-broadband prices.
Chapter 4. ICT prices and the role of competition 158box 4. 2: Panel regression models for fixed-broadband and mobile-cellular prices Two models are used for the regressions:
Data collected by ITU, see Annex 2 for more details on the methodology for the collection of fixed-broadband prices.
Data collected by ITU, see Annex 2 for more details on the methodology for the collection of mobile-cellular prices.
Data sourced from Informa. Herfindahl-Hirschman Index for mobile cellular (voice and SMS: with Li being the number of mobile-cellular subscriptions of firm i,
Data sourced from Informa. 159 Measuring the Information Society Report 2014 Box 4. 2: Panel regression models for fixed-broadband and mobile-cellular prices (continued) Regulatory variable:
Data collected by ITU, see www. itu. int/tracker for more information. Fixed-broadband speed:
Speed of the entry-level fixed broadband plan in Mbit/s. Data collected by ITU.
if the fixed-broadband plan includes a cap in the monthly data allowance, and 0 otherwise.
Data collected by ITU. Descriptive statistics of the dependent variables: Average Standard deviation Minimum Maximum 2008 2013 2008 2013 2008 2013 2008 2013 GNI p. c. 14'707 16'439 18
Descriptive statistics calculated for 124 economies that have complete data for the two models. Source: ITU.
plans with data caps are linked to prices 31%lower than unlimited plans Fixed-broadband speed 0. 016 (0. 023) Not significant Constant 4. 304 (0
The model also suggests that the application of data caps by fixed-broadband service providers is correlated with cheaper entry-level fixedbroadband plans,
The ITU data collection considers a minimum of 1 GB Chapter 4. ICT prices and the role of competition 162 monthly consumption for fixed-broadband plans.
whether it includes a data cap of 1 GB or more, or allows unlimited traffic.
In 2013, entry-level fixed-broadband plans included a data cap in 35 per cent of the countries included in the fixed-broadband model, a higher percentage than in 2008 (24 per cent.
fixedbroadband prices are cheaper in countries where data caps were enforced in entry-level fixedbroadband plans.
This suggests that operators can offer cheaper prices in exchange for reduced data consumption, thus indicating that capacity in fixed-broadband networks is still an issue in several countries,
i e. the marginal cost of additional Internet data beyond 1 GB is still nonnegligible in many countries.
such as operators'strategies on data caps, competition in the fixed-broadband market and the ICT regulatory environment, may together be a greater determinant for fixed-broadband prices than the price difference explained by GNI
Another factor that is found to be fixed relevant in-broadband prices is the existence of data caps
*Data correspond to the GNI per capita (Atlas method) in 2013 or latest available year adjusted with the international inflation rates.
GNI p. c. and PPP$ values are based on World bank data. Chapter 4. ICT prices and the role of competition 168 and that policy-makers and regulators may need to look at the actions that can be taken to ease capacity bottlenecks.
Based on the econometric model, it can be concluded that factors that are purely attributable to the telecommunication sector, such as operators'strategies on data caps,
For more information on the PPP methodology and data, see http://icp. worldbank. org. 3 GNI takes into account all production in the domestic economy (i e.
http://data. worldbank. org/indicator/NY. GNP. PCAP. CD. 4 Voice over internet services, such as Skype or Voipbuster, are excluded from the analysis in this section
ITU data for countries accounting for 97 per cent of global fixed (wired)- broadband subscriptions,
see http://data. worldbank. org/about/country-classifications/country -and-lending-groups. 11 The Communications Commission of Kenya (CCK) issued mobile virtual network operator licences to three operators in April 2014.
These data suggest there is limited competition among operators. 13 Advanced mobile technologies refer to standards agreed by the ITU Radiocommunication Assembly for next-generation mobile technologies IMT-Advanced such as
For the details of Verizon's MORE Everything plan, see http://www. verizonwireless. com/wcms/consumer/shop/shop-data-plans/more-everything. html. 24 The details of the different‘4g'plans
-con-el-modem. 25 Data for mobile-broadband services have been collected since 2012 through the ITU ICT Price Basket Questionnaire,
http://data. worldbank. org/about/country-classifications/country-and-lending-groups. 31 Household incomes include wages, salaries, self employment incomes, capital and property income,
http://go. worldbank. org/W3hl5gd710. 35 Differences in the equivalence scales of the source data used in this chapter are corrected roughly using ITU estimates on the number of inhabitants per household for each country.
World bank Povcalnet data on income distribution are published per capita, and figures from the OECD Database on Income Distribution are equivalized using the square root scale. 36 For OECD countries,
household income is estimated by multiplying the equivalized household income by the square root of the average number of inhabitants per household in the country.
According to 2008/9 survey data, mean household consumption in Angola was almost the same as in Uganda,
rather than to an overall higher level of income and consumption expenditure by most households. 38 Data on income distribution are aggregated per household
data on income distribution are averaged per decile. If the price of a fixed-broadband plan represents less than 5 per cent of the average income of the first decile in a country,
In conclusion, the data presented do not determine whether 100 per cent of a population within a country can afford a fixed-broadband plan,
The World bank's Povcalnet data are adjusted using ITU estimates on the average number of inhabitants per household. 40 See Footnote 10.41 For examples regarding broadband markets
ITU Tariff Policies Database 2013 (ICTEYE, http://www. itu. int/net4/ITU-D/icteye. 48 For a detailed description of the ITU ICT Regulatory Tracker, see www. itu. int/tracker. 49 The EU acquis is the body of rights
which price and market share data were available. This includes 95 economies from the developing world
which price and market share data were available for 2013. This includes 99 economies from the developing world
which price and market share data were available. This includes 96 economies from the developing world
whereas dispersion in mobile-cellular prices is of 60 per cent around the mean. 173 Measuring the Information Society Report 2014 Chapter 5. The role of big data for ICT monitoring
and for development 5. 1 Introduction One of the key challenges in measuring the information society has been the lack of upto-date and reliable data, in particular from developing countries.
and quality of those statistics and identify new data sources. In this context the emergence of big data holds great promise,
and there is an opportunity to explore their use in order to complement the existing, but often limited, ICT data.
There is no unique definition of the relatively new phenomenon known as big data. At the most basic level it is understood as being data sets whose volume,
velocity or variety is compared very high to the kinds of datasets that have traditionally been used.
The emergence of big data is linked closely to advances in ICTS. In today's hyper-connected digital world, people and things leave digital footprints in many different forms
and through ever-increasing data flows originating from, among other things, commercial transactions, private and public records that companies and governments collect
and store about their clients and citizens, usergenerated online content such as photos, videos, tweets and other messages,
Big data have great potential to help produce new and insightful information, and there is a growing debate on how businesses,
governments and citizens can maximize the benefits of big data. Although it was the private sector that first used big data to enhance efficiency and increase revenues
the practice has expanded to the global statistical community. The United nations Statistical commission (UNSC) and national statistical organizations (NSOS) are looking into ways of using big data sources to complement official statistics
and better meet their objectives for providing timely and accurate evidence for policy-making. 1 So far,
there is limited evidence as to the value added by big data in the context of monitoring of the information society,
and Chapter 5. The role of big data for ICT monitoring and for development 174 there is a need to explore its potential as a new data source.
While existing data can provide a relatively accurate picture of the spread of telecommunication networks and services
there are significant data gaps when it comes to understanding the development of the information society.
While an increasing number of countries currently collect data on the individual use of ICTS, many developing countries do not produce such information (collected through household surveys
Consequently, not enough data are available about the types of activity that the Internet is used for,
In other areas, such as education, health or public services, even fewer data are available to show developments over time
and that big data have the potential to help realize those efforts. In addition to the data produced
and held by telecommunication operators, the broader ICT sector, which includes not just telecommunication companies but also over-the-top (OTT) service providers such as Google, Twitter, Facebook, Whatsapp, Netflix, Amazon and many others, captures a wide array of behavioural
data. Together, these data sources hold great promise for ICT monitoring, and this chapter will explore the potential of today's hyper-connected digital world to expand on existing access and infrastructure indicators and move towards indicators on use, quality and equality of use.
At the same time, there is a growing debate on the role and potential of big data when it comes to providing new insights for broader social and economic development.
Big data are already being leveraged to understand socioeconomic well-being forecast unemployment and analyse societal ties. Big data from the ICT industry play a particularly important role
because they are the only stream of big data with global socioeconomic coverage. In particular, mobile telephone access is quasiubiquitous,
and ITU estimates that by the end of 2014 the number of global mobile subscriptions will be approaching 7 billion.
At the same time, almost 3 billion people 40 per cent of the world's population will be using the Internet.
and where big data hold great promise for development. However, while there are a growing number of research collaborations and promising proofof-concept studies,
To this end, this chapter will contribute to the debate on big data for development highlight advances, point to some best practices
including in regard to the production and sharing of big data for development. The chapter will first (in Section 5. 2) describe some of the current big data trends and definitions,
highlight the technological developments that have facilitated the emergence of big data, and identify the main sources
and uses of big data, including the use of big data for development and ICT monitoring. Section 5. 3 will examine the range and type of data that telecommunication companies,
in particular mobile-cellular operators, produce, and how those data are 175 Measuring the Information Society Report 2014 currently being used to track ICT developments
and improve their business. Section 5. 4 looks at the ways in which telecom big data may be used to complement official ICT statistics and assist in the provision of new evidence for a host of policy domains,
while Section 5. 5 discusses the challenges of leveraging big data for ICT monitoring and broader development, including in terms of standardization and privacy.
It will also make some recommendations for mainstreaming and fully exploiting telecom big data for monitoring and for social and economic development,
in particular with regard to the different stakeholders involved in the area of big data from the ICT industry. 5. 2 Big data sources,
trends and analytics With the origins of the term big data being shared between academic circles, industry and the media,
the term itself is amorphous, with no single definition (Ward and Barker, 2013). At the most basic level of understanding, it usually refers to large and complex datasets, Table 5. 1:
Sources of big data Sources Some examples Administrative data Electronic medical records Insurance records Tax records Commercial transactions Bank transactions (inter-bank as well as personal) Credit card
transactions Supermarket purchases Online purchases Sensors and tracking devices Road and traffic sensors Climate sensors Equipment and infrastructure sensors Mobile phones Satellite/GPS devices
store and process increasing amounts of data from different data sources. Indeed, one of the key trends fostering the emergence of big data is the massive datafication
and digitization, including of human activity, into digital breadcrumbs or footprints. In an increasingly digitized world,
big data are generated in digital form from a number of sources. They include administrative records (for example, bank or electronic medical records), commercial transactions between two entities (such as online purchases or credit card transactions), sensors and tracking devices (for example
mobile phones and GPS devices), and activities carried out by users on the Internet (including searches
and social media content)( Table 5. 1). Big data is not just about the volume of the data.
describes big data characteristics such as velocity and variety, in addition to volume (Laney, 2001). Velocity refers to the speed at
which data are generated, assessed and analysed, while the Chapter 5. The role of big data for ICT monitoring and for development 176 term variety encompasses the fact that data can exist as different media (text,
audio and video) and come in different formats (structured and unstructured). The three-Vs definition has caught on
A fourth V veracity was introduced to capture aspects relating to data quality and provenance, and the uncertainty that may exist in their analysis (IBM, 2013).
A fifth V value is included by some to acknowledge the potentially high socioeconomic value that may be generated by big data (Jones,
2012)( Figure 5. 1). Included within the scope of big data is the category of transaction-generated data (TGD),
2 also sometimes described as data exhaust or trace data. These are digital records or traces that have been generated as by-products of doing things (such as processing payments,
making a phone call and so on) that leave behind bits of information. The value of this subset of big data is that it is connected directly to human behaviour
and its accuracy is generally high. Most of the data captured by telecommunication companies can be classified as TGD.
As is often the case with technological innovation, it is the private sector that has been Figure 5. 1:
The five Vs of big data Source: ITU. at the forefront of extracting value from this data deluge.
Encouraged by promising results but also reduced budgets, the public sector is turning towards big data to improve its service delivery and increase operational efficiency.
In addition, there are uses for big data in broader development and monitoring, and there is an increasing focus on big data's role in producing timely (even real-time) information,
as well as new insights that can be used to drive social and economic well-being. Big data uses by the private and public sectors Marketing professionals,
whose constant aim is to understand their customers, are now increasingly shifting from conventional methods, such as surveys, to the extraction of customer preferences from the analysis of big data.
Walmart, the world's biggest retailer, has been one of the largest and earliest users of big data.
In 2004, it discovered that the snack food known as Pop Tarts was purchased heavily by United states citizens preparing for serious weather events such as hurricanes.
The correlation analysis revealed a behaviour associated with VOLUME VARIETY VERACITY Vast amounts of data generated through large-scale datafication and digitization of information Different types and forms of data
including large amounts of unstructured data Level of quality, accuracy and uncertainty of data and data sources VALUE Potential of big data for socioeconomic development VELOCITY Speed at
which data are generated and analyzed 177 Measuring the Information Society Report 2014 a specific condition that then led Walmart to improve its production chain in this case,
by increasing the supply of Pop Tarts to areas likely to be affected by a disaster.
Walmart has also made use of predictive analytics, which uses personal information and purchasing patterns to extrapolate to a likely future behaviour,
large-scale automated correlation analysis and predictive analytics are two of the key techniques that have helped unleash the value of big data.
Nor is the private sector's use of big data techniques restricted solely to market research. Companies and whole industries (healthcare, energy and utilities, transport, etc.
Firms that are highly proficient in their use of data-driven decision making have been found to have productivity levels up to 6 per cent higher than firms making minimal to no use of data for decision-making (Brynjolfsson, Hitt and Kim, 2011.
How big data saves energy Vestas Wind Systems improves turbine performance Vestas, a global energy company dedicated to wind energy,
with installations in over 70 countries, has used big data platforms to improve the modelling of wind energy production
By using big data techniques based on a large set of factors and an extended set of structured and unstructured data
Big data have enabled the creation of a new information environment and allowed the company to manage
and location data in ways that were previously not possible. These new insights have led to improved decisions relating not only to wind turbine placement and operation,
while increasing the accuracy of the customer's return-on-investment estimates. Source: ITU, based on IBM (2012.
Telecom operators also use big data techniques to understand and control churn, optimize their management of customer relations
These fundamental shifts in data exploitation to generate new socioeconomic value, coupled with the simultaneous emergence of new rich data sources that can potentially be linked together
and analysed with ease, have sparked also the interest of governments, researchers and development agencies. Encouraged by the potential of big data to produce new insights and slimmer budgets
governments (at all levels) are now looking to exploit big data and increase the application of data analytics to a range of activities, including monitoring and improvement of tax compliance and revenues, crime detection and prediction,
and improvement of public service delivery (Giles, 2012; Lazer et al. 2009). ) To this end, governments, in addition to the data they collect
and generate themselves, Chapter 5. The role of big data for ICT monitoring and for development 178 complement their official statistics by leveraging data from new sources, including crowd-sourced data generated by the public.
In the United states, for example, Boston City hall released the mobile app Street Bump, which uses a phone's accelerometer to detect potholes
while the app user is driving around Boston and notifies City hall. 3 Some of the richest data sources for enabling governments
and development agencies to improve service delivery are actually external. Such external data include those captured
and/or collected by the private sector, as well as the digital breadcrumbs left behind by citizens as they go about their daily lives.
According to a recently published White house report, United states government agencies can make use of public and private databases
and big data analytics to improve public administration, from land management to the administration of benefits. The Department of the treasury has set up a Do Not Pay portal,
which links various databases and identifies ineligible recipients to avoid wrong payments and reduce waste and fraud4 (The White house, 2014).
Big data for development and ICT monitoring One of the richest sources of big data is captured the data by the use of ICTS.
This broadly includes data captured directly by telecommunication operators as well as by Internet companies and by content providers such as Google, Facebook, Twitter, etc.
Big data from the ICT services industry are already helping to produce large-scale development insights of relevance to public policy.
Collectively, they can provide rich and potentially real-time insights to a host of policy domains.
It should be noted that in some countries and regions the use of big data including big data from the ICT industry,
is subject to national regulation. In the EU, for example, a number of directives require data producers to obtain users'consent before gathering any of their personal data. 5 One of the best-known examples of leveraging the online population's digital breadcrumbs for development purposes is Google Flu Trends (GFT.
Following its launch in 2008, GFT was remarkably accurate in tracking the spread of influenza in the United states,
GFT was held up as an outstanding example of big data in action and of the great potential of big data for broader development and monitoring (Mayer-Schönberger and Cukier, 2013;
) This proved to be so successful that it spawned similar efforts focusing on the use of search-engine data to understand dengue fever outbreaks,
The Internet has also been a rich source of big data beyond the realm of user search terms.
Online job-posting data are being used to supplement traditional labour statistics in the United States7 and other countries.
In another effort, an academic project at MIT known as the Billion Prices Project collects high-frequency price data from hundreds of online retailers. 8 The data are used then by researchers to understand a whole host of macroeconomic
a UN initiative to use big data for sustainable development and humanitarian action, has been mining Twitter data from Indonesia (where Twitter usage is high) 9 to understand food price crises.
Global Pulse was able to identify a consistent pattern among specific food-related tweets and the daily food price index.
In fact, it was able to use predictive analytics on the Twitter data to forecast the consumer price index several weeks in advance (Byrne, 2013.
UN Global Pulse is also using Twitter data to understand and compare the relevance of different development topics among countries (Box 5. 2). Box 5. 2:
UN Global Pulse and the Millennium Campaign are using big data and visual analytics to identify the most pressing development topics that people around the world are concerned about
the ICT sector is itself using the Internet as a source of big data for monitoring purposes.
Regulators and others are now using the Internet to crowdsource quality of service (Qos) data on broadband quality.
and address coverage and quality issues in different areas. 10-200'000 400'000 600'000 800'000 1'000'000 1'200'000
and oceans Phone and Internet access Equality between men and women Chapter 5. The role of big data for ICT monitoring and for development 180 Mobile data Despite the rapid growth in Internet access,
Depending on the source of Internet data, results may also be biased more or less. A 2013 study into the characteristics and behaviour of Facebook users, for example, revealed that
On the one hand, for example, the American Facebook user's relationship status of married on Facebook is very similar to real life (census) data on the average age
non-Internet-related mobilenetwork big data seems to have the widest socioeconomic coverage in the near term,
Mobile data are already being utilized for research and policy-making not only in developed but also in developing economies.
and forecast economic developments. 13 Data are also being used to improve responsiveness in the event of natural disasters or disease outbreaks.
Call records have also been merged with epidemiological data to understand the spread of malaria in Kenya (Pindolia et al.
Mobile network big data have been utilized to great effect in the area of transportation, helping to measure and model people's movements (even in real time) and understand traffic flows (Wu et al.,
) It is evident from the examples given that big data from the ICT sector, and especially those available to telecommunication operators, have wide applicability for informing multiple public policy domains.
Leveraging such data to complement official statistics and facilitate broader development will enable governments as well as development agencies to better serve their citizens and beneficiaries.
Less use has thus far been made of telecommunication big data with a view to understanding its potential for producing additional information and statistics on the information society.
it is first important to better understand the type of data that can be made available. 5. 3 Telecommunication data
and their potential for big data analytics Fixed and mobile telecommunication network operators, including Internet service providers (ISPS), are an important source of data and for the purpose of this chapter, all forms of telecommunication big data (either volume,
velocity or variety) are being considered. Most telecommunication data can be considered as TGD, 14 that is, the result of an action undertaken (such as making a call,
sending an SMS, accessing the Internet or recharging a prepaid card). Since the service with the widest coverage and greatest uptake and popularity is the mobilecellular service,
data from mobile operators have the greatest potential to produce representative results and reveal developmental insights on the population,
including in developing countries and, increasingly, low-income areas. Not surprisingly, the big data for development initiatives (outlined in Section 2. 2) have drawn mainly on mobile network big data rather than on those from fixed-telephone operators or ISPS.
Figure 5. 2 illustrates some of the similarities and differences in the type of information that mobile network operators,
Telecommunication data The mobile telecommunication data that operators possess can be classified into different types, depending on the nature of the information they produce.
They include traffic data, service access detail records, location and movement data, device characteristics, customer details and tariff data.
For a more detailed overview of these types of data, see Chapter 5 Annex. To collect traffic data,
operators use a range of metrics to understand and manage the traffic flowing through their networks,
including the measurement of Internet data volumes, call, SMS and MMS volumes, and value-added service (VAS) volumes.
Internet service providers can also use deep packet inspection (DPI), 15 which is a special process for scanning data packages transiting the network.
Service access detail records, including call detail records (CDRS), are collected by operators whenever clients use a service.
These data are potentially also very useful for building a rich profile of customers, as outlined in this section.
The degree of accuracy of this information depends on a number of factors, including the network used and device generation,
passive and active positioning data, with the latter providing more detailed and precise location information. Since mobile user devices used to access mobile telecommunication services come with an international mobile station equipment identity (IMEI) number,
which in turn can provide some insight as to that Chapter 5. The role of big data for ICT monitoring and for development 182figure 5. 2:
An overview of telecom network data Source: ITU, adapted from Naef et al. 2014). ) Traffic data Fixed operator Mobile operator ISP Data volume Call volume SMS/MMS volume Erlang DPI data Timestamp of use Contact
network Duration of use Applicable charges Handset type Technology utilized (2g, 3g, DSL/ADSL, etc.
Billing address Passive positioning data (e g. cell ID) IMEI Active positioning data (e g. cell triangulation, GPS) MAC address Customer demographics (e g. age, gender
, national ID card number) Payment history (postpaid) Billing address Recharge history (prepaid) Service order history Tariff sheet Service access detail records Location data Device
characteristics Customer details Tariff data Transaction generated data Stored warehouse data 183 Measuring the Information Society Report 2014 subscriber's purchasing power (see below for more details.
Finally, operators maintain tariff data in the form of billing records for their current and past services,
How mobile operators currently use data to track service uptake, business performance and revenues Operators use their TGD to monitor the uptake
On the basis of the detailed service-usage data collected, telecommunication operators can produce a range of detailed indicators relating to service consumption.
data upload volumes, data download volumes, level of use of different VAS, and level of use of different OTT services.
These data can be reported as averages (over time or for different categories of user), as well as at various levels of aggregation (again over time or for different categories of user).
Finally, service consumption data are used to produce revenue data and projections at various levels of disaggregation or aggregation.
often associate revenue data with resource allocation to ensure that Qos at the base stations used by their premium customers is maintained at the highest possible level.
but also consultancy firms and others, use aggregated revenue data to track and benchmark countries'ICT developments, monitor the evolution of the information society
The telecom industry's use of big data Telecommunication companies are actively seeking to intensify their use of big data analytics
For operators, big data open up opportunities for better understanding of their customers, which in turn leads to improved sales and marketing opportunities.
big data can help optimize network operations and create new revenue streams and business lines, for example when selling data.
Customer profiling Telecom operators capture a range of behavioural data about their customers. Chapter 5. The role of big data for ICT monitoring and for development 184 Customer profiles include details about customers'mobility patterns, social networks and consumption preferences.
Collectively, these digital breadcrumbs enable operators to profile and segment their customers based on a variety of metrics (Figure 5. 3). Depending on the country or region,
there may be different privacy and data regulations governing the manner in which operators may keep and/or use such data.
This being the case the extent to which behavioural profiling is used by operators may vary greatly.
Customer interests: these can be captured, or in some cases inferred, on the basis of usage levels (time spent and/or volume) for different VAS and OTT services.
Big data, on the other hand, can help to enhance that classification by enabling analysis of the levels of consumption of different services,
Big data techniques can help operators understand churn better by enabling them to model the likelihood of customers leaving the network
Customer profiling using telecom big data Source: ITU. CUSTOMER INTERESTS SOCIOECONOMIC CLASS LEVEL OF INFLUENCE OF CUSTOMERS LIKELIHOOD OF CHURN MOBILITY PROFILE 185 Measuring the Information Society Report 2014 in competitor networks.
In the Republic of korea, for example, SK Planet, a subsidiary of SK TELECOM, uses big data to help its parent company to cut churn
and has used data mining to achieve a fourfold improvement in churn forecasting. The operator found that customers planning to quit their current package tend to use specific search phrases, such as data plan or operator benefits, at least three to seven days before taking action.
When operators suspect that customers may be looking elsewhere, they may try to keep them by providing them with a tailored offer. 18 Network planning
operators may also seek to monetize the data they hold. The simplest way of doing this is to sell (anonymized) data to third parties.
The customer insights obtained through the analysis of usage data can also help create new business lines,
either through innovation (e g. new types of VAS) or by partnering with other businesses, including credit-scoring and related financial services.
One example is based the US big data startup Cignifi, 19 which obtains data from mobile operators
and financial institutions to build credit profiles and evaluate customer creditworthiness (see Box 5. 8). Cross-promotions with brick-andmortar businesses are a potentially high-growth area in
which the detailed mobility profiles available to operators are leveraged. 5. 4 Big data from mobile telecommunications for development and for better monitoring In 2013, the United nations High-level Panel of Eminent Persons on the Post-2015
Development Agenda called for a data revolution that draws on existing and new sources of data for the post-2015 development agenda (United nations, 2013).
In March 2014, the forty-fifth session of UNSC, the highest decision-making body for international statistical activities, presented a report on big data and modernization of Chapter 5. The role of big data for ICT
and proposed the creation of a big data working group at the global level (UNSC, 2013). 20 Current uses of big data to complement official statistics are still exploratory,
but there is a growing interest in this topic, as evidenced by the numerous initiatives being pursued by the United nations, as well as by others,
There are many big data sources that can be used to monitor and assess development results. In a world where mobile telephony is increasingly ubiquitous,
it is not surprising that mobile telecommunication big data have unique potential as a new data source,
even among the poorest and most deprived, making them particularly valuable by comparison with other types of telecommunication data.
Indeed, when referring to the data revolution, the United nations High-level Panel cited the example of mobile technology
This section will present some of the existing (and growing) evidence for the role of mobile big data in achieving development goals in various policy areas,
In addition to their use for development, telecommunication big data have potential as a source to enable monitoring of the information society,
however, ITU is exploring the potential of big data to complement its existing, and often limited, set of ICT statistics.
This section presents a first attempt to help identify some of the areas in which mobile telecommunication big data could complement existing ICT indicators to provide a more complete
Mobile phone big data for development Mobile data offer a view of an individual's behaviour in a low-cost, high-resolution, realtime manner.
In addition to the fact that these data are uniquely detailed and tractable, the information captured cannot easily be derived from other sources on such a scale.
The fact that the format of the data is relatively similar across different operators and countries creates a huge potential for the global scaling of any application found to have significant benefits.
Box 5. 4 illustrates the potential of mobile data for development in a number of different areas.
Big data for disaster management and syndromic surveillance21 Mobility data collected immediately after a disaster can in many cases help emergency responders to locate affected populations
One application of such mobility data is for syndromic surveillance, especially to model the spread of vector-borne22 and 187 Measuring the Information Society Report 2014 Box 5. 4:
Using mobile data for development A recently published report (Cartesian, 2014) explores the potential of mobile data for development.
It points to three primary types of analysis-ex-post evaluation, real-time measurement, and future predictions and planning-in a number of areas (including health, agriculture and economic development),
through mobile data Migration monitoring Text analysis economic downturn prediction Text analysis commodity fluctuation prediction Assessment of mobility restrictions Disease containment targeting Migratory p
opulation tracking Predicting outbreak spread Mobile data to track food assistance delivery Geo-targeted links between Ag suppliers/purchasers Pests,
bad harvest Ag yield/shock predictions Campaign effectiveness Social network delineated market areas Predictive algorithms to anticipate prod. churn Social network targeted marketing Post-disaster refugee reunification
Pioneering research in Kenya combined passive mobile positioning data with malaria prevalence data to identify the source and spread of infections (Wesolowski et al.
) Similar work in Haiti showed how mobile phone data was used to track the spread of cholera after the 2010 earthquake (Bengtsson et al.
2011, see Box 5. 5). The integration of mobility data from mobile networks with geographic information frameworks,
Chapter 5. The role of big data for ICT monitoring and for development 188 Big data for better transportation planning A data-centric approach to transportation management is already a reality in many developed economies.
Transportation systems are being fed with sensor data from a multitude of sources such as loop detectors axle counters, parking occupancy monitors, CCTV, integrated public transport card readers and GPS data derived not only from phones but also from public transport and private vehicles (Amini, Bouillet, Calabrese, Gasparini
and Verscheure, 2011. One advantage of mobile networks is that even the least developed mobile network infrastructure generates passive positioning data,
which, despite its limited spatial accuracy (cell ID), has great potential for transportation planning. For example, IBM researchers used CDR data from mobile operator Orange to map out citizens'travel routes in Abidjan, the largest city in Côte d'ivoire,
and show how data-driven insights could be used to improve the planning and management of transportation services, thereby reducing congestion (Berlingerio et al.,
2013). ) By simply extending one bus route and adding four new ones, overall travel time was reduced by ten per cent.
Passive mobile positioning data has also been used for transportation planning and management in Estonia (Ahas and Mark, 2005),
and has provided reliable results in Sri lanka (Lokanathan et al.,2014, see Box 5. 6). Box 5. 5:
How mobile network data can track population displacements an example from the 2010 Haiti earthquake The Figure below shows the number of people estimated to have been in Port-au-prince (Pap) on the day of the 2010 Haiti earthquake,
This map was produced on the basis of mobile network data to show the potential of big data in tracking population movements.
Leveraging mobile network data for transportation and urban planning in Sri lanka Very similar findings between the results of an official household survey assessing mobility patterns (right-hand map) with the results of a big data analysis using mobile-phone
data (left-hand map) underscore the merits of big data. The image on the left, based on mobile-phone data, depicts the relative population density in Colombo city
and its surrounding regions at 1300 hours on a weekday in 2013, compared to midnight the previous day.
Mobile big data (left) versus official survey data (right) Both passive and active positioning data are used to analyse traffic conditions, particularly in urban areas with higher base-station density.
Active positioning data (especially GPS) produce higher precision in location data and are therefore the most useful.
Operators may offer such specialized services (based on passive or active location data) either directly, or by providing data to third parties.
Mobile network data are less expensive, are in real time and are less time-consuming to produce than survey data,
particularly in urban and peri-urban areas where base-station density tends to be high. In another example
the analysis of mobility flows between two Spanish cities derived from three different data sources mobilephone data,
geolocated Twitter messages and the census showed very similar results, and although the representativeness of the Twitter geolocated data was lower than the (real-time) mobile-phone and census data,
the degrees of consistency between the population density profiles and mobility patterns detected by means of the three datasets were significant (Lenormand et al, 2014).
Chapter 5. The role of big data for ICT monitoring and for development 190 Big data for socioeconomic analysis Data from mobile operators can provide insights in the areas of economic development and socioeconomic status
Big data techniques can therefore complement official statistics in the intervals between official surveys, which are usually relatively expensive
In many cases, insights derived from big data sources may help to fill in the gaps, rather than replace official surveys.
It should also be noted that mobile network big data are one of the few big data sources (and often the only one) in developing economies that contain behavioural information on low-income population groups Frias-Martinez et al.
2012) developed a mathematical model to map human mobility variables derived from mobile network data to people's socioeconomic and income levels.
and income-related data derived from official household surveys, and the results showed that populations with higher socioeconomic levels are associated more strongly with larger mobility ranges than populations from lower socioeconomic levels.
the study suggested that it was possible to create a model to estimate income levels based on data from mobile network operators.
used two types of mobile network data, namely subscriber communication data and airtime credit purchase records,
Poverty mapping in Côte d'ivoire using mobile network data In Côte d'ivoire, researchers used mobile network data (specifically communication patterns,
One of the challenges has to do with operator sensitivity regarding revenue data and the difficulty this poses for outside parties wishing to obtain such data.
The use of mobile-operator TGD can also foster financial inclusion by facilitating the provision of credit to the unbanked.
A compelling example of how mobile big data can be used for the unbanked is Cignifi, a big data startup that uses the mobile phone records of poor people to assess their creditworthiness
when they apply for a loan (Box 5. 8). Big data for understanding societal structures Social-network studies relying on self-reporting relational data typically involve both a limited number of people
and limited number of time points (usually one). As a result, social-network analysis has generally been confined to the examination of small population groups through a small number of snapshots of interaction patterns.
By examining social communication patterns based on telecommunication data, it has become possible to obtain insights into societal structures on a scale that was previously unavailable.
Using mobile-phone data to track the creditworthiness of the unbanked Cignifi, a big data startup, has developed an analytic platform to provide credit
and marketing scores for consumers, based on their mobile-phone data. The Cignifi business model is founded on the idea that Mobile phone usage is not random it is highly predictive of an individual consumer's lifestyle and risk.
Based on the behavioural analysis of each mobile-phone user phone calls, text messages, data usage and, extrapolating from these,
lifestyles the company identifies patterns and uses them to generate individual credit risk profiles. This information could help many of the world's unbanked to have access to insurance
and respond to changes in customer activity as the data are refreshed, usually every two weeks. In addition to updating a person's creditworthiness,
and verified the findings from the model against historical lending data from approximately 40 000 borrowers using the mobile operator Oi's lending business,
Chapter 5. The role of big data for ICT monitoring and for development 192 have been used to study the geographic dispersion
) However, telecommunication data are also revolutionizing the study of societal structures at the micro level.
2009) show that it is possible to assess friendship using data from mobile network operators, and that the accuracy is compared high
when with self-reported data. Leveraging these behavioural signatures to obtain an accurate characterization of relationships in the absence of survey data could also enable the quantification
and prediction of macro and micro social-network structures that have thus far been unobservable. Big data to monitor the information society There is a case to be made for analysing data captured by telecommunication operators in the interests of improving the current range of indicators used for monitoring the information society.
An internationally-accepted and widely-adopted list of indicators is the core list of ICT indicators developed by the Partnership on Measuring ICT for Development,
a multistakeholder initiative launched in 2004.24 This list includes, among others, the key-infrastructure, access and individual-use indicators that ITU collects
Some of these indicators are amenable for augmentation using big data analytics. 25 The core indicators on ICT infrastructure
One of the main issues with mobile-cellular and mobile-broadband subscription data is that they do not refer to unique subscriptions
making it important for operators to monitor the time during which a SIM CARD remains inactive.
whereas the number of connections per 100 population far exceeded 100 per cent. 26 Survey-based data,
for example on Internet users and mobile-phone users, do not entail the same issues as subscription data.
Survey-based data can also be broken down by individual characteristics, including gender, age, educational level and occupation,
which substantially increase the data's added value. However, the main challenge with these data is that they are not widely available (in particular,
many developing countries do not yet collect data on individual use of, or household access to, ICTS), are relatively expensive to produce,
and are much less timely than subscription data (often with a time lag of one year.
Consequently, data on users of the information society and the types of online service they consume are limited,
and in many 193 Measuring the Information Society Report 2014 cases outdated. Against this background, mobile networks and mobile big data could be used to identify alternative,
less costly and faster ways of carrying out representative surveys (Box 5. 9). Given the shortcomings of existing administrative data from operators and survey data collected by NSOS,
it is particularly interesting to assess some of the ways in which big data can be used to overcome the shortcomings of existing key ICT indicators
and to provide additional insights into ICT access and use, user behaviour, activities and also the individual user.
Big data could help in obtaining more granular information in several areas, and big data techniques could be applied to existing data to produce new insights.
In particular, operators'big data could produce information in the following areas: Individual subscriber characteristics: Additional categorization across both time and space are possible for subscription indicators,
and big data could provide additional information on gender, socioeconomic status and user location. Information on gender or age, for example, could be derived from customer registration information (notwithstanding a number of challenges and privacy issues,
as discussed later in Section 5. 5). The socioeconomic status of the person linked to a subscription could be derived from big data techniques applied to users'consumption information,
as well as other data contained in customer registration information. In addition, the analysis of customers'mobility patterns will often allow for an understanding of important locations (work
and home being the two most important) and of the use of mobile services in rural versus urban areas.
All subscription data could provide information as to location. In the case of fixed-telephone and fixed-broadband subscriptions,
or to link those characteristics to other (administrative) databases in order to Box 5. 9: Using mobile big data
and mobile networks for implementing surveys An important measurement for assessing the development of the information society is the extent to
Given the need for continued recourse to surveys for collecting the corresponding data, and the declining response rates where traditional surveys are concerned (Groves, 2011),
mobile operators could develop platforms to facilitate the collection of survey data. This could include targeting a wide variety of respondents covering the full spectrum of appropriate demographic profiles,
followed by a process of extrapolation using big data analytics. In 2011 for example, UN Global Pulse partnered with Jana, 27 a mobile-technology company,
Chapter 5. The role of big data for ICT monitoring and for development 194 create new information.
Particularly rich possibilities exist where data from mobile-cellular and mobile-broadband subscriptions are concerned, since they are linked to mobility profiles.
The indicators for such subscriptions could be broken further down to understand the utilization of services including voice, data and VAS over time,
In addition, mobile-operator data could be combined with customer information from popular online services, such as Facebook, Google or other, local (financial, social etc.
This could be done by using probabilistic analyses to match the profiles developed using data from online services with customer profiles generated from analyses of mobile-operator data.
By applying big data techniques to survey data and administrative data from operators, new insights could be derived, in particular, in respect of the following:
Big data techniques could help extrapolate the actual number of unique mobile subscribers or users, rather than just subscriptions,
and by taking into account usage patterns or data from popular Internet companies such as Google or Facebook.
By linking data collected from different sources and combining subscription data and usage patterns, a correlation algorithm could be developed to reverse engineer approximate values for these indicators,
and Virseda (2012) on estimating socioeconomic variables using mobile-phone usage data, as described in greater detail at the beginning of this section.
big data methods only complement existing surveys rather than replacing them completely (see Section 5. 5 for a further discussion of this).
In sum, relatively simple big data techniques can help analyse and provide complementary information on existing ICT data,
and provide new insights into the measurement of the information society. This includes information on the use of different services and applications, intensity, frequency,
Given multiple SIM usage and the fact that users will in many cases be using ICT services from more than one operator or device
Such techniques will often include combining data from surveys with big data to build new correlation and predictive analytic techniques.
and for other big data for development projects, big data analysis cannot replace survey data, which is needed to build
and test correlations and to validate big data results. While the opportunities discussed above present what is analytically possible,
data access and privacy considerations are nuanced complex and, and therefore place constraints on what is practically feasible or advisable.
and the way forward Attempting to extract value from an exponentially growing data deluge of varying structure
The most pressing concerns are associated those with the standardization and interoperability of big data analytics as well as with privacy and security.
Addressing such privacy and other concerns with respect to data sharing and use is critical, and it is important for big data producers
and users to collaborate closely in that regard. This includes raising awareness about the importance and potential of producing new insights,
and the establishment of public-private partnerships to exploit fully the potential of big data for development.
Data curation, standardization and continuity Data curation and data preparation help to structure, archive, document
and preserve data in a framework that will facilitate human understanding and decision-making. Traditional curation approaches do not scale with big data
and require automation, especially since 85 per cent of big data are estimated to be unstructured (Techamerica Foundation, 2012.
Dealing with large heterogeneous data sets calls for algorithms that can understand the data shape
while also providing analysts with some understanding of what the curation is doing to the data (Weber, Palmer and Chao, 2012).
Telecom network operators themselves have to contend with interoperability issues arising from the different systems (often from different vendors) they employ.
It is not uncommon for operators to write customized mediation software to overcome potential inter-comparability issues among data from different systems.
when one has to take account of secondary third-party users that may seek to leverage the data.
The framework used by an NSO to organize data would be different from that used by a network engineer or a marketing or business intelligence specialist.
Naturally, telecom network operators have curated their data based on their needs. To be able to use telecom big data for development and monitoring
and to guarantee its continuity, the creation of a semantic framework would require greater consensus among the many diverse stakeholders involved (telecom operators, network equipment manufacturers, system developers, developmental practitioners and researchers, NSOS, etc.).
Chapter 5. The role of big data for ICT monitoring and for development 196 Accessing and storing data,
and data philanthropy Big data for development is still in its nascent stages and, as such, comes with its share of challenges, not least
of which is obtaining access to what is essentially private data. Private corporations would hesitate to share information on their clients
and their business processes in case such sharing is illegal, precipitates a loss of user confidence and/or accidentally reveals competitive business processes.
Until holders of big data become more comfortable about their release it is going to be difficult for third-party research entities to gain access.
Researchers (mainly from developed countries, with some exceptions such as LIRNEASIA) have succeeded recently in obtaining mobile network big data,
All the parties to such agreements have to address the necessary parameters as to how data are to be used,
both researchers and operators face costs arising from the technical challenges associated with extraction of the data,
Some mobile operators are taking tepid steps towards sharing data more publicly. Orange, for example, hosted a Data for Development Challenge,
releasing an aggregated anonymized mobile dataset from Côte d'ivoire to researchers and convening a conference at MIT in early 2013,
this time using Orange data from Senegal, is planned for 2015.30 In 2014, Telecom italia initiated a similar challenge,
making data from the territories of Milan and the Autonomous Province of Trento available to researchers for analysis. 31 It has gone,
it partnered with other data providers to curate and release additional big datasets containing weather, public and private transport, energy, event and social network data.
In both the Orange and Telecom italia cases, researchers had to go through an approval process in order to gain access.
Organizations such as UN Global Pulse are seeking to popularize the concept of data philanthropy, aimed at systematizing the regular and safe sharing of data by building on the precedents being created by the ad hoc activities outlined earlier.
Such efforts by UN Global Pulse as well as by other organizations such as LIRNEASIA, that seek to bring different stakeholders to the same table,
remain critical to the efforts being made to open up private-data stores in order to obtain actionable development insights.
and sharing of such data are to become a reality. Cross-sector and crossdomain collaboration would benefit greatly from facilitators
or intermediaries capable of addressing issues related to standardization and data-curation practices when pooling data from multiple sources.
and privacy burdens faced by data providers, effectively acting as a gatekeeper to ensure that data are used transparently
and in a way that contributes to overall scientific knowledge generation, while ensuring that any safeguards that may be applicable in respect of private information are applied.
which decided to share all of its clinical trial data. To facilitate the process, they hired Yale university's Open Data Access (YODA) Project 197 Measuring the Information Society Report 2014 to act as gatekeepers (Krumholz, 2014).
YODA undertakes the necessary scientific review of any proposals (from scientists around the world) to make use of the data
and ensures that necessary privacy and data usage guidelines are followed. 32 The question remains as to who is placed best to act as gatekeepers
and standard-bearers when it comes to telecom network big data. Some have argued that NSOS are placed well to ensure that best practices are followed in the collection and representation of big data,
and to provide a stamp of trust for potential third-party data seekers. Telecom operators, for their part
are regulated mostly by sectorspecific regulators who can also have purview and dictate terms governing the privacy
and data-reporting responsibilities of operators. Ultimately, however, the decision as to who takes on the gatekeeper
and standardization function requires the confluence of multiple actors. It is here that organizations such as ITU,
UN Global Pulse and others have a greater role to play in building an institutional model for data sharing and collaboration, in consultation with all stakeholders.
The sharing (subject to appropriate privacy protocols) of privately held data such as mobilephone records can be mutually beneficial to both government and private sector.
Such a collaborative early-warning and earlyaction system shows how data sharing could be considered a business risk mitigation strategy for operators in emerging markets.
it should be noted that the emergence of big data is linked closely to advances in the ICT sphere,
including the falling cost of data storage. Depending on the data volume, storage can still be costly,
especially where privacy considerations preempt the use of specialized third-party cloud-based services. But as storage prices continue to fall,
Privacy and security As social scientists look towards private data sources, privacy and security concerns become paramount.
all stakeholders must see tangible benefits from such data sharing. These stakeholders include not just the public and private sectors
but also, significantly, the general public, who in many cases are the primary producers of such data through their activities.
It is also the public that must ultimately decide on how the data they produce may be used.
and the private and public sectors in today's new data ecosystem. 33 Discussions must address the individual's privacy expectations,
OECD, for example, defines personal data as any information relating to an identified or identifiable individual (data subject)( OECD, 2013.
and consent practised by most companies to inform users of what data are Chapter 5. The role of big data for ICT monitoring
however, that in a big data world the inform and consent approach is woefully inadequate and impractical,
Secondly, in the big data paradigm, the greatest potential often lies in secondary uses, which may well manifest long after the data was collected originally.
It is thus impractical for companies to have a priori knowledge of all the potential uses and to seek permission from the user every time such a new use is found.
Given the volumes of data that individuals are now generating, companies would find themselves struggling to maintain meaningful control.
when data from one source is combined with data from other sources to reveal/infer new data and insights.
This blurs the lines between personal and nonpersonal information, allowing seemingly nonpersonal data to be linked to an actual individual (Ohm, 2010.
Digitized behavioural data crumbs may in fact greatly diminish personal privacy. The use of DPI, for example can technically reveal all of a user's online activity.
Data anonymization35 (i e. methods designed to strip data of personal information), employed by computational social scientists,
the authors point out that the data could in fact be de-anonymized completely by cross-referencing them with other data sources.
SIM resellers may pre-register the SIMS they sell under their own name, and SIMS that are registered by one family member may be used by other members of the same family.
Sri lankan operators, for example, see a great mismatch between the person registering a subscription and the person using it.
These safeguards must also ensure that data are kept secure. Data breaches undermine consumer confidence and hinder efforts to exploit big data for the greater social good.
Encryption, virtual private networks (VPNS), firewalls, threat monitoring and auditing are some potential technical solutions that are employed currently,
199 Measuring the Information Society Report 2014 but they need to be mainstreamed (Adolph, 2013).
but it will be some time before a consensus is achieved on the most appropriate method (s). In response to the growing trend to unlock socioeconomic value from the rising tide of big data,
with the principles arising from a new approach that shifts governance from the data per se to its use;
Given the complexity of the questions related to privacy and data protection in a big data world, the danger is that these questions may take too long to resolve
and further delay the potential use of big data for broader development. Hence, a balanced risk-based approach may be required in the context of
i e. the use of telecom big data for monitoring and development. This does still require the confluence of appropriate stakeholders.
research into the use of big data for development can be sandboxed, with appropriate privacy protections imposed on researchers,
or GIGO for short, is a computer science concept that refers to the fact that the veracity of the output of any logical process depends on the veracity of the input data.
In the big data paradigm, it is easy to overlook that concept, given the expectation that when dealing with vast volumes of (often unstructured) data from a multitude of sources,
messiness is to be expected. As Mayer-Schönberger and Cukier (2013) note, What we lose in accuracy at the micro level we gain in insight at the macro level.
This common conception can often be misleading. Data quality and their provenance do matter, and the question is important in establishing the generalizability of the big data findings.
Data provenance and data cleaning Understanding data provenance involves tracing the pathways taken by data from the originating source through all the processes that may have mutated,
replicated or combined the data that feed into the big data analyses. This is no simple feat.
Nor given the varied sources of data that are utilized, is it always as feasible as the scientific community would wish.
However, at the very least it is important to understand some aspects of the origin of data.
For example, the fact that some mobile network operators choose to include the complete routing of a call that has been forwarded means that there may be multiple records in the CDRS for the same call.
If that is not taken into consideration, the subsequent social network analysis could contain errors (overstating or understating tie strength, for example).
While it may not be possible to establish data provenance as envisaged by scientists, it is at the very least important to understand the underlying processes that may have created the data.
Data cleaning remains a key part of the process to ensure data quality. It is important to verify that the quantitative and qualitative
(i e. categorical) variables have been recorded as expected. In a subsequent step, outliers must be removed, using decision-tree algorithms or other techniques.
However, data cleaning itself is a subjective process (for example, one has to decide which variables to consider)
Are the data representative? Related to the question of data provenance is the issue of understanding the underlying Chapter 5. The role of big data for ICT monitoring
and for development 200 population whose behaviour has been captured. The large data sizes may make the sampling rate irrelevant,
but they do not necessarily make it representative. Not everyone uses Twitter, Facebook or Google. For example, ITU estimates suggest that 40 per cent of the world's population uses the Internet.
when considering how telecommunication data may be used for monitoring and development. While the potential benefits to be gained from leveraging mobile network operator data for monitoring
and development purposes hinges on the large coverage, close to the actual population size, it is nevertheless not the whole population.
Questions such as the extent of coverage of the poor or the levels of gender representation among telecom users, are all valid considerations.
While the registration information might provide answers, the reality is that the demographic information on telecom subscribers,
Hence, the big in big data does not automatically mean that issues such as measurement bias and methodology,
internal and external data validity and data interdependencies can be ignored. These are fundamental issues not just for small data but also for big data (Boyd and Crawford, 2012.
Behavioural change Digitized online behaviour can be subject to self-censorship and the creation of multiple personas,
so studying people's data exhaust may not always give us insights into real-world dynamics. This may be less of an issue with TGD,
where in essence the data artefact is itself a by-product of another activity. Telecom network big data,
which mostly fall under this category, may be less susceptible to self-censorship and persona development, but the possibility of these phenomena cannot be ruled out.
In a way, big data analyses of behavioural data are subject to a form of the Heisenberg uncertainty principle, whereby as soon as the basic process of an analysis is known,
and understanding the real-world context therefore remains important when considering big data analyses for monitoring purposes.
when CDR data from Rwanda showed low mobility in the wake of flooding, he theorized that this was due to an outbreak of cholera.
when it comes to the generalizability of telecomdata analyses based on big data. For example, prior research had established a power-law distribution between the frequency of airtime recharges
Causation versus correlation It is easy to confuse correlation with causation in the big data paradigm
Big data draws many of its techniques from machine learning, which is primarily about correlation and predictions. 40 Big data are by their very nature observational
and can measure only correlation and not causality. Supporters of big data have predicted the end of theory and hypothesis-testing, with correlation trumping causality as the most relevant method (Anderson, 2008;
Mayer-Schönberger and Cukier, 2013. However, such predictions may be premature. The behavioural economist Sendhil Mullainathan notes that inductive science
(i e. the algorithmic mining of big data sources) will not drown out traditional deductive science (i e. hypothesis testing), even in a big data paradigm.
Among the three Vs in the traditional big data definition, volume and variety produce countervailing forces.
More volume makes big data induction techniques easier and more effective, while more variety makes them harder and less effective.
(i e. deductive science) Chapter 5. The role of big data for ICT monitoring and for development 202 rather than merely predicting it (Mullainathan,
) Causal modelling is possible in a big data paradigm by conducting experiments. Telecom network operators themselves use such techniques when rolling out new services or, for that matter, for pricing purposes.
The role of traditional small data in verification The documented failures of GFT also point to the importance of traditional statistics as corroborating evidence.
the true value of GFT is realized only through its pairing with small data, in this case the statistics collected by the Centers for Disease Control and Prevention (CDC).
when combined with small data, Greater value can be obtained by combining GFT with other near real time health data.
Where data from mobile network operators are used for syndromic surveillance, as in the case of malaria in Kenya (Wesolowski et al.,
2012a), big data are most useful as a basis for encouraging timely investigation, rather than as a replacement for existing measures of disease activity.
Even when engaging with the broader question of how telecommunication network data could be used for monitoring,
surveys and supplemental datasets will remain important to sharpen the analyses and especially to verify the underlying assumptions.
For instance, Blumenstock and Eagle (2012) ran a basic household survey against a randomized set of phone numbers prior to data anonymization to build a training dataset.
Similarly, Frias-Martinez and Virseda (2012) needed census data to build their algorithms and provide training data for their algorithms to reverse engineer approximate survey maps.
Official statistics will thus continue to be important to building the big data models and for periodic benchmarking
so that the models can be tuned fine to reflect ground realities. Transparency and replicability The issues with GFT also illustrate transparency and replicability problems with big data research.
The fact that the original private data may in many cases not be available to everyone underscores the importance of opening up such private-data sources (in a manner that addresses potential privacy concerns)
or of peer reviews that can hone and improve the analyses. Instead, consumers of such research have no option
and extracting value from big data calls for a combination of specialized skills in the areas of data mining, statistics and domain expertise,
as well as data preparation, cleaning and visualization. NSOS may have deep statistical skills in house, but this is not enough
when it comes to working with large volumes of big data calling for computer science and decision-analysis skills that are emphasized not in traditional statistical courses (Mcafee and Brynjolfsson,
and identified intensive training and capacity development of their staff as a prerequisite to being able to exploit new big data sources (UNSC, 2013).
i e. data scientists. Mckinsey predicts that by 2018 the demand for data-savvy managers and analysts in the United states will amount to 450 000,
whereas the supply will fall far short of this, at only 160 000 (Manyika et al.,2011).
) This suggests that organizations wishing to leverage big data for development will face competition from the private sector
which stand to benefit the most from the use of telecommunication big data to complement official statistics,
The way forward Current research suggests that new big data sources have great potential to complement official statistics and produce insightful information to foster development.
Future efforts to mainstream and derive full benefit from the use of big data will have to overcome a number of barriers.
Very limited information is available on opportunities for using big data to complement official ICT statistics.
Although this report highlights some of the big data sources and techniques that could be used, further research is needed to understand
and confirm the usefulness of big data sources for monitoring the information society. As with other official statistics, it is paramount for big data producers
and big data users to collaborate and to initiate a dialogue to identify opportunities and understand needs and constraints.
Since many of the big data sources lie within the private sector close cooperation between NSOS, on the one hand,
and telecommunication operators and Internet companies, including search engines and social networks, on the other, is necessary
Operators and Internet companies Business interests will naturally provide operators and Internet companies with the incentive to talk to commercial vendors of big data analytics.
and researchers to understand how to leverage big data for different purposes. Such engagement will also broaden their understanding of the limitations and assist them in the development of new methodologies
Indeed, where the applications of data use for development are concerned, operators also have an interest in maximizing the economic well-being of their customer base.
They are placed well to develop a Chapter 5. The role of big data for ICT monitoring and for development 204 privacy framework, in consultation with other stakeholders.
and share their data with those from other sources (including from competitors), but this is something that is worth exploring.
Combining big data sources has great potential to increase added value and produce new insights. There is scope for exploring established models for such pooling for example, the sharing by banks of some of their customer data with credit bureaux.
Governments Governments have different opportunities and different roles to play in the exploitation of big data for monitoring and development.
They can use big data to identify areas where rapid intervention may be necessary to track progress
and make sure their decisions are based evidence, and to strengthen accountability. More and more governments are recognizing the importance of big data
and have set up communities of practice and working groups to study their use and potential impact (UNSC 2013).
and take a lead in setting big data standards. To this end, national regulatory authorities (NRAS) and NSOS, in consultation with other national stakeholders, are placed best to lead the corresponding discussions
and big data clearing houses that promote analytical best practices in relation to the use of big data for complementing official statistics and for development.
would also have to encompass best practices in relation to data curation and metadata standards. To this end, NSOS must also prioritize the upgrading of the in-house technical skills they require
in order to handle big data, while at the same time investing in the necessary computational infrastructure. As the main regulatory interface to the telecom sector, NRAS are placed well to co-champion the national discussion on how telecommunication big data may be leveraged for social good.
Regulators have a role to play in facilitating the introduction of legislation that addresses privacy concerns while encouraging data sharing in a secure manner.
The following recommendations were made in a recently published ITU draft paper (ITU 2014): ) Establishing mechanisms to protect privacy:
In return, data users would be permitted to reuse personal data for novel purposes where a privacy assessment indicates minimal privacy risks.
While the use of big data can help better decision-making through probabilistic predictions, this information should not be used against citizens.
which government agencies and others can utilize big data predictions. Fostering big data competition and openness: Regulators could foster big data competition in increasingly concentrated big data markets,
including by ensuring that data holders allow others to access their data under fair and reasonable terms. 205 Measuring the Information Society Report 2014 International stakeholders International stakeholders including UN AGENCIES and initiatives (such as
ITU and UN Global Pulse), the Partnership on Measuring ICT for Development, ICT industry associations and producers of big data (Google, Facebook, etc.)
have an important role globally. More work is needed to understand fully the potential of big data
and examine the challenges and opportunities related to big data in the ICT sector. To this end, the key international stakeholders have to work together to facilitate the global discussion on the use of big data.
UN Global Pulse, as one of the main UN initiatives exploring the use of big data,
can do much to inform and motivate the discussion on global best practices and the use of big data for development.
Where using big data for monitoring the information society is concerned new partnerships, including public-private partnerships between data providers
and the ICT statistical community, including ITU, could be formed to explore new opportunities and address challenges, including in the area of international data comparability and standards.
As one of the main international bodies working on issues related to the telecommunication sector, ITU could leverage its position to facilitate global discussion on the use of telecom big data for monitoring the information society.
Together, ITU and UN Global Pulse could facilitate the work that needs to be done by NRAS and NSOS, through awareness raising and engagement on privacy frameworks
data sharing, and analytical global best practices. ITU could help reduce the transaction costs associated with obtaining telecommunication big data,
for example by facilitating the standards-setting process. Standardized contracts for obtaining data access as well as standards on how the data are stored,
collated and curated can collectively reduce the overall transaction costs of accessing and leveraging telecommunication big data for social good.
Academia, research institutes and development practitioners The research into how telecom data may be used to aid broader development is being done mainly by academia, public and private research institutes and, to a lesser degree,
development practitioners. This makes them important stakeholders in defining the state of the art with respect to leveraging big data for development.
They, more than others, have been the first to engage with telecommunication operators with a view to using their data for development.
They therefore understand the potential and challenges from multiple perspectives. Their collective experiences will be valuable as big data for development becomes mainstreamed. 207 Measuring the Information Society Report 2014 Chapter 5 Annex The mobile-telecommunication data that operators possess can be classified into different types,
depending on the nature of the information they produce. They include traffic data service access detail records, location and movement data, device characteristics, customer details and tariff data.
Traffic data Operators use a range of metrics to understand and manage the traffic flowing through their networks.
These include: Data volume: both uplink and downlink volumes for Internet traffic can be captured at various levels of disaggregation down to the individual subscriber,
or even to the level of a base station (in the case of a mobile operator) or local switch (in PSTN networks).
These can be analysed to understand subscriber demand for data at both an individual level and at aggregate levels,
and the understanding thus gained can be used for billing purposes and for network management. Erlang:
a dimensionless metric used by mobile network operators to understand the offered and utilized network load. 41 Erlang data are used to understand the load on a base station at any given time.
Call, SMS and MMS volumes are used for a variety of purposes from billing to customer relationship management, as well for network planning.
Deep packet inspection42 (DPI) is used to scan the information that goes over a network. Operators employ DPI to varying degrees,
and it is not always feasible for the entire data stream to be captured and stored, owing to the storage requirements that would be needed and also to privacy concerns.
Often only the header information which includes originating and recipient Internet protocols (IPS), is captured for a variety of purposes,
including to manage the network and understand the demand for particular applications and websites. Service access detail records
The most common use of such data is for basic billing purposes, in addition to which they can be used to build a rich profile of customers,
as outlined in Section 5. 3. Chapter 5. The role of big data for ICT monitoring
and for development 208 Location and movement data Mobile networks can, depending on their sophistication, capture a range of movement and location variables,
passive and active positioning data (see Annex Box 1). Annex Box 1: Active versus passive positioning data Passive positioning transaction generated data (TGD) is generated automatically by the network
and captured in the operator's logs for billing and network management purposes, to understand network load
Active positioning data (which is of relevance only to mobile networks) is location and movement data that is captured in response to a specially initiated network query to locate a handset using either network or handset-based positioning methods.
GPS location data can also be considered as active positioning data. Active positioning data Active positioning data can be generated using either devicecentric or network-centric methods
as well as via satellite (i e. GPS. The use of these methods has developed either in response to national regulations requiring operators to capture higher-precision location data,
and/or to a business case for providing location-based services. The large-scale capture of such higher-resolution data is undertaken mainly by operators in developed economies.
Operators in developing economies use some of these methods, but often on a case-by-case basis,
and not for their entire subscriber base. 43 However, this trend is currently changing, and an increasing number of regulators are
Passive positioning data Passive location data are contained in the subscriber's registration information which includes some form of mailing address (billing address in the case of postpaid).
While fixed-telephone network operators have access only to static location data, mobile networks have much richer and dynamic location data.
CDRS, SMS detail records and Internet access records are the main sources of passive positioning data for mobile operators
and reside in their data warehouses. These records include the ID of the antenna (cell ID),
which in turn has a geolocation, an azimuth (i e. antenna orientation information) and an angular tilt. 45 It is also possible to obtain such data in real time through data-mediation services,
but these are implemented not universally. Passive location data from the billing records are obviously sparse
and generated only when the phone is used and when the network knows which cell a particular handset is connected currently to.
However, many operators choose not to archive these data if they do not have a business case to justify the additional storage costs.
such cellhandoff data provide a time-stamped sequence of cells that the phone was attached to,
Passive positioning data based on cell IDS is compared inexpensive when to active positioning data, but the tradeoff lies in their lower precision, usually at the level of network cells.
This lower-resolution location estimate can range from a few hundred metres in urban areas with a higher-density base station coverage
to a few kilometres, especially in rural areas with sparse coverage. Furthermore, handsets are served not always by the nearest antenna,
for a variety of reasons associated with signal strength, topography and saturation loads at the nearest antenna during peak times.
Despite these location errors or limitations that can occur in analyses using such passive location data
and/or spatial) these data remain very valuable. Source: ITU. Device characteristics All mobile user devices used to access mobile telecommunication services come with an international mobile station equipment identity (IMEI) number.
Active positioning methods Method Description Cellular triangulation using angle of arrival (Aoa) The Aoa method uses data from base stations that have been augmented with arrays of smart antennas.
so it is suited more for rural and semi-urban areas with sparser antenna coverage. GSM-GPS and assisted GPS (A-GPS) Both of these utilize the network (mainly via triangulation from multiple base stations) to augment the satellite signal.
Such location data have high spatial resolution, but are costly for operators to implement. Enhanced observed time difference (E-OTD) E-OTD is a device-centric positioning technique requiring the handset to make the necessary location calculation, based on the signals from three or more synchronized
The accuracy of such techniques range from 50 to 125m Note/Source: For more information regarding these methods, refer to CGALIES (2002.
Customer details Telecom network operators capture various items of demographic data during the customer registration process.
and usage profile that the operator can leverage for a variety of purposes (see Section 5. 3). Tariff data Operators maintain the complete tariff sheet
Mobile operators can associate such data with traffic data to understand the revenue that is being generated by specific network elements (e g. base stations),
Chapter 5. The role of big data for ICT monitoring and for development 210 1 The report of the UN Secretary-general's High-level Panel of Eminent Persons on the post-2015 Development Agenda
calls for a data revolution that would draw on existing and new sources of data to fully integrate statistics into decision-making
and use of, data and ensure increased support for statistical systems. This suggests that efforts to improve the availability of,
and complement, official statistics have turned to the search for new data sources, including big data. In addition, the European Statistical System Committee (ESSC) in 2013 adopted the Scheveningen Memorandum on Big data and Official Statistics,
which acknowledges that Big data represents new opportunities and challenges for Official Statistics, and which encourages the European Statistical System
and its partners to effectively examine the potential of Big data sources, see: http://epp. eurostat. ec. europa. eu/portal/page/portal/pgp ess/0 docs/estat/SCHEVENINGEN MEMORANDUM%20final%20version. pdf. 2 This term was discussed first
in 1991, although the term then used was generated transaction information (Mcmanus, 1990) 3 For more information, see http://www. cityofboston. gov/doit/apps/streetbump. asp. 4 See http://www. donotpay. treas. gov/About. htm. 5 In Europe,
the collection and processing of personal data or information is regulated currently by Directive 95/46/EC on the protection of individuals with regard to the processing of personal data and on the free movement of such data (Data protection Directive) 1 and Directive
2002/58/EC, as amended by Directive 2009/136/EC, on privacy and electronic communications (the eprivacy Directive),
which personal data cannot legitimately be processed without the consent of the data subject, except if necessary to preserve public order or morality,
Building on this principle, Article 5 of the eprivacy Directive further provides that the processing of personal data can be effected only with the consent of the data subject
http://policyreview. info/articles/analysis/big data-big-responsibilities. 6 See https://www. google. org/denguetrends/.
/7 A good example of this is the Conference Board Help Wanted Online (HWOL) data series that measures the number of new,
corporate boards and smaller job sites in the United states. More information can be found at http://www. conference-board. org/data/helpwantedonline. cfm. 8 See http://bpp. mit. edu
/11 ITU World Telecommunication/ICT Indicators database, 17th edition, 2014, available at: http://www. itu. int/en/ITU-D/Statistics/Pages/publications/wtid. aspx. 12 See http://blog. stephenwolfram. com/2013/04/data science-of-the-facebook-world/.
and regional economic development in Côte d'ivoire. 14 The term metadata is used also quite extensively to refer to TGD from telecommunication operators. 15 Deep packet inspection (DPI) is a process that utilizes specialized software to scan all of the data
and analysis of health data about a clinical syndrome that has a significant impact on public health,
with the data in question being used to drive decisions about health policy and health education. 22 A vector-borne disease is a disease that is transmitted through an agent (person, animal or microorganism).
23 A geographic information framework is a representation framework that codes the components of geospatial data in a standardized manner so as to facilitate analyses
and data exchange. 24 See http://www. itu. int/en/ITU-D/Statistics/Pages/intlcoop/partnership/default. aspx for more information regarding the partnership. 25 For the latest list
514426,00. html. 30 More information about the Data for Development (D4d) challenges using Orange data can be found at http://www. d4d. orange. com/home. 31 See http://www. telecomitalia. com
/tit/en/bigdatachallenge/contest. html. 32 More information about the Yale university Open Data (YODA) project can be found at http://medicine. yale. edu/core/projects/yodap/index
or two people talking for 30 minutes each. 42 DPI is a process that utilizes specialized software to scan all of the data packets traversing a particular IP network.
and would have to have an accuracy estimate of up to 50m within three years of the new licence coming into effect.
of the sample regulation on the Indian Department of Telecommunications website (http://dot. gov. in/sites/default/files/Unified%20licence 0. pdf). Chapter 5. The role of big data for ICT monitoring and for development
For more information, see http://en. wikipedia. org/wiki/MAC ADDRESS. 47 An international mobile subscriber identity (IMSI) number is a 15-digit number unique to the particular SIM in a subscriber's handset.
the data deluge makes the scientific method obsolete. Wired Magazine, 16 (7). Askitas, N. and Zimmermann, K. F. 2009), Google econometrics and unemployment forecasting.
http://ftp. iza. org/dp4201. pdf. Azman, A s.,Urquhart, E. A.,Zaitchik, B. and Lessler, J. 2013), Using Mobile phone Data to Supercharge Epidemic Models
Data for Development: Net Mobi 2013. Bengtsson, L.,Lu, X.,Thorson, A.,Garfield, R. and von Schreeb, J. 2011), Improved response to disasters and outbreaks by tracking population movements with mobile phone network data:
a post-earthquake geospatial study in Haiti. PLOS Medicine, 8 (8), e1001083. doi: 10.1371/journal. pmed. 1001083.
and optimizing public transport using cellphone data. Data for Development: Net Mobi 2013. Blumenstock, J. E. 2011), Using mobile phone data to measure the ties between nations.
Proceedings of the 2011 iconference (pp. 195 202. New york, USA: ACM Press. doi: 10.1145/1940761.1940788.
Blumenstock, J. E. 2012), Inferring patterns of internal migration from mobile phone call records: evidence from Rwanda.
Bollier, D. 2010), The Promise and Peril of Big data. WASHINGTON DC. Bouckaert, J.,Van dijk, T. and Verboven, F. 2010), Access regulation, competition, and broadband penetration:
Boyd, D. and Crawford, K. 2012), Critical questions for big data. Information, Communication & Society, 15 (5), 662 679. doi:
How Does driven Data Decisionmaking Affect Firm Performance? Retrieved from: http://papers. ssrn. com/sol3/papers. cfm?
Byrne, C. 2013), How The UN's New Data Lab in Indonesia Uses Twitter to Preempt Disaster, Fast Company.
http://www. fastcolabs. com/3007178/open-company/how-uns-new-data-lab-indonesia-uses-twitterpreempt-disaster.
Caceres, N.,Wideberg, J. P. and Benitez, F. G. 2007), Deriving origin destination data from a mobile phone network.
Calabrese, F.,Di Lorenzo, G.,Liu, L. and Ratti, C. 2011), Estimating origin-destination flows using mobile phone location data.
Global Mobile Data Traffic Forecast Update, 2013-2018, White paper, February 2014. Available at: http://www. cisco. com/c/en/us/solutions/collateral/service-provider/visual-networking-index-vni/white paper c11-520862. pdf. CCK (2010), Communications
http://www. cck. go. ke/resc/downloads/Sector statistics report q2 201314. pdf. David, T. 2013), Big data from Cheap Phones.
. 2009), Inferring friendship network structure by using mobile phone data. Proceedings of the National Academy of Sciences of the United states of america, 106 (36), 15274 8. doi:
, M. S. and Brilliant, L. 2009), Detecting influenza epidemics using search engine query data. Nature, 457 (7232), 1012 4. doi:
Groves, R. M. 2011), A possible data future for the observational social sciences. COSSA 30th Anniversary meeting, Washington, D c. Gruber, H. and Koutroumpis, P. 2013), Competition enhancing regulation and diffusion of innovation:
https://wiki. gsmaintelligence. com/gsma kb/images/e/e0/Mobile for development country overview sri lanka. pdf. GSMA (2013b), The Mandatory Registration of Prepaid SIM CARD Users, White paper.
Harford, T. 2014), Big data: are we making a big mistake? Financial times. pp. 7 11. Retrieved 28 march 2014 from:
The real-world use of big data. How innovative enterprises extract value from uncertain data. Retrieved from:
http://public. dhe. ibm. com/common/ssi/ecm/en/gbe03519usen/GBE03519USEN. PDF. IBM (2013), The Four V's of Big data.
Retrieved from: http://www. ibmbigdatahub. com/sites/default/files/infographic file/4-Vs-of-big data. jpg.
ICT Qatar (2014), Qatar's ICT landscape 2014 households and individuals. Available at: http://www. ictqatar. qa/en/documents/download/Qatar's%20ict%20landscape%20report%202014-Household%20and%20individuals 12. pdf. ITU (2006), Security in Telecommunications
http://www. itu. int/ITU-D/ict/publications/wtdr 10/material/WTDR2010 EXECSUM-en. pdf. ITU (2011), Handbook for the collection of administrative data on Telecommunications/ICT.
http://www. itu. int/en/ITU-D/Statistics/Documents/publications/mis2012/MIS2012 WITHOUT ANNEX 4. pdf. ITU (2013), Big data:
Report 2014 ITU GSR discussion paper (2014), Big data-Opportunity or Threat. Retrieved from: http://www. itu. int/en/ITU-D/Conferences/GSR/Documents/GSR2014/Discussion%20papers%20and%20presentations%20-%20gsr14/Session3 gsr14-Discussionpaper-Bigdata
. pdf. Jones, S. 2012), Why Big data is the fourth factor of production, Financial times. Retrieved 27 december 2012 from:
Krumholz, H. M. 2014), Give the data to the People, New york times. Retrieved 2 february 2014 from:
http://www. nytimes. com/2014/02/03/opinion/give-the-data-to-the-people. html? hp&rref=opinion& r=1. Kumar, K. and Muhota, K. 2012), Can Digital Footprints Lead to Greater Financial Inclusion?(
Laney, D. 2001), 3d data management: Controlling data volume, velocity and variety. Gartner, 6. Lazer, D.,Kennedy, R.,King, G. and Vespignani, A. 2014), Big data.
The parable of Google Flu: traps in big data analysis. Science (New york, N y.),343 (6176), 1203 5. doi:
10.1126/science. 1248506. Lazer, D.,Pentland, A.,Adamic, L.,Aral, S.,Barabasi, A l.,Brewer, D.,Van alstyne, M. 2009), Computational social science.
Science (New york, N y.),323 (5915), 721 3. doi: 10.1126/science. 1167742. Lenormand, M, Picornell, M.,Cantu-Ros, O.,Tugores, A.,Louail, T.,Herranz, R.,Barthelemy, M.,Frias-Martinez, E.,Ramasco, J. Cross
Lokanathan, S.,Silva, N. de, Kreindler, G.,Miyauchi, Y. and Dhananjaya, D. 2014), Using mobile network big data for informing transportation and urban planning in Colombo.
Manyika, J.,Chui, M.,Brown, B.,Bughin, J.,Dobbs, R.,Roxburgh, C. and Byers, A. H. 2011), Big data:
Data for Development: Net Mobi 2013. Mayer-Schönberger, V. and Cukier, K. 2013), Big data: A revolution that will transform how we live,
work, and think. Houghton Mifflin Harcourt. Mcafee, A. and Brynjolfsson, E. 2012), Big data: the management revolution.
Harvard Business Review, 90 (10), 60 6, 68,128. Retrieved from: http://www. ncbi. nlm. nih. gov/pubmed/23074865.
What big data means for social science. Headcon'13 Part I. Retrieved from: http://edge. org/panel/headcon-13-part-i. Naef, E.,Muelbert, P.,Raza, S.,Frederick, R.,Kendall, J. and Gupta, N. 2014
), Using mobile data for development. Gates Foundation Whitepaper. Narayanan, A. and Shmatikov, V. 2008), Robust de-anonymization of large sparse datasets. 2008 IEEE Symposium on Security and Privacy (sp 2008)( pp. 111 125.
Pindolia, D. K.,Garcia, A j.,Wesolowski, A.,Smith, D. L.,Buckee, C. O.,Noor, A m.,Tatem, A j. 2012), Human movement data
2012), Demystifying Big data: A Practical Guide To Transforming The Business of Government. Retrieved from: http://www. techamerica. org/Docs/filemanager. cfm?
f=techamerica-bigdatareport-final. pdf. The White house (2014), Big data: Seizing Opportunities, Preserving values. Executive Office of the President.
Using New Data to Understand Emerging Vulnerability in Real-time. UN High-level Panel (2013), A new global Partnership:
http://unohrlls. org/custom-content/uploads/2013/10/State-of-the-LDCS-2013. pdf. UNSC (2013), Report of the Secretary general on Big data and modernization of statistical systems.
UNSC (2014), Big data and modernization of statistical systems, December 2013. Available at: http://unstats. un. org/unsd/statcom/doc14/2014-11-Bigdata-E. pdf. Van damme, E. 2002), The European UMTS-auctions, European Economic
10.1016/S0014-2921 (01) 00219-7. Varian, H r. 2013a), Beyond Big data. NABE Annual Meeting. San francisco, CA. Varian, H r. 2013b), Big data:
New Tricks for Econometrics. Wallsten, S. 2001), An econometric analysis of telecom competition, privatization, and regulation in Africa and Latin america.
A. 2013), Undefined By Data: A Survey of Big data Definitions, 2. Databases. Retrieved from: http://arxiv. org/abs/1309.5821.
Waverman, L. and Koutroumpis, P. 2011), Benchmarking telecoms regulation: The Telecommunications Regulatory Governance Index (TRGI.
Weber, N m.,Palmer, C. L. and Chao, T. C. 2012), Current Trends and Future Directions in Data Curation Research and Education.
Rewards and Risks of Big data (B. Bilbao-Osorio, S. Dutta and B. Lanvin, Eds..References 220 Wesolowski, A. and Eagle, N. 2010), Parameterizing the Dynamics of Slums.
Using Cellular network Data. Data for Development: Net Mobi 2013. Xia, J. 2011), The third-generation-mobile (3g) policy and deployment in China:
Current status, challenges, and prospects, Telecommunications Policy, Volume 35, Issue 1 february 2011, pp. 51-63. doi:
and the results of the sensitivity analysis. 1. Indicators included in the IDI The selection of indicators was based on certain criteria,
including relevance for the index objectives, data availability and the results of various statistical analyses such as the principal component analysis (PCA).
Data for all of these indicators are collected by ITU. 2 1. Fixed-telephone subscriptions per 100 inhabitants Fixed-telephone subscriptions refers to the sum of active analogue fixed-telephone lines,
It excludes subscriptions via data cards or USB modems, subscriptions to public mobile data services,
in megabits per second (Mbit/s). It is measured as the sum of used capacity of all Internet exchanges offering international bandwidth.
but should be considered a household asset. 3 Data are obtained by countries through national household surveys
There are certain data limits to this indicator, insofar as estimates have to be calculated for many developing countries
as more data become available, the quality of the indicator will improve. 5. Percentage of households with Internet access The Internet is a worldwide public computer network.
Data are obtained by countries through national household surveys and are provided either directly to ITU by national statistical offices (NSO),
There are certain data limits to this indicator, insofar as estimates have to be calculated for many developing countries
as more data become available, the quality of the indicator will improve. b) ICT use indicators The indicators included in this group capture ICT intensity and usage.
Data for all of these indicators are collected by ITU. 4 1. Percentage of individuals using the Internet Individuals using the Internet refers to people who used the Internet from any location and for any purpose, irrespective of the device and network used, in the last three months.
Data are obtained by countries through national household surveys and are provided either directly to ITU by national statistical offices (NSO),
There are certain data limits to this indicator, insofar as estimates have to be calculated for many developing countries
as more data become available, the quality of the indicator will improve. 2. Fixed (wired)- broadband subscriptions per 100 inhabitants Fixed (wired)- broadband subscriptions refers to the number of subscriptions for high-speed access to the public Internet (a TCP IP connection).
It also excludes mobilebroadband subscriptions where users can access a service throughout the country wherever coverage is available.
Active mobile-broadband subscriptions refers to the sum of standard mobilebroadband subscriptions and dedicated mobile-broadband data subscriptions to the public Internet.
Standard mobile-broadband subscriptions refers to active mobile-cellular subscriptions with advertised data speeds of 256 kbit/s
and which have been used to set up an Internet data connection using Internet Protocol (IP) in the past three months.
Standard SMS and MMS messaging do not count as an active Internet data connection, even if the messages are delivered via IP.
Dedicated mobile-broadband data subscriptions refers to subscriptions to dedicated data services (over a mobile network) that allow access to the greater Internet
either as a standalone service (e g. using a data card such as a USB modem/dongle) or as an add-on data package to voice services
Annex 1. ICT Development Index (IDI) methodology 224 c) ICT skills indicators Data on adult literacy rates and gross secondary and tertiary enrolment
regardless of age, expressed as a percentage of the eligible official school age population corresponding to the same level of education in a given school-year. 2. Imputation of missing data A critical step in the construction of the index is to create a complete
data set, without missing values. There are several imputation techniques that can be applied to estimate missing data. 6 Each of the imputation techniques,
like any other method employed in the process, has its own strengths and weaknesses. The most important consideration is to ensure that the imputed data will reflect a country's actual level of ICT access
usage and skills. Given that ICT access and usage are correlated both with national income, hot-deck imputation was chosen as the method for imputing the missing data,
where previous year data are not available to calculate the growth rates. Hotdeck imputation uses data from countries with similar characteristics, such as GNI per capita and geographic location.
For example, missing data for country A were estimated for a certain indicator by first identifying the countries that have similar levels of GNI per capita
and that are from the same region and an indicator that has known a relationship to the indicator to be estimated.
For instance Internet use data of country A was estimated by using Internet use data of country B from the same region with similar level of GNI per capita and similar level of fixed Internet and wireless-broadband subscriptions.
The same logic was applied to estimate missing data for all indicators included in the index. 3. Normalization of data Normalization of the data is necessary before any aggregation can be made
in order to ensure that the data set uses the same unit of measurement. For the indicators selected for the construction of the IDI,
it is important to transform the values to the same unit of measurement, since some values are expressed as a percentage of the population/of households,
the data were transformed first to a logarithmic (log) scale. Outliers were identified then using a cut off value calculated by adding two standard deviations to the mean of the rescaled values, resulting in a log value of 5. 90.
After normalizing the data, the individual series were rescaled all to identical ranges, from 1 to 10.
(which tops the IDI 2013). 6. Sensitivity analysis Sensitivity analysis was carried out to investigate the robustness of the index results,
the data were transformed first to a logarithmic (log) scale. The ideal value of 787'260 bit/s per Internet user is equivalent to 5. 90 if transformed to a log scale.
including the selection of individual indicators, the imputation of missing values and the normalization, weighting and aggregation of the data.
This confirms the results conveyed by the IDI. 229 Measuring the Information Society Report 2014 1 Principal component analysis was used to examine the underlying nature of the data.
The ICT Development Index'report (ITU, 2009). 2 More information about the indicators is available in the ITU‘Handbook for the collection of administrative data on telecommunications/ICT'2011,
on 4-6 june 2013, see http://www. itu. int/en/ITU-D/Statistics/Documents/events/brazil2013/Final report EGH. pdf). As some of the data used in the calculation of the IDI
however, the data may not necessarily reflect these revisions. 4 More information about the indicators is available in the ITU Handbook for the collection of administrative data on telecommunications/ICT'2011,
Endnotes 231 Measuring the Information Society Report 2014 Annex 2. ICT price data methodology 1. Price data collection and sources The price data
The data were collected through the ITU ICT Price Basket questionnaire, which was sent to the administrations and statistical contacts of all 193 ITU Member States in October 2013.
Through the questionnaire, contacts were requested to provide 2013 data for fixed-telephone, mobile-cellular, fixed-broadband and mobile-broadband prices;
The collection of price data from ITU Member States and the methodology applied for the IPB was agreed upon by the ITU Expert Group on Telecommunication/ICT Indicators (EGTI) 1 and endorsed by the eighth World
Annex 2. ICT price data methodology 232 The fixed-telephone sub-basket does not take into consideration the onetime connection charge.
such as the onetime charge for a SIM CARD. The basket gives the price of a standard basket of mobile monthly usage in USD determined by OECD for 30 outgoing calls per month in predetermined ratios plus 100 SMS messages. 4 The cost
this should be indicated in a note. 6. The same price plan should be used for collecting all the data specified.
This can present a challenge for data collection, since it may not be possible to isolate the prices for one service.
the mobilecellular sub-basket therefore corresponds to a basic, representative (low-usage) package Annex 2. ICT price data methodology 234 Annex Table 2. 1:
For comparability reasons, the fixedbroadband sub-basket is based on a monthly data usage of (a minimum of) 1 Gigabyte (GB.
For plans that limit the monthly amount of data transferred by including data volume caps below 1 GB,
the cost for the additional bytes is added to the sub-basket. The minimum speed of a broadband connection is 256 kbit/s. 235 Measuring the Information Society Report 2014 Annex Box 2. 2:
Annex 2. ICT price data methodology 236annex Box 2. 3: Rules applied in collecting fixed-broadband Internet prices 1. The prices of the operator with the largest market share (measured by the number of subscriptions) should be used. 2. Prices should be collected in national currency,
7. The same price plan should be used for collecting all the data specified. For example, if a given Plan A is selected for the fixedbroadband service,
the volume of data that can be downloaded, etc. 8. Prices should be collected for regular (non-promotional) plan
This can present a challenge for price data collection, since it may not be possible to isolate the prices for one service.
preference is given to the cheapest available connection that offers a speed of at least 256 kbit/s and 1 GB of data volume.
If providers set a limit of less than 1 GB on the amount of data that can be transferred within a month,
then the price per additional byte is added to the monthly price so as to calculate the cost of 1 GB of data per month.
Preference should be given to the most widely used fixed (wired)- broadband technology (DSL, cable, etc.).
for the first time, ITU collected mobilebroadband prices through its annual ICT Price Basket Questionnaire. 8 The collection of mobilebroadband price data from ITU Member States was agreed upon by the ITU Expert Group
and revised in 2013 by EGTI in view of the lessons learned from the first data collection exercise.
To capture the price of different data packages, covering prepaid and postpaid services, and supported by different devices (handset and computer),
mobile-broadband prices were collected for two different data thresholds, based on a set of rules (see Annex Box 2. 4). For plans that were limited in terms of validity (less than 30 days),
for extending a plan limited in terms of data allowance (or validity). The customer:(i) continues to use the service and pays an excess usage charge for additional data10 or (ii) purchases an additional (add-on) package.
The plans selected represent the least expensive offers that include the minimum amount of data for each respective mobile-broadband plan.
and could purchase given the data allowance and validity of each respective plan. Annex 2. ICT price data methodology 238annex Box 2. 4:
Rules applied in collecting mobile-broadband prices11 1. Prices should be collected based on one of the following technologies:
the plan satisfying the indicated data volume requirement should be used. 8. Where operators propose different commitment periods for postpaid mobile-broadband plans,
A note should be added in case only longer commitment periods are offered. 9. Price data should be collected for the cheapest plan with a data volume allowance of a minimum of:
Data volumes should refer to both upload and download data volumes. If prices are linked to‘hours of use'and not to data volumes
this information should be added in a separate note. Note: ITU will most likely not be able to include these cases in a comparison. 10.
A validity period of 30 days should be chosen. If this is not available, 15 days should be used.
Preference should be given to packages (including a certain data volume. Pay-as-you-go offers should be used
since most often there are limits in the data volumes, either applied by throttling (limiting the speed)
(i e. data collected from operators), as well as to discuss outstanding methodological issues and new indicators. EGTI is open to all ITU members and experts in the field of ICT statistics and data collection.
It works through an online discussion forum (http://www. itu. int/ITU-D/ict/Expertgroup/default. asp) and face-to-face meetings.
in such cases, the advertised price is used. 8 Data for fixed-telephone, mobile-cellular and fixed-broadband have been collected since 2008 through the ITU ICT Price Basket Questionnaire,
which is sent out annually to all ITU Member States/national statistical contacts. 9 See footnote 1. 10 Some operators throttle speeds after the data allowance included in the base package has been reached.
Data in italics refer to ITU estimates. For further notes, see p. 248. Source: ITU World Telecommunication/ICT Indicators database.
Annex 3. Statistical tables of indicators used to compute de IDI 244 Percentage of individuals using the Internet Fixed (wired)- broadband subscriptions per 100 inhabitants Wireless-broadband subscriptions per 100 inhabitants
Data in italics refer to ITU estimates. For further notes, see p. 248. Source: ITU World Telecommunication/ICT Indicators database.
Annex 3. Statistical tables of indicators used to compute de IDI 246 Skills indicators Gross enrolment ratio Adult Seconday Tertiary literacy rate Economy 2012 2013
Data in italics refer to ITU estimates. Source: ITU World Telecommunication/ICT Indicators database. Annex 3. Statistical tables of indicators used to compute de IDI 248 Access indicators Fixed-telephone subscriptions per 100 inhabitants, 2012: 1) Incl. 524 958 WLL
subscriptions. 2) Incl. payphone, excl. VOIP. 3) Incl. ISDN channels measured in ISDNB channels equivalents. 4) Incl.
policy of the biggest WLL operator. 13) Data excluding own (NRA) consumption. 14) Excl. voice-over-IP (Voip) subscriptions,
45 609.17) February 2013 NTA MIS. 18) Based on 2013q3 data. 19) Refers to active Fixed Wired/Wireless lines. 20) Per June 2013.21) Operators'data
Definitive data (annual report) may change because quarterly reports use a smaller sample of operators than annual report. 24) Fixed and fixed-wireless subscriptions. 25) Excl. internal lines
Data for the third quarter of 2013. Mobile-cellular subscriptions per 100 inhabitants, 2012: 1) Numbers are down due to data cleanse. 2) ACMA Communications Report 2011-12.2) incl. payphone, excl.
Voip. 3) Active subscriptions. 4) Bhutan Telecom and Tashi Cell are the only two service providers in Bhutan. 5) Activity criteria:
2 082 589.8) Excl. 2 720 698 prepaid cards that are used to provide Travel SIM/World Mobile service. 9) Excl. data-only SIM CARDS
Including PHS and data cards, undividable. 15) Decrease was due to registration of SIMS. 16) Figure obtained from all five mobile (GSM
& CDMA) operators currently providing service in the country. 17) Incl. inactive subscriptions. 18) Preliminary. 19) Active subscriptions (87.85%total).
Incl. inactive. 22) Incl. data-only subscriptions (not possible to disaggregate the information at this point.
excl. 226 827 M2m subscriptions. 24) Incl. active (in the last 6 months) prepaid accounts. 25) Registered SIM CARDS (incl. inactive:
from this year only counting prepaid subscriptions used in the last 90 days. 27) No differentiation between active and non-active subscriptions. 28) Incl. data dedicated subscriptions. 29) Decrease due to the closing
but excl. mobile data subscriptions (via data cards, USB modems and M2m cards). 7) Estimate. 8) Incl. data-only subscriptions. 9) Data based on NRA estimates.
Excl. data-only SIM CARDS and M2m cards. 10) Incl. fixed wireless local loop (WLL) subscriptions. 11) Incl.
PHS and data cards, undividable. 12) Active subscriptions (after clearing simcards that became inactive. 13) Figure obtained from all four mobile (GSM
& CDMA) operators currently providing service in the country. 14) Break in comparability. Incl. only active subscriptions. 15) Active subscriptions (83.44%total.
16 february 2013 NTA MIS. 17) Based on 2013q3 data. 18) Measured using subscriptions active in the last 90 days. 19) Per June 2013.20) Excl. 463 646
M2m subscriptions. 21) Incl. active (in the last 6 months) prepaid accounts. 22) Registered SIM CARDS (incl. inactive:
Definitive data (annual report) may change because quarterly reports use a smaller sample of operators than annual report. 26) Data for the fourth quarter of 2013.27) Incl. data dedicated subscriptions
. 28) Reduction is due to implementation of sim cards registration. International Internet bandwidth Bit/s per Internet user, 2012: 1) Refers to a survey conducted with the following companies:
MOT. 11) Data obtained from nine service operators. 12 may 2012 purchased capacity. Lit capacity: 43 096 471 Mbit/s. 13) Incoming capacity;
peak Notes 249 Measuring the Information Society Report 2014 weekly incoming capacity, averaged over 4 weeks in December 14) Preliminary. 15) SLT Data. 16) Refers to the total
4 may be revised with comprehensive data from mobile broadbnad providers. 5) Estimate. 6 june. 7) Sum of incoming capacity of all ISPS in the country. 8) Activated external capacity. 9) By September 2013.10) Data obtained from eight service operators. 11
) 1st april 2013 purchased capacity. Lit capacity: 70 464 304 Mbit/s. 12) Incoming capacity, average peak incoming capacity for December 13) Not update for 2013.14) Refers to the total capacity of the international bandwidth. 15) Potential (installed) capacity.
or laptop. 4) Data correspond to dwellings (not households). 5) Ghana Living Standards Survey 2012/2013.
and electronic dictionary. 7) Estimate. 8) From Household Socioeconomic survey-2012.9) Census data. 10) Computer includes the number of personal computer, Notebook,
laptop or a tablet. 4) Data correspond to dwellings (not households). 5) Ghana Living Standards Survey 2012/2013.
Percentage of households with Internet access, 2012: 1) Estimated based on 2011 proportion of households with internet and using estimated annual growth rate of 2. 8%.2) Preliminary. 3) Data
Refers to access at home, on cell phone or other mobile device and via mobile modem. 10) Census data. 11) Excl. households
. 3) Data correspond to dwellings (not households. 4) Ghana Living Standards Survey 2012/2013. The estimate is based on households who own
Population 5+.10) 12+years. 11) Population 5+.Direct response from individuals 15 years and above. 12) The methodology depends basically on the number of internet users using hard indicators instead of data
based on the data provided by 89.1%of operators. 6) The figure is corrected. The previous figure was 1'636'700.7) Only ADSL, excl. cable modem. 8) Speeds greater than,
/s. 14) Operators data/ictqatar estimate. 15) Incl. subscriptions at downstream speeds equal to, or greater than, 144 kbit/s (the number of subscriptions that are included in the 144-256 range is insignificant.
16) Q3. 17) Excl. 3203 Wimax subscriptions. 18) Excl. corporate connections. 19) Data reflect subscriptions with associated transfer rates exceeding 200 kbps
no specific data collected for=256 kbit/s. 6) CRC estimation as of 31.12.2013.7) Estimate. 8) Data based on NRA estimates. 9) Only ADSL,
These are the subscriptions with the minimum download speed of 512 kbps. This is as per the revised definition of Broadband in India with effect from 18th july 2013.12) December 13) ADSL and Leased lines. 14) Based on 2013q3 data
. 15) Per June 2013.16) Operators'data. 17) Incl. subscriptions at downstream speeds equal to, or greater than, 144 kbit/s (the number of subscriptions that are included in the 144-256 range is insignificant).
Definitive data (annual report) may change because quarterly reports use a smaller sample of operators than annual report. 19) Estimate.
Refers to March 2013.20) Excl. 3175 Wimax subscriptions. 21) Excl. corporate connections. 22) 2013 data is an estimate as of June 30, 2013.
Data reflect subscriptions with associated transfer rates exceeding 200 kbps in at least one direction, consistent with the reporting threshold the FCC adopted in 2000.
BWA and active mobile subscriptions. 13) Estimate based on partial SIT data and ITU estimates. 14) Speeds greater than,
/Data refer to the sum of fixed wireless broadband and active mobilebroadband subscriptions. 16) Incl. mobile broadband
but data are not available. 23) Operators data/ictqatar estimate. 24) Refers to active mobile-broadband subscriptions only. 25.dec 26) Incl. 4125165 active mobile-broadband subscriptions plus 3203
512 Kbps. 15) subscriptions with minimum download speed of 512 kbps. This is as per the revised definition of Broadband in India with effect from 18th july 2013.16) Data refer to the sum of fixed wireless
broadband and active mobile-broadband subscriptions. 17) 2013 data is an estimate as of June 30,
2013.18) Incl. mobile broadband and Wimax. 19) Estimate based on 1. Standard mobile subscriptions using data services 2. Dedicated data subscriptions 3. Add on data
packages. 20) Based on 2013q3 data. 21) Per June 2013.22) Mobile broadband only. Fixed wireless and satellite exist
but data are not available. 23) Operators'data. 24) As at Dec 2013.25) Q4 report.
Definitive data (annual report) may change because quarterly reports use a smaller sample of operators than annual report. 26) Wireless Broadband services are not being offered in St vincent as yet.
Overtext Web Module V3.0 Alpha
Copyright Semantic-Knowledge, 1994-2011