<%BANNER%>

A Cross-Country Analysis of Ubiquitous Broadband Deployment

Permanent Link: http://ufdc.ufl.edu/UFE0022611/00001

Material Information

Title: A Cross-Country Analysis of Ubiquitous Broadband Deployment Examination of Adoption Factors
Physical Description: 1 online resource (147 p.)
Language: english
Creator: Lee, Sangwon
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

Subjects / Keywords: adoption, broadband, mobile, policy
Journalism and Communications -- Dissertations, Academic -- UF
Genre: Mass Communication thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Broadband infrastructure is a key component of the knowledge economy. Broadband connections on both fixed and mobile networks are becoming an indicator of the knowledge economy. A growing body of scholarship details contributing factors that may lead to broadband adoption. In spite of the growing body of literature about broadband adoption, these previous studies have the following limitations: 1) small number of independent variables; 2) insufficient number of observations; 3) lack of theoretical background; 4) focus on only fixed broadband technology; and 5) inconsistent empirical results. Employing the largest secondary data set, this study examines adoption factors of fixed and mobile as well as ubiquitous broadband (fixed and mobile). The result of nonlinear and linear regression analysis of fixed broadband deployment suggests local loop unbundling (LLU) policy, platform completion between different broadband technologies and other diverse industry, ICT (Information and Communication Technology) and demographic factors influenced fixed broadband diffusion. Specifically, the regression analysis of fixed broadband penetration found different types of LLU policies and previous fixed broadband penetration are significant factors of fixed broadband deployment. Some of the significant factors of fixed broadband deployment are different in the developed countries and developing countries. The result of linear regression analysis of mobile broadband deployment suggests market mediated standardization policy, income, and 1G and 2G mobile penetration are significant factors of mobile broadband deployment. Also, the result of linear regression analysis of ubiquitous broadband deployment suggests with other industry, ICT, and demographic variables, network competition between fixed and mobile and interactions of platform completion in fixed broadband markets and multiple standardization policy in mobile markets are significant factors of ubiquitous broadband deployment. Some of the significant factors of ubiquitous broadband deployment were different in the developed countries and developing countries. Considering the result of this study, countries fostering broadband deployment need to adopt LLU policy for broadband, but the costs and benefits of LLU policy should be carefully considered. The results of this study also implies for the initial 4G mobile markets, whereby fixed and mobile broadband networks will be converged, governments need to be open to diverse competitive standards instead of government-mandated standards. However, in the long term, industry-wide coordination and mutual learning processes are more important.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Sangwon Lee.
Thesis: Thesis (Ph.D.)--University of Florida, 2008.
Local: Adviser: Brown, Justin.
Local: Co-adviser: Chan-Olmsted, Sylvia M.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2009-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0022611:00001

Permanent Link: http://ufdc.ufl.edu/UFE0022611/00001

Material Information

Title: A Cross-Country Analysis of Ubiquitous Broadband Deployment Examination of Adoption Factors
Physical Description: 1 online resource (147 p.)
Language: english
Creator: Lee, Sangwon
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

Subjects / Keywords: adoption, broadband, mobile, policy
Journalism and Communications -- Dissertations, Academic -- UF
Genre: Mass Communication thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Broadband infrastructure is a key component of the knowledge economy. Broadband connections on both fixed and mobile networks are becoming an indicator of the knowledge economy. A growing body of scholarship details contributing factors that may lead to broadband adoption. In spite of the growing body of literature about broadband adoption, these previous studies have the following limitations: 1) small number of independent variables; 2) insufficient number of observations; 3) lack of theoretical background; 4) focus on only fixed broadband technology; and 5) inconsistent empirical results. Employing the largest secondary data set, this study examines adoption factors of fixed and mobile as well as ubiquitous broadband (fixed and mobile). The result of nonlinear and linear regression analysis of fixed broadband deployment suggests local loop unbundling (LLU) policy, platform completion between different broadband technologies and other diverse industry, ICT (Information and Communication Technology) and demographic factors influenced fixed broadband diffusion. Specifically, the regression analysis of fixed broadband penetration found different types of LLU policies and previous fixed broadband penetration are significant factors of fixed broadband deployment. Some of the significant factors of fixed broadband deployment are different in the developed countries and developing countries. The result of linear regression analysis of mobile broadband deployment suggests market mediated standardization policy, income, and 1G and 2G mobile penetration are significant factors of mobile broadband deployment. Also, the result of linear regression analysis of ubiquitous broadband deployment suggests with other industry, ICT, and demographic variables, network competition between fixed and mobile and interactions of platform completion in fixed broadband markets and multiple standardization policy in mobile markets are significant factors of ubiquitous broadband deployment. Some of the significant factors of ubiquitous broadband deployment were different in the developed countries and developing countries. Considering the result of this study, countries fostering broadband deployment need to adopt LLU policy for broadband, but the costs and benefits of LLU policy should be carefully considered. The results of this study also implies for the initial 4G mobile markets, whereby fixed and mobile broadband networks will be converged, governments need to be open to diverse competitive standards instead of government-mandated standards. However, in the long term, industry-wide coordination and mutual learning processes are more important.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Sangwon Lee.
Thesis: Thesis (Ph.D.)--University of Florida, 2008.
Local: Adviser: Brown, Justin.
Local: Co-adviser: Chan-Olmsted, Sylvia M.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2009-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0022611:00001


This item has the following downloads:


Full Text

PAGE 1

1 A CROSS-COUNTRY ANALYSIS OF UBI QUITOUS BROADBAND DEPLOYMENT: EXAMINATION OF ADOPTION FACTORS By SANGWON LEE A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2008

PAGE 2

2 2008 Sangwon Lee

PAGE 3

3 To my parents, parents-in -laws, and wife, Eunjeong.

PAGE 4

4 ACKNOWLEDGMENTS First, I thank God for his unconditional love fo r m e. Without his love, I could not complete my study and dissertation in the doctoral progra m. Second, I am sincerely grateful to my dissertation chair, Dr. Justin Brown. He supported my study dur ing three years while I am studying telecommunication policy and media economics field at Univ ersity of Florida. Without his support, help, and advices, I could not achieve anything duri ng my Ph.D. study. Third, I truly appreciate my dissertation co-c hair, Dr. Sylvia Chan-Olmsted. From her I learned a lot of valuable direction and creative ideas of my research in media economics and management field. It was truly honor for me to write seven resear ch papers with her. Forth, I thank Dr. David Ostroff. He gave me theoretical approaches about new media systems, which I havent familiar with. These theoretical perspe ctives are truly useful for my dissertation and research. Also, I truly appreciate Dr. Sanford Berg s comments and ideas, which were very influential for dissertation. Wit hout his comments and ideas, I c ould not write be tter dissertation. I also thank Dr. Mark Jamisons comments and supports for my research. From Dr. Jamison, I got a lot of valuable information about broadband issues. Without his support, I could not receive a research grant from the NET Institute. In addition, I gratefully ac knowledge a research grant from the NET Institute, which supported part of my dissertation research. Also, I thank my parents, parents-in-law, and brotherin-law, Kue Cheol, for their l oving encouragement. Finally I tr uly thank my wife, Eunjeongs unconditional sacrifice and support for my st udy.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ............................................................................................................... 4LIST OF TABLES ...........................................................................................................................8LIST OF FIGURES .......................................................................................................................10ABSTRACT ...................................................................................................................... .............11 CHAP TER 1 INTRODUCTION .................................................................................................................. 13Broadband Deployment and Knowledge Economy ...............................................................13Technologies for Broadband Communications ...................................................................... 14Fixed Broadband .............................................................................................................14Mobile Broadband ...........................................................................................................15Portable Internet ..............................................................................................................16Current Status of Global Broadband Deployment ........................................................... 17Significance of the Study ........................................................................................................21Purpose of the Study .......................................................................................................... .....232 LITERATURE REVIEW .......................................................................................................25Theoretical Backgrounds of Broadband Adoption ................................................................. 25Micro-Individual Level Approaches ............................................................................... 25Platform Competition ...................................................................................................... 26Network Effect ................................................................................................................28Digital Divide and Leapfrogging Theory ........................................................................29Path Dependence .............................................................................................................31Drivers of Ubiquitous Broadband Concept ............................................................................ 33Application and Service Convergence ............................................................................ 34Technological Innovation and Convergence ................................................................... 34Industry Convergence and Multiple Play Strategy .......................................................... 36Policy Convergence .........................................................................................................36Consumer Demand .......................................................................................................... 37Research on Fixed-broadband Adoption ................................................................................37Policy Factors ................................................................................................................ ..37Industry Factors ...............................................................................................................39Demographic Factors .......................................................................................................42ICT Factors ......................................................................................................................43Research on Mobile Broadband Adoption .............................................................................44Policy Factors ................................................................................................................ ..44Demographic Factors .......................................................................................................52ICT Factors ......................................................................................................................53

PAGE 6

6 3 ANALYTICAL FRAMEWORK AND RESEARCH QUESTIONS ..................................... 57Analytical Framework and Research Questions ..................................................................... 57Policy Factors ................................................................................................................ ..59Industry Factors ...............................................................................................................60Demographic Factors .......................................................................................................62ICT Factors ......................................................................................................................63Factors of Digital Divide and Network Effect ................................................................64Proposed Empirical Models ....................................................................................................65Fixed-Broadband ............................................................................................................. 65Mobile Broadband ...........................................................................................................70Ubiquitous Broadband ..................................................................................................... 714 RESEARCH METHOD .........................................................................................................74Measurement, Data and Statistical Me thods for Fixed-broadband Deployment .................... 74Policy Factors ................................................................................................................ ..74Industry Factors ...............................................................................................................75Demographic Factors .......................................................................................................76ICT Factors ......................................................................................................................76Other Factors ...................................................................................................................77Measurement, Data and Statistical Met hods for Mobile Broadband Deployment ................. 79Policy Factors ................................................................................................................ ..80Industry Factors ...............................................................................................................80Demographic Factors .......................................................................................................80ICT Factors ......................................................................................................................81Measurement, Data and Statistical Met hods for Ubiquitous Broadband Deployment ........... 82Policy Factors ................................................................................................................ ..83Industry Factors ...............................................................................................................83Demographic Factors .......................................................................................................84ICT Factors ......................................................................................................................845 RESULTS ....................................................................................................................... ........87Data and Descriptive Statistics ...............................................................................................87Regression Analysis of Fixe d Broadband Deployment .......................................................... 89Nonlinear Regression Model ...........................................................................................89Linear Regression model ................................................................................................. 97Results of Fixed-Broadband Deployment fo r Developed and Developing Countries ........... 98Regression Analysis: Developed Countries ............................................................................ 98One-way ANOVA Analysis of Fi xed-Broadband Deployment ...........................................100Regression Analysis of Mob ile Broadband Deployment ..................................................... 102Extended Model .............................................................................................................102Reduced Model ..............................................................................................................102One-Way ANOVA Analysis of Mobile Broadband Deployment ........................................103Regression Analysis of Ubi quitous Broadband Deployment ............................................... 104Model with Network Competition Variable .................................................................. 105

PAGE 7

7 Model with Different Platform Competi tion-S tandardization Policy Variables ........... 106Results of Ubiquitous Broadband Deployment for Developed and Developing Countries .108Regression Analysis: Developed Countries .................................................................. 108Regression Analysis: Developing Countries ................................................................. 110One-Way ANOVA Analysis of Ubi quitous Broadband Deployment .................................. 1116 DISCUSSION AND CONCLUSION .................................................................................. 113Summary of Results and Analysis ........................................................................................113Effects of Policy Factors on Broadband Deployment ................................................... 113Effects of Industry Factors on Broadband Deployment ................................................ 119Effects of Demographic/ICT Factors on Broadband Deployment ................................ 124Digital Divide and Broadband Deployme nt in Developed and Developing Countries .................................................................................................................... 126Implications .................................................................................................................. ........130Theoretical Implications ................................................................................................ 130Policy Implications ........................................................................................................134Limitations and Suggestions for Future Research ................................................................ 136LIST OF REFERENCES .............................................................................................................138BIOGRAPHICAL SKETCH .......................................................................................................147

PAGE 8

8 LIST OF TABLES Table page 1-1 Fixed-Broadband Penetration (Top 20 OECD countries) by Technology, December ...... 18 1-2 Mobile Broadband Penetration (Top 20 IT U m embership countries), December 2005 .... 19 1-3 Ubiquitous Broadband Penetration (Top 20 ITU m embership countries), December 2005....................................................................................................................................20 2-1 Cross-National empirical studies on fixed-broadband adoption factors ........................... 45 2-2 Cross-National empirical studi es on mobile adoption factors ........................................... 55 4-1 Variables, measurement and data so urces for fixed-broadband deploym ent .................... 78 4-2 Variables, measurement and da ta Sources for fixed-broadband ........................................ 79 4-4 Variables, measurement and data sour ces for ubiquitous broadband deploym ent ............85 5-1 Descriptive statistics for fixed broadband deploym ent ......................................................90 5-2 Descriptive statistics for mobile broadband deploym ent ................................................... 90 5-3 Descriptive statistics for ubiquitous broadband deploym ent ............................................. 91 5-4 Selected countries examined for fixed broadband deployment (ITU, 2005) ..................... 92 5-5 Countries examined for mobile broadband deploym ent (ITU, 2005) ................................93 5-6 Selected countries examined for ubiqu itous broadband deploym ent (ITU, 2005) ............ 94 5-7 Results of regressions of fixed broadband deploym ent .....................................................96 5-8 Results of regressions of fixed broadband penetration for developed and developing countries ..................................................................................................................... ......100 5-9 Difference in fixed broadband penetrati on and fixed broadband penetration growth rate by incom e, region, triple -play offerings and LLU .................................................... 101 5-10 Results of regression analysis of m obile broadband deployment .................................... 103 5-11 Difference in mobile broadband penetr ation and mobile broadband penetration growth rate by incom e, region, quadrupleplay offerings and licensing policy .............. 104 5-12 Results of regressions of Total (Ubiquitous) broadband deploym ent ............................. 107

PAGE 9

9 5-13 Results of regressions of total (Ubiquitous) broadband penetration for developing and developed Countries ..................................................................................................109 6-1 Significant policy factors of broadband deployment* .....................................................117 6-2 Significant industry factors of broadband deploym ent* .................................................. 122 6-3 Significant demographic/ICT f actors of broadband deployment* ................................... 127 6-4 Common and Different Significant Factors of Broadband Deploym ent** ..................... 130

PAGE 10

10 LIST OF FIGURES Figure page 2-1 Concept of ubiquitous broadband ......................................................................................35 3-1 Analytical framework for fixed-broadband deploym ent ...................................................57 3-2 Analytical framework for m obile broadband deployment ................................................. 58 3-3 Analytical framework for ubiquitous broadband deploym ent ........................................... 58 3-4 Broadband penetration 19992005 in Japan and Korea. .................................................... 66

PAGE 11

11 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy A CROSS-COUNTRY ANALYSIS OF UBI QUITOUS BROADBAND DEPLOYMENT: EXAMINATION OF ADOPTION FACTORS By Sangwon Lee August 2008 Chair: Justin S. Brown Cochair: Sylvia M. Chan-Olmsted Major: Mass Communication Broadband infrastructure is a key com ponent of the knowledge economy. Broadband connections on both fixed and mobile networks are becoming an indicator of the knowledge economy. A growing body of scholarship details cont ributing factors that may lead to broadband adoption. In spite of the growing body of litera ture about broadband adoption, these previous studies have the following limitations: 1) small nu mber of independent variables; 2) insufficient number of observations; 3) lack of theoretical background; 4) focus on only fixed broadband technology; and 5) inconsistent em pirical results. Empl oying the largest seconda ry data set, this study examines adoption factors of fixed and mob ile as well as ubiquitous broadband (fixed and mobile). The result of nonlinear and linear regressi on analysis of fixed broadband deployment suggests local loop unbundling (LLU) policy, plat form completion between different broadband technologies and other diverse industry, ICT (Information and Communication Technology) and demographic factors influenced fixed broadband diffusion. Specifica lly, the regression analysis of fixed broadband penetration found different types of LLU policies and previous fixed broadband penetration are signifi cant factors of fixed broadband deployment. Some of the

PAGE 12

12 significant factors of fixed broadband deployment are different in the developed countries and developing countries. The result of linear regression analysis of mobile broadband deployment suggests market mediated standardization policy, income, and 1G and 2G mobile penetr ation are significant factors of mobile broa dband deployment. Also, the result of linear regression analysis of ubiquitous broadband deployment suggests with other industry, ICT, and demographic variables, network competition between fixed and mobile and interactions of platform completion in fixed broadband markets and multiple standardization policy in mob ile markets are significant factors of ubiquitous broadband deployment. Some of the significant factors of ubiquitous broadband deploy ment were different in the developed countries and developing countries. Considering the result of this study, countries fostering broadband deployment need to adopt LLU policy for broadband, but the costs and benefits of LLU policy should be carefully considered. The results of this study also implies for the initial 4G mobile markets, whereby fixed and mobile broadband networks will be converged, governments need to be open to diverse competitive standards instead of government-manda ted standards. However, in the long term, industry-wide coordination and mutual le arning processes are more important.

PAGE 13

13 CHAPTER 1 INTRODUCTION Broadband Deployment and Knowledge Economy During the last part of the twen tieth century, there has been a steady growth in Internet and mobile adoption around the globe. Conti nuous technological innovations in the telecommunication industry enab le society to enter the era of convergence between broadband Intern et, wireless networks, and multimodal content and services. Broadband communications and infrastructure lie at the h eart of this trend. Broadband conn ections, on both fixed and mobile networks, are now recognized as indicators of the so-called knowledge economy. Widespread and affordable broadband access encourages in novation and economic growth in an economy, and attracts foreign i nvestment (ITU, 2003a). Although there exist various definiti ons of broadband, the International Telecommunication Union (ITU) defines broadba nd as a network offering a combined speed of equal to, or greater than, 256 kbit/s in one or both directions (ITU, 2005; ITU, 2006).1 According to the International Telecommunication Union (ITU), as of December 2005, telecommunication providers in more than 166 co untries offered fixed-broadband services and roughly 68 nations launched mobile-broadband se rvices (ITU, 2006). Fixed broadband may be defined as transmission capacity with sufficien t bandwidth to permit combined provision of voice, data, and video, with no lower limit th rough a fixed line (ITU, 2003b). 3G mobile systems, which provide higher transmission rates than possible in second generation wireless technologies, supporting data transport rates of at least 256 kbit/s for all radio environments, are 1 Initially broadband was defined as communication technologies that provide high-speed, always-on connections to the Internet for large numbers of residential and small-bus iness subscribers (Crandall, 2005; Fransman, 2006; ITU, 2003). This definition of broadband focuses on the fixed broadband technologies such as DSL and cable modem. The definition of broadband by the ITU network offering a combined speed of equal to, or greater than, 256 kbit/s in one or both directionsmay include more diverse broadband technologies such as mobile broadband.

PAGE 14

14 commonly referred to as mobile broadband (ITU, 2006; ITU, 2003b; Shelanski, 2003). Successful diffusion of fixed and mobile broadba nd is necessary for the provision of advanced IP-based services such as VoIP (Voice over Inte rnet Protocol) and IP TV (Internet Protocol Television) as well as mobile televi sion (Lee et. al, 2007; Lee & Brown, 2007). Technologies for Broadband Communications For broadband connectivity, either fixed m obile and portable Internet technologies may be employed. Fixed broadband is mainly implem ented through technologies such as digital subscriber line (DSL), cable modem, and fi bertothehome (FTTH) (ITU, 2003b). Mobile broadband is mainly implemented through tech nologies such as W-CDMA, CDMA 2000 1x EVDO, and HSDPA (ITU, 2006). Main portable Intern et technologies are wireless local area networks (WLAN), wireless metropolitan area networks (WMAN), and IEEE 802.16 (WiMAX). Fixed Broadband Thus far, for fixed-broadband, globally the dom inant platforms are DSL (64.34 %) and cable modem (29.89 %), though other platforms, su ch as fiber-to-the-home and other platforms serve around 6 percent (ITU, 2006). DSL can bring high-bandwidth information to homes and small businesses over ordinary copper telephone lines (ITU, 2005). DSL is distance-sensitive meaning th at speeds and signal qualities are influenced by the distance between the subscriber and lo cal exchange carriers nearest switching office (ITU, 2003b). Typically, the download speed of DSL services range from 256kbit/s to 6 Mbit/s, depending on DSL technology (Crandall, 2005). The most popular DSL technology is Asymmetric Digital Subscriber Line (ADSL), which is used mainly for Internet access, video on demand, and remote LAN access (ITU, 2006; Crandall, 2005). In terms of market share, DSL is typically the dominant fixed-broadband technology in most countries (OECD, 2007; ITU, 2006).

PAGE 15

15 Cable modems may provide high-speed interact ive services, including internet access, to be delivered over the cabl e television infrastructure, to give subscribers Internet speeds up to 1.2Mbit/s (ITU, 2006). Cable modem is a domin ant fixed-broadband technology in the United States and Canada (OECD, 2007; ITU, 2006). Fibertothehome (FTTH) generally refe rs to broadband telecommunications systems deployed on fiber-optic cables directly to homes or business (ITU, 2006). Fibertothehome is an enabling technology, which can offer the hi ghest speed fixed-br oadband connections (Crandall, 2005). In Japan, new FTTH subscr ibers outnumbered new DSL subscribers in 2005 (ITU, 2006). Mobile Broadband For m obile broadband, the dominant standa rds in operation are W-CDMA (60.04 %) and CDMA 2000 1x EV-DO (39.95 %) (ITU, 2006). W-CD MA (Wideband Code Division Multiple Access) is a third-generation mobile standard under the IMT-2000 banner, first deployed in Japan, also referred to as UMTS in Europe (ITU, 2006). Theoretically W-CDMA can achieve a data rate of 2 Mbit/s for low-mobility envir onment and 384 kbit/s for mobile systems (ITU, 2006). These are adequate speeds for broadband application such as downloading music and video to a handset (ITU, 2003b; ITU, 2006). HSDPA (High-Speed Downlink Packet Access) is an enhanced protocol to W-CDMA netw orks which boosts network capacity up to 14.4 Mbps (GSM Association, 2007; ITU, 2006). HSDPA technologies were first deployed in 2005 by AT&T in the United States (GSM Association, 2007). CDMA 2000 (Code Division Multiple Access 2000) is another third-generation digital cellular standard under the IMT-2000 banner, first deployed in Korea, which in cludes CDMA2000 1x and 1xEV-DO (Evolution, Data Optimized) (ITU, 2005). CDMA 1xEV-DO is current ly the dominant technology in the U.S. and

PAGE 16

16 Korea. Theoretically it can achieve a data rate of up to 2.4 Mbit/s and supports advanced data applications, such as MP3 transfers, video conferencing, and video downloads (ITU, 2006). Portable Internet Portable Internet can be defined as a platform for high-speed data access using Internet Protocol (IP) (ITU, 2004). Compared to the fixed and mobile broadband, portable Internet technologies can offer a better level of mobility than fixed-broadband services but a higher level of speed than mobile broa dband services (ITU, 2006). For portable Internet, wireless local area networks (WLAN), wire less metropolitan area networks (WMAN), and IEEE 802.16 (WiMAX) are main technologies (ITU, 2006). WLAN is a wireless network whereby a user can connect to a local area network (LAN) through a wireless connection, as an alterna tive to a wire-based local area ne twork. The most popular standard for WLAN is Wi-Fi (wireless fidelity) (ITU, 2006; ITU, 2004). WLAN technology enables mobile devices connect to a fixed-broa dband network via radi o links with an access point (ITU, 2006). WMAN is a wireless communications network that covers a geographic area such as a city or suburb (ITU, 2006; ITU, 2004). By increasing signa l power of the base stations, the technology can reach mobile devices at a c onsiderable distance (ITU, 2006). WiMAX is a fixed-wireless standard IEEE 802.16 that allows for long-range communications at 70 Mbit/s over 50 kilometres (ITU, 2004). It can be used as a backbone Internet connection to rural areas. WiMAX coul d eventually be combined with 3G mobile broadband to offer more customized high-speed environments (ITU, 2006). WiMax is being considered as a pre-4G standard with WiBro.2 In Korea, the first commercial mobile application 2 System beyond IMT-2000 or 4G mobile technology will be able to provide a comprehensive IP solution where voice, data and streamed multimedia (ITU, 2004). 4G w ill be achieved after wired and wireless technologies converge and will capable of providing 100 Mbit/sec and 1 Gbit/sec speeds both indoors and outdoors with premium quality and high security (Wikipedia, 2007).

PAGE 17

17 for WiMAX certified products took place with the launch of WiBro services (OECD, 2007d). WiBro employs the licensed 2.3 GHz frequenc y band with an 8.75 MHz channel bandwidth (OECD, 2007d). Both WiBro and Mobile WiMAX us e OFDMA, but the number of sub-channels and the frame structure differs between the two standards (OECD, 2007d). Current Status of Global Broadband Deployment Many countries are still in the early stages of broadband, as evidenced by the differences in deploym ent between countries. According to th e latest Organization fo r Economic Co-operation and Development (OECD) penetration data (December 2006), Denmark, Netherlands, Iceland, Korea, and Switzerland are leading broadband economies among OECD c ountries (see Table 11). According to the International Telecommunicat ion Union (ITU), there were approximately 216 million fixed-broadband subscribers and just over 60 million mobile broadband subscribers at the end of 2005 (ITU, 2006). According to the Organization for Economic Co-operation and Development the OECD, fixed-broadband adoption ove r the first 10 years is faster than previous services like cellular and dial-up services across OECD countries (OECD, 2006). There exists a wide range of mobile broadba nd diffusion levels across countries. As of December 2005, Korea, Italy, Japan, Portugal an d Hong Kong were the top five mobile broadband economies in terms of the 3G mobile penetration rate (ITU, 2006; see Table 1-2). WCDMA and CDMA 2000 are the two main standards for 3G wireless technologies (Gandal, Salant, & Waverman, 2003). Most of the European Community adopted WCDMA for 3G wireless services. By November 2005, almost 92 percent of the European 3G mobile customers subscribed to WCDMA technology-based services (ITU, 2005). On the other hand, many countries in the Americas, Asia, and Africa adopted CDMA 2000 or both CDMA 2000 and WCDMA in their 3G markets.

PAGE 18

18 Table 1-1. Fixed-Broadband Penetration (T op 20 OECD countries) by Technology, December 2006 DSL Cable Fiber/LAN Total Rank Total Subscribers Denmark 19.6 9.4 2.8 31.9 1 1,590,539 Netherlands 19.5 12.0 0.4 31.8 2 5,192,200 Iceland 28.8 0 0.2 29.7 3 87,738 Korea 11.4 10.7 7.0 29.1 4 14,042,728 Switzerland 18.8 8.8 0 28.5 5 2,140,309 Norway 21.7 3.8 1.5 27.7 6 1,278,346 Finland 23.5 3.5 0 27.2 7 1,428,000 Sweden 16.0 5.2 0 26.0 8 2,346,300 Canada 11.4 12.3 0 23.8 9 7,675,533 Belgium 14.0 8.4 0 22.5 10 2,353,956 U.K. 16.5 5.1 0 21.6 11 12,993,354 Luxembourg 18.2 2.2 0 20.4 12 93,214 France 19.1 1.1 0 20.3 13 12,699,000 Japan 11.1 2.8 6.2 20.2 14 25,755,080 United States 8.5 10.3 0.3 19.6 15 58,136,577 Australia 15.0 3.3 0 19.2 16 3,939,288 Austria 10.6 6.4 0 17.3 17 1,427,986 Germany 16.4 0.5 0 17.1 18 14,085,232 Spain 12.1 3.1 0 15.3 19 6,654,881 Italy 13.8 0 0.4 14.8 20 8,638,873 Note. Data were derived from Organizati on for Economic Co-operation and Development (2007a). For the measurement of fixed-broa dband penetration ra te, the total number of fixed-broadband subscribers per 100 inhabitants was employed. The ranking is based on the fixed-broadb and penetration rate. Source: OECD broadband st atistics. Paris: OECD.

PAGE 19

19 Table 1-2. Mobile Broadband Penetration (Top 20 ITU membership countries), December 2005 Mobile Broadband Penetration Rate Rank Total Mobile Broadband Subscribers Korea 25.95 1 12,530,945 Italy 17.67 2 10,262,000 Japan 13.89 3 17,792,600 Portugal 8.79 4 922,560 Hong Kong, China 8.19 5 576,500 Brunei Darussalam 8.11 6 30,000 U.K. 7.60 7 4,536,800 Sweden 7.31 8 660,790 Austria 6.84 9 559,000 Luxembourg 5.76 10 26,480 Ireland 4.94 11 205,200 Australia 3.97 12 801,100 Singapore 3.11 13 131,920 Germany 2.77 14 2,289,000 Israel 2.71 15 187,000 France 2.62 16 1,583,000 Norway 2.49 17 115,000 Denmark 2.27 18 123,210 New Zealand 2.24 19 90,300 Spain 2.20 20 939,000 Note. Data were derived from the Interna tional Telecommunication Union (2006). For the measurement of mobile broadband penetra tion rate, the total number of mobile broadband subscribers per 100 inhabitants was employed. The ranking is based on the mobile broadband penetration rate. Source: ITU Internet Reports 2006. Geneva: ITU. Thus, with the deployment of higher-speed mob ile networks such as 3G, users can access networks at any time, though always-on connect ivity by fixed and mobile technologies (ITU, 2005). Now ubiquitous broadband connections th rough both fixed and mobile networks are becoming the norm in the knowledge economy (ITU, 2006).3 Also, in the near future, it is expected that all wired and wireless comm unications will converge with Next Generation Networks (NGN) (OECD, 2007c). In this contex t, it is necessary to examine the extended definition of broadband-ubiquit ous broadband including both fixed and mobile broadband. 3 Vision of a ubiquitous broadband network is broadband connection anytime, anywhere, by anyone and anything (ITU, 2005).

PAGE 20

20 Table 1-3. Ubiquitous Broadband Penetration (Top 20 ITU membership countries), December 2005 Total Broadband Penetration Rate Rank Total Broadband Subscribers Korea 51.20 1 24,721,656 Hong Kong, China 31.80 2 2,235,598 Japan 31.40 3 40,157,748 Italy 29.40 4 17,082,048 Sweden 27.60 5 2,498,790 Netherlands 27.20 6 4,437,000 Iceland 26.50 7 78,017 Switzerland 24.50 8 1,830,446 Norway 23.90 9 1,106,352 Finland 23.80 10 1,251,700 United Kingdom 23.60 11 14,076,700 Austria 21.20 12 1,733,000 Luxembourg 21.00 13 97,713 Canada 20.90 14 6,741,699 Taiwan, China 20.60 15 4,716,103 Israel 20.50 16 1,416,626 Portugal 20.30 17 2,134,594 Belgium 19.30 18 1,992,791 Singapore 18.40 19 794,420 France 18.30 20 11,048,600 Note. Data were derived from the Interna tional Telecommunication Union (2006). For the measurement of total broadband penetration ra te, the total number of fixed and mobile broadband subscribers per 100 inhabitants was employed. The ranking is based on the total broadband penetration rate. Source: ITU Internet Reports 2006. Geneva: ITU. There were over 275 million total broadband subscribers including fixed and mobile broadband at the end of 2005 (ITU, 2006). In term s of the total broadb and penetration rate including fixed and mobile broadband subscribers, there were 4.3 subscribers per 100 inhabitants at the end of 2005 (ITU, 2006). As of Decembe r 2005, Korea, Hong Kong, Japan, Sweden and Netherlands were the top broadband economies in te rms of the total broadband penetration rate (see table 1-3) (ITU, 2006).

PAGE 21

21 Significance of the Study A growing body of scholarship details contributing factors that m ay lead to broadband adoption. Some empirical studies find that plat form competition (inter-modal competition), local loop unbundling (LLU), income, and demographic va riables such as population density increase fixed-broadband diffusion (Dis taso et al., 2006; Garcia-Mur illo, 2005; Grosso, 2006; Lee, 2006b). Many countries have considered local loop unbundling (LLU)4 and facilities-based competition as important policy in itiatives to promote rapid fixe d-broadband diffusion. Platform competition (facilities-based comp etition among several different broadband platforms) is often thought to be crucial for reducing prices, improving the quality of service, increasing the number of customers and promoting investment a nd innovation (ITU, 2003b; DotEcon and Criterion Economics, 2003). There is a growing body of em pirical research about fixed-broadband diffusion. In their study of 30 OECD count ries, Cava-Ferreruela and Alabau-Mu oz (2006) find that technological competition, low cost of deploying infrastructures, and predilection to use new technologies are key factors for broadband s upply and demand. Analyzing data from 14 European countries, Distaso et al. (2006) argue that inter-platform competition drives broadband diffusion, but that competition in the DSL market does not play a significant role. Using logit regression, Garcia-Murillo (2005 ) finds that unbundling an incumbents infrastructure only results in a substantial increas e in broadband deployment for mi ddle-income countries, but not for their high-income counterparts. Kim and others (2003) suggest the pr eparedness of a nation and the cost conditions of de ploying advanced networks are the most consistent factors 4 Local loop unbundling refers to the process of requiring incumbent operators to open, wholly or in part, the last mile of their telecommunications networks to competitors (ITU, 2003b; OECD, 2003).

PAGE 22

22 explaining broadband uptake in OECD countries. Using generalized least squares, Grosso (2006) finds that competition, income, and unbundling increase broadband diffusion. Despite the growing body of literature about broadband adoption, these previous studies have the following limitations: 1) small number of independent variables; 2) insufficient number of observations; 3) lack of th eoretical background; 4) focus on only fixed-broadband technology; and 5) inconsistent empirical results. The previous empirical studi es on fixed-broadband adoption employed only a limited number of indepe ndent variables with insufficient data. Also, previous studies have not covered some important independent variables such as institutional environment, international Internet bandwidt h, telecommunication infras tructure investment, content, and age. Previous studies did not pr opose refined and diverse theoretical frameworks like platform/network competition, network effects, path dependence, and leapfrogging. Also, previous studies focused only on fixed-broadband technology a nd didnt include the mobile broadband from the analysis. In addition, the results of these empirical studies are not consistent.5 Convergence in application and services technology, industry, policy, consumer demand and multiple-play strategies have lead to th e expansion of the definition of broadband. As broadband technology is developed, the definitio n of broadband is extended as a network offering across various types of platforms incl uding both fixed and mobile broadband access at the same time (ITU, 2006). Also fixed and mobile networks will be converged in the Next Generation Networks (NGN) (OECD, 2007c). To incorporate this trend, this study also exam ines influential factors that lead to the deployment of this extended definition of br oadband-ubiquitous broadband, which includes 5 For instance, the effects of local loop unbundling policy s till are not clearly understood and the results of empirical studies are consistent.

PAGE 23

23 both fixed and mobile broadband. Currently no cr oss-cultural empirical work exists that incorporates and measures factor s that affect fixed and mobile broadband adoption. Also, it is interesting to examine whether network comp etition between fixed and mobile broadband network has influenced broadband deployment. Employing more independent variables based on refined theoretical framework with the largest data sets, this study examines adoption fa ctors of fixed and mobile as well as ubiquitous broadband (fixed and mobile). Purpose of the Study Current study exam ined why broadband adoption differs among countries. Diverse macrolevel factors such as policy, industry, technology, consumer demographics, ICT, and institutional environment might influence new media tec hnology adoption. In particular, regulation and competition policy are important macro-level factors of new media technology adoption. Regarding fixed-broadband deployment, ma ny countries have considered local loop unbundling (LLU) and facilities-based competition as important policy initiatives to promote rapid fixed-broadband diffusion. It is widely held that platform competition (facilities-based competition among several different broadband pl atforms) is crucial for reducing prices, improving quality of service, increasing cust omers and promoting investment and innovation (DotEcon & Criterion Economics, 2003). Regarding mobile deployment, there have b een debates on whether single or multiple standards policy is more effective for faster adoption of mobile. Competition among different mobile standards might be a key driver of mobile broadband deployment (Lee & Marcu, 2007). Regulation and competition policy are also importa nt factors in the deployment of ubiquitous broadband (including fixed and mobile broadb and at the same time). LLU and network competition between fixed and mobile networ ks with other factors ICT, technology,

PAGE 24

24 consumer, industry, and institutional factors might be key drivers of ubiquitous broadband deployment. Using a large data set from the ITU (International Telecommunication Union) and the OECD (Organization for Economic Co-operation and Development), this empirical study examines determinants of the global fixed, mobi le and ubiquitous broadband deployment. Based upon the results of this empirical research, this dissertation suggests policy and strategy implications. Also this study has implications for the deployment of future 4G or Next Generation Networks (NGN), whic h fixed and mobile broadband services will be offered over the same network.

PAGE 25

25 CHAPTER 2 LITERATURE REVIEW This section discusses scholar ship that addresses theoretical background and influential factors of broadband adoption. Typically broadband a doption research either exam ines factors at a micro-individual level or macro-national level (b etween countries) or case analysis, employing both quantitative and qualitative methodologies. While this review focuses on empirical studies regarding broadband adoption between countries, micro-individual level and case stud ies are also discussed. Based upon the classification of exis ting research, policy, indus try, demographic, and Information and Communication Technology (ICT) a nd institutional factors may influence fixed and mobile broadband deployment. This section al so presents theory and research related to network effects, compatibility, path dependence, leapfrogging and di gital divide issues. Theoretical Backgrounds of Broadband Adoption Micro-Individual Level Approaches In the m icro-individual level, socio-psychological factors may influence broadband adoption. Rogers theorized that innovations would spread through society in an S-curve, as the early adopters select the tec hnology first, followed by the majority, until a technology or innovation is common (Rogers, 2003). In explaini ng the diffusion of innovation, Rogers (2003) identified five perceived char acteristics of an innovation su ch as relative advantage, compatibility, complexity, trialability and obser vability. The relative ad vantage of an innovation refers the degree to which an innovation is percei ved as being better than the idea it supersedes and this construct is one of th e best predictors of the adop tion of an innovation (Rogers, 2003). Compatibility can be defined as the degree to which an innovation is perceived as consistent with the existing values, past experiences and needs of potential adopters (R ogers, 2003). Complexity can be defined as the degree to which an innovation is perceived as relatively difficult to

PAGE 26

26 understand and use (Rogers, 2003). Trialability is the degree to which an innovation may be experimented with on a limited basis (Rogers, 20 03). Observability refers the degree to which the results of an innovation are vi sible to others (Rogers, 2003). These five perceived characteristics may influence the broadband adoption in the individual level. Relating to fixed-broadband adoption, Oh et al. (2003) finds that innovation attributes, such as compatibility, visibility a nd result demonstrability and perceived usefulness, perceived ease of use and perceived resources based on the extended technology acceptance model have influenced fixed-broa dband adoption in Korea. The Technology Acceptance Model (TAM) suggest s that perceived us efulness (PU) the degree to which a person believes that using a pa rticular system would enhance his or her job performance and perceived ease-of-use (PEOU) the degree to which a person believes that using a particular system would be free from effort may influence technology acceptance by users (Davis, 1989). Through employing the Technology Acceptance M odel (TAM) model, Pagani (2004) found perceived usef ulness, ease of use, price, a nd speed are the most important determinants of adoption of 3G multimedia mobile services. Recently, based on the uses and gratifications perspective, Chang et al. (2006) found that perceived needs for passing time was a significant factor in explaining the difference between adopters and non-adopters of online games. However, this individual level of analys is on the new media technology adoption cannot explain why there is a difference in broadband diffusion between countries. Platform Competition In the m acro-national level of analysis, dive rse dimensions of competition are useful in explaining broadband adoption. Porter (1979)s five forces model could be a micro-level basis for macro-level analysis.

PAGE 27

27 Porter (1979) suggested five forces may dete rmine the competitive intensity and therefore attractiveness of a market. Porter referred to these forces as the micro-environment, to contrast it with the more general te rm macro-environment (Porter, 1979). Five forces include three forces from horizontal competition, namely threat of substitute products, threat of established rivals, and threat of new entrants. The two remaini ng forces are based upon v ertical competition the bargaining power of suppliers and bargaini ng power of customers (Porter, 1979). Among Porters five forces, the concept of horizontal competition is a useful micro-level basis for explaining different dimensions of competition at the macro-level. In the broadband markets, the threat of substitute products can be transformed into platform competition (intermodal competition-competition between different technologies ). The threat of established rivals can be transformed into market competition, and the threat of new entrants could be related to the intramodal competition thro ugh open access policy. Platform competition is an important theoretic al basis of this study. Platform competition occurs when different technologies compete to provide telecommunication services to end-users (Church & Gandal, 2005). Platform competition in network industry involves competition between technologies that are not only differentiated, but also are competing networks (Church & Gandal, 2005). Strong platform competition among different technologies may lead to lower prices, increased feature offerings, and more extensive broadband networks (ITU, 2003a). Regarding mobile broadband, platform co mpetition could be related to the marketmediated standard policy. Though market-mediate d standardization policy might lead to limited network externalities and economies of scale, multip le standards and different types of services across technologies enable the ex istence of diverse competing systems, which may lead to more and better services in the ma rket (Gruber & Verboven, 2001b). Market-mediated standard policy

PAGE 28

28 may also bring differentiated services and lo wer prices to customers (Gruber & Verboven, 2001b; Church & Gandal, 2005). Relating to ubiquitous broadband, which in cludes both fixed and mobile broadband, platform competition could be related to the competition between fixed and mobile network. This competition between networks might lead to lower prices, improving the quality of service, increasing the number of customers and prom oting investment and innovation (ITU, 2003b; DotEcon and Criterion Economics, 2003). With other dimensions of competition such as intramodal competition and market competition (mar ket concentration), platform competition might be a key driver of broadba nd adoption in many countries. Network Effect Network effect m ay also be related to th e broadband adoption. Network effect is the circumstance in which the net value of an acti on is affected by the number of agents taking equivalent actions (Liebowitz a nd Margolis, 1994). In other words, network effect means the fact that higher usage of certain products or services makes them more valuable. One consequence of network effect is that the purchase of a good by one individual indirectly benefits othe rs who own the good. This t ype of side-effect in a transaction is known as an externality in economics, and externaliti es arising from network effects are known as network externalities (Church & Gandal, 2005). The resulting bandwagon effect is an example of a positive feedback loop (Rohlfs, 2001). For products characterized by network effects the decision by consumers regarding which network to join will depend not only on relative product characteristics and prices, but also the expected size of the network (Church & Gandal, 2005). The role of the size of the exis ting installed base in determining the size of the network in the future arises because positive network effects give rise to positive feedback effects (Shapiro,

PAGE 29

29 & Varian, 1999). These positive feedback effects create a strong tendency for the strong grow stronger in a virtuous cycle. The greater the installed base, the greater network benefits, the more attractive the network to adopters, the grea ter adoption and the greater the installed base, etc. (Shapiro, & Varian, 1999; Church & Gandal, 2005). If network effect exists in the use of broadband, new subscriber s joining a broadband network might influence the utility of current subscribers (Madden et al., 2004). Network effects might suggest that current subscription is positively correlated to the previous subscription (Economides & Himmelberg, 1995). Madden et al. (2004) found that these network effects have influenced mobile telephony subscription. Howeve r there is no empirical work to test the existence of network effects on fixed-broadband subscription. Rohlfs (2001) suggested a form of network e ffects that he calls bandwagon effects. He suggests that as more and more people subscr ibe new media technology such as VCRs, personal computer, mobile and broadband Internet, others are attracted to it (Rohl fs, 2001; Haring et al., 2002). Recently Jang et al. (2005) found that the pattern of diffusion of mobile telecommunications for OECD countries is gene rally characterized by an S-shaped curve; nevertheless, significant differences exist in the spread of the Scurve, largely due to differences in the magnitude of the network externality coefficient. Digital Divide and Leapfrogging Theory Digita l divide is also related to the broa dband adoption issue. The development of new media technology that is valuable to consumers might create concerns for consumers who cannot or will not purchase it (Crandall, 2003). In gene ral, digital divide could be defined as the dichotomy between those who have computer s and Internet access and those who do not (Warschauer, 2003). Rice and Katz (2003) identi fies and analyzes three kinds of digital divides for both the Internet and mobileusers/nonuser, veteran/recent, and continuing/dropout.

PAGE 30

30 Employing a national representative telephone survey of American s in 2000, they found similarities and differences among those digital di vides that are based on demographic variables (Rice &Katz, 2003). Through an analysis of Australian household survey data, Madden et al. (1996) found there is information inequality (d igital divide) resulted from broadband network access. Recently using comprehensive U.S. da ta covering all forms of fixed-broadband technology, Prieger (2003) fi nds unequal broadband availa bility in areas with high concentrations of poor, minority, or rural househo lds. More recently using a panel data from 161 countries over 1999-2001 period, Chi nn and Fairlie (2006) finds global digital divide is mainly explained by income differentials. However ther e is no empirical study on the broadband digital divide that employs a crosscountry data analysis. Digital divide concerns of infrastructure deployment of communication technologies are also related to the l eapfrogging theory, whereby developing countries ma y skip inferior, less efficient, m ore expensive or more polluting tech nologies and industries and move directly to more advanced ones. According to Steinm ueller (2001), leapfrogging can be defined as bypassing stages in capacity building or invest ment through which countries were previously required to pass during the pro cess of economic development. Developing countries generally can not afford to create or rebuild basic ICT infrastructure for new communication technologies. However, becau se of the lack of investment in legacy systems, hardware, and software, they can be in a good position to leapfrog over some of these incremental steps (Lee, 2006). Si nce Internet technology makes it eas ier to develop ICT services, current technologies offer great potential to de veloping countries looking to introduce ICT into their region (Steinmueller, 2001).

PAGE 31

31 Relating to the new media technology, leap frogging can be applied to the new media technology adoption such as fi xed-broadband. Broadband may have an effect on the digital divide between countries: as developed countries adopt broadband, the current trend is for them to pull even further ahead of developing ones, thus increasing the di gital divide (ITU, 2003). However, as the unit costs of providing broadba nd become cheaper, some developing countries may be able to use the technology to leapfrog ah ead, providing an integrated voice, video and data network using the same infrastructure (ITU, 2003). The technology of wireless LANs, in particular, can be used successfully to extend broadband access (ITU, 2003). However, in developing countries, introduci ng more advanced new media technology such as 3G mobile is a consider able burden (Lee, 2006). One pr oblem hindering the successful introduction of future 3G mobile in developing countries is the relativ ely higher cost of 3G services than 2G services for cons umers in those countries (ITU, 2001) Recently an empirical study suggested that an established infrastructure for relevant information and communication tec hnologies is an influential f actor for 3G mobile diffusion (Lee et al., 2007). This result of the empirical study might be different from the initial diffusion of 1G or 2G mobile systems which exhibited the phenomenon of leapfrogging in many countries where the mobile penetration rates significantly outpaced the demand for ot her ICT services (Lee et al., 2007). More refined empirical research on the leapfrogging and new media technology adoption such as broadband is necessary. Path Dependence Path dependence theory can be related to the policy c hoice for broadband diffusion. In general, path dependence refers to the dependenc e of a system or network on past decisions of producers and consum ers (Margolis & Liebow itz, 1990). Margolis and Liebowitz (2000) distinguish between different t ypes of path dependence. Some types of path dependence do not

PAGE 32

32 imply inefficiencies and, while they may be interesting to study for other reasons, do not challenge the policy implicati ons of neoclassical economics.1 The case of the QWERTY keyboard (David, 1985) is a well-known example of pathdependence theory. The current dominance of the QWERTY keyboard today is not thought to be due to its superiority for typing but because it was invented earl ier than the Dvorak keyboard(David, 1985; Bensen & Farrell, 1994; Katz & Shapiro, 1994; Shapiro, & Varian, 1999). Paul David called this pat h dependence, and argued that in ferior standards can persist simply because of the legacy they have built up. The claim of path dependence, at least as it applies to public policy, is that people often either i gnore those interconnecti ons or only look at them in a narrow and myopic manner, and so they get locked into bad solutions. Relating to the mobile communications standa rdization policy, in spite of EUs success story of mandated standard (GSM ) in initial stage of 2G, if market-mediated standard policy leads to faster adoption of 3G mobile in the markets, it implies that the EU 3G standard policy that may have assumed WCDMA or locked opera tors into WCDMA might be a very costly public policy decision (Gandal et al., 2003). In this situation, EU countries would be better off if the countries adopt multiple standards simultane ously, but it would be difficult for them to coordinate involves such a ch allenge (Margolis & Liebowitz, 2000). Path dependence induces an inefficiency arising from small differences in initial public policy-making, which lead to outcomes that are likely to be costly to change (Liebowitz & Margolis, 1995) 1 They contrasted this category both to first-degree path dependence, which ha s no implications for efficiency, and to second-degree path dependence, where transactions cost s and/or the impossibility of foresight lead to outcomes that offer lower payoffs than some hypot hetical -but unattainable -alternativ e. Only what they call third degree path dependence a situation where society would be better off if everybody switched standards simultaneouslymay lead to inefficiencies.

PAGE 33

33 Drivers of Ubiquitous Broadband Concept A purpose of this study is to exam ine adoption factors of ubiquitous broadband. The concept of ubiquitous computing was first us ed in 1991 by Marc Weiser (1991). He pointed to the invisibility of technology through the tr ansformation of everyday items into small computers (Weiser, 1991). In the first paradigm of computing, mainframes were shared by many people (one computer for ma ny people) (ITU, 2006). Now we are moving from the personal computer era (one computer per person) to procee d to the phase of the ub iquitous computing era (many computers per person) (ITU, 2006). With the development of mobile technologies, this notion of ubiquitous computing can be applied to the ubiquitous broadband access, wh ich means multiple broadband accesses per person with multiple devices. Under the ubiqui tous broadband environment, broadband access anytime, anywhere, by anyone and anything through both fixed and mobile broadband is possible (see figure 2-1). For the ubiquitous br oadband, network competition between fixed and mobile networks might be a driver of broadba nd deployment. Also, for the ubiquitous broadband, platform competition among different fixed br oadband technologies and standard competition between different mobile standards as well as intra-modal competition in a technology are possible. These multiple modes of competition may suggest implications for competition policy and regulation for broadband infrastructure. Multiple play strategy, which is a marketing co ncept for delivering multiple services over a single access network, is also a driver of ubi quitous broadband concept. For instance, many telecommunications companies and cable operato rs launched quadruple play service, which offers fixed-broadband access, television, fixe d telephone, and mobile broadband service in a bundled package.

PAGE 34

34 Also, in the near future, in the environm ent of the Next Generation Networks (NGNs) which will be achieved after fixed and mobile networks are converged by fully IP (Internet Protocol)-based integrated system, the notion of ubiquitous broadband could be replaced by Converged Ubiquitous Broadband (Cubiquitous Broadband) over a network (see figure 2-1). In this NGN environment, single or multiple sta ndardization policy is po ssible in a country. In this context, examination of adoption fact ors of ubiquitous broadband is necessary. Underlying forces of this notion of ubiquitous broadband may be attributed to convergence in services, technology, industry, policy, consumer demand, and multiple play strategy. These underlying forces might be drivers of ubiquitous broadband concept. Application and Service Convergence Initially the m ain application of fixed-br oadband was Internet access, but now fixedbroadband offers more diverse appl ications and services such as voice, data and video services (e.g. IP TV) (ITU, 2003). Also, in the early develo pmental stage of mobile main application and services was voice, but current 3G mobile can pr ovide more diverse applications and services such as Internet, data and video se rvices (e.g. Mobile TV) (ITU, 2001). Currently both fixed and mobile broadband services offer the same or similar application and services, which might be differentiated. In addition, in fixed and mobile converged environment, this trend of application and se rvice convergence will be accelerated (OECD, 2007c). Technological Innovation and Convergence W ith the development of technology, higher bandwidth and technological convergence is possible (ITU, 2004; ITU, 2005; ITU, 2006; Ki m, 2005). Technological innovation and higher bandwidth lead to broadband access anytime, anywhere and by anyone and anything through

PAGE 35

35 Figure 2-1. Concept of ubiquitous broadband Fixed Telephone Cellular Phone Mobile Broadband Portable Internet Fixed Broadband PC/ Narrowband Ubiquitous Broadband (Fixed, Portable Internet, and Mobile Broadband Accesses) Network (Infrastructure) Competition 1G 2G 2.5G 3G W-CDMA HSDPA CDMA-2000 (Single or Multiple Standardization Policy) Wi Fi Fiber DSL Cable Modem ( Platform com p etition ) Ubiquitous One computer Computing for one person One broadband access for one person Multiple broadband accesses for one person with multiple devices Converged Ubiquitous Broadband (Cubiquitous Broadband) 4G Mobile Technology Fixed Mobile Convergence (FMC) Next Generation Networks (NGNs) Packet-based IP Network (Single or Multiple Standardization Policy) Voice Data Video PSTN (Mobile TV) Video Voice Data (IP TV) (VoIP)

PAGE 36

36 both fixed and mobile broadband (ITU, 2006). Technological innovation also may lead to the ubiquity of digital technology (many computer s for one person) (ITU, 2006; Weiser, 1991).As fixed-mobile convergence develops, voice, data, and video services using the fixed broadband and mobile networks become interrelated (OECD, 2007c). Industry Convergence and Multiple Play Strategy Horizontalization of businesses m ay be a ke y factor of converged, ubiquitous broadband. Collis et al. (1997) suggested convergence is transforming the multimedia industry from three vertical businesses (telephone, te levision, and computer) to five horizontal segments such as content, packaging, transmission network, manipulation infrastructu re, and terminals. Friedman (2006) suggests that businesses had to begin collaborating horizontally as opposed to vertically, which meant that companies and people had to start collaborating with other departments or companies in order to add value creation or innovation. Multi-play strategy might also lead to industr y convergence and be a driver of broadband diffusion. The strategy of triple or quadruple play has been touted as one of the most significant strategic moves for both cable ope rators and telephone companies in the last few years (Time Warner, 2007). In this context, bundling strategy might be a f actor in broadband deployment. Multi-play services through bundling strategy continue to gain popularity in global telecommunication markets. By September 2005, tr iple play services were available from 48 providers in 23 OECD countries and quadruple-play offers (i ncluding mobile voice) were available in 10 OECD countries (OECD, 2006b). It was said that bundling might increase customer retention and average reve nue per user (ARPU) (Krauss, 2006). Policy Convergence Wirth (2006) suggests that because technology innovation and standardization, changing consumer characteristics, search for syner gy, bandwagon effect, and content repurposing, policy

PAGE 37

37 is a driver for convergence. Policy and regulatory responses to media convergence and new media technology may lead to integration of polic y and regulatory system, which might lead to converged services (Sawng et al., 2005). Consumer Demand Consum er demand for diversity, mobility, highe r speeds, interactivit y, personal access, and converged multimedia services may be factors lead ing to ubiquitous broadband. In general, end users are interested in services and applications when underlyi ng technology (platform) is not important to end users (Kim, 2005). Dowling et al. (1998) suggests that there are three dimensions of convergence: consumer demand, industry/firm supply, and technology. It was also put forward that a converged product that can fu lfill new or existing consumer needs in a novel, more convenient, and/or less expensive way is crucial to the success of its offering (Picard, 2000). Thus, consumer demand may influe nce converged ubiquitous broadband. Research on Fixed-broadband Adoption Policy Factors On the supply side, m any countries have c onsidered local loop unbundling regulation as important policy initiatives to promote ra pid broadband diffusion. Lo cal loop unbundling (LLU) which refers to the process by which incumbent carriers lease, wholly or in part, the local segment of their telecommunications network to competitors has been considered an important policy to stimulate intra-modal competition (OECD, 2003). Implementation of LLU widely differs among countries. Types of LLU full unbundling2, line sharing3 and bit stream 2 Full unbundling (physical acce ss to raw copper) exists where an incumbent provides full access to its raw copper (OECD, 2003). With full unbundling new entrants take full control of the copper pairs and can provide both voice and DSL services. However, the incumbent still retains ow nership of the unbundled loop and is responsible for maintenance (OECD, 2003).

PAGE 38

38 access4 and LLU prices are different across count ries (OECD, 2003). There are arguments for and against local loop unbundling. LLU policy might introduce competition in the DSL markets and prices might fall when incumbent carriers are compelled to open up their networks to competitors (ITU, 2003a). Thus, LLU may generate c onsumer benefits in the near future through open access to competitors (Frieden, 2005a). However, LLU may confiscate incumbents property and reduce their incenti ves to invest in advanced telecommunication technologies (Frieden, 2005a). There have been a lot of debates on the e ffects of LLU in the United States. Hausman (2001, 2002) claims LLU regulation in the U.S. ha s impeded the incumbents deployment of the network facilities require d for DSL, conveying competitive advantages and market share to cable operators providing broadband cable modem services. Through an empirical analysis of CLECs investment plans and an event study that explores the impact of the Ta uzin-Dingell bill on share prices, Glassman and Lehr (2001) found that reduction of network unbundling for broadba nd deployment places downward pressure on the competitive carriers equity prices, there by reducing investment by entrants in network facilities. Employing logit regression analysis from selected ITU countries, Garcia-Murillo (2005) found unbundling an incumbents infrastructure only results in a substantial improvement in broadband deployment for middle-income countries, but not for their high-income counterparts. Distaso et. al. ( 2006) also found LLU price is an explanatory variable of fixed3 Line sharing (shared access) allows an incumbent to ma intain control over copper pairs while new players can lease part of the copper pair spectrum for data services including Integrated Services Digital Network (ISDN) and DSL (OECD, 2003). However, there are some techni cal concerns such as line noise (OECD, 2003). 4 Bit stream access (wholesale access) is a type of wholes ale arrangement in which new entrants have no managing control over the physical lines and are not allowed to install their own equipment(OECD, 2003). The new entrants generally do not favor this type becau se, unlike full unbundling and line sharing, they can on ly provide th e services that the incumbent designates (OECD, 2003).

PAGE 39

39 broadband adoption. Recently thr ough regression analysis of OECD countries data, Grosso (2006) found LLU have influenced fixed-broadband deployment. Throug h their empirical study with 179 observations, Wallsten (2006) found unbund ling is a key driver of fixed-broadband adoption in OECD countries. In spite of a growing body of literature on the effects of LLU policy on broadband deployment, the effectiveness of different types of LLU policies such as full unbundling, line sharing, and bit stream access and th e impact of LLU price regulation have not been clearly understood. Cross-ownership also might influence broadband deployment. An ITU report suggests some broadband markets have struggled with broadband deployment because of cross-ownership issues (ITU, 2003a). If a provider owns portio ns of a cable company and telecommunication company, the provider has disincentives to ro ll out both DSL and cable modem (ITU, 2003a). For successful broadband deployment, government role is also important. Es pecially in the rural regions, there is often insufficient private investment for network construction. In these high cost regions, full public funding for infrastructure or public-owned network is often the only option (ITU, 2003b). Institutional environment might also infl uence broadband deployment. Using ordinary least-squares hierarchical regr ession analysis, Andonova (2006) fi nds institutional environment such as political rights and civi l liberties are correlated with deployment of th e Internet. In spite of these previous studies, there was no empirical study, which tests the infl uences of cross-media regulation, and institutional environmen t on fixed-broadband diffusion. Industry Factors A few studies argue th at inter-modal comp etition (platform comp etition among different technologies) with other factor s in the supply side of the br oadband market increase broadband adoption. Aron and Burnstein (2003) suggest that broadband availabil ity in a state is driven by

PAGE 40

40 inter-modal competition and cost factors, but not by the raw availability of broadband services. Using U.S. state data in 2000, they found that the independent effect of direct, inter-modal competition is associated with increased househol d subscription to broadband services (Aron & Burnstein, 2003). Recently, through two different econometric anal yses (time-series analysis and multipleregression analysis) using data from 50 states, Lee (2006) suggests platform competition and the availability of different broa dband platforms have influenced broadband diffusion in the United States. Through panel data analysis, Denni and Gruber (2005) find that inter-platform competition, intra-platform competition in the DSL market, and telecommunication density have a positive impact on broadband diffusion in the United States. Beyond research that assesses industry fact ors that contribute U. S. broadband adoption, several studies compare multiple factors of broadband adoption among countries. From the analysis of EU membership countries data, a report from DotEcon & Criterion Economics shows that inter-modal competition among platfo rms rather than access-based market entry increases the adoption of broadband. This repor t suggests broadband penetration tends to be higher in European countries wh ere DSL and non-DSL platforms ha ve similar market share, but the report was not supported by statistical methods (DotEcon & Criterion Economics, 2003). Based upon analysis of data from 14 European countries, Distaso, Lupi and Maneti (2006) demonstrate that inter-platform competition drives broadband a doption, but that competition in the DSL market does not play a significant role. Through a comp arative study of Korea and the United States, Lee and Chan-Olmsted (2004) suggest a combination of policy, consumer demands, and technological factors supporte d by broadband-related industry could make differences in broadband deployment among co untries. Through statistical analysis of

PAGE 41

41 approximately 100 countries, Garcia-Murillo (20 05) found fixed-broadband price, income (GDP per capita), competition have been influen tial factors of fixed-broadband adoption. Cava-Ferreruela and Alabau-Mu oz (2006) suggest technological competition, low cost of deploying infrastructures, and pred iction of the use of new technologies might be key factors for broadband supply and demand, respectively. Gr osso (2006) found competition and LLU influenced broadband penetration among OECD countries. More recently Fransman (2006) suggested disruptive competitors are an im portant determinant of global broadband performance.5 More recently employing multivariate analysis of 110 country data, Lee and Brown (2007) find platform competition, broadband speed, and cont ent contribute global broadband adoption. They also find the impacts of platform competiti on are strong when market share of dominant technology and non-dominant technology is similar (Lee & Brown, 2007). Recently Ridder (2007) found low fixed-broadband price is correlated to the high level of broadband diffusion. More recentl y, Atkinson et al. (2008) also found low level of broadband price is factor of broadband adoption in OECD countries. Higher bandwidth6 also might be correlated with th e broadband adoption. Growth in demand for higher capacity is a key driver of broadband diffusion (ITU, 2006). Fansman (2006) suggests capacity of broadband is a measure of national performance in broadband. In spite of 5 Disruptive competition can be defined as existing when competitors to the incumbent in the markets have been so aggressive with their pricing that they even do not cover their marginal costs in the markets and end up making short-run losses (Fransman, 2006). Fransman (2006) argues that a main reason for superior broadband performance of Korea and Japan is results of disruptive competition. 6 Bandwidth is a measure of how fast data flows on a gi ven transmission path, and determines the quantity and the speed of information transmitted (ITU, 2003b).

PAGE 42

42 importance of bandwidth, there was no empiri cal work, which tests correlation between bandwidth and broadband deployment. Telecommunication infrastructure investment from private and public sector is a contributing factor of telecommunication network deployment (ITU, 2003b). Some top broadband economies such as Korea and Sweden employed national deployment strategies to promote infrastructure investment from public and private sector (ITU, 2003a). Demographic Factors Som e empirical studies on fixed-broadband depl oyment suggest demographic factors such as income and population density have influenc ed fixed-broadband adoption. In addition to the supply-side research, several em pirically-driven studies illustra te the demand side of broadband adoption in the U.S. Through data analysis of a national sample of U.S. households, Rappoport, Kridel, Taylor and Alleman (2001) found that pri ce elasticity of demand for broadband service is much greater than narrowband service. Using an estimation of an economic model based on statistical data from 2000 to 2001, Crandall, Sida k and Singer (2002) showed that the decision to use a broadband connection depends on the opportunity cost of time for the user and intensity of Internet use. Kim, Bauer and Wildman (2003) suggest the prepare dness of a nation and population density as a cost condition of deploying advanced networ ks are the most consistent factors explaining broadband uptake in OECD countries. More recently, through a nationwide U.S. surv ey, Savage and Waldman (2005) found that preference for high-speed access is apparent among higher income and college-educated households. Through data analysis of U.S. na tional surveys from 2002 to 2005, Horrigan (2005) claims the intensity of online use is the cri tical factor in underst anding the home broadband adoption decision and suggests the intensity of Internet use is a function of connection speed and years of online experience. Horrigans more r ecent survey demonstrates younger age, higher

PAGE 43

43 education and income, and urban living share of population may lead to higher level of broadband adoption (Horrigan, 2007). In addition, the United States Government Accountability Office (2006) found consumers with higher incomes and college de grees are significantly more likely to adopt fixed broadband. Chaudhuri, Flamm and Horrigan (2005) found the influences of traditional sociodemographic variables like income and educat ion on broadband deployment are strong in the U.S.. They also find substantial variation obser ved in access price may largely have a spatial explanation of Internet acce ss (Chaudhuri et al., 2005). Recen tly, through a household-level analysis, Clements and Abramowitz (2006) found income, age, educational attainment, and the presence of children influence adoption of broa dband service in the United States. Using data from 50 states in the United States, Lee (2006) also suggests income have influenced broadband deployment in the United States. Recently Gr osso (2006) found income measured by GDP per capita is related to the broadband penetration among OECD countries. Wallsten (2006) also found income and urbanization are factors of broadband adoption in OECD countries. In his empirical study, Turn er (2006) found income and poverty rate are influential factors of broadband deployment. Recently Ridder (2007) and Atkinson (2008)s empirical study found age is negatively correlated to the broadband adoption in OECD countries. More recently Trkman et al. (2008) found popula tion density and education are influential demographic factors of fixed-broadba nd deployment in EU countries. ICT Factors Recent stud ies on broadband diffusion provide ICT factors such as infrastructure and teledensity have influenced fixed-broadband adoption. Through a comparative study of broadband deployment in Canada, Japan, Korea, and the United States, Frieden (2005) argues the role of government in Information a nd Communication Technology (ICT) incubation is

PAGE 44

44 important for rapid broadband deployment. Kim, Bauer and Wildman (2003) suggest the preparedness of a nation is a factor of broadband deployment. Using panel data analysis of the U.S. states, Denni and Gruber (2005) also find that telecomm unication density has been an influential factor of broadband deployment in the United States. Recently Wallsten (2006) also found teledensity is a factor of broadband a doption in OECD countries. Through a multivariate analysis of ITU membership countries, Lee and Brown (2007) find ICT infrastructure is a significant factor of global broadband adoption. More recentl y, through a factor analysis, Trkman et al. (2008) found that communicati on technology expenditure, household PC access rate, Internet penetra tion, and fixed phone penetration are factors of fixed-broadband deployment in EU countries. Despite existing research efforts to better understand broadband adoption, the influence of important variables on global broadband adop tion across countries such as platform competition, LLU, population density, urban population, ICT infrastructure, broadband price, content, age and broadband speed have not be en clearly understood in a single systematic study (see Table 2-1). However, there is no empi rical study which test effects of institutional environment, international bandwidth, and telecommunication infrastructure investment. Also, there is no empirical work, which tests network effects and digital divide related to the diffusion of fixed-broadband. Research on Mobile Broadband Adoption Policy Factors Many stakeholders in the m obile industry such as policymakers, mobile service operators, content providers, and end-users play an in fluential role in shaping mobile broadband deployments at different phases. Saugstrup and Henten (2004) propose that regulation and market competition are important factors affecting the deploy ment of the new 3G technology.

PAGE 45

45 Table 2-1. Cross-National empirical studi es on fixed-broadband adoption factors Study Main independent variables Significant variables Kim et al. (2003) Broadband pric e Preparedness of a nation Dial-up service price Population density 30 countries 30 observations Income Preparedness of a nation Competition Population density Policy (unbundling, cross ownership, government funding) Garcia-Murillo (2005) Broadband price Broadband price Approximately 100 countries Income Income Observations varies Education Population density depending on the model (18-92) Competition Competition Population density Internet access Policy (unbundling, cross ownership) Unbundling Content Personal computers Internet access Distaso et al. (2006) Intra-modal competition Inter-modal competition EU countries Inter-modal competition LLU price 158 observations (15 time periods) Rights of way LLU price Price of leased line Price of ten minutes call Cava-Ferreruela and Broadband price Competition Technological competition Alabau-Mu oz (2006) Infrastructure investment Cost of deploying infrastructures Telecom services penetration Economic indicators

PAGE 46

46 Table 2-1 Continued. 30 countries 90 observations Internet indicat ors Demographic indicators (3 years: 2000-2002) Economic indicators Demographic indicators Education indicators Social indicators Grosso (2006) Competition Competition 30 countries Income Income Unbundling 117 observations (4 years: 2001-2004) Unbundling Fixed Internet penetration Wallsten (2006) Income Income 30 countries Unbundling Unbundling 179 observations Teledensity Teledensity (5 years: 1999-2003) Ur banization Urbanization Turner (2006) Price Income 30 countries Income Poverty rate 30 observations Population density (2005) Education Poverty rate Urbanization Ridder (2007) Price Price 30 countries Income Age 30 observations Age Urbanization (2005) Education Saturation Saturation Weather Urbanization Competition Policy Atkinson et al. (2008) Market concentration Urbanicity 30 countries Urbanicity Income 30 observations Home ownership Internet users (2007) Income Price Temperature Median age Median age Internet users Education Income inequality Price Internet users

PAGE 47

47 Table 2-1 Continued. Trkman et al. (2008) El ectronic purchasing Communication technology expenditures 25 countries Internet users Household PC access 23 observations Information technology expenditures Internet penetration (2006) Communication technology expenditures GDP per capita Household Income Household PC access PC users Internet penetration GDP per capita Price Fixed phone penetration Population density Education Fixed phone penetration Population density Education Specifically, Gans, King, and Wri ght (2004) suggest that sta ndardization policy plays a significant role in the success of wireless comm unication. From a historical perspective, regulations such as standardization with the intent to safeguard consumers have sometimes confined the development of new telecommunications services. Maeda, Amar, and Gibson (2006) argue that because of the typical progres sion of a new media technology market from the initial monopoly dominance of a large firm, to th e addition of many smaller competitors in the growth stage, to the few, surviving strong compe titors after a phase of c onsolidation, inflexible regulation such as mandated standardizati on may not actually benefit the consumers. Many studies in the economics of standard s have focused on the private and social incentives for standardization (G andal, 2002; David & Greenstein, 1990 ). Most of the literature suggests that compatibility and standardization ma y lead to efficient outcomes in the market. Theoretically, a single standard tends to de liver better economies of scale and network externalities. Nevertheless, Ga ndal et al. (2003) claim that th e aforementioned benefits of standardization are unclear in the mobile market. It was argued that as long as the mobile

PAGE 48

48 networks are interconnected and coverage is eff ective, there is little need for compatibility (Gandal et al., 2003). Roaming (i.e., using ones cellular phone outside th e providers coverage area) is a main network benefit for mobile communication. Thus, there are few network externalities that may justify a government mandated standard in the mobile market (Gans et al., 2004). In essence, there are both advantages and disadvantages in market mediated multiple standards and government mandated single standard. Though market-mediated standards might lead to limited network externalities and econom ies of scale, multiple wireless standards and different types of services acro ss technologies enable the existen ce of diverse competing systems which may lead to more and better m obile services (Gruber & Verboven, 2001). There are a few empirical studies addressi ng the standardization policy in the mobile industry. Gruber and Verboven (200 1) find that the early diffusi on of digital technologies in mobile markets was faster in Europe where most countries had adopted a single standard. Kioski and Kretschmer (2002) empirically estimate the effects of standardization through two alternative approaches. They conc lude that standardization has a positive but insignificant effect on the timing of initial entry of 2G services but can also lead to higher prices because it dampens competition. It seems that while a government-mandated standard was useful in stimulating mobile adoption in the initial stage (e.g., first generation m obile), as the mobile technology becomes more advanced, standardization policies become less relevant and even limiting (Cabral & Kretschmer, 2004). Cabral and Kretschmer (2004) examine the effectiveness of public policy in the context of competing standards with ne twork externalities and discovered that current mobile diffusion levels are quite similar between the United States (multiple standards) and Europe (mostly single standards). Recently Kauffman and Techatassanasoontorn (2005) found that market-mediated standardization policy is a contributing factor of in ternational diffusion of

PAGE 49

49 digital mobile technology. More recently, Rouvine n (2006) investigate the factors affecting the diffusion of digital mobile telephony across developed and developing countries. It was concluded that mandated standard s actually hinder competition in both groups. The review of literature thus far points to the changing role of standardizati on policy as mobile technology and markets continue to develop. It al so reveals that while there have been studies on the effects of standardization in early 1G and 2G mobile industry, no empirical work has ventured into the new 3G mobile markets concerning the e ffects of standardization policy. In regards to 3G licensing policy, since th e methods of licensing adopted by a country impact the nature and number of service prov iders in a mobile market, an effective policy framework for licensing 3G operators is important to the diffusion of 3G m obile services. Xavier (2001) suggests that, for the 3G mobile to develop successfully in a country, the licensing framework must reflect the high levels of investment required for a 3G-network rollout and nurture the growth of new and innovative 3G services. Granting individual 3G licenses ty pically involves some form of selection process. As the wireless business becomes more attractive to investors, the competition for the right to operate a certain 3G spectrum has intensified. In general, three main selection approaches auctions, beauty contests, and a hybrid system have been used to choose licensees by governmental agencies (Xavier, 2001; ITU, 2001; Bauer, 2003). Many countries like Canada, Germany, and the U.K. used auctions to assign 3G licenses. Supporters of the auctioning approach argue that auctions are transparent and fair, and can be designed to incorporate a wide range of public policy goals (McMillan, 1995). In addition, the auctioning of 3G spectrum licenses can generate substantial revenues for governments and may in crease competition with existing services (Kwerel & Strack, 2001). Auctioni ng also allocates 3G spectrums to those operators that value

PAGE 50

50 the spectrum most highly, and who can thus be e xpected to make the most economically efficient use of the spectrum (Xavier, 2001). However, it is difficult to translate theoretically efficient auction designs into practice because they assume that all potential bidders have full access to information on 3G markets (Xavier, 2001). Additionally, government control over spectrum allocation means there are strong government incen tives to manage the resource in a way that maximizes their own revenue and the procedure of auction design could be quite complex (ITU, 2004). Noam (2003) also argues that the U.S. sp ectrum auctioning system generated costs of deployment delay far in excess of any efficiency gains and permitted other countries to leapfrog the United States in mobile technology, a pplications, and consumer satisfaction. In other countries such as Japan, Korea, No rway, and Sweden, the beauty contest or comparative evaluations method has been used for 3G licensing (ITU, 2001; Xavier, 2001). With the beauty-contest method, the government accepts a pplications from eligible applicants and decisions to award licenses are typically base d on multiple criteria (ITU, 2004). The advantage of the beauty contest is that it awards 3G licen ses to those applicants th at would make the best use of the license from a social point of vi ew (ITU, 2004). However, the beauty-contest method can be time-consuming and unfair. Some countries such as Au stria, Italy, and Hong Kong adopted a combined or hybrid method for awarding 3G licenses (ITU, 2001; Xavier, 2001). The hybrid method requires applicants to pre-qualify under mu ltiple criteria similar to those established for the conventional beauty contest in order to bid (Xavier, 2001). Licenses are then allocated on the basis of an auction. It is plausible that the selection process that a countr y chose to select 3G mobile licensees might play a role in the adoption of these services.

PAGE 51

51 Institutional environment might also influe nce mobile broadband deployment. Employing regression analysis, Andonova (2006 ) finds institutional environmen t such as political rights and civil liberties are correlated with deployment of m obile. In spite of these previous studies, there was no empirical study, which tests the influenc es of institutional environment on fixedbroadband diffusion. Industry Factors In addition to policy issues, th e characteristics of the mobile industry in a country affect the marketing of 3G services in that country and thus its rate of deployment. Steinbock (2003) suggests that the thrust of change in the mob ile industry has shifted from technology to markets in the 3G era. In a separate study, Lee (2006) finds that market-based standardization policy and competition empirically correlate with mobile gr owth rates. Furthermore, through a longitudinal analysis of 25 Asian countries from 1986 to 1998, Burki and Aslam (200 0) find that digital mobile competition did indeed promote mobile diffusion. More recently, in his investigation of the factors affecting the diffusi on of digital mobile telephony across developed and developing countries, Rouvinen (2006) conclude that market competition promotes mobile diffusion in both groups. It is intuitive that the cost of mobile services would affect the demand for such services. A number of empirical studies have investigated the possible caus al relationship between mobile service cost and mobile deployment. Ahn and L ee (1999) study the determinants of demand for mobile telephone networks usi ng observations from 64 ITU memb er countries and found that price was not a strong predictor of demand. On the other hand, using a panel data set of 56 countries, Madden et al. (2004) ex amine the economic factors that affect the growth of global mobile telephony and conclude that lower cost contributes to mobile diffusion. There seems to be conflicting results concerning the role of pric ing in the development of mobile phone services.

PAGE 52

52 Similar to the factor of mob ile phone cost, the prices of nonvoice mobile applications in each country may also affect the consumer acceptan ce and usage of enhanced mobile services in that country. Studies have suggest ed that mobile applications and services that ex ploit the value of a wide-area wireless network and customers needs and desires are important drivers of success for 3G mobile systems (Foster, 2003). Th e availability of app ealing, diverse mobile applications such as multimedia messaging and mob ile internet is likely to contribute to the growth of 3G mobile (Wilska, 2003; Nobel, 2004) Specifically, Foster ( 2003) proposes that the development of mobile applicati ons and services that leverage the unique capabilities of the wireless environment such as personal and ubiqu itous capabilities would directly promote 3G mobile deployment. More recently using multivariate analysis of 55 countries data, Lee et al. (2007) find lower pricing of mobile ap plication such as SMS services is co rrelated with the higher level of 3G mobile deployment. Demographic Factors Many studies have em pirically supported the importance of national economic health in stimulating the demand for mobile services. Ahn and Lee (1999), in their study of the determinants of demand for mobile telephone netw orks, find that the prob ability of subscribing to the telephone networks was positively correlat ed with per capita GDP. Using a panel data set of 56 countries to investig ate the economic factors influenc ing the growth of mobile phone services, Madden et al. (2004) conclude that higher income and a large user base tend to promote mobile diffusion. In a separate empirical study ab out the same topic, A ndonova (2006) also find GDP per capita to be major contribu ting variable to mobile diffusion. Communication research has l ong identified the importance of demographic factors as antecedents in new media adoption (Atkin and LaRose, 1994). It was suggested that early

PAGE 53

53 adopters tend to have higher socio-economic status (Rodgers, 2003). Va rious studies have reported a positive relationship between education and new media technology adoption (Lin, 1998; LaRose &Atkin, 1992). Wareham et al. (2004) report that education is a steady indicator of wireless phone diffusion because achieving higher education has a positive association with being comfortable with higher technology use. In addition to socioeconomic factors, there seems to be a positive linkage between population density and mobile penetration rate s as evidenced in areas like Hong Kong and Luxemburg that have been garnering mobile penetration rates over 120 and 150 percent (ITU, 2005). Through data analysis of approximately 100 countries, Garcia-Muril lo (2005) finds that population density has positive effects on the number of broadband subscribers. Kim and et al. (2003) also suggest that population density should be considered the cost conditions of deploying advanced networks and is one of the influe ntial factors in explai ning broadband uptake. However, some empirical studies examining the relationship between mobile diffusion and share of urban population find no significant correlation between the two va riables (Gruber, 2001; Koski & Kretschmer, 2002). Some studies emph asize the role of socio-cultural factors in explaining broadband deployment. In his compar ative study of broadband deployment in Asia, Aizu (2002) argued social and cult ural factors were important e xplanatory variables for widely differing diffusion rates in Asian countries. He su ggests killer applications like online gaming is an influential factor of broadba nd adoption in Korea (Aizu, 2002). ICT Factors The existing inform ation and communication infra structure in a country is also a potential factor that might affect the a doption of 3G mobile services. Some countries might be more prepared than others to adopt new communica tions technology beca use of their existing

PAGE 54

54 information and communications technology (ICT) development and consumers experience with relevant information/communication services. Some ITU Internet reports have suggested that the countries that already have high PC and Internet penetration have seen users embrace broadband services more readily (ITU, 2003b; ITU 2003c). Kim et al. (2003) found that the preparedness of a nation, which is measured by the attitudes of a nation towards advanced information technol ogy and the availability of complementary technologies, such as computers, is one of the important factors in explaining new media technology adoption like broadband services. Specifically, through a panel data analysis, Denni and Gruber (2005) find that telecommunicati on density has a positive impact on broadband diffusion. In spite of a growing body of literature th at addresses the factors contributing to mobile adoption at the country level, few empirical studies have focused on the factors that affect mobile broadband adoption globally (see Table 2-2). In addition, mobile broadband is not a brand new innovation, but an evolving new co mmunication/media technology that might be affected differently by policy, industry, ICT a nd demographic factors than fixed-broadband. There is no empirical work, which tests whethe r fixed-broadband is a complement to or substitute for mobile broadband. Also, there is no empirical study regardi ng the potential mobile broadband digital divide.

PAGE 55

55 Table 2-2. Cross-National empirical studies on mobile adoption factors Study Main independent variables Significant findings Gruber (2001) Income Late mobile adoption Urban population Multiple operators 140 countries Fixed penetra tion High fixed penetration Wait time Wait time Digital mobile competition Number of mobile operators Market transition index Gruber and Income Competition Verboven (2001) Fixed penetration Single standard Digital mobile Incumbent pre-empt 140 countries Standard sequential entry Competition Koski and Income Between and within standards Kretschmer (2002) Urban population Competition Competition Lower user cost 32 countries Analog mobile penetration Dominant digital mobile standard Mobile operators (dummy) Liikanen et al. (2004) Inco me Digital mobile introduction Uraband population hinders analog mobile diffusion 80 countries Fixed penetratio n Generation-specific results Number of analog/digital standards Years since introduction Standard (dummy) Mobile telephony operation Age-dependency ratio differ from generic results Kauffman and Techatassanasoontorn (2005) Years since introduction Income Standard (dummy) Number of digital mobile phone standards 46 countries Mobile telephony operation Mobile service price Analog mobile phone penetration Analog mobile phone service prices Standardization policy Licensing Policy

PAGE 56

56 Table 2-2 Continued. Age-dependency ratio Income Income Fixed phone penetration Number of digital mobile phone standards Rouvinen (2006) Income Standards Population Competition 165 countries Standard Network effects Fixed Penetration/user cost Development Technology Democracy

PAGE 57

57 CHAPTER 3 ANALYTICAL FRAMEWORK AND RESEARCH QUESTIONS This section discusses the anal ytical fra mework and proposed research questions for this study. This study examines adoption factors of fixed, mobile, and ubiquitous broadband that collapses both fixed and wireless into one category. This empiri cal study proposes four different empirical models based on the four different sets of proposed research questions. Analytical Framework and Research Questions This study tests adoption factor s of fixed, m obile and ubiqui tous broadband deployment. This study mainly examines whether policy, indus try, demographic, and ICT factors influenced broadband deployment (see figure 3-1, 3-2, and 3-3) The first analytical framework employs both non-linear and linear regre ssion and tests whether macronational level factors have influence fixed-broadband depl oyment (see figure 3-1). Figure 3-1. Analytical framework for fixed-broadba nd deployment The second analytical framework employs linear regression to test th e adoption factors of mobile broadband deployment. The second analy tical framework empirically tests whether macro-national level factors have influenced m obile broadband deployment (see figure 3-2). The third analytical framework employs linear regression to test th e adoption factors of ubiquitous broadband deployment. The third anal ytical framework empirically tests whether Fixed Broadband Deployment Industry Platform competition Price Speed Bandwidth Investment Demographic Income Education Urban population Population density Age ICT Factors PC penetration Content Internet usage Teledensity Policy LLU Political Freedom Economic Freedom

PAGE 58

58 macro-national level adoption factors have in fluenced ubiquitous broadband deployment (see figure 3-3). Figure 3-2. Analytical framework for mobile broadband deployment Figure 3-3. Analytical framework for ubiquitous broadband deployment These three analytical frameworks include f our categories of indepe ndent variables: (1) policy factors, (2) industry fact ors, (3) demographic factors, and (4) ICT factors. Choice of individual independent variables is based on previous empirical stud ies. Most of the independent variables are similar. Technology-specific i ndependent variables are LLU for fixed and ubiquitous broadband, platform competition for fixed-broadband, standa rdization policy for mobile broadband and ubiquitous broadband, ne twork competition for ubiquitous broadband, Ubiquitous Broadband Deployment (Fixed and Mobile) Industry Network competition Price Bandwidth Investment A pp lication p rice Demographic Income Education Urban population Population density Age ICT Factors PC penetration Content Internet usage Teledensity Policy LLU Standardization Political Freedom Economic Freedom Mobile Broadband Deployment Industry Price Bandwidth Investment Application price Demographic Income Education Urban population Population density Age ICT Factors PC penetration Content Internet usage Teledensity 1G/2G penetration Policy Standardization Licensing Political Freedom Economic Freedom

PAGE 59

59 mobile application price for mobile and ubiquitous broadband, and 1G and 2G penetration for mobile broadband. Policy Factors For policy f actors, LLU policy (for fixed and ubiquitous broadband), political freedom (for all broadband technologies), economic freedom (for all three models), and standardization policy (for mobile broadband and ubiquitous broadband ) were selected for the empirical test. Many countries have considered local loop unbu ndling policy as important policy tools for promoting rapid broadband adoption. Though LLU ma y reduce their incentives to invest in advanced telecommunication technologies (Fried en, 2005a), some previous empirical studies suggest LLU policy might be positively correl ated with broadband deployment (Grosso, 2006; Garcia-Murillo, 2005; Distaso et. al., 2006, L ee & Brown, 2007). Considering fixed-broadband is the dominant type of broadband technology, LLU also might be positively correlated to the ubiquitous broadband deployment. Institutional environment such as politic al freedom and economic freedom might be positively correlated to the higher level of fixe d, mobile and ubiquitous broadband deployment. In the previous empirical study, Andonova (2006) suggest political rights and civil liberties are correlated with deployment of the Internet. With political freedom this study also tests the relationship between economic freed om and broadband deployment. Multiple standardization policy might be pos itively related to rapid mobile broadband. Though multiple standardization policy might l ead to limited network externalities and economies of scale, multiple standards and different types of services across technologies may lead to better and differentiated services in the market (Church & Gandal, 2005). One recent empirical study suggests that multiple standardiza tion policy contributes to the higher level of mobile broadband deployment (L ee, et. al, 2007). Based on this previous study, it is also

PAGE 60

60 expected that with the effects of platform co mpetition standardization policy might lead to the higher level of ubiquitous broadba nd diffusion. Also, different 3G licensing methods might also affect mobile broadband de ployment (Lee, et al., 2007). Based on the literature reviewed, following research questions (RQs) are proposed: RQ1: Have policy factors, specifically Local Loop Unbundling (LLU), and institutional environment such as political and economic freedom contributed to the adoption of fixedbroadband services? RQ2: Have policy factors, market mediated stan dardization policy, and institutional environment such as political and economic freedom contri buted to the adoption of mobile broadband services? RQ3: Have policy factors, specifically Local Loop Unbundling (LLU) and standardization policy, and institutional environment such as political and economic freedom contributed to the adoption of ubiquit ous broadband services? Industry Factors For industry factors, platfor m (network) competition (for fixed and ubiquitous broadband), price (for all broadband technologies), speed (for fixed-broadband), bandwidth (for all broadband technologies), telecommunication networ k investment (for all fixed and ubiquitous broadband), mobile network investment (for mo bile broadband) and multiple strategy (for all broadband technologies) were select ed for the empirical test. Platform (network) competition among diffe rent technologies (n etworks) might be positively correlated with the fixed and ubi quitous broadband adoption. A growing body of previous literature suggests th at facilities-based competition among several different broadband platforms (networks) may lead to improved quality of service, higher level of broadband adoption and network investment (ITU, 2003b; DotEcon and Criterion Economics, 2003; Aron & Burnstein, 2003; Lee, 2006; Di staso et. al, 2006 ). However, there was no empirical study which tests the relationship between network competition between fixed and mobile broadband and ubiquitous broadband adoption.

PAGE 61

61 Lower price for fixed, mobile and ubiquitous broadband could be negatively related to the broadband diffusion. However, there is no empi rical study, which found relationship between price variables and fixed-broa dband adoption. Recent OECD report suggests there is a non-linear relationship between price vari ables and broadband adoption (O ECD, 2007e). Recently an empirical study found lower levels of mobile app lication price could be correlated with higher levels of mobile broadbanf diffu sion (Lee et al, 2007). Also fi xed and mobile broadband service could be a substitute for or complementary to one a nother. It is necessary to test whether price of fixed or price of mobile broadband ha ve influenced mobile or fixed-broadband diffusion. Higher speed, bandwidth, and higher level of telecommunication (mobile network investment for mobile broadband) network invest ment might be positivel y correlated with the fixed, mobile and ubiquitous broadband adoption. However, data for the speed variable were available only for fixed-broadband. One recent empirical study found higher speed is positively correlated with the fixed-broadband deploymen t (Lee & Brown, 2007). Growth in demand for higher capacity and telecommunicat ion infrastructure investment from private and public sector might lead to higher level of broadband di ffusion (ITU, 2003b; ITU, 2006). But there was no empirical work yet. Also, multiple play strategy might be a key driver of broadband adoption. As an ITU report suggested, corporate strategy is an infl uential factor of broadband deployment (ITU, 2003b). It is interesting to examin e whether triple-play strategy ha ve influenced the growth of fixed-broadband and quadruple-play strategy have influenced th e growth of mobile broadband. Also, triple-play and quadrupleplay strategy might influence the diffusion of ubiquitous broadband.

PAGE 62

62 RQ4: Have industry factors, specifically pl atform competition, fixed-broadband price, broadband speed, bandwidth, telecommunication in frastructure investment, mobile service price, and triple-play stra tegy influenced the adoption of fixed-broadband services? RQ5: Have industry factors, specifically pl atform competition, fixed-broadband price, bandwidth, mobile network investment, mobile se rvice price, mobile application price, and quadruple-play strategy influenced the adoption of mobile broadband services? RQ6: Have industry factors, specifically ne twork competition, fixed-broadband price, bandwidth, telecommunication infrastructure i nvestment, mobile service price, mobile application price, and multiple play strate gy influenced the deployment of ubiquitous broadband services? RQ7: Is mobile broadband a complement to or substitute for fixed-broadband? RQ8: Is fixed-broadband a complement to or substitute for mobile broadband? Demographic Factors For dem ographic factors, income, educati on, urban population, population density, and age were selected for the empirical test. All these demographic factors are employed for fixed, mobile, and ubiquitous broadband. Previous st udies on new media technology adoption suggest higher level of income, education, urban popula tion share, and population density might be positively correlated with the higher levels of fixed, mobile, and ubiquitous broadband. Some empirical studies on fixed-broadb and find high level of population density is related to rapid fixed-broadband deployment (Kim et al., 2003 ; Garcia-Murillo, 2005; Lee & Brown, 2007). An empirical study suggests high level of income c ontributed to high level of mobile adoption (Madden et al., 2004). Most of studies on th e communication sector suggest younger age is correlated with the higher leve l of new media technology adoption. Recently De Ridder finds the 35-39 and 40-44 year old groups were positiv ely correlated with fi xed-broadband adoption (OECD, 2007e). Based on this previous OECD study, it is interesting to examine whether age groups such as 15-24, 25-34, and 35-44 are corre lated with fixed, mobile, and ubiquitous broadband diffusion.

PAGE 63

63 RQ9: Have demographic factors, specifically income, education, urban population share, population density and age influenced the deployment of fixedbroadband services? RQ10: Have demographic factors, specifically income, education, urban population share, population density and age influenced the deployment of mobile broadband services? RQ11: Have demographic factors, specifically income, education, urban population share, population density and age influenced the depl oyment of ubiquitous broadband services? ICT Factors For ICT factors, PC penetration, content, Internet usage, and teledensity w ere selected for the empirical test. These ICT factors are empl oyed for fixed, mobile, and ubiquitous broadband. For mobile broadband deployment, to capture the effects of previous generation mobile technology, 1G and 2G penetration wa s also selected for the empiri cal test. Liiknen et al. (2004) found that 1G has a positive effect on 2G. The existing information communication infrastructure in a country might be important factor that might affect the adoption of fixed, mobile and ubiquitous broadband services. Frie den (2005b) suggests ICT incubation is a key driver of broadband adoption. Kim et al. (2003) suggest the prepar edness of a nation is a driver of fixed-broadband deployment. A recent empirical study suggests content is factor of fixedbroadband deployment (Lee & Marcu, 2007). It is e xpected that high level of PC penetration, content, Internet usage, and teledensity are pos itively correlated with the high level of fixed, mobile, and ubiquitous broadband deployment. 1G and 2G penetration could be positively or negatively correlated to the mobile broadband diffusion. RQ12: Have ICT factors, specifically PC penetrati on, content, Internet usage, and teledensity influenced the deployment of fixed-broadband services? RQ13: Have ICT factors, specifically PC penetrati on, content, Internet usage, and teledensity, 1G and 2G mobile pe netration influenced the deploymen t of mobile br oadband services? RQ14: Have ICT factors, PC penetration, content, In ternet usage, and tele density influenced the deployment of ubiquitous broadband services?

PAGE 64

64 Factors of Digital Divide and Network Effect For factors of digital divide, this study sele cts incom e and region for the empirical test. These factors are employed for fixed, mobile and ubiquitous broadba nd. Previous studies suggest income and region might be factors of digital divide (Crandall, 2003; Rice & Katz 2003; Madden et al., 1996; Chinn and Fairlie, 2 006). Also, to capture the different contributing factors of broadband diffusion for developed and developing countries, it is interesting to examine whether there are common or different factors in the diffusion of broadband in developed and developing countries. Based on the pr evious literature, this study also examines generation as a factor of digital divide. RQ15: Is there digital divide in fixed-br oadband deployment among group of countries originated from income and region? RQ16: Is there digital divide in mobile br oadband deployment among group of countries originated from income and region? RQ17: Is there digital divide in ubiquitous broadband deploy ment among group of countries originated from income and region? RQ18: Are there common or different factors in the diffusion of fixed and ubiquitous broadband in developed and developing countries?1 By using log linear regression model, this study also test whether previous broadband subscription has influenced current broadband subscription. If network e ffect exists, current subscription might be positively correlated w ith the previous subscription of new media (Economides & Himmelberg, 1995). Network effect also might exist in the adoption of mobile broadband and ubiquitous broadband, but limitati on of data, this study examines whether previous subscription has infl uenced current subscription fo r fixed-broadband technology. 1 Currently only small number of data mobile broadband related data for developing countries are available. Therefore, this study excluded this res earch question for mobile broadband.

PAGE 65

65 Based on the literature reviewed, the follo wing research question (RQ) is proposed: RQ19: Have previous subscripti on of fixed-broadband influenced current subscription? Proposed Empirical Models Fixed-Broadband To exam ine determinants of the global fixed-broadband deployment, this study employs both non-linear and linear re gression models. This logistic regression model2 (non-linear regression model) employs 240 observations for broadband services from OECD (Organization for Economic Co-operation and Development) c ountries. This study also estimates a linear regression model of fixed-br oadband penetration. The linea r regression model employs approximately 380 observations for fixed-broadband services from the ITU (International Telecommunication Union) membership countries. Non-linear model of fi xed-broadband diffusion. For the estimation of fixed-broadband diffusion, a logistic model of technology diffu sion is employed. Gruber and Verbovens (2001b) logistic model of mobile diffusion is applied to the diffusion model for fixed-broadband. The logistic specification is appropriate for capturing the existenc e of network externalities (Gruber and Verboven, 2001b). With network externality, higher adoption of fixed-broadband services makes subscribers more valuable. In some c ountries fixed-broadband penetration pattern is nonlinear and standard S-shaped curve (see figure 3-4). 2 This logistic regression model is based on the previous study by Lee and Marcu (2007).

PAGE 66

66 A) B) C) D) Figure 3-4. Broadband penetration 1999-2005 in Japan and Korea. A) DSL in Japan. B) CableModem in Japan. C) DSL in Kor ea. D) Cable-Modem in Japan Letting ity denote the percentage of country i s population that has broadband access to the Internet by time t the standard logistic diffusion equation is: ) exp(1*tba y yitit it it (3-1) where ita, itb, and* ity are parameters, as discussed below. 0% 1% 2% 3% 4% 5% 6% 7% 8% 9% 19982000 20022004 2006 Year 0% 2% 4% 6% 8% 10% 12% 14% 16% 1998 2000 2002 2004 2006 Year 0% 1% 1% 2% 2% 3% 3% 1998200020022004 2006 Year -2% 0% 2% 4% 6% 8% 10% 12% 1998 2000 20022004 2006 Year

PAGE 67

67 Not all individuals in a country adopt a new media technology, such as broadband, regardless of how inexpensive the technology may become. This is captured in the model by* ity, the long run expected fraction of subscribers (the ceiling pa rameter, or saturation point).3 The parameteritain equation (3-1) is a constant of integra tion that gives the initial value of fixedbroadband penetration.4 A positive value shifts the S-shaped function upwards while a negative one shifts it downwards, wit hout modifying the S-shape. The parameter itbin equation (3-1) captures the spee d of fixed-broadband diffusion. This can be seen by differentiating equa tion (3-1) with respect to time: *1it itit it it ity yy b ydt dy (3-2) Equation (3-2) shows that itb is equal to the growth rate in the number of adopters, relative to the fraction of potential s ubscribers who have not yet adopted the technology. The speed of fixed-broadband diffusion varies with policy variablesj itD and country socioeconomic characteristicsitXin linear fashion. In the previous literatures two broad classes of logistic diffusion models have been proposed: the variable-ceiling logistic a nd the variable-speed logistic (Fernandez-Cornejo & McBride, 2002) Letting the ceiling vary with country characteristics poses significant estimation problems. There is no guarantee that the parameter will stay at theoretically justifiable levels, or that the m odel will converge. The variable-speed 3 Note that tasyyitit *. 4 Note that 0 1* tas e y yita it it.

PAGE 68

68 logistic model is easier to estimate and the sp eed of adoption can be positive or negative, depending on the movement of exogenous factors. The logistic regression is sy mmetric and imposes an inflect ion point halfway between zero and the saturation point. The inflection point is crucial in determining the saturation point (Bewley and Griffiths, 2003). The saturation point is estimated from the observations of early adopting countries that have passed the midway point, such as Japan and South Korea. To the extent the saturation points of lagging countries differ from those of forerunners, holding the ceiling parameter fixed across countries may bias the expected saturation point for lagging countries. This is somewhat mitigated by the add ition of an error term to equation (3-1) for the purpose of estimation. it J j j it j itXD b 1 0 (3-3) The country characteristics included initXare variables that may influence the supply and demand for fixed-broadband services. The demand for fixed-broadband services is expected to increase with the higher level of income, educa tion, PC penetration, bandwidth and Internet content. Higher population density and percenta ge of urban population decrease deployment cost, increasing the supply of broadband. The effects of policy variables on fixed-broadband penetration are main interests of this nonlinear regression. The main polic y variables included in this study are dummy variables cap turing the effects of different types of LLU such as full bundling, line sharing, bit stream access, LLU price regulation (regulatory approval for line rental charges). Interaction du mmy variables of these differen t types of LLU with LLU price regulation are included in this model with plat form competition and institutional environments such as political and economic freedom.

PAGE 69

69Linear model of fixed-broadband diffusion. To capture more diverse determinants of global broadband deployment, a multiple regression analysis (linear model) is also implemented. To examine the influences of qua ntifiable variables on the diffus ion patterns of fixed-broadband, this study formulates the following linear regression model. Since the distribution of dependent variable and many independent variables in this linear regression model is positively skewed, data transformation with logarithm was employed. Ln Yt (BPR) = 0 + 1(ln Platform Competition) + 2(ln Previous Penetration) + 3(ln Political Freedom) + 4(ln Economic Freedom) + 5(ln Fixed-broadband Price) + 6(ln Mobile Price) + 7(ln Speed) + 8(ln Bandwidth) + 9(ln Investment) + 10(ln Income) + 11(ln Education) + 12(ln Population Density) + 13(ln Urban Population) + 14(ln Age) + 15(ln PC Penetration) + 16(ln Content) + 17(ln Internet usage) + 18(ln Teledensity) + t (3-4) The empirical model (3-4) for multivariate analysis was a composite model from previous empirical studies. In the mode l, the dependent variable (Yt) is fixed-broadband diffusion. This study included independent variables such as platform competition, previous penetration, political freedom, economic freedom, fi xed-broadband price, speed, bandwidth, telecommunication network investment, income, education, population de nsity, urban population, age, PC penetration, content, Internet usage, and teledens ity. To examine whether mobile broadband is a complement to or a substitute for fixed-broadband, mobile price was also included in the linear regression model. One-way ANOVA is used to test the digital di vide research questions. The digital divide between high/medium/low income countries and between regions is analyzed by the one-way ANOVA by comparing a few groups of countries. One-way ANOVA is also used to test the

PAGE 70

70 effects of LLU and triple-play strategy. 159 observa tions were available to test the effects of LLU and 29 observations were avai lable to test the effects of triple-play strategy on the growth rate of fixed-broadband penetration. Also, based on the result of regression analysis, significant factors of fixed-broadband diffusion for deve loped countries (high income countries) and developing countries (medium or low income countries) are compared. Mobile Broadband To examine determinants of global mobile broadband deployment, this study employs a log linear regression model. In th e model, the dependent variable (Yt) is mobile broadband diffusion. The linear regression model employs approximately 106 observations for mobile broadband services from the ITU (Internati onal Telecommunication Union) membership countries. Linear model of mobile broadband diffusion. To examine the influences of quantifiable variables on the diffusion pattern s of mobile broadband, this st udy formulates the following log linear regression model. Since the distribution of dependent variable and many independent variables in this linear regression model is positively skewed, data transformation with logarithm was utilized. Ln Yt (MBPR) = 0 + 1(Standardization Policy) + 2(ln Political Freedom) + 3(ln Economic Freedom) + 4(ln Fixed-broadband Price) + 5(ln Mobile price) + 6(ln Bandwidth) + 7(ln Investment) + 8(ln Mobile Application Price ) + 9(ln Income) + 10(ln Education) + 11(ln Population Density) + 12(ln Urban Population) + 13(ln Age) + 14(ln PC Penetration) + 15(ln Content) + 16(ln Internet usage) + 17(ln Teledensity) + 18(ln 1G and 2G Penetration) + t (3-5)

PAGE 71

71 The empirical model (3-5) for multivariate analysis was a composite model from previous empirical studies about mobile diffusion. This study includes i ndependent variables such as standardization policy, political freedom, economic freedom, mobile price, bandwidth, telecommunication investment, mob ile application price, income education, population density, urban population, age, PC penetr ation, content, Internet usag e, teledensity, and 1G and 2G penetration. To examine whether fixed-broadband is a complement to or a substitute for mobile broadband, fixed-broadband price was also in cluded in the linear regression model. One-way ANOVA is also used to test the infl uence of licensing policy and determinants of digital divide. The digital divide between high/medium/low income countries and between regions is analyzed by the one -way ANOVA by comparing a few groups of countries. One-way ANOVA is also used to test th e effects of quadruple-play stra tegy on the growth of mobile broadband penetration. 69 observations were availa ble to test the effects of licensing policy and 29 observations were available to test the effects of quadruple-play strategy on the growth rate of mobile broadband penetration. Ubiquitous Broadband As already discussed in the previous chapters, underlying forces of the notion of ubiquitous broadband are convergence in servi ces, technology, industry, and policy and consumer demand. These underlying forces may in fluence the macro-leve l adoption factors of ubiquitous broadband. A vision of ubiquitous broadband is that broadband access is possible anytime, anywhere, by anyone and anything thr ough both fixed and mobile broadband. In this context, examination of adoption factors of ubiquitous broadba nd is necessary.

PAGE 72

72 To examine determinants of broadband adopti on, this study employs a log linear regression model. The linear regression model employs approximately 216 observations for broadband services from the ITU (Inter national Telecommunication Uni on) membership countries. Linear model of ubiquitous broadband diffusion. To examine the influences of quantifiable variables on broadba nd adoption, this study formulat es the following log linear regression model. Since the distri bution of many variables in th is linear regression model is positively skewed, data transformati on with logarithm was utilized. Ln Yt (BPR) = 0 + 1(ln Network Competition (or Standardization Policy Type) + 2(ln Speed) + 3(ln Political Freedom) + 4(ln Economic Freedom) + 5(ln Fixed-broadband Price) + 6(ln Mobile price) + 7(ln Bandwidth) + 8(ln Telecommunication Investment) + 9(ln Income) + 10(ln Education) + 11(ln Population Density) + 12(ln Urban Population) + 13(ln Age Group) + 14(ln PC Penetration) + 15(ln Content) + 16(ln Internet usage) + 17(ln Teledensity) + t (3-6) The empirical model (3-6) for multivariate analysis was a composite model from previous empirical studies about fixed and mobile broadb and deployment. In the empirical model, the dependent variable (Yt) is broadband adoption that accounts for both fixed and mobile services. Independent variables are policy factors such as standardization policy type, political freedom, and economic freedom. Three different standardization policy type s (Policy Type I: platform competition for fixed-broadband without mobile br oadband services, Policy Type II: platform competition with single standard for mobile broadband services, Policy Type III: platform competition with multiple standards for mobile br oadband services) are examined in the model. Also industry factors such as speed, network competition, telecommunication infrastructure investment, fixed-broadband pric e, bandwidth, and mobile servic e price, demographic factors such as income, education, population density, ur ban population, and age, and ICT factors such

PAGE 73

73 as PC penetration, Internet us age, content, and teledensity. Since some of policy types for standardization policy might be conceptually a nd empirically correlated, only one variable (network competition or policy types for standard ization policy) were included in an empirical model. One-way ANOVA is also used to test th e digital divide research questions. The digital divide between high/medium/low income countries and between regions is analyzed by the oneway ANOVA by comparing a few groups of countries The effects of mu lti-play strategy also were analyzed by one-way ANOVA by comparing two groups of countries (countries with triple-play and without triple-p lay and countries with quadrup le play and without quadrupleplay). One-way ANOVA is also used to test the effects of LLU policy. 83 observations were available to test the effects of LLU and 29 observations were available to test the effects of multiple-play strategy on the growth rate of fixed-broadband penetration.

PAGE 74

74 CHAPTER 4 RESEARCH METHOD This chapter discusses data, m easurement, a nd statistical methods to test the adoption factors of global broadband de ployment. This study uses sec ondary datasets and employs quantitative methodologies such as non-linear and linear regression analysis, and one-way ANOVA to investigate the role of the contribut ing factors in affecting fixed, mobile, and ubiquitous broadband adoption at the national level. Employing secondary analysis may save time, money, and personnel and it is effici ent (Wimmer & Dominick, 2006). Secondary analysis1 also allows researcher to expend the scope of the study considerably (Stewart, 1984). Measurement, Data and Statistical Met hods for Fixed-broadband Deployment Table 4-1 and 4-2 shows the variables, thei r measurement, and the corresponding data sources for fixed-broadband deployment. The dependent variable fixed-broadband deployment, was measured by the number of broadband subscribers per 100 inhabitants. As detailed in the literature review, there are many potential independent variables involving policy, industry, ICT and demographic factors that may in fluence fixed-broadband adoption. Policy Factors LLU (Local loop unbundling) might be a key dr iver of the fixed-broadband deployment (ITU, 2003b; Garcia-Murillo, 2005; Distaso et. al., 2006). To capture the effects of different types (full unbundling, line sharing, and bit st ream access) of LLU and LLU price regulation (regulatory approval for line rent al charges), dummy variables (1 for with full unbundling 0 for otherwise; 1 for with unbundling 0 otherwise; 1 fo r with bit stream access 0 for otherwise; 1 for with price regulation 0 for no pric e regulation) are also employed. Some previous studies used 1 Not all information obtained from secondary sources is equally reliable or valid. For evaluation of secondary data, information must be evaluated according to its recency and cred ibility (Stewart, 1984).

PAGE 75

75 dummy variable as a measure of LLU (Garci a-Murillo, 2005; Lee & Brown, 2007). For the actual nonlinear model of regression, interaction variable s of these dummy variables are used to prevent multicollinearity issue. Three types of LLU policy were identified from the interactions of these dummy variablesLLU Policy I (full unbundling, line sharing and bit stream access without price regulation), LLU Po licy II (full unbundling, line sharing, no bit stream access with price regulation), and LLU Polic y III (full unbundling, line shar ing and bit stream access with price regulation). For the one-way ANOVA, dummy variable (1 for with LLU, 0 for no LLU) is used. Political freedom is measured by the inve rse of the score on civ il liberties (originally ranging from 1 to 7; Andonova, 2006). For the meas urement of economic freedom, the index of economic freedom index has been used. The index of economic freedom is defined by multiple rights and liberties such as business freedom, trade freedom, monetary freedom, and freedom from government (Beach & Kane, 2007). Industry Factors Platform competition is an important industry variable in which the broadband market is served by competing platforms. In the previous studies platform compet ition could be measured by HHI (Herfindahl-Hirshman-Index) or dummy variable (0 or 1) (Distaso et. al., 2006; Lee & Marcu, 2007). A report from DotEcon & Criter ion Economics (2003) suggested broadband penetration tends to be higher in European co untries where DSL and n on-DSL platforms have similar market share. This study employs more generalized measures for platform competition by the HHI (Herfindahl-Hirshman-Index) betwee n different fixed-broa dband technologies. Fixed-broadband price arguabl y has been a key industry f actor in promoting broadband demand. Successful broadband economies are char acterized by low prices as a result of flourishing competition and innovativ e pricing schemes to attract a wide variety of customers (ITU, 2003a). Broadband price is measured by br oadband monthly charge (in U.S. Dollars).

PAGE 76

76 Broadband speed is also considered important in dependent variable that might influence global broadband adoption. It is measured by broadba nd download speed (kilobit per second). As a product differentiation strategy in the br oadband access market, broadband speed might influence broadband demand. For the measurem ent of bandwidth, international Internet bandwidth (bits per inhabitant ) is employed. For the meas urement of te lecommunication infrastructure investment, annual telecommunicati on investment is employed. For mobile price, per minute charge (in U.S. Dollars) for a local call during peak time is used. Demographic Factors Demographic variables such as income, e ducation, population density, urban population and age might influence fixed-broadband deploy ment. For the measurement of income, GDP per capita is used. Many studies employed GDP per cap ita for the measurement of income (Kim et al., 2001; Garcia-Murillo, 2005; Grosso, 2006; Ridder, 2007). Level of education is measured by the UNDP education index. The United Nati ons Development Programme (UNDP) education index measures a countrys relative achievement in both adult literacy and combined primary, secondary and tertiary gross enro lment. Initially an index for a dult literacy and one for combined gross enrollment are calculated and then these tw o indices are combined to create the education index, with two-thirds weight given to adult literacy and one-third weight to combined gross enrolment (UNDP, 2005). Population density is measured by population density per km2. Urban population is measured by the percentage of urban population. Th is study has interest in particular segment of age groups, so age is measured by percen tage of age between 15 and 34. ICT Factors Internet content may be rela ted to the diffusion of broadband. For the proxy measurement of content, Internet ho sts per 10000 inhabitants is employed Internet usage is measured by

PAGE 77

77 Internet users per 100 inhabitants. Teledensity is measured by main telephone lines per 100 inhabitants. To measure the PC infrastructure, estimated PCs per 100 inhabitants is used. Other Factors This study also examines the independent vari able of income and regions using categorical variable. Previous subscription could influence the diffusion of fixed-broadband. To test the influence of network effects on the fixed-broa dband deployment, this study includes previous years fixed-broadband subscribers pe r 100 inhabitants in the model. This study employs different samples for non-li near and linear regression model for fixedbroadband deployment. Through employing two differe nt samples, this study can include more independent variable for the model. For non-lin ear regression model 30 OECD countrys data from 1999 to 2006 are used. A total 240 observati ons are employed for non-linear model. For linear model ITU membership countries data from 2002 to 2006 are employed. Approximately 380 observations are available for the linear model. One-way ANOVA is also used test the digital divide research questions and the effects of triple-play strategy. The digita l divide between high/medium/low income countries and between regions (Africa, Americas, Asia, Europe, a nd Oceania) is analyzed by the one-way ANOVA by comparing a few groups of countries. This categor y of regions is based on the categorization of regions by the ITU (International Teleco mmunication Union). For the one-way ANOVA approximately 380 observations are available. Dummy variable is used for the measurement of triple-play strategy (1 for country with triple-play strategy, 0 for country without triple-play strategy). 29 observations are available for the one-way ANOVA analysis. To capture common and different factors of broadband diffusion, th is study also compares significant factors of broadband diffusion for developing and developed countries. 214 obser vations were available for

PAGE 78

78 Table 4-1. Variables, measurement and da ta sources for fixed-broadband deployment (Non-linear regression model) Variables Measurement Data Sources Fixed-broadband deployment Fixed-broadband subscribers per 100 inhabitants OECD (1999-2006) Income GDP per capita ITU (1999-2006) PC Infrastructure Estimated PCs per 100 inhabitants ITU (1999-2006) LLU Policy I Dummy (1 for with full unbundling, line sharing, bit stream access, no LLU price regulation, 0 for otherwise) OECD (1999-2006) LLU-Policy II LLU-Policy III Population density Dummy (1 for with full unbundling, line sharing, bit stream access, with LLU price regulation, 0 for otherwise) Dummy (1 for with full unbundling, line sharing, bit stream access, with LLU price regulation, 0 for otherwise) Population density (per km2) OECD (1999-2006) OECD (1999-2006) OECD (1999-2006) ITU (1999-2006) Internet content Internet hosts per 10000 inhabitants ITU (1999-2006) Platform Competition Education HHI (Herfinall-Hirschman Index) for different fixed-broadband platforms UNDP Education index OECD (1999-2006) UNDP (1998-2007) Bandwidth Political freedom Economic freedom International Internet Bandwidth (Bits per inhabitant) Inverse of the scor e on civil liberties Index of economic freedom ITU (1999-2006) Freedom House (19992006) Heritage Foundation (1999-2006)

PAGE 79

79 Table 4-2. Variables, measurement a nd data Sources for fixed-broadband (Loglinear regression model) Variables Measurement Data Sources Fixed-broadband deployment Fixed-broadband subscribers per 100 inhabitants ITU (2002-2005) Income GDP per capita ITU (2002-2005) PC Infrastructure Estimated PCs per 100 inhabitants ITU (2002-2005) Platform Competition HHI (Herfinall-Hirschman Index) for different fixed-broadband platforms ITU (2002-2005) Population Density Popul ation density (per km2) ITU (2002-2005) Internet Usage Internet user per 100 inhabitants ITU (2002-2005) Internet Content Internet hosts per 100 inhabitants ITU (2002-2005) Mobile Price Speed Education Per minute local call (USD) peak charge Broadband speed (Kbit/s) UNDP education index ITU (2002-2005) ITU (2002-2005) UNDP (2002-2005) Urban Population Telecommunication Infrastructure Investment Teledensity Previous Penetration Bandwidth Age Political Freedom Economic Freedom Fixed-broadband Price Percentage of urban population Annual telecommunication investment (USD) Main telephone lines per 100 inhabitants Previous years fixed-broadband subscribers per 100 inhabitants International Internet bandwidth (Bits per inhabitant) Percentage of age between 35-44 Inverse of the scor e on civil liberties Index of economic freedom Lower speed monthly charge (USD) Euromonitor (2002-2005) ITU (2002-2005) ITU (2002-2005) ITU (2001-2004) ITU (2002-2005) World Bank (2002-2005) Freedom House (20022005) Heritage Foundation (2002-2005) ITU (2002-2005) developing countries and 166 observa tions were available for develo ped countries. Most of data were collected from ITU, OECD, and World Bank. Measurement, Data and Statistical Methods for Mobile Broadband Deployment Table 4-3 shows the variables, their measurement, and th e corresponding data sources. Mobile diffusion can be measured either at the household or individual level. Wareham and Levy (2002) used the proportion of households that owned a mobile telephone to measure mobile diffusion. Most other mobile studies (Madden et al. 2004; Koski & Kretschmer, 2002; Gruber 2001; Ahn & Lee, 1999) used mobile penetration rates at the individu al level for such a

PAGE 80

80 measurement. In the context of this study, mobile broadband diffu sion rate (dependent variable) is measured by the number of mobile broadband subscribers per 100 inhabitants in a country. Policy Factors To examine the standardization policy factor, a dummy variable (0 or 1) is employed (i.e., if a country employed multiple standards, 1 wa s coded). Licensing approaches is coded as a categorical variable of beauty contest, auction, or hybrid licen sing system. Political freedom is measured by the inverse of the score on civil liberties (originally ranging from 1 to 7; Andonova, 2006). For the measurement of economic freedom the index of economic freedom index has been used. Industry Factors Because of the variabilit y of mobile services and thus th eir pricing, this study adopted per minute local call peak charge (USD) to measure th e cost of mobile services in each country. Previous studies have used the local call peak charge (USD) per minute (or per month) measure to indicate the relative level of prices for resi dential mobile voice services (ITU, 2005; Rouvinen, 2006). Regarding the factor of non-voice mobile appl ications, the cost of short message services (SMS) is employed as the price proxy for mobile broadband rele vant applications. SMS is a feature available in many new digital phones that le ts users receive and se nd short text messages. Fixed-broadband price is measured by broadband monthly charge (in U.S. Dollars). This study also includes telecommunication infrastructure invest ment in the empirical model. It is measured by annual telecommunication investment (USD ). For the measurement of bandwidth, international Internet bandwidth (b its per inhabitants) is used. Demographic Factors In terms of demographic variables, level of education could be meas ured by illiteracy rate and average education/degree level (Garcia-Murillo, 2005; Clements & Abramowitz; 2006). For

PAGE 81

81 the measurement of education, this study uses the UNDP education index. The UNDP education index measures a countrys relative achievement in both adult literacy and combined primary, secondary and tertiary gross en rollment. Studies have used a share of urban population to measure the demographic aspect of population de nsity (Gruber, 2001; Liikanen et. al, 2004; Koski & Kretschmer, 2002). In this study, popula tion density is measured by population per km2. Urban population is measured by the percentage of urban population and age is measured by percentage of age between 35 and 44. In the previous study, Ridder (2007) suggested only age groups 35-39 and 40-44 were correlated with fixed-broadband deployment in his correlation study. For the measurement of income, the typical GDP per capita is used. ICT Factors Teledensity is measured by main telephone lines per 100 inhabitants. To assess PC infrastructure, estimated PCs per 100 inhabitants are used. For the proxy measurement of content, Internet hosts per 10000 inhabitants is employed. Internet usage is measured by Internet users per 100 inhabitants. Previous study by Liikan en et al. (2004) sugge sted 1G (2G) has a positive(negative effect) on 2G (1G) diffusion. Based on this empirical result, this study examines relationship between 3G (mobile broadband) mobile penetr ation and 1G and 2G penetration. 1G and 2G pene tration is gauged by 1G and 2G mobile subscribers per 100 inhabitants in a country. This study also examines the independent variab le of income and regions using categorical variable to examine the digital divide issue. For the measurement of quadruple play strategy, dummy variable (1 for country with quadruple-play0 for country no multiple play strategy) was used. Data were collected from ITU, OECD, World Bank, IDAT E, Haritage Foundation and Freedom House. This study applied the statistical analyses of log linear regression analysis, and

PAGE 82

82 one-way ANOVA to assess the influential factors of mobile broadband deployment. A total of 106 observations were available for regr ession analysis and one-way ANOVA. Table 4-3. Variables, measurement and da ta sources for mobile broadband adoption Variables Measurement Data Sources Mobile broadband deployment Mobile broadband subscribers per 100 inhabitants ITU (2004-2006), IDATE (2006) Economic Freedom Index of economic freedom Haritage Foundation (2004-2006) Political Freedom Licensing Policy Standardization Policy Price of Fixedbroadband Service Mobile Service Price Inverse of the scor e on civil liberties Categorical variable (Auction, beauty contest, Hybrid approach) Dummy (1 for multiple standard, 0 for single standard) Lower speed broadband monthly charge (USD) Per minute local call (USD) peak charge Freedom House (2004-2006) ITU (2001) ITU (2004-2006) ITU (2004-2006) ITU (2004-2006) Income GDP per capita ITU (2004-2006) PC Infrastructure Estimated PCs per 100 inhabitants ITU (2004-2006) Education UNDP educati on index UNDP (2004-2006) Population Density Popul ation density (per km2) ITU (2004-2006) Internet Usage Urban Population Telecommunication Infrastructure Investment Teledensity Mobile Application Cost Age Content Bandwidth 1G and 2G Penetration Internet user per 100 inhabitants Percentage of urban population Annual telecommunication investment (USD) Main telephone lines per 100 inhabitants Cost of SMS service Percentage of Age between 35-44 Internet hosts per 100 inhabitants International Internet bandwidth (bits per inhabitant) 1G and 2G mobile subscribers per 100 inhabitants ITU (2004-2006) Euromonitor (20042006) ITU (2004-2006) ITU (2004-2006) ITU (2004-2006) World Bank (20042006) ITU (2004-2006) ITU (2004-2006) ITU (2004-2006) Measurement, Data and Statistical Methods for Ubiquitous Broadband Deployment Table 4-4 shows the variable s, their measures, and the corresponding data sources for ubiquitous broadband adoption. In this study broa dband services can be deployed through fixed and mobile networks. Therefore this study also examines adoption factors of total broadband

PAGE 83

83 deployment (fixed plus mobile). Total broadband subscribers are obtaine d by summing the totals of fixed and mobile broadband subscribers. To tal broadband subscribers per 100 inhabitants are the broadband penetration obtained by dividi ng by the total populati on and multiplying by 100 (ITU, 2006). Policy Factors DSL is a dominant fixed-broadband technology in most of countries. Open access policy (LLU) in DSL markets might be a main driver of the ubiquitous broadband deployment (ITU, 2003b). For the measurement of standardization/platform competition policy, three different types were categorized: platform competition on ly in fixed-broadband market and no mobile broadband services (Policy I); platform co mpetition in fixed-broadband markets and single standardization policy in mobile broadband markets (Policy II); and platform competition in fixed-broadband markets and multiple standardization policy in mobile broadband markets (Policy III). For all three types of LLU policy, dummy variables are employed (1 for with LLU, 0 for no LLU). Political freedom is measured by the inverse of the score on civil liberties (Andonova, 2006). For the measurement of economic freedom, the index of economic freedom index has been used. Industry Factors Network competition between fixed and mobile networks might be an influential factor of ubiquitous broadband deployment. Ubiquit ous broadband access means broadband access anytime, anywhere with anything through fixed and mobile broadband (ITU, 2006). In the empirical model, network competition is measured by dummy variable (1 for DSL and cable modem is available for fixed-broadband and multiple standards for mobile broadband, 0 for otherwise). This study adopted per minute local ca ll (USD) peak charge to measure the cost of mobile services in each country Regarding the factor of non-voice mobile applications, the cost

PAGE 84

84 of short message services (SMS) is used as the price proxy for mobile broadband relevant applications. Fixed-broadband price is measured by broadband monthly charge (in U.S. Dollars). Telecommunication infrastructure investment is measured by a nnual telecommunication investment. For the measurement of bandwidth, international Internet bandwidth (bits per inhabitant) is employed. Demographic Factors In terms of demographic variables, level of education could be meas ured by illiteracy rate and average education/degree level. For the meas urement of education, this study employs the UNDP education index. A share of urban population is used to m easure the demographic aspect of population density (Gruber, 2001; Liikanen et al, 2004; Koski & Kretschmer, 2002). In this study, population density is measured by population per km2. Age is measured by percentage of age between 35-44. For the measurement of income, the GDP per capita is used. ICT Factors To measure the PC infrastructure, estimated PCs per 100 inhabitants are used. Teledensity is measured by main telephone lines per 100 inhab itants. For the proxy measurement of content, Internet hosts per 10000 inhabitants is employed. Internet usage is measured by Internet users per 100 inhabitants. This study also examines th e independent variable of income and regions using categorical variable. For the measurement of multiple play strategy, dummy variable (1 for country with multiple play strategy triple-pla y or quadruple-play 0 for country no multiple play strategy) was used. Most of the secondary data ha s been collected fr om the International ITU, OECD, World Bank, Haritage Foundation, and Freedom House. This study employs the stat istical analyses of linear regression analysis and one-way ANOVA to assess the influential factors of broadband

PAGE 85

85 deployment. A total of 216 observations were av ailable for log linear regression analysis and one-way ANOVA. Table 4-4. Variables, measurement and data sources for ubiquitous broadband deployment Variables Measurement Data Sources Total broadband deployment Total broadband subscribers per 100 inhabitants ITU (2004-2005) Policy I Policy II Policy III LLU Economic Freedom Dummy (1 for with platform competition in fixed-broadband markets and no standardization in mobile broadband markets, 0 for otherwise) Dummy (1 for with platform competition in fixed-broadband markets and single standardization policy in mobile broadband markets, 0 for otherwise) Dummy (1 for with platform competition in fixed-broadband markets and multiple standardization policy in mobile broadband markets, 0 for otherwise) Dummy (1 for with LLU, 0 for no LLU) Index of economic freedom ITU (2004-2005) ITU (2004-2005) ITU (2004-2005) OECD (2004-2005) Haritage Foundation (2004-2005) Political Freedom Network Competition Price of fixed-broadband Service Mobile Service Price Inverse of the scor e on civil liberties Dummy (1 for DSL and cable modem for fixed-broadband and multiple standards for mobile broadband are available, 0 for otherwise) Broadband monthly charge (USD) Per minute local call (USD) peak charge Freedom House (20042005) ITU (2004-2005) ITU (2004-2005) ITU (2004-2005) Income GDP per capita ITU (2004-2005) PC Infrastructure Estimated PCs per 100 inhabitants ITU (2004-2005) Education UNDP Educati on Index UNDP (2004-2005) Population Density Popul ation density (per km2) ITU (2004-2005)

PAGE 86

86 Table 4-4 Continued. Internet Usage Urban Population Telecommunication Infrastructure Investment Teledensity Age Content Bandwidth Speed Internet user per 100 inhabitants Percentage of urban population Annual telecommunication investment Main telephone lines per 100 inhabitants Percentage of Age between 35-44 Internet hosts per 10000 inhabitants International Internet bandwidth (bits per inhabitant) Broadband speed (Kbit/s) ITU (2004-2005) Euromonitor (2004-2005) ITU (2004-2005) ITU (2004-2005) ITU (2004-2005) World Bnak (2004-2005) ITU (2004-2005) ITU (2004-2005) ITU (2004-2005)

PAGE 87

87 CHAPTER 5 RESULTS This chapter provides the results of the data an alysis. The first part provides the description of the secon dary data set and descriptive statistics for fixed, mobile and ubiquitous broadband. In the second part of this chapter, the result of regression analys is for fixed broadband deployment is provided. The third part provides common and different factors of fixed broadband deployment for developed and developing countries. In the fourth part of th is chapter, the result of one-way ANOVA for fixed broa dband deployment is provided. The fifth part provides the result of regression analysis for mobile broadband deployment. In the sixth part of this chapter, the result of one-way ANOVA for mobile broadband is provided. The seventh part provides the result of regression analysis for ubiquitous broadband deployment. Th e eighth part provides common and different factors of ubiquitous broadband deployment for developed and developing countries. In the final part of th is chapter, the result of one-way ANOVA for ubiquitous broadband deployment is provided. Data and Descriptive Statistics This study employs secondary data for all different empirical models of broadband deployment. Data are mostly collected from international organizations such as the ITU (International Telecommunication Union), OECD (Organization for Economic Co-operation and Development, and World Bank. For the nonlin ear regression model of fixed broadband deployment, a total of 240 observations were av ailable, which covers data from 1999 to 2006. This nonlinear model of fixed broadband depl oyment is based on the observations from 30 OECD countries. This data set has advantages and disadvantages for the data analysis. This OECD data set allows more years of observations (e.g. 8 years period), bu t the variation of data by country income is small since OECD count ries consist of 30 comparatively developed

PAGE 88

88 countries. However, this data set allows us to examine variables such as different types of LLU policies such as full unbundling, line sharing, an d bit stream access, which other data sets (e.g, ITU membership countries) cannot provide. Data set for the nonlinear model allows 12 different independent variables. For the linear regression model of fixed broadband diffusion, 380 observations were available. This linear regression model is based on the ITU data set, which allows 4 years of observations. For the ITU data set, the variation of data by country income is larger than that of data from the OECD countries. Table 5-4 provides the list of selected countries examined for the linear model of fixed broadband deployment. This sufficient variation of data by country allows us to examine differences in the significant factors of fixed broadband deployment between developed and developing countries. The data se t for the linear regression model allows observations of 18 different inde pendent variables. Table 5-1 pr ovides descriptive statistics for the linear model of fixed broadband deployment. Mean of the fixed broadband penetration among ITU membership countries between 2002 a nd 2005 is 4.33 per 100 inhabitants and the standard deviation for the fixed broadband penetration is 6.16. For the linear regression model of mobile broadband, a total of 106 observations were available. This linear regression model of mobile broadband is based on the ITU data set, which allows 3 years of observations. Table 5-5 provides the list of selected countries examined for the linear model of mobile broadband deployment. For this data set, the number of observations for low income countries was small, since m obile broadband is comparatively new media technology for developing countries This insufficient variation of data by country income didnt allow for the examination of differences in the significant factors of mobile broadband deployment between developed and developing count ries. The data set for the linear regression

PAGE 89

89 model of mobile broadband provide s observations of 18 different i ndependent variables. Table 52 offers descriptive statistics for the linear mode l of mobile broadband deployment. Mean of the mobile broadband penetration am ong ITU membership countries (for countries which mobile broadband is available) 4.65 per 100 inhabitant s (standard deviation for the mobile broadband penetration: 9.10). For the linear regression model of ubiqu itous broadband deploy ment, 216 observations were available. This linear regression model of ubiquitous broadband di ffusion is also based on the ITU data set, which allows 2 years of observations between 2004 and 2005. Table 5-6 provides the list of selected countries examin ed for the linear model of ubiquitous broadband diffusion. This data set provides sufficient variati on of data by country income and allows us to examine differences in the significant fact ors of ubiquitous broa dband diffusion between developed and developing countries. The data se t for the linear regression model allows observations of 17 different indepe ndent variables. Table 5-3 offers descriptive statistics for the linear model of ubiquitous broadband diffusion. Mean of the ubiquitous broadband penetration among ITU membership countries between 2004 a nd 2005 is 6.96 per 100 inhabitants and the standard deviation for the ubiqui tous broadband penetration is 10. 79. For all empirical models, some observations for a portion of the independent variables were missing. Therefore the number of observations employed for the real regressi on analysis was smaller than the number of observations proposed. Regression Analysis of Fixed Broadband Deployment Nonlinear Regression Model Data for nonlinear regression model covers all 30 OECD countries from 1999 to 2006. This study estimates the variable -speed logistic model describe d in equations (1) and (3) by

PAGE 90

90 Table 5-1. Descriptive statistic s for fixed broadband deployment Variables Mean Standard Deviation Fixed broadband deployment 4.33 6.16 Income 11973.28 12316.92 PC Infrastructure 23.59 21.64 Platform Competition 7181.97 2603.45 Population Density 520.07 2390.51 Internet Usage 26.62 21.36 Internet Content 435.36 877.99 Mobile Price Speed Education 0.35 896.48 0.88 0.33 2285.13 0.10 Urban Population Telecommunication Investment Teledensity Previous Penetration Bandwidth Age Political Freedom Economic Freedom Fixed Broadband Price 67.91 5E+010 31.25 2.86 1997.81 14.19 2.49 64.81 100.09 20.34 3.5930E+011 18.85 4.76 4336.00 2.37 1.64 9.71 343.76 Table 5-2. Descriptive statistics for mobile broadband deployment Variables Mean Standard Deviation Mobile broadband deployment 4.65 9.10 Economic Freedom 67.83 9.17 Political Freedom Standardization Policy Price of Fixed Broadband Service Mobile Service Price 1.89 0.27 67.83 0.35 1.41 0.44 123.61 0.19 Income 21252.75 16057.63 PC Infrastructure 38.18 25.93 Education 0.92 0.087 Population Density 344.40 1124.79 Internet Usage Urban Population Telecommunication Investment Teledensity Mobile Application Cost Age Content Bandwidth 1G and 2G Mobile Penetration 39.48 72.70 1E+010 70.15 0.10 14.79 854.53 4253.74 73.60 22.09 14.78 6.452+010 32.10 0.07 2.04 1246.77 6442.72 30.79

PAGE 91

91 Table 5-3. Descriptive statistics for ubiquitous broadband deployment Variables Mean Standard Deviation Total Broadband Deployment 6.96 10.79 Policy I Policy II Policy III Economic Freedom 0.26 0.32 0.11 64.00 0.44 0.47 0.32 9.60 Political Freedom Network Competition Price of Fixed Broadband Service Mobile Service Price 2.51 0.43 118.54 0.30 1.64 0.49 446.53 0.21 Income 11955.70 13275.03 PC Infrastructure 24.91 23.85 Education 0.87 0.10 Population Density 503.73 2506.82 Internet Usage Urban Population Telecommunication Infrastructure Investment Teledensity Age Content Bandwidth Speed 27.72 66.82 6E+010 29.45 14.09 455.77 2247.34 918.18 21.69 20.26 4.2800E+011 18.45 2.31 959.82 4886.65 2012.05

PAGE 92

92 Table 5-4. Selected countries examined for fixed broadband deployment (ITU, 2005) Total Total Total fixed fixed fixed broadband broadband broadband subscribers subscribers subscribers per 100 per 100 per 100 Country inhabitants Country i nhabitants Country Inhabitants Algeria 0.59 Germany 13.06 Panama 0.54 Andorra 15.4 Greece 1.44 Peru 1.26 Antigua 7.02 Guatemala 0.22 Philippines 0.15 Argentina 2.4 Guyana 0.27 Poland 2.45 Aruba 12.34 Hong Kong, China 23.56 Portugal 11.11 Australia 8.87 Hungary 6.45 Puerto Rico 2.99 Austria 14.36 Iceland 26.54 Qatar 3.23 Bahrain 2.95 India 0.12 Romania 3.48 Barbados 11.87 Ireland 7.77 Russia 1.11 Belgium 19.13 Israel 17.82 San Marino 4.52 Belize 1.86 Italy 11.74 Saudi Arabia 0.28 Bermuda 28.81 Jamaica 1.7 Senegal 0.15 Bosnia 0.35 Japan 18.19 Seychelles 1.18 Brazil 2.35 Jordan 0.42 Singapore 15.3 Brunei 2.17 Korea 25.24 Slovak Republic 3.36 Canada 19.84 Kuwait 0.93 Slovenia 10 Cape Verde 0.2 Latvia 2.63 South Africa 0.35 Chile 4.54 Lebanon 3.63 Spain 11.79 China 2.84 Liechtenstein 24.76 St. Vincent 3.06 Colombia 0.7 Lithuania 6.83 Suriname 0.25 Costa Rica 1.08 Luxembourg 15.08 Sweden 21.21 Croatia 2.55 Macau 14.79 Switzerland 21.77 Cyprus 3.82 Malaysia 1.95 TFYR Macedonia 0.61 Czech Republic 6.94 Maldives 1.08 Taiwan, China 19.06 Denmark 24.75 Malta 11.14 Thailand 0.16 Dominica 0.74 Mauritius 0.25 Tonga 0.64 Ecuador 0.2 Mexico 1.8 Trinidad & Tobago 0.83 Egypt 0.13 Moldova 0.25 Tunisia 0.17 El Salva 0.61 Morocco 0.82 Turkey 2.17 Equatorial Guinea 0.04 Netherlands 25.15 Tuvalu 1.43

PAGE 93

93 Table 5-4 Continued. Estonia 13.48 New Caledonia 4.05 United Arab Emirates 2.88 Faroe Islands 12.48 New Zealand 8.22 United Kingdom 16.58 Fiji 0.83 Nicaragua 0.18 United States 16.1 Finland 22.37 Norway 21.46 Uruguay 1.88 France 15.66 Oman 0.32 Venezuela 1.33 Note. Data were derived from the International Telecommunication Union (2006). Source: ITU Internet Reports 2005. Geneva: ITU. Table 5-5. Countries examined for mobile broadband deployment (ITU, 2005) Country 3G mobile subscribers per 100 inhabitants (2004) Number of 3G mobile subscribersCountry 3G mobile subscribers per 100 inhabitants (2004) Number of 3G mobile subscribers Angola 1.598 225000 Latvia .044 1000 Argentina .052 20000 Luxembourg .828 3810 Australia 3.862 76900 Mauritius .041 500 Austria 2.488 20200 Mexico .019 20000 Bahrain .128 950 Moldova .070 3000 Belgium .013 1391 Netherlands .166 27000 Brazil .945 1706660 New Zealand 18.721 732000 Canada 23.314 7400000 Nicaragua .357 20000 Chile .485 74730 Norway .154 7000 China .066 8711300 Peru 1.088 300000 Colombia .011 5000 Poland .003 1000 Czech Republic .479 49000 Portugal .983 99000 Denmark 2.317 124650 Romania 1.254 279408 Ecuador 1.516 200000 Russia .127 181000 Finland .141 7361 Slovenia .318 6300 France .063 38000 South Africa .007 3000 Germany .297 245000 Spain .187 77000 Greece .084 9250 Sweden 3.241 288150 Guatemala .395 50000 Switzerland .140 10000 Hong Kong 2.949 210000 Taiwan 1.318 300000 Indonesia .022 50000 Thai .969 615000 Ireland .075 3000 UAE .176 5370 Israel 27.790 1823000 UK 4.765 2832000 Italy 4.882 2800000 USA 16.681 49550000 Japan 20.110 25700000 Uruguay .006 193 Kazakh .649 100000 Venezuela .764 200000 Korea 57.370 27509000 Note. Data were derived from the International Telecommunication Union (2005). Source: ITU Internet Reports 2005. Geneva: ITU.

PAGE 94

94 Table 5-6. Selected countries examined fo r ubiquitous broadband deployment (ITU, 2005) Country Total broadband subscribers per 100 inhabitants (2005) Country Total broadband subscribers per 100 inhabitants (2005) Country Total broadband subscribers per 100 inhabitants (2005) Algeria 0.7 Greece 2.5 Paraguay 0.1 Andorra 14.8 Grenada 0.6 Philippines 0.1 Argentina 2.2 Guyana 0.3 Poland 3.3 Australia 14.4 Hong Kong, China 31.8 Portugal 20.3 Austria 21.2 Hungary 6.5 Puerto Rico 0.6 Bahamas 4 Iceland 26.5 Qatar 3.2 Bahrain 4.3 India 0.12 Reunion 7.2 Barbados 11.8 Ireland 11.5 Romania 4 Belgium 19.3 Israel 20.5 Russia 1.1 Belize 1.6 Italy 29.4 Senegal 0.15 Bolivia 0.1 Japan 31.4 Seychelles 0.7 Bosnia 0.3 Jordan 0.2 Singapore 18.4 Brazil 1.9 Korea (Rep) 51.2 Slovak Republic 2.6 Bulgaria 0.5 Kuwait 0.9 Slovenia 9.9 Canada 20.9 Kyrgyzstan 0.05 Solomon Islands 0.09 Cape Verde 0.2 Latvia 2.3 Spain 13.9 Chile 4.5 Lebanon 3.6 Sri Lanka 0.1 China 2.9 Lithuania 6.8 St. Kitts and Nevis 1.1 Colombia 0.5 Luxembourg 21 St. Vincent 3.1 Costa Rica 0.9 Macao, China 14.8 Suriname 0.2 Croatia 2 Malaysia 2.1 Sweden 27.6 Cyprus 3.2 Maldives 0.6 Switzerland 24.5 Czech Republic 5.1 Malta 11.1 Taiwan, China 20.6 Denmark 27.1 Martinique 1.5 TFYR Macedonia 0.6 Dominica 4.6 Mexico 2.2 Thailand 0.1 Dominican Rep. 0.6 Moldova 0.19 Trinidad & Tobago 0.8 Ecuador 0.2 Morocco 0.8 Tunisia 0.2 Egypt 0.2 Netherlands 27.2 Turkey 2.2 El Salvador 0.6 New Caledonia 4 United Arab Emirates 3.1

PAGE 95

95 Table 5-6 Continued. Country Total broadband subscribers per 100 inhabitants (2005) Country Total broadband subscribers per 100 inhabitants (2005) Country Total broadband subscribers per 100 inhabitants (2005) Estonia 13.5 New Zealand 10.5 United Kingdom 23.6 Faroe Islands 11.7 Nicaragua 0.12 United States 18 Finland 23.8 Norway 23.9 Uruguay 0.8 France 18.3 Palestine 0.2 Venezuela 0.8 French Polynesia 4.2 Panama 0.6 Vietnam 0.25 Germany 15.7 Peru 1.2 Zimbabwe 0.09 Note. Data were derived from the Inte rnational Telecommunication Union (2006). Source: ITU Internet Reports 2005. Geneva: ITU. nonlinear least squares, after adding disturbances to equati on (1) (see pp. 66-68). Table 5-7 provides the results (see nonlinear model part in table 5-7). PC penetration was associated with higher broadband penetration levels. PC penetration was statistically significant at the .01 level. High level of education, population density, and Internet content (the number of Internet hosts per 10000 inhabitants) were statistically significant at the .05 level. The main inte rest of this nonlinear model of fixed broadband diffusion is the effect of LLU policy. All three types of LLU policy vari ables are statistically significant at the 1 percent level. This may mean that LLU policy I (with full unbundling, line sharing, bit stream access and without LLU price regu lation), LLU policy II (with full unbundling, line sharing, no bit stream access and with LLU price regulatio n) and LLU policy III (with full unbundling, line sharing, bit stream access and with LLU price re gulation) have contribut ed high level of fixed broadband penetration.1 1 Most of OECD countries have two types of LLU policy. One major type of LLU was LLU policy, which has full unbundling, line sharing, and bit stream access and the other major type of LLU in OECD countries was LLU policy, which has full unbundling and line sharin g without bit stream access. With tw o different types of LLU, for this empirical study, interaction of these two types and LLU price regulation. Only three cases of LLU policies were

PAGE 96

96 Table 5-7. Results of regression s of fixed broadband deployment Nonlinear Model Log linear (Extended Model) Log linear (Reduced Model) Variable Coefficients B t-stat Coefficients B t-stat Coefficients B t-stat Constant Ceiling 26.14 19.85*** .62 .92 .02 .13 Initial level Parameter -3.52 -13.08*** Speed .04 .80 Fixed broadband price -.09 -.03* -.10 -1.9* Mobile price .08 1.8* .06 1.4 Education .47 2.11** -.08 -1.7 Internet use .35 3.98*** .31 4.28*** Population density B<.001 2.07** .002 .067 Bandwidth 1.07E-06 0.30 .13 3.75*** .13 4.15*** Content B<.001 2.77** -.05 -1.47 Political freedom Economic freedom Urban population Age (35-44) Platform competition Previous penetration Telecom investment Teledensity Income PC Penetration -.023 -.004 B<-.001 -1.13E-06 .004 -.40 -1.61 -1.07 -0.63 3.09*** -.18 -.60 .07 .27 -.07 .65 -.007 -1.85* -1.58 .58 .78 -1.84* 21.5*** -.41 -.10 -.08 .63 -1.3 -2.1** 23.01*** LLU Policy Type I .21 3.39*** LLU Policy Type II LLU Policy Type III .19 .15 2.81*** 3.00*** R-Squared Number of observations 0.93 217 0.92 255 0.91 282 Statistically significant at the 10% level ** Statistically significant at the 5% level ***Statistically significant at the 1% level High level of platform competition, which is measured by HHI (Herfindall-Hirschman Index) was related to high level of fixed br oadband penetration, but it was not statistically identified and included in the model after careful check of correlation between these different types of LLU cases, which might not lead to multicollinearity in the model.

PAGE 97

97 significant at the .10 level (p -value:.29). Bandwidth political freedom, and economic freedom were not significant in the model. To check multicollinearity issue in this model, correlation analysis was also conducted. Based on the 80 benchmark, there were no highly correlated independent variables in the model. R-squared for this model was .932 and adjusted R-squared was .930. Linear Regression model A total of 380 observations were analyzed employing the multiple regression analysis. Extended and reduced model were identified from the data analysis. Note that dependent variable and independent variables were transfor med using logarithmic function since data were positively skewed. Extended model. Initially, all eighteen independent variables were included for the multiple regression analysis. Multicollinearity issue might occur when independent variables are highly correlated, a correlation an alysis conducted to check potenti al multicollinearity problems. To assess the strength of corre lations, the .80 Pearson correlat ion criterion was employed. PC penetration, teledensity, and income were rem oved from the initial model because of its high correlation with other in dependent variables. Table 5-7 sh ows the ANOVA table of the extended regression model, which illustrates the models significance at the .01 level (F statistic: 199.15, P <. 001). Specifically, Internet use, bandwidth, and previous penetration were statistically significant at the .01 level. Other independe nt variables such as fixed br oadband price, political freedom, mobile price, and platform competition were st atistically significant at the .1 level. Other variables such as speed, educat ion, population density, content, urban population, age (35-44), telecommunication investment were not statistically significant. R-squared for the extended model was .926.

PAGE 98

98Reduced model. To check the stability of results in the empirical study, non-significant variables such as speed, educat ion, population density, content, urban population, age (35-44), and telecommunication investment were removed fr om the reduced model. The reduced model is significant at the .01 level (F statistic: 357.76). In this model, mobile price was positively related to the dependent variable, but it was not statistically significant at the .1 level. Internet use, bandwidth, and previous penetratio n were statistically significant at the .01 level, and platform competition was significant at the .05 level. Also lower price of fixed broadband was associated with higher level of fixed broadband penetrati on. R-squared for the reduced model was .915. Table 5-7 provides the results of the reduced model from the regression analysis. Results of Fixed-Broadband Deployment fo r Developed and Developing Countries A total of 132 observations were analyzed employing the multiple regression analysis for developed countries and a total of 148 observations were analyzed for developing countries. Rsquared for the model for developed countries was .90, and R-squared for the model for developing countries was .81. De pendent variable and independent variables were transformed using logarithmic function since data were positively skewed. Regression Analysis: Developed Countries In the initial model, all eighteen independent variables were included for the multiple regression analysis. PC penetrati on, teledensity, and Internet us e were removed from the initial model because of its high correla tion with other independent variables. For the extended model, some insignificant variables such as political freedom, economic freedom urban population, age (35-44), and telecommunication investment were removed from the model. In the reduced model, other insignificant variables such as speed, fixed broadband price, bandwidth, and platform competition were also removed from the model. Table 5-8 provides the ANOVA table

PAGE 99

99 of the reduced regression model, which illustrates the models significance at the .01 level (P <. 001). In the reduced model, education and previous penetration were statistically significant at the .01 level. Other independent variables such as income, population density, and content were statistically significant at the .05 level. Mobile price was negatively associated with high level of fixed broadband diffusion at the .1 level in the developed countries. The results of analysis of developed countries were consiste nt with the results of analys is of OECD countries. In both models, high level of education, population density, and Internet content were statistically significant. Regression Analysis: Developing Countries Initially, all eighteen independent variable s were included for the multiple regression analysis. PC penetration, teledensity, and income were removed from the initial model because of its high correlation with other variables. For the extended model, some insignificant variables such as education, population density, content, economic freedom, urban population, age (35-44), and telecommunication investment were removed fr om the model. For th e reduced model, other insignificant variable such as speed was also removed from the model. Table 5-8 provides the results of the extended and reduced regression model, which illustrates the models significance at the .01 level (P <. 001). The result of analysis for developing countri es was very different from the result of developed countries. In the reduced model, bandwidth and previous penetration were statistically significant at the .01 level. Internet use was statis tically significant at the .05 level. Mobile price was positively associated with high level of fixed broadband diffusion at the .1 level in the developing countries instead of negative association in the developed countries. Fixed broadband price and HHI was negatively asso ciated with the high level of fixed broadband penetration.

PAGE 100

100 Table 5-8. Results of regressions of fixed br oadband penetration for developed and developing countries Developed Countries Developing Countries Variable Extended model Reduced model Extended model Reduced model Coefficients B t-stat Coefficients B t-stat Coefficients B t-stat Coefficients B t-stat Constant -.17 -.40 -.48 -1.52 -.28 -.66 -.20 -.70 Speed .04 1.46 .18 1.35 Fixed broadband price -.10 -1.61 -.14 -1.80* -.13 -1.67* Income Mobile price .13 -.06 1.43 -1.79* .16 -.06 2.13** -1.69* .13 1.75* .14 1.83* Education 1.91 2.96*** 2.28 3.79*** Internet use .19 1.66* .24 2.14** Population density .03 1.68* .04 2.50** Bandwidth .03 .90 .11 1.97* .15 2.71*** Content .05 1.90* .06 2.55** Political freedom Economic freedom Urban population Age (35-44) Platform competition Previous penetration Telecom investment Teledensity PC Penetration -.05 .62 -.74 20.71*** .65 24.27*** -.23 -1.55 -.10 -1.90* .61 13.89*** -.23 -1.54 -.09 -1.72* .61 14.09*** R-Squared Number of observations 0.90 132 0.90 132 0.81 148 0.81 148 Statistically significant at the 10% level. ** Statistically significant at the 5% level. ***Statistically significa nt at the 1% level One-way ANOVA Analysis of Fixed-Broadband Deployment Table 5-9 provides the data analysis em ploying one-way ANOVA. Mean difference of fixed-broadband penetration between high, medium and low income countries was statistically significant (F-statistic: 123.49, P <.01). Higher income countries tend to have higher fixedbroadband penetration. This result may suggest that there is digital divide between countries by income.

PAGE 101

101 Mean difference of fixed-broadband penetrati on between regions (Africa, America, Asia, Europe, and Oceania) was also st atistically significant (F-statistic: 11.75, P <.01). European and Asian countries tend to have highe r fixed-broadband penetration th an other countries in Africa, America, and Oceania. This result suggests that there is digital divide between countries by region. Mean difference of fixed-broadband penetration growth rate between countries with tripleplay strategy and without triple-play strategy was also significant (see table 5-9). This result may suggest that triple-pla y is a contributing factor of globa l fixed-broadband deployment. Also, mean difference of fixed-broadband penetration between countries with LLU policy and without LLU policy was also significant (see table 59). This result suggests that LLU policy is a contributing factor of global fixed-broadband deployment. Table 5-9. Difference in fixed broadband penetration and fixed broadband penetration growth rate by income, region, trip le-play offerings and LLU Dependent Variable Source SS df MS F Variable Mean Fixed Broadband Penetration Income 5702.61 2 2851.30 123.49** High Income 8.72 Error 8704.38 377 23.09 Medium Income Low Income 1.06 .06 Region 1605.32 4 401.33 11.75** Africa .59 Error 12801.6 7 375 34.13 America Asia Europe Oceania 2.43 4.59 6.67 2.64 Fixed Broadband Penetration Growth Rate Tripleplay 13.94 1 13.94 8.23** Triple-play 4.47 Error 47.40 28 1.69 No Triple-play 2.86 Fixed Broadband Penetration LLU 1796.99 1 1796.99 47.86** LLU 6.90 Error 5894.45 157 37.54 No LLU 4.13 Note. Categorization of countries by income a nd region are based on the ITUs categorization of countries by income and region. **Statistically significant at the 1% level

PAGE 102

102Regression Analysis of Mobile Broadband Deployment A total of 106 observations were analyzed employing the multiple regression analysis. Extended and reduced model were identified for the analysis. Note that dependent variable and independent variables were transformed using logarithmic function since data were positively skewed. Extended Model In the initial model, all eighteen independent variables were included for the multiple regression analysis. A correlation analysis was conducted to check potential multicollinearity problems. To assess the strength of correlati ons, the .80 Pearson correlation benchmark was employed. PC penetration, Internet use, bandw idth, and content were removed from the initial model because of its high correlation with ot her independent variables. Some independent variables, which have unexpected sign, such as teledensity, mobile price, political freedom, and urban population were removed from the model. Also, some insignificant variables such as education and cost of mobile application were rem oved from the initial model. Table 5-10 provides the ANOVA table of the extended regression model, wh ich illustrates the models significance at the .01 level. Specifically, multiple standardization policy and income were statistically significant at the .01 level. 1G and 2G mobile penetration was st atistically significant at the .1 level, which suggests mobile broadband is a substitute of 1G and 2G mobile. Other variables such as fixed broadband price, telecommunicatio n investment, economic freedom population density, and age (35-44) were not statistically significant. R-squared for the extended model was .46. Reduced Model To check the stability of resu lts in the empirical study, nonsignificant variables such as fixed broadband price, telecommunication i nvestment, economic freedom, population density,

PAGE 103

103 and age (35-44) were removed fr om the reduced model. The reduced model is significant at the .01 level. In the reduced model, multiple standa rdization policy, and inco me were statistically significant at the .01 level. 1G and 2G mobile penetration was negati vely correlated to the dependent variable and was signifi cant at the .05 level. Also, higher level of population density was associated with higher level of mobile br oadband penetration. R-squared for the reduced model was .44 (see Table 5-7). Table 5-10. Results of regression anal ysis of mobile broadband deployment Log linear (extended model) Log linear (reduced model) Variable Coefficients B t-stat Coefficients B t-stat Constant Standardization policy -3.71 .89 -1.23 4.85*** -4.54 .86 -7.19 5.03*** 1G and 2G penetration -.85 -1.68* -.95 -2.12** Income 1.28 4.33*** 1.33 5.63*** Fixed broadband price -.06 -.22 Age (35-44) 1.69 .93 Telecom investment .07 .86 Economic freedom Population density -1.81 .19 -1.01 1.50 .21 1.82* R-Squared Number of observations 0.46 101 0.44 105 Statistically significant at the 10% level. ** Statistically significant at the 5% level. ***Statistically significa nt at the 1% level One-Way ANOVA Analysis of Mo bile Broadband Deployment Table 5-11 offers the data analysis employing one-way ANOVA. Mean difference of mobile broadband penetration between high, medium and low income countries was statistically significant at the .1 level (F-sta tistic: 2.75, P <.1). Higher income countries tend to have higher mobile broadband penetration. This result may suggest that there is digital divide between countries by income (see Table 5-11). Mean difference of mobile broadband penetra tion between regions (Afr ica, America, Asia, Europe, and Oceania) was also stat istically significant at the .1 le vel (F-statistic: 2.38, P <.1).

PAGE 104

104 Asian countries have the highest mobile broadband penetration than other countries in other regions. This result suggests th at there is digital divide be tween countries by region. Mean difference of mobile broadband penetration growth rate between countries with quadruple-play strategy and without quadruple-play strategy was not signi ficant (see Table 5-11). This result suggests that quadruple-play is not a contributin g factor of global mobile broadband deployment yet. Also, mean difference of mobile broadba nd penetration between countries which have different licensing policy system wa s not significant (see table 5-11). Table 5-11. Difference in mobile broadband pe netration and mobile broadband penetration growth rate by income, region, quadruple-play offerings and licensing policy Dependent Variable Source SS Df MS F Variable Mean Mobile Broadband Penetration Income 442.06 2 221.03 2.75* High Income 6.14 Error 8262.58 103 80.21 Medium Income Low Income 2.06 .51 Region 750.29 4 187.57 2.38* Africa .39 Error 7954.34 101 78.75 America Asia Europe Oceania 3.10 9.44 3.55 7.19 Mobile Broadband Penetration Growth Rate Quadruple -play 73.17 1 73.17 .86 Quadruple-play 7.14 Error 2394.72 28 85.53 No quadrupleplay 10.86 Mobile Broadband Penetration Licensing Policy 654.67 3 218.22 2.09 Auction Beauty Contest Hybrid 4.18 9.15 11.97 Error 6662.07 64 104.09 Other .48 Note. Categorization of countries by income a nd region are based on the ITUs categorization of countries by income and region. Statistically significant at the 10% level Regression Analysis of Ub iquitous Broadband Deployment A total of 216 observations were analyzed employing the multiple regression analysis. For the ubiquitous broadband deployment, this st udy examines both the model with network competition variable and the model with different platform completion-standardization policy

PAGE 105

105 type variable.2 Extended and reduced model were identi fied from the data analysis for both models. Note that dependent variable and i ndependent variables we re transformed using logarithmic function since data were positively skewed. The results of both models show similar results. Model with Network Competition Variable Extended model. In the initial model, seventeen inde pendent variables were included for the multiple regression analysis. Multicollinear ity problem might occur when independent variables are highly correlate d, a correlation analysis c onducted to check potential multicollinearity problems. To assess the strength of correlations, the criterion of .80 Pearson correlations was employed. PC penetration, teledensity, Intern et use, and bandwidth were removed from the initial model because of its high correlation with other independent variables. The extended regression model was significant at the .01 level. Specifically, network competition, income, and c ontent were statistically significant at the .01 level. Fixed broadband price, speed, and politi cal freedom were statistically significant at the .05 level. Other variables such as mobile price, education, population de nsity, urban population, age (35-44), telecommunication investment, a nd economic freedom were not statistically significant. R-squared for the extended model was .81. Reduced model. In the reduced model, non-significant variables such as mobile price, education, age (35-44), teleco mmunication investment, and economic freedom were removed from the reduced model. The reduced model was significant at the .01 le vel. In the reduced 2 One goal of this empirical model was to examine effective policy variables, which might influence higher level of ubiquitous broadband diffusion. Network competition variable and different platform competition-standardization policy could be included in a single model. However, network competition variable and some of platform competition-standardization policy type were correlated with each other. Therefore this study examines two different models the model with network competition variable and the model with platform competitionstandardization policy variable.

PAGE 106

106 model, income and political freedom were statistically significant at the .01 level. Fixed broadband price, speed, content, urban population, and network co mpetition were statistically significant at the .05 level. Also, population density was statistically significant at the .1 level. Rsquared for the reduced model was .79. Table 5-12 provides the results of th e regression analysis. Model with Different Platform Compet ition-Standardization Policy Variables Extended model. Main interest of this model is the effects of different platform competition-standardization policy types. Three different types were categorized for the platform competition-standardization policy types: platfo rm competition only in fixed-broadband markets and no mobile broadband services (Policy type I); platform competition in fixedbroadband markets and single standardization policy in mob ile broadband markets (Policy type II); and platform competition in fixed-broadband markets and multiple standardization policy in mobile broadband markets (Policy type III). For the extended model, initia lly, all nineteen independent variables were included for the multiple regression analysis. A correlatio n analysis conducted to check potential multicollinearity issues. Based on the benchmark of .80 Pearson correlations, PC penetration, teledensity, Inte rnet use, and bandwidth were removed from the initial model because of its high correlation with other indepe ndent variables. Also, for the extended model, other insignificant independent va riables such as mobile price, education, age (35-44), and economic freedom were removed from the initial model. The extended model was significant at the .01 level. In particular, Policy type III (platform competition in fixed-broadband markets and multiple standardization policy in mobile broadb and markets) and income were statistically significant at the .01 level. Policy type II (pla tform competition in fixedbroadband markets and single standardization policy in mobile broa dband markets), speed, political freedom, and population density were statistica lly significant at the .05 leve l. Policy type I (platform

PAGE 107

107 competition only in fixed-broadband markets and no mobile broadband services) was significant at the .1 level. Other inde pendent variables such as fi xed broadband price, content, telecommunication investment, and urban population were not statis tically significant. R-squared for the extended model was .80. Reduced model. To check the stability of results in the empirical study, non-significant independent variable telecommunication investme nt was removed from the reduced model. The reduced model is significant at the .01 level. In the reduced model, Policy type III, income, and political freedom were statistically significant at the .01 level. Policy type II, speed, and population density were statistically significant at the .05 level. Policy type I, fixed broadband price, and urban population were significant at the .1 level. Content was not statistically significant. R-squared for the reduced model was .80. A total of 190 observations were available for this model. Table 5-12 provides the results of the reduced model from the regression analysis. Table 5-12. Results of regressions of Total (Ubiquitous) broadband deployment Model 1 (with network completion variable) Model 2 (with policy I, II, III) Variable Extended model Reduced model Extended model Reduced model Coefficients B t-stat Coefficients B t-stat Coefficients B t-stat Coefficients B t-stat Constant -4.74 -3.50*** -4.11 -7.74*** -4.61 -7.64*** -4.41 -8.29*** Speed .23 2.16** .25 2.55** .22 2.03** .23 2.36** Fixed broadband Price -.19 -2.11** -.17 -2.11** -.12 -1.45 -.13 -1.69* Income Mobile price .67 -.04 5.65*** -.491 .71 7.48*** .76 7.42*** .79 8.32*** Education -.89 -1.16 Content .20 3.05*** .12 2.34** .09 1.62 .07 .17 Political freedom -.37 .18** -.50 3.27*** -.40 -2.42** -.50 -3.35*** Age (35-44) .32 .43 Telecom investment .02 .03 .03 1.08

PAGE 108

108 Table 5-12 Continued. Economic freedom Population density Urban population Teledensity PC Penetration Internet use Bandwidth Network competition Policy Type I Policy Type II Policy Type III .06 .08 .40 .23 .08 1.54 1.50 2.67*** .09 .49 .18 1.96* 2.05** 2.32** .11 2.24** .41 1.64 .18 1.86* .25 2.46** .56 4.33** .12 2.48** .45 1.96* .16 1.84* .21 2.28* .53 4.33** R-Squared Number of observations 0.81 169 0.79 191 0.80 177 0.80 191 Statistically significant at the 10% level ** Statistically significant at the 5% level ***Statistically significant at the 1% level Results of Ubiquitous Broa dband Deployment for Developed and Developing Countries A total of 73 observations were analyzed employing the multiple regression analysis for developed countries and a total of 120 observations were analyzed for developing countries. Rsquared for the final reduced model for developed countries was .77, and R-squared for the final reduced model for developing countries was .56. To examine effects of different platform competition-standardization policy types, Policy t ype I, II, and III were included for the model. Dependent variable and independe nt variables were transformed using logarithmic function since data were positively skewed. Regression Analysis: Developed Countries In the initial model, all nineteen independent variables were included for the multiple regression analysis. PC penetra tion was removed from the initial model because of its high

PAGE 109

109Table 5-13. Results of regressions of total (Ubiquitous) broadband penetration for developing and developed Countries Developing Countries Developed Countries Variable Extended model Reduced model Extended model Reduced model Coefficients B t-stat Coefficients B t-stat Coefficients B t-stat Coefficients B t-stat Constant -2.71 -1.12 -3.48 -6.57 -1.53 -.75 -.20 -5.73*** Speed .08 .33 .13 1.69* .14 2.15** Fixed broadband price -.18 -1.60 -.11 -1.11 -.04 -.25 Income Mobile price .57 .17 2.62** 1.08 .79 4.24*** .30 1.19 -.05 -.65 Education .41 .21 Content .07 .95 Political freedom -.03 -.12 -.23 -.96 -.29 -2.18** Age (35-44) -.94 .84 Telecom investment .01 .31 .004 .07 Economic freedom Population density Urban population Teledensity PC Penetration Internet use Bandwidth Network competition Policy Type I Policy Type II Policy Type III .11 .12 .25 .05 .31 .23 .36 .36 .10 .35 1.02 .29 2.62** 1.86* 2.66*** 1.96* -.001 .25 .19 .33 .39 -.005 2.96*** 1.74* 2.73*** 2.13** -1.25 -1.22 .14 2.94*** .24 .44 .86 2.90*** .14 1.3 .37 1.85* .29 1.50 .59 2.94*** .09 2.58** .91 3.65*** .21 3.05*** .25 1.75* .23 1.72* .54 4.01*** R-Squared Number of observations 0.69 94 0.56 120 0.79 69 0.77 73 Statistically significant at the 10% level. ** Statistically significant at the 5% level. ***Statistically significant at the 1% level.

PAGE 110

110 correlation with other in dependent variables. Also, for the extended model, some insignificant variables, which have unexpected sign like teledensity and age, were removed from the model. In the reduced model, other insignificant vari ables such as fixed broadband price, income, mobile price, education, content, teleco mmunication investment, urban population, and economic freedom were also removed from th e model. Table 5-13 provides the ANOVA table of the reduced regression model, which illustrates th e models significance at the .01 level (P <. 001). In the reduced model, Policy type III, ba ndwidth, and Internet use were statistically significant at the .01 level. Speed, political fr eedom, and population density were statistically significant at the .05 level. Also, policy type I and policy II were si gnificant at the .1 level in the developed countries. Regression Analysis: Developing Countries Initially, all nineteen indepe ndent variables were included fo r the regression model. Some insignificant variables such as education, content, and population density, which have unexpected sign, were removed from the model. Al so, insignificant independe nt variable Internet usage was removed from the model. In the reduce d model, other insignifica nt variables such as speed, political freedom, age (35-44), telecommu nication investment, economic freedom, urban population, and PC penetration we re removed from the model. Ta ble 5-13 provides the results of the extended and reduced regression model. The result of analysis for developing countries was different from the result of developed countries. In the reduced model, policy type II, income, and bandwidth were statistically significant at the .01 level. Policy type III was sta tistically significant at the .05 level. Policy type I was significant at the .1 level in the devel oping countries. Consideri ng significance level of Policy type II and policy type III, it appears that in the developing countries policy type II (platform competition in fixed-broadband mark et and single standard for mobile broadband

PAGE 111

111 market) is more effective than policy type III (platform competition in fixed-broadband market and multiple standards for mobile broadband market). In the developed countries, policy type III was more significant variable than policy type II. One-Way ANOVA Analysis of Ubi quitous Broadband Deployment Table 5-14 provides the result of one-way ANOVA analysis of ubiquitous broadband deployment. Mean difference of ubiquitous broa dband penetration between high, medium, and low income countries was statistically significant (F -statistic: 69.80, P <.01). Higher income countries tend to have higher ubiquitous broadb and penetration. This result may suggest that there is digital divide in the ubiquitous broadband access between countries by income. Mean difference of ubiquitous broadband pene tration between regions (Africa, America, Asia, Europe, and Oceania) was also statisticall y significant (F-statistic: 5.87, P <.01). European and Asian countries tend to have higher ubiquitous broadband penetration than other countries in Africa, America, and Oceania. This result suggests that there is digital di vide between countries by region. Mean difference of ubiquitous broadband penetration between countri es with multi-play strategies such as triple-play and quadruple-play offerings without multi-play strategies was also significant (see table 5-14). This result suggests th at multi-play strategy such as triple-play and quadruple-play offerings is a contributing fact or of global ubiquitous broadband deployment including fixed and mobile broadband. Also, me an difference of broadband penetration between countries with LLU policy and without LLU polic y was also significant (see table 5-14). This suggests that LLU policy is a contributing factor of global ubiquitous broadband deployment.

PAGE 112

112Table 5-14. Difference in total (ubiquitous ) broadband penetration and total broadband penetration rate by income, region and multiple play offerings Dependent Variable Source SS df MS F Variable Mean Total Broadband Penetration Income 9917.0524958.52 69.80***High Income 15.12 Error 15130.5921371.03 Medium Income 1.52 Low Income .22 Region 2509.464627.36 5.87*** Africa .88 Error 22538.19211106.81 America Asia Europe Oceania 3.63 8.17 10.53 7.89 Total Broadband Penetration Triple-Play 1192.7511192.75 14.20**Triple-play 21.67 Error 2351.612883.99 No Triple-play 6.76 Quad-play 517.881517.88 4.791*Quad-play 24.53 Error 3026.4728108.09 No Quad-play 15.47 Total Broadband Penetration LLU 2441.2212441.22 17.40***LLU 537.05 Error 11362.2181140.27 No LLU 914.11 Note. Categorization of countries by inco me and region are based on the ITUs categor ization of countries by income and region. Statistically significant at the 10% level. ** Statistically significant at the 5% level. ***Statistically significa nt at the 1% level

PAGE 113

113 CHAPTER 6 DISCUSSION AND CONCLUSION This study exam ines adoption factors of fixed, mobile, and ubiquitous broadband and helps explains why there are differences in the diffusion of broadband communication technology between countries. This chapter summarizes the empirical results and analysis. Implications of this study and limitations and suggestions for future re search are also discussed in this chapter. Summary of Results and Analysis Effects of Policy Factors on Broadband Deployment This study examined several policy factors of broadband diffusion. For fixed-broadband, the effects of interactions of different types of LLU policies and LLU price regulation on fixedbroadband deployment were examined. For mobile broadband, the effects of market-mediated standard policy and the influence of diffe rent 3G licensing assignment types on mobile broadband diffusion were tested For ubiquitous broadband, the effect of LLU policy (LLU dummy variable) and influences of different types of platform competition-standardization policies on the diffusion of ubiquitous broadba nd were examined. Also, for all broadband technologies, the effects of in stitutional environment such as political and economic freedom were tested. Effects of LLU policy on broadband deployment. The results of nonlinear regression analysis suggests the effects of interactions of different types of LLU and LLU price regulation were very significant factors of fixed broadband diffusion in OE CD countries (see Table 6-1). Most OECD countries adopted two types of LLU. The first type of LLU in OECD countries is implementation of full unbundling, line sharing, and bit stream access all together. Some OECD countries adopted full unbundling and line sharing only without bit stream access (OECD, 2003). Also, many OECD countries have LLU price regulation such as regulatory approval for line

PAGE 114

114 rental charges (OECD, 2003). The result of nonl inear regression study suggests that LLU policy type I (full unbundling, line sharing, and bit st ream access without LLU price regulation for line rental charges), LLU policy type II (full unbundli ng, line sharing, and no bit stream access with LLU price regulation for line re ntal charges), and LLU policy type III (full unbundling, line sharing, and bit stream access with LLU price regul ation for line rental charges) were significant factors of fixed broadband diffusion in OECD countries. The result of one-way ANOVA also suggests LLU policy has been a key driver of fixed broadband diffusion in many ITU membership countries (see Table 6-1). Also, th e result of one-way ANOVA suggests ubiquitous broadband penetration of countries with LLU policy is higher than that of countries without LLU policy (see Table 6-1). Consider ing fixed broadband is more popular in most of countries and competition effect of LLU policy might influence other platforms (e.g. cable) market behavior for competition, it is expected result. Based on these results, in general, LLU policy can be an effective policy tool for improving broadband diffusion.1 Considering that DSL is the major source of residential broadband delivery in most countries, intra-m odal competition in the DSL market through LLU policy might have contributed to a greater a doption of fixed-broadband (Lee & Brown, 2008). LLU policy might simulate the competitive e ffect by opening up an incumbent network for competitive access. Effective intramodal competition through LLU policy may bring real choice for customers and reduce DSL prices (Lee, 2006; DotEcon & Crite rion Economics, 2003). However, implementation of LLU policy and LLU price widely differs among countries (OECD, 2003). In spite of significance LLU policies effect on broadband deployment, strong LLU 1 These results are consistent with the results of other previous studies by Ridder (2007), Wallsten (2006), and Grosso (2006).

PAGE 115

115 regulation may confiscate incumbents property a nd reduce their investme nt incentives in new telecommunication technologies (Frieden, 2005a). Considering these costs and benefits of LLU policy, more refined LLU policy should be recommended. Effects of other policy factors on mob ile and ubiquitous broadband deployment. The log linear regression results of mobile broadband penetration for all countries suggest that a market-based multiple standards policy significan tly contributes to the diffusion of mobile broadband services (see Table 6-1) This finding is consistent with the findings by Cabral and Kretschmer (2004), which assert, as mobile t echnology becomes more ma ture, standardization and its scale and efficiency benefits seem to become less relevant. This finding might also suggest that the importance of marketmediat ed multiple standards when a new technology evolves into a different stage of development characterized by more advanced, differentiable features (Lee et al., 2007b). Also, the log linear regression results of ubiqu itous broadband penetration for all countries suggest that both the interacti ons of DSL-cable platform comp etition availability in fixedbroadband markets and single and multiple standard ization policy significantly contributes to the diffusion of ubiquitous broadband pe netration (see Table 6-1). However, can these results of log linear regression of mobile broadband penetration and ubiquitous broadband penetration for all countries be applied to both developed and de veloping countries? Unfort unately, since mobile broadband is a very new medium for most de veloping countries, only a small number of observations for mobile broadband penetration were available. In spite of the small number of observations for the mobile broadband penetra tion, the result of log linear regressions of ubiquitous broadband penetration for developed countries and developi ng countries gives us more ideas about this policy i ssue. The log linear regression results of ubiquitous broadband

PAGE 116

116 penetration for developed countries suggest that th e interaction vari able (Policy type II) of DSLcable platform competition availability in fixe d-broadband markets and single standardization policy in mobile broadband markets was not stat istically significant (see Table 6-1). However, the interaction variable (Policy type III) of DSL-cable platform competition availability in fixedbroadband markets and multiple standardization policy in mobile broadband markets was statistically significant (see Table 6-1). In the deve loping countries, the situa tion seems different. In the developing countries both interaction variables (Policy type II and Policy type III) were statistically significant (see Tabl e 6-1). Moreover, in developing countries, considering the Pvalue and t-statistic for Policy type II and Policy type III variables, Policy type II variable (with DSL-cable platform competition in fixed-broadband markets and single standardization policy in mobile broadband markets) was more significantly associated with the diffusion of ubiquitous broadband (see Table 6-1). Considering that in mo st of developed countries currently DSL-cable modem platform competition is available, the re sults of these empirical studies for mobile broadband diffusion and ubiquitous br oadband diffusion at least suggest that in the initial status of mobile broadband deployment, market-mediated standardization policy is more effective for developed (high income) countries, which basic ICT infrastructure is already deployed and might have more mature mobile industry and consumers.

PAGE 117

117 Table 6-1. Significant policy fact ors of broadband deployment* Model/Policy Factors LLU LLU LLU LLU Multiple P-S P-S P-S Political EconomicType of Policy I** Policy II Policy III DummyStandards Policy I*** Policy II Policy III FreedomFreedom Licensing Fixed-Broadband Nonlinear(N=217) Yes Yes Yes No No OECD Countries (p<.01) (p<.01) (p<.01) Fixed-Broadband Linear(N=282) No No All Countries Fixed-Broadband Linear(N=132) No No Developed Countries Fixed-Broadband Linear(N=148) No No Developing Countries Fixed-Broadband One-way ANOVA Yes (N=158) All Countries (p<.01) Mobile Broadband Linear (N=105) Yes No No All Countries (p<.01) Mobile Broadband

PAGE 118

118 Table 6-1 Continued, One-way ANOVA No (N=65) All Countries Total Broadband Linear (N=191) Yes Yes Yes Yes No All Countries (p<.1) (p<.1) (p<.05) (p<.01) Total Broadband Linear (N=73) Yes Yes Yes Yes No Developed Countries (p<.1) (p<.1) (p<.01) (p<.05) Total Broadband Linear (N=120) Yes Yes Yes No No Developing Countries (p<.1) (p<. 01) (p<.05) Total Broadband One-way ANOVA (N=82) Yes (p<.01) All Countries *For the significant variables, variables which are statistically significant at the .1 level were included in this table. This result is based on the final reduced model. ** LLU Policy Type I: Full unbundling + line sharing + bit stream access + No LLU price regulation LLU Policy Type II: Full unbundling + line shar ing + no bit stream access + LLU price regulation LLU Policy Type III: Full unbundling + line shar ing + bit stream access + LLU price regulation ***P-S Policy Type I: Cable modemDSL platform competition + no standard ization policy for mobile broadband P-S Policy Type II: Cable mode m-DSL platform competition + single standardization policy for mobile broadband P-S Policy Type III: Cable modem-DSL platform co mpetition + multiple standardization policy for mobile broadband

PAGE 119

119 It appears that, in developed countries, technological diversity, in a new mediums early stage, is likely to foster innova tive applications and be tter consumer choices, which initially lead to faster deployment of technology like mobile broadband. This study also found high level of political freedom, which is measured by the level of civil liberty in a country, is associated with th e high level of diffusion of ubiquitous broadband. This result suggests, in the diffu sion of 4G mobile technology, political freedom might be an influential factor. Also, the result of one-way ANOVA of mob ile broadband diffusion suggests 3G mobile licensing assignment method is not a significant factor of mo bile broadband diffusion (see Table 6-1). Effects of Industry Factors on Broadband Deployment Platform/network competition is a significant factor of broadband deployment. The result of log linear regression of fi xed-broadband diffusion suggests the higher level of platform competition measured by HHI (Herfinall-Hirschman I ndex) is associated with the higher level of fixed broadband penetration. Though coefficient B of platform competition shows its relationship with fixed-broadband diffusion, it was not a signifi cant factor of fixed-broadband diffusion in developed countries (s ee Table 6-2) in the log linear m odel. This result is consistent with the result of the nonlin ear model of fixed broadband diffusion for OECD countries. Considering OECD countries consist of 30 devel oped countries, this result appears robust (see Table 6-2). However, in developing countries, platform competition was an influential factor of fixed-broadband diffusion. With these results, considering most of developed countries have higher level of fixed-broadband penetration, it ap pears platform competition is mainly effective for countries in the initial diffusion stage of fixed-broadband deployment.

PAGE 120

120 Also, for ubiquitous broadband, network competition between fixed and mobile broadband, which is measured by a dummy variable, was a significant factor of ubiquitous broadband deployment (see Table 6-2). In the log linear model of fixed-broadba nd deployment, this study also found fixed broadband price, which is measured by lower speed monthly charge (USD), was negatively associated with the higher level of fixed-broa dband diffusion. As Ridder (2007) suggested, many previous empirical studies on fixed broadband c ouldnt find this negativ e association between fixed-broadband price and fixed-broadband deployment.2 Considering the relationship between price and normal goods, negative associati on between fixed-broa dband price and fixedbroadband deployment is an expected result. Inte restingly, the result of log linear model of ubiquitous broadband also provide s that negative correlations of price and ubiquitous broadband deployment (see Table 6-2). Considering fixe d-broadband technology is a dominant technology in most of countries for broadband acce ss, this is a plausible result. The result of log linear re gression analysis also suggests download speed, which is measured by the Kbit/s, is a significant factor of ubiquitous-broadband deployment (see Table 62). The result suggests that fast download speeds could lead to more broadband subscribers in the market. This result also may imply consumers who want fast broadband speed will more readily migrate to costly broadband services, if ther e are higher levels of throughput speed offered by broadband service providers (Lee & Brown, 2008). The log linear regression of fixed-broadba nd deployment also suggests that bandwidth, which is measured by bits per inhabitants, is a driver of fixed broadband deployment. 2 Garcia-Murillo (2005) found association of fixed broa dband price, which is measured by monthly price per megabite, with the number of fixed broadband subscribers. However, in the study, fixed broadband price was positively correlated to the number of subscribers.

PAGE 121

121 Considering bandwidth may determine the quantity of information transmitted and the offering of diverse broadband applications, the positive association between bandwidth and fixed broadband deployment is an expected result. The effect of mobile price on fixed-broadband de ployment is very interesting. In the linear regression analysis for all ITU membership countries, mobile price is not a significant factor of fixed-broadband diffusion. Also, the result of linear regression analysis of mobile broadband suggests fixed-broadband price is not a significant factor of mobile-broadband diffusion. Based on these results, in the linear regression model for all countries, mobile service (and mobilebroadband service) is not a complement or subs titute to fixed-broadba nd services and fixedbroadband service is not a complement or subst itute to mobile-broadband service. However, surprisingly, in developed countri es, mobile price is negatively associated with the higher level of fixed broadband, but, in developing countries, mobile price is positively associated with the higher level of fixed broadband. This result might suggest that in deve loped countries mobile service is a complement to fixed-broadband services, but in developing countries mobile service is a substitute to fixed broadband. Considering consumers in low-income countries tend to have limited budget for media and telecommunication services, this result is possible. This finding may also suggest that when mobile broadband services become popular in most countries, this trend could be continue d for both developed c ountries and developing countries. If the data for mobile broadband servic es in developing countries become sufficiently available, this relationship should be tested in the future research. The result of one-way ANOVA for fixed a nd mobile broadband suggests triple-play strategy have contributed higher level of fixed broadband penetra tion growth rate, but quadrupleplay strategy have not significantly influenced the growth rate of mobile penetration yet. This

PAGE 122

122 Table 6-2. Significant industry f actors of broadband deployment* Model /Industry Factors Platform Fixed Broadband Speed Bandwidth Telecom Mobile Mobile Application Cost Network Competition TripleStrategy Quadruple -Strategy Competition Price Investment Price Fixed-Broadband Nonlinear(N=217) No No No OECD Countries Fixed-Broadband Linear(N=282) Yes Yes No Yes No No All Countries (p<.05) (p<.1) (p<.01) Fixed-Broadband Linear(N=132) No No No No No Yes Developed Countries (p<.1) Fixed-Broadband Linear(N=148) Yes Yes No Yes No Yes Developing Countries (p<.1) (p<.1) (p<.01) (p<.1) Fixed-Broadband One-way ANOVA Yes All Countries(N=158) (p<.01) Mobile Broadband Linear (N=105) No No No No No All Countries Mobile Broadband One-way ANOVA No All Countries(N=65) Total Broadband Linear (N=191) Yes Yes No No No Yes All Countries (p<.05) (p<.05) (p<.05) Total Broadband

PAGE 123

123Linear (N=73) No Yes Yes No No Developed Countries (p<.05) (p<.01) Total Broadband Linear (N=120) No No Yes No No Developing Countries (p<.01) Total Broadband One-way ANOVA Yes Yes All Countries(N=29) (p<.05) (p<.1) *For the significant variables, variables which are statistically significant at the .1 level were included in this table. This result is based on the final reduced model.

PAGE 124

124 result may suggest that in the initial multiple pl ay markets the quadruple-play strategy is not as successful as triple-play strategy.3 The result of one-way ANOVA for ubiquitous broadband also suggests quadruple-play strategy have not significantly influe nced the ubiquitous broadband penetration yet as triple-play strategy.4 Effects of Demographic/ICT Factors on Broadband Deployment The results of both linear regression anal ysis of mobile broadband and ubiquitous broadband suggest higher level of income measured by GDP per capita is associated with higher levels of mobile and ubiquitous broadband deployment (see Table 6-3).5 These results suggest consumers with higher incomes are more likel y to purchase broadband services. Broadband service providers in hi gh-income markets may consider pos sible segmentation strategies by income. Also, higher levels of education are a ssociated with higher le vels of fixed-broadband deployment in both nonlinear and linear regression models. As Rodgers (2003) suggests, many early adopters tend to have highe r socio-economic status (e.g. high level of education). High level of population density and urbanization are consid ered as supply factors for broadband diffusion (Ridder, 2007) The results of nonlinear and linear regression analysis of fixed broadband suggests higher leve ls of population density are asso ciated with higher levels of fixed-broadband deployment and linear regr ession analysis of ubiquitous broadband shows population density is also a signi ficant factor of ubiquitous br oadband deployment. This result 3 One of the reasons might be the high switching cost for mobile services. To switch over mobile providers, the consumer should pay high cancellation fee. 4 Table 5-14 provides this result. The result of one-way ANOVA suggests that the effects of triple-play strategy on the diffusion of ubiquitous broadband are statistically significant at the .05 level, while the effects of quadruple-play strategy on the diffusion of ubiquitous broadband are statistically significant at the .1 level only. 5 This result is consistent with the result by Madden et al. (2004) and Andonova (2006), which suggested income is a significant factor of mobile diffusion. Also, some previous studies on fixed broadband suggested income is an influential factor of broadband diffusion (see Wallste n (2006); Grosso (2006); Garcia-Murillo (2005)).

PAGE 125

125 may imply more densely populated country have a dvantages in the cost conditions for network deployment. Also, higher levels of urban population share are associated with higher levels of ubiquitous broadband deployment, which may imply that more urbanized countries have better cost conditions for broadband deployment. The result of nonlinear regression analysis suggests that higher levels of PC penetration are associated with the higher levels of fixed broa dband penetration in OECD countries (see Table 63). This result might suggest that an already well-established IC T infrastructure may lead to more rapid fixed-broadband deployment. Also the result of linear regression for all ITU membership countries suggests high levels of Internet usage in a country is correlated with high levels of fixed broadband diffusion (see Table 6-3). As Frieden (2005) contends, these results of ICT factors might suggest that IC T incubation, which is supporte d by ICT infrastructure or ICT use such as PC infrastructure and Internet us e, may be key drivers of broadband deployment. In addition, Internet content, as measur ed by the number of Internet hosts per 10000 inhabitants, was a significant f actor for fixed broadband penetration in OECD countries (see Table 6-3). This result implies that the amount of compelling content, services and applications within a nation is an important factor of fixed-broadband diffusion (Lee & Brown, 2008). This study also examined the causal relations hip between 1G and 2G mobile penetration and mobile broadband penetration. Considerin g the negative relationship between these two variables in the final reduced model of mobile broadband diffusion, mobile broadband can be considered a substitute of 1G and 2G mobile services. While this result of inter-generation effects on mobile diffusion is different from th e previous study by Liikne n et al. (2004), which found that 1G has positive effect on 2G, it is nevertheless an expected result because many

PAGE 126

126 mobile users prefer 3G mobile se rvices that offer more diverse m obile applications than 1G or 2G mobile services. Also, the result of linear regression analys is of fixed-broadband diffusion suggests previous fixed-broadband penetr ation is positively associated with current fixed-broadband penetration, which may imply existence of ne twork effects. As Econo mides and Himmelberg (1995) suggest, if there are ne twork effects, current subscription of new media is positively correlated with previous subscrip tion of new media. Based on th e result of log linear regression study of fixed-broadband deployment, this st udy at least suggests that between 2002 and 2005, new fixed-broadband subscribers joined broadband network influenced the utility of current fixed-broadband subscribers.6 Digital Divide and Broadband Deployment in Developed and Developing Countries The result of one-way ANOVA for fixed, mobile, and ubiquitous broadband diffusion, suggests there are significant diffe rences in the diffusion of br oadband by income (between high, medium, and low income countries) and by region (between Africa, America, Asia, Europe, and Oceania). This result is consistent with the resu lt of previous by Chinn and Fairlie (2006), which found that the global digital divide is mainly explained by income differentials. This study also examined common and diff erent factors of broadband deployment in developing and developed countries.7 Table 6-4 provides summary of the results for developing and developed countries. For fixed-broadband deployment, common significant factors of broadband diffusion for both developing and devel oping countries are fixedbroadband price and 6 If more broadband penetration data, which covers more periods, are available, using different nonlinear model, more refined research to test the impact of network effect on broadband deployment will be possible. 7 Because of very small number of observations of mobile broadband for developing countries, this study examined only fixed and ubiquitous broadband deployment for developing and developed countries.

PAGE 127

127Table 6-3. Significant demographic/IC T factors of broadband deployment* Model /Factors Income Education Urban Population Population Density Age (35-44) PC Penetration Content Internet Usage Teledensity 1G/2G Penetration Previous Penetration Fixed-Broadband Nonlinear(N=217) No Yes Yes Yes Yes OECD Countries (p<.05) (p<.05) (p<..01) (p<.05) Fixed-Broadband Linear(N=282) No No No No No NoNo Yes No Yes All Countries (p<.01) (p<.01) Fixed-Broadband Linear(N=132) Yes Yes No Yes NoNo Yes No No Yes Developed Countries (p<.05) (p<.01) (p<.1) (p<.05) (p<.01) Fixed-Broadband Linear(N=148) No No No No No NoNo Yes No Yes Developing Countries (p<.05) (p<.01) Fixed-Broadband One-way ANOVA Yes All Countries(N=158) (p<.01) Mobile Broadband Linear (N=105) No No No No No Yes All Countries (p<.05) Mobile Broadband One-way ANOVA Yes All Countries(N=65) (p<.1) Total Broadband Linear (N=191) Yes No YesYes No No No No No All Countries (p<.01) (p<.1) (p<.05) Total Broadband Linear (N=73) No No No Yes No No No Yes No

PAGE 128

128Developed Countries (p<.05) (p<.01) Total Broadband Linear (N=120) Yes No No No No No NoNo No Developing Countries (p<.01) Total Broadband One-way ANOVA Yes All Countries(N=82) (p<..01) *For the significant variables, variables which are statistically significant at the .1 level were included in this table. This result is based on the final reduced model.

PAGE 129

129 previous fixed-broadband penetration. For both developing and deve loped countries, fixedbroadband price is negatively associated with fixed-broadband deployment and previous fixedbroadband penetration is positively associated with current fixed-broadband penetration (see Table 6-4). This result implies lower pricing and network effects contributed to more rapid fixedbroadband diffusion in both developing and deve loped countries. Mobile price was positively associated with fixed-broadband diffusion in de veloping countries, but negatively associated with fixed-broadband diffusion in developed coun tries. This result implies that in developing countries mobile service is a substitute for fixe d broadband service, but in developed countries it is a complement for fixed broadband service. Table 6-4 also suggests that Internet use, bandwidth and platform competition are significan t factors of fixed-broadband diffusion only for developing countries and income, education, po pulation density, and content are significant factors only for developed count ries. This result implies for fixed-broadband diffusion, ICT infrastructure (Internet use), industry competitio n (HHI), and technological factors such as bandwidth are more important va riables in developing countries. For ubiquitous-broadband deployment, common significant factors of broadband diffusion for both developing and developi ng countries are bandwidth a nd Policy type III (A country which platform competition between cable mode m and DSL platform is available and has multiple standardization policy for mobile broadband).8 Table 6-4 also sugge sts that income is a significant factor of ubiquitous broadband diffusion only for developing countries and speed, political freedom, population density, and Internet use are significant factors only for developed countries. Policy type II (A country which pl atform competition between cable modem and DSL platform is available and has single standardizat ion policy for mobile broadband) is a significant 8 Policy type I was a significant factor for both developing and developed countries at the .1 level.

PAGE 130

130 factor for developing countries, but significant only at the .1 leve l for developed countries. This result may imply the impacts of a single standard and its benefits such as network externalities for mobile broadband may be larger in developing countries than in developed countries. Table 6-4. Common and Diffe rent Significant Factors of Broadband Deployment** Broadband Technolog y Common Significant Factors Significant Factors: Significant Factors: Developing Countries Only Developed Countries Only Fixed broadband price Mobile price (+)* Mobile price (-)* Previous broadband penetration Internet use Income Fixed Broadband Bandwidth Education Platform competition* Population density Content Bandwidth Income Speed Policy Type III (p <.05) Policy Type I* (p < .1) Policy Type II (p < .01) Political freedom Ubiquitous Broadband Population density Internet use Policy Type II* (p <.1) **For the significant variables, variables which are statistically significant at the .05 level were included in this table. This result is based on the final reduced model. Variable with is significant at the level .1 ***Policy Type I: Cable mode m-DSL platform competition + no standardization policy for mobile broadband Policy Type II: Cable modem-DSL platfo rm competition + single standardization policy for mobile broadband Policy Type III: Cable modem-DSL plat form competition + multiple standardization policy for mobile broadband Implications Theoretical Implications One of the main goals of this study is to examine the effects of platform (or standard/network competition) on broadband depl oyment. Platform competition occurs when different technologies (platforms) compet e to provide similar or differentiated

PAGE 131

131 telecommunication services to end-users (Chur ch & Gandal, 2005). Platform competition in network industry involves competition between t echnologies that are not only differentiated, but also involve competing networks (Church & Gandal, 2005). Using ITU membership countries data, this study tested the impacts of platform competition between cable modem, DSL and other platforms on fixed-broadband penetration. The result provides competition between different fixed-broadband platforms is an influential factor of fixed-broadband deployment in ITU membership countries.9 Interestingly, the result of nonlin ear regressions of fixed-broadband penetration suggest high levels of platform competition are related to high levels of fixedbroadband penetration, but the effects of platform competition are not statistically significant in OECD countries. This result is consistent with the result of linear regression analysis of developed countries (high income ITU membership countries). C onsidering OECD countries are composed of 30 developed countries with co mparatively high GDP per capita, it seems this result is robust. Considering all results of statistical analysis of fixed-broadband and previous empirical studies on fixed broadband deployment, it appears the effects of platform competition are strong in the initial deployment (e.g a developing country with low level of fixed broadband penetration) of fixed-broadband, but the effects of platform competition are decreasing when the broadband market size is sufficiently large or broadband market is mature.10 Strong platform competition among different technologies may lead to lower prices, increased feature offerings, and more extensive broadband networks (ITU, 2003a), but, it seems, afte r the initial deployment of fixed-broadband, these effects of platform competition are decreasing. In the future, with 9 It appears this result was similar in developing countries (see Table 5-8). 10 In some OECD countries such as Korea and Japan, the diffusion pattern of fixed broadband already shows s-shape curve, as Rodgers (2003) suggested in the diffusion of innovations theoretical framework.

PAGE 132

132 larger numbers of data and observation periods, the effects of platform competition should be continuously examined.11 In spite of this observation of the effects of platform competition on fixed-broadband diffusion, the results of log linear regression analysis of ubiquitous broadband and mobile broadband suggest theoretical framework of platfo rm competition are stil l useful in explaining the deployment of broadband. The result of da ta analysis of ubiquitous broadband suggests network competition between mobile and fixed broadband is a significant factor of ubiquitous broadband deployment. Also, this study also sugg ests competition between different mobile standards (e.g. competition between W-CDMA and CDMA 2000) in mo bile industry is a significant factor of mobile broa dband diffusion. Considering the st atus of mobile and ubiquitous broadband deployment is still in early stage in mo st of countries, these re sults support the result of fixed broadband deployment research, which the effects of platform (or network competition) are strong in the initial deploym ent of broadband deployment. With greater numbers of data and observation periods, the results of this study should be further a ssessed as mobile and ubiquitous broadband continues to grow.12 This study also examined whether network effects are involved in the diffusion of broadband. For the test of networ k effect, a long period of observations with sufficient number of data is necessary. For the nonlinear model of fixed broadband, this study employs Gruber and Verboven (2001)s model. Since nonlinear model of fixedbroadband di ffusion already assume network externality, this study te sted whether previous subscrip tion of fixed-broadband is a 11 H ffler (2007) found negative side of platform/network competition. He suggested that comparing additional social surplus attributable to cable competition with the cable investments, without significant positive externality, infrastructure competition has probably not been welfare enhancing (H ffler, 2007). 12 However, considering very initial deployment status of mobile broadband in most of countries, this new hypothesis can be tested after significa nt time period is passed.

PAGE 133

133 significant factor of current subscription in the log linear regression model of fixed broadband.13 As expected, previous fixedbroadband penetrat ion was an influential factor of current fixedbroadband deployment in all ITU membership countries, whether characterized as developed or developing countries. Considering impacts of platform competition in ITU membership countries, it appears currently in many countries that ne twork effects and the effects of platform competition co-exist. The result of log linear re gression analysis suggests, for fixed broadband, new subscribers joining a broadband network might influence the ut ility of current subscribers. This network effect in fixed-broadband markets might beco me significant after a certain broadband subscription percentage has achieved critical mass. This study has not examined critical mass point for fixed-br oadband deployment; however, netw ork effect and critical mass point may be captured in fixed-broadband deployment pattern in some countries like Korea, Japan and UK (Lee & Marcu, 2007). Research shoul d continue to examine network effects and the effects of platform competition, thereby capturing how broadband diffusion patterns change over time. Path dependence theory, which refers to the dependence of a system or network on past decisions of producers and consumers, is also related to mobile broadband deployment. The result of log linear regression analysis of mobile broadband suggests that market-mediated standard policy is a significan t factor of mobile broadband diffu sion. It appears that marketmediated multiple standards are important when a new technology evolves into a different stage of development characterized by more advanced, differentiable features (Lee et al., 2007). The result of the regression study of mobile broadban d suggests that, in spite of EUs success story of mandated standard (GSM) in initial stage of 2G, the 3G standard policy that may have 13 Because of short period of observa tions, this hypothesis was not tested for mobile and ubiquitous broadband.

PAGE 134

134 assumed WCDMA or locked operators into WC DMA might be a very costly public policy decision (Gandal et al., 2003). However, considering the positive effects of platform competition on broadband diffusion decrease as the market matures, the applicability of path dependence theory is limited and the effect of EUs ma ndated single standard policy on mobile broadband diffusion is not clear in the long term. Also, the result of this empirica l study suggests that an estab lished infrastructure (e.g. PC penetration, ICT use, previous fixed broadband penetration) for relevant information and communication technologies is an influential factor for fixed-broadband diffusion (Lee et al., 2007). This result may imply the phenomenon of leapfrogging in developing countries cannot be easily applied to the diffusion of fixed-broadband. More refined st udies about the ap plicability of leapfrogging theory to broadband diffusi on are necessary in the future. Policy Implications This study examined the effects of LLU polic y on fixed-broadband diffusion. There have been a lot of debates on the eff ects of LLU policy, but the type of LLU policy and LLU price are very different across countries. The result of nonlinear regressi on study suggests that LLU policy type I (full unbundling, line sharing, and bit st ream access without LLU price regulation for line rental charges), LLU policy type II (full unbundli ng, line sharing, and no bit stream access with LLU price regulation for line re ntal charges), and LLU policy type III (full unbundling, line sharing, and bit stream access with LLU price re gulation for line rental charges) were all significant explanatory variables of fixed broadband diffusion in OECD countries. Apparently it seems this result supports the effectiveness of LLU on fixed broadband in many countries. Effective LLU policy may generate consumer bene fits in the near future through open access to competitors (Frieden, 2005a). Considering the re sult of this study, countries fostering broadband deployment need consider adopting LLU policy for the fixed-broadband market. However, LLU

PAGE 135

135 might reduce incumbents incentives to invest in new telecommunication technologies (Frieden, 2005a). Considering these costs and benefits of LLU policy, it ma y be better if countries might pursue light-touch regulation such as line-shar ing and/or bit stream access instead of full unbundling at a reasonable LLU price. A previous study suggests, the uptake of these light forms of LLU has been relatively succe ssful (de Bijl & Peitz, 2005). This study also found significant effects of platform competition on fixed-broadband diffusion in the initial deploymen t of fixed broadband and initial success of market-mediated standard policy in mobile broadband markets. Also, this study found positive effects of network competition and interactions of platform competition and multiple standards policy on ubiquitous broadband deployment. This result implies, at le ast in the initial broa dband markets, regulation across different platforms should be as competiti vely neutral as possible for sustaining strong platform competition. Considering positive effects of network externality and the possibility of decreasing effects of platform/network/standar d competition on broadband diffusion in the long term, it is still important to note that concepts of efficiency and ease of integration are critical for future broadband markets. These discussions on policy factors of broadband deployment have policy implications for the diffusion of 4G mobile tec hnologies (or pre 4G mob ile technologies). The result of this study implies in the initial 4G mobile markets, governments need to be open to diverse standards for competition instead of government-mandated sta ndards. In the long term, as Noam (2003) suggested, industry-wide coordi nation and mutual learning processes is important. In this cooperative voluntary standard set ting situation, compatibility is achieved by agreeing to a standard (Church & Gandal, 2005). In doing so, in the long term, broadband service providers may suppress competition between networks in favor of competition on a network.

PAGE 136

136 Also, this study has different policy implica tions for developed and developing countries. The results of fixed broadband deployment study suggest that mobile service could be a substitute for fixed broadband se rvice in developing countries, but it could be a complement for fixed broadband service in developed countries. This result may imply, in the long term, when mobile broadband services are mature in many countries, deployment of fixed (mobile) broadband might positively influenced mobile (fixed) broadband in developed countries. Considering leapfrogging theory cannot easily applied to broadband deployment in the developing countries, this result of statistical analysis for de veloping countries may suggest without sufficient previo us ICT experiences and better economic status14, it is not easy to deploy fixed and mobile broadband at the same time. Also, this study suggests, for the ubiquitous broadband deployment, policy type II (platform competition in fixed broadband plus single standard for mobile broadband) or policy type III (platform competition in fixed broadband plus multiple standards for mobile broadband) would be recommendable in the developing countries, but, only policy type III would be re commendable for developed countries.15 Limitations and Suggestions for Future Research This study has some limitations. For the fixed-broadband diffusion model, because of data availability, more diverse independent variable s cannot be included for the nonlinear regression model. Also, since the nonlinear model already assumes network externality, the impacts of network effects on fixed broadband could not be tested in the model. When more data and observations over a longer period are available, and, with different nonlinear model such as Gompaz model, more refined analysis on the effects of platform competition and network effects 14 It appears that insufficient income and limited budget for mobile and fixed broadband services are reasons of substitute relationship between mobile service and fixe d broadband services in developing countries. 15 In spite of this general policy recommendation, each countrys situation and environment such as size of a country should be considered for the best policy choice.

PAGE 137

137 will be possible. Also, for the analysis of the e ffects of LLU policy, if diverse data about the effects of line sharing and bit stream access are available with sufficient observations, more refined comparison of the effects of different type of LLU policies are possible. For the mobile broadband diffusion model, data for developing countries were not sufficient to analyze the difference between deve loping countries and deve loped countries. Also, with a small number of observations over comp aratively short period, nonlinear nature of broadband diffusion was not easily captured fo r mobile and ubiquitous broadband diffusion. When more observations over a longer time period are available, more refined analysis will be possible. Also, if different multiple measurements for ubiquitous broadband deployment and data are available, more improved research for th e diffusion of ubiquitous broadband deployment, which has greater implications for 4G mobile deployment, will be available. Finally, this study could not in clude any impacts of sociocultural variables and other policy variables on broadband deployment in a count ry. For instance, a countrys culture such as life style and policy factors such as crossownership regulation of media and spectrum availability16 could be another influen tial factors of broadband depl oyment. When more refined measurement and data for these variables are avai lable, these variables can be included in the empirical model. 16 For instance, the scarcity of frequency spectrum may influence the performance of the mobile telecommunications industry (Gruber, 2001c). Efficient use of spectrum and optimal licensing fee could be important for the deployment of 3G mobile and mobile price.

PAGE 138

138 LIST OF REFERENCES Ahn, H. & Lee, M. (1999). An econometric analysis of the demand for access to mobile telephone networks. Information Economics and Policy, 11, 297-305. Aizu, I. (2002). A comparative study of broadb and in Asia: Development and policy. Paper presented at RIETI Sympos ium, Tokyo. Andonova, V. (2006). Mobile phones: the Inte rnet and the institutional environment. Telecommunications Policy, 30, 29-45. Aron, D. J. & Burnstein, D. E (2003). Broadba nd adoption in the United States: An empirical analysis. In A.L. Shampine (Ed.), Down to the wire: Studi es in the diffusion and regulation of telecommunications technologies (pp.119 -138). Haupaugge, NY: Nova Science Publishers. Atkinson, R.D., Correa, D.K. & Hedlund, J.A. (2008). Explaining inte rnational broadband leadership. Washington, DC: The Informa tion Technology and Innovation Foundation. Cabral, L. M. B. & Kretschmer, T. (2007). Standa rds battle and public polic y. In S. Greenstein & V. Stango (Eds.), Standards and public policy (pp.329 -344). Cambridge, MA: Cambridge University Press. Cava-Ferreruela, I. & AlabauMu oz, A. (2006). Broadband policy assessment: A cross-national empirical analysis. Telecommunications Policy, 30, 445-463. Church, J. & Gandal, N. (2005). Platform comp etition in telecommunications. In M., Cave, S., Majumdar & I. Vogelsang (Eds.), The handbook of telecommunications vol. 2.: Technology evolution and the Internet., Armsterdam, The Netherlands: North-Holland. Clements, M & Abramowitz, A. (2006, Sept ember). The development and adoption of broadband service: A household level analysis. Paper presented 35th Research Conference on Communication, Info rmation and Internet Policy, Arlington, VA. Chang, B., Lee, S. & Kim, B. (2006). Explorin g factors affecting the adoption and continuance of online games among college students in South Korea. New Media and Society 8(2), 295-319. Chaudhuri, A., Flamm, K.S. and Ho rrigan J. (2005), An analysis of the determinants of internet access. Telecommunications Policy, 29(9-10),731-55. Chinn, M.D. & Fairlie, R.W. (2006). The determinants of the global digital divide: A crosscountry analysis of computer and Internet penetration. Oxford Economic Papers, 59, 1644.

PAGE 139

139 Collis, D.J., Bane, P.W., & Bradley, S. P. ( 1997). Winners and losers: Industry structure in converging world of telecommunications, com puting and entertainment. In D.B. Yoffie (Ed.), Competing in the age of digital convergence (pp.159-200). Boston: Harvard Business School Press. Crandall, R. W., Sidak, J.G. & Singer, H.J. (200 2). The empirical case against the regulation of broadband access. Berkeley Technology Law Journal, 17(3), 953-87. Crandall, R. W. (2005). Broadband Communications. in Cave, M., Majumdar, S. and Vogelsang, I. (Eds.), Handbook of telecommunications economic s, Volume 2: Technology evolution and the Internet(pp.156 -191), Amsterdam, Neth erlands: North-Holland. David, P., & Greenstein, S. (1990). The economics of compatibility sta ndards: An introduction of recent research. Economics of Innovation and New Technology, 1, 3-41. Davis, F.D. (1989). Perceived Usefulness, Per ceived Ease of Use, and User Acceptance of Information Technology. MIS Quarterly, 13(3), 319. De Bijl, P.W.J. & Peitz, M. (2005). Local loop unbundling in Europe: Experience, prospects and policy challenges. Communications & Strategies, 57, 33-57. Denni, M. & Gruber, H. (2005). The diffusion of broadband tel ecommunications: The role of competition. Paper presented at International Telecommunication Conference, Pontevedra. Distaso, W., Lupi, P & Maneti, F.M. (2006). Plat form competition and br oadband uptake: theory and Empirical evidence from the European Union. Information Economics and Policy, 18 (1). 87-106. DotEcon & Criterion Economics (2003). Competiti on in broadband provision and its implication for regulatory policy: A report for the Brussels Round Table. London: DotEcon & Criterion Economics. Retrie ved July 7, 2006, from http://www.dotecon.com Dowling, M., Lechner, C., &T hielm an, B. (1998). Convergen ce Innovation and change of market structures between te levision and online services. Electronic Markets, 8(4). Economides, N. & Himmelberg, C. (1995). Critical mass and network size with application to the US fax market. Discussion paper EC-9511, Stern School of Business, New York University. Fernandez-Cornejo, J & McBride, W. D. (2002). Adoption of bioe ngineered crops. US Department of Agriculture Agri cultural Economics Report, 810. Fransman, M. (2006). Introduction. In M. Fransman (Ed.), Global Broadband Battles; Why the U.S. and Europe Lag While Asia Leads (pp. 1-58). Standford University Press, Stanford, CA.

PAGE 140

140 Freedom House (2007). Freedom in the worl d country ranking. Retrieved May 10 from http://www.freedomhouse.org Frieden, R. (2005a). Unbundling th e local loop: A cost/benefit analysis for developing nations. Info, 7(6), 3-15. Frieden, R. (2005b). Lessons fr om broadband development in Canada, Japan, Korea and the United States. Telecommunications Policy, 29(8), 595-613. Friedman T.L. (2006). The world is flat: A brief hist ory of the twenty first century. New York, NY: Farrar, Straus and Giroux. Gandal, N. (2002). Compatibility, Standardization, and network effects: Some policy implications. Oxford Review of Economic Policy, 18 (1), 80-91. Gandal, N., Salant, D. & Waverman, L. (2003) Standards in wirele ss telephone networks. Telecommunications Policy, 27, 325. Garcia-Murillo, M. (2005). In ternational broadband deployment: The impact of unbundling. Communications & Strategies, 57. 83-108. Glassman, J. & Lehr, W. (2001). The economics of Tauzin-Dingell: Th eory and evidence. Retrieved December 3 from http://ebusiness.mit.edu/research/papers Grosso, M. (2006). Determinants of broadband pene tration in OECD nations Paper presented to the Australian Communications Policy and Research Forum. Gruber, H. (2001a). Competition and innovation: Th e diffusion of mobile telecommunications in Central and Eastern Europe. Information Economics and Policy, 13(1), 19-34. Gruber, H. & Verboven F. ( 2001b). The evolution of market s under entry and standards regulation: The case of globa l mobile telecommunications. International Journal of Industrial Organization 19, 1189-1212. Gruber, H. (2001c). Spectrum limits and competition in mobile markets: the role of license fees. Telecommunications Policy, 25, 59-70. GSMA (2007). How to realize the benefits of mobile broadband t oday. Retrieved November 26, 2007 from http://hspa.gsmworld.com Haring, J., Rohlfs, J.H. & Shooshan, H.M. (2002). Propelling the broadband bandwagon. Bethesda, MA; Strategic Policy Institute. Heritage Foundation (2008). 2008 Index of econo mic freedom. Retrieved May 10, 2008 from http://www.heritage.org/research/ features/index/downloads.cfm

PAGE 141

141 Hausman, J. A. (2001). Regulation by TSLRIC: Economic effects on investment and innovation in Sidak, J., Engel, C. and Knieps, G. (Eds.), Competition and Regulation in Telecommunications: Examining Germany and America (pp. 51-68). Boston, MA: Kluwer Academic. Hausman, J. A. (2002). Internet related servi ces: The results of asymmetric regulation, in Crandall, R.W. and Alleman, J.H. (Eds.), Broadband: Should we regulate High-Speed Internet (pp.129-56), AEI-Brookings Joint Center for Regulatory Studies, Washington, DC. Horrigan, J.B. (2005). Broadband a doption at home in the United States: Growing but slowing. Paper presented at 33rd Research Conf erence on Communication, Information and Internet Policy, VA: Arlington. Horrigan, J.B. (2007). Home broadba nd adoption 2007, Retrieved November 26, 2007 from http://www.pewinternet.org International Telecommunication Union (2001). 3G mobile licensing policy: From GSM to IMT2000a comparative analysis. ITU Workshop on Licensing 3G Mobile, September 1921, 2001, Geneva: Internationa l Telecommunication Union. International Telecommunication Union (2003a). Promoting broadband: Background paper for workshop on promoting broadband. Retrieved November 11, 2005 from http://www.itu.int International Telecommunication Union (2003b). Birth of broadband. Geneva, Switzerland: ITU. International Telecommunication Un ion (2003c). Mobile overtakes fi xed: Implications for policy and regulation. Geneva, Switzerland: Author. International Telecommunication Union (2004). The portable Internet. Geneva, Switzerland: Author. International Telecommunication Union (2005). The Internet of Things. Geneva, Switzerland: Author. International Telecommunication Union (2006). Digital.life. Geneva, Switzerland: Author. Jang, S., Dai, S. & Sung, S. (2005). The pattern and externality effect of diffusion of mobile telecommunications: The cas e of OECD and Taiwan. Information Economics and Policy, 17, 133-148. Kang, M.H. (2002). Digital Cabl e: Exploring Factors Associated with Early Adoption. Journal of Media Economics, 14(3), 193. Katz, Michael & Shapiro, C. (1985). Network Ex ternalities, Competition and Compatibility. The American Economic Review, 75 (3), 424-440.

PAGE 142

142 Kauffman, R.J. & Techatassanasoontorn, A.A. ( 2005). International diffusi on of digital mobile technology: A couple-hazard state-based approach. Information Technology and Management, 6, 253-292. Kim, D. (2005). Direction of mobile communica tion in the age of convergence. Retrieved October 13, 2007 from http://www.etri.re.kr Kim, J. H., Bauer, J.M. & Wildman, S.S. ( 2003). Broadband uptake in OECD countries: Policy lessons from comparative statistical analysis. Paper presented at the 31st Research Conference on Communication, Informati on and Internet policy. VA: Arlington. Koski, H. & Kretschmer, T. (2002). Entry, sta ndards and competition: Firm strategies and the diffusion of mobile telephony. ETLA Discussion Papers, 827. Krauss, J. (2006). Market drivers for trip le play. Retrieved March 14, 2008, from http://www.keymile.com Lee, C. & Chan-Olmsted, S.M. (2004).Compe titive advantage of broadband Internet: A comparative study between South Korea and the United States. Telecommunications Policy, 28 (9-10), 649-677. Lee, S. (2006). Broadband deployment in the United States: Examining the impacts of the platform competition. The International Journal on Media Management, 8(4), 173-181. Lee, S. (2006, August). The Impacts of Mark et-based Standardization Policy on Mobile Deployment in OECD Countries: An Em pirical Analysis. Paper presented Association for Education in Journalism and Mass Communication Annual Convention, San Francisco, CA. Lee, S., Chan-Olmsted, S.M. & Kim, H. (2007, August). The Deployment of Third-Generation Mobile Services: A Multinational Analysis of Contributing Factors. Paper presented to the 2007 Association for Education in Journalism and Mass Communication Annual Convention, Washington D.C. Lee, S., & Marcu, I. M. (2007, August). Fixed and mobile broadband deployment: An empirical analysis of adoption factors. Paper presented to the 35th Research Conference on Communication, Informa tion and Internet Policy, Arlington, VA. Lee, S., Brown, J. (2008). Examining Broadba nd Adoption Factors: An Empirical Analysis between Countries. Info: the Journal of Policy, Regulation, and Strategy for Telecommunication, Information, and Media, 10 (1), 2539. Lee, W. (2006, August). Mobile phone diffusion in developed and developing countries: Digital divide, factor, and diffe rence. Paper presented Association for Education in Journalism and Mass Communication Annual Convention, San Francisco, CA.

PAGE 143

143 Lehr, W., Osorio, C., Gillett, S. & Sirbu, M. (2005). Measuring broadb ands economic impact. Paper presented at 33rd Research Conf erence on Communication, Information and Internet Policy, Arlington, VA. Liebowitz, S.J. & Margolis, S. (1990). The fable of the keys. Journal of Law & Economics, 30(1), 1-26. Liebowitz, S.J. & Margolis, S. (1994). Ne twork externality: An uncommon tragedy. Journal of Economic Perspective, 8 (2). 133-150. Liikanen, J., Stoneman, P. & Toivanen, O. (2004). Intergenerational effect s in the diffusion of new technology: The case of mobile phones. International Journal of Industrial Organization, 22 (8/9). 1137-1154. Lin, C.A. (1998). Exploring pers onal computer adoption dynamics. Journal of Broadcasting and Electronic Media, 42 (1). 95-112. Madden, G., Coble-Neal, G. & Dalzell, B. (2004). A dynamic model of mobile telephony subscription incorporating a network effect. Telecommunications Policy, 28, 133-144. Noam, E. M. (2003). The next frontier for openne ss: Wireless communications. In E. M. Noam, & D. Steinbock (Eds.), Competition for the Mobile Internet (pp. 21-37). Norwell, MA: Kluwer Academic Publishers. Oh, S., Ahn, J. & Kim, B. (2003). Adoption of broadband Internet: The ro le of experience in building attitudes. Journal of Information Technology, 18, 267-280. Organization for Economic Co-ope ration and Development (2001). The development of broadband access in OECD countries. Paris: OECD. Organization for Economic Co-ope ration and Development (2003). Development in local loop unbundling. Paris: OECD. Organization for Economic Co-ope ration and Development (2005). Communications outlook 2005. Paris: OECD. Organization for Economic Co-operation and Development (2006a). OECD broadband statistics. Paris: OECD. Retrieved July 5, 2006 from http://www.oecd.org/document Organizatio n for Economic Co-operation and De velopment (2006b). Multiple play: Pricing and policy trends. Paris: OECD. Organization for Economic Co-operation and Development (2007a). OECD broadband statistics. Paris: OECD. Retrieved July 26, 2007 from http://www.oecd.org/document

PAGE 144

144 Organization for Economic Co-ope ration and Development (2007b). Communications outlook 2007. Paris: OECD. Organization for Economic Co-operation and Deve lopment (2007c). Mobile multiple play: New service pricing and policy im plications. Paris: OECD. Re trieved October 29, 2007 from http://www.oecd.org Organization for Economic Co-operation and De velopment (2007d). Fixed-Mobile convergence: Market development and policy issues. Paris: OECD. Retrieved October 29, 2007 from http://www.oecd.org Organization for Economic Co-operation and De velopment (2007e). Catching-up in broadband: What will it take?. Paris: OECD. Retrieved October 2, 2007 from http://www.oecd.org Owen, B. M. (2002). Broadband my steries. In Crandall, R.W. and Alleman, J.H. (Eds.), Broadband: Should we regulate High-Speed Internet (pp.9-38), AEI-Brookings Joint Center for Regulatory St udies, Washington, DC. Pagani, M. (2004). Determinants of adoption of third generation mobile multimedia services. Journal of Interacti ve Mobile Marketing,18 (3),46-59. Picard, R.G. (2000). Changing business models of online contents services: There implication for multimedia and other content producers. International Journal on Media Management, 2, 60-68. Prieger, J.E. (2003). The supply side of the digital divide: Is there equal availability in the broadband Internet access market. Economic Inquiry, 41(2), 346-363. Rappoport, P. N., Kridel, D. J., Taylor, L. D. and Alleman, J. (2001). Residential demand for Access to the Internet. In Madden, G. (Ed.), Emerging telecommunications networks: The international handbook of telecommunications economics, Volume II. (pp.1-20), Edward Elgar Publishers, Cheltenham, UK. Reese, S.D. (1988). New Commu nication Technologies and the Information Worker: The Influence of Occupation. Journal of Communication, 38(2), 59. Rice, R.E. & Katz, J.E. (2003). Comparing Internet and mobile phone usage: Digital divides of usage, adoption, and dropouts. Telecommunications Policy, 27, 597-624. Ridder, J. D. (2007, August). Catching-up in br oadband: What will it ta ke?. Paper presented to the 35th Research Conference on Communication, Information and Internet Policy, Arlington, VA. Rohlfs, J.H. (2001). Bandwagon Effects in High -Technology Industries. Cambridge, MA: The MIT Press. Rogers, E. M. (2003). Diffusion of innovations. New York, NY: Free Press.

PAGE 145

145 Rouvinen, P. (2006). Diffusion of digital mobile telephony: Are developing countries different? Telecommunications Policy, 30, 46-63. Savage, S. J. & Waldman, D. (2005). Broadband Inte rnet access, awareness, and use: Analysis of United States household data. Telecommunications Policy, 29 (8), 615-633. Shapiro, C. & Varian, H. R. (1999). Information Rules. Boston, MA: Harvard Business School Press. Time Warner (2007). Time Warner cable news. Retr ieved March 29, 2008, from http://www.timewarnercable.com Turner, S.D. (2006). Broadband reality check II: The truth behind Americas digital decline.Washington, DC: The Free Press. United Nations Development Program (2004). Human development report 2005. New York, NY: UNDP. United Nations Development Program (2005). Human development report 2006. New York, NY: Author. Viard, B., (2007). Do switching costs make mark ets more or less competitive: The case of 800number portability. Rand Journal of Economics, 38, 146-163. Wallsten, S. (2006). Broadband and unbundling regulations in OE CD countries. Working paper 06-16, AEI-Brookings Joint Center for Regulatory Studies. Warschauer, M. (2003). Technology and social incursion: Rethinking digital divide. Cambridge, MA: The MIT Press. Weiser, M. (1991, September). The computer for the 21st Century. Scientific American. Wikipedia. (n.d.). Retrieved October 11, 2007, from http://en.wikipedia.org Williams, M. (2006). Cutting the cord to analog phone. Retrieved March 31, 2006, http://www.networkworld.com/n ews/2006/081006-no-analog-phone.html Wirth, M. (2006). Issues in media convergence in Albarran, A., Chan-Olmsted, S.M., & Wirth, M. (ed). Handbook of Media Management and Economics. Mahwah, New Jersey: Lawrence Erlbaum & Associates. Sawng,Y., Lee, J. & Han, H.(2005). Market expansion and customer acceptance for IT convergence related new business. Information and Telecommunication Policy Research, 12 (1), 189-221.

PAGE 146

146 Shelanski, H. (2003). Competition policy for 3G wireless services. In E. M. Noam & D. Steinbock (Eds.), Competition for the Mobile Internet (pp. 39-54). Norwell, MA: Kluwer Academic Publishers. Xavier, Patrick (2001). Licensing of third generation (3G) mobile: Briefing Paper. ITU Workshop on Licensing 3G Mobile, September 19-21, 2001, Geneva, International Telecommunication Union.

PAGE 147

147 BIOGRAPHICAL SKETCH Sangwon Lee received his bachelo rs degree in public adm inist ration at Yonsei University in Seoul, Korea. He also earned a masters de gree in telecommunicati on at George Washington University in Washington, DC. He has worked for consultant and analyst in telecommunication sector. Lee conducted research in the area of telecommunication policy, media economics, media management, and new media technology and regulati on. His work has been published in journals such as International Journal on Media Management Info: the Journal of Policy, Regulation, and Strategy for Telecommunication, Information and Media, and Internationa l Journal of Mobile Marketing.