TeleGeography 1999 Global Telecommunications Traffic Statistics and Commentary This work is based upon sources believed to be reliable, but the publisher does not warrant the accuracy or completeness of any information for any purpose and is not responsible for any errors or omissions. This work is for the confidential use of subscribers. Neither the whole nor any part of this publication may be reproduced or transmitted in any form or by any means, electronic, mechanical, photocopied, recorded or otherwise, with- out prior written consent from TeleGeography, Inc. All rights reserved. Copyright © TeleGeography, Inc. 1998 First Printing--October 1998 Second Printing--November 1998 Third Printing--January 1999 This is copy 102944 Do not reproduce. ISBN ! -886142-15-7 Printed in the United States of America TeleGeography, Inc. ¯ 1730 Rhode Island Avenue, NW ¯ Suite 1205 ° Washington, DC 20036 USA Tel. ÷! 202 467 0017 Fax ÷! 202 467 0851 E-mail: info@telegeography.com http://www.telegeography.com TeleGeography 1999 © TeleGeography, Inc. 1998 TeleGeography, Inc. This report was prepared by TeleGeography, Inc., an independent research and pub- lishing company based in Washington, D.C. The TeleGeography 1999 research and production team included: Gregory Staple .................. Editor in Chief Jason Kowal ................. Managing Editor Timothy Stronge ................. Deputy Editor Mickie Kennedy .............. Research Associate Janice Gentile ............... Research Associate Jonathan Palmatier ........... Research Associate Caroline Bouhdili ............. Research Assistant Kenneth Cukier .............. Contributing Writer Zachary Schrag ................. Map Designer Irena Slage ..................... Cover Design The following also are available from TeleGeography, Inc.: New International Carriers 1998 Americas Edition (ISBN 1-886142-14-9) New International Carriers 1998 Europe Edition (ISBN 1-886142-13-0) New International Carriers 1998 Asia/Pacific Edition (ISBN 1-886142-12-2) Telecom Map of the World- 1997 Edition (ISBN 1-86186-096-X) TeleGeography 1997/98 (ISBN 1-886142-10-6) Direction of Traffic 1996 (ISBN 1-92-61-06291-1 ) For all inquiries, please contact: TeleGeography, Inc. 1730 Rhode Island Avenue, NW Suite 1205 Washington, DC 20036 USA Tel. 1 202 467 0017 Fax +1 202 467 0851 E-maih info@telegeography.com For more information, please visit our web site: http://www.telegeography.com © TeleGeography, Inc. 1998 TeleGeography 1999 Contents List of Figures ..................................................................... 4 Preface .......................................................................... 7 Introduction ...................................................................... 9 Carriers International Carrier Evolution ..................................................... 26 Regional Analysis International Telephony & Asian Network Expansion .................................. 41 International Long Distance in Latin America ....................................... 44 Emerging North American Carriers .............................................. 47 International Services of U.S. RBOCs ............................................ 50 European Carrier Financing .................................................. 53 Market Shares of Competing International Carriers, 1988-1997 .............................. 54 The Top 40 International Carriers ................................................... 56 Pricing Elements of an International Calt .................................................... 59 International Call Costs for U.S. Carriers .......................................... 60 The Accounting Rate Regime in Transition .............................................. 62 International Settlement Rates ................................................. 66 FCC Settlement Benchmarks .................................................. 68 National lnterconnection Rates ................................................ 70 Wholesale Prices Trading Minutes and Bandwidth ............................................... 71 What is Sparking the New Markets for Power? ...................................... 77 Retail Prices Retail Prices for a Three Minute Call ............................................. 80 U.S. and U.K. Tariff Maps, 1994-1998 ........................................... 82 Origin of U.S. Carrier Revenues ..................................................... 84 Facilities International Submarine Cables .................................................... 86 The Rise of Private Cable Systems .............................................. 92 New Cross-Border Networks in Europe ........................................... 94 International Circuit Usage by U.S. Carriers ............................................ 98 A Primer on Bits .............................................................. 101 International Communications Satellites .............................................. 102 New Global Mobile Satellite Services ........................................... 106 The Internet The Global Internet: A Primer .................................................... 112 lnternet Backbone Maps .................................................... 115 Internet Hosts by Economy, 1993-1997 ............................................. 146 Journey to the Center of the Web .................................................. 150 Traffic Outgoing International Traffic Tables by Country ........................................ 155 Global Traffic Review ........................................................... 253 International Traffic by Region ................................................ 255 International Traffic by Country ............................................... 259 International Traffic by Route ................................................. 262 Alternative Traffic Arrangements ............................................... 265 Methodology ................................................................ 268 Global Reference: Blue Pages National Telecommunications Indicators .............................................. 270 International Dialing Codes ...................................................... 275 End Pages Calling the Planets ............................................................ 286 About TeleGeography, Inc ........................................................ 290 TeleGeography 1999 © TeleGeographyo Inc. 1998 List of Figures Introduction F~gure Carriers 1. The Global Market. Someone Is Always Trading When You’re Not ....................... 10 Figure 2. Living m a Two Super-Power World .............. 11 Figure 3. Clocks are Maps--Time is an Atlas ............... 12 Figure 4. Brand~ng the World: The Best Known Telecom Brands .. 15 Figure 5. Global Numbers for Global Markets .............. 16 Figure 6. The OSI Protocol Stack .................... 17 Figure 7. Virtual Places, Real Money .................. 18 Figure 8. Next Generation Networks: Betting on the Future .... 19 Figure 9 Next Generation Networks Will Cost Less ........... 20 F~gure 10. Data Eclipses Voice--Will Revenue Follow? ......... 21 Figure 11. Mars Calhng: How NASA May Keep in Touch with the Red Planet ...................... 22 International Carder Evolution Figure 1. The International Carder Boom ............... 26 Figure 2. Countries with Multiple International Carders ....... 27 Map of International Carder Competition ................ 28 Figure 3. Typical Infrastructure of an International Call ....... 30 F~gure 4. Alternative Call Delivery Methods .............. 31 Figure 5, The Call Turn-Around Effect ................ 32 Figure 6. International Camer Affihates of Cable & Wireless ..... 33 Figure 7 Selected Multinational CarderAuthodzations ...... 34 Figure 8. International Camer Mergers and Acquisitions, Sept 1997-Oct. 1998 ............ 35 Figure 9. Upcoming International Carder Pdvat~zations ........ 36 Figure I0. Traffic Base of Incumbent and New Carders ...... 37 Map of WTO Commitments ......................... 38 International Telephony & As~an Network Expansion Figure 1. Domestic and International Telecom Revenue, 1997 .41 Figure 2. S~mplified IRR of New Line Investment ........ 42 International Long D~stance ~n Latin Amedca F~gure I. Comparative Latin Amedcan Long Distance Rates .... 44 F~gure 2. Debt and Equity Raised in Latin America, 1998 ..... 45 Emerging North Amencan Carners F~gure 1. Performance of Emerging International Long Distance Carders ......................... 47 F~gure 2. Financmgs of Emerging International Long D~stance Carders .......................... 48 International Services of U.S. RBOCs Figure 1. RBOC International Services Authorized by the FCC .... 50 Figure 2. Changing Landscape of U.S. Regional Bell Operating Companies .................... 5I F~gure 3. RBOC Section 271 Applications ................ 52 European Carder Fmancings F~gure 1. Major European Financings, January 1997- October 1998 ........................ 53 Market Shares of Competing International Carders ................. 54 The Top 40 International Camers ........................ 56 Pricing Elements of an International Call F~gure 1. International Call Components .................. 59 International Call Costs for U.S. Carders ..................... 60 The Accounting Rate Regime in Transition Figure 1. How Accounting Rates Work ................. 62 Figure 2 The Logic Behind Refile ..................... 6:3 Figure 3. FCC Settlement Benchmarks (USS) .............. 63 F~gure 4 FCC Benchmark Deadlines Approaching .......... 64 Figure 5. ITU Benchmarks ............................ 64 Figure 6 Termination Rates Compared .................. 65 International Settlement Rates ............................... 66 FCC Settlement Benchmarks .............................. 68 National Intereonnection Rates ............................... 70 Trading Minutes and Bandwidth Figure 1. Buy Your Minutes Online ..................... 71 Figure 2. Arbmet and Company ........................ 72 Figure 3. The Band-X Index of U.K International Wholesale Pdces, 1997-1998 ...................... 7:3 Figure 4. Why Settlement Pohcy Matters ................ 74 F~gure 5 Wave of the Future? ....................... 75 What is Sparking New Markets for Power?. Figure 1. U.S. Wholesale Electricity Trades ............... 77 F~gure 2. Power Markets Overview ............... 78 Retad Prices for a Three Minute Call ......................... 80 International Tadff Maps, 1994 & 1998 ...................... 82 Odgin of U.S. Carder Revenues Figure I. Revenues for International Voice Service, 1997 ...... 84 Figure 2. International Call Revenue by Destination, 1997 ..... 84 Facilities International Submarine Cables Map of Global Submarine Cable Capacity ................ 86 Figure 1. Map of Major Submarine Cables in Asia/F~cific ..... 88 F~gure 2 Map of Major European Cables ................. 89 Figure 3 Map of Major Caribbean Cables ............. 89 Figure 4. Major International Submarine Cables by Region ..... 90 Figure 8, Out of Service Trans-Oceanic Cable Systems ...... 91 The R~se of Private Cable Systems Figure 1. The History of Submarine Cable Capacity .......... 92 Figure 2. International Submarine Cable Capacity and Cost ..... 93 New Cross-Border Networks in Europe F~gure 1. MCI WoridCom’s European Network .............. 94 Figure 2. New European International Networks ............ 95 Figure 3. Hermes’ European Network .................. 97 International Circuit Usage by U.S, Carders Figure I. International Circuit Usage by Region, 1995-1997 ..... 98 © TeleGeography, Inc. 1998 F~gure 2. Idle Orcuits of U.S Carders by Region, 1995N1997 ..................... 99 Figure 3. International Orcuit Usage for Selected Routes, 1995-1997 .................... 100 A Primer on Bits Figure 1. Measuring Bytes Bit by Bit .................... 101 Figure 2. Measunng Telecommunications Bandwidth--DS-0 to OC-192 .......... Internal Communications Satelhtes Map of International Satellite Capacity Figure 1. Orbital Map of Major Geostationary Satelhtes ....... 103 F~gure 2. Major International Telecommumcations Satellites in Geostat~onary Orbit ................... 104 Figure 3. Intemat~onal Satelhte Capacity and Cost .......... 105 The New Global Mobile Satellite Services Figure 1. Network Architecture for Global MSS ............. 106 F~gure 2. Chaper by the Minute ...................... 107 Figure 3. Firet-Generat~on MSS Operators ............... 107 Figure 4. Market Forecasts for Second-Generation Global MSS,1998-2002 ..................... 108 The Internet The Global tnternet: A Primer F~gure 1. Geographic D~stribution of Intemet Hosts and Content ......................... 113 Figure 2. The Intemet m 1971--Across the U.S. at 56 K~ps .. .114 Figure 3. Major U.S. internet Backbones ~n 1998~155 Mbps to 2 Gbps ....................... 115 Figure 4 Asian Intemet Backbone Connectivity ........... 116 Figure 5. European Intemet Backbone Connectivity ...... 11 F~gure 6. African Internet Backbone Connectivity .......... 118 F~gure 7. Backbone Connectivity for Latin Amenca and the Middle East ................ 119 F~gure 8. Internet Traceroute Maps, July 1998 ............ 120 Figure 9. The World’s Top ISPs .................... 122 Figure 10. The U.S. Controls the Internet~s Directory System .... 123 Figure 11. The lnternet~s New Governors ............... 124 Figure 12. Power Shifts in lnternet Governance. 1982-1998 ...125 Figure 13. On the Net, Bandwidth Determines Distance ....... 126 Figure 14. Distance and Delay on the Cable & Wireless Backbone ....................... 127 F~gure 15. A Primer on Peering .................... 128 Figure 16. Regional lntemet Growth ................ 129 F~gure 17. Lessons of the MCI-WorldCom Merger ....... 130 Figure 18. Internet Growth in Western Europe ............ 131 F~gure 19. Intemet Host Growth 1981-1998 .......... 132 Figure 20 NAP Traffic as a Measuring Stick ............. 133 Figure 21. Start on the Net: B~g Crowds, But Not for Long .... 134 Figure 22. A Signaling Channel for the Internet? ............ 135 Figure 23. A Slow On-Ramp. Internet Users and Access Speeds ........................... 136 F~gure 24. Why Broadband Matters .................. 137 Figure 25. Satelhtes and the Internet .................. 38 Figure 26. Internet Telephony: Build Once, Run Anything? ..... 139 TeleGeography 1999 Figure 27. The fill and the Internet ................... 140 Figure 28. Future Scenarios ....................... 14t Figure 29. Voice’s Shrinking Role ..................... 142 F~gure 30 Internet Protocol Developments on the Horizon ..... 143 Figure 31. Quelle Langue is the Internet? ............... 144 Internet Hosts by Economy, 1993-1997 Figure 1. Internet Host Computers by Economy, A-D ......... 146 Figure 2. Intemet Host Computers by Economy, D-L ......... 147 F~gure 3. Internet Host Computers by Economy, L-S ......... 148 Figure 4. Internet Host Computers by Economy, S-Z ....... 149 Journey to the Center of the Web F~gure 1. Long and Medium Range Web Scans ............ 150 Figure 2. Short Range Web Scan .................. 151 Traffic Outgoing International Traffic Tables, by Country Albania .................................... 155 Andorra ...................................... 156 Argentina .................................. 157 Armenia ....................................... 158 Australia .................................... 159 Austria ............................... 160 Bahamas ....................................... 161 Bahrain ..................................... 162 Bangladesh ................................. 163 Belarus ....................................... 164 Beloum .................................... 165 Bolivia ..................................... 166 Botswana ................................ 167 Brazil .................................. 168 Brunei .............................. 169 Canada .................................... 170 Chile .............................. 171 China ........................................ 172 Colombia .................................. 173 Commonwealth of Independent States ............. 174 Costa Rica ................................ 175 Cyprus ................................. 176 Czech Republic ........................... 177 Denmark ..................................... 178 Dominican Reput~l~c ....................... 179 El Salvador ............................... 180 Estonia ................................. 181 Finland ............................... 182 France ...................................... 183 French Polynesia .............................. 184 Germany .................................. I85 Greece .................................... 186 Guyana ................................... 87 Hong Kong ................................ 188 Hungary ...................................... 189 Iceland ................................ 190 India ....................................... 191 Indonesia ..................................... 192 TeleGeography 1999 © TeleGeographyo Inc. 1998 List of Figures Iran .................................... 193 Ireland ...................................... 194 Israel ........................................ 195 Italy ........................................ 196 Japan ....................................... 197 Jordan ....................................... 198 Kazakhstan .................................... 199 Republic of Korea ............................... 200 Kuwait ........................................ 201 Luxembourg ............................. 202 Macau ...................................... 203 Macedonia .................................. 204 Malaysia ...................................... 205 Malta ..................................... 206 Mexico ....................................... 207 Moldova ...................................... 208 Namibla .................................... 209 Netherlands ................................ 210 New Zealand .................................. 211 Nicaragua ................................... 2] 2 Norway ...................................... 213 Oman ...................................... 214 Pakistan ....................................... 215 Panama ............................... 216 Papua New Guinea ..................... 217 Paraguay ................................. 218 Peru ................................ 219 Philippines .................................. 220 Poland ................................. 221 Portugal ............................... 222 Romania ..................................... 223 Russia ................................ 224 Saudi Arabia .................................... 225 Singapore ................................... 226 Slovak Republic .................................. 227 Slovenia ................................. 228 South Africa .............................. 229 Spain ..................................... 230 Sd Lanka ................................ 231 Sweden .................................... 232 Switzerland ............................... 233 Syria ..................................... 234 Taiwan .................................. 235 Thailand .................................. 236 Turkey ....................................... 237 Ukraine ..................................... 238 United Arab Emirates ....................... 239 United Kingdom--Outgoing ........................ 240 United ~ngdom--lncommg ........................... 241 (continued) U.K Top 100 Correspondents .................... 242 Umted States--Outgoing .......................... 243 United States--Incoming ......................... 244 U.S. Top 100 Correspondents ......................... 245 U.S.--Traffic by Carrier ............................ 246 U.S.--Resale Traffic .............................. 247 Uruguay ....................................... 248 Uzbek~stan .................................... 249 Venezuela .................................... 250 ~iemam ................................... 251 Yugoslavia ...................................... 252 Global Traffic Review Figure 1. International Traffic and Mmn Line Growth ......... 253 Rgure 2. International Traffic, Revenue, and Subscriber Growth ............................ 253 Rgure 3. Intercontinental Traffic Flows, 1995 & 1997 ....... 254 Rgure 4. Interregional Traffic Flows, 1997 ............... 255 Figure 5. International Traffic by Origin, 1997 ............. 255 Rgure 6. Traffic Growth by Reoon, 1996-1997 ........... 255 Figure 7. European Telecommunications Traffic Flows, 1997 .... 256 F~gure 8. Latin American Telecomrnumcations Traffic Flows, 1997 .......................... 257 Figure 9. Asian Telecommunications Traffic Flows, 1997 ....... 258 Figure I 0 International Traffic Growth for Selected Countnes, 1996-1997 .................. 259 Rgure I 1 Traffic Balances for Selected Countries, 1997 ...... 259 Figure 12. International Traffic Indicators, 1997 ........... 260 Figure 13 The Top 50 International Routes, 1997 ........ 262 Figure I4. Traffic Imbalances on U.S. Routes, 1997 ......... 263 Figure 15. Traffic Imbalances on Non-U.S. Routes, 1997 ...... 263 Figure 16. International Outbound Routes with Rapidly Growing Traffic, 1996-1997 ........... 264 Figure 17. International Outbound Routes w~th Declining Traffic, 1996-1997 ............. 264 Figure 18. Alternative Traffic Arrangements as a Share of Global International Traffic, 1996-1997 ....... 265 Rgure 19. Selected Odgin Countries of Call-Back Traffic, 1997 ....................... 265 Figure 20. Reported U.S.-Hubbed Refile and Calling Card Traffic ............................ 266 Rgure 21. Illicit Bypass Traffic, 1997 .................. 266 Figure 22. International IP Telephony Forecasts ............. 267 Global Reference~Blue Pages National Telecommumcations Indicators ............... 270 tnternat=onal Dialing Codes ....................... 275 Map of Wodd Dialing Codes ........................ 278 North American Area Codes ........................ 280 Map of North Amencan Area Codes ................... 282 © TeleGeographyo Inc. 1998 TeleGeography 1999 Preface Beyond the Beginning Today, there is no doubt that telecommunication carriers and customers have reached the shores of a new world. Like Columbus in 1492, after his ships landed in the West Indies, we too have survived the uncertainty of the open seas. We have now reached the terra firma of a new age. And we have some idea, based on experience, of the con- tours and characteristics of the new terrain. We also know the vast opportunity pro- vided by the interior into which we enthusiastically move. Deregulation, privatization and liberalization not only have been predicted and antici- pated, but also experienced. The WTO agreement has evolved from an exciting news story. It now serves as the underpinnings of business plans for new competing telecom companies all over Europe. The Internet is not just a trendy novelty with great poten- tial, but an increasingly routine part of everyday life and commerce. The migration from switched networks to IP/data packets has moved from arresting futurolog3, to conven- tional wisdom substantiated by fact. So, with a bow to Winston Churchill, we may be able to look back to the close of the twentieth century, not as anything approaching the end, or even the beginning of the end, but as the end of the beginning--of the Communications Age. As we solidify our beachhead and prepare a forward march into the new telecommuni- cation world and the opportunities that surely await, a reliable guide is essential. That’s why MCI WorldCom and Stentor are proud to sponsor TeleGeographuy 1999. For cus- tomers and providers, for regulators and investors, even for wandering citizens who want to understand the new world, this industry standard supplies trustworthy data and guideposts that are essential for the trip. Michael J. Rowny, President and CEO International Ventures, Alliances & Correspondent MCI WorldCom, Inc. Carol Stephenson President & CEO Stentor Resource Centre Inc. TeleGeography 1999 © TeleGeography, Inc. 1998 Acknowledgements W e wish to thank the numerous carriers, government departments, regulators and international organizations from around the world who responded to our requests for information. This report would not exist without the help of the dedicated people at these organizations who took the time to ensure that the data reported here are as current and accurate as possible. We would also like to thank the people who helped review the final draft of this book, including: Stephan Beckert, Nancy Dysart, Cindy Obenstine, Jonathan Starzyk, and lan Watson. TeleGeography 1999 was supported, in part, by a publication grant from MCI WorldCom, Inc. and Stentor Resource Centre Inc. As in the past, however, this grant was made without any pre- condition; TeleGeography, Inc. is solely responsible for the report’s editorial contents. The Editors © TeleGeography, Inc. 1998 TeleGeography 1999 Introduction by Gregory Staple, TeleGeography, These are the days of miracles and wonder This is the long distance call... The way the camera follows us in slow-too The way we look to us all. Paul Simon, Graceland Last year, we tracked the telephone industry’s shift from carr- er club to open market. The 1997 World Trade Organization (WTO) agreement liberalizing market entry by foreign tele- phone companies capped the old regime. This year we consider how this new market-oriented regime is changing the industry’s economic geography and vice versa. Telecom networks affect the global economy which, in turn, impacts the telecom industry’s future prospects. In Part II we review some of the trans-national economic strategies--brand- ing, service portals, and backbone networks--being pursued by major telecom companies. The Internet’s cross-cutting power is given special attention. We close in Part Ill with a short note on the challenge of making tomorrow’s network services not only fulty international but interplanetary too. Let us start, though, with some down-to-earth economics. 1. The Networked Economy Many readers of TeleGeography will have experienced that rush of emotion--panic, fear, depression--which accompanies the sharp drop in the price of one’s listed shares. It can happen overnight (see Figure 1). The world’s round-the-clock currency and securities markets have teft few nations untouched and have dramatized the economic spillover from the last decade’s investments in cross-border telecom facilities. International telecom networks relay the successes and failures of once distant countries worldwide, and often in unpredictable and unprecedented ways. That has made the world smaller but, at the same time, less manageable, as poticymakers strug- gle to account for the networked effect of hundreds of new places and actors. Telegeography may thus lead even the most powerful countries to lose control over social and economic matters which were thought to be purely local. So large are today’s currency markets, so great are the cross- border flows of capital, and so powerful are the networks which oil these transactions that no one country can intervene effec- tively. "It’s just like the lnternet," as one foreign affairs writer put it, "nobody is in charge" of the global economy (see Figure 2). Whether or not governments are able to change that, as some propose (e.g., by creating a new global reserve bank and more flexible financial safety nets), it is now clear that the telecom sector itself will not escape unscathed from the present eco- nomic crises. Incumbent carriers and those newcomers fortu- nate enough to have issued shares before mid- 1998 are now in a much stronger position. Other market entrants have been crippled. The cutback in lndonesia’s foreign-led network expansion program and the uncertain future of network com- petition in Brazil are but two examples. The roll-out of mobile satellite services may be particularly hard hit if the pumhasing power of the world’s least wired markets (Asia, Africa) remains low just as new systems, such as Iridium and Globalstar, begin service. Most telecom operators see themselves as innocent bystanders in this unfolding economic drama. In some ways, they are: international carriers did not cause the "melt-down" in Asian currencies or South American share prices. Nor can they sin- gie-handedly reflate the worlds’ economies. On the other hand, without global networks, the breadth of the recent market shocks, and the speed of their transmission to other economies--the so-called financial contagion--almost certainly would have been more limited. Likewise, without the global currency and securities markets which telecom networks make possible, any economic recovery will be far more pro- tracted and uneven. It would thus be a mistake to discount the role which telecom networks have played in bringing us to the current economic juncture or in surmounting it. To see why, a brief historical reprise may be useful. Since the late 1970s, first in the U.S. and Western Europe, and later more generally, telecom policy makers have advanced a new model for the industry, one based on private ownership and competition rather than state-owned monopolies. The ini- tial stimulus for reform came largely from corporate users, especially in the financial community, and the computer indus- try. They feared that the enormous potential of computer-to- computer connections (data networks, electronic trading sys- tems) would be frustrated if telecom sector rules (licensing, tar- iffs) applied. Businesses also wanted the right to self-provision networks, to buy their own terminal equipment, and to obtain private or leased lines on flexible terms. By the early 1980s, the reformers’ agenda began to win wider political support. Governments admitted the poor performance TeleGeography 1999 © TeleGeography, Inc. 1998 Figure 1. The Global Market: Someone Is Always Trading When You’re Not October 22-23, 1997: Hong Kong Market Drop Pulls New York Down 17697.61 Tokyo -3% 11637.77 5149.8 17151.55 Hong K, ong -10% 10426.30 London -3°/~ 4991.5 ; 8031.65 ~t N~w York (DJ ?,) -2% 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 p.m. Oct. 22 a.m. Oct. 23 p.m. November 16-17, 1997: Japanese Surge Sparks Worldwide Rally 15082.52 Tokyo +7.96°A 16283,32 4741.80 i i Londl =n +2.64% 5"’ 7572.48 4~67.09 New York (DJIA) +1.66% 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 p m. Nov.16 a.m. Nov. 17 p.m. Note: Figures show Hon9 Kong, Tokyo, London, and New York (Dow Jones Industrial Average) stock exchange indexes. The Tokyo and Hong Kong exchanges close for lunch. Sourc e: The Wa//Street Journa/, © Dow Jones, Inc. 1998 © TeleGeography, Inc. 1998 TeleGeography 1999 Figure 2. Living in a Two Super-Power World Few mainstream journalists have paid much attention to the ways in which electronic networks are shaping the world economy. One who has is Thomas L. Friedman, a foreign affairs columnist for The New York Times, now on sabbatical to finish a book about globalization, Some of Friedman’s thoughts are excerpted below, The first except is taken from an Apdl 6, 1998 speech in Washington, D.C, to the Anti- Defamation League (ADL); The second is from a August t5, 1998 column in The Times. "The global economy is an entirely anonymous force. It’s just like the Internet. Nobody is in charge ... Last year, I spoke at a conference on globalism in Morocco ... A former Algerian Prime Minister was there and he got up when I was done and said: ’Mr. Friedman, I have to tell ~tou, this global economy ~lou talk about, this globalization, is just another conspiraql to keep us down. It’s like Zionism and Colonialism and Westernism .... Mr. Prime Minister, I said, I have to tell you something. It’s much worse than ~lOU think ... I wish t could tell you that we were back there turning the dials and pulling the levers ... But we’re not. We’re not thinking about you at all. That’s what’s really sca~j .,. We’re too caught up under the same pressures to downsize, to streamline, to get our economy in order, wornjing about the bond market and the stock market and the man from Moody’s. We live in a two super-power world. There is the United States and Moody’s bond rating service. The United States can destroy you by dropping bombs and Moody’s can destroy you by down-grading ~tour bonds. It doesn’t matter whether you are the United States ... or Canada or Algeria ..,, "1 had an unusual interview the other day. It was with Dejair Birschner, the 48 gear old Ma~lor of Una, a town of 52,000 inhabi- tants on the edge of Brazil’s Atlantic rain forest, [I toured] the rain forest ecopark near Una that is supposed to provide jobs so .... peo- ple won’t strip the forest bare and destroy one of the world’s great ecosystems. Mayor Birschner said he understands that logging is not sustainable anymore but he also knows that his town really isn’t prepared for life without logging... Mayor Birschner represents a whole generation of people ... who are trapped in a no man’s land, between the computer generation that their kids, ff they’re lucky, might get up to speed for and their parents" generation that enjoy the stable existence from logging and farming. I knew what [the Mayor] was trying to ask me: "M~I villagers can’t five off the forest anymore, and we’re not equipped to live off computers. What are we supposed to do?" ,,. There are a lot of [people] out there trying to avoid becoming road kill on the information highway. They are the villagers in Una, the pensioners in Russia, the unemployed in China, Indonesia, South Korea and Thailand. Analysts have been wonder- ing for a while now whether those left behind by globalization will develop an alternative ideology to liberal, free market capitalism ... I don’t think there will be an alternative ... [Those] who can’t keep up will just eat the rain forest--each in his own way, without t~ing to explain It or justi~ It ... if governments don’t develop safety nets for the left-behinds, to protect them from hitting bottom and to help lift them into the game ..." ©Thomas L Friedman and TheNew York ~imes, 1998 of many state-owned operators and the very large amounts of capital needed to expand and to modernize their networks. The growing telecom demands of the service sector, already the largest employer in some countries, provided a further con- stituency for reform. Encouraged by such information-intensive businesses, most governments came to view telecom sector reform as a precondition for success in the information econo- my. The victory of this reform agenda is well documented. Starting with British Telecommunications and Nippon Telegraph & Telephone in the mid-1980s, dozens of carders have been pri- vatized, numerous markets have been opened to competition, and sectoral investment has soared. The change in network access--in global connectivity--is truly astonishing, and the financial services industry has been among the greatest benefi- ciaries. Some of the statistics are worth reviewing because they show just how novel and, thus, how unprecedented is today’s networked economy. Since 1985, carriers have installed over one half of the world’s fixed telephone network (over 400 million lines) and almost all of the world’s mobile telephone facilities (another 175 million lines). Further, in less than 15 years, developing countries have added more telephone lines than the world’s industrialized countries (excluding the U.S.) installed during the first 100 years after the telephone’s invention (circa 1876). Similarly, but 15 years ago, the largest state-of-the-art trans- Pacific fiber optic cable, TPC-2, could carry just 1,700 simulta- neous calls; next year a new China-U.S. cable will be able to carry almost 5 million calls at once, and will increase trans- Pacific capacity more than sixfold. Call volumes already reflect the rapid change in connectivity. Within the next week, Americans will spend more time on the phone to Russia and China than they did during all of 1985. And, this year, inter- national telephone traffic from South Korea to Japan will more than triple Korea’s calls to the entire world in 1985; the same will be true for Korean traffic to China and the United States. And then, of course, there is the Internet. In 1988 it connect- ed a few hundred thousand people, mostly in the U.S. Today the Internet links over 100 million users in more than 150 countries, and as e-commerce gains wider consumer support, the lnternet is rapidly becoming a marketplace of choice for trading shares and numerous other products. TeleGeo~raphy 1999 © TeleGeograph/j, Inc. 1998 Figure 3. Clocks are Maps--Time is an Atlas The Circadian Geography of the Network t~erals indicate London Time gain Shanghai 40 20 20 80 loo Paris People (5.6 billion = 100 on scale} Main telephone lines (462 million = 100 on scale) Cellular subscribers (82 million = 100 on scale) Internet hosts (13 million = 100 on scale) © TeleGeography, Inc. 1998 Global networks make time rather than distance the main barrier to communication. Time zones, networked together during office hours, have become continents. But because network terminals are not evenly distributed around the world, the size and position of these new continents varies. This map shows the proportion of the wodd’s people, telephones, cell phones, and Intemet hosts which may be connected during the working day (9 a.m. to 5 p.m.) in three cities: Shanghai, Pads and New York. At 8 a.m. London time, the beginning of Pads’s workday, the largest number of people and phone lines are accessible as Europe and Africa join their Asia colleagues at the office. But, as China goes home (after 9 a.m. in London) network connections drop until the Americans begin to arrive at work at 2 p.m. (I 4:00) London time. During the next few hours, available Intemet hosts and cell phones peak only to taper offas New York goes home (22:00 London time). Each day, while the demand for connections moves clockwise from east to west with the sun, available network capacity appears to move counterclockwise as network resources are idled during the night. © TeleGeography, In€, 1998 © TeleGeogroph~l, Inc. 1998 TeleGeo~raphy 1999 Figure 3. Clocks are Maps--Time is an Atlas (continued) Indexed Population and Network Terminals By l~me Zone, 1996 100 ¯ Population (1.34 billion = 100) 80 ¯ Main Lines (169 million = 100) ;~ Cellular Subscribers (30 million = 100) 60 Internet Hosts (6 million : 100) ~o Key The map on the facing page was created by summing data for each indicator (population, telephones, etc.) up to nine whole time zones (the period from 9 a.m. to 5 p.m.). The data were also scaled. Raw data for individ- ual time zones are shown on this page. Data are for 1996. Data for the United States and Canada have been scaled according to the population of each state and province; other coun’~’)es that span multiple time zones have been treated as if the total population and all ter- minals were found in the capital city. Data source: International Telecommunication Union. "Clocks are Maps" was designed and produced by Gregory Staple and Zachary Schrag. © TeleGeography, Inc. 1998. All rights reserved. Population and Network Terminals By ~me Zone, 1996 (millions) Main Ceil Internet Time Population Lines Phones Hosts London 12:00 181 34 8 1 Paris 13:00 609 169 26 3 istanbul 14:00 452 62 4 0 Moscow 15:00 387 31 1 0 Dubai 16:00 6 1 0 0 Kabul 16:30 21 0 -- -- Tashkent 17:00 28 2 0 0 Bombay 17:30 1,180 23 1 0 Almaty 18:00 27 3 0 0 Rangoon 18:30 46 0 0 -- Jakarta 19:00 347 10 2 0 Shanghai 20:00 1,366 77 12 0 Tokyo 21:00 314 83 30 1 Sydney 22:00 23 10 4 1 Noumea 23:00 1 0 0 0 Auckland 24:00 5 2 0 0 Tonga 1:00 0 0 0 0 Honolulu 2:00 2 1 0 0 Eastern Pacific 3:00 .... Los Angeles 4:00 45 29 7 2 Phoenix 5:00 16 10 3 1 Chicago 6:00 191 53 12 3 New York 7:00 274 114 27 6 Caracas 8:00 55 7 1 0 Newfoundland 8:30 1 0 0 0 Rio de Janiero 9:00 197 22 3 0 Mid-Atlantic 10:00 .... Cape Verde 11:00 0 0 -- -- TeleGeography ! 999 © TeleGeography, Inc, 1998 These new global networks have not only helped the world’s market makers. By plugBing into a network, small countries which are far from the world’s major markets need no longer come second to more centrally located competitors. In addi- tion, every new phone line and lnternet port added in one country can be used to leverage the power of hundreds of mil- lions of network terminals and billions of investment dollars elsewhere. What matters today is whether you are connected and how; network access and bandwidth are the great equaliz- ers (see "On the Net, Bandwidth Determines Distance" at page 126). But high speed network access is a two-way street. It empow- ers traders at both ends of the link, and transmits the hopes and fears of "bulls" and "bears" alike. And the networks are global, so that a once safe harbor can quickly become a storm- washed promontory. As the world’s financial markets became network-based, providing a ready dial tone for traders across the globe, almost any country may find itself battered by dis- tant financial gales. Put in more conventional terms, global telecom networks may simultaneously foster both deflation and inflation on a wide geographic scale, local monetary and fiscal policies notwith- standing. Over the medium to long term, greater connectivity is likely to act as a deflationary force by reducing production and distribution costs, and by expanding the number of buyers and sellers in the market, thus increasing competition. Yet, in the short run, the expansion of telecom capacity may have exactly the opposite effect. By sharply cutting transaction costs, especially for financial services, electronic networks may inject (and later extract) huge amounts of liquidity into the economy. They may likewise radically decentralize the creation (and destruction) of credit by private actors (banks, hedge funds, corporations) in ways we do not understand. Hence the common belief that no one is in charge of the global economy. Global networks also can make economic sentiment (i.e., mar- ket confidence) as contagious as wildfire, leading to worldwide bouts of optimism and gloom. Against this background, tradi- tional economic policy tools (interest rates, government spend- ing) may be of limited value in creating economic stability let alone renewed growth. The right approach probably will not be found until we know far more about how the tools we already have are affected by the networked economy described here. Finally we must keep in mind that the global network ties all of its users to a distinct daily rhythm. Though networks make dis- tance to market less important, they can make time more so. As a result, the ability to leverage one of the Net’s three main time zones or telecontinents may become as important to eco- nomic success as network access (see Figure 3). To sum up then, our thesis is straightforward: the triumph of telecom reform--including deregulated telephone services, unreBulated markets for data communications, and the result- ing network investments--is partly responsible for today’s eco- nomic policy dilemmas. It follows that any serious attempt to address these dilemmas must take this telecom legacy into account. II. The Economy’s New Networks The borderless, round-the-clock markets advanced by modern telecom networks have had a profound impact on the organi- zation of the telecom industry itself. Until quite recently, polit- ical geography dictated telegeography and national borders defined the market for telecommunication operators. Now a new te]egeography is emerging---one where market boundaries are mainly drawn by technology and business strategy rather than by politicians. The old geography was static and homogenous. With few exceptions, carriers were locally owned, monopolized their home market and routed international traffic directly to other countries on symmetric terms (landing fees were the same at each end). The new geography is less straightforward. It is dynamic and heterogenous. Pan-national carriers and service arrangements are commonplace, and the flow of traffic and revenues is often anything but direct, making conventional country-by-country settlement arrangements and routing rules less relevant. Instead, company and service-specific terms are popular And while dozens of smaller countries still adhere to the old regime, tomorrow’s telegeography will be distinguished by more and more local variation. A. Agents of Change The industry’s post-national geography is being shaped by sev- eral factors. Three stand out. Of prime importance is the Internet. Its packet-switched architecture and protocols have become the medium of choice for new international networks. Internet traffic patterns differ markedly from those on the pub- lic-switched telephone network (PSi-N). There are no fixed routes between service providers. International traffic typically is sent indirectly over several networks to take advantage of cheaper or less congested bandwidth. The flow of much Internet traffic also is asymmetric--reflecting the Internet’s hybrid communications model, which is like telephony (one to one) and broadcasting (one to many) at the same time. The few key strokes needed to enter the name of a far-off Web site may then generate a torrent of bits in the opposite direction. The financial terms on which Internet service providers hand off traffic amongst themselves are different as well. Politics has also played a role in changing the industry’s geog- raphy. As with other economic sectors, governments have opened most of the largest telecom markets for competition, and the WTO agreement generally prevents politicians from later closing the door This has prompted foreign carriers to © TeleGeography, Inc. 1998 TeleGeography 1999 enter new markets in droves. (See our New International Carriers directory for the details.) International joint ventures and end-to-end service arrangements are growing. Prices are falling. And investment capital has been plentiful. (That may not last, of course, given the recent downturn in equity mar- kets.) All of this has affected cross-border traffic patterns. The rapid buildout of mobile telephone networks is a third fac- tor underlining the breakup of the old world. Mobiles still gen- erate only a small portion of all international traffic. But the spread of wireless terminals, like the Internet, has changed many people’s perception of place. Once fixed and site-specif- ic, offices and homes are becoming transient--a function of per- sonal mobility and available telecom networks, a social rather than a physical construct. Every additional mobile adds to this new sense of place and to the overall demand for borderless communications, anytime and anywhere. It is much too soon to be sure about the various ways the lntemet, competition, and mobile service networks will change the geography of international communications. The process is uneven and changes in some countries are well along while the status quo continues elsewhere. B. Post-National Strategies As national boundaries cease to define the traffic base, facilities and routing options for international operators, how will the industry adapt? What will distinguish the winners among tomorrow’s stateless operators? 1. Brands One response to the declining power of physical geography is the promotion of a strong virtual substitute: a brand. For the largest telephone companies with global ambition, branded services may be paramount, especially as falling prices make discounting schemes less viable. Global brands already define the market for many products which, unlike telecommu- nications, have not been regulated closely. Brand names for the top-selling sport shoes, cars, drinks, and cosmetics--Nike, Ford, Coca-Cola0 Revlon--are better known than most coun- tries. Indeed, for many consumers, brands are countries. They provide identity, security, status, and a passport to success, regardless of nationality. Only a handful of tetecom operators have global brands today. AT&T is reportedly the best known (see Figure 4). As monopo- lies, most telecom operators paid little attention to their brand name. Many incumbent carriers still have weak brands. It is the new entrants, often mobile carriers, that have made brand- ing a priority, although success has often eluded them too. A recent survey of European consumers by Andersen Consulting and the French company, lpsos Opinion, found that less than ten percent of users could name a telephone company outside their home market (e.g., only three percent of Germans named Figure 4. Branding the World: The Best Known Telecom Brands Note: In the latest edition of The World’s Greatest Brands (New York University Press, 1997), a quadrennial survey by Interbrand, an international leader in branding ser- vices, only one telephone company, AT&T, was ranked among the world’s 100 most popular brands. And even AT&T ranked 55th, well below such would-be competitors as M=crosoft (gth) and IBM (10th). Source: Ioterbrand (wwrw.interbrand.com), TeleGeography, © TeleGeography, Inc. 1~t98 TeleGeography 1999 © TeleGeography, Inc. 1998 Figure 5. Global Numbers for Global Markets However you label it--toll-free, free-call, or 800 service--universal freephone service is off to a slow start. In its first year (before ser- vices were launched) only around 16,000 numbers were registered with the ITU. Following the pattern of national freephone services, the system is more popular in some areas than others. Currently, around 90 percent of all national freephone numbers are registered in North Amedca, and a large majority of universal intemational freephone numbers (UIFNs) are roistered there as well. To receive a UIFN, a customer applies through a Recognized Operating Agency (ROA), which includes all companies recognized by a telecom regulatory authority. Once approved, the customer must implement the number in at least two countries within 90 days of reg- istration, or the number will immediately return to the pool of avail- able numbers. Unlike registerint~ an Intemet domain name, a cus- tomer pays a one-time fee (currently around 200 Swiss Francs/US$140) to register and to maintain the number rather than paying annual or bi-annual fees. UIFNs are roistered on a first-come first-served basis, with no extra charge involved for "vanity" numbers. If a duplicate application is received, pdority will be given if one of the applicants has the same number registered for national freephone service. Once registered, the number is completely portable, allowin{~ the cus- tomer to retain the same global number when chan0n~ service providers or adding new mutes. The cost of the incoming calls, vary- ing by carrier, range from about $I to $2 a minute in the Currently, universal freephone service is available in the following countries: Australia, Belgium, Denmark, France, Germany, Hong Kong, Japan, Netherlands, New Zealand, Norway, Switzerland, the United Kingdom and the United States. Other countries have agreed to participate in the program, but have not yet implemented the sys- tem on their networks. © TeleGeography, Inc. 1998 BT and only one percent of British consumers named Deutsche Telekom). By comparison, the competition in waiting--the Internet--has a growing roster of global brands. Netscape, Microsoft, AOL, Yahoo, IBM and even Sun (Java) probably have wider name recognition than do Deutsche Telekom or Teleglobe. Then there is the television industry, increasingly Internet savvy (visit www.msnbc.com), and with a large stable of global marks: ABC (Disney), NBC (GE), M’IV (Viacom), CNN (Time Warner). Telecom operators probably have two choices. They can pur- chase, or ally, with the strongest brands available in the broad- er communications industry. Or they can spend heavily to cre- ate their own brands. The financial value of such an exercise should not be underestimated. Based on a survey by Interbrand, Telecom New Zealand (TNZ) restated the value of its intangible assets in 1998 to include NZ$3 billion for brands --almost 40 percent of its total assets. Though some accoun- tants might find such a valuation suspect, TNZ’s approach reflects the post-national world of telecom competition. When brands trump nations, branded services will be the price of sov- ereignty. Various cross-market branding strategies are also becoming common. Excel (now owned by Teleglobe) and other telcos have already announced flat rate long distance ~elephone ser- vices on the home pages of various lnternet companies. Some observers suspect that the recent wave of big mergers in the U.S. (AT&T-TCI, GTE-Bell Atlantic, SBC-Ameritech) is ddven in part by the desire to develop national consumer brands for bundles of different telecom services. In Europe, efforts are afoot to create multinational marketing identities. For exam- ple, BT, Telenor, and Tele Danmark recently named their con- sortium Telenordia as part of their strategy to tackle the Scandinavian market. Advertising expenses are likely to rise as more companies launch branding (or rebranding) campaigns. But carriers may be helped by two proven pan-geographic marketing tools: access numbers and calling cards. Each let carriers stake out a virtual territory far beyond their own physical networks. Single or double digit access codes are particularly easy to brand. Hence, the spirited legal battle in France over the rules for allotting the 7 remaining single digit access codes for new carriers. Even where access codes are longer and less scarce, however, as in the U.S., a memorable number (e.g., call "10- 10-321" to save 50 percent) can even provide a well-known company with a leading brand (in this case, MCI). Put the number on a calling card and its reach is even greater. Most telephone access codes currently are, at best, national brands. Contrast the Internet, where a memorable address offers an instant global brand; you can click on it anywhere in the world and get to the same local site. One day soon more and more phone calls may start that way, with telephone num- bers and Internet addresses automatically translated by the service providers’ software. Branding global access codes will then be much easier (among other things). In the meantime, operators seeking a global access number must make do with the new "free phone" codes launched last year by the International Telecommunication Union (FLU) (see Figure 5). 2. Portals While strong brands will help some companies tap a wider mar- ket, others may go global by promoting new service gateways © TeleGeography, Inc. 1998 TeleGeography 1999 or portals. And while brands are all about creating one’s own identity, success in the portal business is likely to depend on partnerships. As Randall Rothenberg related in Wired magazine, "The theory that profits lie in serving a gateway through which consumers pass ... took off about last March--around the time Yahoo affil- iated with MCI." Disney then bought the rest of Starwave to become a portal and Lycos did a deal with AT&T to become one, too. In May, a little known enterprise called Zap Corporation, formerly in the seafood-processing market, made an unsuccessful $1.7 billion offer for Excite, one of the last independent, high-profile search engines. Portal pandemonium continued in June with NBC and Disney buying into, respectively, CNET’s Snap and Infoseek. AT&T and AOL could not agree on a merger, however, and AT&T turned around and bought one of yesterday’s choicest portals, the cable TV giant TCI. Watching these deals unfold ted Rothenberg to conclude that "becoming a portal involved no special skill set." And, he’s right, of course. In cyberspace, if you don’t have an attractive front door or a good location, build it. If your business doesn’t generate a lot of consumer traffic, find a partner or two or three who does. This is information architecture, after all. We’re not talking about national parks. Like other virtual artifacts, por- tals have no natural beauty; their landmark status and crowd drawing ability are all artificial (see Figure 7). For telecom operators, therefore, portals might be thought of as a generic term--a virtual construct rather than a finished product. Markets are becoming global and customers are demanding easy access to both voice and data services. New network gateways--let’s keep calling them portals for now--are likely to play a key role in helping carriers meet this twin chal- lenge. Portals also are a magic word for investors. Even though the stock market has cooled dramatically, a typical portal company such as AOL or amazon.corn is still trading at a much higher earnings multiple than is common for major international card- ers (see Figure 7). So how do telephone companies become portals? For the most part, they already are, but without the cachet. Think: access codes, directory assistance, free phone numbers and calling cards. Branding services like these can make any telephone keypad into an effective portal for many voice services. But to be a portal, carriers will need to find ways to leverage their current traffic streams. They must also duplicate their brands on-line so as to draw a wider audience and to offer more services (e-commerce). That is where partners come in. Telephone companies are just learning to manage web-based information and entertainment services. By joining forces with widely known lnternet sites they are likely to leverage the reach of all concerned. 3. Backbones To some readers the preceding discussion of brands and portals may seem beside the point. Brands don’t deliver services-- networks do. Without networks, portals are a dead end. Network facilities have always set the bounds for telecom mar- kets, and now that governments have stopped artificially parti- tioning networks along national or even sub-national lines, net- works will define whether a carrier is sovereign or not. Anyone who doesn’t realize that has got their OSI protocol stack upside down (see Figure 6). Services run on top of networks, not vice versa. This engineering perspective is compelling in many ways. And it enjoys as much favor, perhaps more, in the investment world as the sales (brands) and marketing (portals) visions discussed above (see Figure 8). The rise of Internet Protocol (IP) based networks presents tele- phone carriers with the biggest economic and technical chal- lenge since national voice networks were first knit together in the early 20th century. Packet-based communications net- works using IP technology are angling to take over the services offered by established telephone companies as never before. The new IP-centric or "next-gen" telcos are likely to have radi- cally lower cost structures than the incumbents (see Figure 9). And they can efficiently transport video and data services as well as voice traffic. Thus, as Randall Hancock and Charles Gerlach of the Gemini consultancy, Cambridge, Massachusetts, wrote in a recent White Paper: "This IP revolution is real. It is happening now, and it will fundamentally restructure the $500 billion global telecommunications industry over the next decade." The big question, of course, is where and how fast? Figure 6. The OS! Protocol Stack The Open Systems Intemonnection (OSI) standards provide a seven level design framework for communication networks so that equipment from different vendors can interoperate. The first three levels of the OS1 stack are the Physical, Data Link and Network layers, which roughly correspond to the network infrastructure and routing protocols. The last three layers are Session, Presentation and Applications which affect specific user services. The fourth level-- Transmission--typically provides a software interface between the top three and bol~om three layers. However, bright line boundaries between different OSI levels do not always exist [e.g., some Level 3 functions may be per- formed by a network at Level 2 or © TeleGeography, tnc, 1998 TeleGeography 1999 © TeleGeography, Inc. 1998 Figure 7. Virtual Places, Real Money The astonishing market value which portals now command may have taught the telecom industry a valuable lesson (see chart at right). Virtual places can have every bit the economic clout of physical places and sometimes more. This may seem illogical. We are accus- tomed to describing the places we know in terms of their physical attributes--the street where we live, the countryside we saw on hol- iday. Yet, places are always a mix of hard (physical) and soft (social) factors. The character of a neighborhood or a resort is determined not simply by its physical characteristics, but by our interaction with others and with the place itself. Likewise, the lntemet runs over thousands of kilometers of cable circuits and mydad computers in tens of countries, but it is nothing if not a social network. A new generation of geographers is trying to make sense of the power held by both virtual and physical places. Paul Adams, a professor at the State University of New York, Albany, argues that place is best understood as a function of communications and human action--that is, as a histodca! process. Adams (with Barney Wharf) recently co- edited a special issue on cybergeography for the American Geographical Review. He contends that the pattern of communica- tions defines the existence and particularity of place just as do walls, furniture or sight lines. Adams wdtes: "a communication system is to communicators as a place is to inhabitants." That may also help to explain why the spread of new communication systems, particu- Price/Earnings Ratios of Portals vs. Telcos 350 150 I00 # Yahoo,: f,, Telecom Italia AT&T Sep- Oct- Nov- Dec- Jan- Feb- Mar- Ap~’- May- Jum Jun- Jul- Aug- Sep- lady the lntemet, has been accompanied by so many geographical metaphors from "chat room" to "information highway" to "home page." © TeleGeography, Inc. 1998 The international phone network and the lnternet already con- nect. But IP network gateways to the PSTN are still relatively scarce. By and large, today’s Internet is a network within other networks, built from various carrier-owned transmission circuits and largely non-carder switches and terminals. It primarily transmits non-voice traffic. In contrast, although the PSTN pro- vides local dial-up gateways for Internet users, in most of the world the PSTN is still used primarily for voice traffic. The functional differences between the telephone network and the lnternet are breaking down, however. The main reason is data traffic. Demand for data communications is rising much faster than for voice, and new technologies make it ever cheap- er to convert any traffic stream into a digital (i.e., data) format. Thus, as data traffic becomes predominant--that probably hap- pened this year for some U.S. long distance carriers (see Figure l O)--the core transmission function of the Internet and the PSTN will converge and so will protocols, say network engi- neers. Both networks will be primarily data networks with IP or IP-compatible traffic and, consequently, they will also be direct competitors. International telephone services face the most immediate chal- lenge from IP services. Fax traffic accounts for at least 20 per- cent of international minutes on the switched network and on many trans-Pacific routes the share exceeds 50 percent. IP fax programs already are integrated into word processing and e- mail applications; they can also be downloaded from IP fax car- tiers, such as FaxSav or FaxWeb. While Internet fax remains a niche business for now, international IP services are quickly moving into the mainstream. The head-on challenge which 1P fax and later IP telephony pose for existing operators has led many of them to buy or build sim- ilar low cost, IP-friendly networks. Most incumbents already control the all important "last mile" of the network. Long haul IP backbone facilities have thus become the real geographic imperative. As the IP revolution gathers force, it is the reach and capacity of these networks that will define one carder’s sphere of influence vis-a-vis its competitors. For example, Sprint has announced an ambitious plan for a new broadband integrated data network (see Figure 10). That plan was apparently triggered, in part, by WorldCom’s pending acquisition of MCI, a deal which WorldCom said would provide it with an unrivaled network for "corporate" (read: data) cus- tomers plus an enhanced Internet backbone. Competition authorities later ruled that MCI had to divest its lntemet busi- ness as a precondition to the merger given the large lnternet backbone business WorldCom already had (i.e., UUNet). But the buyer of MCI’s Internet business, Cable & Wireless (C&W), was permitted to lease transmission capacity from the new MCI WorldCom; both MCI and WoddCom kept their existing net- works, which are considerable. WorldCom’s European fiber © TeleGeography, Inc. 1998 TeleGeography 1999 Figure 8. Next Generation Networks: Betting on the Future Performance of Next-Gen Network Company Market:’13cker IPO Date Global Crossing, Ltd. NASDAQ:GBLX Aug. 13, 1998 Equant NYSE:ENT July 20, 1998 IXC Comumcations, Inc. NASDAQ:IIXC July 2, 1998 Level 3 Communications, Inc. NYSE:LLL May 19, 1998 Global TeleSystems Group, Inc. NASDAQ:GTSG Sept. 26, 1997 RSL Communications, Ltd. NASDAQ:RSLCE Aug. 25, 1997 13west Communications, Inc NASDAQ:QWST June 24, 1997 COLT Telecom Group NASDAQ:COLTY Dec. 10, 1998 Stocks Since IP0 Offering Offer Price Sept. 17, 1998 $399M $22.00 $19.75 $541 M $27.00 $47.75 $89M $16.00 $30.06 $132M $22.00 $35.25 $222M $20.00 $29.50 $158M $22.00 $22.13 $297M $22.00 $32.50 $300M $18.10 $44.38 © TeleGeography, Inc, 1998 optic net (Ulysses) is now linked to the U.S. via the new 30 Gbps Gemini cable, and a WofidCom backed trans-Pacific cable (Southern Cross) is scheduled for service in 2000. A high speed (200 Gbps) IP backbone network is also at the center of the AT&T-BT joint venture for international services announced this July. The new venture will combine both com- panies’ existing facilities. Additionally, it will spend $1 billion over five years--in part to buy technology from others--to develop a "carrier strength" IP network linking 100 major cities. Corporate executives shepherding this new venture have also hinted that within a few years this new IP network could largely replace the two companies’ international switched net- works, and all the conventional interconnection and settlement fees that go with them. What AT&T and BT are now whispering has already been shout- ed by today’s upstart IP telephone operators. Yet few have cared to listen, perhaps because the implications are so unset- tling and also so uncertain. The international telephone busi- ness is just coming to grips with the revenue and pricing con- sequences of changing from a century-old reBime of standard, country-specific call termination charges or settlement fees to a system of network access fees which not only vary by country, but often by city and carrier too (see page 62). International IP networks now threaten to orphan this infant regime by adopting a new family of intemonnection arrangements based on lnternet traffic flows. What that will ultimately mean for the balance of payments between carriers or the bottom line is still anybody’s guess. Peering (sender keep all) arrangements have been popular among most Internet networks in the past. But larger networks now seem keen on promoting pay-as-you-go terms for all but a handful of other large networks with which they will peer pri- vately. And whereas international carriers have long shared the cost of transnational cimuits, each paying their way to a mid- point, there is no such rule for the Internet and no agreed mid- points. More alarming still, unlike international telephony, retail customers are typically billed for Internet services at a fiat monthly rate regardless of the duration or distance of the con- nection (though outside North America per minute local call charges often apply). In these cimumstances, those interna- tional operators willing to contemplate an IP network future seem brave indeed. Nevertheless, the AT&T-BT vision is a multi-billion dollar endorsement--some might say, gamble--on the direction being taken by the world’s IP-centric telephone companies. Two U.S. companies, Qwest Communications and Level 3 Communications, have attracted the most attention, although non-U.S, telcos such as Bell Canada, which has announced its own North American 1P backbone, have similar plans. Qwest has almost completed a 16,000 mile long distance fiber net- work in North America with more bandwidth~8 fiber pairs, each operating at 9.6 Gbps (OC-192)--than AT&T, MCI WorldCom, and Sprint now use combined. Over a fraction of this capacity, Qwest has begun selling flat rate, 7.5 cents a minute, long distance calls in the U.S. It has also sold or swapped capacity on its network to GTE, Frontier, and Teleglobe, among others. Qwest’s willingness to sell "dark fiber" to its competitors has been a ray of light for many smaller telcos which understand the IP network imperative but lack the resources to act inde- pendently. Most telecom regulators have recognized that resale opportunities, like interconnection fights, are crucial to the development of fully competitive telephone service mar- kets. But data networks historically have been able to pick and choose their network partners and resellers, and the network access which Qwest has offered its competitors thus far could well be denied downstream. Furthermore, other IP networks may set different policies. TeleGeography 1999 © TeleGeography, Inc. 1998 Like the new AT&T-BT venture, the eponymous Level 3 (recall the OSI stack) hopes to build its own high capacity IP-based network which can interoperate easily with the PSTN. Level 3 wants to deliver end-to-end services, not merely trunking capacity, as Qwest does today, and thereby capture the full cost advantages of its network. To do so, Level 3 has allied with major equipment vendors such as Cisco (which will also help switch Spdnt’s ION traffic), Nortel and Lucent. Level 3 and its allies are pushing Internet engineers and the ITU’s standards body to adopt IP Device Control (IPDC) soft- ware for gateway switches to control access to both IP and switched networks, thus creating a standard interface for voice and IP traffic exchange. In simplified terms, with IPDC, a net- work’s Internet and PSTN gateway would be under the unified control of a media gateway controller (MGC). It would use the telephone industry’s existing Signaling System 7 (SS7) to ana- lyze an incoming "call" and then apply IPDC software to instruct an Intemet or voice switched gateway, as appropriate, to carry the traffic. Every major telco supplier is anxious to gain an edge in installing such next-gen switching technologies even if that means buying up data networking suppliers to get there. Nortel bought Bay Networks and is now known as Nortel Networks, while Alcatel, one of Europe’s champions, merged Figure 9. Next Generation Networks Will Cost Less The data-centric long distance networks now being built by Sprint, MCI WoridCom, Level 3 and others are likely to have far lower delivery costs [70 percent lower, says Sprint) for most voice and data services. The reasons are: (I) a better performance/cost ratio for data switches and touters vs. conventional central office switches and (2) a 1,000 fold capacity increase for fiber optic networks since the late 1980s, Switching Costs For example, Level 3 CEO James Q. Crowe contends that IP packet switching has a better performance/cost ratio than Asynchronous Transfer Mode (ATM) or conventional class 4/5 PSTN switches, and the difference is growing. (ATM is a protocol for the high speed transmission of data packets.) To support this claim, Crowe cites research by Peter Sevcik, a senior associate at Northeast Consulting Resources Inc. (Boston). Sevcik has found that each successive new generation of switching technology cuts the performance/cost doubling time in half. That is, conventional central office switches double their performance/cost ratio every 80 months, according to Sevcik, while ATM switches do it in 40 months. Even better, packet switches and routers double their performance/cost ratio every 20 months. For example, Cisco Systems 7000 Series, introduced in 1993, can switch 45 megabits per second at a cost of around $320/Mb/s. Cisco’s 12000 Series, introducd last year, can switch 2.4 gigabits per second at a cost of around $30/Mb/s. Transport Costs New optical transmission technologies permit very large volumes of digitized voice or data traffic to be carried on a single strand of fiber optic cable by dividing the available bandwidth into multiple, frequency-specific channels of light. This technique, known as Dense Wave Division Multiplexing [DWDM), already permits carriers simultaneously to route over 1.5 million phone calls on a single fiber pair. The cost of routing calls is diminishing exponentially. In a few years, the capacity of a fiber pair is likely to be 15 times that or over .500 Gbps. DWDM will also permit similar very high bit rates on the next generation of trans-oceanic submarine cables, such as TAT-14, which will go into service in 2000 and beyond [see page 93). The per minute cost of carrying a voice call on such cables is minuscule. Economics of Alternative Switching Technologies 10,000,000 1,000,000 100,000 10,000 i 1,000 Pocket Sw~ch (Frame) Circuit Sw~tch (ATM) C~rcu~ Sw~ch (Telephone) .~o~ 1980 1985 1990 1995 2000 (log scale) 1~ Cost Trends in Optical Fiber Tranmission 45 Mbps Mbps ps 7 6bps e s 6bps 0.01 L 1975 1980 1985 1990 1995 2000 Source: Peter J. Sevcik, Senior Associate, Northeast Consulting Resources, Inc., Boston, sevcik@ncri.com. Originally in "Why Circuit Switching is Doomed," Business Communications Review, Sept. 1997. Source: AT&T data reproduced by the World Bank © TeteGeography, Inc. 1998 © TeleGeography, Inc. 1998 TeleGeography 1999 with Texas-based DSC Communications Corp. In addtition, Lucent (formerly an arm of AT&T) has bought several smaller datacom vendors too, and is actively promoting its own "voice- grade" switches for IP networking. The fear of Nortel et al is, if anything, more pronounced than their core customers: as IP networks handle ever more of the world’s telecom traffic, the revenue stream from manufacturing and installing today’s cen- tral office switches may fall off abruptly. The strategic implications of the IP network revolution are hard to understate. Ever rising volumes of data traffic, and the new economics of fiber optic cables and high-capacity data switch- es, make data-centric networks compelling even for smaller car- tiers. The next-gen facilities of a Qwest or a Level 3 may seem geographically limited today. But a clutch of new undersea mega-cables will let them, and their offshore cousins, reach around the globe tomorrow. In the short run, end-to-end IP backbones can be used by voice services for self-correspondence (to bypass accounting rates). Later, the same backbone capacity will be crucial to competing for new IP services, whatever they might be. The economic logic of next-gen networks also looks compelling for any carrier that must compete on price or service quality. Data traffic is a bandwidth "hog" and those carriers with the largest backbone networks are most likely to be able to make the transition from the Internet’s current "best efforts" packet delivery service to a 99.9 pement reliable carrier-grade model. Moreover, only 1P networks which can deliver similar grades of service are likely to be offered "private" sender-keep-all peer- ing arrangements. Networks which cannot meet the grade may end up paying by the packet for interconnection which, in turn, is likely to mean higher prices for customers or lower margins for the operator. I11. From Cyberspace To Outer~pace This essay began by considering the legacy of telecom policy reform and network investments on the economy--the so- called "network effect." Viewed from a distance though, say the mid 21st century, the most lasting legacy of today’s global telecom networks may not be the worldwide diffusion of a mar- ket culture but the germination of an interplanetary one. How so? Satellites and other space vehicles operate at such a speed and at such a remove from the earth that radio communication is always an essential part of the enterprise. Space to Earth links are one part of the infrastructure. So are the communication packages onboard every space probe. Global networks also are needed to link the wide array of ground stations used for com- mand and control functions, as well as for downloading data and images. That is actually where the real telecom story begins. Since the 1960s, successive advances in space communication networks have brought us face-to-face with the cosmos. Remember the Earth rise filmed by the Apollo astronauts on the Moon; the eerie familiarity of a rock-strewn desert relayed by the Viking lander on Mars; the wondrous haze of distant galax- Figure 10. Data Eclipses Voice---Will Revenue Follow? 8.5 80 7.5 7.0 6.5 6.0 5.5 5.0 4.5 4.0 3,5 3.0 2.5 Data Traffic Transition For U.S. Long Distance Industry (1988-2010) 2.0 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 1998 appears to be the year when the volume of data traffic eclipsed voice traffic for some U.S. long distance carriers. Although conclusive statistics are not yet available, this crossover was anticipated by an influential 1997 analysis of U.S. long distance traffic by two MIT researchers, Philip Mutooni (now with I.R Phusion Technologies, Inc.), and Dr. David Tennenhouse (see chart). In June 1998, Sprint seemed to confirm the MIT study in announcing that it would replace its current U.S. long distance network with a new broadband Integrated On-demand data Network (ION) for the simultaneous delivery of voice, video and data services. Overall, however, voice traffic on U.S. local exchan{~e and long distance networks is still estimated to be several times the volume of data traffic. See A. Odlyzko, "The lnternet and Other Networks: Utilization Rates and Their Implications." (wvvw.research.att.com/--amo/doc/networks.html) Source: Modeling the Communication Ne~work’s Transition to a Oata-Centric Model by Phitip Mutoeni and David Tennenhouse. Capacity f~gures derived from data on mix of trunk lines at selected AT&T and MCI points of presence (POPs), For details, see: http://ksgwww.harvard edu/i~p/iicornpol/Papers/Mutooni,htm. © TeleGeegraphy, Inc. 1998 TeleGeography 1999 © TeleGeographyo Inc. 1998 ies lensed by the Hubble telescope; and, just last summer, Pathfinder’s stunning new panoramas from Mars. As image after image from Pathfinder was posted on the web site of the National Aeronautical and Space Administration (NASA), it became one of the most popular lnternet destinations on Earth (see http://mars.ivv.nasa.gov). Millions of people looked to cyberspace for the best view of outer space. All this publicity may not be lost on future mission planners. Most space communication systems have been narrowly tai- lored to a mission’s scientific payloads. New space nets may be engineered with a broader purpose, especially as the number of interplanetary missions rises. For instance, the Jet Propulsion Laboratory (JPL) near Pasadena, California, a prime NASA con- tractor, already has an ambitious plan to bring the Internet to outer space by developing a more delay tolerant IP protocol to support high bandwidth communications. It also hopes to engi- neer a number of IP gateways in space that would relay com- munications from the earth to both planetary and deep space vehicles. The JPL scored something of a coup earlier this year when it appointed Vint Cerf, one of the lnternet’s founders, as a distinguished visiting scientist. As Cerf told the annual lnternet Society Conference (ISOC) in Geneva last July, the time Figure 11. Mars Calling: How NASA May Keep in Touch with the Red Planet Rover/Science Mars-Synchronous Telecommunications R~tay Satellite In Sltu Spacecraft Dedicated Lunar and Mars Su~nstwork Using Single or Arrayed Antennss Source: A, Co ben, Report of the 90-day Study on Human Exploration of the Moon and Mars, NASA, Washington, DC, November 1989, © TeleGeography, Inc. 1998 TeleGeography t999 is right for developing an interplanetary Internet because plan- ning for Pathfinder’s follow-on missions is already well advanced. Although Cerf’s speech may have seemed far-fetched to many participants, the lnternet’s ability to deliver Martian vistas to people’s desktop PCs seems to have sparked a broad new interest in space exploration. A month later, over 700 Mars enthusiasts from a dozen countries gathered in Colorado to found the Mars Society (www.marssociety.org). The founding convention was largely organized by Robert Zubrin, an inde- pendent astronautical engineer. His 1996 book, The Case For Mars, offers a blueprint for a comparatively low cost human mission to Mars by 2010. As the images relayed by future lan- ders reach us, Zubrin argues that "we will see Mars as truly another world, no longer a notion, but a destination." Or as other space activists put it, "Space is a place, not a program." The Mars Society is intent on raising private funds to get to the red planet, if governments decline to underwrite the mission. Some supporters have even hinted at selling media rights, Olympics-style. And, yes, ever so quietly, a few big name tele- corn brands have started to compete for the fight to connect the first telephone call on Mars (see Figure 11). Our review of interplanetary communications continues at the back of this volume. For those interested in more local net- works, a new edition of our best selling wall map on world telecommunications will be published in January 1999. As always, we welcome your comments and suggestions. ~i~ Gregory Staple is President of TeleGeography, Inc. and a part- ner in the Washington, DC communications law firm Koteen & Naftalin. He can be reached by email at gstaple@telegeogra- phy.com. For Further Reading The Networked Economy The economic impact of global networks was mooted a decade ago in The Borderless World: Power and Strategy in the Interlinked Economy by Kenichi Ohmae (Harper Business, 1990). For more recent accounts, see The Death of D~stance by Frances Cairncross (Harvard Business Press, 1997); Market Unbound, Unleashtng Global Capitalism by Lowell L. Bryan (John Wiley & Sons, 1996), and Money/Space: Geographies of Monetary Transformation, edited by Andrew Leyshon and Nigel Thrift (Routledge, 1996]. For background, see Gregory Staple, "TeleGeography and the Explosion of Place, Why the Network That is Bringing the World Together Is Pulling It Apart," in E.M. Noam and A. J. Wolfson (Editors), Globalism and Localism in Telecommunications (Elsivier Science, B.V. 1997). Recent U.S. proposals to stabilize the global economy were outlined by President Clinton in a September 14, 1998 speech to the Council on Foreign Relations, New York (www2.whitehouse.gov/MH/New/html/ 19980914-520.html). See also the October 6, 1998 address by Michael Camdessus, Chairman, International Monetary Fund, to the Fund’s Governors, www.im f.orgiexternalin p/speeches/1998/ 100698.htm. For an instructive review of the debate on delinking weak economies from the global market by, for example, restricting capital inflows, see Martin Wolf, "Wisdom of Free Flows Questioned," The Financial Times, October 2, 1998; and the record of the October 2, 1998 IMF Economic Forum at www.imf.orgJexternal/np/tr/1998/ tr981002a.htm. Next Generation Networks The impact which IP-centric networks will have on circuit-switched tele- phone companies is profiled in a June 1998 White Paper by Randall S. Hancock and Charles L. Gertach, "The IP Revolution," Gemini Strategic Research Group, Cambridge, Massachusetts. For an engineering view, "The Cook Report On lnternet" (www.cookreport.com), published monthly by Gordon Cook, is invaluable. Also of interest are the April 1998 issue of tele.com on the convergence of television and the Internet (www.tele.com.com/0498/Features/tdc0498covers.html), and Andrew Odlyzko’s papers on the data/voice network crossover (www.research.att.com/tildeamo/doc/networks.html) Interplanetary Communications The wireless modems which relayed Pathfinders Martian vistas over 120 million miles are profiled in "Mars: 2400 Baud Away," by Todd Wallock, Network World, July 14, 1997 (www.nwfusion.com). For background on future mission requirements, see Michael A. Jordan, Eric O. Basgues, and Patricia E. Gould, "Communications Needs For Mars Exploration: Operational Implications," in Thomas R. Meyer, The Case For Mars IV.. The International Exploration of Mars, (American Astronautical Society, 1997). A profile of NASA’s current Deep Space Network for interplanetary communications can be found at http://deepspace.jpl.nasa.gov. TeleGeography 1999 © TeleGeography, Inc. 1998 Carriers TeleGeography 1999 © TeleGeography, Inc. 1998 International Carrier Evolution Overview: Boom Times For New Carriers As of July 1998, over 1,000 facilities-based international carriers were operational worldwide. Two years before, there were less than 500 (see Figures 1 and 2). These car- riers, old and new, are all authorized to own and to operate international transmission facilities. How did they all arise? Some owe their existence to politics--the abolition of monopolies. Others were helped more by new technologies (e.g., call-back switchesl that made it easier for new entrants to gain a foothold in the market. Together, these changes have made it possible for carriers to service niche markets and later to begin building their own networks. When examined country-by-country, competition has occurred on different schedules. On a regional basis though, change seems to come in waves: North America With over 200 new international carri- ers entering the U.S. market between July 1997 and July 1998, the U.S. leads the world in new carriers. This is due both to the streamlined authorization process in the U.S. and America’s status as the world’s premier traffic hub (not to mention a strong domestic economy). Soon the Canadian market will expand as well, as Teleglobe’s monopoly on overseas calls will end later this year. For more analysis, see "Emerging North American Carriers" on page 47. Europe Spurred by the 1998 market-opening agenda of the European Union (EU), more than 400 facilities-based carriers now compete to provide international service. Competition is also taking hold in non-EU states: Switzerland will soon have at least 20 carriers; Iceland and Norway offer fully open telecom markets; and many Eastern European markets are preparing for liberaliza- tion. But the greatest growth has occurred in the U.K., where full competition was introduced in 1996. As of July 1998, there were more than 140 authorized inter- national operators in the U.K., including both full facili- ties-based carriers and international simple resale (ISR) carders (see below for more on ISR). Asia/Pacific Due to regulatory constraints and ongoing financial crises, competition has lagged in Asia. The notable exceptions are Australia and Japan, which Figure 1. The International Carrier Boom Global Growth of International Carriers, July 1995 July 1998 1200 1042 1000 600 200 367 586 0 1995 1996 1997 1998 Regional Growth of Authorized International Carriers, 1995-1998 Europe ~~ Asia-Pacific Africa ¯ July 1998 July 1997 :;July 1996 July 1995 too 200 300 400 Number of Competitors 500 Note: Figures include all carriers authorized to provide ~ntemational facilities-based service or international simple resale, Source: Adapted from TeleGeography’s 3-volume directory of international carriers, New International Carriers. © TeleGeography, Inc. 1998 © TeleGeographyo Inc. 1998 TeleGeography 1999 Rank Country July 1998 July 1997 July 1996 July 1995 1. United States 393 175 115 65 2. United Kingdom 144 100 65 35 3. Germany 32 1 1 1 4, France 29 1 1 1 5. Russia* 29 25 21 18 6. Netherlands 23 3 1 1 7. Canada 21 21 19 18 8. Switzerland 21 1 1 1 9. Mexico 15 9 16. Australia 14 10 11. Austria 13 1 12. Japan 13 3 13. Sweden 13 11 14. El Salvador 12 1 15. Philippines 12 9 16. Belgium 11 1 17. Denmark 11 9 18. New Zealand 11 9 1 8 1 3 9 1 9 1 7 9 1 8 1 3 7 1 9 1 1 2 19. Chile 20. Italy 21, Spain 22. Finland 23. Norway 24. Ireland 25. Malaysia 26. Hong Kong 27. Colombia 28. Dominican Republic 29. Israel 30. Kazakhstan 31. Korea 32. Luxembourg 9 9 9 8 7 5 5 4 3 3 3 3 3 3 9 1 1 8 1 3 5 1 1 3 3 1 2 1 9 1 1 8 1 3 5 1 1 3 1 1 2 1 9 1 1 5 1 1 4 1 1 3 1 1 2 1 33. Bermuda 34. Brunei 35. China 36. Ecuador 37. Ghana 38. Indonesia 39. Monaco** 40. Portugal 41. Uganda 42. Ukraine 2 2 2 2 2 2 2 2 2 2 2 2 2 1 1 2 2 2 1 2 2 2 2 1 1 2 1 2 1 2 1 2 2 1 1 2 1 2 1 2 CAGR 95-98 (%) 82.2 60.2 217.5 207.2 17.2 184.4 5.3 175.9 146.6 20.5 135.1 63.0 22.9 128.9 10.1 122.4 122.4 76.5 0.0 108.0 108.0 17.0 91.3 71.0 7.7 58.7 44.2 0.0 44.2 44.2 14.5 41 26.0 0.0 0.0 26.0 26.0 0.0 26.0 0.0 26.0 0.0 Note: Figures include all carriers authorized to provide facilities-based international service or international simple resale as of July 1 for each year. Estimates include carriers authorized to provide service m only certain mumcipalities. ** Although more than one carrier operates in the country, carriers do not dzrectly compete. Source: Adapted from TeleGeography’s 3-volume directory of international carriers, New International Carriers, © TeleGeography, Inc. 1998 TeleGeography ! 999 © TeleGeography, Inc. 1998 Map of International Carrier Competition International Carrier Competition ¯ More than 10 carriers i~ Between 2 and 10 carriers Monopoly All figures current to July 1998. Data include all carriers authorized to provide facilities-based international service in each country. © Te]eGeography, Inc 1998 opened their markets to full competition in 1997 and 1998, respectively. Both countries now have more than a dozen carriers competing to provide international ser- vices. To read more about recent developments in the Asian market, see "International Telephony & Network Expansion" on page 41. Latin America Chile and the Dominican Republic, once the only two competitive markets in this region, have been joined by Mexico, Ecuador, Colombia, and El Salvador, which together have hatched almost 30 new carriers over the last year and a half. Soon the biggest Latin market, Brazil, will have competition, with a new long distance concession for sale in late 1998. To read more about Latin America, see "International Long Distance in Latin America" on page 44. The proliferation of carriers is far from over. The February 1997 World Trade Organization (WTO) agreement on basic telecom services set market opening schedules for over 50 countries, many of which have yet to go into effect (for a summary of commitments, see "Map of WTO Commitments" on page 38). If competition continues apace, despite inevitable consolidation, we expect more than 2,000 facilities-based carriers will compete to carry international calls by 2002. Heavy and Light Carrier Models Without question, the rapid growth of international carriers is impressive. But to lump all carriers together is a bit mis- leading. The truth is there are very few newcomers that can be directly compared with the incumbent former monopo- lies. While new companies and old both have similar regu- © TeleCeography, Inc. 1998 TeleGeography 1999 Nalfierlands United Belgium ~o~embourg Switzerland Au$~’ie Yugoslavie ~j/ Oominicen Republic Po~ugel Z"" Ecuador Chile Although Rostelecom holds a legal monopoly in Russia, more than two dozen carri- ers provide international service through agreements with municipalities. Source: Adapted from TeleGeography’s 3-volume directory of international carriers, New International Carriers. © TeleGeography, Inc. 1898 latory authorizations, many of the challengers exist (and thrive) by reselling, repackaging, and reprogramming the offerings of established carriers. In past editions of TeleGeography, we termed this the "Light" carrier model. The "Heavy" carrier model, by comparison, typically involves the construction or purchase of substantial network facilities. That can be a costly enterprise, of course, and thus the Heavy carrier model is more typical of established telephone companies. New market entrants, on the other hand, tend to rely on Light carrier strategies either as a first step toward creating a network or as an end in itself. Although the distinction between Heavy and Light carriers is helpful, it can also be deceptive. The line between Light and Heavy is most noticeable in markets which lack facilities- based competition. In competitive markets, even former monopolies engage in Light carrier practices such as resale, call-back, refile, and other forms of untraditional call rout- ing. The Old Regime To better understand the current international service mar- ket, it is useful to look at the origin of the Light carrier model. The old regime governing international telecommu- nications was characterized by state-owned national monopolies, typically Post Telegraph & Telephone (Frl-I-) operators. Historically, these Heavy carriers did not operate directly in other countries but provided international service by connecting their cable and satellite "half-circuits" with the matching facilities of a foreign carrier (see Figure 3, "Typical Infrastructure of an International Call"). The PTTs compensated one another by a 50/50 division of a whole- TeleGeography 1999 © TeleGeographg, Inc. 1998 International Cable/Satellite Half-Circuits sale facilities charge, or accounting rate, per minute of traf- fic. Each carder recouped this charge, plus a mark-up, which might be 150 percent or more, through its own inter- national tadff. For a full explanation, see "The Accounting Rate Regime in Transition" on page 62. The first tdckle of Light carder services came in the 1980s when companies were authorized to resell the international switched services of established carders. These companies focused on their home markets and profited by passing through to subscribers a portion of the volume-based dis- count available from facilities-based carders. Later, facili- ties-based competition began when a few new carriers, such as MCl in the U.S. and Mercury in the U.K., were authorized to build their own international transmission facilities. Throughout the 1980s, however, competition was limited and the options for customers sparse. The old regime remained intact, albeit with threats looming on the horizon. Alternative Call Routing At the beginning of the 1990s, new technology, rapidly falling transmission costs, and the rising demands of multi- national users brought a challenge to the old regime. In addition to using basic service resale, carriers began to explore a number of new call delivery methods, all of which departed from the old regime’s method of exchanging inter- national traffic (see Figure 4, ’~dternative Call Delivery Methods"). In each case described below, the new methods exploit, or bypass altogether, the accounting rate system: International Simple Resale (ISR) ISR bypasses per- minute charges of the accounting rate system by using private lines rather than the public switched telephone network (PSTN) to transport calls. Broadly defined, ISR refers to the wholesale purchase or lease of international private line (IPL) capacity from a facilities-based carder which is then resold to customers for switched telephone service. (An IPL is a leased circuit whose capacity is ded- icated to the lessee.) Telephone service is typically pro- vialed by interconnecting the ISR carrier’s circuits to the PSTN at one or both ends. ISR was first authorized in the U.S. and U.K. in 1992 and soon after gained acceptance in Australia, Canada, and Sweden. Today, ISR is legal in over 25 countries and can be used to carry calls either directly between two countries or from an originating country to a hub country and then to its final destination. In ! 997, ISR accounted for an estimated two percent of global traffic (around 1.5 billion minutes), most of which was carried on the U.S-Canada and U.S.-U.K. routes. Call-back With a small investment in hardware and soft- ware, call-back carders can change the direction and thus the cost of an international call. A basic call-back arrangement works like this: A pre-subscdbed customer in country A dials a call-back operator in country B and, after a certain number of rings, hangs up. (The customer does not pay for this initial uncompleted call.) The call- back company then uses its switching software to initiate a call to the subscriber, and when the caller in country A answers, she receives a dial tone from the call-back com- pany’s switch. The customer can then place a call in country B (e.g., the U.S.) orto a third country (for details on the economics of call-back, see Figure 5, "The Call- Turnaround Effect"). Call-back companies, such as USA Globalink and Telegroup in the U.S., originated approxi- © TeleGeographyo Inc. 1998 TeleGeography 1999 Call-back % 1. Customer dials international access ~:~ number, waits for ring, then hangs up. 2. Carrier calls customer from call-back hub country. 3. Customer answers phone and has dial- tone from call-back country. ¯ 4. Customer then calls a final destination inside or outside the call-back country. Refile 1. Customer dials international number. 2. Originating carrier sends call to refile carrier. 3. Refile carrier terminates call in refile hub country (U.K.). 4. Call is re-originated and sent to final destination via refile carrier. Internet Telephony (phone to phone) 1. Customer dials international number. 2. Call is routed to carrier’s server via PSTN; server converts analog voice call to IP format and sends to Internet gateway. 3. Call travels over the Internetto a gate- way in terminating country; call is re- converted to analog format. 4. Call is delivered to its destination via PSTN in terminating country. Note: The d~agrams above are not drawn to scale and call paths are drawn for =llustratJve purposes only. Call-back, refile, and IP telephony can all take different forms depending on origin, destination, technology, and company. IP telephony, for example, =s also possible beWveen t~vo personal computers (PCs) or between a PC and a telephone-based destination. Source: TeleGeography, Inc. © TeleGeography, Inc. 1998 TeleGeography 1999 © TeleGeography, Inc. 1998 Figure 5. The Call-Turnaround Effect International telephone traffic is a bit like water; it always tends to follow the path of least resistance. Other thin~s being equal, the direction of traffic will follow price differentials in the same way that flows of water reflect underlying gradients. For international telephone calls, there are really two prices: a retail price paid by consumers and a wholesale price a~eed by the opera- tars providing the service. Historically, thanks to the accounting rate system, there was effectively no gradient in the wholesale pdce because the same rate (the accounting rate) was applied in both directions. Thus, insofar as there was a pdce differential, it was in the prices charged to end-users (the collection charge) and the mark-up that this represented over the accounting rate. In competitive mar- kets with significant economies of scale, such as the U.S., the margin between the retail pdce and the wholesale pdce tended to be lower than in other countries, so that marginally more traffic originated from the U.S. than from other countries. In the early 1990s, two things happened to change that picture. First, computer technology became available which made it easier to reverse the direction of a call, through call-back, calling cards or country-direct services. Second, wholesale carriers in the U.S. began selling outbound capacity at rates either at, or just below, the settle- ment rate. They were able to do this because a bizarre U.S. regula- tion-proportionate return of traffic--meant that they could afford to lose money on outbound traffic in order to gain proportionately more return traffic and the associated per minute settlement payments. Thus proportionate return created an artificial gradient in the settle- ment rate on the U.S. route which made it relatively more profitable to terminate traffic in foreign countries. As a result of these devel- opments, call-turnaround is now a multi-billion dollar industry (see charts below). Developing countries have made angry-sounding noises about call- back and many of them have tded to ban it. But the reality is that by reversing the direction of traffic from poor countries, call-back sends developing countries more settlement payments. For a coun- try such as India, call-turnaround probably generated around 200 million minutes of traffic in 1997 and it contributed to India’s net set- tlement in-payment of US$517 million from the U.S. that year. But what would happen if a real gradient were created in the settle- ment rate? What would happen if India charged $0.23 per minute to land traffic while U.S. carders charged only $0.07 to terminate traf- fic? This proposition is not as far fetched as it may seem, because even thou{~h India is a member of the WTO, and therefore eligible to enter the U.S. market, it has not agreed to open its market to foreign carders. Hence, because it may soon become more profitable to ter- minate traffic in the U.S. than in India, the direction of call-turn- around may be reversed. Even if one ignores proportionate return for the moment, which the Federal Communications Commission will soon abolish on many routes, a switch located on Indian territory would be able to offer U.S. residents a rate only slightly above $0.07 per minute to call India whereas U.S. based carders could only com- pete at rates above $0.23 per minute. Of course, the Indian opera- tor offering the call-back service would have to make a net settlement payment towards the U.S., but this should be easily covered by its collection charges to the U.S. Perhaps those developing countries which are currently eager to ban call-back ought to think a little more seriously about this market opportunity before foreclosing their options. This box was contributed by Dr. Tim Kelly, Head of Operations Analysis at the ITU. The views expressed are his own and do not necssarily reflect the opinions of the ITU or its membership. Traffic on U.S.-Hong Kong Route, 1988-1997 700 U.S. to J around 200 H.K. 1988 1989 1990 1991 1992 1993 1994 1995 1998 1997 Traffic on U.S.-India Route, 1988-1997 U.S. to/ 200 ~" Callturn- around 100 Inl~]a to U.S. 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 Note: Call turn-around figures are estimates based on change in traffic balance on each route. Source: TeleGeography, Inc. and International Telecommunication Union, © TeleGeography, Inc. 1998 © TeleGeographyo Inc. 1998 TeleGeography 1999 mately 2 billion minutes of call-back traffic in 1997, accounting for 2.5 percent of global traffic. Refile Refile is a form of indirect call routing which, like call-back, seeks to arbitrage non-cost-based differences in wholesale call prices to turn a profit. Refile generally is profitable for a carrier if it is cheaper to send a call from country A to country C via country B because the cost of paying the wholesale landing fee from A to B plus the fee for B to C is cheaper than paying to send the traf- fic from A to C directly. Refile thus involves two calls; one to the refile country and another, re-originated call, to the final destination. Unlike traditional transit traffic (A to C via B), however, the terminating carrier is unaware of the call’s true point of origin and has not given its con- sent. Thus the term "traffic smuggling" has been used to describe refile. Although refile is difficult to measure due to its routing, we estimate that five percent of the world’s traffic (4 billion minutes) was refiled through third coun- tries in 1997. lnternet Although the routing methods described here may decline as accounting rates become more cost- based, one new method should flourish: the global Internet. In combination with privately-owned interna- tional lines and gateways, some new carriers are avoid- ing accounting rates by pushing traffic onto the Internet, the same network that carries web pages and email mes- sages. For a more detailed discussion of lnternet tele- phony issues, see "The Global Internet: A Primer" on page 112. Multinational Carriers The element which most clearly marks the difference between the old regime and the new is the disintegration of national boundaries which once dictated the direction of traffic flows and facilities ownership. Due in part to com- mitments made in the VVTO agreement, operators may now hold stakes in foreign international carriers in over 70 coun- tries. In addition, over 20 countries now allow foreign car- riers to open wholly-owned operations on tuff once reserved for national carriers only. In many countries, today’s carri- Figure 6. International Carrier Affiliates of Cable & Wireless I Nor~ Ameoca 20 I V,tacom Corp, 50 IPnvateTransatianhc Telecom (Cabte) Cable &Wireless Inc (USA) 49 TSTT (Tr,mdad & Tobago) 180 I C&WBermuda 100 i C&W Ascens=on is! 180 I C&W Falklands St. K=tts & Nev=s Telecornmunlcatlons 80 I C&W gormmca Cable & Wireless plc ) I I Europe /M,dd,e East, India ~and Af .... .~_~ C&W (Schwe=z) AG Sw=~zer]and I C&W Europe (Belgium) 180 I C&W Russ,a r 65 I Jama=ca Dig=por~ international Ltd, .~ C&W (Wl) Ltd (Sa=nt Lucia) _~ c&w(wl) Ltd (St V=ncent & the Grenadmes) .~ C&W (Wl) Ltd (Antigua & Barhuda) C&W (Wl) Ltd (Vlrg=n Islands) 13 1 I Bezeq (israel) ._~ Sakhahn Telecorn (Rusem) 50 Nakhodka Telecom (Russm) ~J I TOJ(J ...... 49 I caw Panama 180 I CaW West Indies 180 I C&WCaymanlsl 85 BET(Barbados) i caw Grenada Asla/Pac=fic 1758 I IDC (Japan) _.~ CWJ CSL(Japan Type 2 carrier) ) Note: Figures include alI international carriers ovcned ~n whole or in part by Cable & Wireless plc. Source: Adapted from TeleGeography’s 3-volume directory of international carriers, New/ntematlona/Carriers, © TeteGeography, lnc, 1998 TeleGeography 1999 © TeleGeography, Inc. 1998 Figure 7. Selected Multinational Carrier Authorizations Parent Carrier Headquarters FaciliCom International, LLC United States Hermes Europe Railtel BV Belgium Long Distance International, Inc. Primus Telecommunications Group, Inc. United States United States Wholly-Owned Affiliate Affiliate Market FCI Austria Austria FCl Belgium Belgium FCI Finland Finland FCI France France FC] Italy Italy FCI Benelux BV Netherlands FCI Norway Norway FCI Switzerland Switzerland FaciliCom International Ltd. United Kingdom Hermes Europe Railtel (France) SARL France Hermes Europe Railtel (Germany) GmbH Germany Hermes Europe Railtel BV Netherlands Hermes Europe Railtet Switzerland Hermes Europe Railtel (UK) Ltd Umted Kingdom LDI Denmark NS Denmark LDI (France) France LDI (Germany) Germany LDI (Italy) Italy LDI (Sweden) Sweden Long Distance International Ltd. Switzerland Long Distance International Ltd. United Kingdom Primus Telecommunications Pty Ltd. Australia Primus Telecommunications, Inc. Canada Primus France France Primus Telecommunications Deutschland GmbH Germany Primus Netherlands Netherlands Primus Telecommunications Ltd. United Kingdom RSL Communications, Inc. United States RSL COM Australia Pry Ltd. Australia RSL COM Danmark A/S Denmark RSL C0M Finland 0y Finland RSL C0M France SA France RSL COM Deutschland GmbH Germany RSL COM Japan KK Japan RSL CGM Nederland BV Netherlands RSL COM Sweden AB Sweden RSL C0M UK Ltd. United Kingdom Teleglobe Inc. Canada Teleglobe Australia Pty Ltd. Australia Teleglobe France SAS France Teleglobe GmbH Germany Te~eglobe Italia SpA Italy Teleglobe Japan Inc. Japan Teleglobe BV Netherlands Telegtobe Norge AS Norway Teleglobe International (UK) Ltd. United Kingdom Teleglobe USA Inc. United States Viatel, Inc. United States Viatel Belgium SA Belgium Viatel SA France Viaphone GmbH Germany Viatel SRL Italy V=atel Global Communications BV Netherlands Viatel UK Ltd. United Kingdom MCI WoddCom, Inc. United States WofldCom Australia Australia WorldCom France France WorldCom Telecommunication Services GmbH Germany WorldCom Ireland Ireland WorldCom SpA Italy WorldCom Japan Inc. Japan WorldCom Netherlands Netherlands WorldCom AB Sweden WorldCom International, Ltd. United Kingdom Note: Affihates include all wholly owned affiliates authorized to provide famines-based international serv=ces as of July 1998. Affiliates author=zed to provide inter- natqonat simple resale (ISR) service are also listed. All parent companies are also authorized to provide service in their home markets. Source: Adapted from TeteGeography’s 3-volume d~recto~ of internat~ona~ carriers, New International Carriers. © TeleGeography, Inc. 1998 © TeleGeography, Inc. 1998 TeleGeography 1999 Figure 8. International Carrier Mergers and Acquisitions, Sept. 1997-0ct. 1998 Buyer Country Target Country Date Value (millions) Advanced Comm. Group U.S. Feist Long Distance, Inc. U.S. May 98 Undisclosed U.S. Firstel, Inc. U.S. May 98 Undisclosed Ameritech Corp. U.S. Tele Danmark A/S (34.4%; grown to 42.2%) Denmark Jan 98 $3,200 AT&T Corp. U.S. Tele-Communications Inc. U.S. Pending $45,000 U.S. Telepor~ Communications Group U.S. Jul 98 $11,000 Bell Atlantic Corp. U.S. GTE Corp. U.S. Pending $66,600 British Telecommunications plc U.K. Binariang Bhd (33.3%) Malaysia Pending $436 Call-Net Enterprises Inc. Canada Fonorola Corp. U.S. Jul 98 $1,220 China Telecom China Hang Kong Telecommunications Ltd. (7.8%) Hang Kong Feb 98 $1,184 China Hang Kong Telecommunications Ltd. (9.5%) Hang Keng Apr 98 $2,232 Deutsche Telekom AG Germany France T~l~com (2%) France Aug 98 $1,300 Esprit Telecom Group plc U.K. IMS Telecom Netherlands Nov 97 Undisclosed Excel Communications, Inc. U.S. Telco Communications Group U.S. Oct 97 $1,200 FaciliCom International, LLC U.S. TeleOne Ltd. Finland May 98 Undisclosed France T~l~com France Deutsche Telekom AG (2%) Germany Aug 98 $1,670 Global Crossing Ltd. Bermuda Neptune Communications Corp. U.S. Jul 98 Undisclosed Inter-Americas Comm. Corp. U.S. lusatel Chile SA Chile Dec 97 $7 Japan Telecom Co. Ltd. Japan International Telecom Japan Ltd. Japan Oct 97 Undisclosed Kokusai Denshin Denwa Co., Ltd. Japan Swiftcall UK U.K. May 98 Undisclosed Japan Teleway Japan Corp. Japan Pending $424 MCI Corp. U.S. Embratel SA (20% equity; 51.8% voting) Brazil Jul 98 $2,260 News Corp. U.S. PLD Telekom Inc. (38%) Russia May 98 $81 Primes Telecommunications, Inc. U.S. TresCom International, Inc. U.S. Feb 98 $135 U.S. USFI, Inc. U.S. Oct 97 $12 Qwest Communications Intl. Ltd. U.S. LCI International Telecom Carp. U.S. Jun 98 $4,400 RSL Communications, Ltd. U.S. Callcom AG (78.5%) Switzerland Dec 97 $2 U.S. European Telecom SA (90%) Belgium Jan 98 Undisclosed U.S. Newtelco Telekom AG (90% stake) Austria Sep 97 Undisclosed U.S. Westel Telecommunications, Inc. Canada Pending $38 U.S. Westinghouse Communications U.S. Apr 98 $90 SBC Communications, Inc. U.S. Ameritech Corp. U.S. Pending $65,400 U.S. SNET U.S. Pending $5,700 Scottish Telecommunications Ltd. U.K. Demon Internet Ltd. U.K. Apr 98 $110 STAR Telecommunications, Inc. U.S. PT-I Communications, Inc. U.S. Pending $195 SwissCom Switzerland UTATelekom AG Austria Jul 98 Undisclosed Technology Control Services Inc. U.S. Interglobe Telecom. International, Inc. U.K. Mar 98 $17 Teleglobe Communications Corp. Canada Excel Communications, Inc. U.S. Pending $3,100 Teleport Communications Group U.S. ACC Corp. U.S. Apr 98 $1,100 Viatel, Inc. U.S. Fiat Rate Communications, Inc. U.S. Mar 98 Undisclosed U.S. Jazztel (7.5%) Spain Jun 98 $6 WorldCom, Inc. U.S. MC! Corp. U.S. Oct 98 $41,800 U.S. TCL Telecommunicat=ons Ireland Oct 97 $29 WorldPort Communicat=ons, Inc. U.S. AACR Dam. Rap. Pending Undisclosed U.S. EnerTel NV Netherlands Jun 98 $110 Note: ~1 transactions revolve 100 percent of targe’(s shares unless otherwise noted. Dates indicate time of legal completion of transaction rather than announce- ment of intention to merge or to acquire. Source: Adapted from TeteGeography’s 3-volume directory of international carriers, New International Carriers. © TeleGeography, Inc. 1998 ers may fully or partly own facilities on both ends of an international cal!. This type of end-to-end facilities-based service model is still the exception rather than the rule, but it is seriously affecting investment decisions and making accounting rates less and less relevant (when a carrier owns facilities at both ends, settlement rates became an internal transfer payment between two affiliates). To take advantage of these options, many new carriers are establishing beachheads in multiple markets---either by establishing a new affiliate or by purchasing the operations of an existing carrier (see Figure 7, "Selected Multinational Carrier Authorizations" and Figure 8, "International Carrier Mergers"). Companies like Primus, RSL, Esprit, and LDI represent the first wave of new "multinational" carriers. They may have a national headquarters, but they are essen- tially stateless carriers. Europe’s unique economic geogra- phy has made it a prime target for the multinational opera- tions of such new carriers, including COLT, Facilicom, Hermes, and Tele2. TeleGeography 1999 © TeleGeographyo Inc. 1998 Country Australia Bahamas Bulgaria Croatia Czech Republic Denmark Ecuador Egypt Estonia Finland France Breece Honduras Iceland India Japan Jordan Kenya Korea Kuwait Macedonia, FYR Malta Moldova Mongolia Netherlands Nicaragua Oman Pakistan Papua New Guinea Poland (a) Romania Saudi Arabia Singapore Sri Lanka Switzerland Taiwan Tanzania Thailand Thailand Turkey (b) Uganda Ukraine Company Telstra BATELCO BTC HPT SPT Telecom Tele Danmark Emetel (Andinatel Pacifictel) Telecom Egypt Estonian Telecom Sonera (Telecom Finland France T~l~com OTE Hondutel Iceland Telecom Ltd. VSNL N’R" JTC KPT Korea Telecom KCC Macedonia Telecommunications Maltacom Moldtelecom MTC PTI" Telecom BV Enitel BTO PTCL Telikom PNG TPSA RomTelecom SA Saudi Telecom Singapore Telecom Sri Lanka Telecom Limited Swisscom Chunghwa Telecom Tanzania Telecom CAT TOT T[Jrk TelekomBnikasyon Uganda Telecom Utel Gov’t Share Sale 1996 Traffic Share to Sell Date (million min.) 66% 66% 1999 9,452.0 100% 49% 1999 56.7 100% 51% 1999 83.5 100% 25% 1999 242.4 51% 51% 1999 210.4 9% 9% 1998 573.0 100% 35% 1998 48.2 100% 20% 1998 113.0 100% 49% 1999 58.5 100% 20% 1998 219.2 75% 12% 1998 3,116.0 75% 10% 1998 516.0 100% 49% 1998 41.6 100% 51% 2000 32.5 65% 11% 1999 384,2 65% n.a. 1999 1,710.0 100% 40% 1998 74.6 100% n.a. 1999 20.8 71% 33% 1999 520.0 100% 51% 2000 140.7 100% 33% 1998 51.0 100% 40% 1998 31.7 100% 37% 1998 50.2 100% 100% 1998 2.4 62% 62% 2001 1,534.0 100% 40% 1999 29.4 100% 40% n.a. 62.6 88% 26% 1999 77.0 100% n.a. 1999 24.5 100% 25% 1998 437.2 100% 35% 1998 91.5 100% 80% 2000 584.0 88% 12% n.a. 942.0 62% 62% 2001 29.3 100% 34.5% 1998 1,936.0 100% 75% 1999 674.0 100% 100% 1999 5.9 100% n.a. 1999 247.4 100% n.a. 1999 9.5 100% 20% 1998 473.0 100% 51% 1998 5.4 51% 25% 1998 340.8 Note: Table includes all international carrier pdvatizations announced as of September t998. Almost 140 other carriers are at least part government-owned but have not made recent prival~zat]on p}ans. Traffic is volume (in minutes) of outgoing interna~enat cells. (a) Poland plans to sell up to 25 percent of TPSA through an IP0 in 1998, with an additional 30 percent stake to be made available to a strategic investor in early 1999. (b) Turkey plans to sell 20 percent of T(Jrk Telekom~nikayson to a strategic investor in November t998, with an additional stake to be made available through an IPO in 1999. Source: Adapted from TeleGsography’s 3-volurne directory of international carriers, New International Carde[s. © TeleGeography, Inc. 1998 © TeleGeographyo Inc. 1998 TeleGeography 1999 Figure 10. Traffic Base of Incumbent and New Carriers Traffic Base of Top 20 Carriers vs. Rest of World, 1997 Traffic Base of Carriers Beginning Operations after 1989 I0.0 11% 8,0 5.0 4.0 2.0 1.0 O0 0.03% 0.08% 1990 1991 Global Market Share of New Carriers 3% 2% 0.31% 1992 1993 1994 1995 1996 1997 Note: "New Carrier Traffic" only includes carriers that began facilities-based operation after 1989. Therefore, the figures exclude traffic originated by MCI, Sprint, Cable & Wireless Communications, IDC, and Japan Teiecom. Source: TeleGeography, tnc. © TeleGeography, Inc. 1998 Established carriers are also expanding their global footprint by acquiring strategic stakes in foreign state-owned carriers. Spain’s former monopoly, Telef6nica de Espafia, for exam- pie, has purchased stakes in CTC in Chile (43.6%), Telef6nica de Argentina (16.9%), Telef6nica del PetS, and may be negotiating for part of MCI WorldCom’s controlling stake of Brazil’s Embratel. Another European incumbent attempting to gain a foothold in Latin America is Telecom Italia, which has invested in Entel Chile (20%), Telecom Argentina (33%), Entel Bolivia (50%), as well as Cuba’s international carrier, Etecsa (30%]. Future privatizations are likely to offer incumbents further opportunities to glob- alize their business because many governments still hold a large stake in their national carriers. As of September 1998, 30 of the top 50 international carriers were still part- ly government-owned (see Figure 9, "Upcoming Carrier Privatizations"). Another way to become a global carrier is to inherit it. U.K.- based Cable & Wireless (C&W), the company which has been bringing telephone service to exotic (and not so exot- ic) locations since 1929, owns stakes in over 40 interna- tional carriers, including Hong Kong Telecom and Optus Communications in Australia. Major operators bearing the C&W name also exist in the U.S. and the U.K. For a com- plete picture of C&W’s international carrier assets, see the ownership chart in Figure 6. Last Stand of the Incumbents? The rapid growth and diversification of carriers has perma- nently altered the landscape of international telephony. But, though they be great in number, the market power of most newcomers is still limited. The world’s top ten carriers still carry half of all international traffic (see Figure 10, "Traffic Base of Incumbent and New Carriers"). The prima- ry issue is not whether incumbent carriers will survive--they most certainly will. At issue is their way of doing business. And as new carriers continue to gather steam, their new weapons (ISR, call-back, refile, the lnternet, and multi-coun- try affiliates) are likely to forever change the rules of the game. ~i~..’~ TeleGeography 1999 © TeleGeograpl~y, Inc. 1998 Map of WTO Commitments [] 0therWTO Members Key to Country Commitments under the w’ro Agreement ¯ Full Market Access in 1998 O Partial Market Access in 1998 ¯ Full Market Access in 1999 or After O Partial Market Access in 1999 or After O No Commitments Offered Note: Countw commitments under the 1997 WTO Agreement on Basic Telecommunica- tion Services include the right to offer Local (local telephone exchange), International (international telephone service), and Satellite (mobile satellite-based telephone service; commitments on fixed satellite ser- vices not shown). Foreign ewnerzhip refers to the right of a WT0 member to invest in a foreign telecom carrier; some countries have exempted specific carriers. Antigua & Barbuda Argan~na Ansi’alia Austria Foreign Foreign Local Interna~onal Satefl~ Ownership Local International Satellite Ownerzhip ¯ Dominica ¯ ¯ ¯ ¯ Dominican Rapublic Brunei ¯ ¯ ¯ ¯ Germany ¯ ¯ ¯ ¯ B.l,eria ¯ ¯ ¯ ¯ Gh.ea 0 0 0 0 Canada ¯ ¯ ¯ 0 Greoco ¯ ¯ ¯ ¯ Chila 0 ¯ ¯ Gr.nada ¯ ¯ 0 ~o~.,e 0 0 0 0 ~ue,a..,a ¯ ¯ ¯ ¯ CSte D’lvoire ¯ ¯ ¯ ¯ Hong Kong Q 0 0 © TeleCeography, Inc. 1998 TeleGe%oraphy 1999 Colom Chila ~Jblic /Antigua & Barbuda Iceland 1 Netherlands 2 Belgium 3 I.uxem bc~rg 4 Czech Republic 5 Slovak Republic 6 Austria 7 Hungs~/ 8 Romania 9 Bulgaria United Kingdom Portugal C6te d’lvoire Ghana ’ © TeleGeography, Inc. 1998 Designed by Gregory C. Staple and Timothy J. Stronge Local Foreign Foreign Foreign International Satellite Ownership Local International Satellite Ownership Local International Satellite Ownership Slovak Hungary ¯ ¯ ¯ ¯ Mexico ¯ ¯ ¯ 0 Republic ¯ ¯ ¯ ¯ Iceland ¯ ¯ O ¯ Mongolia ¯ ¯ ¯ ¯ South Africa 0 (~ 0 India 0 0 Netherlands ¯ ¯ ¯ ¯ Spain ¯ ¯ ¯ ¯ Indonesia 0 _ ~1 ~:’ 0 New Zealand ¯ ¯ ¯ ¯ Sri Lanka (~ 0 Ireland ¯ ¯ ¯ ¯ Norway ¯ ¯ ¯ ¯ Sweden ¯ ¯ ¯ ¯ israel 0 0 0 Pakistan S 0 0 Thailand ¯ ¯ ¯ 0 Italy ¯ ¯ ¯ ¯ Peru ¯ ¯ ¯ ¯ Trinidad ,Tobage ¯ ¯ ¯ ¯ Jamaica ¯ ¯ 0 Philippines ¯ ¯ !--’ 0 Tunisia ¯ : +: -:: 0 Republic of Korea Luxembourg Malaysia ¯ ¯ ¯ Poland ¯ ¯ ¯ 0 Turkey 0 0 0 0 ¯ ¯ 0 Portugal ¯ ¯ ¯ 0 United Kingdom ¯ ¯ ¯ ¯ ¯ ¯ ¯ R .... ia ¯ ¯ ¯ ¯ United s,.,es ¯ ¯ ¯ ¯ ~, 0 +enoga, 0 0 0 Vanazua,a ¯ ¯ ¯ ¯ ¯ ¯ ¯ Singapore 0 ¯ 0 Note: Belize, Brazil, Ecuador, M ....... Papua New G ....... d Swttzerland have made seaarate commitments TeleGeography 1999 © TeleGeography, Inc. 1998 Sources for the WT0 Map The World Trade Organization (WTO) Basic Telecommunications Agreement opens the telecommunication markets of over 60 countries to foreign carriers and investors. But the Agreement is not a single document. It is based upon certain general prinu ciples and procedural rights in two umbrella treaties--the Agreement Establishing the World Trade Organization and Annex 1B thereto, the General Agreement on Trade in Services (GATS). The GATS also has an "Annex on Telecommunications" and an "Annex on Negotiation On Basic Telecommunication." Both treaties were concluded in 1994 at the close of the Uruguay Round of trade negotiations begun at Punta del Este, Uruguay in 1986. The text of these documents forms an inte~ gral part of the "Final Act Embodying the Results of the Uruguay Round of Multilateral Trade Negotiations." The Basic Telecommunications Agreement also includes detailed countryMby-country commitments to liberalize the provision of certain telecommunication services pursuant to the GATSo These national "Schedules of Specific Commitments and Lists of Exemptions"--the exemptions refer to the Most Favored Nation (MFN) obligation in Article II of the GATS~-are annexed to the "Fourth Protocol" to the GATS, adopted in 1996. These Schedules often include an "Additional Commitment" to abide by the regulatory principles stated in a "Reference Paper" adopted in 1996 by the initial Negotiating Group on Basic Telecommunication (NGBT~. The chart at the bottom of the map overleaf summarizes each country’s Schedule of Specific Commitments on some of the most significant areas for liberalization: foreign ownership rules and local, mobile satelliteMbased and international market access. The full text for all Schedules and GATS documents can be found at http://www.wto.org/wto/services/tel.htm. © TeleGeography, Inc. 1998 TeleGeography 999 International Telephony & Asian Network Expansion by Craig Irvine, Merrill Lynch & Co., Inc. Introduction The Asian financial crisis has changed the economics of interna- tional telephony and its role in network financing. Net settle- ment revenues are now cherished, while a capital shortage favors incumbents and has altered the traditional economics of network expansion. The New Financial Challenge For most Asian incumbents, international telephony has histor- ically accounted for 30 percent to 60 percent of revenue, as shown in Figure 1. A growing settlement imbalance, particular- ly with the United States, has increased net settlement rev- enues and resultant pressure for lower settlement rates. A year ago, falling settlement rates and the Federal Communications Commission’s (FCC’s) settlement benchmarks were the top strategic priority problems for many Asian carriers. We believe this is no longer the case. Shrinking access to investment capital has overtaken depen- dence on international settlement payments as the key financial challenge for many Asian carriers. The depreciation of Asian currencies has transformed a key concern--falling net settle- ment revenues--into a key positive--U.S, dollar revenues. Pressure on settlement rates continues, but it has been over- come by the shortage of investment capital as debt spreads have widened and new equity funding has dried up. This is not to suggest that carriers have surrendered in the settlement rate battles. Rather, many have shifted their focus to a new area of greater concern: where to apply their scarce investment capital. An acute shortage of capital enhances incumbents’ competitive position. Established players enjoy distinct--and now more sustainable--competitive advantages over new entrants. Expensive local loop assets remain a key route to capturing international traffic. A competitive advantage based on net- work coverage--which is unsustainable given adequate time and resources--becomes relatively sustainable when capital is scarce. Indirect access to alternative international carriers has not overcome this obstacle: access charges in many countries remain stubbornly high, leaving insufficient margin for the new entrant to finance a meaningful toehold in the local loop. Incumbent profitability is also based primarily on pre-crisis investment costs. A few cases of successful tariff rebalancing notwithstanding, local currency tariffs remain at or below cost, while equipment remains largely imported and thus consider- ably more expensive today than a year ago in most of Asia. So while an incumbent can typically stow investment and harvest cash, the new entrant is spending precious capital at ever-high- er prices just to catch up, let alone pull ahead. Figure 1, Domestic and International Telecom Revenue, 1997 Telekom Malaysia Singapore Telecom Philippines Long Distance Telephone Hong Kong Telecom 0% 20% 40% 60% 80% 100% Percent of Total Gross Telecom Revenue Source: Merrill Lynch & Co., Inc. © TeleGeography, Inc. 1998 TeleGeography 1999 © TeleGeogrophg, Inc. 1998 Foreign strategic capital can turn this dynamic on its head. A fresh injection of capital can rejuvenate the financial viability of the new entrant. Not surprisingly, virtually every new entrant in Asia (wireline and wireless) is now attempting to court one or more foreign strategic partners, though we believe transactions will be sparse in the current environment. So does this mean incumbents are securely riding a virtuous cycle where a dominant share of end user connections feeds them international traffic and keeps net settlement revenues rolling in? Probably not. While competitive pressures may be easing, pressure on profitability clearly remains. Incumbents enjoy strong cost advantages over new entrants, but in the cur- rent environment, they must still apply this advantage wisely to support overall returns. The currency crisis has directly affected the incentives to expand basic networks. Figure 2 shows a simple estimate of the internal rate of return [IRR) of new line investment now and eighteen months ago. For simplicity, we assume revenue per line and operating mar0ns remain constant (ignoring cellular and other distortions), and that the incremental investment cost of a new line is a constant US$1,000 per line with a seven year life. Clearly, this cost is even higher for new entrants still installing core infrastructure. Decreasing IRRs have thus lowered the incentive to expand net- works, which further implies less need to defend high settle- ment rates. So are we now seeing greater willingness to nego- tiate down settlement rates? Yes and no. Most carriers con- tinue to resist falling rates, but Hong Kong Telecom (HKTJ would like to see the pace of decline accelerate. The cross subsidy has evolved into a different form in Hong Kong. Rather than subsidize network expansion, high interna- tional telephony margins have been used to subsidize compet- itive entry. For three years now, new entrants have shared their delivery fees with consumers and captured enough of HKT’s core business that earlier this year HKT agreed to give up its international monopoly. Hong Kong Telecom is now preparing for the post-monopoly world by negotiating settlement rates to below the same FCC benchmarks (US$0.15 per minute) which a year ago they were aggressively resisting. Hong Kong may be able to become the first Asian country to do away with settlement rates on some routes where international simple resale [ISR) is lega!--perhaps even without a major discontinuity in profitability or ability to finance network expansion. One of HKT’s goals is to reclaim this traffic for the switched network and, in effect, to offer an international direct dial ODD) service that is price-competitive with ISR and call-back, while betting on broadband services, superior cost efficiencies, and abundant capacity to maintain profitability. AS such, reform in Hong Kong has moved closer to the com- petitive ideal than any other Asian market. 1SR will be legal (with licenses "on demand") on reciprocally legal routes from January 1999 with facilities-based competition from January 2000. Other regulatory initiatives support development of free Figure 2. Simplified IRR of New Line Investment 120% 100% 8O% 6O% 4O% 20% 0% -20% ¯ Dec. i996 [] Sept. 1998 Hong Kong India (MTNL) Philippines Long Singapore Tetekom Malaysia Telecorn Distance Telecom Indonesia Telephone (PT Telkorn) Note: Internal Rate of Return (IRR) is the theoretical discount rate at which the net present value (NPV) of new line investment equals zero. Devaluations in Asian currency have had a serious impact on these projections, most notably in Indonesia, Source: Merrill Lynch & Co., Inc. © TeteGeography, Inc. 1998 © TeleCeography, Inc. 1998 TeleCeography 1999 competition, including mandatory number portability and shar- ing of certain local loop facilities. InterestinBly, Hong Kong is pushing ahead with these reforms when international traffic (ex- China) is falling, thereby creating an interesting case study in accelerating reform during difficult times. If carriers are short on internally generated capital, what can they do? Unfortunately for most carders, the regional crisis has slowed external capital raisings to a virtual halt. Exceptions include a few debt refinancings, but new equity issuance has slowed dramatically. Many carders thus seek to obtain new capital from each other--through mergers and/or strategic investments. Market observers have predicted consolidation in the Asian telecom industry for several years now. However, virtually all regional mergers and acquisitions have been confined to in- country deals, partly because foreigners are generally not inter- ested. Foreign telcos are less than interested in strateBic investment in Asia for several reasons. In most Asian countries, foreign own- ership limits have prevented foreign investors from obtaining direct contro!. And past difficulties in obtaining meaningful operating influence on a minority basis have made injecting fresh capital into struggling new entrants difficult to justify on current terms. While global players would likely pay a strate- gic premium to be able to guarantee quality connectivity to global multinational carriers, they have generally been unable to obtain this degree of direct operating control. Moreover, strategic investors can rarely justify acquisitions on a replace- ment cost or inexpensive capacity argument. Returns on tete- corn assets have historically been closely tied to their physical location and hence, to the economic activity of the area. Foreign carders must thus be cautious about viewing distressed Asian networks merely as cheap capacity. Privatizations have similarly stalled. Deep vested interests in heretofore unusual market structures have persisted, and in some cases, actually deepened. Several countries still lack an independent regulator, even where the governing ministry has an economic stake in unusual market structures or revenue sharing arrangements. When will this improve? Most likely when governments and/or dominant private shareholders lower their valuation and control expectations. This is likely to happen only when the pain gets intense enough, which in turn is integrally related to issues like the degree of creditor forbearance (for privately-owned card- ers) and depletion of government fiscal reserves (for state- owned carriers). Strategic Responses In this environment, over the next 1 2-18 months, carriers have several strategic alternatives, including: Develop a "hub" strategy For carriers less reliant on inter- national service to finance buitdout, we expect "hub" strate- gies. For example, having effectively lost the high margin of switched international traffic, HK~ is attempting a transition to a high volume, low margin hub business. In one sense, this strategy attempts to turn telecom services into a "trad- able" good. Rather than be dependent on traffic originating from or terminating in the host country, an efficient carder with no IDD margin to protect can compete for the entire global market--but only if their internal efficiencies, corre- spondent relationships and transmission capacity are strong enough. Acquire struggling competitors We expect incumbents to be on the lookout for regional acquisitions. In this respect, the Asian crisis has created some attractive value for inter- national carders seeking to diversify their revenue bases away from the pricing risks of the settlement rate system. We believe acquisitions of local loop assets--at the right price--will develop into a key strategy, particularly for Asian international carders with abundant cash. Grow or acquire cellular customer bases A corollary to this approach is to focus on cellular, which represents a rapidly Browing proportion of total "lines." As a key source of inter- national traffic, cellular is thus Browing in strategic impor- tance to diversifying international carders. Conclusion Financing Asian network expansion remains a complicated game. International tariffs continue to fall while local currency investment costs rise. Most Asian carriers still face a difficult transition to fully rebalanced tariffs, which despite the growing financial pressure to liberalize more fully, is still probably sever- al years away. ~i~..~ Craig Irvine is Vice President!Asian Telecommunications Analyst at Merrill Lynch & Co., Inc. He can be reached by email at craig_irvine@hk.mL com. TeleGeography 1999 © TeleGeography, Inc. 1998 International Long Distance in Latin America by Myles Davis, Luiz Carvalho, and Josh Milberg, Morgan Stanley Dean Witter & Co. The editors of TeleGeography have asked us to address several questions: First, what are the prospects for the region’s inter- national long distance providers--both incumbents and new entrants? Second, what were the key events of 1998 and what do we expect next year and beyond? Third, how much debt and equity was raised in 1998, and what’s on the calendar after 19987 Finally, what will be the important issues for inter- national carders in the next year?. Telecommunications in Latin America is stepping across the threshold of liberalization. Some countries have moved in advance of the rest of the region--Chile and Mexico, in partic- ular-and others have, through referenda or other means, only begun to open their telecom sectors. For the majority of nations in Latin America, however, 1998 and 1999 are the years in which the first major steps are being taken toward lib- eralized, competitive telecom markets for local service and both national and international long distance. In our view, two main issues have moved to the forefront. First, regulators and governments are focused on their countries’ need for infrastructure, reflecting the relatively low phone pen- etration rates in Latin America compared to those in North America or Western Europe. This has led to buildout require- ments for new licensees in most Latin American nations. Second, the laws of supply and demand suggest that fewer licenses awarded should tend to create a higher price per license. A desire to maximize value~and income~has, in our view, played a part in keeping low the number of new entrants at this stage of liberalization. The pace from here, then, is deliberate, and offers a relatively benign scenario for incumbent providers. Chile’s experience with full-scale competition in national and international long distance might be seen as a cautionary tale for the large Latin American regulators. Although Chilean outbound international traffic has grown an average of 49 percent per year since full competition was introduced in late 1994, rates have fallen over 18 percent per year in the same time frame (on a weighted average basis) and Entel’s market share has fallen to approxi- mately 36 percent. Following the initial "free for all," consoli- dation has begun, leaving Chile with three major long distance competitors. In this environment, many countries have decided on a duop- oly or oligopoly as the first stage of market liberalization. Brazil, for example, has outlined plans that will create one long distance competitor with a nationwide license to compete with Embratel, the incumbent long distance (domestic and interna- tional) provider Embratel itself was purchased in this year’s Figure 1. Comparative Latin American Long Distance Rates $1.60 $1.20 $1.oo .~_ ~~ $0.80 $0.40 $0.20 $0.00 13 Domestic Long Distance Avg. ¯ International Long Distance Avg. E]Avg. Incoming Settlement Argentina Brazil Chile Mexico Peru Venezuela U.S. Note: Figures based on carrier estimates and are current to August 31, 1998. Source: Morgan Stanley Dean Witter & Co. © TeleGeography, Inc. 1998 44 © TeleGeography, Inc. 1998 TeleGeography 1999 privatization auction by a consortium led by MCI. The new entrant must fulfill significant requirements for infrastructure deployment. Argentina, according to its liberalization schedule, will license two long distance competitors to the incumbents Telecom and Telef6nica before November 1999. As in Brazil, there is a requirement for network buildout. There are excep- tions to this trend--El Salvador, one of the smaller markets in the region, has awarded at least eight provisional licenses to long distance competitors since November 1997. Current Trends What do these trends signify for new and incumbent providers of international tong distance in Latin America? First, we expect a more gradual transition to competition than we have seen in the U.S. or Europe. We have mentioned the infrastructure deployment required of most new licensees in Latin America, as well as the factors contributing to small num- bers of awarded licenses in many countries. These in essence raise high barriers to entry to Latin markets, and keep the num- ber of competitors limited. Second, the resale market will develop more slowly. The regu- latory orientation we have described--and the low phone pen- etration that drives it--also create another significant difference between the advent of competition in Latin America and that in the U.S. or Europe. In the U.S., resale became a part of the competitive mix in a relatively short time, due to the availabil- ity of network elements and the strong regulatory structure in the U.S. As Europe moved toward full liberalization in 1998, network elements were also available for lease. In Latin America, however, there is little network capacity available for resale--with Chile as the notable exception--and little interest on the part of regulators to make it available, given their focus on increasing overall phone penetration and infrastructure buildup. Third, we forecast only a ~radual reduction in the current rela- tively high rates for international calls in many Latin American markets. These rates have declined over the past few years (see Figure 1) but remain among the highest in the world. We expect only a gradual fall in collection rates in markets when there are only one or two competitors. In many countries, the driving force behind lower international rates will continue to be call-back operators and Internet telephony providers, whether legal or not. We must add one more variable in the foregoing analysis: the "global contagion" of currency crises which has battered Latin American stock markets in 1998. This environment could lead to slower growth in Latin American economies, in turn, affect- ing the outlook for telecom growth. We have begun to see the results of this, as at least one potential competitive entrant has deferred its planned investments in Latin America until both the economic and the regulatory environments stabilize. From con- versations we have had with other operators, the general con- sensus is that meanin~ul entry by competitive operators will not take place in many Latin American countries at least until 2000. By this time, they believe, the economic difficulties fac- ing the region in 1998 will be improving, and the limited corn- Figure 2. Debt and Equity Raised in Latin America, 1998 Date Operator]Country Description Price Paid ($US) April 98 Tricom Sole alternative provider of diversified telephone services (ILD, $70 million (Dominican Republic) DLD, cellular, and local)in Dominican Republic. Raised $70 million in IP0 on New York Stock Exchange, only public offering of equity by Latin American telecom in 1998. June 98 Impsat Leading competitive provider of satellite and terrestrial private $225 million (Argentina, Colombia, line and data services throughout South America, primarily others) to large corporate customers. $225 million high-yield offering. July 98 Telebras The Brazilian government sold its stake (representing 51.8 percent $2.65 billion (Brazil) of voting shares)in each of 12 newly formed companies carved (Embratel) out of Telebras, to a number of strategic investors. The 51.8 percent stake in international carrier Embratel was bought by a consortium led by MCI for $2.65 billion. July 98 CTE France TOlOcom acquired 51 percent of CTE, a corporation formed $275 million (El Salvador) from ex-monopoly ANTEL. CTE had 380,000 fixed lines in service, 180 million MiTF a year, and an as-yet-uncommerciatized license for PCS 1900 MHz. CTE has no exclusivity. France T~lOcom’s offer exceeded the only other bid, made by Telmex, by $4 million. July 98 Intelsa Telef6nica acquired 51 percent of Intelsa, a corporation holding $41 million (El Salvador) as-yet-uncommercialized licenses for cellular and fixed telephony. Source: Morgan Stanley Dean Witter & Co. © TeleGeography, Inc. 1998 TeleGeol~raphy 1999 © TeleGeography, Inc. 1998 petition phase of liberalization will be coming to an end, allow- ing more competitors into the markets. The Future What lies ahead? Four key issues face both incumbent and new international operators in Latin America. First, the presence of a strong regulatory body. Latin America does not have the equivalent of a European Union (EU) to define and to press for relatively uniform rules of liberalization and competitive entry. Although Memosur, the regional trade alliance between Brazil, Bolivia, Chile, Argentina, Paraguay, and Uruguay has played this role to some extent, it does not encompass all of Latin America, nor are its member states as firmly tied to the organization as are the members of the EU. Given this, whether individual nations are able to create regu- latory bodies that can serve as strong and effective arbiters of competition--safeguarding the interests of those operators that "build" while clearing an equitable path toward eventual full competition--will serve as a defining test of national commit- ment to competition. In the absence of a strong regulator, the road toward open telecom markets will be long and slow, given the role that most incumbents play in the region’s national economies--and stock markets. Second, how well and how quickly the region--and the world-- is able to pull itself out of the current economic crisis. Economists here at Morgan Stanley Dean Witter have argued ¯ forcefully that most Latin economies are strong relative to other emerging economies and do not deserve the punishment they are receiving. Foreign direct investment has continued at a strong pace even through the turmoil of mid- 1998--the strong demand for the Brazilian telecom pdvatization being a major example. But mutual fund flows into the region have slowed. And global debt and equity markets as a whole have become much more difficult for most smaller international players. The availability of capital is one of the major drivers behind the rapid pace of competitive entry in Europe, and, without it, reaping the benefits of liberalization in Latin America even in the best regulatory climate would take quite some time. Major global telecom players--MCl, Telef6nica and Telecom Italia-- have made substantial investments in the region, but these for- eign firms have largely invested in incumbents, and we believe it is the next generation of entrants, operating on a smaller scale, that will bring truly competitive services to the region. Two things must occur for this next generation to move: first, access to capital in world markets in general must improve, and second, the riskiness, real or perceived, of investing in Latin America must diminish. Third, infrastructure development. As mentioned, Latin America has relatively low teledensity rates. Extending tele- phone networks to a larger percentage of the population, cre- ating backbone networks to carry voice and data, and deploy- ing switches and touters to handle the traffic are much needed. And infrastructure requirements are part of most liberalization and licensing structures in the region. Western Europe, by com- parison, was easily adaptable to competition; phone penetra- tion was high, and leased circuits, though an order of magni- tude more expensive than in the U.S., were available. Fourth, and finally, the decline in international settlement and collection rates. This is more of a medium term than near term issue for Latin America, as we expect higher-than-average rates to persist longer in this region than in the U.S. or Europe. Over time, however, both rates to customers and termination rates will move toward cost, and strategies based on rate arbitrage will result in narrower and narrower margins. This will have three effects: First, at the wholesale level, countries which are net traffic importers--those that receive more calls than they make--should lose income from the decline in settlement rates. Second, the decline in collection rates should result in less income to carders as each minute costs less to the caller Lastly, this decline in cost should spur traffic growth, as we have seen in Chile, where, as we mentioned, rates have declined 18 percent per year while traffic has risen 49 percent per year Ultimately, the health of any international operator will be dependent on an efficient, low cost network in the target coun- try and a sound scheme enforced by a strong regulator--for interconnection to the local loop at reasonable rates. This net- work might be owned, or leased from another carrier if a coun- try has developed sufficient infrastructure, as in Chile. But competitive local access is probably a distant point down the road for many countries in Latin America. Nevertheless, facil- ities-based competition has taken hold in the U.S., and Europe seems to be moving in a similar direction--but at a more rapid pace. Eventually, therefore, we expect Latin America will do the same--probably sooner rather than later Myles Davis and Luiz Carvalho are research analysts, and Josh Milberg is a research associate, at Morgan Stanley Dean Witter & Co. They can be reached by email at davismy@rns.com, carvalh@rns.corn, and milberg@ms.com, respectively. © TeleGeographyo Inc. 1998 TeleGeography 1999 Emerging North American Carriers by Joe Noel and Mark Langner, Harnbrecht & Ouist, LLC In 1998, the North American telecommunications market saw a new group of international carriers emerge as a market rome. These emerging international long distance (E-ILD) operators have taken advantage of a number of trends, including dereg- ulation of telecom markets worldwide, steady growth in tele- phone usage, and an expanding number of second and third tier domestic carriers who need international connections. Emerging ILDs have raised large amounts of capital to support their growth. In so doing, most have moved from a "Light Carrier" resale model to a "Heavy Carrier" facilities-based model (see "International Carder Evolution" on page 26). This shift, in turn, has raised barriers to entry for other potential entrants to the international telephony market. New submarine cable companies, in part supported by E-ILD players, also wilt bring massive amounts of international capac- ity on-line. That will put further pressure on prices. It will also aid the emergence of high-bandwidth, data-oriented carriers, which have started the U.S. with companies such as Qwest and Level 3. We discuss these developments in more detail below. Market Dynamics In North America, the carriers pursuing the E-ILD market have changed over the last two years. Most are now bigger and more firmly established, with more of their own facilities, oper- ations, and intemonnection agreements than at the beginning of 1997. The growth of the early entrants has also effectively closed the window of opportunity for aspiring providers to "bootstrap" their way into the E-1LD market. In contrast, from 1990 to 1996 or so, a low capital market entry path existed, based on the resale and "call-back" services which some cur- rent E-1LD players used to launch their businesses. E-ILD carriers are now intent on increasing their direct inter- connections and deployed facilities in order to lower costs, par- ticularly on less competitive routes outside of the pdmary focus of incumbent providers. At the same time, E-ILDs are increas- ing their product lines and retail sales channels. We see retail sales as a necessary complement to the E-ILD provider’s tradi- tional wholesale business so as to manage growth and to main- tain margins. Figure 1. Performance of Emerging International Long Distance Carriers S&P Long Distance Index E-ILD Index I Jan. 1, 1997 Jan. 1,1998 2O0 180 160 140 120 100 6O Note: Index prices are from January 1,1997 to September 14, 1998. The benchmark of 100 equals the value of each index on January 1, 1997. E-ILD Index comprises: Star Telecom--STRX, IDT Corp.--IDTC, Pacific Gateway Exchenge--PGEX, Primus Tetecommunications--PRTL, RSL Communlcations--RSLCF, Startec Global CemmunlcaUons--STGC, end Telegroup--TGRP. Source: Hambrecht & (luist, LLC © TeleGeography, Inc. 1998 TeleGeography 1999 © TeleGeography 1999 Figure 2. Financings of Emerging International Long Distance Carriers Public Debt Company Ticker Private Equity IPO Follow on Startec Global STGC $1,003,259 $39,330,000 -- STAR Telecom STRX $13,649,000 $35,550,000 $157,545,000 I DT I DTC $2,411,600 $46,000,000 $126,707,031 Primus PRTL $29,241,328 $60,375,000 -- RSL C0M RSLCF $97,787,000 $182,160,000 (c) Telegroup TGRP $56,961 $46,000,000 -- Pacific Gateway Exch. PGEX $941,734 $59,280,600 (e) -- $100,000,000 $375,000,000 $988,969,600 $97,000,000 Notes: (a) Includes $160M private debt placement. (b) Includes $7.5M of Conve~ble Debt ($6.5M of which was subsequently conver~ed). (c) In August, RS L COM fired a registration statement for tz $253M follow on offering. (d) Includes $20M of Senior Sub Debt ($1Z4M redeemed with IPO proceeds); includes $25M conve~ble notes. (e) Includes $2t.6M from sale of shores to individual investors at lPO price. Source: Hambrecht & Quist, LLC Long TermDebt $160,266,000 (a) $27,775,000 $15,936,27 (b) $21,461,000 $45,263,118 (d) © TeleGeography, Inc. 1998 Market Performance The performance of E-ILDs in the public market since 1997 has trailed more established segments of the telecom industry. Figure 1 shows the performance of the E-ILD group relative to the S&P RBOC and Long Distance telecom indices. The E-ILD providers’ stock values have suffered particularly since the stock market moved off of its historic high in July 1998. However, if we took at the performance of the individual com- panies over this same period we see that certain E-ILD providers have experienced strong stock price growth--particu- larly IDT and STAR Telecom--which approximately doubled in value and outperformed the S&P 500. In this analysis, com- panies with business models which have flexibility and which have focused on building facilities have fared better than those carriers more reliant on wholesale and/or resale business mod- els. Ultimately we believe that the E-ILD companies which are best able to manage the disparate environments of network building (a long term process), and the constantly changing day-to-day nature of pricing international services, will be most successful. In today’s volatile stock market, it is no longer enough to man- age one side or the other of this equation. Looking to 1999 Although the E-ILD market is already mature, we see the mar- ket becoming even more complex due to the influence of data and the Internet on international telephony. Additionally, con- tinuing market volatility will have a tremendous impact on the strategies of international carriers based on the availability of capital to fund acquisitions and growth. While not discussed in detail here, in 1999 we expect Internet protocol (IP) and larger packet-based telephone services on the networks of new carriers, more than telephony on the public Internet, to play a larger role. For the most part, we expect these types of services to feed into the existing structure of wholesale telephony as another transmission choice for inter- national service retailers, with price, quality and availability as the factors determining the speed at which these types of ser- vices grow. We expect E-ILD players to become players in the IP telephony market. Some E-ILDs have already begun IP ser- vices, such as IDT’s Net2Phone, and RSEs Delta Three. Additionally, we expect that current E-ILD providers will also be likely distributors of IP telephony services sold on a wholesale basis. The Rise of the International Data Carrier During 1997 and 1998, the North American market has seen the strong growth of data-oriented carriers such as Qwest, Level 3, IXC Communications, and others. In 1999, this trend will begin to move offshore, supported by the vast amounts of undersea cable being brought on-line by companies such as Global Crossing, FLAG, MC] WorldCom, and others. An early indication of the strong demand for international data services is reflected in the IPO of Equant earlier in 1998, leaving the carrier capitalized at over $6 billion. This trend will have its greatest impact on incumbent providers as these companies’ blue chip customers are most eager for global data solutions. In the medium term (two to five years) we expect the increased amount of capacity that these data carriers will bring to the market will have a profound effect on traditional international telephony markets. The resultant drop in voice traffic, howev- er, is expected to be concentrated on tier one countries, which are markets where the E-ILD players have already seen signifi- cant competition, and on customer groups that the E-ILD play- ers have not traditionally pursued (the largest of corporate cus- tomers). The Role of Capital as a Competitive Weapon Based on current trends, we believe that managing capital will be an increasingly important part of the E-ILD players’ strate- © TeleGeography, Inc. 1998 TeleGeography 1999 gies. Figure 2 shows the group of EqLD providers that have been most active in the capital markets. In recent months, however, funding for most communications carriers has dried up. In the U.S. there has not been a signifi- cant high-yield debt offering completed during the summer of 1998. Hence, if markets continue to be volatile, there will likeM ly be a significant impact on new car~ers’ ability to raise funds using the same avenues available to prior entrants. In contrast, carriers with capital--and the necessary cash flow to service debt--are likely to strengthen their market positions and be able to take advantage of acquisition opportunities. Conclusions Overall, we continue to see international telecom as an a~rac- rive business for established players. We believe this is due to continued strong growth fundamentals for international teie- phony services in general. We expect North American incumbents to see continued ero- sion in overall market share. The one cartier that is likely to escape this trend is Teleglobe. This carrier has "changed its stripes" and acts more like an E-ILD provider than an incum~ bent (though we expect Teleglobe’s Canadian share to continue to erode). For the E-ILD players, margins will continue to shrink on routes to developed countries. The effect will be offset, however, by continued success in other, less competitive markets, where the E-ILD players have valuable relationships and cimuit capacity, where vast amounts of capacity are not likely to come on line soon, and where incumbent players and other retail-oriented providers will continue to need capacity from EqLDs. Joe Noel is senior analyst for telecom markets and Mark Langner is a research associate for communications services at Hambrecht & Quist LLC (San Francisco), an investment bank. They can be reached at ÷ 1 (415) 439 3300, or via email at jnoel@hamquist.com and mlangner@hamquist.com. Te~eGeography 1999 © TeleGeography, Inc. 1998 International Services of U.S. RBOCs When will the RBOCs compete for international services? The short answer is "they already do," but primarily as resale car- riers (see Figure 1). And even though some of the Regional Bell Operating Companies [RBOCs) may acquire their own interna- tional facilities, they won’t become major competitors until they can sell international services to their own local customers. In most states that will not occur until 1999, at the earliest. To understand why, it is helpful to review briefly America’s historic communications reform law, the Telecommunications Act of 1996 ( 1996 Act). The 1996 Act was motivated largely by two interrelated objec- tives. First, the U.S. Congress sought to foster greater compe- tition for local telephone services by, among other things, allow- ing the country’s major long-distance carders--AT&T, MCI and Sprint--to compete directly for local services with incumbent carders, such as the RBOCs. The second goal--and the political quid pro quo for the first-- was to free the RBOCs from the antitrust constraints imposed in 1984 when they were divested from AT&T. Once freed, the RBOCs would be able to provide interexchange, including inter- national, service in direct competition with their former parent. The RBOCs are by far the largest local exchange carriers (LECs) in the United States. Each RBOC serves at least 15 million access lines, and collectively the RBOCs account for approxi- mately 85 percent of all U.S. access lines. The 1996 Act per- mits RBOCs wishing to provide international service for calls originating outside of their local service regions to do so by sire- ply filing a standard application under Section 214 of the Communications Act. In contrast, for in-region international service, an RBOC must obtain Section 214 authority and file an application, state-by- state, under Section 271 of the Communications ACt. The Federal Communications Commission (FCC) may not grant an RBOC Section 271 authority until it has consulted with the U.S. Department of Justice and the agency is satisfied that three competitive safeguards have been met. First, for each state in which the RBOC seeks to provide service, the RBOC must have entered into a connection agreement with at least one unaffiliated, facilities-based (or predominantly facilities-based) competitor Alternatively, an RBOC may pub- lish its general terms for access and interconnection, which must have been approved by the relevant state utilities com- mission. Second, the RBOC’s interconnection agreement or its published terms must satisfy a competitive checklist. Specifically, inter- connection must: (l) be unbundled and cost-based; [2) include access to poles and rights of way; (3) include access to emergency and directory services; (4) provide universal direc- tory listings; (5) provide access to telephone numbers; (6) pro- vide for local dialing parity; (7) offer number portability; (8) offer reciprocal compensation arrangements; and (9) permit resale. Third, once this checklist is satisfied, the FCC may only autho- rize an RBOC to offer in-region long-distance service if it is pro- Figure 1. RBOC International Services Authorized by the FCC Out of Region IMTS In Region IMTS Switched Resale Facilities-Based Switched Resale Facilities-Based Ameritech Corp. July 19, 1998 July 9, 1997 -- -- Bell Atlantic Corp. July 19, 1996 Feb. 7, 1997 -- -- (NYNEX) July 19, 1996 Feb. 6, 1997 pending -- BellSouth Corp. June 3, 1996 June 3, 1996 pending -- SBC Communications, Inc. Oct. 25, 1996 -- -- -- (Pacific Telesis) Feb. 13, 1997 Sept. 5, 1997" pending pending U S West, Inc. Dec. 27, 1996 -- -- -- Note: IMTS is International Message Telephone Service. Each application for international service was filed by an RBOC subsidiary separate from the local service provider. All dates are for the earliest application granted. Data current to Sept. 1998. ¯ Indicates route restrictions apply where the RBOC has a foreign carrier affiliate, © TeleBeography, Inc. 1998 © TeleGeography, Inc. 1998 TeteGeography 1999 Figure 2. Changing Landscape of U.S. Regional Bell Operating Companies RBOC Territories in 1994 Ameritech RBOC Territories in 1999 US West Note: At this printing, SBC’s mergers with Amentech and SNET were still pending final approval. Map does not display territory gained by the pending Bell At~ant~c-t3TE merger. GTE, although not an RBOC, is the third-ranked local service provider m the U.S, with service territories spread across 28 states. © TeleGeography, Inc. 1998 TeleGeography 1999 © TeleGeography, Inc. 1998 Figure 3. RBOC Section 271 Applications State Status Ameritech Corp. Michigan denied (8/97) Bell Atlantic Corp. -- -- BellSouth Corp. South Carolina denied (12/97) Louisiana denied (10/98) SBC Communications, Inc. Oklahoma denied (6/97) US West, Inc. -- -- Note: Each application for international service was filed by an RBOC sub- sidia~/separate from the local service provider. Data currentto Oct 1998. © TeleGeography, Inc, 1998 vided through an independent affiliate with separate officers, directors, employees and accounts. This separate affiliate requirement "sunsets" after three years. Ameritech was the first RBOC to file a Section 271 application for in-region long-distance authority with a January 1997 application to serve Michigan. However, Ameritech’s applica- tion was later dismissed because its interconnection agreement had not been given final state approval. Ameritech later refiled its application, but the FCC rejected it again in August 1997, this time because the company’s interconnection agreement did not satisf3, three items in the competitive checklist. Later Section 271 applications by BellSouth and SBC also have been rejected (see Figure 5). BeltSouth has appealed the denial of its South Carolina appli- cation in a court case that is currently pending. In a novel arBu- ment to the U.S. Court of Appeals, BellSouth claims that Section 271 is unconstitutional because its conditions for long distance market entry apply only to the RBOCs. Under this theory, Section 271 amounts to a "Bill of Attainder"--that is, a law that seeks to punish certain individuals--which is prohibit- ed by the U.S. Constitution. Surprisingly, this argument has been met with some support in the Court of Appeals and is being given serious consideration. Developments in the BellSouth appeal, therefore, bear close watching. Frustrated in their attempts to receive Section 271 approval, the RBOCs have sought creative ways to provide their cus- tomers with bundled local and long distance service offerings. For example, two RBOCs--US West and Ameritech--recently entered into an arrangement with Qwest Communications to provide long distance services to their in-region customers. However, in September 1998 the FCC struck down these arrangements with Qwest, saying that they violated the requirements of Section 271. Qwest and US West promptly filed a court appeal. Bottom line: Absent judicial intervention, the FCC is unlikely to authorize the RBOCs to offer in-region long distance and international services until they make a good faith effort to comply with the competitive requirements of Section 27 I. Until the RBOCs have authority to provide long distance ser- vices in key states, they will not be able to market internation- al service to their core customers--business and high volume residential customers within their local service regions. The FCC’s Section 271 proceedings (and related local interconnec- tion proceedings) thus will require continuing review by anyone interested in the RBOCs’ future as international carriers. This overview is adapted from a paper prepared by Koteen & Naftalin, LLP, entitled "The RBOCs Enter the Market for Domestic and International Long-Distance Services." Koteen & Naftalin, LLR is one of Washington DC’s leading communica- tions law firms. Founded in 1953, its clients now include U.S. and foreign companies in the telecommunications, data net- working, electronic equipment, broadcasting and entertainment industries. For further information, contact Greg Staple at +1 202 467 5700 (voice); +1 202 567 5915 (fax); greg.staple@koteen, com. © TeleGeography, Inc. 1998 TeleGeography 1999 European Carrier Financings Figure 1. Major European Financings, January 1997-October 1998 Principal Amount Offer Date Issuer Issuer Nation Type of Security (US$ millions) 1/10/97 Telenor AS Norway Notes $109.5 1/27/97 France Telecom SA France Notes $142.2 2/17/97 Telefonica Spain ADRs $210.8 2/27/97 Esprit Telecom Group United Kingdom ADRs IPO $17.1 4/8/97 British Telecom United Kingdom Notes $1,000.0 5/13/98 British Telecom United Kingdom Bonds $1,000.0 10/9/97 Portugal Telecom SA Portugal ADRs $301.4 10/20/97 France Telecom SA France Ordinary Shares IPO $2,455.9 10/25/97 Telecom Italia Italy Common Shares $1,968.0 11/7/97 Esat Telecom Group Ireland ADRs IPO $39.0 11/13/97 MagyarCom Hungary Ordinary Shares IPO $385.8 11/20/97 COLT Telecom Group United Kingdom Common Shares $114.5 11/21/97 COLT Telecom Group United Kingdom Global Bonds $199.1 12/9/97 Energis United Kingdom Ordinary Shares IPO $119,7 12/12/97 Esprit Telecom Group United Kingdom Bonds $148.0 3/11/98 Telenor AS Norway Notes $100.6 3/20/98 SPT Telecom AS Czech Republic Common Shares $30.4 4/3/98 Telefonica Spain Ordinary Shares $2,746.0 4/23/98 SPT Telecom AS Czech Republic Notes $278.1 5/1/98 Phone Systems & Network France Ordinary Shares IPO $3.3 5/5/98 SPT Telecom AS Czech Republic Notes $140.5 5/6/98 France Telecom SA France Bonds $500.0 5/15/98 Maltacom Malta Global Dep. Rec. IPO $46.0 5/18/98 Telenor AS Norway Notes $67.3 5/20/98 VersaTel Telecom BV Netherlands Bonds $225.0 6/2/98 OTE Greece Ordinary Shares IPO $74,3 6/7/98 NETnet International Sweden Ordinary Shares $2.6 6/11/98 Cable & Wireless United Kingdom ADRs $1,033.3 6/17/98 Helsingin Puhelin Finland Ordinary Shares $168.6 6/18/98 Esat Telecom Group Ireland ADRs $93.0 6/19/98 Esprit Telecom Group United Kingdom Senior Notes & Bonds $198.6 6/26/98 Tele Danmark AS Denmark Bonds $276.3 7/23/98 Telenor AS Norway Senior Notes $112.0 7/24/98 COLT Telecom Group United Kingdom Bonds & Stock $336.9 10/5/98 Swisscom Switzerland Ordinary Shares IPO $5,500.0 Source: Securities Data Corp. (http://www.securitiesdata.com) © TeleGeography, Inc. 1998 TeleGeography 1999 © TeleGeography, Inc. 1998 Market Shares of Competing International Carriers Percentage of Outgoing MiTT Country/Carrier 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 United States AT&T 89.1 83.3 78 4 74 8 70.3 62.2 6(31 MCl 7.0 10.2 14 6 17 8 21.2 24.8 26 5 Sprint 3 5 5.8 6 4 6.3 7 3 10 3 11.1 Worldcom 0 6 2 1 54.3 50.2 45.3 28.5 28.4 26 0 11.3 13.2 12.2 3,5 4 5 6.2 33 2.4 3.7 7.0 Pacific Gateway Exch. Others 0.4 0 7 0 7 1 1 1 2 2.1 0 2 United Kingdom BT 95 5 91.0 86.0 81.0 76.8 74 2 68 6 67 7 60.0 54 9 C&W Comm 4.5 9.0 14.0 19.0 23.2 24 0 28.1 25 8 26.8 30 3 WorldCom 6.6 5.1 GlobalOne 3.1 1.5 ACC 3.O 3 6 Others 2.2 3.3 6.5 1 4 6 ~pan KDD 93.3 88.0 73.3 69.7 66.9 66.3 66 2 63.9 61 0 IDC 3.7 6.5 13.3 15.3 16.9 17.3 17.3 18.7 19 2 Japan Telecom 3.0 5 5 13.4 15 0 16.2 16 4 16 5 17.5 19.8 New Zealand TNZ 92.0 82~0 80.0 78.4 74.8 78.0 78.2 74.6 ClearCom 8,0 18.0 20.0 21.6 25 2 22.0 19.8 20 2 Others 2.0 5 2 Republic of Korea Korea Telecom 79.9 74.5 68 7 72.6 73.5 69 Dacom 20.1 25.5 31 3 27.4 26.5 27 Onse 4 Chile Entel Chile 80.0 57.5 40.0 40.6 37.3 33 Chdesat 20 0 25.0 19.7 19 4 15.2 17 VTR Telecom 1.0 17.5 10.2 10.3 9.3 11 CTC-Mundo 21.0 20 7 222 22 BelISouth CNle 6 6 6.8 10.0 10 lusatel 1 2 1 2.8 3 CNT 1 1 1 1 Transam 1 1 2.8 3 Philippines PLDT Phihppne Global Corn Eastern Telecom Capitol W~reless Bayan Tel Smart Digitel Philcom Islacom 91.6 84.2 69 8.4 15 8 23 7 1 68 23 6 1 1 1 78 71 6 3 5 7 1 1 4 5 1 2 2 3 2 7 1 1 Australia Telstra 98.0 87.0 76 3 73.4 62 0 55 Optus 2 0 13.0 21 9 23.4 27,0 26 AAPT 11 Primus 3 Others 1.8 3 2 11.0 5 B0 8 12 Canada (Canada-U.S. route only) Stentor AT&T Canada Long Distance Sprint Canada Fonorola ACC Others 83 8 13 9 3 4 93 2 57 8 15 12 4 4 ©TeleGeography, 56 9 16 7 5 7 Inc. 1998 © TeleGeography, Inc. 1998 TeleGeography 1999 Country/Carrier 1988 1989 Dominican Republic Codetel Tricom All America Cables and Radio, Inc. (AACR) Sweden Telia AB Tale-2 Others Percentage of Outgoing MiTT 1990 1991 1992 1993 1994 1995 1996 1997 >90 85.8 830 77,0 73.8 ha. 6.7 75 128 129 na 7.5 95 10.2 133 92 87 76 69 66 8 13 21 22 22 3 9 12 Finland Telecom Rnland Finnet International Telia Others 90 72.8 66.0 58.9 5 19,1 24.2 28 2 3 7.7 8.8 9 3 2 0.4 0.9 3 5 Indonesia PT Indosat PT Satelindo 99 5 95.4 89.5 84 8 0.5 4.6 11.5 14 2 Denmark Tale Danmark Tale2 Teha A/S Other 92.5 82 4.0 10 35 8 1 Malaysia Telekom Malaysia TRI Others 90 8 2 81 15 4 Mexico Telmex Avantel Alestra Others 84 0 75 75 1.0 Ireland Telecom E~reann Esat Telecommumcatzons WorldCom Ireland Others 91 5 3 1 Israel Bezeq Barak Golden Dnes 83 10 7 Netherlands PTT Telecom (KPN) Enertel Others Notes: MiTE is Minutes of Telecommumcat~ons Traffic Data based on outgozng rnterna- tzonal traffzc for the pubhc switched net~vork only covenng the full calendar or cal year. Some data aggregated ~n "Others" rows include market shares for car- riers shown zndzvidually ~n later years Market shares may not total to 100 percent due to roundzng. Umted States: Market shares for U.S. carners prior to 1993 exclude traffic to Canada and Mexzco; for the traffic base of second tmr U.S. carriers, see page 246 The 1996 figures for WorldCom reflect its acquzsztion of MFS. Umted Kzngdorn: Carriers’ traffic to ireland zs excluded prior to 1994 The figures for Cable & W~reless Comrnumcat~ons reflect data for Mercury prior to its April 1997 merger with Bell Cablernedia, Videotron, and NYNEX CableComrns. Market shares based on fiscal year reporting. Data for second tier carriers may ~nclude traffic refiled wa the U.K., thus overstating actual market shares of U.K.-ongmat- ed traffic. 95 3 2 Japan: The figures for Japan Telecom reflect data for ITJ prior to ets October 1997 merger w~th domestec long distance carrier Japan Telecom Co. Market shares based on fiscal year repo~ng and are for international Direct D~al traffic only. New Zealand: Market shares for New Zealand carriers prior to 1996 exclude resellars and are based on fiscal year reporting. Chile: The 1994 market shares for Chile are based on traffic for the month of December only Australia: Market shares for 1994 and 1995 are based on traffic for October to December quarters only and reflect wholesale minutes for facihSes-based carri- ers only. Market shares in 1996 and 1997 are from fiscal year ending 30 June. Indonesia: PT Satelindo began international service ~n September 1994. Netherlands. Cornpet~tors to P’iF Telecorn did not begin service until the second half of 1997 ©TeleGeography, Inc 1998 TeleGeo{~raphy 1999 © TeleGeography, Inc. 1998 The lop 40 International Carriers Outgoing Traffic 1987 Revenue (millions of minutes) (US$ billions) Rank Company Origin Country 1997 1996 Change 96-97 Total Int’l Service 1. AT&T (a) United States 10,290 9,459 8.8% $51.30 $5.63 2. MCI (a) United States 5,907 5,356 10.3% $19.65 $2.62 3. Deutsche Telekom (a) Germany 5,333 5,100 4.6% $37.72 $5.25 4. BT(a) United Kingdom 3,735 3,158 18.3% $25.87 $2.57 5. France T~l~com France 3,545 3,118 13.8% $26.17 $2.11 6. Sprint (a) United States 2,759 2,480 11.2% $14.87 $0.81 7. Telecom Italia Italy 2,352 2,184 7.7% $24.30 $1.80 8. Swissc om Switzerland 2,164 1,936 11.8% $6.95 $1.70 9. C&W Communications (a,c) United Kingdom 2,085 1,411 n.a. $3.78 $1.22 10. Stentor (d) Canada 1,778 1,650 7.8% $4.06 n.a. 11. Hong Kong Telecom (a,b) Hong Kong 1,718 1,739 -1.2% $4.21 $2.25 12. China Telecom China 1,632 1,433 13.9% $18.53 $4.10 13. KPN (a) Netherlands 1,535 1,534 0.1% $15.19 $2.20 14. WorldCom (a,e) United States 1,400 846 65.5% $7.35 $0.11 15. Belqacom (a) Belgium 1,340 1,228 9.1% $4.13 $0.55 16. Telefbnica (a) Spain 1,355 1,189 14.0% $15.58 $0.89 17. Singapore Telecom (a,b) Singapore 1,161 942 23.3% $2.95 $1.22 18. KDD (a) Japan 1,105 1,103 D.0% $3.09 $2.5 19. Teleglobe (a,f) Canada 1,112 915 21.5% $1.37 $0.78 20. Telmex (a) Mexico 1,009 1,071 -5.8% $7.53 $1.7 21. Telekom Austria (a) Austria 996 960 3.7% $5.30 $0.90 22. Rostelecom (a) Russia 939 851 10.3% $1.96 $0.31 23. Telstra (b) Australia 835 829 0.1% $11.29 $0.90 24. Chunghwa Telecom Taiwan 789 674 17.1% $5.13" n.a. 25. Telia Sweden 747 706 5.8% $5.90 $0.70 26. Pacific Gateway Exch. (a) United States 743 166 346.8% $0.30 $0.01 27. Etisalat United Arab Emirates 690 589 17.1% $1.03" n.a. 28. Saudi Com. Ministry Saudi Arabia 660 584 13.0% $2.11" n.a. 29. Telecom Eireann (a,g) Ireland 635 580 9.5% $1.94 $0.60 30. Korea Telecom Korea, Rep. of 610 520 17.3% $8.52 n.a. 31. OTE Greece 594 516 15.2% $2.87 $0.60 32. Telekom Malaysia (a,b) Malaysia 589 571 3.2% $7.17 $1.18 33. TUrk Telekomiinikasyon Turkey 558 473 17.8% $2.57* n.a. 34. Tele Danmark(a) Denmark 513 573 -10.5% $4.25 $0.36 35. Telekomunikacja Polska Poland 501 437 14.6% $2.36 n.a. 36. UTEL (a,h) Ukraine 487 341 n.a. $1.11" n.a. 37. Telenor Norway 481 444 8.5% $3.49 $0.60 38. Telebras (a) Brazil 459 367 25.1% $14.16 $0.90 39. WorldxChange United States 430 63 682.5% n.a. $0.04 40. VSNL(b) India 421 384 9.4% $1.35 $1.20 Note’ Traffic figures are for public switched network circuits only (service resale is excluded). Data for U.S. and U.K. carriers include International S~rnple Resale (ISR). International service revenues generally reflect net of revenues after adOng or subtracting for settlement payments All revenue figures con- verted from original currency at conversion rate current to year end reported Figures =n itahcs are est=rnates based on previous year serv=ce revenue ratios a. Data based on billing pmnt of call, not originating pmnt. b. Data are for the fiscal year ending 31 March. Telstra FY ends 30 June c. Data for C&W Comrnumcat~ons ~n FY 1996/97 include only traff=c carried by Mercury prior to its April 1997 merger with U K. cable compames. d. Stentor traffic is for the U.S. only, of which approximately 70 percent is originated by Bell Canada. e. 1996 WorldCorn data reflect data from MFS acquisition f. Teleglobe data are for Canada only The company originated a reported 952 million rn~nutes of traff=c outside Canada. g. Telecorn E=reann data for 1996 exclude traffic to Northern treland. h. UTEL 1996 data 1996 exclude traffic to countries outside the Commonwealth of Independent States. Revenue data for 1996. Source: TeleGeography, Inc. and company reports © TeleGeography, Inc. 1998 Pricing TeleGeography 1999 © TeleGeography, Inc. 1998 © TeleGeographyo Inc. 1998 TeleGeography 1999 Elements of an International Call Now that competition has finally arrived, what will happen to prices? This section answers that question by breaking down the cost ele- ments for completing an international call. Referring to the table on pages 60 to 61, let’s use a call from New York to Berlin as an exampte. Not including call-back, refile, and other forms of non- traditional traffic switching, a U.S. carrier has four basic methods of transporting its customer’s call to the destination in Germany: I. Carrier Settlement. To switch the call from the customer’s telephone to its own long distance network, the international carrier first pays the Local Exchange Carrier (LEC) in New York an origination fee of 1.6¢ per minute. Next, the carrier moves the call along a "backhaul" route; that is, from its national network to the undersea cable landing station on the Atlantic Ocean shoreline. The carrier shifts the call onto the interna- tional "half cimuit" it owns, then pays the German carrier a settlement fee to transfer the call onto its matching half circuit and to the final destination. The U.S. carrier’s marginal cost of using its own backhaul and international half circuit is insignificant: 0. I¢ per minute. The settlement rate, at I0¢ per minute, is not so cheap. Total cost: 12¢ per minute. 2. Carrier Interconnect. New competition rules in Germany permit foreign carriers to interconnect directly with the domestic telephone network. Rather than financing a half cir- cuit and paying a settlement fee, a U.S. carrier can purchase a whole circuit all the way to an international gateway in Germany, then pay the German carrier a 2.8¢ per minute fee to switch the call to Berlin. Total cost, including origination and backhauh 5¢ per minute. 3. International Simple Resale (ISR). A carrier is not required to own its own circuits. Instead, it can switch traffic onto U.S.- Germany private lines leased from other carriers. Total cost, including origination, backhaul, private line lease, and inter- connection in Germany: 6¢ per minute. 4. Service Resale. A telephone service provider may wish to avoid carrying its own traffic to Germany altogether. To offer its customers access to Berlin, the U.S. company can purchase the minutes transported over another carrier’s network in bulk, and market those minutes as its own. The two charges required for end-to-end service resale include a 4¢ per minute "wholesale rate" covedng origination and U.S. domestic long distance, plus a 7¢ per minute fee for the underlying carrier’s international transport and termination charges. Total cost: I I ¢ per minute. The following pages examine the component costs of provisioning an international call in more detail. Statistics on settlement rates and interconnection charges are also provided. Articles on tele- corn service and electricity trading examine the new models for purchasing minutes and bandwidth. Finally, tables and maps on retail prices review how carriers are passing along their costs to end users. Figure 1. International Call Components Countty A Count~ B Half Circuit Local National Backhaul International Exchange Transport Transport Network Elements Covered by Settlement Rate H ai.-" Clrc air Nadonai, National Sw~tch N e~,~,o rk Network Elements Covered by Settiement Rate © TeteGeography, Inc, 1999 TeleGeo~raphy 1999 © TeleGeographyo Inc. 199~ International Call Costs for U.S. Carriers Per Minute Cost (U.S, Cents) U.S.-End Int’l Foreign-End International Origination Half Circuit Half Circuit Whole Circuit Settlement Interconnect Wholesale Cost Ownership Ownership Lease Rate Rate Rate Americas U.S.-Canada (Toronto) Retail Price -- Carrier Settlement 1.7 Carrier Interconnect 1.7 ISR 1.7 Wholesale for resellers 4.0 U.S.-Mexico (Mexico City) Retail Price -- Carrier Settlement 1.8 Carrier Interconnect 1.8 ISR 3.2 Wholesale for resellers 4.0 Europe U.S.-Germany Retail Price -- Carrier Settlement 1.7 Carrier Interconnect 1.7 ISR 1.7 Wholesale for resellers 4.0 U.S.-U.K. Retail Price -- Carrier Settlement 1.7 Carrier Interconnect 1.7 ISR 1.7 Wholesale for resellers 4.0 Total ...... $O.lZ 0.2 -- -- 10.0 -- -- $0.12 0.2 0.2 -- -- 5.3 -- $0.07 -- -- 1.3 -- 5.3 -- $0.08 ..... 5.0 $0.09 ...... $0.44 0.5 -- -- 35.0 -- -- $0.37 0.5 0.5 -- -- 5.7 -- $0.09 -- -- 1.2 -- 5.7 -- $0.10 ..... 18.8 $0.23 ...... $0.27 0.1 -- -- 10.0 -- -- $0.12 0.1 0.1 -- -- 2.8 -- $0.05 -- -- 1.9 -- 2.8 -- $0.06 ..... 7.0 $0.11 ...... $0.12 0.1 -- -- 10.0 -- -- $0.12 0.1 0.1 -- -- 1.9 -- $0,04 -- -- 1.6 -- 1.9 -- $0.05 ..... 5.0 $0.09 Notes: All cost components expressed in U.S. cents and are excluswe of taxes. Cost totals expressed =n U.S. dollars. Rates are based on ~nternational calls originating from Washington, D.C. only. Data do not reflect costs for international calls terminating m the U.S. Actual carrier costs may vaN. 0rig=nation cost includes access charges paid to Local Exchange Carrier (Bell Atlant=c) and U.S. domestic network costs for transmittzng calls to international gate- way. Assumptions for calculating per minute costs of using international networks include: Each 64 kbps circu=t is compressed at a 5:1 ratio and is used for ten years. Each voice path is used 4 hours 1240 m=nutes) per day. Settlement rates are for peak rate traffic terminated by largest foreign carrier. Direct interconnection by formgn carriers to the domestic public switched network is not permitted =n China and India Retail rates are based on the MCI One International Calhng Plan. All rates current to July 1998. © TeleGeography, Inc. 1998 TeleGeography 1999 Per Minute Cost (U.S. Cents) U.S.-End Int’l Foreign-End International Origination Half Circuit Half Circuit Whole Circuit Settlement Interconnect Wholesale Cost Ownership Ownership Lease Rate Rate Rate Total Asia U.S.-Australia Retail Price ....... $0.47 Carrier Settlement 1.8 0.3 -- -- 15.0 -- -- $0.17 Carrier Interconnect 1.8 0.3 0.3 -- -- 1.6 -- $0.04 ISR 2.7 -- -- 7.3 -- 1.6 -- $0.12 Wholesale for resellers 4.0 ..... 7.0 $0,11 U.S.-China Retail Price ....... $1.31 Carrier Settlement 1,7 0.6 -- -- 68.0 -- -- $0.70 Carrier Interconnect ... n.a, ISR n.a. Wholesale for resellers 4.0 ... 42.0 $0.46 U.S.-India Retail Price ....... $1.22 Carrier Settlement 1,7 0.6 -- -- 79.0 -- -- $0.81 Carrier Interconnect ... n.a. ISR n.a. Wholesale for resellers 4.0 ... 60.0 $0.64 U.S,-Japan Retail Price ....... $0.35 Carrier Settlement 1.7 0,1 -- -- 14.0 -- -- $0.16 Carrier Interconnect 1.7 0.1 0.1 -- -- 6.0 -- $0.08 ISR 1.9 -- -- 11.1 -- 6.0 -- $0.19 Wholesale for resellers 4.0 ..... 18.0 $0.22 Sources Federal Communications Commission, MCI WorldCom, Inc., Star Telecommumcatmns, Inc., Band-X, Ltd.; TAT 13 and SEA ME WE 3 cable marke~ng; Global Crossing, Ltd ; concept TeleGeography, Inc © TeleGeography, Inc. 1998 TeleGeography 1999 © TeleGeography, Inc. 1998 The Accounting Rate Regime in Transition Accounting rates provide a common method of compensating originating and terminating carriers for carrying telephone calls over both networks. The system is largely transparent to users and generally works as follows: 1. International carriers negotiate accounting rates on a route-by-route basis. A per minute rate is agreed for land- ing traffic in either direction. Theoretically, this rate is based on the sum of both carriers’ costs, although the cost-linkage is often quite loose. The rate is commonly stated in U.S. dol- lars or Special Drawing Rights (SDRs), a monetary unit whose value reflects a basket of major currencies. 2. On any given route, one carrier pays settlements to another carrier only to the extent that there is a traffic imbalance--that is, one carrier has terminated a greater vol- ume of telephone minutes than the other carrier. The origi- nating and terminating carriers usually divide the accounting rate 50/50 to determine the per minute settlement rate. 3. A carrier’s net revenue for international service is a func- tion of the accounting rate as well as the collection charge /see Figure 1). If traffic is balanced, the value of the accounting rate is essentially irrelevant since no settlement is necessary and each carrier’s revenue wilt depend directly on its collection charge. 4. Where traffic is imbalanced, the accounting rate may have a significant effect on the commercial options of the two carriers. If a carrier has a significant traffic deficit, the Figure 1. How Accounting Rates Work Country A sends 100 i Country B sends 150 minutes to Country B minutes to Country billed at $1 00/m~n I bdled at $0 80/m~n Seffiement rate ~s $025/mm I (accounting rate of$O 50/m=n spi~t 50/50) Courmy A Co/lec~sStO000 Pays $2500to Count~B Rece~ves $37.50 from Country B RetamsSt12.50 Country B has a defimt 1 Co/lects$120.O0 Pays$3750 to CourmyA Source: Adapted from Direction of Traffic 1996, ITU/TeleGeography, Inc. © TeleGeography, Inc. 1998 62 settlement payments which it must make to its foreign cor- respondent limit its ability to reduce its collection charges. Conversely, a carrier with a net traffic surplus has little incen- tive to operate more efficiently or to reduce the accounting rate because of the net settlement benefits it receives under the status quo. Few international carriers are entirely satisfied with the present accounting rate regime. Some want it abolished outright so that they can strike the best commercial deals possible. Others are content to exploit the system (e.g., by supporting call-back carriers or refiling traffic through third countries) so as to lower rates while profiting along the way (see Figure 2). Even carri- ers with large settlement surpluses know reform is essential, if only to ensure a "soft landing" rather than an abrupt fall-off in settlement payments. FCC Settlement Rate Benchmarks Since 1996, spurred by the $8 billion annual settlement out- flow of U.S. carriers, the U.S. Federal Communications Commission (FCC) has been the most active proponent of set- tlement rate reform. The centerpiece of the FCC’s policy is an August 1997 order establishing a set of "benchmark" or model settlement rates which all U.S. international carders must respect beginning in 1999 (see pages 68 to 69). The bench- mark rates range from SO. 15 to $0.23 per minute--often 50 percent or more below the current rate---and will be phased in (see Figure 3). U.S. carders must negotiate benchmark rates with carders from richer countries by January 1999; bench- mark rates must be implemented with carders from middle income and poorer countries between 2000 and 2003. While these new benchmarks primarily apply to public switched tele- phone services, the FCC’s order also prohibits International Simple Resale [ISR) on a route unless 50 percent of the U.S. outbound traffic is already settled at or below the benchmark rate. The FCC calculated its new benchmarks using a controversial model for estimating foreign carriers’ actual costs in terminat- ing U.S. calls. The estimates are based on the per minute tar- iff or tariff proxy for three foreign network components: the international transmission facility (cable/satellite half-circuit); the international gateway facility; and the national extension (domestic transport and local termination). Tariff component prices (TCPs) were calculated for 65 countries and these coun- tries were then divided into four economic groups based on their 1995 Gross Domestic Product (GDP) per capita. The © TeleGeography, Inc. 1998 TeleGeography 1999 Figure 2. The Logic Behind Refile U.S. and U.K. Settlement Rates to Selected Countries, July 1998 $1.50 $1.25 $1.00 $0.75 $0.50 $0.25 ~A~,alPa~i~ic [] Middle East & Africa O Europe Cheaper from Cheaper from U.S. Venezuela $0.00 $0.00 $0.25 $0.50 $0.75 $1.00 $1.25 $1.50 U.K. Settlement Rate ($/mln) Source: FCC, Office of Telecommunications (OFTEL), TeleGeography, Inc. Each marker represents the U.t~ and U.S. settlement rate to one international call destination. If U.S. se~lement rates exactly tracked U.K. rates, destination coordinates would align along the 45° diagonal line. Cleady, the 45° rule fails to hold. Notice, for example, how settlement rates to countries in the Americas tend to be significantly cheaper from the U.S. than from the U.K. A refiler can use these price differences to its own advantage. In July 1998, the actual settlement rate with Venezuela was 40¢ from the U.S. and 82¢ from the U.K. By char~ing British carders a rate somewhere between the U.S. and U.K. costs-- say, 70¢--to send traffic to Venezuela, a refiler could turn a hefty profit. Imagine that the refiler’s costs equal 60¢ per minute---5€ to transfer traffic from the U.K. to the U.S., 15¢ to move the call from the U.S. to Venezuela, and 40¢ to land the call abroad. In our hypothetical example, the British customer saves money by refiling, and the refiler reaps a l 0¢ per minute profit. It’s a shame about the Venezuelan carrier, though--it earns 42¢ less settlement revenue than if the U.K. carrier had sent traffic directly. © TeleGeography, Inc. 1998 average TCP for each economic group was then adopted by the FCC as the benchmark for all countries within a given income category (again, see Figure 3). Despite the one to five year transition period for U.S. carriers, on January 1, 1998, the FCC began to apply its new settlement benchmarks to foreign carriers wishing to enter the U.S. mar- ket. Since then, any foreign-affiliated U.S. carrier seeking to provide telephone service to its home country must satisfy the applicable benchmark. Foreign affiliated carriers already oper- ating in the U.S., however, were Oven a reprieve pending fur- ther review. Some foreign-affiliated U.S. carriers have asked the FCC to waive the benchmarks unless the U.S. carder con- trois 25 percent or more of U.S. traffic on a route or its affiliate controls bottleneck facilities at the foreign end. More than a dozen foreign carders have challenged the FCC’s benchmark rules in the U.S. courts. They argue, among other things, that the agency overstepped its jurisdiction because the benchmarks amount to a de facto price cap on the termination charges of foreign operators. The case was argued in September 1998 and a decision is expected by January 1999. Most observers expect the FCC’s order to be upheld, although the court may direct the agency to reconsider the level of the benchmarks for some countries. Even if the FCC’s order is upheld by the courts, lower settle- ment rates must still be negotiated by U.S. carriers. The FCC’s order is not self executing; it only holds out the threat of FCC intervention if private carrier-to-carrier negotiations fail to pro- duce more cost-based settlement rates. Since August 1997 though, U.S. carriers have had considerable success in negoti- ating lower rates. As of September 1998, U.S. rates were at or below the SO. 15 settlement benchmark on approximately 20 routes reflecting over 40 percent of U.S. outbound traffic. And carders in several other foreign countries (notably, Mexico, Egypt, Turkey, Venezuela, Korea, Dominican Republic, and South Africa) had agreed to meet the benchmarks within the FCC’s stated time frame. In the year since the benchmarks order was adopted, the FCC has also taken other steps to place economic pressure on car- Figure 3. FCC Settlement Benchmarks (US$) Income Group GDP per capita Benchmark Effective Date Low 0-$726 $0.23 1 Jan. 2002 Lower Middle $726-$2,985 $0.19 1 Jan. 2001 Upper Middle $2,896-$8,955 $0.19 1 Jan, 2000 High $8,956 $0.15 1 Jan. 1999 For countries with less then 1 telephone hne per 100 people, the effectzve date is I Jan. 2003. Wa=vers of the effective date may also be requested ~n certain cases. Source: FCC © TeleGeography, Inc. 1998 TeleGeography 1999 © TeleGeography, Inc. 1998 Figure 4. FCC Benchmark Deadlines Approaching $1.00 $0.75 $0.50 $0.25 ¯ 1997 Settlement Rat~ [] 1998 Settlement Rate [] FCC Benchmark Rate $0.00 Germany Japan Singapore Brazil Mexico Dominican Indonesia Philippines China India Republic Jan. 1999 Deadline Jan. 2000 Deadline Jan. 2001 Deadline Jan. 2002 Deadline Note: Settlement rates for 1997 are an average for the entire year; t998 rates are current to July 1998. Source: TeleGeography, Inc. ©TeleGeography, Inc. 1988 Hers which persist in maintaining significantly above-cost set- tlement rates. First, as soon as the benchmark rate has been met on a given U.S. route, the FCC has opened it to ISR. Because ISR traffic uses international private lines, it bypasses the accounting rate regime. Thus, each of the 12 new U.S. routes opened to ISR since 1997--routes include Japan, France, Germany, and Italy--has placed downward pressure on the settlement rates char~ed to neighboring destinations. Second, in August 1998, the FCC proposed to permit U.S. car- riers dispense with accounting rates altogether when exchang- ing international traffic with a non-dominant foreign carrier (see www.fcc.gov/Bureaus/International/Notices/1 996/ FCC98190.pdt~. The FCC’s proposed rules would also permit U.S. international carriers to negotiate the best termination rates they can with any foreign carrier in a country where 1SR is permitted. In addition, the FCC has proposed to allow U.S. car- tiers to offer ISR services to countries which do not meet the FCC’s benchmark rates if the service would only involve a com- paratively small percentage of traffic. These FCC proposals have been widely supported by both U.S. and foreign ca~ers, and are likely to be adopted by Spring 1999. The economic thrust of these FCC reforms was given a further boost on October 1, 1998 when the Canadian Radio-Television Commission (CRTC) adopted a "light handed" set of settlement Figure 5. ITU Settlement Benchmarks Settlement Rate Categories (US$) Country Teledensity Benchmark Example (Phone Lines per 100 people) Rate Countries >50 0.043 SDRs ($0.0571 Germany, U.S. 35-50 0.088 SDRs ($0.118) Belgium, Italy 20-35 0.118 SDRs ($0.158) Bahrain, Hungary 10-20 0.162 SDRs ($0.216) Chile, South Africa 5-10 0.210 SDRs ($0.281) Mex=co, China 1-5 0.251 SDRs ($0.335) India, Philippines 1 0.327 SDRs ($0.437) Bangladesh, Ghana Source: tTU ITU and FCC Benchmark Settlement Rates for Selected Countries [] FCC Benchmark Rate [] ITU Benchmark Rate $0.00 India Ch=na Mexico Argentina Japan France © TeleGeography, Inc. 1998 © TeleGeography, Inc. 1998 TeleGeo~raphy 1999 rules for new competitive international carriers in Canada. Under the new rules, an unlimited number of carriers will be permitted to compete with the Canadian incumbent, Teleglobe. Settlement terms will be a matter for commercial negotiation, subject to general competition rules; there will be no settlement benchmarks and ISR will be permitted on any route where it is lawful at the foreign end (see www.crtc.gc.ca/ENG/telecom/deci- sion/1998/d9817-o.txt). Thus, given that ISR is already per- mitted on the U.S.-Canada route, the CRTC’s action gave both U.S. and Canadian carriers a significant new option for raffling traffic to and from "high cost" overseas carriers. Enter the ITU A focus group within the International Telecommunication Union (ITU), the rules-making body for global telecommunications, is considering its own proposals for settlement rate benchmarks (see www.itu.int/intset/focus/index.html). The 1TU’s plan differs from the FCC benchmarks policy in a number of ways: 1. Benchmarks are assigned to a country according to that country’s teledensity (number of telephone lines per 100 inhabitants) rather than GDP (see Figure 5). The ITU pro- posals include an additional benchmark category (39 U.S. cents per minute) for small island nations. The ITU targets are based on current published settlement rates for 224 countries, with the target rate established as the average of the lowest 20 percent in each category. 2. FCC benchmarks may be phased in from 1999 to 2002, depending on a country’s per capita GDP. ITU recommen- dations set December 31, 2001 as the date for achieving target settlement rates, but propose a longer transition peri- od-up to the year 2005--for countries where net settle- ment payments constitute a significant source of telecom- munications revenue. 3. The ITU’s proposed rates cover a much wider range than the FCC’s prescribed band--from 6 to 44 cents per minute compared to the FCC’s 15 to 23 cents. 4. While the FCC benchmarks relate to fewer than 250 out- going mutes from the U.S., ITU target rates cover more than 40,000 international mutes. Settlement rates between two countries in different teledensity categories would be set at or below the target rate for the country with lower teleden- sity. The ITU proposals also suggest the use of asymmetric accounting rates--with a variation by a few percentage points from the traditional 50/50 division--for countries with low teledensities. The ITU focus group has also offered target rates for routing traffic through third countries. Transit rates would vary accord- ing to the amount of telephone traffic sent over a route, rang- ing from 4 cents per minute on routes with more than 1.5 mil- lion minutes of international traffic to 6 cents per minute for routes with less than 350,000 minutes each year. Figure 6. Termination Rates Compared Australia (Telstra) Finland Japan(KDD) ¯ Settlement Rate [] Interconnectmn Rate Sweden U.K. (BT) $0.00 $0.05 $0.10 $0.15 $0.20 $025 Termination Rate from United States ($/mln) Note: Rates valid as of July 1998. Interconnection rate includes domestic tong d=stance and local switching, but excludes cost of foreign half circuit. Source: EU Commission, FCC © TeleGeography, Inc. 1998 Despite attempts to reach consensus, the ITU’s proposals may have little effect. The FCC’s regulations have teeth because most countries enjoy a traffic surplus with the U.S. and hence any rules limiting U.S. outpayments are almost certain to bite. In contrast, the ITU cannot exert financial pressure to enforce compliance; target rates remain subject to the bilateral agree- ment by both parties on an international route. Whether car- ders will choose to adopt these rates remains to be seen. Direct Interconnection Reform efforts notwithstanding, settlement rates are of declin- ing importance to many carriers because competitive markets typically offer the option of obtaining access to the "foreign" phone network directly. Where market entry has been liberal- ized, instead of paying settlement charges, many carriers pre- fer to acquire a whole circuit to the destination country and pay a domestic interconnection fee (see Figure 6). Fees are often unbundled so that the more local facilities a foreign carrier owns, the tess it has to pay for the "last mile." The table on page 70 illustrates how these pricing bands apply. Regulatory action is reinforcing the movement toward low inter- connect fees. For example, the Commission of the European Union (EU) has authorized national regulatory authorities in the fifteen EU member states to order lower interconnect fees charged by incumbent carriers if the fees they charge fail to reflect their actual costs. The EU Commission has also pub- lished recommended "best practice" rates to encourage cost transparency (see www.ispo.cec.be/infosoc/telecompolicy). TeleGeography 1999 © TeleGeography, Inc. 1998 International Settlement Rates United States Destination 1996 1997 1998 Andorra 0.31 0.29 0.28 Argentina 0.72 0.63 0.38 Australia 0.23 0.21 0.15 Austria 0,22 0.21 0.14 Bahamas 0.30/0.15 0.30/0.15 0.30/0.15 Bahrain 0.80 0.80 0.72 Bangladesh 1.00 1.00 0.80 Belarus 0.60 0.50 0.43 Belgium 0.28 0.20 0.14 Bolivia 0.63 0.60 0.48 Brazil 0.52 0.52 0.35 Canada 0.12/0.10 0.12/0.10 0.12/0.10 Chile 0.50 0.55 0.55 China 1.07 0.89 0.68 Colombia 0.63 0.59 0.50 Costa Rica 0.58 0.58 0.40 Croatia 0.51 0.34 0.27 Cyprus 0.65 0.55 0.47 Czech Republic 0.36 0.34 0.27 Denmark 0.15 0.14 0.11 Dominican Republic 0.45 0.40 0.40 El Salvador 0.55 0.55 0.44 Finland 0.26 0.25 0.20 France 0.18 0.13 0.10 French Polynesia 1.25 1.25 1.25 Germany 0.12 0.11 0.10 Ghana 0.50 0.50 0.50 Greece 0.51 0.48 0.36 Guyana 0.85 0.85 0.85 Hong Kong 0.47 0.40 0.35 Hungary 0.51 0.42 0.27 I c eland 0.47 0.44 0.37 India 0.80 0.79 0.79 Indonesia 0.70 0.65 0.53 Iran 1.50/1.25 1.50/1.25 1.05 ireland 0.18 0.17 0.16 Israel 0.59 0.59 0.30 Italy 0.26 0.!9 0.11 Japan 0.46 0.43 0.14 Jordan 0.75 0.75 0.75 United Kingdom 1996 (US$) 199e (os$) 0.26 0.22 0.85 0.86 0.27 0.17 0.24 0.20 0.33 0.38 1.04 0.82 1.04 0,99 0.34 0.35 0.15 0.10 0.91 0.90 0.70 0.49 0.14 0.08 0.94 0.90 1.14 1.08 0.88 0.90 1.02 0.69 0.28 0.33 0.31 0.25 0.21 0.21 0.10 0.10 0.58 0.67 1.30 1.54 0.18 0.15 0.12 0.11 1.39 1.64 0.10 0.08 0.56 0.66 0.30 0.29 0.84 0.90 0.42 0.45 0.17 0.18 0.26 0.23 0.96 0.95 1.07 1.21 1.02 1.21 0.13 0.16 0.35 0.25 0.27 0.16 0.63 0.59 1.02 1.21 Notes’ 1, All rates expressed in US$ Equ=valent dollar values are presented for accounting rates that are established =n Special Drawing Rights (SDRs), gold francs, or pounds sterling. 2. The average U.S, accounting rate for 1996 and 1997 ~s weighted by the total minutes between the U.S. and each location ~n that year. U.K. 1996 rates date to October 1996, U.S. and U.K. 1998 rates current to July 1998. 4 U.K. 1996 rates some’ames appear lower than July 1998 rates due to the lower 1998 value of the U.S. dollar against pound sterling. 5. Where two rates are shown, there are peak/off-peak rates or growth-based rates (traffic above a benchmark level is eligible for a lower rate). B, Rates are for largest carrier serving the route. D~fferent accounting rates may apply to competing carriers. © TeleGeography, Inc. 1998 TeleGeography 1999 United States Destination 1996 1997 1998 Kazakhstan 1.05 1.05 0.67 Korea, Republic of 0.62 0.49 0.43 Kuwait 0.84 0.78 0.77 Luxembourg 0.29 0.21 0.14 Macau 0.68 0.68 0.60 Malaysia 0.45 0.45 0.40 Mexico 0.34 0.35 0.35 Moldova 1.04 1.04 1.00 Netherlands 0.18 0.17 0,10 New Zealand 0.22 0.14 0.14 Norway 0.15 0.14 0.11 Oman 1.20 1.13 1.11 Pakistan 1.10/0.70 1.00/0.60 0.90/0.50 Panama 0.63 0.60 0.51 Paraguay 0.73 0.73 0.50 Peru 0.62 0.57 0.43 Philippines 0.50 0.50 0.41 Poland 0.48 0.35 0.35 Portugal 0.42 0.34 0.24 Russia 1.06 1.06 0.40 Saudi Arabia 1.10 1.01/0.81 1.00/0.80 Singapore 0.45 0.42 0.36 Slovak Republic 0.65 0.42 0.34 Slovenia 0.36 0.34 0.34 South Africa 0.50 0.50 0.40 Spain 0.32 0.30 0.14 Sri Lanka 1.00 1.00 1.00 Sweden 0.09 0.08 Switzerland 0.28 0.19 0.14 Taiwan 0.60 0.60 0.50 Thailand 0.75 0.75 0.53 Turkey 0.58 0.55 0,37 Ukraine 0.70 0.65 0.55 United Arab Emirates 1.00/0.65 1.00/0.65 1.00/0.65 United Kingdom 0.18/0.11 0.10/0.07 0.10/0.07 United States n.a. n.a. n.a. Uruguay 0.85 0.85 0.45 Uzbekistan 0.85 0.85 0.85 Venezuela 0.50 0.50 0.40 Vietnam 1.15/1.00/0.93/0.85 1.15/1.00/0.93/0.85 1.15/1.00/0.93/0.85 Yugoslavia 0.58 0.56 0.37 United Kingdom 1996 (US$) 1998 (US$) 1.10 0.82 0.70 0.64 0.70 0.82 0.20 0.23 0.56 0.53 0.50 0.51 0.56 0.45 0.50 0.30 0.20 0.07 0.35 0.20 0.17 0.07 0.77 0.82 1.11 0.66 1.02 0.77 0.90 0.90 0.90 0.74 0.77 0.49 0.26 0.28 0.29 0.23 0.36 0.37 0.91 1.27 0.50 0.59 0.20 0.19 0.24 0.16 0.90 0.66 0.29 0.16 0.77 0.90 0.10 0.12 0.14 0.07 0.77 0.57 0.84 0.82 0.27 0.32 0.95 0.31 1.02 0.49 n.a. n.a. 0.17 0.08 0.82 0.97 0.97 0.99 0.82 0.82 1.46 1.31 0,33 0.28 Source" FCC; Office of Telecommumcatlons (OF’TEL) © TeleGeography, Inc 1998 TeleGeography 1999 © TeleGeography, Inc. 1998 FCC Settlement Benchmarks Benchmarks Methodology These Tariffed Component Prices (TCPs) were calculated by the staff of the Federal Communications Commission (FCC) and were used to derive average benchmark settlement rates for U.S. international tele- phone carriers in the FCC’s Report and Order IB Docket No. 96-261, FCC 97-280, released August 18, 1997 (Benchmarks Order). Implementation of the Order is staggered over several years, accord- ing to national incomes, from January 1, 1999 for high income coun- tries to January 1, 2003, for low income countries (see "The Accounting Rate Regime in Transition" on pages 62 to 65). The TCP for each country is derived from the prices for the three net- work elements used to provide international phone service as identi- fied by Recommendation D.140 of the International Telecommunication Union Telecommunication Standardization Sector (ITU-T). These elements are: (1) International transmission facilities (cable/satellite half circuits); (2) International switching facilities; (3) National extension (domestic transport and termination). The FCC used 1996 tariff rates for the largest carrier in each country to calculate the price for the international transmission and national extension elements. For the international transmission portion, the FCC used the rate for a high capacity (1.5 Mbps or more) interna- tional private line, assuming 4/1 compression on each 64 kbps circuit, and a usage level of 8,000 minutes per 64 kbps circuit per month. For the national extension, the FCC relied upon national long distance tariffs, making some adjustments for the expected distribution of inbound traffic by time of day and distance. The per minute cost of the international switching element was dedved from the accounting rate share figures stated in the ITU-T Recommendation D.300R for the international exchange component. Details on the FCC’s methodology can be found in Appendix E to the Benchmarks Order. See also the December 1996 "Foreign Tariffed Components Prices" report prepared by the FCC’s International Bureau, at Appendices C and D, which contains the relevant interna- tional private line and domestic long distance tariffs. Tariffed Component Prices for FCC Benchmarks International International National Tariffed Component FCC Settlement Country Transmission (US¢) ÷ Switching (US¢) Extension (US¢) Price (US¢ Total) Benchmarks(US¢) Upper Income Bracket: Effective 1 January 1999 Australia 4.8 1.9 12.0 18.7 15.0 Austria 8.1 1.9 21.4 31.4 15.0 Bahamas 5.2 1.9 12.8 19,9 15.0 Belgium 3.0 1.9 9.2 14.1 15.0 Denmark 5.9 1.9 6.6 14.4 15.0 France 2.9 1.9 12.7 17.5 15.0 6ermany 4.3 1.9 13.6 19.8 15.0 Hong Kon9 5.1 1.9 0.0 7.0 15.0 Ireland 2.7 1.9 13.4 18.0 15.0 Israel 4.2 1.9 2,4 8,5 15.0 Italy 4.8 1.9 11.5 18.2 15.0 Japan 6.5 1.9 11.3 19.7 15.0 Kuwait 7.1 1.9 0.0 9.0 15.0 Netherlands 2.6 1.9 5.3 9.8 15.0 New Zealand 5.7 1.9 16.2 23.8 15.0 Norway 3.2 1.9 6.5 11.6 15.0 Portugal 4.6 1.9 17.4 23.9 15.0 Singapore 5.0 1.9 0.7 7.6 15.0 Spare 4.8 1.9 11.4 18.1 15.0 Sweden 3.6 1.9 4.5 10.0 15.0 Switzerland 4.4 1.9 14.3 20.6 15.0 Taiwan 5.7 1.9 6,3 13.9 15.0 Umted Arab Emirates 3.3 1.9 2.5 7.7 15.0 United Kingdom 2.4 1.9 82 13.0 15.0 © TeleGeography, Inc. 1998 TeleGeog~phy 1999 Tariffed Component Prices for FCC Benchmarks (continued) International International National Tariffed Component Country Transmission (US¢) Switching (US¢) Extension (US¢) Price (US¢ Total) FCC Settlement Benchmarks (US¢) Upper Middle Income Bracket: Effective 1 January 2000 Argentina 6.7 3.4 22.0 32.1 19.0 Barbados 8.6 3.4 0.0 12.0 19.0 Brazil 6.6 3.4 17.8 27.8 19.0 Chile 2.9 3.4 12.3 18.6 19.0 Czech Republic 8.1 3.4 7.5 19.0 19.0 (3reece 5.2 3.4 14.4 23.0 19.0 Hungary 6.1 3.4 4.9 14.4 19.0 Korea 5.1 3.4 4.3 12.8 19.0 Malaysia 6.6 3.4 12.4 22.4 19.0 Mexico 0.9 3.4 12.5 16.8 19.0 South Africa 5.2 3.4 8.3 16.9 19.0 Trinidad 3.6 3.4 7.6 14.6 19.0 Uruguay 12.7 3.4 6.2 22.3 19.0 Lower Middle Income Bracket: Effective I January 2001 Colombia 5.1 4.8 8.6 18.5 19.0 Costa Rica 3.3 4.8 2.2 10.3 19.0 Dominican Repubhc 3.6 4.8 6.1 14.5 19.0 Ecuador 2.9 4.8 2.6 10.3 19.0 El Salvador 5.9 4.8 1.1 11.8 19.0 Guatemala 3.1 4.8 2.4 10.3 19.0 Indonesia 6.8 4.8 23.9 35.5 19.0 Jamaica 2.9 4.8 1.0 8.7 19.0 Jordan 15.9 4.8 2.3 23.0 19.0 Panama 4.7 4.8 9.9 19.4 19.0 Peru 5.8 4.8 5.5 16.1 19.0 Philippines 6.5 4.8 12.6 23.9 19.0 Poland 4.7 4.8 15.1 24.6 19.0 Russia 5.4 4.8 25.2 35.4 19.0 Thailand 4.0 4.8 8.3 17.1 19.0 Turkey 5.4 4.8 7.7 17.9 19.0 Venezuela 3.7 4.8 15.3 23.8 19.0 Lower Income Bracket: Effective 1 January 2002 China 8.7 4.8 4.2 17.7 23.0 Egypt 10.4 4.8 2.0 17.2 23.0 Guyana 6.6 4.8 0.6 12.0 23.0 Haiti 8.6 4.8 17.0 30.4 23.0 Honduras 3.1 4.8 8.7 16.6 23.0 India 8.1 4.8 18.3 31.2 23.0 Kenya 25.5 4.8 12.3 42.6 23.0 Nicaragua 3.8 4.8 3.7 12.3 23.0 Pakistan 14.7 4.8 7.2 26.7 23.0 Vietnam 9.3 4.8 10.6 24.7 23.0 Source: FCC © TeleGeography, Inc. 1998 TeleGeography 1999 © TeleGeography, Inc. 1998 National Interconnection Rates Local Termination Regional Termination National Termination (US cents) (US cents) (US cents) Australia 1.62 5.30 9.90 Austria 2.00 2.00 2.63 Belgium 1.23 2.33 3.26 Canada (Toronto) n.a. 5.33 n.a. Denmark 1.09 2.02 2.46 Finland 1.56 1.58 3.12 France 0.78 1.90 2.80 Germany 1.10 1.88 2.86 (3reece 2.01 2.01 2.87 Italy 1.68 2.74 n.a. Ireland 2.44 4.61 5.75 Japan 0.71 1.73 5.98 Luxembourg 2.23 2.23 2.23 Netherlands 1.30 1.78 2,29 Portugal 1.33 2.63 19.98 Spain 1.65 1.65 4.63 Sweden 1.27 1.96 2.68 Switzerland n.a. 2.72 3.73 U.K. 0.68 0,96 1.88 U.S. (Washington, D.C.) n.a. 0.79 n.a. Notes: All interconnectzon charges are for peak perzod and are current to July 1998. Local termination ~s the lowest level of interconnection, typically g~wng a carrier access to a single town or part of a city. Regional termination generally gzves a carrzer access to all subscribers w~thzn a metropohtan area or a North American area code U.S. termznation fees vary according to Local Exchange Carrier (LEC). U.S. national average was 1.49¢ as of July 1998. Source: European Umon Commission, FCC, CRTC (Canada), MPT (Japan), OFCOM (Switzerland) © TeleGeography, Inc. 1998 © TeleGeography, Inc. 1998 TeleGeography 1999 Trading Minutes and Bandwidth Telecom wholesale markets--where carriers buy and sell min- utes and capacity from each other--can save telephone com- panies substantially. Rather than building links to numerous countries, a start-up operator can cut costs by purchasing other carders’ bandwidth on a route-by-route, as needed basis. And large carriers benefit from reselling capacity on their networks that otherwise would have lain idle. Internet-based "bandwidth exchanges," which permit tetcos to buy and sell network access online, bring a new level of effi- ciency to wholesale telecom markets. This article takes a nuts- and-bolts look at how bandwidth markets operate. Will these markets soon make telecom capacity a global commodity. Perhaps. But bandwidth exchanges must first overcome some significant obstacles before they can live up to their potential. How They Work Numerous bandwidth exchanges have sprung up on the lnternet (see Figure 2). Band-X, Ltd., which began service in July 1997, was among the first. Rather than serving as a spot market, the London-based exchange first saw the light of day as an lnternet dating service of sorts for telcos, providing a cen- tral meeting place for introducing potential buyers and sellers of international bandwidth and wholesale minutes. Here’s how a sale is conducted using the standard Band-X service: 1. A carrier wishing to sell minutes carried on its network posts an offer on the Band-X web site. The seller provides details on the origin, destination, number of minutes avail- able, and per minute rate. The carder’s identity remains undisclosed. 2. A buyer interested in the carder’s offer e-mails a response to Band-X management. 3. Band-X management introduces buyer and seller. 4. On their own, buyer and seller establish the details for interconnecting their networks and determine how many minutes are to be sold. 5. If an agreement is reached, the seller pays Band-X a com- mission within seven days of receiving payment. The com- mission rate varies according to the size of the sale~2.25 percent on the first $200,000 invoiced, 1.125 percent on Figure 1. Buy Your Minutes Online A Login to Band-X B. Review latest bids and offers C. Call up details on routes of interest D. Ernail Band-X Source: Bond-X, inc. (vwvvv.band-x.com) © TeleGeography, Inc. 1998 TeleGeography 1999 © TeleCeography, Inc. 1998 the next $200,000, and 0.625 percent on all payment thereafter 6. Upon receipt of commission, Band-X passes on 25 per- cent of its fee to the buyer. By holding out the promise of this sum, Band-X gives the buyer an incentive to see that the seller carries through with payment of commission. When a buyer posts a bid to purchase minutes or capacity, the process works in reverse--the buyer pays commission and sell- er receives the 25 percent refund. In addition to offedng its standard service as a bandwidth bro- ker, Band-X is introducing a new service that more closely mir- rors the functions of a true spot market, such as instant and anonymous exchange of money for network access. Band-X first introduced a preliminary version of its "Band-X Switched" service in the summer of 1998. When fully operational, a sale will proceed as follows: 1. Band-X establishes switches at prominent international switching hubs such as London’s Telehouse and New York’s 60 Hudson Street. Potential buyers and sellers connect to the Band-X switch. 2. A carrier wishing to sell wholesale minutes posts details on the destination and per minute rate at the Band-X web site. 3. An interested buyer pays Band-X a deposit equal to the value of minutes it wishes to buy. The buyer’s traffic is immediately routed onto the seller’s network. Unlike the standard Band-X service, both parties remain anonymous, even after the deal is concluded. Because capacity is sold on a first-come, first-served basis, buyers and sellers have an incentive to pre-establish network interconnection and payment accounts with Band-X. Figure 2, Arbinet and Company In the late nineteenth century, when substantial improvements in trans- portation had opened new avenues for supply but had not yet created an integrated national market, more than 1,200 commodity exchanges operated in the United States. The market for international wholesale telecom service has attracted its fair share of Internet-based market places, each with its own strategy for al~racting potential buyers and sellers. Cape Saffron specializes in Internet telephony; IP telephony guru Jeff Pulver plans to launch a Voice over IP minutes exchange [www.min-X.com); Amsterdam- based InterXion is the first band- width exchange on the European continent; RateXchange has estab- lished real-time switching in New York and Los Angeles. www.arbinet.com vwvw, capesaffron.com www.ratexchange.com Perhaps the most interesting exam- ple is the eponymous Arbinet. The company owes its moniker to the unique service it provides-- ARBltrage NETworks. Arbinet’s Global Clearing Network (GCN) is an automated, real-time switching service that allows carriers to draw on an international least cost routing database to find the cheapest way to send an international call abroad. Arbinet manufactures and sells its own switches to carder participants which communicate with the routing database. Carriers wishing to sell capacity enter into the www.interxion.com database when they wish to open their networks to the GCN, speci- fying route quality details and rates. Buyers then refile their traffic via the carrier participating in the Global Clearing Network, accessing sellers’ networks via the Arbinet switches on carder premises. In a sense, then, Arbinet offers the reverse of the "Band-X Switched" ser- vice: whereas Band-X encourages carriers to interconnect with cen- trally located switches, Arbinet sends the switches to the carders. © Teleeeography, Inc. 1998 © TeleGeography, Inc. 1998 TeleGeography 1999 Figure 3. The Band-X Index of U.K. International Wholesale Prices, 1997-1998 110 100 ....... U.K.-India 90 8O Composite Index U.K.-Netherlands Sep Oct Nov Dec Jan Feb Mar Apt May Jan Jul 1997 1998 Note: The BandoX U.K. index measures relative price movement for the top Wventy UoK. outbound routes, The composite ~ndex is an average of price changes to those twenty designations, with each route weighted according to its proportion of U.K. outgoing traffic as reported by TeleGeography. Source: Band-X (wv~v.band-x, com) ©TeleGeography, Inc. 1998 4. Band-X passes on payment to the cartier after charging a two pement fee to the buyer and a three percent fee to the seller 5. Using call quality monitoring software, Band-X periodi- cally tests the voice grade on the seller’s network to ensure that call quality meets the minimum standard established for the route. Back to Basics In their current form, Band-X and its peers greatly facilitate the exchange of network access among carders. Some observers have even speculated that these exchanges may soon resemble commodity markets, where one unit is fully interchangeable with any other, and anonymous buyers and sellers trade at a price determined by the forces of supply and demand. In addi- tion to serving as "spot" markets---on which goods are traded for immediate delivery--commodity exchanges often feature a secondary derivatives market where investors can purchase standardized contracts for delivery at a specified future date and price. A derivatives market for telecom services would per- mit carders to hedge their future revenues and costs. A review of the economic conditions underlying successful com- modity markets may tell us where bandwidth exchanges may be heading: ¯ Large number of buyers and sellers. A market in which only a handful of players operates does not require a cen- tralized exchange; instead, companies can rely on industry acquaintances to conclude deals. A growing pool of buyers and sellers encourages companies to seek out central mar- ket places for exchange. Also, a large number of market participants helps guarantee that prices are not arbitrarily established by a handful of powerful players, but rather are set by the forces of supply and demand. ¯ Transparent price information. Commodity markets fea- ture equal access to price information by buyers and sellers. This condition is important if market forces are to determine a single price for a good. ¯ Homogenous product. For a single price to emerge for a Oven commodity, all units of that commodity must also be interchangeable. To enforce homogeneity, most commodity exchanges feature pre-established, standard contracts that specify quality, quantity, and delivery terms. o Confidence in market mechanisms. Face-to-face negoti- ation helps buyers to determine that they will get their money’s worth. In the anonymity of a commodity market, however, buyers need to trust that they will get what they paid for before they engage in large-scale purchases. ¯ Volatile prices. Some price movement is necessary to spur interest in a spot market. Otherwise, buyers would simply purchase long-term contracts from suppliers. The likelihood of price fiuctuation, which Bives buyers and sellers an incentive to hedge against damaBing price swings, is especially important in creating a role for futures markets. TeleGeography 1999 © TeleGeography, Inc. ! 998 Figure 4. Why Settlement Policy Matters Wholesale Tariffs and Settlement Rates for One Minute U.S. International Calls to Sample Destinations $1.00 0 $O.75 o $0.50 o South Africa .-, o O o o o ,o- o -0 0 0 ," 0--o o ~~o o e° o $0.25 O O .-- Trendline $0.00 ...................................................... $0.00 $0.25 $0.50 $0.75 $1.00 $1.25 Settlement Rate (U.S.$/min) Note: Each circle represents the settlement rate and wholesale tsrifffor one call destination. For example, the U.S. settlement rate to South Africa is $0.40 and the wholesale rate is $0.5t. A statistical test between settlement and wholesale rates yields a correlation (r=) of 0.8205. That is, 82 percent of the variance among coun- tries" wholesale tariffs is explained by the variance in the settlement rate to those countries. Source: FCC, RateXchange (vwvw.ratexchange.com) © TeleGeography, Inc. 1998 How Bandwidth Exchanges Fit the Bill So how does the market for wholesale international telephone service stack up? Presently, wholesale telecom service trading is oriented around pdvate, negotiated contracts. Taken point by point, it is clear that bandwidth exchanges address a num- ber of the present system’s shortcomings: ¯ Large number of buyers and sellers. Telecom liberaliza- tion has created hundreds of licensed operators interested in buying and selling network access from other carriers. At least initially, however, these newcomers do not always have the means of surveying all sources of supply or of being noticed by potential customers. Deals with established players or time-consuming search and negotiation with newer participants are their choices. Helping to introduce new market players represents a chief function of bandwidth exchanges. The implementation of real-time switching and exchange will boost the volume of carder interaction even further. ¯ Transparent price information. Selling network access on a private, contract-by-contract basis offers suppliers scope to discriminate in favor of their langest buyers. In contrast, bandwidth exchanges increase transparency by publishing all available bids and offers for a given international route. ¯ Homogenous product. When trading raw bandwidth, product homogeneity is rarely a concern. One E-1 circuit between Telehouse (London) and 60 Hudson (New York) is pretty much like any other E-1 connection between the two global switching centers. With minutes trading, however, ~uaranteeing call quality is a real problem. Such factors as method of switching (circuit versus packet), number of net- works traversed, and degree of digital compression can affect the quality of a call. Fortunately, bandwidth exchanges can enforce product homogeneity by using call testing software. Bandwidth exchanges will use the soft- ware to ensure that traded minutes measure up to a pre- established benchmark quality rating for calls. ¯ Confidence in market mechanisms. Although many new sellers of wholesale minutes exist, the reliability and quality of their products is often unknown. Face-to-face negotiation thus remains an important component of minutes deals. For a reseller concerned about the call quality its customers will experience, the choice of suppliers is self-limited to carders it trusts. Trusted suppliers can command a premium. © TeleGeography, Inc. 1998 TeleGeography 1999 A reliable quality testing system by the bandwidth exchanges would go a long way toward encouraging buyers to make purchases from anonymous suppliers. Overcoming the trust barrier would enhance the efficiency of telecom ser- vice trading by expanding the pool of market participants with whom a telco can reliably trade and by eliminating time-consuming negotiation in favor of instant exchange. ¯ Volatile prices. When combined with pdce transparency, a shift from negotiated contracts to instant and anonymous exchange could increase price volatility by speeding market reaction to changes in the underlying ~undamentals of sup- ply (e.g., a new undersea cable comes on-line) and demand (e.g., a government permits new resellers to enter its inter- national market). Remaining Problems If the creation of an active forward market for minutes and bandwidth is the ultimate goal for bandwidth exchanges, two significant obstacles remain. ¯ Fragmented global market. Commodity markets in which worldwide prices move together attract greater liquidity than exchanges with a narrower geographic scope. Coffee represents one commodity whose market has a global reach. A bumper coffee crop in Brazil, for example, affects Indonesian coffee growers by lowering the pdce of coffee traded in New York. This price link works because Brazilian and Indonesian coffee beans are close substitutes. Unfortunately, calls to different international destinations make poor substitutes. A U.K.-India and U.K.-Netherlands call minute are hardly interchangeable--and the recent price trends for calls on these international routes show it (see Figure 3.) By arbitraging price differences among different routes, aggressive least cost routing would go some way toward tying international wholesale call prices together (see "The Logic Behind Refile" on page 63). But this link would likely remain only tenuous, largely because national government policies still exert a considerable influence over termination rates and competition within country barriers. Although real factors of supply, such as increased international capacity, do influence prices, wholesale tariffs mostly have fallen to destinations where the national government has enforced a transition to lower interconnection rates and competition. (For example, see Figure 4 on the correlation between set- tlement rates and wholesale prices.) Until telecom deregu- lation is more widespread and mature, prices for different routes will continue to move out of sync with each other, and no single, global market for international network access will develop. Figure 5. Wave of the Future? Prices for February 1999 Pork Belly Futures, May-September 1998 Jul-Oz-g8 Aug-03-98 ra.~o-ng-98 Selo-25-98 Open 4b 90~ High 49,500 Net Change -0.s,~5 Volume 1306 Low 46.100 Last 48,900s Prev Volume 1232 Delay 10rain Note: This char~ shows day-to-day variance ~n the price for a pork belly futures contract traded on the Chicago Mercantile Exchange. The smooth line from August through September shows a 50 day moving average. The traded commodity is not actually pork belhes, but rather standard contracts to obtain a set quan- tity (40,000 pounds) of pork bellies at a future date (February). Source: QuoteWatch (vwwv.quotewatch.com) TeleGeo~raphy 1999 © TeleGeography, Inc. 1998 ¯ Insufficient price volatility. Even if wholesale prices for most international routes began to move in tandem, pre- dictability of price movements could dampen interest in a futures market. In most commodity markets, pdces move up and down. On many routes in the wholesale minutes market, however, prices have moved only downward. Theoretically, a buyer of telecom minutes might hedge against the volatility of the degree of price decrease. Without the possibility that prices might actually rise, how- ever, a buyer often lacks the psychological imperative to rely on a derivatives market to cover his or her future position. Prospects By introducing the growing number of buyers and sellers to each other, offering improved price transparency, providing cen- tral points for instantaneous traffic switching, and enforcing measurable standards for call quality, bandwidth exchanges promise to increase significantly the efficiency of the interna~ tional telecom services market. Commodity markets, they are not. As long as the international market for telephone service remains fragmented by national boundaries, a central bench- mark that tracks prices on multiple routes wilt not emerge. Of course, an exchange need not offer worldwide scope to become an active commodity market. Many mercantile exchanges are regional. In the medium term, a market may converge around traffic to and from countries with advanced competition regimes, where further government deregulation has less scope to alter costs radically and where prices already do appear to be bottoming out. Alternatively, carriers might soon offer futures contracts for delivery on a limited, route-by-route basis. Enron Corp., the Texas-based conglomerate with experience in wholesale energy trading, announced in September 1998 that it was holding pre- liminary talks with Band~X regarding standard futures contracts for telecom services. (For more on the parallels between elec- tricity and telecom service trading, see "What is Sparking New Markets for Power?." on pages 77 to 79.) The lack of liquidity attracted by a unified global benchmark price for international wholesale minutes and bandwidth, how- ever, will hamper the development of an active derivatives mar- ket. Until liquidity grows and prices bottom out, investors wish- ing to take positions in a fully developed commodity exchange will have to turn to more traditional markets. Pork bellies may be the best bet. © TeleGeography, Inc. 1998 TeleGeography 1999 What is Sparking New Markets by Donald Weightman for Poweff When Indiana utilities turned off power for large industrial cus- tomers in the midwestern U.S. last June 25, some of the blame went to familiar causes: surging demand from a heat wave, out- ages at key generating plants, a transmission line brought down by lightning. But other parts of the story are new--a turbulent wholesale power market where a supply squeeze and panicked traders sent prices from $30 to more than $7500 per megawatt hour The June price shock in the U.S. market is evidence of some-- but not all-of the dramatic changes in once pervasively-regu- lated wholesale power exchanges. Besides the spot markets, there are also new paper markets in electricity futures and options. Each of these markets--spot and paper--will be sur- veyed in turn, with an eye toward identifying the structural problems and prospects that might face a market in bandwidth. Some critical facts about the electric utility industry have not changed. Electricity, unlike wheat, oil, and silver, cannot be stored, and the transmission grid continues to have the economies of scale and other characteristics associated with the "natural monopoly" rationale for regulation. But decades of rising power prices--in the face of declining fuel costs--have brought on a customer-driven push toward competition. What’s New At Wholesale There are new players. Aside from sales to retail utilities with- out generation of their own, wholesale exchanges were largely the province of vertically inte~ated utilities. Regulatory policy favoring fuel diversity brought in new sellers of generation after the energy crisis of the 1970s. Under new U.S. laws, since 1992, almost anyone can trade kilowatts at wholesale, and as of 1997 some 100 power marketers--a few dozen utility affili- ates, the rest independent--were active traders. Where once there were regulated rates based on costs (with seasonal or emergency exchanges between utilities at nominal markups), now prices are set by bargains struck over the tele- phone, based on reports from market news wires, or reported futures trades from the New York Mercantile Exchange (NYMEX). Operating needs--emergencies, spikes in demand, or the chance to share savings by dispatching efficient units~clrove deals under the old regime. But trading was also constrained Figure 1. U.S. Wholesale Electricity Trades 1000 9OO 800 700 600 50O 4O0 300 200 100 0 Spot Market Growth, 1995-1997 1995 1998 1997 100 Futures Market Growth, 1996-1998 80 Palo Verde 40 Border Market ~; 20 lO 1996 1997 1998 Source: Edison Electric Institute, Power Marketers Yearbook--1997 Source: NewYork Mercantile Exchange (NYMEX) Note: Spot market totals include all U.S, wholesale electricity trades. Futures for only four regional markets (including the California-0regon Border and Palo Verde exchanges) are traded on NYMEX. TeleGeography 1999 © TeleGeography, Inc. 1998 Figure 2. Power Markets Overview Method of Price Establishment Basis of Prices Participants Reliability of Supply Purpose of Transaction Contractual Terms Traditional Wholesale Spot Futures Exchanges Markets Markets Regulation Negotiation Auction Costs Supply & Demand Trader Expectations Utilities Registered Marketers Open Standards for Reserves Set by Contract Physical Delivery Offered through Exchange Operating Problems & Operations or Speculation Hedging or Speculation Shared Efficiencies Complex Complex Simple Tailored Standardized Standardized Negotiated Intermediaries Involved Posted on Exchange Source: Donald Weightman by scarce transmission and by the fact that utilities were not eager to open their power lines and switchyards to competitors. The 1992 statute, aiming to spark competition, opened up the transmission grid. Although there are still emergency and sim- ilar operational exchanges, there is a rapidly growing wholesale market for resale. Dependable service had been the touchstone of utility practice: reserve supplies were required by reliability councils, and paid for by ratepayers. As "excess inventory," such reserves are for- eign to competitive markets. "Firm" power assured for deliv- ery, however, commands a hefty price premium, especially when demand surges. As the June 1998 outage suggests, the risk of non-delivery is now allocated by contract, often by liq- uidated damages clauses where the defaulting seller must pay the cost of replacement. Even under regulation, there were spot markets growing up around the delivery points where utilities exchanged seasonal or emergency reserves. The cdtical next step was to open these markets to new competitors with access to the grid. Surging Spot Markets Traders work over the telephone. The difference after the 1992 statute is in how transmission is arranged if deals are struck. Federal regulations now require utilities to post data on avail- able transmission capacity at ]nternet web sites. Under the Open Access Same Time Information System (OASIS), being developed in phases, traders needing transmission for power may use passwords to log in and query utility data bases for next day capacity. The user may then reserve capacity by flit- ing out a data template, much as one would now make a plane reservation elsewhere on the Net. In another phase under dis- cussion, the OASIS system would increase posted information to include purehaser and price data. Because the utilities owning transmission for a regional market also do business there, competitors claim that data from the templates is being used to spot and poach customers. These critics also call for more dgorous structural separation between power transmission and power generation, even to the point of mandating divestiture of the transmission gdds to independent operators. Market Interest in Futures? The growth in spot markets and speculative power trading has created interest in laying off the risk of price swings. Utilities recovering costs under traditional rate regulation had little con- cern for hedging against volatile prices. But high volume buy- ers and sellers with bottom line exposure to large scale price movements are potential customers for forward and derivative contracts to hedge their risks. Electric power thus meets some of the classic conditions for derivative markets in other com- modities: large numbers of price-taking buyers and sellers and price volatility. Indeed, some observers reckon that wholesale power price swings will become more frequent as natural gas-- itself a commodity with volatile prices--becomes the fuel of choice for new electric generation. To exploit these possibilities, NYMEX in March 1996 intro- duced trading in futures contracts, using prices at transfer points for two Western U.S. spot markets (each with a history © TeleGeography, Inc. 1998 TeleGeography 1999 of sharp swings in prices) for reference. NYMEX has since added contracts based on prices in two Eastern U.S. regional markets, and futures and options trading for other spot mar- kets is in the works at NYMEX and other exchanges. But the paper markets, like their physical counterparts, remain local- ized to regional rather than national prices, and trading has not grown at the rate some had projected. Points of Resistance Some of the obstacles facing spot and futures markets seem susceptible to relatively straightforward resolution. Technical problems facing the OASIS system, for example, look to be worked out with design changes and regulatory pressure to support trader confidence. Redesign of futures contracts from 30 day blocs of peak capacity to different terms may offer hedges better suited to short-term price swings. On the other hand, even more powerful computers may not be enough to map the power grid into real time information sys- tems, and so spot markets could always face some degree of lag. And faster computing and better contractual design may not offset the central facts of market life. Storage is not prac- tical for power; for both technological and competitive reasons, transmission will be a scame resource for some time to come. In contrast to telecom markets, here the enduring barriers appear to be technical rather than legal. Transmission bottle- necks mean that both spot and futures markets may remain regional rather than national. Traders will aim to arbitrage the price differences between markets, but, as Morgan Stanley’s John Woodley notes, "transportation is inherently an issue in any forward market." Without centralized price discovery, futures markets in electrons may remain too thin to manage risk and supply liquidity for the underlying regional spot markets. The economic and political factors which opened the U.S. grid to market forces have not, however, gone away. Notwithstanding the June 1998 scare, U.S. wholesale spot markets in electricity seem likely to continue growing. Yet to come are ways for participants to manage the uncertainties now turning up here as in other commodities markets. Donald Weightman is a Washington, D. C. attorney. His practice concentrates on competition issues for the electric utilities and telecommunications industries. He can be reached by email at dweightman@radix.net. Te~eGeography 1999 © TeleGeograph~l, Inc. !998 Retail Prices for a Three Minute Call From/To Australia Belgium Australia (Telstra) peak n.a. 3.35 Australia off peak n.a. 2.44 Belgium peak 3.25 n.a. Belgium off peak 3.25 n.a. Czech Rep, peak 4.39 2.16 Czech Rep. off peak 3.49 1.71 France peak 2.34 0,88 France off peak 1.90 0.71 Germany (DT) peak 3.17 1.23 Germany off peak 3.17 1.11 Greece peak Greece off peak 1.37 Italy peak 3.70 Italy off peak 3.34 Japan (KDD) peak 4.78 Japan off peak 3.74 Switzerland peak 3.02 Switzerland off peak 2.33 Canada Denmark 2.62 2.99 1.75 2.26 1.62 1.22 1.62 0.88 3.64 2.16 2.90 1.71 0.94 0.94 0.77 0.77 1.23 1.23 1.11 1.11 1.71 1.52 1.71 1.52 1.17 1.37 1.17 1.60 1.60 1.60 1.31 1.45 1.31 5.40 3.12 5.40 4.36 2.56 4.36 1.70 1.45 1.70 1.39 1.13 1.39 Finland France Germany Greece Ireland Italy 2.99 2.82 2.82 3.72 2.55 2.73 2.26 2.26 2.08 2.55 1.73 1.73 1.62 1.22 1.22 1.62 1.62 1.62 1.42 0.88 0.88 1.42 1.42 1.42 2.82 2.16 1.49 2.82 2.90 2.16 2.82 1.71 1.19 2.82 2.30 1.71 1.04 n.a. 0.88 1.04 1.04 0.88 0.85 n.a. 0.71 0.85 0.85 0.71 1.41 1.23 n.a. 1.23 1.23 1.23 1.11 1.11 n.a. 1.11 1.11 1.11 1.52 1.52 1.52 n.a. 1.52 1.52 1.17 1.17 1.17 n.a. 1.17 1.17 1.60 1.60 1.60 1.60 1.60 n.a. 1.31 1.31 1.31 1.31 1.31 n.a. 5.40 5.33 5.33 5.40 5.40 5.40 4.36 4.29 4.29 4.36 4.36 4.36 1.70 1.45 1.45 2.02 2.02 1.45 1.39 1.13 1.13 1.51 1.51 1.13 Brazil peak 5.44 4.39 4.16 Brazil off peak 4.35 3.49 3.32 India peak 4.41 4.41 5.29 India off peak 3.94 3.94 4.41 Malaysia peak ~.68 3.86 2.19 Malaysia off peak 1.60 3.06 1.75 S0~th Afnc~ p-~ak 1.77 4.~6 1.77 South Africa off peak 1.55 3.86 t.55 UA~ pe~ak ..... 3~~ ~.8~ 3.77 UAE off peak 2.07 4.60 2.07 U.K. (BT) peak 2.05 1.19 0.99 U.K. (BT) off peak 1.75 1.11 0.94 4.38 4.38 4.38 4.38 5.06 4.38 4.38 3.49 3.49 3.49 3.49 4.05 3.49 3.49 4.41 4.41 4.41 4.41 4.41 4.41 4.41 3.94 3.94 3.94 3.94 3.94 3.94 3.94 3.28 3.93 3.28 4.37 5.46 3.93 3.86 2.62 2.62 2.62 2.91 3.64 3.06 2.62 z~.-3t~ 3.86 3.1~- 3.15 4.36 ~.~- 4.36 3.86 3.26 2.43 2.43 3.86 2.43 3.86 5.07 -4~08 3.77 4.45 4.45 3.77 3.77 4.20 3.68 2.62 3.68 3.68 2.62 2.62 1.19 -i.Gi --~i~9~ 1.19 1.48 0.96 1.48 1.11 1.45 1.11 1.11 1.24 0.82 1.24 U.K. (ACC) peak 1.25 0.72 0.59 0.72 0.98 0.72 0.72 0.90 0.63 0.90 U.K. (ACC) off peak 1.06 0.68 0.57 0.68 0.88 0.68 0.68 0.75 0.54 0.75 U.S. (MCI) basic flat 8.25 6.24 1.59 5.94 6.84 5.85 5.94 6.66 4.11 4.56 U.S. (MCI One) 1.14 0.87 0.36 0.87 0.87 0.87 0.87 0.87 0.87 0.87 U.S. (AT&T) basic flat 4.53 4.65 1.47 4.38 4.44 3.99 3.75 5.85 4.05 4.50 U.S. (AT&T One Rate) 1.44 0.87 0.36 0.87 0.87 0.87 1.17 0.87 0.87 0.87 U.S. (USA Global Dnk) 0.89 1.13 0.74 0.93 1.19 0.96 0.89 1.47 1.42 1.35 U.S. (Excel WorldRate One) 0,87 0.78 0.33 0.87 0.87 0.78 0.75 1.02 0.84 0.84 U.S. (Delta Three IP Telephony) 0.57 0.81 0.38 0.69 0.90 0.69 0.69 1.26 0.96 0.96 Notes: 1. Al! rates are =n US $ and exclus=ve of taxes 2 Rates were current on July 31, 1998 3. Rates have been calculated m reel t=me using meter step (rounded up to next meter step for a 3 m=nute call} 4 Fees are $3 per month for AT&T One Rate, MCI One, and Excel WorldRate One plans Excel Prime Bus~ness has a m~mmum monthly bilhng of $100 5 Rates for calls from the U.S to Canada and Mexico are fromWashmgton, D.C to Toronto and Mexico C=ty. © TeleGeographyo Inc. 1998 TeleGeography 1999 Japan Korea Mexico Neth’lands Portugal Spain Sweden Turkey U.K. U.S. To/From 3.01 3.37 3.72 2.99 3.72 3.72 2.99 3.72 2.28 2.28 Australia (Telstra) peak 2.46 2,57 3.53 2.26 3.19 3.19 2.61 3.01 1.75 1.75 Australia off peak 3.25 4.46 5.68 1.22 1.62 1.62 1.62 2.43 1.22 1.62 Belgium peak 3.25 4.46 5.68 0.88 1.42 1.42 1.42 2.03 0.88 1.62 Belgium off peak 4.39 5.57 8.18 2.16 3.87 2.82 2.82 2.82 2.01 3.27 Czech Rep. peak 3.49 5.57 8.18 1.71 3.87 2.82 2.82 2.82 1.64 2.68 Czech Rep. off peak 2.46 2.34 2.89 0.94 1.04 0.88 0.94 1.25 0.88 0.94 France peak 1.98 1.90 2.32 0.77 0.85 0.71 0.77 1.02 0.71 0.77 France off peak 3.17 3.17 4.46 1.23 1.41 1.23 1.41 1.64 1.23 1.23 Germany (DT) peak 3.17 3.17 4,46 1.11 1.11 1.11 1.11 1.41 1.11 1.11 Germany off peak 2.74 2.74 2.74 1.52 1.52 1.52 1.52 1.32 1.52 1.71 Greece peak 2.33 2.33 2.33 1.17 1.17 1.17 1.17 0.99 1.17 1.37 Greece off peak 3.70 4.86 4.86 1.60 1.80 1.60 1.60 2.39 1.31 1.60 Italy peak 3.34 4.06 4.06 1.31 1.31 1.31 1.31 1.89 1,23 1.45 Italy off peak n.a. 3.88 5.13 5.40 5.40 5.40 5.40 6.23 5.33 3.12 Japan (KDD) peak n.a. 3.05 4.16 4.38 4.36 4.38 4.36 5.06 4.29 2.56 Japan off peak 3.02 4.03 4.03 1.45 2.02 2.02 1.70 2.46 1.45 1.45 Switzerland peak 2.33 3.02 3.02 1.13 1.51 1.51 1.39 1.95 1.13 1.13 Switzerland off peak 5.44 8.11 4.16 4.38 4.33 4.38 4.38 5.08 4.38 2.75 Brazil peak 4.35 6.49 3.32 3.49 3.46 3.49 3.49 4.05 3.49 2.20 Brazil off peak 4.41 4.41 5.29 4.41 4.41 4.41 4,41 4.41 4.41 5.29 India peak 3.94 3.94 4.41 3.31 3.31 3.31 3.31 3.31 3.31 4.41 India off peak 2.40 2.84 5.46 3.28 5.46 3.86 3.28 3.93 2.33 2.19 Malaysia peak 2.11 2.26 3.64 2.62 3.64 2.62 2.82 2.91 1.82 1.68 Malaysia off peak 4,36 3.86 4.80 3.15 3.70 4.36 4.36 3.86 2.43 1.99 South Africa peak 3.86 3.26 4.31 2.43 3.04 3.86 3.88 3.26 1.71 1.55 South Africa off peak 5.07 4.45 9.19 4.45 5.88 4.45 5.07 5.88 3.77 3.77 UAE peak 4.45 3.88 5.88 3.68 4.80 3.68 4.20 4.60 2.62 2.45 UAE off peak 3.22 4.51 4.51 1.19 1.48 1.48 1.19 2.82 n.a. 0.99 U.K. (BT) peak 3.05 4.29 4.29 1.11 1.24 1.24 1.11 2.39 n.a. 0.94 U.K. (BT) off peak 1.64 2.70 2.70 0.72 0.90 1.00 0.72 1,85 n,a. 0.59 U.K. (ACC) peak 1.43 2.58 2.58 0.88 0.75 0.75 0.68 1.60 n.a. 0.57 U.K. (ACC) off peak 4.41 5.43 3,51 3.93 4.89 4.86 3.90 5.40 3.36 n.a. U.S. (MCl) basic flat 1.05 1.95 1.32 0.87 0.87 0.87 0.87 1.86 0.36 n.a. U.S. (MCI One) 4.35 5.46 3,48 3,93 4,95 4.86 3.93 5.52 3.27 n.a. U.S. (AT&T) basic flat 1.53 1.95 2.88 0.87 0.87 0.87 0.87 1.88 0.36 n.a. U.S. (AT&T One Rate) 1.16 2.14 2.20 1.08 1.64 1.51 0.81 2.19 0.80 n.a. U.S. (USA Global Link) 1.14 1.62 1.14 0.78 1.02 0.87 0.78 1.77 0.33 n.a. U.S. (Excel WorldRate One) 0.84 1.77 1.20 0.69 1.24 1.14 0.45 1.71 0.48 n.a. U.S. (Delta Three IPTelephony) Source Phillips Tarifica Ltd, 40 Furmval St, London EC4A 1JQ, U.K Tel +44 171 4406500 Fax +44 171 8318552 Emad: consult@tar=fica.com ¯ http://wvwv.tar=fica.com Source for U.S rates: TeieGeography, inc. © Ph=lhps Tar=fica Ltd and "[eleGeography Inc., 1998 TeleGeography 1999 © TeleGeography, Inc. 1998 U.S. and U.K. International Tariff Maps United States: 1998 $1 50 $~.25 St,O0 $0 35 $0.50 $0,50 $0.75 $1.0D $1 25 $~- 50" $1 75J ’, ’ ’Hang Kong~ k SydneyK. /tvll~l~?..~.. Bahram~i ~unaKa, GOdT’,l : United States: 1994 $1 Note: These four maps compare the price per minute of making an international telephone call from the U.S. and the U.K. in 1994 and 1998. Countnes are arranged according to the US$ price per minute of calls made from the center country, Call charges are based on June 1994 and July 1998 published peak rates from MCI and BT. © TeleGeography, Inc. 1998 TeleGeography 1999 United Kingdom 1998 ! ,,’ ;, ,,, Seoul ~ashm~o~ Beijin~ Bucharest Ankara’ SOU~.h AS A Kabul Karoc~i United Kingdom: 1994 $1,50 Beijing TeleGeography 1999 © TeleGeographg, Inc. 1998 Origin of U.S. Carrier Revenues Figure 1. U.S. Carrier Revenues for International Voice Service, 1997 Total Receipts (US$ millions) Average Revenue per Minute (US$/minute) AT&T MCI Sprint WorldCom Pac. Gateway Exch. Total Billed Settlement Retained Settlement Net Billed Settlement Retained Settlement Revenue Outpayment Revenue Inpayment Revenue Revenue Outpayment Revenue Inpayment 8,077.0 3,754.5 4,322.6 1,305.4 5,628.0 0.78 0.36 0.42 0.30 4,234.4 2,296.1 1,938.2 681.9 2,620.2 0.72 0.39 0.33 0.30 1,455.8 992.3 463.5 341.6 805.1 0.53 0.36 0.17 0.20 500.0 521.8 -21.8 135.8 114.0 0.36 0.37 -0.02 0.31 174.2 183.2 -9.0 23.3 14.3 0.23 0.25 -0.01 0.20 14,441.5 7,747.9 6,693.5 2,488.0 9,181.5 0.68 0.37 0.32 0.28 Note: This table breaks down international voice service revenue for the five largest U.S. international carriers in t997. For example, Wor|dCom collected $500 million from customers for U.S. international outgoing calls, and paid foreign c arriers $521.8 million to terminate those calls. Thus, the company lost $21.8 million by carrying U.S. outgoing calls. Because FCC regulations entitle each U.S. carrier to terminate incoming calls based on the percentage of U.S. outgoing traffic it originates, WorldCom collected a significant sum ($135,8 million) on foreign settlement inpayments, netting $t14.0 mill=on on international voice service. Source: FCC carderfilings. ©TeleGeography, lnc. 1998 Figure 2. U.S. Carrier International Call Revenue by Destination, 1997 Average U.S. Outgoing Call Revenue bV Region $1.50 $1.25 $1.00 $0.75 $0.5O ¯ Settiemen~ Payments Retained Revenue $0.25 $0.00 Average U.S. Outgoing Call Revenue by Country $1.50 Settlement Payment~ Retained Revenue $1.25 $0.75 $0.50 $0.25 I Note. Charts show average revenue on U.S. international outgoing calls for the four largest U,S. international carriers. Total column height shows the average price for calls on a given route. Retained revenue equals average price minus settlement payment, and includes such components as access fees for origination, network casts, and profit. Source: FCC carrier filings. © TeleGeography, Inc. 1998 Facilities TeleGeography 1999 © TeleGeography, Inc. 1998 International Submarine Cables Note: TAT 14 N0rth/South capacity of~640 Gbps is not clraw~l to s~ale. © TeleGeography, Inc. 1998 © TeleGeography, Inc. 1998 TeleGeog~aphy 1999 Key Common Private Major International Submarine Cables Carrier In-Service 80 40 20 10 5 1 Planned Capacity in gigabits per second (Gbps). One Gbps is equivalent to approximately 70,000 telephone calls. "Planned" cable systems are scheduled to begin operations on or before January 1, 2001. © TeleGeography, Inc, 1998 TeleGeography 1999 page © TeleGeographg, Inc. 1998 Figure 1. Map of Major Submarine Cables in Asia/Pacific pC1 TPC 5 :hina US Note: The color scheme used in these regional focus maps is intended only to help distinguish cable systems. The maps depict major international submarine cables in service or scheduled to begin service before January 2002. Domestic cable systems are omitted with the exception of the Japan Information Highway (JIH), which is under construction with service scheduled to commence March 1999. JIH wes included because of its large cepacity (100 Gbps). © TeleGeography, Inc. 1998 © TeleGeography, Inc. 1998 TeleGeog~aphy 1999 Figure 2: Map of Major European ~bles © TeleGeography, Inc, 1998 Figure 3: Map of Major Caribbean Cables Note: Map depicts major internationa|eubmarine eables:in service or ec~dUled:-to begin serviesbefOre~U~2002~ Regional Cabi~sarsomitted~ ©TeleGeography, Inc. 1998 TeleGeography 1999 © TeleGeography, Inc. 1998 Ready for System 64 Kbps Cost Cable Service Capacity (Gbps) Circuits (US$ millions) Asia Pacific APC 1993 1.68 22,680 332 APCN 1996 10.0 120,960 550 China-U.S. 1999 80.0 967,680 1,400 G-P Cable 1999 5.0 60,480 n.a. GPT 1989 0.56 7,560 178 HAW 4 1989 0.56 7,560 n.a. HAW 5 1993 1.68 22,680 157 Japan-U.S. 2000 80.0 967,600 n.a. NPC 1991 1.68 22,680 425 PacRim East 1993 1.12 15,120 233 PacRim West 1994 1.12 15,120 n.a. PC I 2000 80.0 967,680 1,240 Southern Cross* 2000 40.0 483,840 940 Tasman 2 1991 1.12 15,120 110 TPC 3 1988 0.56 7,560 718 TPC 4 1992 1.12 15,120 373 TPC 5 1997 20.0 241,920 1,240 TVH 1995 1.12 15,120 128 Europe-Mediterranean Eurafrica 1992 2.5 30,240 n.a. Lev 1998 20.0 241,920 66 Indian Ocean FLAB 1997 10.0 120,960 1,500 JASURAUS 1997 5.0 60,480 100 SAFE/SAT 3 2000 20.0 241,920 550 SEA-ME-WE 2 1994 1.12 15,120 n.a. S EA-M E-WE 3 1998 20.0 241,920 1,300 North Atlantic AC 1 1998 40.0 483,840 750 CANTAT 3 1994 5.0 60,480 302 CANUS-1 1995 5.0 60,480 n.a. Gemini 1997 30.0 362,880 500 PTAT 1 1989 1.68 22,680 510 TAT 8 1988 0.56 7,560 360 TAT 9 1991 1.68 22,680 406 TAT 10 1992 1.68 22,680 300 TAT 11 1993 1.68 22,680 280 TAT 12/13"* 1995 10.0 120,960 750 TAT 14 2000 640.0 7,741,440 n.a. South Atlantic Americas 1 North 1994 5.0 60,480 n.a. Americas 1 South 1994 1.68 22,680 n.a. Americas 2 1999 40.0 483,840 375 Atlantis 2 1999 5.0 60,480 231 BUS 1 1997 2.5 30,240 n.a. CARAC 1990 0.48 6,480 n.a. Col um bus 2 1994 1.68 22,680 337 Columbus 3 1999 10.0 120,960 236 MAC 1999 20.0 241,920 415 MAYA 1999 7.5 90,720 PA C 2000 20. 0 241,920 280 PanAm 1998 I0.0 120,960 311 SAT-2 1993 0.56 7,560 n.a. SAFE/SAT 3 2000 20.0 241,920 550 UNISUR 1994 1.12 15,120 n.a. Note: italics indicate systems that are planned or under construction. A 64 Kbps voice circuit figure assumes capacity of 2.5 Gbps or greater based on Synchronous Transpor~ Module (STM) hierarchy; i.e., 16 STM-t (30,240 64 Kbps voice channels) per 2.5 Gbps plus approximately 23% signaling overhead. A 14% signalling overhead is assumed on cables of lesser capamty. The Hawaii-U.S. mainland leg of this cable will have a capamty of 80 Gbps. ** The TAT 12]13 ring was upgraded to 10 Gbps in 1998 and wilt be upgraded to 15 Gbps in 1998. © TeleGeography, Inc. 1998 © TeleGeography, Inc. 1998 TeleGeography 1999 Figure 5, Out of Service Trans-Oceanic Cable Systems System Capacity Region Lifetime System (Voice Paths) Landing Points Trans-Atlantic Trans-Pacific 1956-1978 TAT 1 89 U.K.-Canada 1959-1982 TAT 2 89 France-Canada 1961-1985 CANTAT 1 160 Canada-U.K. 1963-1986 TAT 3 276 U.S.-U.K. 1965-1987 TAT 4 138 U.S.-France 1970-1993 TAT 5 1,440 U.S.-Spain 1976-1994 TAT 6 8,400 U.S.-France 1976-1995 CANTAT 2 3,800 Canada-U.K. 1983-1994 TAT 7 8,400 U.S.- U.K. 1957-1989 HAW 1 89 California-Hawaii 1963-1984 COM PAC 160 Ca nada-U.S.-Fiji-N.Z.-Australia 1964-1994 HAW 2 1,690 California-Hawaii 1964-1990 TPC 1 167 6uam-Philippines-U.S.-Japan 1974-1993 HAW 3 1,440 California-Hawa ii 1975-1994 TPC 2 1,690 Guam-Japan-U.S. 1983-1997 ANZCAN "D" 6,900 Australia-N.Z.-U.S.-Canada ©TeleGeography, Inc. 1998 TeleGeography 1999 © TeleGeography, Inc. 1998 The Rise of Private Cable Systems The Club Until the late 1980s, all commemial submarine cables for telecommunications were built, used, and paid for by la~e groups of incumbent (usually monopoly) carriers. The systems were designed to work within the half-circuit correspondent relationship common to international telephony through most of the 20th century (for more, see "International Carrier Evolution" on page 26). The "Club" members that built these systems shared costs and worked together without fear of competition from their part- hers. Members would forecast their expected capacity needs on international mutes based on fairly predictable growth rates for telephone services and plan cables accordingly. Capacity was allocated and payment was made before or during con- struction of the system. U.S. and other regulators typically treated Club members as "common carriers;" that is, members were required to sell capacity to non-members on a non-dis- criminatory basis and close to cost. Evidence of Change When international service competition began, the Club system began to break down. Because Club members controlled both supply and price on the cables, capacity for non-members was limited and expensive. In addition, Club systems only provid- ed connectivity from shore to shore, forcing new carriers to lease "backhaul" capacity from the former monopoly in order to connect the cable to their network, usually many miles inland. New international carriers needed to find a more cost- effective way of carrying their trans-oceanic traffic. Figure 1. The History of Submarine Cable Capacity Trans-Oceanic Capacity, 1990-2000 8OO Atlantic Pacific 1990 1992 1994 1996 1998 2000 700 IO0 Source: TeleGeography, Inc. Three years ago, approximately 32 C-igabits per second (Gbps) of trans-oceanic submarine cable capacity landed in the U.S. As of this writing, however, there is over 100 Gbps of additional capacity. By 2001, the total will exceed one Terabit per second (1,000 Gbps)-- enough capacity to carry more than 70 million simultaneous tele- phone calls. To put these numbers in context, a rough rule of thumb is that one Gbps of capacity can provide approximately 70,000 tele- phone channels (this assumes five voice paths can be derived from a Trans-Oceanic Cable Voice Paths, 1988-2001 Trans-Atlantic Trans-Pacific 1988 43,750 1,857 1989 43,750 43,750 1990 175,000 43,750 1991 175,000 135,000 1992 437,500 222,500 1993 568,750 222,500 1994 1,090,625 310,000 1995 1,481,250 310,000 1996 1,481,250 310,000 1997 3,825,000 1,872,500 1998 7,340,625 1,872,500 1999 8,903,125 8,122,500 2000 58,903,125 23,747,500 Note Voice paths representthe number of simultaneous conversa- tions that can be carried on a particular hnk Various encoding and compression techniques are used to carry more than one conversation on each 64 Kbps circuit. Figures in this table assume a 5:1 signal com- pression ratio. Data represents year end totals. Source: TeleGeography, Inc., Euroconsult standard 64 Kbps circuit). The pdmary reason for the growth in bandwidth is technological; the capacity of a pair of fiber optic strands have increased from five to 160 Gbps over the last five years of this decade (see Figure 2). The timing of the bandwidth explosion is cdtical for two reasons. First, Internet growth is generating a very large new demand for trans-oceanic bandwidth. And second, new international carriers need bandwidth to build their own networks. ©TeleGeography, Inc. 1998 © TeleGeographyo Inc. 1998 TeleGeography 1999 Figure 2. International Submarine Cable Capacity and Cost 100,000 10,000 1,000 100 Capacity per Fiber, 1985-2000 DWDM p 160 Gbps (1,935,360 circuits) WOM/20 Gbps (241,920 circuits) _.~ 2 5 Gbps (30,240 c=rcuits) 1990 1995 2000 Submarine Cable Cost per Voice Channel per Year, 1970-2001 $100,000 $10 $1 197o 1975 1980 1985 1990 1995 2000 Note: Circuitfiguresindicatethe number of64Kbps circu}tspius signaling overhead. Source: Atcatel Atsthom (www.atcatel.co,fr) Note: System cost is derived from total construction costs divided bythe sys- tem’s lifetime. Rgures assume compression ratio of 5:t on post-t988 systems, Source: TeleGeography, Inc. and Euroconsult © TeleGeography, Inc. 1998 The first visible sign that the Club system was in decline came when new carriers began building their own cables and landing facilities. PTAT, a 1.68 Gbps trans-Atlantic cable built in 1989 (and now owned by Sprint and Cable & Wireless), was the first "private" (i.e., non-Club) cable system since the entrepreneur- ial telegraph cables of the 19th century. Like Club cables, pri- vate cables sold capacity to other parties, but unlike Club sys- tems, they could choose to whom they sold (i.e., they could dis- criminate). Later came FLAG, a system financed by telecom concerns as well as private investors, which was designed from the start to sell capacity rather than to reserve it for its builders. Today the biggest private cable player is Global Crossing, a publicly trad- ed company with the sole purpose of building a global subma- rine cable network, including backhaul facilities to the 50 biggest cities in the world. Although the first leg of the system, Atlantic Crossing-I, was only partially operational as of this writing, it has already matched bandwidth with the largest (and sold-out) Club system on the Atlantic, TAT 12/13. The Club rules for building and managing submarine cables will never be the same again. Here’s why: Proliferation of new carriers Market liberalization has gen- erated hundreds of new carriers that need low cost options for bandwidth, and many new entrants do not have the up- front capital required to participate in the construction of Club systems. The Internet The growth of data services, both in the form of private intranets and the public Internet, has placed unpredictable yet very large new demands on the cable planning process. Club rules, nonetheless, compel newcom- ers to purchase capacity at the outset in the form of Indefeasible Rights of Use (IRUs), which typically last the lifetime of the cable. For many players, however, the incre- mental acquisition of bandwidth is much more practical. To meet this demand, some private systems, like FLAG, have abandoned the IRU purchase model and are selling capacity in short leases, even as short as three years. Competition among incumbents Former monopoly carders are now competing against one another, both through alliances (e.g., BT/AT&T vs. Global One) and through foreign affiliate operations. Club members are less likely to find the motivation to cooperate on long term projects with partners who may end up becoming competitive threats. Thus they are more willing to let private investors fund new cables. The rise of private cables has not, however, stopped the plan- ning and building of "quasi-Club" systems. In fact, a new quasi-Club Atlantic cable, TAT-14, scheduled to begin service in 2000, will have more than four times the capacity of all exist- ing cables on the route. Future regulation and access terms for this cable (private or common carrier model) will be closely watched (see the map of "International Submarine Cables" on page 86). ~.~ TeleGeography 1999 © TeleGeography, Inc. 1998 New Cross-Border bg Gr~h~rn Finnie, Networks in Europe Yankee Group Europe Of one thing we can be quite sure: Europeans will have access to a lot more cross-border bandwidth in the next two years. But trying to decide exactly how much new bandwidth is being installed or is actually commercially available is like counting bees--difficult at the best of times, and near-impossible when they are flying. Unlike submarine cable systems, which are relatively transpar- ent, Europe has an intricate web of fiber optic cables in the ground. And they are largely owned by a disparate group of national utilities. Information on the web is patchy, and not all of it has been made commercially available. Quite a bit is in the hands of utilities which have signed exclusive deals with one partner, often a foreign telco--an approach which has attract- ed the interest of anti-trust authorities. Meanwhile, new inte- grated networks are being laid alongside this existing patch- work infrastructure, and new technologies are being deployed which greatly boost bandwidth per fiber pair on both new and old cables. On top of all that, a vigorous market in dark fiber has developed, and a market in wavelengths is just beginning to gather speed. Yet neither dark fiber, which is not in service, nor lit fiber, which is, has a determinate bandwidth. Background Europe’s long-distance and cross-border infrastructure market was liberalized in 1996. It looked hugely attractive for poten- Figure 1. MC! WorldCom’s European Network Dublin Source: MCI WortdCom, Inc. (http:#www.wcom.com) Map design by TeteGeography, Inc. © 1998 © TeleGeography. Inc. 1998 TeleGeography 1999 Figure 2. New European International Networks Organizations Major Cities Bandwidth Commercial Type Launch BT/Concert 200 Gbps 1999 Mostly laid cable owned by European domestic partners; estimated 7,000 route kilometers. Cable & Wireless 1998 Carrier One Esprit Flute Global Crossing Hermes Level 3 Viatel MCl WorldCom Amsterdam, Antwerp, Brussels, DUsseldorf, Frankfurt, London, Marseille, Milan, Munich, Paris, Rotterdam, Strasbourg, ZUrich Amsterdam, Barcelona, Brussels, Geneva, London, Madrid, Milan, Paris, ZUrich Plans to add more second-tier cities in Austria, Denmark, France, Ireland, Italy, Portugal, Sweden 1998: Amsterdam, Geneva, Frankfurt, London, Paris, Vienna 1999: Brussels, Copenhagen, Dublin, Madrid, Milan, Stockholm 1998: Amsterdam, London, Paris Planned: Brussels, DUsseldorf, Frankfurt, Lyon, Marseille, Strasbourg, Stuttgart Amsterdam, London 1999: Amsterdam, Antwerp, Brussels, Cologne, Copenhagen, DLisseldorf, Frankfurt, Hamburg, Hanover, London, Paris, Rotterdam, Strasbourg. 2000: Lyon, Marseilles, Milan, Turin, ZUrich Later: Barcelona, Berlin, Madrid, Munich, Rome, Stuttgart, Vienna 1998: Amsterdam, Antwerp, Brussels, D~isseldorf, Frankfurt, Geneva, London, Milan, Munich, Paris, Rotterdam, Strasbourg, Stuttgart Amsterdam, Antwerp, Brussels, DUsseldorf, Frankfurt, London, Paris, Rotterdam, Strasbourg, Stuttgart 1998: Amsterdam, Brussels, Frankfurt, London, Paris 1999-2000: Berlin, Copenhagen, DLisseldorf, Geneva, Hamburg, Lyon, Madrid, Marseille, Milan, Munich, Oslo, Stockholm, Stuttgart 2 x 155 Mbps initially; more planned 34 Mbps initially; more planned 2.5 Gbps initially; upgrade likely 2.5 Gbpsinitially; more in response to demand Total available bandwidth is 9.6 Tbps or more, depending on cable size; imtial lit capacity not known 180 Gbps 20 Gbps imtially; will move to 160 to 320 Gbps w~thin 12 months 40 Bbps; likely to be upgraded 1998 1998(London to Paris) 1999 !999 1998 1999 1999 1998 Leasing dark fiber from third parties where it can; otherwise leasing conventionally. Will not dig. Currently based on leased capacity but intends to lay fiber (national net in Germany) and lease darkfiber on some routes. Based almost entirely on dark fiber. Submarine cable venture with 12 fiber pairs; will lease bandwidth and dark fiber to third parties. Almost entirely laid cable; mimmum 24 fiber pairs, 48 on terrestrial routes, 144 within Germany. Global Crossing will be a carriers’ carrier only, and will link network to other transcontinental and regional networks in the Americas and elsewhere. Will increase capacity to 320 Gbps next year. Planning 5,000 km European network based on laid cable or dark fiber and linking ten mainly financial centers. More laid cable than dark fiber; 3,700 route kilometers in total. More laid cable than darkfiber (over 3,000 route kilometers); laid cable ~s 2x24 fiber pa~rs. Source: Yankee Group Europe © TeleGeography, Inc. t998 TeleGeography 1999 © TeleGeography, Inc. 1998 tial entrants: the price of bandwidth was (and is) very high, and--incredibly--Europe did not (and does not) have a pan- continental network managed and run by a single operator. Rather, it’s a hodgepodge of national networks linked up at bor- der points. Early in the 1990s, under pressure from frustrated multinational end users, Europe’s national monopoly telcos made a belated at-tempt to create a common infrastructure, variously known as GEN and ME-FRAN. But at best it did no more than ameliorate the situation slightly. And prices remained far above costs, despite the passage in 1992 of a so- called Open Network Provision (ONP) directive on leased cir- cuits that called specifically for cost-oriented tariffs. Right up until this year, prices were still typically five to ten times higher for a cross-border circuit in Europe than for comparable circuits in the U.S. Yet despite this tempting backdrop, early entrants soon found that simply setting up an office and laying cable across the con- tinent was easier said than done. Continent-wide fiber cable networks built to a single specification, and entirely owned by one group, do not yet exist, and likely will not for some years yet. Instead, the new players, many of which are consortia owned or operated by more than one company, are using a patchwork of laid and leased cable. The earliest entrant, Hermes Europe Railtel BV, was conceived eight years ago. It planned to lay cable along rights of way owned by railway companies, but the problems of working with a large group of state-owned national railway authorities proved unmanageable. Now, only two railway authorities are in the group (it’s almost 90 percent owned by U.S. entrepre- neur Global TeleSystems), and the network is a mix of laid fiber and dark fiber Fully 70 percent of the network is based on dark fiber leased from a variety of utilities, including railway authorities such as the Belgian Railway Authority and Eurotunnel. Other utilities selling capacity in Europe include highway management companies such as France’s SANEF, gas and pipeline companies, electricity companies, and waterways. Most entrants have followed Hermes’ lead to varying extents. Esprit Telecom, for instance, is building a network that is based almost entirely on leasing dark fiber Cable & Wireless is fol- lowing suit. New technology makes a network based on dark fiber rather than laid cable doubly attractive. Because dense wavelength division multiplexing (DWDM) componentry became widely available in 1998, most entrants don’t believe they need to light more than one or two fiber pairs to supply what they need. Hermes, for example, is using 40-wavelength technology that will allow each fiber pair to operate at 100 Gbps, and will move to 80-wavelength technology next year. It only has two pairs lit, and doesn’t believe it will need to light any more in the fore- seeable future. Viatel has gone further, using Lucent’s new DWDM technology, which allows 40 x 10 Gbps capacity per pair Despite that, some companies clearly have the strategic intent to lay cable if they can. Though MCI WorldCom won’t say how much of its network is based on dark fiber leased from utilities, we believe it’s less than 30 percent, and certainly a lower pro- portion than Hermes. The same is true of Viatel, despite the very high capacity it can achieve per fiber pair And big new entrants with huge ambitions, such as BT and its European partners, will build a network that is based almost entirely on their own laid cable. They, of course, have the advantage that much of the network either already exists or is being laid any- way. Most ambitious of all, Global Crossing Ltd., a Los Angeles- based start-up which raised S400 million in an August 1998 IPO and already has completed a trans-Atlantic cable (AC-1), announced plans in October to build out a pan-European net- work with between 48 and 144 fiber pairs, yielding a minimum 10 Tbps capacity, dwarfing other networks. It will link 13 cities by the end of 1999, and another five the following year Meanwhile, other options are muddying the distinction between owners and lessors of infrastructure. As well as leasing dark fiber to third parties, some cable owners such as Racal in the U.K. (which bought out the British Rail national fiber network) are considering leasing specific wavelengths, or ’lambdas’ in industry jargon. Others (including for example Viatel) are leas- ing capacity to customers for the lifetime of the cable, on the same principle (the so-called indefeasible rights of use or IRUs) as capacity in submarine cables. So How Much Bandwidth Exists? Figure 2 only includes those companies that have firm, declared plans to lay cable or to build a network based on dark fiber, or that have already begun to do so. There are several other companies that had not declared their hand at the time of writing, but are likely or possible entrants. These include in particular Global One, the joint venture between Deutsche Telekom, France T~l~com and Sprint; Global Crossing, the U.S. start-up that has already commissioned Atlantic Crossing and Pacific Crossing; and Qwest, the U.S. national Internet Protocol (IP) network operator that bought one of Europe’s biggest lnternet operators, EUnet, in April 1998. What does all of this mean for bandwidth availability? It’s dif- ficult to be definitive. All that can be said with confidence is that bandwidth potentially available will increase by several orders of magnitude in a very short period. On key routes in northwestern Europe especially to and from Amsterdam, Brussels, Frankfurt, London, and Paris--there is likely to be between 500 Gbps to one Tbps of lit capacity by the end of 1999 compared to no more than ten Gbps before the boom © TeleGeography, Inc. 1998 TeleGeography 1999 began. Beyond that, capacity will increase quickly to major cities to the east, south, and north, such as Zi~rich, Geneva, Milan, DOsseldorf, and Stockholm, and also to many smaller intermediary cities such as Stuttgart and Strasbou~. Some players, including WorldCom and Viatel, are planning to build extensive national networks in the bigger countries like Germany. In principle, then, it sounds like a glut is on the way: even the most enthusiastic of Internet, IP and e-commerce zealots might find it difficult to argue that actual utilization will increase 50- 100 fold in two years. In reality, though, in the same way that quite a bit of bandwidth on new trans-Atlantic routes has van- ished into the vaults of old and new telcos and service providers (and one or two futures entrepreneurs), so the new bandwidth in Europe may not actually be utilized by or avail- able to ultimate end users, but rather hoarded for a whole vari- ety of reasons. That might pose some new issues for regulators in Europe who are anxious to ensure that Europeans get enough long-distance, cross-border bandwidth to allow elec- tronic commerce to flourish and knit the continent’s member states together. Yet even in the worst-case scenario, there’s no going back to the bad old days that beleaguered Europe’s bandwidth junkies. Bandwidth will be plentiful, prices will in time fall to U.S. levels, and most mainstream telcos will need to do some very rapid re- thinking about tariffs. As the falling price of commodity band- width feeds through to the price of commodity international phone calls, rapid revenue attrition is a near-certainty. Cutting dependence on phone usage revenues will be the key to success in the new decade. ~i~ Graham Finnie is Research Director at Yankee Group Europe (YGE) and can be reached at graham@yankee.co.uk. Contact YGE at: (tel) ÷44 1923 24 6511; (fax) ÷44 1923 24 2456. Figure 3. Hermes" European Network Key Sw~tch Facdity O Planned Facility (during 1998) ........ Planned Network Development Int’l. Network .0 Stockholm /." Copenhagen,c~" Rotterdam ~. Hamburg Amsterdam~ "’.0 ," i 1 ........ ]~ene~ Milan : : .." Barcelon~ d o Madrid Source: Hermes Europe Railtel BV (http://wvwv.her.com) Map design by TeleGeography, Inc. © 1998 TeleGeography 1999 © TeleGeography, Inc. 1998 International Circuit Usage by U.S. Carriers Figure 1. International Circuit Usage by Region, 1995-1997 U.S. Carrier 64 Kbps Circuit Usage For Private For Public Switched Total Circuits Idle Total Lines Network In Use Circuits Available Americas 1995 9,489 79,892 89,381 10,789 100,170 1998 38,170 82,801 120,971 9,432 130,403 1997 79,358 102,367 181,725 11,498 193,221 W. Europe 1995 9,997 22,389 32,386 54,593 88,979 1996 33,083 29,536 62,619 33,053 95,672 1997 43,784 34,476 78,260 46,245 124,505 E. Europe 1995 241 2,886 3,127 1,470 4,597 1996 478 3,344 3,822 1,704 5,528 1997 1,328 3,742 5,068 1,719 6,787 Middle East 1995 506 2,560 3,066 268 3,332 1996 908 2,836 3,744 560 4,304 1997 1,432 3,096 4,528 479 5,007 Africa 1995 199 2,051 2,250 181 2,431 1996 406 2,416 2,822 327 3,149 1997 699 2,608 3,307 292 3,599 Asia 1995 5,067 13,185 18,252 26,605 44,857 1996 15,015 16,475 31,490 27,163 58,653 1997 23,545 19,567 43,112 30,830 73,942 Oceania 1995 998 3,125 4,123 1,628 5,751 1996 3,302 3,110 6,412 2,523 8,935 1997 5,430 4,861 10,291 1,690 11,981 Total 1995 26,497 126,150 152,647 n.a. n.a. 1996 91,362 140,518 231,880 74,762 306,642 1997 155,574 170,717 326,291 92,751 419,042 Note: Data based on year-end FCC circuit status reports filed by AT&T, MCt, Sprint, and WorldCom for circuits originating in continental U.S. only. "Idle" cir- cuits are circuits owned by a carrier at year end but not in use. Satellite capacity uSlization is generally not reflected in this data because U.S. carders do not acquire international satellite capacity in advance. The FCC estimates that 25-30 percent of total submarine cable capacity landed in the U.S. is controlled by foreign carriers and thus not reported here. Also, up to t00 percent of used capacity goes unreported because it is reserved for restoraSon purposes only. International Private Line Circuit Usage of U.S, Carriers by Region, 1995-97 80,000 70,000 60,000 50,000 40,000 30,000 20,000 10,000 0 [] 1995 [] 1996 1997 Americas W. Europe Asia Oceania Rest of WorLd Source: FCC © TeleGeography, Inc. 1998 © TeleGeography, Inc. 1998 TeleGeography 1999 1995 Americas 1996 1997 1995 W. Europe 1996 1997 1995 E. Europe 1999 1997 1995 Middle East 1996 1997 1995 Africa 199s 1997 1995 Asia 19~ 1997 1995 Oceania 1996 1997 Circuit Usage and Idle Capacity, 1995-97 ¯ IPLs [] PSTN rq Idle I excess capaci~ 0% 20% 40% 60% 80% 100% Percent of Total 64 Kbps Circuits 80,000 70,000 60,000 50,000 40,000 30,000 20,000 10,000 0 Idle Circuits of U.S. Carriers by Region, 1995-97 Americas [] 1995 [] 1996 1997 W. Europe Asia Oceania Rest of World Source: FCC © TeleGeography, Inc. !998 TeleGeography 1999 © TeleGeography, Inc. 1998 Figure 3. International Circuit Usage for Selected Routes, 1995-1997 U.S. Carrier 64 Kbps Circuit Usage For Private For Public Switched Total Circuits Idle Total Lines Network In Use Circuits Available Canada 1995 5,543 44,172 49,715 1,936 51,651 1996 29,698 41,793 71,491 917 72,408 1997 68,383 50,343 118,726 1,178 119,904 M exi c o 1995 1,653 23,416 25,069 800 25,869 1996 13,312 27, 784 41,096 840 41,936 1997 19,155 36,935 56,090 1,148 57,238 Hong Kong 1995 860 742 1,602 1,036 2,638 1996 1,921 961 2,882 3,722 6,604 1997 3,058 1,221 4,279 1,825 6,104 J a pan 1995 2,241 4,619 6,860 16,259 23,119 1996 7,682 5,354 13,036 17,696 30,732 1997 10,087 6,149 16,236 17,178 33,414 Singapore 1995 521 306 827 593 1,420 1996 1,114 582 1,696 508 2,204 1997 1,617 570 2,187 571 2,758 U.K. 1995 6,048 8,317 14,365 27,001 41,366 1996 18,959 12,648 31,607 10,844 42,451 1997 23,008 14,662 37,670 20,118 57,788 Note: Date based on year-end FCC circuit status reports filed by AT&T, MCI, Sprint, and WorldCom for circuits originating in continental U.S. only. "Idle" cir- cuits are circuits owned by a carrier at year end but not in use. Satelhte capac=ty utilization is generally not reflected in this date because U.S. carriers do not acquire international satellite capacity in advance. The FCC estimates that 25-30 percent of tetel submarine cable capacity landed in the U.S. is controlled by foreign carriers and thus not reported here. Also, up to 100 percent of used capacity goes unreported because it is reserved for restoration purposes only. Circuit Usage and Idle Capacity, 1995-1997 1995 Canada1999 1997 1995 Mexico1996 1997 1995 Hong Kong199~ 1997 I [] IPLs [] PSTN D Idle t 1 0% 10% 20°/0 30% 40% 50% 60% 70% 80°/0 90°/0 100% Percent of Total 64 Kbps Czrcu~s Source: FCC © TeleGeography, Inc. t998 10o © TeleGeography, Inc. 1998 TeleGeography 1999 A Primer on Bits Figure 1. Measuring Bytes Bit by Bit Below are the standard metric prefixes used in the SI (Syst~me International) conventions for scientific measurement. With units of time (e.g., gigabits per second) or things that come in powers of 10, they retain the=r usual meanings of multiplication by powers of 1,000 = 103. When used with bytes (e.g., gigabytes of data storage) or other things that naturally come in powers of 2, they usually denote multiplication by pow- ers of 1,024 = 2l°. Base 10 Base 2 1 Kilobit/s = 1,0001 = 103 = 1,000 1 Kilobyte = 1,0241 = 21° = 1,024 1 Megabit/s = 1,0002 = 106 = 1,000,000 1 Megabyte = 1,0242 = 22o = 1,048,576 1 Gigabit/s = 1,0003 = 109 = 1,000,000,000 1 Gigabyte = 1,0243 = 230 = 1,073,741,824 1 Terabit/s = 1,0004 = 1012 = 1,000,000,000,000 1 Terabyte = 1,0244 = 240 = 1,099,511,627,776 1 Petabit/s = 1,0005 = 1015 = 1,000,000,000,000,000 1 Petabyte = 1,024~ = 25o = 1,125,899,906,842,624 1 Exabit/s = 1,0008 = 1018 = 1,000,000,000,000,000,000 1 Exabyte = 1,0246 = 26o = 1,152,921,504,606,846,976 1 Zettabit/s = 1,0007 = 1021 = 1,000,000,000,000,000,000,000 1 Zettabyte = 1,0247 = 270 = 1,180,591,620,717,411,303,424 1 Yottabit]s = 1,0008 = 1024 = 1,000,000,000,000,000,000,000,000 1 Yottabyte = 1,0248 = 28o = 1,208,925,819,614,629,174,706,176 © TeleGeography, Inc+ 1998 Figure 2. Measuring Telecommunications Bandwidth--DS-O to 0C-192 ~. ¯ Carrier Technology Data Rate(Mbps) Description 64 Kbps Circuits* DS-0 0.064 Base rate in the Digital Signal (DS) ~evel h~erarchy 1 T-1 (DS-1) 1.544 Primary level of the American T-carrier multiplexing 24 system; capacity isthe same as a DS 1 carrier T-2 (DS-2) 6.312 Four times the capacity of T-1 96 T-3 (DS-3) 44.736 28 times the capacity of T-1 672 T-4 (DS-4) 274.176 168 times the capacity of T-1 4,032 E-1 2.048 Primary level of the European E-carr=er multiplexing system 30 E-2 8.448 Carries four multiplexed E-1 signals 120 E-3 34.368 Carries four E-2 signals 480 E-4 139.264 Carries four E-3 signals 1,920 E-5 565.148 Carries four E-4 signals 7,680 OC-1/STS-1 51.840 Basic signaling rate of SONET hierarchy 810 OC-3/STM-1 155.520 Exactly three times the capacity of 0C-1"* 2,430 OC-12/STM-4 622.080 12 times the capacity of 0C-1 9,720 0C-24 1,244.160 24 times the capacity of 0C-I 19,440 0 C-48/STM 16 2,488.320 48 tim e s th e c a pa city of 0 C-1 38,880 0C-192/STM-64 9,953.280 192 times the capacity of 0C-1 155,520 Key °DS" -OC- "STM" "STS" T-carrier system in U S., Canada, and Japan w=th 1.544 Mbps as the primary level (24 vmce channels x 64 Kbps per channel). Dtgttal Stgnal that travels on the T-carner or E-carrier Used in countries other than U.S., Canada, and Japan The h=erarchy was established by the CEPT (Conference Europ~enne des Postes et T~l~communmations) with 2.048 Mbps as the primary level ([30 voice channels 2 channels for overhead( x 64 Kbps per channel). 0pt~cal Carr=er interface des=gned to work with STS-n (Synchronous Transport Signal) s=gnahng rate in a SONET (Synchronous Optical Network) Synchronous Transport Module refers to a large carrier (base s=gna1155.52 Mbps) =n a SONET Synchronous Transport S~gnal is the electrical counterpart to the Optmal Carrier (OC). The number of 64 Kbps is presented for comparat=ve purposes only. The actual number of simultaneous conversations poss=bla over a given carrier may vary depending on the encoding scheme used. ** In the "E" and "T" hierarchies, each higher level is set to be =almost but not exactly" a multiple of the bit rate for the previous order (plesiochronous). To elimi- nate problems associated w=th plesiochronous multiplexing, SONET, a synchronous h=erarchy, was defined m the United States in 1986. As a result, the "OC" and "STM" careers are exact bit-rate multiples of their primary levels, 0C-1 and $TM-1, respectively. Source: TeleGeography, Inc.; Alcatel; Newton’s Telecom Bictiona~y © Tele6eography, Inc. 1999 101 TeleGeo~raphy 1999 © TeleGeograph~l, Inc. 1 998 International Communications Satellites iv iv lV lv ¯ ¯ ¯ 30~W Iv iv my ¯ ¯ ¯ 15°W my m~ IV l’~ Key ¯ Six 36-MHz transponder equivalents for the C-bond (4/8 GHz) ¯ Six 36-MHz tronsponder equivalents for the Ku-band (11-IZ/14 GHz) [] V Outline denotes sotellites under construction as of August 1998 The maps on this poge show 36 of the approximately 180 commercial geosation- ary satellites scheduled to begin operotion by Jonuary 2001. These satellites use C- and Ku-band transponders to carry voice, data, and video. On some sotellites (such as Panamsat’s), a majority of capacity is used for video transmission services. A 36 MHz C-band transponder can generally carry about 2,000 calls, but requires a large dish for TV reception. Smaller Ku-bond dishes are increasingly the technology of choice for direct broadcast satellite recep- tion. Intersputnik satellites carry only one (Gorizont and Raduga) ortwo (Express) Ku- band transponders at 36 MHz, therefore the actual Ku-band capacity for these satellites is somewhat less than suggested on the maps. Source: TeleGeography, Inc., /nternational Satellite Directory 1998, Euroconsult 102 I~°E © TeleGeography, Inc. 1998 TeleGeography 1999 103 TeleGeography 1999 © TeleGeography, Inc. 1998 Figure 2. Major International Telecommunications Satellites in Geostationary Orbit Transponders (36 MHz Equivalent) Satellite Bus Launch Date Orbital Slot C-band Ku-band Panamsat PAS-1 GE 3000 June 1988 315.0°E 24.0 12.0 PAS- 1R HS-702 May 1999 315.0% 36.0 36.0 PAS-2 HS-601 July 1994 169.0°E 25.1 25.1 PAS-3 HS-601 Jan. 1996 317,0°E 25.1 25.1 PAS-4 HS-601 Aug, 1995 68.50E 25.1 24.6 PAS-5 HS-601 HP Aug. 1997 302.0°E 24.0 24.0 PAS-6B FS-13003 Aug. 1997 317.00E -- 36.0 PAS-7 FS-13003 June 1998 68.5°E 14.0 30.0 PAS-8 FS-13003 June 1998 166.0°E 24.0 24.0 Intelsat Intelsat 505 Ford Aerospace Sept. 1982 72.0°E 34,0 12,0 Intelsat 510 Ford Aerospace March 1985 33.0°E 40.0 12.0 Intelsat 511 Ford Aerospace June 1985 330.5°E 42.0 12.0 Intelsat 601 HS-393 Oct. 1991 325.5°E 64.0 24.0 Intelsat 602 HS-393 Oct. 1989 62.0°E 64.0 24.0 Intelsat 603 HS-393 March 1990 335.5°E 64.0 24.0 intelsat 604 HS-393 June 1990 60.0°E 64.0 24.0 Intelsat 605 HS-393 Aug 1991 332.5°E 64.0 24.0 Intelsat 701 FS-1300 Oct. 1993 180.0°E 42.0 20.0 Intelsat 702 FS-1300 Jun. 1994 177,0°E 42.0 20.0 Intelsat 704 FS-1300 Jan. 1995 66.0°E 42.0 20.0 Intelsat 705 FS-1300 March 1995 342,0°E 42.0 20.0 Intelsat 706 FS-1300 May 1995 307.0°E 42.0 28.0 Intelsat 707 FS-1300 Feb. 1996 359.0°E 42.0 28.0 Intelsat 709 FS-1300 July 1996 310,0°E 42.0 20.0 Intelsat 801 AS-7000 Feb. 1997 328.5°E 64.0 12.0 Intelsat 802 AS-7000 June 1997 174.0°E 64.0 12.0 Intelsat 804 AS-7000 Nov. 1997 64.0°E 64.0 12.0 Intelsat 805 AS-7000 June 1998 304.5°E 36.0 6.0 Intelsat 901 (IS-IX1) n.a. Ju/y 2000 60.0% 76.0 20.0 Inte/sat g02 n.a. Q4 2000 62.0% 76.0 20.0 APR- 1 n.a. 04 1998 83.0% 11.0 n.a. Orion Orion-1 Eurostar 2000 Nov. 1994 322.5°E -- 48.0 Orion-2 Eurostar 2000 0.2 1999 343.0% -- 45.0 Orion-3 HS601HP Nov. 1998 139.0% 10.0 44.0 New Skies Satellites, NV** New Skies 513 Ford Aerospace May 1988 183,0°E 42.0 21.3 New Skzes 703 FS-1300 Oct. 1994 57.0°E 42.3 24.5 New Sk=es 803 AS-7000 Sept. 1997 338,5°E 64.2 16.7 New Skies 806 AS-7000 Q4 1998 319.5°E 36.0 6.0 Intersputnik Express 1 Express May 1994 130.0°E 10.0 2.0 Express 2 Express Oct. 1994 346.0°E 10.0 2.0 Express 6 Express Sept. 1996 80.0°E 10.0 2.0 Gorizont 22 Gorizont Dec. 1995 40.5°E 6.0 1.0 Gorizont 25 Gorizont April 1992 103.0°E 6.0 1.0 Gor~zont 27 Gor~zont Nov. 1992 263.5°E 10.0 2.0 Gorizont 35 Gor~zont Nov. 1992 80,0°E 6.0 1.0 Gor=zont 41 Gorizont Nov. 1993 130.0°E 6.0 1.0 Gorizont 42 Gor=zont May 1994 142.5°E 6.0 1,0 Raduga 25 Raduga Jan. 1995 70.0°E 6.0 -- Raduga 30 Raduga Dec. 1993 85.0°E 3.0 -- LMI-1 A2100AX Dec. 1998 75.0°E 28.0 12,0 Columbia TDRS-5~ TRW Aug. 1991 185.7°E 12.0 17.0 TDRS-6t TRW Jan. 1993 313.0°E 12.0 17.0 Columbia 515tt Ford Aerospace Jan. 1989 319,5°E 42.0 21.4 Note: Italics indicate a system under construction atthis printing. tn March 1998, Intelsat’s Assembly of Parties approved the creation of an independent "spin-off" company known as New Skies Satellites, NV, Six satellites, including those listed here, have been transferred from Intelsat to New Skies primardy for direct broadcast satellite (DBS) services. Since 1992, Columbia Communications Corp. has been leasing from NASA 12 C-band transponders on two TDRS satellites for commercial services over the Atlantic and the Pacific zones, In addition, Columbia leases Atlantic capacity on Inteisat 605. ~t Formerly tntelsat 515, Columbia 515 replaced TDRS 4 at 319.5°E in April 1998. Sources: TeleGeography, Inc.; International Sate!lrte Otrectory 199~ Merrill Lynch © TeleGeography, Inc. 1998 © TeleGeography, Inc. 1998 TeleGeography 1999 Figure 3. International Satellite Capacity and Cost Historical Trans-Oceanic Satellite Voice Paths, 1988-1998 Trans-Atlantic Trans-Pacific 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 78,000 93,000 283000 283 000 496 000 620 800 620 800 710 800 710 800 935 800 1,048,300 39,000 39,000 39,000 27,000 27,000 83,300 234,000 234,000 234,000 424,500 469,500 Note: Data prior to 1993 include Intelsat satellites only. After 1989, deployment of D=gltal Code Multiphcatzon Equzpment (DCME) made 5:1 compression possible where only 2:1 had been used prewously. Capac=ty estimates exclude one Intelsat satellite m each region held in reserve. Source: TeleGeography, Inc. and Euroconsult (TeL +33 1 43 38 06 00, Fax +33 1 43 38 12 40 Intelsat Cost per Circuit per Year, 1970-2000 $8,000 $7,000 $6,000 $5,000 s4,ooo ~$3,000 $2,000 ~.-VA $1,000 =S~=mA 1970 1975 1980 1985 1990 1995 2000 © TeteGeography, Inc. t988 105 TeleGeography 999 © TeleGeography, Inc. 1998 The New Global Traffic Routing, Mobile Satellite Services Settlements and Demand Until recently, using a "sat phone" meant carrying around a briefcase-sized unit full of equipment and endudng a half-sec- ond lag in conversation. And only lnmarsat had global cover- age, with TMI Communications and Amedcan Mobile Satellite Communications {AMSC) limited to service in North America. The next generation of global mobile satellite service (MSS) ventures hope to change all that. Iridium will bet~in full com- mercial service in November 1998, offering telephone calls from cellular-sized handsets to or from almost anywhere on the planet. In addition to Iridium, Globalstar and ICO are expect- ed to be the eady leaders in the field and together will operate 124 satellites in low and medium earth orbits. The various orbital configurations and satellite specifications have been well covered in the media, but the critical ground segment deserves a closer look from the international carrier’s perspective: ¯ How will these MSS ventures deliver international calls? ¯ How will they intemonnect with the terrestrial wireline and wireless networks? ¯ What will carriers pay each other for handing off traffic? To explore these questions, it is helpful to distinguish the basic amhitectures and call routing strategies to be employed by the three major MSS players (see Figure 1 ). Figure 1. Network Architecture for Global MSS Diagrams show a sample call completed between a handheld MSS subscriber in central Asia and a PSTN number in Paris, Iridium Globalstar ICO Key i MSS Subscriber Landline Terminal A Gateway Market Entry Iridium 1998 Globalstar 1999 ICO 2OOO ..... Space Segment PSTN/PLMN Terrestrial ICONET No. of No. of Satellites Gateways 66 11 48 65-75 10 12 Note: Diagrams showthe different network architectures of each system; actual gateway locations, terrestrml routing, and cellular coverage may vary, © TeleGeography, Inc, 1998 106 © TeleGeographyo Inc. 1998 TeleGeography 1999 Est. Usage Est. Capacity Pre-Operational Cost per min Cost per min (min/sub/mo) (billions of min/year) Costs ($ Millions) @ 1B min/year @ 3B min/year Iridium 50 1.5 $4,361 $1.65 $0.62* Globalstar 100 6.0 $2,539 $0.82 $0.40 I CO 100 3.0 $4,652 $0.64 $0.28 Note: System traffic of one billion minutes would require more than 830,008 subscribers using 100 minutes per month. All fixed costs are amo~zed over the life of each system’s satellites: Iridium, 7 years; Globalstar, 7.5 years; ICO, 12 years. Source: Merrill Lynch & Co., Dresdner Kleinwor~ Beeson. *This figure is presented for comparative purposes only. © Tele6eography, Inc. 1998 Iridium Iridium relies upon a fleet of 66 low earth orbit (LEO) satellites. Inter-satellite links route a subscriber’s call to the gateway nearest the called party, from which the call is routed via the Public Switched Telephone Network (PSTN) or the Public Land Mobile Telephone Network (PLMN) to its destination. By transmitting calls from one satellite to another, Iridium needs fewer gateways on the ground and bypasses terrestrial tele- phone networks for much of the call route. The Iridium space seBrnent charge ($1.00 $I .-/5 per minute) covers the inter- national charges to the gateway; the gateway operator is then responsible for entering into interconnection arrangements with such national or international carriers as it requires to terminate traffic. Settlement charges may apply, but typically only from the gateway country (not the originating country) to the termi- nation country. Globalstar Globalstar plans to begin service in 1999 and will use LEO satellites to relay signals from its subscribers directly to the ter- restrial gateway closest to the point of call termination. AJI Globalstar traffic will enter the PSTN or PLMN from the local gateway, which will be independently owned and operated. Gateway operators will be compensated by Globalstar and, in turn, will pay to terminate calls. The system is, in effect, wire- less access to the PSTN. This network architecture requires that there be at least one gateway in view of each satellite at all times. Globalstar has, however, opted to install more gateways than the architecture requires in an effort to reduce the distance from the gateway to the call recipient and lower the cost of ter- minating the call. ICO ICO will field its middle earth orbit (MEO) system in 2000. Because of the higher orbit and consequently broader footprint of its satellites, ICO will require fewer gateways than Globatstar. However, where both Globalstar and Iridium hand off their traf- fic to independent gateway operators, losing control--and associated revenue--arising from the ground segment, ICO will own its gateways. ICO is also investing in its own terrestrial network, called ICONET. The company, thereby, hopes to limit interconnection payments to third parties. An Ice satellite will route subscriber traffic to the gateway that offers Ice the low- est-cost terrestrial route. For example, a call made by an Ice user in North Africa to a PSTN number in a neighboring coun- try may be routed through Europe, where long distance opera- tors offer Ice cheaper rates than the host African country. PSTN/PLMN Interconnection Challenges Despite efforts to distinBuish their services, the target markets for these global mobile ventures are essentially the same: cel- lular extension services, basic mobile, aeronautical/maritime specialty services, and fixed site communications. For each of these markets, maintaining a seamless link between the space Figure 3. First Generation MSS Operators Est. 8/98 Usage Market Entry Voice Units* (min/sub/mo) Inmarsat 1991 17,000 90 AMSC 1995 11,000 100 TMI 1998 1,550 80 Voice umts exclude maritime systems. It is tempting to look to first-generation Geostationary Earth Orbit (GEO) MSS providers like Inmarsat, AMSC and TMI to estimate the demand for the services of the global mobile new entrants. However, significant differences in their target customers and appli- cations could make the comparison misleading. Service of AMSC and TM1 is limited to North America and the Caribbean, and over 98 percent of subscribers use the services purely for domestic pur- poses. Mobile data users also far outstrip the base of voice termi- nals, especially for AMSC. Inmarsat does provide global service but it has primarily tangeted MSS users who need fixed or semi- fixed communications in remote areas like hinterland logging sites or offshore oil platforms. In contrast, smaller handheld telephones will allow new MSS carriers to target large numbers of mainstream subscribers for cellular extension and basic mobile services. Thus, despite the very small subscriber base of existing MSS operators, supporters of Iridium et al remain surprisingly bullish. © TeleGeography, Inc. t998 107 TeleGeography 1999 © TeleGeography, Inc. 1998 segment and terrestrial networks (wireless or wireline) is critical and requires that operators develop a globally dispersed set of alliances with carders and manufacturers. Such alliances will prove vital not only for access to the PSTN/PLMN, but for mar- keting and securing regulatory authorizations. There are challenges inherent in these relationships, however MSS operators generally insist that their traffic will not canni- balize existing carder revenue streams and will complement rather than bypass the lucrative international settlement streams which now benefit poorer countries with low teledensi- ty--the very countries, of course, which are viewed as prime targets for new offerings. Even Iridium, whose satellites will capture additional revenues by switching subscribers’ interna- tional traffic via inter-satellite links, contends that this is traffic that would otherwise have been absent from the terrestrial net- works. Potential for conflict exists, though, over the extent to which the traditional accounting rate regime will be applied to MSS traffic as it is commingled with conventional international mes- sage telephone services (IMTS) traffic. The Federal Communications Commission (FCC) has exempted MSS traffic from its International Settlements Policy, which requires the equal division of accounting rates, equal treatment of all U.S. carriers, and proportionate return of inbound U.S. traffic. This allows for private line routings on high-volume routes, bypass- ing conventional accounting rate settlements. But few other governments have yet resolved how global MSS traffic will be settled. In any case, most operators expect that despite their global footprint, more than half of their traffic will be domestic, which will dampen the impact on international settlements whatever they turn out to be. And gateway traffic delivered by Gtobalstar will generally be treated the same as any other inter- national traffic stream (see Figure 4 for projected MSS interna- tional traffic). Trans-border roaming presents another potential conflict affect- ing MSS international traffic. The promise of MSS lies in the ultimate flexibility to reach anyone from anywhere. The cellular model, however, suggests that some nations may refuse to allow the use of transceivers licensed in another country. To avoid the problem, the International Telecommunication Union (ITU) has sponsored a memorandum on free circulation of handsets, which many countries have accepted. By pursuing partnerships with cellular providers, global mobile operators hope to take advantage of their access to an explod- ing pool of target customers--the number of worldwide cellular subscribers has grown to over 200 million and is projected to reach 450 million by 2000 and exceed one billion by 2006. In addition, operators hope to benefit from cellutar’s established billing and customer service resources. These obvious strategic benefits are mitigated by the nature of the ce!lular/MSS rela- tionship. All three major operators have proposed dual- or tri- mode handheld transceivers, which will automatically check if the user is within their partner’s cellular coverage or roaming area and use the less expensive cellular option if available. Any call that an MSS subscriber makes will only be carried on the space segment if the call originates outside the partner’s cellu- lar network or roaming area. Therefore the cellular operator has an incentive to build out its network as far as possible to retain as much traffic--and revenue--as possible. How much will it cost? The current mobile satellite services supported by Inmarsat charge $3-5 per minute for airtime. Retail prices for the national satellite telephone service offered by AMSC and TMI are priced at about $ 1.50 per minute. The next generation of Figure 4. Market Forecasts for Second-Generation Global MSS, 1998-2002 MSS Subscribers (millions) Mobile Satellite Service Operator 1998 1999 2000 2001 2002 Iridium 50 600 1,200 1,800 2,500 Globalstar -- 600 1,510 2,223 3,057 ACeS -- 110 283 403 578 ICO -- -- 588 1,241 2,006 Ellipso -- -- 442 715 1,147 lnmarsat and Regional GEOs 187 268 491 806 1,654 Total Subscribers 237 1,578 4,514 7,188 10,942 Est. MSS Annual Revenue ($m) (CAGR 46.6%) 1,134 3,398 7,123 9,735 13,324 Est. MSS Annual Usage (billions of rain) 0.1 0.3 1.4 3.0 4.7 Total International Traffic (billions of rain) (CAGR 13%) 92.4 104.5 118.0 133.4 150.7 MSS as % of Total International Traffic 0.03% 0.09% 0.37% 0.70% 0.97% Note: MSS traffic is assumed to be 30 percent international Subscribers include voice and data customers. Source: Subscriber and revenue data adapted from Merrill Lynch; total international minutes: TeleGeography, Inc. © TeleGeography, tnc, 1998 108 © TeleGeography, Inc. 1998 TeleGeography 1999 MSS will try to price service below $3.00 per minute to gener- ate the rapid growth necessary to cover their higher fixed costs. Iridium has announced that subscriber rates will be based on the average of the caller’s alternatives in that area--such as hotel rates and payphones--plus an Iridium "mobility premi- um" of 25 to 30 percent; calls will total anywhere between $2 and $7 per minute. As noted earlier, Iridium retains traffic on its inter-satellite network for much of the call route. The com- pany plans to maximize this competitive advantage over its early competitors by targeting subscribers who make high vol- umes of international calls. It is estimated that Globalstar and ICO will charge about $0.85-$1.50 per minute excluding international long distance charges. It is ICO, however, which may provide the lowest over- all price to customers. ICO’s space segment cost is marginally higher than that of Globalstar but ICO should be able to achieve substantial cost savings by taking advantage of com- peting interconnect rates at different gateways on the ICONET. According to a Dresdner Kleinwort Benson report, this savings could amount to as much as 50 percent relative to the full PSTN rate assumed for Globalstar. Demand for Global Mobile MSS operators have identified the importance of quickly load- ing their capacity given the limited lifetimes of their satellites and the "use it or lose it" nature of their minutes for sale. To do this, they have chosen to rely heavily on their own investor base for partnerships that will provide easy access to existing mobile telephone customers. MSS operators must balance the push for early revenues from high prices in "captive" environ- ments with the need to build market share for next generation systems which will be priced competitively with terrestrial cellu- lar. Second generation satellites with increased broadband capaci- ty capable of supporting mobile data applications are planned for 2003. Operators with existing narrowband voice services may have an advantage in marketing the new broadband capa- bilities of their next generation satellites over those who do not. In addition, the next generation satellites will offer operators the opportunity to broaden their target customer base by accommodating the bandwidthqntensive lnternet applications of an increasingly mobile workforce. ~i~ For Further Reading Rob Frieden, "That Pesky Last Mile." Telecommunications Policy, Vol. 22 No. 2, March 1998. Merrill Lynch, Pierece, Fenner, & Smith Inc., Global Satelfite Marketplace 98, April 1998. Dresdner K]einwort Benson, Mobile Satellite Services, March 1998. For more information, visit the providers’ web sites: AMSC (http://www.AmMobile.com) G!obalstar (http://www.globalstarcom) ICO (http://www.ico.com) Inmarsat (http://www.inmarsat.org) Iridium (http://www.iridium.com) TMI (http://www.tmi.ca) 109 Te~eGeography 1999 © TeleOeography, Inc. 1998 The lnternet TeleGeography 1999 © TeleGeography, Inc. 1998 The Global by Kenneth Nell Cukier, Internet: A Primer Communications Week International I. Introduction In Octobe~ 1997, when the elite of France’s high-tech indus- tries held :: ~:pecial meeting at the Senat to discuss "the infor- mation soc,.~.ty,’’ the chief executive of France T~l~com (FT), Michel Bon, downplayed the importance of the Internet. He argued that the country’s proprietary videotext system, Minitel, remained a success. But four months later, F-i- began transitioning Minitel content providers to the Web and by Spdng 1998, the carrier was touting its lnternet telephony research. Then, in September 1998, Deutsche Telekom and FT announced a joint billion dollar network based on lnternet Protocol (IP). It was followed in October by a joint venture with IBM to develop new screen-based telephones for Internet access. Everyone knows that the Internet will change the business of carrying telecom traffic around the world. And several other major carriers, such as AT&T, MCI WorldCom, Cable & Wireless (C&W), and Spdnt, have announced plans to make IP friendly networks the core of their 21st century business. These carri- ers, some a century old, are reacting to upstarts with space-age names, such as Qwest and Level 3, who believe that Internet technology provides a low cost means of building a global com- munications network. And build they will, eyeing data commu- nications-not voice--as the majodty of their traffic and source of future revenue (see "International Circuit Usage by U.S. Carriers" on page 98). Yet to most international carriers, the economics of the lnternet remain uncertain, more threat than promise. The mechanics of the Internet also seem to be in constant flux--who owns what and where? That makes deciding what to build and what to buy, who to connect with and on what terms, more like gam- bling than network planning. The failure to gamble on these decisions, however, just could cost carriers their business. This essay attempts to provide carriers with some of the infor- mation required to make their business decisions less like a roulette game. It maps the Internet’s current reach, sketches how it is changing, and where future opportunities--and dan- gers--lie. Like the Net itself, it scraps national borders and instead takes a global view. It provides possible answers to such questions as: What is dri- ving demand for lnternet bandwidth? What are the major costs for carriers building IP networks? And, how will new lnternet services affect the Net’s physical topography and traffic flows? It also challenges a number of assumptions. For example, new technologies~e.g., Internet telephony--rather than being a major bandwidth driver may instead have only a marginal effect. And caching technologies, seen by some as the band- List of Figures in this Article Figure 1, Geographic Disl~ibution of lntemet Hosts and Content .... 113 Figure 2. The Internet in 1971--Across the U.S. at 56 lOps ....... 114 Figure 3. Major U.S. Internet Backbones in 1998--155 Mbps to 2 Gbps .................................. 115 Figure 4. Asian tntemet Backbone Connectivity ................ 116 Figure 5. European lntemet Backbone Connectivity ............. 117 Figure 6. African lntemet Backbone Connectivity ............... 118 Figure 7. Backbone Connectivity for Latin America and the Middle East .............................. t 19 Figure 8. lntemet Traceroute Maps, July 1998 ................ 120 Figure 9. The World’s Top ISPs ........................... 122 Figure 10. The U.S. Controls the lntemet’s Directory System ....... 123 Figure 11. The lntemet’s New Governors .................... 124 Figure 12. Power Shifts in Intemet Governance: 1982-1998 ...... 125 Figure 13. On the Net, Bandwidth Determines Distance .......... 12.6 Figure 14. Distance and Delay on the Cable 6~ Wireless Backbone ... 127 Figure 15. A Primer on Peering ........................... 128 Figure 16. Figure 17. Figure 18. Figure 19. Figure 20. Figure 21. Figure 22. Figure 23. Figure 24. Figure 25. Figure 26. Figure 27. Figure 28. Figure 29. Figure 30. Figure 31. Regional Internet Growth ....................... 129 Lessons of the MCI-WorldCorn Merger .............. 130 Internet Growth in Western Europe ................ 131 lnternet Host Growth 1981-1998 ................. 132 NAP Traffic as a Measuring Stick .................. 133 Starr on the Net: Big Crowds, But Not for Long ....... 134 A Si~naling Channel for the tntemet? ............... 13S A Slow On-Ramp: Internet Users and Access Speeds .... 136 Why Broadband Matters ....................... 13? Satellites and the tntemet ....................... 138 Internet Telephony: Build Once, Run Anything?. ........ 139 The iTU and the lntemet ........................ 1 Future Scenarios ............................. 141 Voice’s Shrinking Role ......................... 142 tnternet Protocol Developments on the Horizon ........ 1 Quelle Langue is the Intemet? .................... 112 © TeleGeography, Inc. 1998 TeleGeography 1999 width savior of Internet service providers ([SPs) outside the U.S., may only bring ephemeral savings. When all is said and done, this essay remains a primer. The Interoet’s only certainty is constant change. We start with the basic architecture. 11. How Is the lnternet Structured? The answer today is fairly obvious: the Internet is U.S.-centric. Viewed from afar, the Internet’s basic transmission facilities form a big star centered in the U.S. with spokes of light reach- ing around the world--some broad like the beams of a seamh light, others laser thin. Them are many reasons for the architecture, including the U.S. head-start in building infrastructure, the location of lnteroet content, the artificially high cost of cross-border capacity out- side the U.S., and customer demand for lnternet services. Yet, this topology is in many ways unplanned, a reflection of the lnternet’s U.S. origins and its embryonic commercial structure, rather than engineering efficiency. But whether built by acci- dent or design, the Intemet’s architecture matters greatly. It affects the way traffic is routed, how today’s infrastructure is utilized, and the scale and location of the bandwidth required tomorrow. Let us look at history first. When the lnteroet began, it was a national U.S. network (see Figure 2). Later, data networking research institutions outside the U.S. enthusiastically plugged in, leasing international circuits at their own cost to do so. In fact, non-U.S, networks paid for both ends of the circuit, a precedent that still remains and has since escalated to a diplo- matic controversy (see Figure 15). At the time, European and Asian network operators didn’t complain about paying the full cost because the U.S. backbone was "the Interoet." And the data flow was almost entirely one-way--from the U.S. to over- seas locations. American users would only rarely seek to con- nect to non-U.S, networks. Traffic patterns are now beginning to change. More U.S. lnternet users are seeking off-shore content and intra-regional European and Asian traffic is growing, as are facilities there. Yet the lnteroet’s old architecture remains in place, largely because half of Internet users, as well as 58 percent of hosts, are still based in the U.S. and the U.S. is the source of the world’s most popular lnteroet content (see Figure 1). Traffic balances also tell the story. A little over half of lnternet traffic in Europe goes to the U.S., according to two ISPs, Amsterdam-based EUnet and Stockholm’s Telia. In Asia, the figure is 70 percent, according to both Telstra in Australia and Inet Inc. in Korea. That provides a powerful incentive for non- U.S. ISPs to continue buying the international bandwidth to get to the U.S. while at the same time seeking ways to cut associ- ated costs. While users pay for this bandwidth indirectly as Figure 1. Geographic Distribution of lnternet Hosts and Content Internet Host Distribution, Sept. 1997 Oceania 3.1% Europe .~ 24.3%// Africa O.5% Asi~ 6.8% Rest of Americas 7.2% U,S. 58.1% Physical Location of the 100 Most Visited Web Sites Non-U.S. 6% Other U.S. 24% California 4O% Texas 3% Florida 3% Mass. 4% N ew York 16% Washington State 4% Note: Data for host distribution chart d~stributes the generic Top Level Domain (gTLD) registrations between countries according to registrations in September t997. Data for top s~tes chart is based primarily on proxy server legs from strategic locations on the Internet backbone, as welt as on monitoring usage pattern of over 100,000 users (60% U.S., 40% non-U.S.). Source: Host chart adapted by the International Telecommunication Union from data originating with Network W~zards (vwwv.nw.corn), the OECD, and RIPE (ww~v.ripe.net). Top sites chart adapted from Web21 (wvwv.web21.com) data. © TeleGeography, Inc. 1998 TeleGeography 1999 © TeleGeography, Inc. 1998 Figure 2. The lnternet in 1971--Across the U.S. at 56 Kbps ~URROUGHS ............... December 1969 Key September 1971 IMP Terminal IMP Note: The map above reflects data reported in BBN’s "O.uarterly Technical Reports" delivered to ARPA under contract between 1969 and 197t. An Interface Message Processor, or IMP, was a node computer designed to handle the message-sending tasks for several host computers connected in a local network. Source: Adapted from maps drawn by Alex McKenz~e, m Peter Salus’ Casting the Net, Addison-Wesley Publishing Company, 1995, Map design by TeteGeography, Inc. © 1998 part of the monthly subscription fee or leased line cost, it is the service provider who pays up-front. The status quo is also supported by a second economic ratio- nale: ISPs need to intemonnect with one another to furnish connectivity everywhere. It would be impractical for a German ISP to have direct links with every provider in Europe, as well as the Middle East and Latin America. Networks outside the U.S. thus tend to make the U.S. their central switching office (see Figures 4-7). It’s more economical to lease a single line to the U.S. rather than multiple smaller-capacity lines to other for- eign points to connect with other networks. The central role of U.S. networks in switching the world’s lnternet traffic can be highlighted by drawing a series of "traceroutes," a networking tracking procedure to identify the end-to-end path taken by any given lnternet packet. In a recent experiment, packets from Hong Kong to Japan transited Silicon Valley in California, and London to Rome traffic was routed via New York (see Figure 8). This may seem illogical. Shouldn’t traffic between nearby coun- tries be sent by a more direct path? Perhaps. But making 114 sense of Internet traffic patterns is a bit like mastering quantum physics--the longer the route the shorter the time it may take to arrive. There is a major engineering reason, nonetheless, to explain the lnternet’s current architecture. It has to do with the way the addressing system works. Each computer on the lnternet is assigned a unique address or domain name, and to send or receive any information to or from that address, a query is often first sent to the Internet’s electronic directory system or root name servers. Of the 13 root name servers, only three are deployed outside the U.S.--in Tokyo, London, and Stockholm (see Figure 10). If a foreign Internet service provider does not have a direct connection to any of these three cities, chances are that it will send an address query to the U.S. to access the vital domain name system (DNS) data needed to reute the user’s traffic. It’s a serious problem. One request for Web access generates two DNS transactions; one from the customer (to find the server), and one from the server (the reverse look up). Each request may generate a stream of packets. Measured at one California-based exchange © TeleGeography, Inc. 1998 TeleGeography 1999 Figure 3. Major U.S. Internet Backbones in 1998--155 Mbps to 2 Gbps San Francisco ’ San Francisco NAP MAE-WEST FIX-WEST ~.. Los Angeles __ MAE-LA Atlanta Key caw 1 Sprint ~, Major national backbone ..... UUNET 0C-3 to 0C-48 AGIS ,1 Dallas/ Fort Worth :’ MAE-DALLAS Houston MAE-HOUSTON 0C-48 OC 12 Dual 0C-3 0C-3 (2Gbps 622Mbps 31uMbps 155Mbps Chicago i Internet Exchange Point (IX) ,;Atlanta i Other metropolitan area New York Washington MAE-EAST+ FIX-EAST This map shows portions of the lnternet backbones of the four lead- ing Amedcan backbone providers. According to Boardwatch Magazine, these four companies account for at least two-thirds of the backbone connections in the country. Many backbone providers are reluctant to release detailed maps or diagrams of their networks, so it is impossible to show the precise Source: Boardwatch Magazine Directory of lnternet Service Providers (www.boardwatch.com) and company reports. routes or capacities here. Nevetheless, it is clear that a handful of metropolitan areas: New York/Philadelphia; Washington, DC; Atlanta; Chicago; Dallas/Fort Worth; Los Angeles; San Francisco/San Jose; and Seattle--remain the critical nodes, enjoying the fastest con- nections and switching the most traffic, including international traffic. Map design by TeleGeography, Inc. © 1998 point in late 1997, DNS Iookups accounted for 24.4 percent of packet traffic and 9.9 percent of byte volume, according to Bellcore. (A fuller discussion can be found in a recent OECD paper, "lnternet Traffic Exchange: Developments and Policy.") While the statistic sounds suspiciously high, and may not be representative of the global Internet, most observers agree that DNS traffic is a significant proportion of traffic. Hence, accord- ing to some engineers, the current U.S.-dominated root server deployment is not only a symptom of the network’s centricity, but also partially a cause. Will it change? Probably. The coordination of root servers, as well as domain names, IP addresses, and protocol development will soon be overseen by a new nonprofit organization with a board of directors reflecting U.S. and non-U.S, interests. It should get to work earnestly in 1999, and will represent an important power shift on Intemet governance (see Figure 11 ). Back to economics again. There are price and performance incentives to hub traffic via the U.S. The lack of telecom com- petition in Europe and Asia has historically meant that users in those regions paid more for telecom circuits than in the U.S. We’ve heard this before. But the magnitude of the problem is startling: A 300 kilometer circuit at two megabits per second (Mbps) in Europe costs up to four times more than the equiv- alent capacity in the U.S. Try and go across a border with the same speed line and the price differential is 17 times higher. This pdce gap is also growing. In June 1998, the Brussels- based European Internet Service Providers Association (EuroISPA) held a conference for European Commission (EC) officials to underscore the high costs for European long distance 115 TeleGeography 1999 © TeleGeography, Inc. 1998 Figure 4. Asianlnternet Backbone Conne~ivi~ CHINA Key 155 45 Total backbone capacity in Mbps To SA AUSTRALIA Japan Malaysia Korea Trans-Pacific Backbone Capacity (Mbps) 830 Australia 90 Teiwan 90 180 Hong Kong 90 China 45 135 Singapore 90 TOTAL 1,550 Note: Table and map reflect Sept. 98 survey of Asian ISPs with backbone connections above 10 Mbps, Source: Barry Raveendran Greene, Singapore; intra-Asian links from company data; OECD. Map design by TeleGeogrephy, Inc~ © i998 circuits and thus the pressure to transit Internet traffic via the U.S. A circuit from suburbs outside of Washington, D.C., where many U.S. lnternet backbones meet, to Pads or London or Stockholm, costs approximately $30,000 a month per megabit of capacity, said EurolSPA. By comparison, a direct line from any of these European cities to another cost roughly $35,000. By the next month, the presentation was out of date, as some London-based members of EurolSPA were quoted trans-Atlantic pdces of $6,500 per megabit. Part of the reason for this is the power which Europe’s incum- bent operators have over the supply of telecom infrastructure, 116 © TeleGeography, Inc. 1998 TeleGeo~aphy 1999 Figure 5. European Internet Backbone Connectivity Key 450 155 45 Total backbor a capacity m Mbps USA/) Helsinki Stockholm Dublin Londor ¯Prague ~rno Bratislava ¯Madrid Milan "Monaco To Cyprus To Israel Trans-Atlantic Backbone Capacity (Mbps) 9/98 12J98proj. 8/88 12/88proj. United Kingdom 1,063 1,514 Italy 34 68 Sweden 576 776 Russia (not shown) 34 45 Netherlands 250 600 Spain 16 45 France/Monaco 200 245 Poland (not shown) 18 34 Germany 140 215 Switzerland 12 24 Austria 38 49 Portugal 8 12 9/88 lZ/98 proj. Norway 4 8 Other* 100 150 TOTAL 2,457 3,740 *Other countries include: Finland, Belgium, Hunga~/, Croatia, Israel, S. Africa, Macedonia, Yugoslavia, Slovenia, Romania, and Cyprus. Note: Table and map based on Summer 1998 survey of European ISPs with backbone connections above 2 Mbps. Source: Europe data from Robert Cohen, Cohen Communications Group, tel: +1 212 9867720, bcohen@mail.bway.net. Israel figures from Sharon Chan, California State University. Map design by TeleGeography, Inc. © 1998 although this is slowly decreasing as alternative operators gain momentum. As well, new trans-Atlantic capacity has come on- line more quickly than cross-border European capacity. And more capacity usually means better performance. Just as the celebrated book The Death of Distance, by Frances Cairncross, made the point for telecom prices, Internet perfor- mance is frequently distance-insensitive too. In Asia, it is not uncommon for Internet traffic to be quicker if it’s routed via California rather than over a direct link. In Singapore, for example, the round trip time to Phnom Perth is 1,100 millisec- onds but to Los Angeles a scant 400 milliseconds (see Figure 13). Where bandwidth is available and the line speed is con- stant, such as in the United States, distance again becomes a factor (although minor). For example, statistics from C&W’s 117 TeleGeography 1999 © TeleCeogroph~l, Inc. 1998 Figure 6. African Internet Backbone Connectivity To U.K. To France To Belgium To Italy TURKEY SAUDI ARABIA ETHIOPIA Key ~128 Kbps 2 Mbps International Internet Backbone Connectivity I Former French colony F ormer British colony Former Italian colony Former Portuguese colony Former Belgian colony I Former Spanish colony Note: Map reflects summer 1998 data. South Africa’s connection to the U.S., which totals around 37 Mbps, is not drawn to scale. Source: Mike Jensen (http://demiurge.wn.apc.org/africa/index.html) real-time Traffic Report show cross-country delay rates of around 70 milliseconds, roughly three times longer than cases where traffic travels within the same U.S. region (see Figure 14). There is a self-sustaining, some say insidious, industry dynam- ic at play here. Historically, the global Internet was built atop Map design by TeleGeography, Inc. © 1998 excess bandwidth available on existing international cables owned by major telecom carriers. Most carriers planned to fill these cables over ten years or so given that voice traffic growth averaged but 1 3 percent to 1 5 percent annually. Hence, in the short run, leasing capacity for new Internet services seemed like found money. Yet, because U.S. carders picked up or landed 118 © TeleGeography, Inc. 1998 TeleGeography 1999 over 25 percent of the world’s traffic, they tended to build new cable capacity faster than other carders. They therefore had more available capacity to sell to or from the U.S. and at bet- ter rates. This, of course, ted more and more off-shore ISPs to acquire capacity to the U.S. for lnternet services. This in turn led to yet more U.S. carder cables and still better prices. And that, in somewhat simplified fashion, is how the star topol- ogy of the Internet was born and why it still glows so brightly. III. Who Pays For It? "The answer is either really long or really short depending on what you’re trying to say," says Scott Bradner, a leading Internet expert at Harvard University. The Internet does not have a set economic model, so there’s no standard way net- work providers are remunerated for the resources they expend. End of story. The longer answer is more complicated, precisely because the Internet’s provisioning model is not static. In contrast to the public switched telephone network (PSTN), a typical lnternet transmission may involve five or ten different networks rather than two or three. And the connectionless transmission tech- nology on which the lnternet is based also means that the role of each of these networks cannot easily be predicted in advance. Smaller networks increasingly must pay larger net- works by the speed of the leased line for connectivity. The larg- er networks exchange traffic without charge under a peering, sender-keep-all basis (see Figure 15). They seek to recoup their network costs primarily from users and their downstream ISP customers--not from other networks, as in the telephony world. To understand why the Internet’s schemes for funding international networks are so different from the traffic-based settlement arrangements over the PSTN, a brief digression on technology is useful. The (:onnectionless Network Traditional phone networks, built for voice communications, switch or assign a dedicated end-to-end circuit for every call. That is bandwidth intensive and also reliable: every user gets their own circuit. Minute-by-minute and circuit-by-circuit pay- ment methods consequently developed to compensate network providers. The lnternet involves a radical departure: It is based on pack- et-switching. There is no dedicated connection. Any communi- cation is converted to a digital format, broken up into chunks of data called packets, given an address, and sent out into the network. What’s most significant is that the path the packets take is never specified--the network itself determines the route and speed. The packets only know their destination address and how to be reassembled at the end of their trip. To some, it sounded like a crazy idea when it was first proposed in the 1960s by Donald Davies and Paul Baran and later refined in the early 1970s by pioneers like Robert Kahn and Figure 7, Backbone Connectivity for Latin America and the Middle East Latin America Country Connection Via Bandwidth Argentina U.S. 70 Mbps Belize U.S. 125 Kbps Brazil U.S. 106 Mbps Argentina 128 Kbps Uruguay 128 Kbps Caribbean U.S. 1.5-2 Mbps Colombia U.S. 12 Mbps Costa Rica U.S. 4 Mbps Chile U.S. 22 Mbps Dominican Republic U.S. 45 Mbps Ecuador U.S. 2 Mbps Guatemala U.S. 4 Mbps Guadeloupe U.S. 1.5 Mbps Honduras U.S. 64 Kbps Nicaragua U.S. 64 Kbps Panama U.S. 20 Mbps Paraguay U.S. 10 Mbps Peru U.S. 12 Mbps Puerto Rico U.S. 125 Mbps Venezuela U.S. 20 Mbps Middle East Country Connection Via Bandwidth Egypt U.S. 3.5 Mbps France 1.5 Mbps Iran (Tehran) Austria 2 Mbps Israel U.S. 26 Mbps U.K. 2 Mbps Jordan n.a. 512 Kbps Kuwait Philippines 8 Mbps Lebanon U.S. 500 Kbps Italy 500 Kbps Libya n.a, 2 Mbps Morocco U.S. 500 Kbps Italy 2 Mbps France 500 Kbps Turkey U.S. 70 Mbps United Arab Emirates U.S. 45 Mbps Note: Data generally currentto September 1998. Source: TeleGeography research, DTT Consulting, Spotbeam Communications, The Mosaic Group, and Sharon Chan. © TeleGeography, Inc. 1998 Vint Cerf, but it worked. (For an engaging history of the peri- od, see Peter Salus’ book, Casting The Net.) Since the early principles of Transmission Control Protocol/lnternet Protocol (TCP/IP) were advanced by Kahn and Cerf, packet delivery has not changed much. Routers along the way still store and for- ward packets. After forwarding, if the first router doesn’t receive acknowledgment that the packet has arrived safely at the next hop, it resends the packet. The protocol self-adjusts to achieve the best possible service; touters send packets as fast as they can with the least error rate. And in socialistic fash- ion, all packets are treated equally, on a best-effort basis. It wasn’t--and still isn’t--very reliable. If there’s a tot of con- gestion on a single route, packets may be continued on page 122 119 TeleGeography 1999 © TeleGeography, Inc. 1998 Figure 8. Internet Traceroute Maps, July 1998 From California to Washington, DC, direct Hop 5. From Silicon Valley to the Washington area in a single step~,, ¯ To rs7 Ioc gov (140 147 248 7} 1-30 hops, 38 byte packets 1 gate-96maln sjcabove net(207 12696 189)0917/272/530(1 10)rns 2 palx-rnaln-ac3 paa above net (207126 96 122) 1 97/4 64/9 2 (2 65) ms 3 pa~x dra net (198 32 176 40) 2 24/4 16/8 99 (1 85) ms 4 SantaClaral-s6-0 dra net (150147 74 18)3 34/518/7 54 (1 13) ms 5 MaeEaata~5.dra.net (150 147 25 17) 67 7/77 7/93 0 (9 21) ms 6 bbOfCongress-T3 dra net (150147 46 34) 73 4/83 3/95 6 (8 83) ms From Israel to Lebanon via New York and Virginia Hop 4. Across the Atlantic to White Plains, NY. I~ Hop3, From Tel Avivto United Kingdom Hop 13. To Reston, Virginia, then on to Lebanon (hop 14, not shown) To zezna aub edu Ib (193 188 128 14), 30 hops max, 40 byte packets 1 server-czscotlvlbm natal(192 11572 161} 2ms 2ms 1 ms 2 fe7507 tlvlbm net~l(192 116 177 1) 3 ms 3 ms 3 ms 3 portlbrl.pLuk.ibm.net (152 158 16 1) 85 ms 76 ms 8] ms 4 165.87.220,34 (165 87 220 34) 166 ms 172 ms 164 ms 5 1658728113(1658728113) 162ms 166ms 172ms 6 1658723061 (1658723081) 198ms 198ms 207ms 7 165872936(165872936} 2OSrns 181 ms 189ms 8 1658797225(1658797225) 166ms 183ms 195 rns 9 sl-bb11-rly-0-1 spnnthnk net 1144 232 0 33} 180 ms 199 ms 202 ms 10 sl-bb10-pen-7-0 spnnthnk net (144 232 8 154) 182 ms 185 ms 190 ms 11 sl-bbl-pen-8-0-O sprmthnk net (144 232 5 10} 206 ms 190 ms 230 ms 12 gzp-penn-4-fdd~0-0 g~p net (204 59 136 200) 188 ms 190 ms 197 ms 13 204.59.194.18 (204 59 194 18) 543 ms ¯ 432 ms Source: Keynote Systems, Inc Map des=gn by TeleGeography, Inc. © 1998 Traceroute software, originally written by Van Jacobson of Lawrence Berkeley National Laboratory, charts the path of a data packet between two points on the Internet. Launch the packet into the Net, and at every hop--a muter, gateway, or an exchange point--the soft- ware records its location and the time it took to get there. In the first example shown here, the traffic takes the most direct path. After its trans-continental journey from the West Coast of the U.S., the traffic is passed onto another network in Virginia, at the MAE- East exchange point. However, in the second example, when a user in Israel tales to send e-mail to Lebanon, the traffic first passes though the U.K., then goes to New York, down to Virginia, and finally back across the Atlantic to Lebanon. This is most likely because IBM and Sprint, the respective carriers, agreed to intenzonnect at a U.S. exchange point, but not in the Middle East, where no international interexchange points exist. A similar situation occurs in Asia. In the third example, traffic between Hang Kong and Japan transits California. Hang Kong Telecom sends the traffic to Palo Alto, California where the carrier continued on page 121 120 © TeleGeography, Inc. 1998 TeleGeography 1999 Figure 8. Internet Traceroute Maps, July 1998 (continued) From Hong Kong to Japan via Silicon Valley, California Hap 6. From Hang Kong to Pale Alto, To lpsun231 oracte co jp (202 211 136 41), 30 hops max, 40 byte packets 1 web-hk-rtr (203 223 0 1) 1 267 ms 0 814 ms 0 725 ms 2 s4-7a trnh08 hkl net (202 84 224 153) 102 671 ms 285 039 ms 475 136 ms 3 a3-0 yck06 hkt net (205 252 130 82) 297 266 rns 337 731 ms 265 526 ms 4 f5-0 hk-T3 hkt net (205 252 130 207) 297 282 ms 313 282 ms 305 571 ms 5 hssle-0-0 pa~x-T3 hkt net (202 84 128 254) 479 666 ms 234 939 ms 213 439 ms 6 91&NssiS-O,6W1,PAO1,ALTER.NET(157 130 193 133) 188.524 ms 188 033 ms 7 119 ATM3-0 XR1 SCL1 ALTER NET (146 188 144 78) 188 976 ms 306 485 ms 189 928 ms 8 195 ATMg-0-0 6W3 SF01 ALTER NET (146 188 145 237)190 519 ms191 704 ms 190 722 ms 9 att-afo-gwcustomerAtTER NET (157 130 193 62) 191 716 ms 312 471 ms 374 858 ms 10 205 174 74 250 (205 174 74 250) 395 630 ms 294 069 ms 322 915 ms 11 hsd-gate0-fddi3-2.Tokyo.lnterSpin.NET (165 76 0 7)547 064 ms 304 172 ms 392 646 ms 12 202 211 136 98 (202 211 136 98) 533 020 ms 624 000 ms 589 522 rns 13 Ipsun231 oracle ca ]p (202 211 136 41) 681 701 ms 714 300 ms 764 475 ms From London to the Vatican via New York Hop 5. From London New York in lOOms. Hop 10. From Newark to a different London network. Hop 13. From Londor/ to Milan, romaining within Alter. Net Te wwwvat~can va (194 91 153 247), 30 hops max, 40 byte packets 1 ensign-1 front~er-na’avorks co uk (195 200 1 1) 3 ms 3 rns 3 ms 2 central-1 frontier-networks co uk (195 200 12 3) 6 ms 4 ms 7 ms 3 bgp-11 frontier-networks co uk (195 200 12 13} 7 ms 17 ms 7 ms 4 arm-1 frent~er-netwnrks co uk (195 200 12 37) 7 ms 7 ms 8 ms 5 atmO-O-l.nl/¢lgwl.us.ins~et.flet(207 181 1 65) 121 ms 112 ms 96 ms 6 Hss~2-0-0 GW2 NYC4 ALTER NET (157 130 5 229} 117 ms 101 ms 99 ms 7 132 ATM2-0 XRI NYC4 ALTER NET (146 188 178 134} 98 ms 101 ms 107 ms 8 189 ATM3-0-0 XR1 EWR1 ALTER,NET (146 188 178 54) 118 ms 104 ms 123 ms 9 193 ATMll-0-0 CR2 EWR1 ALTER NET (146 188 176 25} 102 ms 103 ms 98 ms 10 105,ATM1.0-0.BR2.LND1.AIter.Net (137 39 30 77) 200 ms 246 ms 215 ms 11 432 ATM6-0-0 CR1 LND1 Alter Net (146 188 5 25) 199 ms 198 ms 218 ms 12 267 ATM6-0-O CR1 LNB2 Alter Net (146 188 4 246) 197 ms 201 ms 262 ms 13 215.Hssi6,-0.CR1.MLN2.~lter.Net (146 188 3 34) 278 ms 234 ms 234 ms 14 Fdd=0-0 GW1 MLN2 Alter Net (146 188 31 35) 237 ms 239 ms 235 ms 15 ITnet-gwcustomerALTER NET (146 188 32 50) 235 ms 242 ms 241 ms 16 Ethernet0 r5 m~01 ITnet (151 1 64 250) 239 ms 273 ms 248 ms 17 fi01-rmlano ITnet (151 1 254 61) 290 ms 248 ms 273 ms 18 roma-fiOl ITnet (151 1 254 41) 258 ms 312 ms 257 ms 19 rome3 ]Tnet (151 1 3 252) 265 ms 288 ms 20 vat~can-gw IT net (151 1 203 26) 318 ms 271 ms 326 ms 21 dnsl vat,can va (194 91 153 5) 265 ms 292 ms 298 ms 22 tins1 vat,can va (194 91 153 5) 265 ms ~H 264 ms 114 278 ms IH Source: Keynote Systems, Inc. Map design byTeleGeography, Inc © 1998 interconnects with UUNet. As the upstream ISP, UUNet exchanges the traffic with the recipient’s upstream ISP, and the traffic is then sent back to Asia. In the last example, a user on one network in London connecting to the ISP hosting the Vatican Web site in Rome has a diabolical ordeal. Since the two downstream networks do not interconnect with one another, the traffic is handled by the upstream ISPs, The first net- work hauls the traffic to New York, where it is picked up by UUNet and, ironically, carried back to London! From there, UUNet routes it to Milan, where the Vatican’s ISP faithfully accepts the traffic. To be sure, not all traffic between these locations follow the same routes: It depends on the peering relationships of different ISPs. Also, on a single network, the path itself dynamically changes to account for network conditions, such as congestion. Nevertheless, the traceroutes shown here, all recorded in July 1998, are illustrative of the general U.S.-centric pattern of the Internet’s traffic flow. To learn more about traceroutes, visit http://www.cyber- geography.com/atlas/routes.html. Or, to run traceroutes from many different locations, see Boardwatch magazine’s compilation of trace- route servers at http://boardwatch.internet.com/traceroute.htmt. 121 TeleGeography 1999 © TeleGeographyo Inc. 1998 dropped; not such a good thing for time-sensitive traffic. But the original designers cleverly built in robustness. To be certain traffic took the most efficient path, reuters would dynamically update information about congestion on the network, querying devices at every subsequent hop. So if a backhoe dug up a cable--or an atom bomb wiped out a city--the router would detect a problem and choose a different path. With network information distributed, the Internet could route around any central point of failure. Once the basic architecture and design principles were in place in 1985, the U.S. National Science Foundation (NSF) began funding data networking pioneers at 13 supercomputing cen- ters across the U.S. A nationwide circuit for the traffic was com- missioned. The academic institutions had to strike deals with local telecom providers to lease local and regional circuits. More institutions sought to be connected to the NSFNET back- bone, which basically was synonymous with the Internet. Things moved quickly. In 1992, after the Internet had proved its commercial viability, the U.S. government wanted out, and the backbone transmission network of the NSFNET was priva- tized. It also began to accept commercial traffic. As well, in 1994, the NSF commissioned four network access points (NAPs), essentially traffic exchange points for ISPs; they were located in northern New Jersey, outside Washington, D.C., Chicago, and San Francisco, and were all run by different tele- corn operators. Thus, in a remarkably short period of time, all the basic ingredients of today’s global lnternet emerged. Of course, an application from outside the traditional lnternet community would dramatically shake things up. The World Wide Web, developed by Tim Berners-Lee and popularized around 1993, was soon followed by the Mosaic browser, fore- runner to Netscape. The exponential growth the lnternet had seen until then--users and host counts generally doubled annu- ally-hit massive proportions and backbone traffic surged. Curiously, what remained constant was different components of the Net’s architecture, albeit on a far more monumental scale. What You Pay Depends On What You Do Back to the economics. As we have seen, although traffic is routed over the Internet on a virtual pathway, without fixed routes or network connections, the physical networks which make up the lnternet--typically leased circuits from telephone companies--do interconnect. And networks do exchange traf- fic. Indeed, the economics of traffic exchange are key to under- standing who pays for what on the lnternet. That said, however, the lntemet industry has matured to such a point that one can now make important distinctions among different categories of network service providers or ISPs. Doing so provides, in part, the answer to how global infrastructure providers, such as telcos, are and will be compensated. The generic term "lntemet service provider" has become mean- ingless. It does not distinguish, for instance, between backbone ISPs that have global infrastructure (such as MCI WorldCom’s UUNet and Sprint), or local ISPs that lease infrastructure in a specific geographic region and require global connectivity, such as EasyNet in Europe (www.easynet.co.uk) and Tokyo lnternet in Japan (www.TokyoNet.ad.jp), or Concentric Network Corp. (www.concentric.net) and EarthLink (www.earthlink.net) in the U.S. Nor does it take into account whether the service provider’s customers are individual users, who tend to request content, or content providers who pay to export data. Better, then, to break down the industry into four classes: (I) backbone ISPs; (2) downstream ISPs; (3) online service Figure 9. The World’s Top ISPs Carrier/ISP MCI WorldCom (UUNet, ANS, CNS) Sprint Market Share of U.S. backbone ~53% 21.19% 5,32% 28.43% U.S. Backbone Speed 622 Mbps 155 Mbps 155 Mbps 155 Mbps Number of POPs 0ve r 1,000 320 375 483 Total Quarterly Revenue $4.96 billion (2Q98) $3.97 billion (2Q98) $8.28 billion (2Q98) $3.12 billion (1Q98) Comment GTE (BBN) Cable & Wireless AG IS 3.57°5 155 Mbps Over 200 n,a, PSIN et 1.84% 622 Mbps Over 400 $44.5 million (1098) Qwest n.a. 622 Mbps Over 400 $393.7 million I (EUnet) (2Q98) I Note: Market share measured by percentage of downstream ISPs served by backbones, Backbone speeds represent fastest nationwide link per single fiber route. Cable & Wireless data includes Intemet infrastructure bought from MCI in September 1998, Source: Boardwatch Magazine’s Directory of lnternet Service Providers, Winter 1998 Spring 1999. © TeleGeography, Inc. 1998 Building national networks in Europe; boosting Asian presence. Global One alliance building pan- European backbone. No owned international infrastructure. Building European network; leases capacity from MCI WorldCom for U.S. backbone. No owned international infrastructure, Bought major ISPs throughout Europe and Asia 1997-1998. EUnet in over 42 countries. © TeleGeography, Inc. 1998 Te~eGeography 1999 providers such as AOL (www.aol.com) and Japan’s NiftyServe (www.niftyserve.or.jp); and (4) firms that specialize in web site hosting, such as Exodus (www.exodus.net). The cost structure and the money flow is determined by the category to which one belongs. Taken together, the worldwide market for Internet access is now big business: it was at $25 billion in 1997 and is predicted to top $100 billion by 2000, according to Zona Reseamh Inc. (www.zonareseamh.com). Web Hosting Companies Web server "farms" emerged from the ISP industry itself but are now somewhat separate since companies have made web hosting into a niche business and are growing rapidly. The important fact is that their traffic flow is mostly uni-directionat. The few bits of data that trickle in when a user requests a web page are overwhelmed by the flood of outgoing audio, video, images, and text. As a result, backbone ISPs demand that Web hosting companies, which typically do not maintain a national Figure The Intemet’s address books--the electronic directories and number- ing systems--are based in the U.S. and supported by government contracts. Amid growing dissent, since 1996 the Intemet community has debat- ed ways to change this, chiefly by creating an international, bottom- up, private sector body. It would take over the functions the Internet Assigned Numbers Authority (IANA) at the University of Southern California (USC, which was headed by Jon Postel until his death in October 1998. 10. The U.S. Controls the Internet’s Directory System In addition to setting policy decisions on the number of top leve! domains, {e.g., .corn, .orB, .edu} the new entity will also allocate the lntemet Protocol (IP} numbers needed for routing. These are a scarce and valuable resource. No plan has yet achieved a consensus. But, the U.S. government and the European Commission, among others, have said they support this private sector initiative. For the U.S., handing off the real estate IANA controls to a private, international entity might be compared to the land grants of the 1800s, only the land value--the cyberspace of tomorrow--is proba- bly infinitely greater As well, the new entity will assume IAN/~s oversight of the wodd’s 13 root name servers. These computers hold the authoritative routing data for the domain name system, and in many cases the data for popular names (.corn, .edu, etc.) too. Without round-the-clock access to these databases, much of the traffic on the lntemet would never find its way. The transfer of IAN~s power has become hotly political. Many nations have come to recognize that IANA represents a central con- trot point in the digital economy, and so seek positions of influence in the new entity. Likewise, large net-savvy companies understand that access to IP numbers affects the magnitude and type of services they can offer downstream. See Figure 11 for further details on the Intemet’s new governors. ID Organization A InterNIC (Network Solutions Inc.) B Information Sciences Institute (USC) C PSINet, Inc. D University of Maryland Computer Science Center E NASA Ames Research Center F Internet Software Consortium G Department of Defense Network Information Center H Army Resource Laboratory I Royal Institute of Technology (NORDUnet) J Information Sciences Institute (USC) K European Network Coordination Center (RIPE NCC) L Information Sciences Institute (USC) M University of Keio (WIDE project) Root Server Administrators Status Private company Academic Private company Academic Government Private company Military Military Academic Academic Non-profit consortium Academic Academic Location Herndon, Virginia Marina del Roy, California Troy, New York College Park, Maryland Moffett F~eld, California Palo Alto, Cahforn~a Vienna, Virginia Aberdeen, Maryland Stockholm, Sweden Marina del Rey, California London, UK Marina del Rey, California Keio, dapan Note: The wodd’s 13 root servers, known by the letters A to M, coordinate the domain name system data that links alphanumeric names to the IP number addresses used for routing. Servers maintained by the U.S. military move among undisclosed locations. Source: http://nic.ddn.mil/DNSIroot-server.h~nl, Internet Assigned Numbers Authority, and root server administrators. © TeleGeography, tnc. 1998 123 TeleGeography 1999 © TeleGeography, Inc. 1998 network, pumhase connectivity from a backbone or down- stream ISP whose customers seek the content. But there’s a problem: The web hosting firm claims that back- bone ISPs are already compensated by their end customers, and thus to seek a payment from the content provider would mean a double payment. The backbone ISP counters that it is forced to haul the content provider’s traffic on its own network to reach its customers--and it wouldn’t need so much infra- structure if the server farm had its own national network. Backbone ISPs thus only agree to accept a server farm’s traffic at a price. In August 1998, a peering dispute erupted between GTE Intemetworking and Exodus over this very issue, and both firms’ customers came close to losing direct connection to one another Downstream ISPs A similar logic is used for downstream ISPs and online service providers, and the price of Intemet connectivity varies by loca- tion and amount of data. For example, a downstream ISP in Cambridge, Massachusetts can lease a 45 Mbps circuit for $2,500 per month. But that only pays for the facilities required to meet the gateway of an upstream backbone ISP. The price to connect with the backbone, which lets the downstream ISP’s Figure 11. The Internet’s New Governors Internet Governance Hierarchy Domain Names i~ ~-G ° v e r n~noen~ tr~ilt eAed vi s ° rY~ I P Addresses Protocols A new structure to replace the lntemet’s unelected governors is slow- ly emerging. The Internet Corporation for Assigned Names and Numbers (ICANN)--if the name sticks--will serve as the umbrella organization for three supporting organizations overseeing domain names, IP addresses and protocols. Al! three groups are expected to retain considerable autonomy. The Domain Name Supporting Organization is made up of registrars for both generic top level domains (TLDs) like .corn, and country- code TLDs like .jp for Japan. The Address Supporting Organization groups the three regional IP address registries, and will include oth- ers as they form. The Protocol Supporting Organization will be rep- resented by the Intemet Engineering Task Force (IETF), probably via the Intemet Architecture Board [IAB). Two so-called Advisory Committees exist. One provides a means for national governments to be represented in ICANN’s operations, while the other is intended to treat issues of domain name system root server deployment (see Figure I0, "The U.S. Controls the Internet’s Directory System"). Who’s Who in Internet Governance: ICANNmlntemet Corporation for Assigned Names and Numbers. Non-profit corporation that acts as the Internet’s central coordinat- ing body, formedy known as the Intemet Assigned Numbers Authority (www.iana.org). Intedm funding may be provided by the IBM-led consortium of major suppliers known as the Global Intemet Project (www.gip.org). wwTLD--World Wide Alliance for Top Level Domains. The umbrella organization to unify top level domain name registrars, including nationat country codes, current registrars of current and possibly future generic TLDs (www.wwtld.org]. ARIN American Registry for lntemet Numbers. Allocates IP addresses for the Americas and the Caribbean [www.arin.net). RIPE NCC--Reseaux lnternet Protocol Europ~ens Network Coordination Center. Allocates IP addresses in Europe, Afdca, and the Middle East (www.ripe.net). APNIC--Asia-Pacific Network Information Center. Allocates IP addresses in the Asia-Pacific region (www.apnic.net). IETF--lnternet Engineering Task Force. The tntemet’s technical stan- dards setting body (www.ietl:.org). IAB~lntemet Architecture Board. Oversight body of the IETF (www.iab.org}. © TeleGeography, Inc. t998 124 © TeleGeography, Inc. 1998 TeleGeography 1999 customers reach other destinations on the Internet--can be as high as $30,000 per month. While the connection fee may seem a crippling cost for U.S.- based ISPs, service providers outside the U.S. must also pay for the cost of an international private line if they wish to connect directly with the lnternet at its core. Such a connection does not come cheaply--trans-Pacific circuits, for example, may cost between $700,000 and $1 million a year for a 45 Mbps line. However, most downstream ISPs and large corporate users that purchase lnternet connectivity do not pay based on their actu- al usage, bit by bit, but based on a usage profile, broken down into different tiers. It would be too expensive and the tools are too embryonic today to meter every data flow and charge for it. According to Internet engineers, the cost of measuring exact usage could put a debilitating premium on Internet service. Such a dilemma is beginning to occur in the U.S. long distance telephone business. With coast-to-coast U.S. rates of $0.10 a minute or less, up to 40 percent of the rate for long distance telephony may reflect the costs of monitoring and monthly billing. So on the Internet, the backbone ISP’s network measures the overall traffic pattern by glancing at the router, octets in and octets out (octets being the 8-bit unit of traffic measurement for packet networks)--and charging the downstream ISP accordingly. This allows a customer to lease a line with much more capacity than is ever used, pay a sum closer to the actu- al usage, and be assured that should traffic spike, the line can meet the demand for an additional fee. The only drawback with Figure 12. Power Shifts in Internet Governance: 1982-1998 1982 The U.S, Department of Defense Advanced Reseamh Projects A~ency (DARPA) begins funding Jon Postel at the Information Sciences Institute at the University of California to coordinate the allocation of Intemet Protocol addresses and assign networking para- meters. The term IANA, lntemet Assigned Numbers Authority, emerges informally in 1988 to descdbe the task. t 98z~ The Domain Name System (DNS) is created to translate IP numbers into user-friendly names, ending in .rail, .gov, .edu, .org and ,com. 1990 New Top Level Domains (TLDs) are created for countries, based on the two-letter country abbreviations of the International Organization for Standards (www.isoJnt). 1992 The U.S. National Science Foundation (www.nsf.gov) begins a cooperative agreement with Network Solutions Inc. (NSI, www.net- sol.corn) to act as a register of domain names and serve as the InterNIC, the Internet Network Information Center (www.intemic.net), which allocates IP numbers. 1996 Jon Postel, head of IANA, proposes a policy to institute new TLDs on a first-come, first-served basis, in response to lntemet entre- preneurs who begin selling self-minted TLDs like ".web". These new TLDs are only routable on a minuscule number of networks (i.e., www.altemic.com). 1997 The International Ad Hoc Committee (www.iahc.org), com- prised of representatives of ]ANA, the lntemet Architecture Board [www.iab.org}, the U.S. Federal Networking Council (www.fnc.gov}, the International Trademark Association {www.inta.org}, the World intellectual Property Organization (www.wipo.int}, and the International Telecommunication Union (www.itu.int}, releases a pro- posal for a new {~oveming system for the DNS and an initial seven new TLDs. In May, 80 companies from around the world sign a Memorandum of Understanding that is deposited at the ITU (www.gtld-mou.org), which creates a Council of Re~strars (CORE) and a Policy Oversight Committee (POC). 1997 The U.S. Department of Commerce, after a sedes of inter- agency task force meetings overseen by the White House, begins soliciting public comments on the issue of DNS reform. 1998 The Department of Commerce issues a "green paper" in January to solicit public comments on a proposed policy to manage the DNS and IP number allocation, as well as oversee root server administration and protocol development. A final "white paper" is issued in June, calling on industry to create the new entity (see htt p://www, ntia.doc.gov/ntiahome/domain name). June 1998 The International Forum on ’the White Paper (www.ifwp.org) forms under the guidance of Boston University law professor Tamar Frankle and is steered by around 30 non-profit orga- nizations specializing in technology. It hosts public meetings in Reston, Virginia, Geneva, Singapore, and Buenos Aires to achieve consensus on the structure of a new body to govern the tntemet’s numbering resources. September 1998 A month before his death,Jon Postel of IANA, on behalf of the Internet community, develops a proposal for a succes- sor body, called the lnternet Corporation for Assigned Names and Numbers (ICANN), along with nine candidates for an interim board of directors. The plan is immediately attacked by a few IFWP partic- ipants when submitted to the U.S. Department of Commerce. October 1998 The Department of Commerce and Network Solutions Inc. extend NSI’s cooperative agreement to register ,corn, .net and .org until 30 September 2000. NSI agrees to transition to a shared registration system, allowing other companies to register names under those TLDs, by June 1, 1999. NSI agrees to stdke a contractual agreement with IANA’s successor for the management of the central "/~’ root server. © TeleGeography, Inc. 1998 125 TeleGeography 1999 © TeleGeography, Inc. 1998 1,800 1,600 Packet Round-Trips from Singapore Telecom Internet Exchange (STIX) 1,400 6OO 1,2oo ,Oco,om O Phnom l~enh, C 1,000 gahru’, Ma Jak~’rta, Kathmandu, Nepal )Lahore, Pakistan Key Capacity of STIX Connection 15~360 4,096 2,048 512 128 L Kbps 64 Lanka odia Noumea, New Caledonia Beijing, China Sydney, Australia San Francisco, CA S. Korea Tokyo, Japa Vancouver, BC 2,000 4,000 6,000 Monaco 8,000 10,000 12,000 14,000 16,000 Distance from Singapore (kin) Note: Packet round-trip ames are averages taken from the Singapore Telecom Intemet Exchange (STIX) web site, http:flvwwv.stix.net, ~n late November to early December 1997 at different times of day. Source: Adapted from Internet Traffic Exchange, OECD 1998. © Tele6eography, Inc. 1998 this approach is that it sets up an incentive for the upstream ISP to overbook capacity, under the hopeful (and reasonable) assumption that all customers do not generate peak loads at once. At MindSpring Enterprises Inc. (www.mindspring.net), chief executive Charles Brewer says 30 pement of company costs dedve from connectivity fees and 13 pement from customer service expenses. He expects a complete reversal over the next five years, as bandwidth becomes a commodity and the ISP must differentiate itself more by the services it offers. Online Service Providers An online service provider, such as AOL, earns revenues not by reselling network transmission service but, by offering propri- etary content and specialized commerce, selling ads, and pro- viding users with the ease-of-use which comes from special soft- ware interfaces and customer hot-lines. They are typically the customers of the backbone ISPs, which also manage the net- work points of presence [PoPs) accessed by dial up retail users. AOL, for example, has outsourced nearly all of its network transmission needs to the MCI WoridCom subsidiary UUNet, though it maintains a multi-vendor strategy to ensure redun- dancy. Even AT&T, when it launched its WoddNet residential lnternet service in 1995, relied mainly on BBN’s network, since acquired by GTE. The online service provider is either paid a flat monthly rate by customers for unlimited service, or charges additional fees after a set usage is exceeded. The real payoff is in the eyeballs and the mouse-clicks--selling specialty content, advertising space and taking a share of e-commerce revenue. The cost of data networking is apparently the highest cost, around 50 percent of revenue. Significantly, however, marketing, customer support and subscriber acquisition represent close to 35 percent of rev- enue. Backbone lSPs All networks are beholden to the backbone ISPs--be they web hosting facilities, downstream ISPs, or online service providers--either to furnish Intemet connectivity or to manage the actual network infrastructure. 126 © TeleGeography, Inc. 1998 TeleGeography 1999 Figure 14. Distance and Delay on the Cable & Wireless Backbone Delay measures the time it takes a paok__~ to tr~x,~l 1Yore one node to another and baok again. The Cable & Wlrele_~s Intemet bac.l~,one aver-ages be~er than 95 milliseoonds ero~s-oountn.,," roundtrip. 8:23:48 PI!I Eastern $~t~ ~tioe are delayed 10 rrirutee, When the anomaly of bad network infrastructure (such as in Asia) does not exist, distance becomes a factor for traffic delay. This is demonstrated by Cable & Wireless’ U.S. Intemet back- Source: Cable &W~reiess, Inc,, htrp://traffic,rncLcorn bone. The roundtrip time of a packet roughly correlates with the distance it must travel. © TeleGeography, Inc. 1998 Internationally, the same dynamic applies. Local downstream ISPs in Asia, Europe and elsewhere need the larger, upstream networks, often the incumbent tetecom provider, for Internet connectivity. And big Internet sharks outside the U.S. find themselves but tiny sardines when they arrive on U.S. shores with their leased circuit dedicated to IP traffic. They must strike an interconnection agreement with one or more of the lnternet backbone networks, just like a regional U.S. ISP. Some off-shore backbone ISPs have begun to acquire their own national U.S. networks to obtain free peering. Most have not. Japan’s NTT hopes to aggregate its traffic with the large U.S.- based backbone ISP Verio, in which it took a ten pement stake, and Qwest was keen to piggyback onto EUnet International’s peering agreements when the U.S. telecom upstart bought the pan-European ISP in March 1998. But it is unclear whether other backbone ISPs will allow either move. No one is certain what is actually going on: backbone interconnection deals are shrouded under "non-disclosure agreements," and only limited bits of information are available (e.g., from a stock prospectus when a company, such as Verio, goes public). The emergence of backbone ISPs to provide connectivity to the Internet--indeed to determine what actually constitutes Internet connectivity--is a relatively recent phenomenon. Not surprisingly, there’s controversy surrounding their role in the lnternet food chain. When the lnternet first evolved, ISPs were generally the same size and swapped traffic freely. Early Net applications, like file transfer protocol, led to more or less symmetrical traffic among ISPs. In contrast, the Web creates a split between the end users, who import data, and content companies, who are data exporters. That’s new. In the early days, the ganglia of net- work interconnections were so complex--since everyone accepted any other network’s traffic--that the only way Inter-net engineers could map the Internet’s topology and traf- fic flow was simply to draw a cloud. Today, however, the enor- mous size differences among ISPs coupled with the terrific infra- structural investment, and major traffic imbalances due to the emergence of Web hosting firms, has meant the practice of set- tlement-free peering is waning. The two noticeable exceptions are that local ISPs peer with their siblings at local exchange TeleGeo{~raphy 1999 © TeleGeography, Inc. 1998 Figure 15. A Primer on Peering Peering is to the Intemet what interconnection is to the telecoms world--the way in which different networks exchange their cus- tomers’ traffic. Without interconnecting, a customer of one ISP can’t reach the customer of another. Over the past two years, a handful of the biggest lntemet service providers have refused to interconnect for free with other tSPs. Instead, they demand smat]er 1SPs pay a fee to compensate for the bandwidth resources larger ISPs must expend to carry the smaller providers’ traffic. Smaller ISPs claim this is discriminato- ry. And regulators reluctantly admit that unless the conflict is settled, they may intervene to assure a competitive market and protect the public interest. The Federal Communications Commission (FCC), for example, in the fall of 1998 solicited pub- lic comment on the issue (see www.fcc.gov/ccb/706). When the lntemet first developed, networks were generally the same size and the traffic flow roughly equal--hence all were "peers." It thus made economic sense to swap traffic freely. Even if there was a slight imbalance, it was tess expensive to set up a free deal than to meter and to charge because both parties ben- efitted by being able to reach more users. But as the Net became commercial, the size of networks began to differ enormously. As a result, large networks such as UUNet and Spdnt began to change the rules. Both small and large ISPs have a point. Large ISPs are not fair- ly compensated, even if the amount of traffic exchanged is equal, because the larger ISP is inherently forced to carry traffic a longer way over its network than the smaller ISR This is because most ISPs use "shortest exit routing," whereby traffic is handed off at the earliest possible network juncture. A larger network has more network interconnection points, and so is constantly in the unen- viable position of having to accept others’ traffic and to carry it farther. On the other hand, large ISPs do not publicly state their peering criteria, and insist that all arrangements be shrouded under non- disclosure agreements. The lack of transparency and objective criteria may put smaller ISPs at a disadvantage. Small ISPs or start-up companies today have little control of their costs, nor can they gauge the direction of market costs unless fair and non- discriminatory peedng criteria exist. Internationally, interconnection takes on a different hue. tSPs outside the U.S. are not considered peers by major U.S. ISPs. In addition, foreign ISPs must pay the full cost for transoceanic cir- cuits required to connect to U,S. backbones. In contrast, domes- tic U.S. ISPs which are treated as peers typically share the cost of the circuits used to join their networks, since they have essen- tially parellel ystems. Off-shore ISPs thus claim they effectively subsidize U.S. Intemet users. Due to "shortest exit routing," the foreign ISP carries the traffic to the U.S. and cardes it back to the in-house re~ion. In the U.S., the long-haul cost is shared by different ISPs, since one is able to carry traffic from one exchange point to another before handing[ it off to the second ISR Discrimination Against Foreign ISPs? The discrimination alleged by off-shoro ISPs has sparked inter- governmental controversy. The Asia Pacific Economic Council in May t 998 a~reed to study the matter, over the objections of AT&T. The International Telecommunications Union (ITU) Study Group 3 is also examining the issue. And an Organization for Economic Cooperating and Development (OECD) report in May t998 stated "the question of cost allocation is a valid one if the current arrangements are not equitable in terms of use made of infrastructure relative to financial contribution." At an OECD Intemet workshop in Osaka, Japan in June 1998, John Hibbard, Vice President of international carder business at Australian carder Telstra, presented the views of non-U.S.-based ISPs: "U.S. demand has increased such that in the case of Telstra, approximately 70 percent of the usage of the link is ddven by Australian users and 30 percent by U.S. users. Yet Tetstra, at the Australian end of the link, pays t 00 percent of the cost .... Our action plan is to promote global awareness, try to negotiate bet- ter solutions with U.S. operators and if that fails, resort to other tactics .... One avenue is to involve the World Trade Organization (WTO) because what is happening wilt stunt the Internet, and hence restrict the potential for electronic commerce." The U.S. private sector, meanwhile, cleverly seeks to capitalize on these disputes. Some companies, like InterNAP and Savvis, are trying to develop a model for metering and billing for intercon- nection traffic, to move the matter from barter to contract. That would bring a de~ree of certainty to the marketplace. And new Net developments, like quality of service assurances, may make much of the controversy moot. To honor service level guarantees between networks, ISPs will need common metrics to meter premium-priced traffic and to charge a settlement fee for handling it. Until then, however, reBulators are keeping a close eye on the Intemet’s commercial structure, especially after the WorldCom-MCl merger (see Figure 17). Moral for the industry--be fair, or be regulated. © TeleGeography, Inc. 1998 128 © TeleGeography, Inc. 1998 TeleGeography 1999 150 Internet User Growth, 1997-2002 135.9 100 50 ¯ Users 2002 Users 1997 22.1 38.8 82.0 16.8 38.7 4.5 4.9 3.8 Rest of World Japan Asia-Pacific Western Europe U.S. (ex. Japan) Source: International Data Corp. Design by TeleGeegraphy, Inc. © 1998 points, and the very biggest backbone ISPs continue to peer among themselves. The point--scale matters. Unless your network is very big and very fast and upgraded continuously, you are a customer of somebody else, and therefore will not be able to control your fixed costs. Which is one reason why regulators, who otherwise loathe to interfere with the lntemet’s dramatic growth, have stepped in to try and keep the backbone market competitive (see Figure 19). But many of the hard questions, such as the regulatory status of backbone networks--basic telecom facility or enhanced ("private") data pipe--remain unresolved. IV. How Fast Is the internet Growing and Why? Thus far we’ve seen that although it is global in reach, the lnternet has a U.S.-centric infrastructure, and that there are strong economic and technical reasons why this remains so. We’ve also seen that the business of providing Intemet facilities and services is becoming quite heterogeneous; there are dis- tinct network tiers with all but the largest ISPs--the backbone ISPs--being customers of the others. This structure affects the economics of the global Internet in terms of the flow of money between downstream and upstream network owners and the incentive to invest in new transmission and switching facilities. Simply put, one ISP’s expenses are another’s revenues. Now, we turn to the question of Internet demand, which is cen- tral to the lnternet’s future structure. Where there is growth, and where not, will determine tomorrow’s revenues and expenses for the ISP industry. Scaling The Net With overall demand for transmission capacity growing at 200 to 300 percent plus per year, the crux of nearly all Internet problems today is scaling--how to take a protocol and infra- structure designed to link a handful of computing centers together to exchange non-real-time data, and make it work internationally, and even carry voice traffic. Keeping this challenge in mind we took at the following four dri- vers of demand: (1) the growth of users and what they do online today; (2) tomorrow’s applications and devices; (3) the effect of local access technologies on international bandwidth consumption; and (4) new technologies that, against the pre- vailing wisdom, may not increase demand very much. Follow the Data We start with a reprise of some basic statistics. Since its incep- tion, the lnternet has doubled annually by many measures. As a rule of thumb, computers linked to the ]nternet or hosts have increased by 100 percent per year, as have domain names and users. For a more detailed review of the current numbers, see Figures 16 and 18. What about Internet backbones? In terms of traffic, the growth rate appears to vary from 200 to 1,000 percent a year, depending on the network. And some ISPs have provisioned for even greater growth. To get a better idea of the magnitude of bandwidth growth, consider the Internet’s evolution. In 1986, the NSFNET backbone ran at 56 Kbps--the rate of many user’s modems today (recall Figure 2). In 1988 it was upgraded to 1.54 Mbps or T-1. In 1991, it was upgraded again, to 45 Mbps or T-3. At this point, the commercial sec- 129 TeleGeography 1999 © TeleGeography, Inc. 1998 Figure 17. Lessons of the MCI-WorldCom Merger On September t4, 1998, the $37 billion MCI-WorldCom me~er cleared its final regulatory hurdle. The U.S. Federal Communications Commission (FCC) approved the deal subject to the pdor divestiture of M¢l’s lntemet business, as European and U.S. antitrust regulators had proposed (the FCC order can be found at www.fcc.gov). MCI promptly agreed and the new MCI WortdCom began business the next day. After 11 months of negotiations, regulators had extracted their pound of flesh--MCl Internet--but the terms of the divestiture and the competition inquiry which preceded raised more questions than were answered. The buyer of MCI’s Intemet assets, Cable & Wireless plc (C&W), paid SI.75 billion for MCI’s Intemet customers (corporate and retail), network ports, routers and peedng agreements--but not MCI’s backbone transmission circuits. C&W agreed to lease the capacity it needed to support its new Internet business from the seller. To some observers, that made the divestiture a Pyrrhic vic- tory for competition given the original regulatory concerns reegard- ing undue concentration in the market for Intemet backbone ser- vices. On the other hand, officials from the European Commission (EC), who were the most adamant about divestiture, privately said that the scope of the divestiture may still have been too broad. And, informally, at least one U.S. antitrust official agreed. Yet whether the outcome was correct, what regulators learned dur- ing the review--the first time competition and telecom watchdogs seriously examined the Net--wilt have future implications. Regulators focused on three broad issues raised by the combina- tion of the MCI and WorldCom networks: the supply of raw trans~ Atlantic circuit capacity; the combined companies’ Intereet market share as measured by the percentage of backbone traffic; and the structure of peering arrangements. Most worrisome to regulators was the amount of traffic the com- panies carried. An essential distinction, however, is that "market share" isn’t really about backbone traffic--that’s just a proxy for the real concern--the number of routes an ISP services. These downstream routes represent a network’s customers where traffic either ofisinates or terminates. The more routes you have, the better the chances are that you can terminate your customer’s traffic without having to hand it off to a dval ]SR That provides better control over the quality of service and costs. But if you get ve~ large, it means that a greater num- ber of other networks are more beholden to you to carry their cus- tomers’ traffic, than you are to them. Regulators understood the mer~er could upset the existing balance of power among backbone networks. As recently as 1996, it was known as the Group of Five--UUNet, BBN, Sprint, MCI, and ANS. The deal would have added MCI to WoddCom’s ownership of UUNet and ANS, reducing the cartel-like club to three. So EC reg- ulators laid down conditions for the merger--and by doing so, advantaged a home-town favorite, U.K.-based C&W. There are three immediate legacies of the investigation. The first is that it has set the s~age for, but not yet settled, ground rules for peedng arrangements. The FCC ducked the issue as apparently did the EC and the U.S. Department of Justice [DOJ) (i.e., the post- merger company can still peer on its own terms). But any other la~e Intemet merger can expect the issue to be front and center. Second, the investigation was distinctive because it used antitrust law, rather than telecom policy principles, to address competitive concerns in communications. Third, there was significant cooperation between the EC’s merger task force and the DOJ throughout the two agencies’ separate investigations. When the deal was announced in October 1997, few observers thousht that the companies would be subject to much scrutiny in Europe, where WorldCom had relatively few assets and MCI even less. Nevertheless, subsequent events make it clear that even minimal assets in foreign jurisdictions, especially the EU, may subject a deal to local approval requirements. In this case, the broad reach of the EC’s competition guidelines eventually allowed European regulators to take the lead, not the DOJ. Indeed, that may have been intentional, allowing the DOJ to avoid a politically sensitive case. No U.S. agency wants to risk political censure as the first bureaucrats that tded to regulate Internet capacity or prices. Finally, the MCl-WortdCom deal may be but the opening salvo in the lar~er conflict over the regulatory status of lnternet backbones. If the backbones are treated as telecommunications infrastructure, that is, as common carrier facilities, then companies supplying "lnternet carriage" probably will be required to provide fair and non-discriminatory terms to ISPs seeking interconnection. If not, they may be able to pick and to choose with whom they connect and on what terms, as do large data networks today, because they are considered "enhanced" service providers. Enhanced services are still subject to antitrust laws. But, as the MCl-WorldCom docket showed, when it comes to the messy details of peering and network access, antitrust authorities seem as willing as telecom officials to let the industry sort them out for itself. © TeleGeography, Inc, 1998 © TeleGeogruphy, Inc. 1998 TeleGeo{~raphy 1999 Figure 18. Internet Growth in Western Europe "~., Internet User Growth, 1998-2002 18 17 18 14 Users 2002 Users 1998 4.8 9.5 10 11.4 Finland Sweden Spain Italy France United (3ermany Kingdom Source: IDATE, The WorldAt]as of the lnternet, http~//www.idate.fr Design by TeleGeography, inc. © I998 tor took over as the key bandwidth vendor. BBN, the compa- ny that created and first operated the forerunner to the lnternet, the ARPANI~F, had an Internet backbone running at T- 1 speeds of between one and three Mbps in 1993. In 1996 its backbone ran at between 150 and 200 Mbps or OC-3. In early 1998, the company deployed OC-12 Dines at 622 Mbps rates. By late 1998, certain routes on the backbone were being upgraded to OC-48 lines, moving traffic at 2.5 Gbps. BBN anticipates having OC-192 lines installed by late 1999 which can handle around 10 Gbps. The international lnternet, outside the U.S., has trailed the magnitude yet matched the rate of U.S. growth. Significantly, the number of international pdvate lines (IPLs) used by U.S. carriers jumped from 26,000 in !995 to over 155,000 in 1997 (see "International Circuit Usage by U.S. Carriers" on page 98). The great majority of this growth appears to be due to capacity acquired by ISPs or corporate intranets. DANTE, the pan-European research network, had a two Mbps backbone in 1996, upgraded to 34 Mbps in 1997, then 155 Mbs by 1998. Carriers state simitar growth rates. France Telecom’s data networking division, Transpac, reports that traf- fic on its 155 Mbps backbone increases at a rate of 15 percent a month, which translates to 400 percent per year, and the net- work will be upgraded to 622 Mbps by 1999. Pan-Pacific links have grown equally fast. There were four 34 Mbps lines from Asia to the U.S. in the first half of 1996 and ten such links a year later. Asia started 1998 with 13 links of 45 Mbps connecting the region to the U.S., and by the second half of the year the number grew to 35 Internet links of 45 Mbps. Within Asia, however, the dedicated Internet bandwidth growth lagged due to high infrastructure costs and national rivalries to become a regional "hub." Instead, the four major Asian backbones--Singapore’s SingNet, Japan’s A-Bone, Australian carrier Telstra’s Big Pond and Hong Kong’s Net Plus--still mostly swap traffic in California. The amount of traffic passed through these telecom pipes also generally doubles each year, barring the one-off explosion of Internet traffic, which occurred as soon as the web and the Mosaic browser caught the imagination of mainstream comput- er users. (By one count, web traffic in 1993 grew by 341,634 percent, overshadowing the then-killer Internet application called gopher, which that year grew by a mere 997 percent.) AS late as 1994, the Web still ranked behind file transfer protocol (FTP) as the most popular lnternet application in terms of traf- fic. Today, around 75 percent of traffic is generated by the Web. The number of Web sites has exploded. It went from an esti- mated 130 sites in June 1993, to 2,738 in June 1994. Twelve months later there were 23,500. In June 1996 there were 252,000; by June 1997 there were I, 117,255. At this point, the growth began to slow to the tried-and-true annual Internet doubling. By May 1998 the number was estimated at 2,215,195. Traffic data showed a similar skyrocketing effect. However, the oft-cited figure that traffic doubles every three months seems misleading--it does not specify on what mutes, nor whether the growth is an aberration or a standing trend. Rather, Andrew Odlyzko and K.G. Coffman of AT&T Labs conclude that growth TeleGeography 1999 © TeleGeography, Inc. 1998 rates are "lower than is often cited [yet] still about 100 percent a year." Public lnternet traffic doubled annually from 1990 to 1994. But from 1994 to 1996 it increased by 1,000 percent. AT&T calls it an "anomalous period," noting that traffic returned to 100 percent annual growth in 1997. Will the growth of the global Internet follow these trends or not? Consider this: Today’s bandwidth consumption and Internet growth has been maintained almost exclusively through adding more users, and through current users doing more. Corporate customers, who represent around half of all Internet users, gen- erated 0.4 Mbps of traffic during peak usage in 1996. By 1998, consumption grew to 1.7 Mbps per customer, says The Yankee Group (www.yankeegroup.com). But ever since the Web popularized the Internet, the growth has not been from more bandwidth-intensive services, other than data-heavy graphics. All this bandwidth consumption is happening over e- mail and the Web. There’s a lot of room for more growth in the number of users, today estimated at around 100 million people. Other band- width drivers, such as online software distribution, online music purchasing, and Internet "bots" that crawl the Web looking for the best priced product, have scarcely penetrated the mass market. Nonetheless, current drivers of demand like video, music, and networked-based applications likely represent a major portion of tomorrow’s lnternet uses. "Silicon Cockroaches" Mike O’Dell, Chief Scientist at UUNet, wagers that the key dri- ver for bandwidth wilt come from what he calls "silicon cock- roaches"--the always-on devices such as mobile phones, per- sonal digital assistants (PDAs) and yet-to-be-conceived multi- purpose devices that will require connectivity. That the Net’s appetite for bandwidth depends upon such things as IP tele- phony and streaming video, are "notions bound up with the immediate interaction of a human," he says. "Those become thinner and thinner assumptions" in the future. "Things that think want to link," says Nicholas Negroponte, the director of MIT’s Media Lab and author of Being Digital. Whether they are small, portable devices or a nationwide net of linked vending machines, ubiquitous connections may also make these silicon cockroaches hungry bandwidth hogs. For instance, grandma wants to buy a sweater so she launches a shopping robot to compare pdces on thousands of Web sites from Bogota to Bucharest while she naps. There will also be a symbiosis between wireless and fixedqine networks. For example, a car’s radio-fed global positioning system (GPS) will constantly maintain its geographic coordi- nates so if the driver becomes lost, or is involved in an accident, the speediest assistance can be had. The wireless base-station monitoring progress along the route might be continually pass- ing the information over terrestrial cables to query map data- Figure 19. Internet Host Growth 1981-1998 100,000,000 10,000,000 1,000,000 100,000 10,000 1,000 Internet Hosts, 1991-1998 36,739,000 hosts July 1998 1,776,000 hosts d u ly 1993 ,~,0~" ,,~’d~3,000 hosts July 1988 100 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 Source: Network Wizards Design by TeleGeography, Inc. © 1998 In years past, each host on the Intemet represented one computer. But the definition of hosts has chanl~ed such that a sinl~le computer can act as many hosts with many names and many addresses all at once. Network Wizards, a company tryinB to keep track of lnternet host Browth, admits that "it is not possible to determine the exact size of the Intemet" but does publish biannual host count data on its web site. For more details, visit www.nw.com/zone/host-count-histo- ry. What about the number of lntemet users? Unfortunately, there is no sinl~le definitive number, larBely due to differing methods of counting who is online when. Some estimates for t997 lnternet users include: 68.7 million International Data Corp., www.idcresearch.com 100.5 million Nua, www.nua.ie 102 million MIDS, www.mids.orB More recent numbers for the U.S. lnternet population include: 70.5 million Nielsen/CommerceNet, www.commercenet.com 67.1 million -Intelliquest, www.intelliquest.com 56.9 million MediaMetdx, www.mediametrix.com 56.0 million International Data Corp., www.idcresearch.com 132 © TeleGeography, Inc. 1998 TeleGeography 1999 Figure 18. Internet Growth in Western Europe 18 Internet User Growth, 1998-2002 17 16 14 1.9 ¯ Users 2002 Users 1998 9.5 10 11.4 Finland Sweden Spain Italy France United Germany Kingdom Source: IDATE, The WoddAt]as ofthe Intemet, http://www.idate.fr Design byTeleGeography, Inc. © 1998 tor took over as the key bandwidth vendor. BBN, the compa- ny that created and first operated the forerunner to the Intemet, the ARPANET, had an lntemet backbone running at T- 1 speeds of between one and three Mbps in 1993. In 1996 its backbone ran at between 150 and 200 Mbps or OC-3. In eady 1998, the company deployed OC-12 lines at 622 Mbps rates. By late 1998, certain routes on the backbone were being upgraded to OC-48 lines, moving traffic at 2.5 Gbps. BBN anticipates having OC-192 lines installed by late 1999 which can handle around 10 Gbps. The international lnternet, outside the U.S., has trailed the magnitude yet matched the rate of U.S. growth. Significantly, the number of international pdvate lines (IPLs) used by U.S. carriers jumped from 26,000 in 1995 to over 155,000 in 1997 (see "International Cimuit Usage by U.S. Carders" on page 98). The great majority of this growth appears to be due to capacity acquired by ISPs or corporate intranets. DANTE, the pan-European research network, had a two Mbps backbone in 1996, upgraded to 34 Mbps in 1997, then 155 Mbs by 1998. Carders state similar growth rates. France Telecom’s data networking division, Transpac, reports that traf- fic on its 155 Mbps backbone increases at a rate of 15 percent a month, which translates to 400 percent per year, and the net- work will be upgraded to 622 Mbps by 1999. Pan-Pacific links have grown equally fast. There were four 34 Mbps lines from Asia to the U.S. in the first half of 1996 and ten such links a year later. Asia started 1998 with 13 links of 45 Mbps connecting the region to the U.S., and by the second half of the year the number grew to 35 Internet links of 45 Mbps. Within Asia, however, the dedicated Internet bandwidth growth lagged due to high infrastructure costs and national rivalries to become a regional "hub." instead, the four major Asian backbones--Singapore’s SingNet, Japan’s A-Bone, Australian carder Telstra’s Big Pond and Hong Kong’s Net Plus--still mostly swap traffic in California. The amount of traffic passed through these tetecom pipes also generally doubles each year, barring the one-off explosion of Internet traffic, which occurred as soon as the web and the Mosaic browser caught the imagination of mainstream comput- er users. (By one count, web traffic in 1993 grew by 341,634 percent, overshadowing the then-kilter lnternet application called gopher, which that year grew by a mere 997 percent.) AS late as 1994, the Web still ranked behind file transfer protocol (FTP) as the most popular Internet application in terms of traf- fic. Today, around 75 percent of traffic is generated by the Web. The number of Web sites has exploded. It went from an esti- mated 130 sites in June 1993, to 2,738 in June 1994. Twelve months later there were 23,500. In June 1996 there were 252,000; by June 1997 there were 1,117,255. At this point, the growth began to slow to the tried-and-true annual Internet doubling. By May 1998 the number was estimated at 2,215,195. Traffic data showed a similar skyrocketing effect. However, the oft-cited figure that traffic doubles every three months seems misleading--it does not specify on what routes, nor whether the growth is an aberration or a standing trend. Rather, Andrew Odlyzko and K.G. Coffman of AT&T Labs conclude that growth 131 TeleGeography 1999 © TeleGeography, Inc. 1998 rates are "lower than is often cited [yet] still about 100 percent a year." Public lntemet traffic doubled annually from 1990 to 1994. But from 1994 to 1996 it increased by 1,000 percent. AT&T calls it an "anomalous period," noting that traffic returned to 100 percent annual growth in 1997. Will the growth of the global lnternet follow these trends or not? Consider this: Today’s bandwidth consumption and lnternet ~tained almost exclusively through adding =oh currem ,~sers doing more. Corporate ,ent arour nil of all lnternet users, gen- -raffic du peak usage in 1996. By ew to 1 .~ }ps per customer, says ,’ankeegro om). But ever since lnternet, growth has not been f; more bandwidth-.’~.’.ensive services, other than data-het’.,y graphics. All this bandwidth consumption is happening over e- mail and the Web. growth has been more users, and customers, who r( erated 0.4 Mbp, 1998, consumpt~.. Yankee Group (w Web popularized There’s a lot of room for more growth in the number of users, today estimated at around 100 million people. Other band- width drivers, such as online software distribution, online music purchasing, and Internet "bots" that crawl the Web looking for the best priced product, have scarcely penetrated the mass market. Nonetheless, current drivers of demand like video, music, and networked-based applications likely represent a major portion of tomorrow’s Internet uses. "Silicon Cockroaches" Mike O’Dell, Chief Scientist at UUNet, wagers that the key dri- ver for bandwidth will come from what he calls "silicon cock- roaches"--the always-on devices such as mobile phones, per- sonal digital assistants (PDAs) and yet-to-be-conceived multi- purpose devices that will require connectivity. That the Net’s appetite for bandwidth depends upon such things as IP tele- phony and streaming video, are "notions bound up with the immediate interaction of a human," he says. "Those become thinner and thinner assumptions" in the future. "Things that think want to link," says Nicholas Negroponte, the director of MIT’s Media Lab and author of Being Digital. Whether they are small, portable devices or a nationwide net of linked vending machines, ubiquitous connections may also make these silicon cockroaches hungry bandwidth hogs. For instance, grandma wants to buy a sweater so she launches a shopping robot to compare prices on thousands of Web sites from Bogota to Bucharest while she naps. There will also be a symbiosis between wireless and fixed-line networks. For example, a car’s radio-fed global positioning system (GPS) will constantly maintain its geographic coordi- nates so if the driver becomes lost, or is involved in an accident, the speediest assistance can be had. The wireless base-station monitoring progTess along the route might be continually pass- ing the information over terrestrial cables to query map data- Figure 19. Internet Host Growth 1981-1998 100,000,000 10,000,000 1,000,000 100,000 10,000 1,000 Internet Hosts, 1991-1998 100 1981 1983 1985 1987 36,739,000 hosts J uly 1998 1,776,000 hosts // July 1993 /°,~3,000 hosts July 1988 1989 1991 1993 1995 1997 1999 Source: Network Wizards Design toy TeleGeography, Inc. © 1998 In years past, each host on the Intemet represented one computer. But the definition of hosts has changed such that a single computer can act as many hosts with many names and many addresses all at once. Network Wizards, a company trying to keep track of Intemet host Brewth, admits that "it is not possible to determine the exact size of the Intemet" but does publish biannual host count data on its web site. For more details, visit www.nw.com/zone/host-count-histo- ry. What about the number of Intemet users? Unfortunately, there is no sinBle definitive number, largely due to differing methods of counting who is online when. Some estimates for 1997 Intemet users include: 68.7 million International Data Corp., www.idcresearch.com 100.5 million Nun, www.nua.ie 102 million MtDS, wv~v.mids.org More recent numbers for the U.S. Internet population include: 70.5 million Nielsen/CommerceNet, www.commercenet.com 67.1 million -Intelliquest, www.intelliquest.com 56.9 million MediaMetrix, www.mediametrix.com 56.0 million International Data Corp., www.idcresearch.com 132 © TeleGeography, Inc. 1998 TeleGeography 1999 bases thousands of miles away. The databases might be feed- ing graphic-intensive maps back to the car. But it will require bandwidth. Today, the Israeli-based company MailPush (www.mailpush.com), backed by BellSouth, offers to contact customers via phone or fax when new e-mail ardves. It does so, by querying customers’ mail servers every seven minutes. Even where people are involved, the demands for bandwidth will increase as new consumer electronic devices are introduced. The Internet’s past growth has been bound up with the com- puter. This probably will not last long. Internet access devices other than PCs, such as %1 set-top boxes, web-enabled phones and devices like address books, planners, and videogame con- soles are expected to triple annually. By around 2004 or 2005, the number of web-enabled devices sold is expected to exceed PCs. The still bulky Nokia 9000 cell phone is probably a harbinger, with its web browser, keyboard and address book. Network Applications Not to be discounted are the applications. Although the Web represents the most traffic, the most commonly used applica- tion is e-mail, with 4.5 terabytes a day accounting for over 400 Mbps of bandwidth. E-mail attachments also matter because there is some evidence they are getting longer and thus require more bandwidth. New word processing and graphics programs appear to be the main reason. Another application that will affect bandwidth demand is elec- tronic commerce. In addition to clicking on more web pages, many purchases will likely use digital certificate technology. A digital certificate is a piece of cryptographic software that attests to the identities of both buyer and seller. The certifi- cates themselves are issued by a certification authority, which would probably be linked with other such authorities higher up in a hierarchy, for cross-certification. As a result, the authenti- cation needed for a single purchase might generate scores of traffic sessions and database queries around the world. In the future, therefore, entering a web site’s URL may entail at least two domain name system lookups, multiple cache queries over a cache network to find a locally-stored version of the web page without having to go to the actual server, and finally an exchange of cross-certified digital certificates---all before the desired web page actually appears in a browser’s window. High Speed Local Access A faster set of "on ramps" for the Internet will also affect the demand for long-haul bandwidth. Today the "last mile" of the Intemet is typically the slowest: 56 Kbps and below. It follows Amdahl’s law, that the speed of a system is determined by its slowest component¯ (Gene Amdahl, was the father of the first mainframe computer, the IBM System/360, and inventor of the parallel processor in 1967.) New local access technologies will not only improve performance for end-users, but in so doing, it Figure 20. NAP Traffic as a Measuring Stick George Glider’s monthly newsletter for high tech investors, the Gilder Technology Report (GTR), publishes one of the few publicly available estimates on the growth of Internet traffic. Each month Glider’s team collects statistics on the packets that move through selected major U.S. Network Access Points (NAPs) and Metropolitan Area Exchanges (MAEs). Based on these data sets, GTR claims that its estimates account for only 20 percent of the Intemet’s total traffic. GTR’s traffic estimates build on the model once used to measure traf- fic on the NSFnet backbone. Like the NSFnet data, it combines the traffic totals for each exchange point. Thus, bits transiting through two or more NAPs/MAEs on a single tdp may be duplicated. On the other hand, this data underrepresents total traffic because it does not include traffic (a) within an individual network; (b) between net- works with private peering points (up to 60 percent of all traffic); or (c) at other exchanges which are outside the U.S. The future of GTR’s metric looks gloomy. In April 1998, the NAPs stopped reporting traffic and it is unlikely that new data will be gen- erated. For information on how to subscribe to the GTR, visit www.gildertech.com or call them at 1 413 274 021 I. 1,400 1,200 -o 1,000 _a 800 NAP Traffic Growth, 1991-1998 600 578.9 Tb 400 163.8 Tb 200 31.4 Tb Source: Gilder Technology Report (wvwv.gildertec h.com) 133 TeleGeography 1999 © TeleGeography, Inc. 1998 Figure 21. Starr on the Net: Big Crowds, But Not for Long When the U.S. Congress decided to post the Independent Counsel’s report on President Clinton’s White House affairs, some pundits pre- dicted an Internal crash. Not so, according to Matdx Information and Directory Services (MIDS, www.mids.org), a company which tracks Internet traffic jams: While there have been discernable slowdowns at the web sites where the Start report has been posted, Internal traffic on a nationwide basis has not been significantly impacted by the release of the Independent Counsel’s report. The accompanying graphs were generated from data we collected at our San Francisco beacon on Friday, September I 1, t998, the day the Starr Report was released, and the preceding Friday, September 4, The Starr day graph shows a decided spike at noon Pacific Daylight Time (PDT). The most interesting lines on the graphs are the two averages. The spike appears in both these averages. The curve labelled "m" (and shown in green in the online versions of the figures) depicts the arith- metic mean, which is the most common kind of average. The curve labelled "M" (and shown as the line between the two light shadings) depicts the median, which is a kind of average favored by Internal performance researchers for this type of data. These averages are of latency, or round trip time from our beacon to the destination nodes, for ping (ICMP ECHO), as indicated by the left hand scale, which is in milliseconds. The graphs also have a black line labeled "n’; which shows the number of responding nodes divid- ed by I0. The nodes pinged for this data were web servers. While the "normal" Friday shows a peak in the morning, tapering off in the afternoon and evening, Friday the 11th shows a tremendous peak at noon PDT, which is 5PM EDT. The Starr Report document was "officially" released at 2:50PM EDT. We cannot of course say for sure that this spike in Internal web server latencies was caused by the release of the Start Report, because for example there could have been some other popular information released at about the same ~me. We can, however, say that such a spike does not normally occur and the coincidence of timing is unusual enough to be inter- prated as con~efation. We observed the same spike from other bea- cons, so it is not an artifact of that particular beacon or its vantage point. The spike indicates a real event in the Internal. Nothing in recent times has caused a spike quite like that’, not the Olympics (Nagano and Atlanta); not the beginning or end of the World Cup. To contact MIDS, send email to jsq@mids.org. Traffic on a Regular Friday MIDS IWR Stats: 980904.00 980904.23 Source: Matrix Information & Directory Services, http:fiwww.rnids.org/press/starr.htrnl Traffic on Day of Start Report’s Release MIDS IWR Stats: 980911.00 980911.23 134 © TeleGeography, Inc. 1998 TeleGeography 1999 will greatly influence the demand placed on IP backbones and their global architecture. Today, access between home and office is fairly evenly split. For residential users, the majority access the Internet with analog modems at speeds between 14.4 and 56 Kbps (see Figure 23). A plethora of new local access technologies is slowly invading the market: cable modem access, DSL (digital subscriber line) and, in Europe, ISDN (integrated services digital network). It is important to note that while backbone speeds have risen by a factor of 13 over the past five years, local access technologies have not been as quick to improve, albeit they have almost doubled annually for the past few years. When the growth of online service providers and Internet access spurted due to the mass appeal of the Web in 1995, modems of 14.4 Kbps were standard fare. In 1996, it doubled to 28.8 FOps, and by 1998 rates of 56 FOps became the norm, despite their actual throughput being closer to 33 Kbps. The lnternet industry is now bracing for a discontinuous jump in local access rates: Cable modems offer high-speed access any- where between 1.5 and, theoretically ten Mbps. DSL offers rates between four and eight Mbps. Another high speed local access technology may be the electdc power grid, if Nortel’s Digital PowerLine project is successful. The technology, being trialed in Europe and Asia, transmits Internet traffic over elec- tricity lines at speeds exceeding one Mbps. For users, these technologies offer great promise. A short video clip, for example, that today takes 46 minutes to trickle in over a 28.8 Kbps modem would zoom in over a ten Mbps cable modem in eight seconds flat (see Figure 24). Impact On IP Backbones The network architectures of these high speed access technolo- gies affect the design and demands placed on upstream IP backbones. DSL is a telecom technology; it travels across the same twisted pair a regular analog modem uses. That is one reason it is such an attractive service--it requires no new local infrastructure, just a new (but expensive) box of electronics on the customer’s premises and at the tetco’s central office which can then be linked to the Internet. But since it uses the infra- structure plant of the telecom universe, DSL technology is starkly different from the hierarchical architecture of cable net- works (and electrical power grids)--and this affects the back- bone infrastructure needed to support the local access speeds. Cable networks resemble "leaf and branch" networks. Unlike the telecom system’s point-to-point links, cable networks are trees, which rely on a sedes of capillaries and customer end points. The challenge of making these networks lnternet-ready is formidable. For instance, it is necessary to protect users at Figure 22. A Signaling Channel for the Internet? Today’s smart telephone networks set up calls with a technology known as Signaling System 7 (SS-7]. The system uses two channels for each PS’FN call: one for the conversation and a separate, dedi- cated, channel which transmits data for call set up, addressing, deliv- ery, and metering. Will an Intemet equivalent to the phone network’s Signaling System 7 emerge? Some trends suggest it may. The Intemet’s addressing system, the domain name system (DNS), consumes bandwidth and congests touters on the transmission infra- structure, as special servers translate the alphanumeric domain names into Intemet Protocol (IP) number addresses for routing. It may be more efficient to route the DNS on a separate channel-- especially if plans called "secure DNS" move forward. Spearheaded by lntemet gurus Carl Malamud, John Gilmore and the Intemet Software Consortium, it applies digital signature encryption technol- ogy to ensure that web sites are the dghtful party they say they are- a critical step in bdn~ing sound security to the underlying infrastruc- ture. With a plethora of encryption keys being invisibly passed between user and web site, taking DNS off the transmission network and onto a separate, more reliable channel may make sense. The same logic-- and the same channel~pplies for digital certificate technology, which may form the bedrock of electronic commerce. It requires con- tinual use of network-based certification authorities to authorize transactions between parties. Them are other reasons to dedicate a channel to network informa- tion. Today, the "Squid" system of caching hierarchies commercialized by Mirrorlmage (www.mirrorimage.net) already acts akin to signaling. If a local cache does not have a copy of a requested web page, the server sends out a query to other peer caches to see if they do. This might entail messages zipping among 20 caches at a time. A new technical model has the top cache server in the hierarchy constantly querying the contents of downstream caches, to act as a sort of index that knows which caches have what pages. Like signaling, all this information is used by the network, not the end user. Finally, lnternet telephony may push the Net to adopt a signaling channel. The Internet Engineering Task Force, a standards body, is at work to meld IP addresses and telephone numbers. And lntemet tele- phony companies have formed an industry working group to create an lnternet version of the telephone system’s "call data record" (CDR) structure, the first step in defining call parameters, service quality and billing, and in tracking interconnections among operators. Taken together, the Intemet just might learn a b~t of intelligence from the telecom world. © TeleGeography, Inc. 1998 135 TeleGeography 1999 © TeleGeography, Inc. 1998 Figure 23. A Slow On-Ramp: Internet Users and Access Speeds Worldwide Home PC Internet Access, 1997 Other ISDN 1% 6% Analog Modems 93% ISDN 17% Worldwide Home PC Internet Access, 2002 Cable Other Modems 2% 8% DSL 9% ~,nalog Modems 64% Source: Adapted from International Data Corp. (www.idcresearch.com),Telecommunications and Networking programs, 1998. Design by TeleGeographyo Inc. © 1998 branches farther from the main soume of the data from suffer- ing poorer service due to bandwidth-hungry customers on branches slightly ahead. Importantly, the deployment of these local access technologies may subtly change the distributed architecture of the Intemet backbone upstream. In the case of cable and electrical power networks, there will be an engineering imperative to build new exchange points to keep as much of that fast traffic off of the backbones as possible. And caching technology can be easily deployed where trunk meets branch. It was an obvious move then for @Home, the ISP affiliate of cable operator TC1 (which AT&T acquired in 1998), to forge an alliance with a cache tech- nology vendor. And to say these types of networks lend them- selves well to multicasting Internet content becomes a tautol- ogy: cable service and electrical power provision are multicast- ing. Real-Time Applications What will people do on the Net when the "last mile" bottleneck is broken? Real-time applications are one possibility. AOL users currently transmit 15 million e-mails on peak days--but 1 16 million "Instant Messages" are sent. Then there’s voice traffic, the first area where real-time traffic flows appeared over lntemet infrastructure. Yet voice isn’t a big bandwidth con- sumer, as the next section explains. Instead, think big: think time-sensitive graphics applications and video. Big bandwidth is required for three-dimensional modeling and "white-boarding"--the ability for many users to work on a doc- ument at the same time. Corporate pioneers like Boeing Corp. 136 already use these applications; in fact, the latest Boeing 777 aircraft was mocked up entirely by engineers collaborating dig- itally worldwide. Moreover, Microsoft’s NetMeeting software, which allow for multi-person video and audio conferencing over the Internet, is being trialed by ISPs. Kesmai Studios (www.kesmai.com), a multi-user Internet-based games company, counted 1.5 million users in the fall of 1998, and doubled its bandwidth usage over the course of the year But what’s most remarkable is that 90 percent of its users access the Internet on analog modems with rates between 28.8 and 56 Kbps. Today, the company engineers its products to take this into account, but expects the real bandwidth adven- ture to begin when faster local access exists. Online games may be a niche service for 14-year-olds today, but the technology required for handling multiple point-to- point real-time traffic can be directly cross-applied to broader commercial uses, such as in medicine, education, and business. Today, many working groups of the Internet Engineering Task Force (1ETF), the Net’s standards setting body, are broadcast over the M-Bone, or "Multimedia Backbone," an experimental virtual network atop the Intemet that is engineered to support real-time video traffic. If there’s enough bandwidth generally, the entire lnternet may support these applications. And as the IETF protocol developers say: "If you build it, they will come." Bandwidth Conservation Technologies Before we conclude that faster lntemet "on-ramps" and new applications will radically boost traffic and hence the bandwidth demanded of the lntemet’s spoke-like architecture, some cau- © TeleGeography, Inc. 1998 TeleGeography 1999 Figure 24. Why Broadband Matters Download ~me for a 3.5 Minute Video Clip Using Different Technologies 28.8 Kbps modem 128 Kbps ISDN modem~~~~,,~ ~i 10 minutes 4 Mbps cable modem 20 seconds 8 Mbps ADSL router I 10 seconds 10 Mbps cable modem 1 8 seconds 0 500 1000 1500 2000 2500 3000 Transfer Time (seconds) Source: FCC and TeleGeography, Inc. © TeleGeography, Inc. 1998 tion is necessary. Fast on-ramps or not, some new services may well reduce the overall demand for global bandwidth. The lntemet is like a living organism. It has feedback mecha- nisms to protect it if too much stress occurs. In this case, as congestion grows due to the bandwidth consumption of new services, the Internet community tends to discover ways of delivering more efficient routing and transmission. For example, caching lnternet content closer to the user--say, at a local ISP--eliminates the need for some of the bandwidth that otherwise would have been used to access Web content over long-haul routes. And deploying IP multicast technology which aggregates flows, means that each new viewer of stream- ing media does not add much to the overall tong-haul band- width consumption, though it certainly will at the last mile. Finally, taking fax and voice traffic off the PSTN may increase tntemet bandwidth, but it also frees up capacity on another part of the network (in this case, the PSTN which may share the same long haul transmission facilities). All three services require a closer look. Internet Caching Caching is misunderstood, believes Barry Raveendran Greene, a Cisco engineer in Singapore. On the surface it saves band- width by locally storing or coding the most frequently request- ed Web pages rather than satisfying each new request for the same page from a remote web server One company, Mirrorlmage Inc. (www.mirrorimage.net), claims to save between 30 to 40 percent on local caches and up to 75 per- cent of bandwidth consumption when local caches are linked to an upstream cache. David Peterschmidt, CEO of rival cache company Inktomi Corp. (www.inktomi.com), claims one European carder realizes up to 80 percent bandwidth cost say- ings. Greene himself admits that Singapore’s government-man- dated proxy servers, which also act as caches, effectively dou- ble the bandwidth of the city-state’s three ISPs, which other- wise would be on a trans-Pacific route to the U.S. or Canada. But those figures are misleading. A cache simply permits the end user to access the page faster The result: "Users click more," says Greene. And eventually, they click through cached content and into non-cached Web pages. A cache saves on bandwidth costs and makes networks more efficient, but it doesn’t reduce the need for bandwidth per se. There is no evi- dence that carriers are provisioning less bandwidth due to implementing caches. In fact, carders that have deployed caches, such as Australia’s Telstra, follow the industry trend in bandwidth consumption; in Telstra’s case, ten percent growth per month, corresponding to 200 percent a year, according to John Hibbard, Telstra’s vice president of international carrier business. To be sure, caching is becoming an essential part of network design for international carders. Since 1997 it has gone from small scale deployments with individual ISPs to large scale out- lays by major carriers. But caching is really about network evo- lution, not about saving money or bandwidth. In the same way as computing went from mainframes to PCs--changing the architecture of where information is stored based on the cost/performance factors of memory--so too is caching merely adding a storage capability to network design. And caching proponents are also looking skyward--at satellites. Since more than one ISP may need access to the content, satel- lites are being called in to beam the content to local proxy servers on an ISP’s network. Teteglobe and lntelsat teamed up 137 TeleGeography 1999 © TeleGeography, Inc. 1998 in late 1998 to trial a such a service for ISPs outside the United States (see Figure 25). Streaming Media Streaming media, the term for bringing audio and video onto the lnternet by streaming its packet flow, is often predicted to bring the Net to its knees. It may actually lighten the lnternet’s load. Currently, each request for a web page~including video or audio content--generates its own unique flow of data. Two users in the same office building, for instance, get two separate streams. However, some networks that host sites specializing in real-time media content are able to deploy multicasting tech- nology that cleverly aggregates the flows to the optimal point, before sending it onward to different users. Once multicasting protocols are standardized, the service will operate over numerous backbone networks. One flow instead of hundreds on the long-haul lines saves bandwidth. Webcasting and streaming video and audio are increasing in popularity. NetRadio (www.netradio.com), which offers 150 Figure 25. Satellites and the Internet While futuristic broadband satellite systems like Celestd, Skybfidge and Teledesic get ready to light up the sky with Internet access, today’s birds are beaming bits now. There are actually some clever advantages to using satellites for the Intemet, especially for lntemet service providers outside the United States. They provide connectivity in a single direction, and are well- suited for point-to-muitipoint uses, like streaming media. Also, satel- lites offer an inexpensive way to transfer large amounts of regular data, like the current 20 megabits per day of USENET data. Some ISPs in Europe and the U.S. provision them to assure redundancy in case terrestrial links fail. In the future, using satellites for individual Intemet connectivity is set to become standard fare (see "The New Global Mobile Satellite Systems" on page 106). Satellites found their first niche in providing Intemet service in regions untouched by international fiber lines, such as the Middle East, Afdca and most of South Amedca. For many countries in those regions, satellites are the only means of gaining Intemet access. For most other countries with a choice of infrastructure, a satellite is the last resort since the performance lags far behind terrestda! fiber. When it makes sense, it is often because it permits ISPs to bypass the artificially high prices of the national carrier. Even in deregulated Australia, the ISP Ozemail Ltd. was able to cut costs by 37 percent by using a 45 Mbps satellite connection across the Pacific, skirting Telstra Corp. and Optus Communications Ltd., the country’s subma- rine cable duopoly. Yet there are many difficulties with using satellites for tnternet ser- vice. They have much greater delay than fiber, which engineers have to take into account. Also, many satellite systems contain technical limitations to how much bandwidth they can transmit at any time. The barrier for the best birds is 155 Mbps, but most satellites are transmitting under 45 Mbps. Finally, the base TCP/IP protocols have to be jerry-rigged. Intemet traffic flows customarily send an acknowl- edgment that each packet arrived safely at the next hop--but back- channeling all these "acks" over the Earth to get to the previous hop wastes expensive capacity, so it is not done. That said, satellite use for the Net--estimated at under five percent of total Intemet capacity--is more than doubling each year for some powerful commercial reasons. Satellite links, like the Intemet itself, are asynchronous. Leasing a trans-oceanic circuit offers a two-way channel, yet the majority of Intemet traffic tends to be inbound from the U.S. A typical web surfer tdcldes out a couple of bits to request content from a Web site, and back at him streams graphics, music and video clips. Far better--and in some cases cheaper--is to send U.S.-bound traf- fic over cable, and use satellite for the data-heavy return path. Over the past few years a number of providers around the wodd, includ- ing Brazil’s Embratel and UUNet France, have done just that. Interpacket Inc. (www.interpacket.net) of Santa Monica, California has made substantial inroads into Latin America, and the France- based NetSat (www.internetsat.com) has similarly in the Middle East and Afdca. Indeed, by supplying the bandwidth via satellite rather than cable, service providers outside the U.S. are able to turn the tables on U.S.- based ISPs. As it stands today, non-U.S.-based providers must pay for the full cost of circuits into the U.S., unlike the telephony world where the cost is evenly shared. Thus, purchasing one-directional capacity lets the ISP outside the U.S, maintain good service quality for its own users (who access U.S, content), without providing better connectivity to the growing num- ber of U.S. users who wish to access content in the foreign country without paying for the international link. Satellite technology may thus impact the politics of backbone financing and force U.S. back- bones to stdke a new commercial deal with off-shore ISPs. Other innovative Internet satellite uses include content caching, and two companies--SkyCache (www.skycache.net) and Mirrorlmage (www.mirrorimage.net)--already offer such services, by beaming popular web pages to proxy caches to reduce the need for US-bound bandwidth. Finally, as a broadcast technology, satellites are well- suited for multicasting web content to many destinations in a single flow. That’s what Telegiobe had in mind when they began trialing a "push" service in late 1998 to send specified content directly to ser- vice providers outside the U.S. The company believes the key driver will be beaming high-demand, data-heavy, real-time web content worldwide, © Telel3eography, Inc. 1998 138 © TeleGeography, Inc. 1998 TeleGeography 1999 different music and program channels, counts a monthly global audience of three million listeners. Broadcast.corn, with its real-time cybercast events, reported one million new visitors to its web site the day Clinton’s heretofore "secret" video testi- mony hit the Net. Though a small fraction of visitors actually saw the video, that was enough to send the company’s stock price skyrocketing. The numbers of users for the services are growing, too. RealNetworks Inc. (www.realnetworks.com), the most popular streaming media software vendor, said that 3.5 million users downloaded its product in July 1998, double the number from a year earlier Currently 28 million users have installed the soft- ware. The company is aware of the stress web video and audio places on backbones, and has allied with cache companies and ISPs to implement architectures to decrease the long-haul bandwidth consumption. Engineering backbones to handle this amount of traffic is diffi- cult. UUNet, for example, needed to create a sort of virtual overlay network, a service they call UUCast. But with abundant bandwidth, that may not be needed. And the bandwidth sav- ings from aggregating the same traffic flow to many users will decrease the overall bandwidth consumption the service requires. Internet and IP Telephony The migration of fax and voice traffic from the PSTN to the lntemet is also unlikely to be a major driver of new bandwidth. That could make the loss of PSTN revenues more painful to incumbent carriers, however, because off-setting increases in bandwidth fees may not materialize. Research firms love to conjure up statistics about this scenario. London-based Tarifica has predicted that by 2001 incumbent carriers will lose from three to 11 percent of their yearly rev- enue due to Internet telephony. And there’s no doubt that Internet telephony and fax services are growing quickly. A num- ber of new carriers have begun services in 1998, such as RSL, PSINet and ICG Netcom, as well as incumbent carriers like Telnor, Telia, Deutsche Telekom and even AT&T, in Asia. Actual usage, however, appears limited and does not seem to support the most fearful forecasts. Instead, industry executives are now downplaying whether Internet telephony actually costs less than circuit-switching, and are promulgating all the benefits computer-telephony integration will bring (see Figure 26). Importantly, there is a difference--sometimes not understood-- between Internet and IP telephony. The former uses the public Internet’s infrastructure, mixing the voice calls with unpre- dictable web traffic and routing it all over ISP exchange points to swap it with other providers. On the other hand, 1P tele- phony simply uses lnternet Protocol as the means to packetize the voice traffic, often for use over a corporate user’s regular leased lines. In the case of PC-to-phone calls, popularized by VocalTec (www.vocaltec.com), users dial in to their ISP and the traffic is then transmitted over the public lnternet. But the majority of firms in the market seek to keep the calls on their own network for as long as possible, to lower costs and ensure better quality. As a result, the traffic may never actual- ly be on the public Internet per se, as it often makes sense to dedicate a circuit optimized for voice over IR And nascent fed- erations of IP telephony companies are extending their cover- age by striking interconnection agreements among themselves, creating a de facto parallel network. But either on the public Internet or not, in terms of bandwidth consumption the end result is the same--operators and users need to provision bandwidth for IP calls, rather than use the PSTN. Figure 26. InternetTelephony: Build Once, Run Anything? Packets, schmackets. In the shont term, tntemet telephony--phone calls and faxes--is cheaper on some routes because it runs over inter- national private lines that avoid the accounting rate regime. But it is only marginally more efficient, technically speaking. Where the circuit-switched PSTN transmits long distance calls using com- pression algorithms of 32 kilobits per second, IP telephony can com- press data over a connectionless network at eight Kbps. Yet it often results in poorer sound quality. Scrap the accounting rate, and wait for advances in circuit-switched technology to compress voice more efficiently, and the cost difference soon disappears. Mike O’Dell, the chief scientist at UUNet decries IP telephony’s cost advantages as: "one of the great myths of all time." But the best engineering and business minds proclaim IP telephony wilt be a winner since it will enable new communications applications that meld voice and data. That’s the logic behind the new networks of Qwest Communications Inc. and Level 3 Communications Corp., which are building massive, pure-IP networks. One early application of this convergence is MCI WorldCom’s Vault service, which lets users click on web-enabled call centers. "IP Telephony is not always cheaper," admits Jeff Pulver, an IP tele- phony consultant. "The investment that needs to happen [is] so that the service providers have the infrastructure to support the wide range of future tP-based services, tP voice will be just one." Yet there may be cost advantages, nevertheless. Juha Hein~inen, Telecom Finland’s chief IP network architect, believes "the real sav- ings of voice over the Internet is not cheaper phone calls--the sav- ings come from the infrastructure: You don’t need to build more than one [network]." © TeleGeography, Inc. 1998 139 TeleGeography 1999 © TeleGeography, Inc. 1998 Figure 27. The ITU and the Internet The International Telecommunication Union [ITU), long considered irrelevant in the brave new wodd of the Interact, has made a con- certed effort to catch up. It presented its prestigious silver medal to Vint Cerf, the co-inventor of lnternet Protocol, in November I995, followed by one in July 1998 to Jon Postel, an Internet old-timer who headed the Intemet Assigned Numbers Authority. In September 1998, it announced that all new standards work would be immedi- ately reassessed to see how it might dovetail with protocol develop- ments at the Net’s standards body, the Internet Engineering Task Force. That comes despite the ITU’s refusing to recognize IETF spec- ifications for years because it is not a government-sanctioned stan- dards group. At the tTU’s quadrennial plenipotentiary conference in Minneapolis in October t998, U.S. Vice President AI Gore called on governments and industry to boost Intemet penetration, especially in the devel- oping wodd, to promote democracy and electronic commerce: Today, t want to pose five great challenges. Together, they make up a Digital Declaration of Interdependence. First, we must improve access to technology so everyone on the planet is within walking distance of voice and data telecommunica- tions services within the next decade. Second, we must overcome our language borders and develop technology with real-time digital translation so anyone on the planet can talk to anyone else. Third, we must create a Global Knowledge Network of people who are working to improve the delivery of education, health care, agn’cultur- al resources and sustainable development---and to ensure public safety. Fourth, we must use communications technology to ensure the free-flow of ideas and support democracy and free speech. Fifth, we must use communication technology to expand economic oppor- tunity to all families and communities around the globe. The full text of Vice President Gore’s speech can be found on the ITU’s web site at http://www.itu.int/newsroom/press/PP98/Documents/ Statement_Gore.htmL © TeleGeographyo Inc, 1998 And what about the impact on total bandwidth? The lower price of Internet telephony and fax will boost consumption, which in turn increases the need for bandwidth dedicated to IP. But the result may be that a carrier simply leases an extra cir- cuit to a firm offering Internet telephony. More traffic is likely to be squeezed onto that circuit than before when it was allo- cated to PSTN traffic. The result could be a net decrease in total bandwidth required. What happens in practice thus will bear close watching because, in theory, IP telephony seems unlikely to be a significant factor in changing the bandwidth provided by global telecom carriers. Rather, IP telephony is more likely to have its largest impact in driving down prices, the mainstay of traditional carriers’ rev- enues. IP telephony avoids the accounting rate on internation- al calls and faxes since it moves over private lines; in the U.S. long distance market, it bypasses local access fees. Executives at pre-paid calling card companies say privately that on many routes, the calls already are routed via Internet gateways. In some cases, IP telephony just makes technological sense. Faxes are better suited to the Internet’s store-and-forward data networking architecture than to a dedicated circuit engineered for real-time transmission. Yet, very few companies use the public lnternet’s infrastructure today because there is essentially no way to control quality of service, especially if the traffic travels between different net- works. Common standards for IP telephone services could change that (see Figure 27). In the interim, three models are emerging. One is to offer IP telephony as an end-to-end ser- vice on a single provider’s network. That costs a lot of money 140 to build, and only the operators with the largest global footprint can manage it. But that’s also the best way to control costs, since no interconnection is required. It was no coincidence, then, that PSlnet (www.psinet.com) began offering corporate users an IP telephony service at the same time as it announced the purchase of numerous Asian ISPs, after having bought out a fleet of ISPs in Europe over the course of the preceding 12 months. Another model is for "federations" of ISPs, interested in being IP telephony operators, to emerge. This formula is championed by GRIC Communications Inc. (www.gric.com). The company began as an "ISP roaming" service that allowed users of differ- ent service providers to use infrastructure in areas where their own network lacked coverage. It now brings disparate providers together under a single roof. GRIC licenses the tech- nology, and oversees a settlement between the different ISPs whose network are used--all the while taking a small commis- sion for its efforts. Finally, true IP telephony clearinghouses may spring up, such as ITXC Corp., the Internet Telephony Exchange Carrier run by for- mer AT&T executive Tom Evslin (www.itxc.com). Unlike GR1C, it is less an alliance and more of an independent service. It hopes to serve as the third-party broker among all the different IP telephony operators. In the first model, the traffic stays on a single provider’s IP net- work. The other models have traffic swapped between numer- ous providers--in GRIC’s case, over their regular public lnternet infrastructure, and in ITXC’s case, most likely over separate IP networks dedicated to lnternet telephony traffic. © TeleGeography, Inc. 1998 TeleGeography 1999 Figure 28. Future Scenarios Three Possible Future Scenarios 1. Complete network convergence: ¯ A single network that melds voice and data emerges. ¯ Rising bandwidth demand due to new applications keeps backbone prices up. ¯ Long distance voice calls are (nearly) free. ¯ New regional infrastructure reduces US-centric Internet. 2. Heterogeneous networks: ¯ Two networks co-exist, one optimized for voice, one for data. ¯ Capacity exceeds demand, with drastic fall in backbone prices. ¯ Cheap trans-oceanic bandwidth discourages regional infrastructure; US-centric Internet continues. ¯ Service-oriented firms exploit cheap bandwidth, stealing carriers’ retail business. 3. Neither and Both: ¯ Separate networks continue due to cost of convergence. ¯ Pace of global Internet adoption slows, mirroring take up of telephony. ¯ Excess bandwidth drops prices; capacity lit up incrementally to keep prices higher. ¯ The trend towards convergence remains just a trend, as multimedia services stay primarily on broadcast and cable TV networks. © TeleGeography, Inc. 1998 Regional Exchange Points The demand for international bandwidth will also be affected by the deployment of more private and regional lnternet exchanges. Here, too, there are strong conservation incentives. As we have seen, international ISPs are switching traffic via the U.S. based on a marketplace anomaly--the high cost of tete- corn capacity in their region due to the historic lack of compe- tition. As that changes, infrastructure prices should fall, mak- ing it advantageous to keep more local traffic local. But for that to happen, more local exchange points will be needed. In the U.S., a number of companies have made private exchange points the mainstay of their business model. lnterNAP (www.internap.com) and Sawis (www.savvis.com) are perhaps the most prominent. In Savvis’ case, the ISP oper- ates nine private NAPs so as to terminate around 90 pement of their lnternet traffic directly. By providing customers with "the shortest, fastest, and ’cleanest’ route to almost any location on the Web," Savvis believes content providers and downstream ISPs wil! sign on. The companies deploying private NAPs pay to peer with the larger backbone networks--indeed, they rely on the backbones for transit to reach other networks and customers downstream in the rare cases where they lack a direct connection. Both companies seek out the most efficient routes for traffic flows, which results in more efficient bandwidth consumption on their domestic backbones. If the model were applied international- ly, trans-oceanic bandwidth would be reduced as carriers that send intra-regional traffic across an ocean to exchange it, would instead keep it local. The percentage of traffic in Asia and Europe that stays local is already changing. As the number of Internet users outside the U.S. grows (see Figure 16), so too will new lnternet content companies and electronic commerce firms that attract their traffic. And with that, the number of exchange points that bypass the U.S. will grew. Over the past 18 months, a nation- al NAP has been built in almost every European country. But these are generally used for national traffic. In Asia, exchange points don’t really exist: traffic is routed at the national loca- tion of each regional backbone initiative--be it Singapore, Japan, Hong Kong or Australia. However, as we’ve already seen, in almost all cases the different backbones themselves do not exchange traffic inside the region but haul it to the U.S. V. Conclusions: What Happens Next? The Economist magazine once called the Internet the "acciden- tal superhighway." Keep that in mind in thinking about where it will go from here. No one really knows. The Internet is both the catalyst and embodiment of a new world--one that little resembles the present one. Consider the 1998 World Cup soccer tournament. The event, which attract- ed the highest number of television viewers ever, took place in France. But of the four computers hosting the Cup’s web site, only one was located in Paris. The others were on the east coast, west coast, and middle of the United States, mirroring the topology of the Internet rather than the geography of the event. In a world of materialism and industry, geography mat- ters; transporting raw materials, goods and people from place to place is central for businesses. But in a digital economy, TeleGeography 1999 © TeleGeography, Inc. 1998 information is the main product of value, and connectivity is what matters. We’ve seen that here. The largest 1P backbone providers are at the top of the lnternet food chain, benefiting from downstream ISPs who are their customers. But being on top has a cost-- the backbone carriers are forced to make continual massive investments in infrastructure to keep up with demand. In an era of bandwidth scarcity, that sounds like a terrific business model. Yet if bandwidth becomes plentiful (see Figure 29), these same ISPs may find themselves on the losing end as con- nectivity becomes a commodity. But, let us consider a less extreme case. How might the future differ if long haul bandwidth is less plentiful because of large new demands, say from IP entertainment services? What of the Internet’s future then? There are at least three possible sce- narios (see Figure 28). The first is almost complete network convergence, which many "next-gen" carriers are hoping for. In this world, voice is a small fraction of traffic. So it is subsumed into a single network, called the Internet, that has enough capacity to carry any real- time applications. New types of communication, melding voice and data, emerge. These new applications treat bandwidth as abundant so they never seek to compress or optimize it, result- ing in a continual demand for bandwidth that keeps prices high. However, the price of voice calls plummets to its provisioning cost of near zero. The fast pace of both demand and new capacity continues making the commercial environment more predictable. The economies of scale derived from a single network, and the lower cost of entry for new network construction due to new technologies, also encourage new regional infrastructure. That fosters more direct connections between countries, dramatical- ly decreasing the U.S.-centricity of the Internet. The radical jump in demand between 1994 and 1998; the growth of capacity from 1999 to 2003; and the phenomenon of U.S. hubbing for Internet traffic, all come to be seen as historical aberrations. Figure 29. Voice’s Shrinking Role U.S. Carrier Trans-Oceanic Circuit Usage (Gbps), 1995-2000 1200.00 14% 1000.00 800.00 E 600.00 400.00 "-’1 200.00 PSTN Share of Total Capacity 0.00 ==’==~ 1995 1996 1997 Idle Circuits/IRUs ¯ IPL Capacity [] PSTN Capacity Reserve [] Foreign Owned 1998 1999 2000 12% 10% 8% 6% 4% 2% o% Note: Data based on TeleGeography submarine cable research and FCC circuit status reports filed by U.S, carriers (see "U.S. Carrier Circuit Usage" in the International Facilities section of this report). "Reserve" capacity assumes that 100 percent of used capacity is also held in reserve for restoration. "Foreign-owned" capacity includes circuits owned by foreign carriers, which the FCC estimates at 25 to 30 percent of total capacity. "Idle Circuits/IRUs" includes circuits owned by carriers at year end but not in use as well as unsold capacity on submarine cables. Satellite capacity utilization is generally not reflected by this data because U.S. carriers do not acquire international satellite capacity in advance, © TeleGeegraphy, lnc, 1998 142 © TeleGeography, Inc. 1998 TeleGeography 1999 In the second scenario, networks are heterogeneous: voice is a small fraction of traffic, yet its different technical characteristics makes a separate network tenable. The huge increase in avail- able capacity is not matched by a corresponding rise in demand, leading to a bandwidth glut, with a drastic fall in pdces. And since the cheapest bandwidth is trans-oceanic, the U.S.-centdc nature of the Internet continues, which reinforces the economic disincentive to build out regional lnternet infra- structure. In this world, companies that focus on services above the trans- port layer, many of them small and innovative, can profit from cheap connectivity at the expense of the carriers who built it, and who focus mainly on transmission. The final scenario is a combination of both, yet it is distinct. The cost and technical means of transitioning the network costs more and takes longer than the optimists predict. This leads to a mixed network and application environment for the next decade. And just as over half the world has never made a tele- phone call, so too does the pace of Internet use, and the demand on the network, progresses slowly relative to the avail- able bandwidth. Capacity greatly outstrips demand so pdces fall, yet new capacity is only lit up incrementally in a bid by car- riers to maintain (somewhat) higher prices than otherwise would be the case. Hence, the Intemet’s topology continues to be largely U.S.-centric, because provisioning high capacity backbones via the U.S. is still cheaper than many direct con- nections. Similarly, the trend towards convergence is just that, a slowly progressing trend. But there’s one "wild card" that must be on every carrier’s radar screen: the role of entertainment. Web users are watch- ing less television. It might not be because Internet users are smarter than the rest of us. It’s because the Net is fast devel- oping into an entertainment medium. Soon, as cable modems, Internet-ready TV appliances and sites such as Broadcast.corn gain mainstream acceptance, the difference between the TV and the computer may vanish. But will the Net’s infrastructure morph accordingly? Probably not, though it is far too early to say. What is certain is that it wilt keep the regulators busy in their efforts to modernize national and international rules for network convergence. Despite the growth of satellite TV transmission, local proBram- ming still remains the most popular around the world, barring noted exceptions like Dallas, Michael Jordan, and The X-Files. CNN and Dow Jones’ Wall Street Journal know this: They’ve invested heavily in local editions of their products. Thus, despite the "world wide" name of the Web, as the Net becomes more entertainment-oriented, we should expect more local content, not less (see Figure 31). That should boost regional traffic on Internet backbones, not trans-oceanic vol- umes. So stay tuned. The future, as it’s often said, may not be what it used to be. Will telecom carriers be prepared? ~i~ Kenneth Nell Cukier covers the Internet as a Senior Editor for CommunicationsWeek International in Paris. He can be reached by email at kenc@cwi.emap.com. Figure 30. Internet Protocol Developments on the On going changes to the base Internet Protocol are intended to make the Net as reliable and scalable as its voice telephony counterpart. Here are a few additions to the alphabet soup of tntemet abbrevia- tions: OoS Intemet engineers are devising ways to build quality of ser- vice (OoS) into IR which specify a class of service for which the user pays extra. Two models are being studied. One specifies pri- ority traffic and gives those packets precedence to tunnel through congested touters. Another simply drops "out-of-profile" packets in periods of congestion, i.e. "pay up or die..." Either way, the result is the same: not all packets are treated equally. RSVP Another protocol enhancement, called Resource Reservation Protocol (RSVP), reserves bandwidth in touters all along the path, similar to "call set-up" on the PSTN. Users are assured of getting the bandwidth an application or session requires, but pay extra for it, since it effectively denies the band- width to other users even if it is not all used. Yet the technique Horizon doesn~ scale very well: There’s a small limit to the number of simultaneous reservations a router can handle, and there is no way to assure another RSVP-compliant network will respect the bandwidth reservation. Nevertheless, it suggests the telephony and data wodds are drawing closer. IPv6 An Internet Protocol address which assigns unique numbers used for routing traffic, is limited in size to 32 bits which allows for four billion hosts. However, as Christian Huitema, the former head of the Internet Architecture Board, points out, "it would be immoral" not to provision for at least every human--with penhaps I O0 different devices connected, from pace makers to toasters. IPv6, so far only deployed on research networks, uses a 128-bit address, allowing for a quadrillion separate hosts. It also has security, muiticast, and real-time flow enhancements to today’s IPv4 (version 5 exists experimentally). Yet, lacking a sufficient incentive to migrate to IPv6, the marketplace has resisted it. © TeleGeography, Inc. 1998 143 TeleGeography 1999 © TeleGeography, Inc. 1998 Acknowledgments I gratefully thank my editors, Jason Kowal and Greg Staple of TeleGeography, who (patiently) challenged and encouraged me. While scores of conversations helped form this article, the following people deserve special mention: Fred Baker, Erik Bataller, Scott Bradner, Robert Cohen, James Crowe, Mike O’Dell, George Gilder, Barry and Laina Raveendran Greene, Geoff Huston, David Isenberg, Mark McFadden, Sam Paltridge, Robert Pepper, Kevin Werbach, and, of course, Robert Young. Finally, appreciation is due CommunicationsWeek International for providing me the opportunity to report on the Internet. Sources and Further Reading Joseph R Bailey, "The Economics of Internet Interconnection Agreements." In Lee W. McKnight and Bailey, eds., Intemet Economics, MIT Press, Cambridge, Massachusetts, 1997. Boardwatch magazine, "Directory of ISPs." Winter 1998. http://board- watch.internet.com/isp/index.html Brian Carpenter, "RFC 1958: Architectural Principles of the Intemet" (Category: Informational). Internet Engineering Task Force, June 1996. http://ds.internic.net/rfc/rfc 1958.txt Gordon Cook, "WorldCom/MCI Merger: Intemet at Risk?" The Cook Report on the Internet; May 2, 1998. http://www.cookreport.com Frances A. Cairncross, The Death of Distance, Harvard Business School Press, Cambridge, Massachusetts, 1997. Kim Claffy, Greg Miller and Kevin Thompson, "The nature of the beast: Recent traffic measurements from an Internet backbone." Cooperative Association for Intemet Data Analysis, San Diego, California. Presented at INET98, Geneva; July 1998. httP:llwww.caida.org/Papersllnet981 index.html Kenneth Nell Cukier and Camille Mendler, "Technology or Ideology?" CommunicationsWeek International; 22 September 1997. htt p ://www.totaitele.com/cwi/191/191 sr.html Kenneth Nell Cukier, "Peering and Fearing: ISP Interconnection and Regulatory Issues." Presented at Harvard University Kennedy School of Government’s Information Infrastructure Project conference; December 1997. http://ks~vww.harvard.edu/iip/iicompol/Papers/Cukier.html Timothy Denton, "The Distribution of Signals in Cyberspace: An exam- ination of what the Internet means for signal distribution and broad- casting." T.M.Denton Consultants, Ottawa, Canada; 14 August 1998. http://www.canniff.com/tdenton/DistributionTOC.htm Jim Dixon, "The Effect of Telecoms Regulation on the European Internet." European Internet Service Providers Association, Brussels, Belgium. www.euroispa.org Presentation at the OECD Internet Workshop; June 1998 in Osaka, Japan. http://www.oecd.org/dsti/ sti/it/cm/act/dixon, pdf George Gilder, "Bandwidth Debate: WorldCom and Qwest." Gilder Technology Report, March 1998. http://www.gildertech.com Laina Raveendran Greene, "International Internet Infrastructure Financing." GetlT Pie., Singapore (www.getit.org). Presentation at the OECD Internet Workshop, Osaka, Japan; June 1998. http:llwww.oecd.org/dstilstilitlcmlact/Irg.pdf Figure 31. Quelle Langue is the Internet in? Ce n’est pus ~tonnant que l’lnternet est centralis~ aux Etats-Unis when you consider that the majodty of its content is in English. Language matters. The entertainment industry knows well that local programming draws viewers. That’s one reason why media compa- nies produce regional editions of their content. The Web directory Yahoo already boasts 11 non-English-based search engines specializ- ing in local language sites. As more local content is created in local languages, Internet traffic may remain local too. While it is difficult to determine the breakdown of English-language Web sites relative to other tongues, a 1996 survey by the lnternet Society (www.isoc.org) noted that English accounts for 82 percent of home pages on the Web, trailed by German with four percent, Japanese with 1.5 percent, French at 1.5 percent, and Spanish at I. I percent. And a July 1998 study by Network Wizards shows that over half of all domain names are registered in the United States. Luckily, technical obstacles are being overcome. Netscape and Microsoft have both released browsers in Arabic, Japanese, and other Asian and European languages. Sofe~vare products now easily render Web sites based on Unicode, a computer standard capable of coding all the world’s languages that do not use the Latin alphabet, and thus can not be digitized in ASCII (American Standard Code for Information Interchange). As the Web invades new nations, this diversity of language will grow. A symbol of this came in July 1998 when Deja News, a Web site that archives Internet newsgroups, said it would support t 7 different lan- guages. The company’s president, Guy Hoffman, said "as much as 30 percent" of the 50,000 monitored e-mail forums are not in English. The next most popular languages were Chinese, Russian, and German. But politics still gets in the way. A French association dedicated to preserving the language went so far as to sue a U,S, university’s study-abroad program since its site was in English, though the case was latter thrown out on technical grounds. Now, France’s Ministry of Culture funds French-language content creators in an effort to boost the language’s use over the Net. © TeteGeography, Inc. 1998 144 © TeleGeography, Inc. 1998 TeleGeography 1999 Katie Hafner and Matthew Lyon, Where the W;zards Stay Up Late: The Origins of the Intemeto Simon & Schuster, New York, 1996. John Hibbard, "Finding an Equitable and Sustainable Commercial Arrangement for the International lnternet." Telstra Corp., Sydney, Australia. Presentation at the OECD Internet Workshop, Osaka, Japan; June 1998. http://www.oecd.org/dsti/sti/it/cm/act/hibbard.pdf International Telecommunications Union, Challenges to the Network: Telecommunications and the Internet, (Authors: T. Kelly, H. S. Sharifi and B. Petrazzini), Geneva, Switzerland; September 1997. http ://www.itu. ch/itudoc/gs/subscirc/gs/2 2 8- 7_40 7 0 2.html J. K MacKie-Mason and H. R. Varian, "Pricing the Internet." In Brian Kahin and Jim Keller, eds. Public Access to the Internet, MIT Press. Cambridge, Massachusetts, 1995. William Lehr, "Understanding Vertical Integration in the lnternet (Draft)." Presented at EURO CPR’98 Conference, Venice, Italy; March 1998. Toni Mack with Carleen Hawn, "Bell Buster," Forbes, September 7, 1998. http://www.forbes.com/forbes/98/0907/6205154a.htm Joanna Makris, "The Bandwidth Barons," Data Communications, July 1998. http://www.data.com/roundups/barons.html Nicholas Negroponte, "Bandwidth Revisited." Wired, June 1998. http ://www.wi red. co m/wi red/6.06/negro po rite. htm l Andrew Odlyzko and K. G. Coffrnan, "The Size and Growth Rate of the Internet." AT&T Labs, Ftorham Park, New Jersey; July 1, 1998. http://www.research.att.com/-amo/doc/networks.html Organization for Economic Cooperation and Development, "lnternet Traffic Exchange: Developments and Policy." (Author: Sam Paltridge), Paris, France; t997. http://www.oecd.org/dsti/sti/it/tndex.htm Organization for Economic Cooperation and Development, "Webcasting and Convergence: Policy Implications." (Author: Sam Paltridge), Paris, France; May 1998. http://www.oecd.org/dsti/sti/it/ cm/index.htm Larry Press, "Tracking the Global Diffusion of the lnternet." Communications of the ACM, vol. 40, no. 11; November 1997. http://som 1 .csudh.edu/fac/Ipress/gdiff/index.htm Rebecca Quick and Thomas E. Weber, "lnternet Sites Contend with Video of Clinton Before the Grand Jury." The Wall Street Journal; September 21, 1998. Peter H. Salus, Casting The Net: From ARPANET to the Internet and Beyond, Addison-Wesley Publishing Co., Reading, Massachusetts, 1995. Gregory C. Staple, "The Global Bandwidth Boom: Something’s Happening Here ... But What?" Presented at the Columbia Institute for Tele-lnformation, Columbia Business School, New York; 1 May 1998. http://www.telegeography.comANhatsnew/cu_speech98.html Tom Steinert-Threlkeld et al, "The Net: Not Just Data Anymore." (Series of articles), Inter@ctive Week, August 31, 1998. http://www.zdnet.com/intweek/printhigh/83198/thenet.html U.S. Department of Commerce, The Emerging Digital Economy, Secretariat for Electronic Commerce (Authors: L. Marghedo, D. Henry, S. Cooke and S. Montes), Washington, DC; April 1998. http://www.ecommerce.gov/emerging.htm U.S. Federal Communications Commission, "Digital Tornado: The Internet and Telecommunications Policy." (Author: Kevin Werbach), Office of Plans and Policy, Working Paper Series no. 29, Washington, DC; March 1997. http://www.fcc.gov/Bureaus/OPP/working_papers/ oppwp29.pdf Kevin Werbach, "How to Price a Bit." Release 1.0, EDventure Holdings Inc., June 22, 1998. http://www.edventure.com/reiease 1/release1 .html Robert H. Zakon, "Hobbes’ Internet Timeline, v3.3." Updated summer 1998. http://www.isoc.org/guest/zakon/Internet/History/HIT.html 145 TeleGeography 1999 © TeleGeography, Inc. 1998 lnternet Hosts by Economy, 1993-1997 Figure 1. Internet Host Computers by Economy, A-D CAGR Country 1993 1994 1995 1996 1997 1993-97 Afghanistan 0 0 0 0 1 n.a. Albania 0 0 35 79 117 n.a. Algeria 0 10 16 28 49 n.a. Andorra 0 0 10 171 491 n.a. Angola 0 0 0 2 4 n.a. Antigua & Barbuda 0 0 160 169 184 n.a. Argentina 762 3,159 9,468 19,697 36,828 163.7% Armenia 0 0 173 177 443 n.a. Aruba 0 0 0 98 87 n.a. Australia 92,879 169,183 327,127 544,385 736,604 67.8% Austria 16,274 30,157 59,595 99,354 133,811 69.3% Azerbaijan 0 0 16 30 347 n.a. Bahamas 346 864 2,169 3,388 7,888 118.6% Bahrain 0 0 142 841 339 n.a. Bangladesh 0 0 0 0 0 n.a. Barbados 0 0 2 21 23 n.a. Belarus 0 1 18 257 710 n.a. Belgium 8,459 20,885 38,206 77,858 137,557 100.8% Belize 0 0 1 12 258 n.a. Benin 0 0 0 9 13 n.a. Bermuda 75 662 1,019 1,968 3,657 164.2% Bhutan 0 0 0 0 2 n.a. Bolivia 0 0 66 431 552 n.a. Bosnia 0 0 0 36 373 n.a. Botswana 0 0 24 25 552 n.a. Brazil 5,646 10,954 31,195 95,639 162,122 131.5% Brunei Darussalam 0 0 156 206 340 n.a. Bulgaria 17 138 1,060 3,327 6,851 348.4% Burkina Faso 0 0 0 1 45 n.a. Burundi 0 0 0 0 0 n.a. Cambodia 0 0 0 0 48 n.a. Cameroon 0 0 0 0 2 n.a. Canada 128,904 293,184 606,151 996,754 1,784,698 92.9% Cape Verde 0 0 0 0 0 n.a. Central African Bep. 0 0 0 6 0 n.a. Chad 0 0 0 0 0 n.a. Chile 1,478 3,319 9,608 16,865 20,177 92.2% China 14 604 2,223 19,816 16,634 202.0% Colombia 315 1,913 3,985 11,960 17,157 171.8% Cameras 0 0 0 0 0 n.a. Congo 0 0 0 1 4 n.a. Costa Rica 418 1,304 2,605 5,362 7,463 105.6% C6te d’lvoire 0 0 3 202 253 n,a. Croatia 339 1,647 2,467 4,644 8,263 122.2% Cuba 0 0 1 24 51 n.a. Cyprus 0 1 392 1,480 3,023 821.7% Czech Republic 4,582 10,683 22,483 41,904 59,411 89.8% Denmark 11,131 25,335 65,739 132,339 230,912 113.4% Djibouti 0 0 0 4 0 n.a. Dominica 0 0 0 55 76 n.a. Note: An Interact hosttypically represents one computer connected to the Intemet, although many users may be connected through a single host. Data distributes the generic Top Level Domain (gTLD) host registrations between countries according to registrations in September 1997. Source: International Telecommunication Union, Chal/enges to the Network:/nternetforDeve/epment, drawing upon lnternet hast data published by Network Wizards and RIPE, and global Top Level Domain (gTLD) d~stribution data published by Imperative.cam, © TeleGeography, Inc. 1998 146 © TeleGeography, Inc. 1998 TeleGeography 1999 Figure 2. lnternet Host Computers by Economy, D-L CAGR Country 1993 1994 1995 1996 1997 1993-97 Dominican Rep. 96 239 603 3,184 6,976 n.a. Ecuador 246 570 1,042 1,497 3,215 90.1% Egypt 63 285 850 2,464 2,978 162.4% El Salvador 0 0 23 132 196 n.a. Equatorial Guinea 0 0 0 0 0 n.a. Eritrea 0 0 0 0 0 n.a. Estonia 439 1,160 3,590 8,000 15,880 145.2% Ethiopia 0 0 1 1 78 n.a. Faroe Islands 0 0 52 96 285 n.a. Fiji 5 5 52 75 92 107.1% Finland 33,878 70,244 220,046 321,464 504,411 96.4% France 62,042 104,050 196,504 313,332 538,788 71.7% French Guiana 0 0 0 27 120 n.a. French Polynesia 0 0 0 25 190 n.a. Gabon 0 0 0 0 5 n.a. Gambia 0 0 0 0 0 n.a. Georgia 0 0 57 213 414 n.a. Germany 124,011 229,825 542,239 806,328 1,407,274 83.5% Ghana 0 0 6 203 252 n.a. Greece 1,952 4,070 8,954 18,784 33,048 102.8% Greenland 0 3 88 215 294 n.a. Grenada 0 0 0 0 1 n.a. Guadeloupe 0 0 0 7 57 n.a. Guam 0 0 55 122 77 n.a. Guatemala 48 121 291 720 1,736 n.a. Guernsey 0 0 0 5 22 n.a. Guinea 0 2 2 2 0 n.a. Guinea-Bissau 0 0 0 0 11 n.a. Guyana 0 0 0 52 67 n.a. Haiti 0 0 0 10 0 n.a. Honduras 0 0 0 408 74 n.a. Hong Kong 10,048 23,253 41,391 89,133 162,681 100.6% Hungary 3,196 7,192 16,594 31,192 71,137 117.2% Iceland 1,782 4,579 8,425 11,736 18,986 80.7% India 148 383 841 3,191 7,390 168.2% Indonesia 702 1,932 6,197 16,077 25,192 n.a. Iran 0 18 271 285 204 n.a. Iraq 0 0 0 0 0 n.a. Ireland 2,886 6,912 16,559 32,164 52,529 106.5% Israel 7,110 14,686 32,104 56,813 104,966 96.0% Italy 21,077 40,330 101,580 192,071 360,520 103.4% Jamaica 0 76 164 249 267 n.a. Japa n 42,769 96,632 269,327 734,406 1,168,956 128.6% Jersey 0 0 0 6 3 n.a. Jordan 117 294 662 1,225 2,856 n.a. Kazakhstan 0 7 188 809 1,213 n.a. Kenya 0 0 17 274 459 n.a. Kiribati 0 0 0 0 0 n.a. Korea, D.P.R. 0 0 0 0 0 n.a. Korea, Republic of 11,646 24,703 43,884 90,850 181,027 98.6% Kuwait 139 221 1,236 2,925 4,070 132.8% Kyrgyzstan 0 0 0 0 147 n.a. Lao P.D.R. 0 0 0 0 0 n.a. Latvia 60 526 1,326 5,789 7,110 230.0% Lebanon 80 199 524 1,337 2,903 n.a. Note: An Internet host typically represents one computer connected to the Internet, although many users may be connected through a single host. Data distributes the generic Top Level Domain (gTLD) host registrations between countries according to registrations in September 1997, Source: International Telecommunication Union, Challenges to the Network: lnternet for Development, drawing upon Intemet host data published by Network Wizards and RIPE, and global Top Level Domain (gTLD) distribution data published by Imperative.com. © TeleGeogrephy, Inc. 1998 147 TeleGeography 1999 © TeleGeographyo Inc. 1998 Figure 3. Internet Host Computers by Economy, L-S CAGR Country 1993 1994 1995 1996 1997 1993-97 Lesotho 0 0 0 1 0 n.a. Liberia 0 0 0 0 1 n.a, Libya 0 0 0 0 1 n.a. Lithuania 1 128 461 1,735 4,057 n.a. Luxembourg 427 838 2,608 4,746 7,695 106.0% Macau 0 12 65 179 151 n.a. Macedonia 0 0 90 194 501 n.a. Madagascar 0 0 0 27 17 n.a. Malawi 0 0 0 0 0 n.a. Malaysia 946 2,883 6,992 29,919 43,611 160.6% Maldives 0 0 0 33 52 n.a. Mall 0 0 0 15 0 n.a. Malta 0 0 69 494 824 n.a. Marshall Islands 0 0 0 0 2 n.a. Martinique 0 0 0 0 12 n.a. Mauritania 0 0 0 0 0 n.a. Mauritius 0 0 0 122 202 n.a. Mayotte 0 0 0 0 0 n.a. Mexico 4,155 8,124 17,003 35,264 54,696 90.5% Micronesia 0 0 0 38 60 n.a. Moldova 0 0 5 6 246 n.a. Mongolia 0 0 0 10 13 n.a. Morocco 0 0 230 470 1,409 n.a. Mozambique 0 0 0 31 69 n.a. Myanmar 0 0 0 0 0 n.a. Namibia 0 0 11 263 642 n.a. Nepal 0 0 19 60 139 n.a. Neth. Antilles 83 208 455 813 1,846 n.a. Netherlands 48,796 98,554 199,396 317,115 503,235 79.2% New Caledonia 0 0 1 23 82 n.a. New Zealand 6,140 32,131 55,618 87,918 177,403 131.8% Nicaragua 0 49 141 532 506 117.5% Niger 0 0 0 5 2 n.a. Nigeria 0 0 0 4 49 n.a. Northern Marianas 0 0 0 0 6 n.a. Norway 30,673 50,817 89,669 159,197 314,172 78.9% Omen 0 0 1 1 672 n.a. Pakistan 0 0 18 513 1,293 n.a. Panama 0 0 0 0 0 n.a. Papua New Guinea 0 0 0 1 53 n.a. Paraguay 0 0 0 187 299 n.a. Peru 0 172 816 5,196 3,426 821.7% Philippines 372 1,264 3,809 7,065 12,573 n.a. Poland 4,983 11,038 23,679 53,856 90,866 106.6% Portugal 3,819 5,771 13,347 26,132 48,817 89.1% Puerto Rico 0 82 82 82 260 n.a. Qatar 0 0 0 21 190 n.a. R~union 0 0 0 0 1 n.a. Romania 69 527 1,749 7,832 13,611 274.9% Russia 1,673 7,143 23,268 60,331 113,972 187.3% Rwanda 0 0 0 1 0 n.a. Sao Tom~ & Principe 0 0 0 0 12 n.a. Samoa 0 0 0 0 0 n.a. Saudi Arabia 302 758 1,683 3,069 6,751 117.4% Senegal 0 0 14 69 117 n.a. Note: An lntemet host typically represents one computer connected to the Intemet, although many users may be connected through a single host, Data distributes the generic Top Love] Domain (gTLD) host registrations between countries according to registrations in September 1997. Source: International Telecommunication Union, Challenges to the Network: Internet for Developmen~ drawing upon Internet host data published by Network Wizards and RIPE, and global Top Level Domain (gTLD) distribution data published by Imperative.com. © TeleGeogrephy, Inc. 1998 148 © TeleGeography, Inc. 1998 TeleGeography 1999 Figure 4. Internet Host Computers by Economy, S-Z CAGR Country 1993 1994 1995 1996 1997 1993-97 Seychelles 0 0 0 1 1 n.a. Sierra Leone 0 0 0 0 0 n.a. Singapore 3,665 7,481 27,653 37,129 77,403 114.4% Slovak Republic 377 1,482 3,136 8,182 15,107 151.5% Slovenia 676 1,787 5,966 14,400 20,920 135.8% Solomon Islands 0 0 9 154 17 n.a. Somalia 0 0 0 0 0 n.a. South Africa 11,576 28,601 51,698 105,053 135,891 85.1% Spain 19,386 40,669 80,139 161,606 312,675 100.4% Sri Lanka 0 0 7 350 681 n.a. St. K~tts and Nevis 0 0 0 2 5 n.a. St. Lucia 0 0 0 21 14 n.a. St. Vincent 0 0 0 0 10 n.a. Sudan 0 0 0 0 0 n.a. Suriname 0 0 0 4 1 n.a. Swaziland 0 0 1 226 331 n.a. Sweden 46,911 89,037 177,651 293,165 481,596 79.0% Switzerland 35,715 58,213 98,526 163,945 263,728 64.8% Syria 0 0 0 0 0 n.a. Ta iwa n 7,995 14,679 25,858 34,877 177,382 117.0% Tajikistan 0 0 0 0 11 n.a. Tanzania 0 0 0 3 25 n.a. Thailand 956 3,428 7,779 15,526 29,473 135.6% Togo 0 0 0 5 37 n.a. Tonga 0 0 2 8 728 n.a. Trinidad & Tobago 71 177 443 796 2,493 143.5% Tunisia 0 54 79 41 51 n.a. Turkey 1,809 5,099 12,595 29,390 83,587 143.5% Turkmenistan 0 0 0 0 3 n.a. Uganda 0 0 58 17 30 n.a. Ukraine 185 534 2,420 6,595 14,039 n.a. United Arab Emirates 0 1 366 1,804 1,946 821.7% United Kingdom 127,090 266,573 528,475 868,951 1,347,322 80.4% United States 1,323,375 2,753,210 5,220,695 8,706,416 17,247,802 90.0% Uruguay 1 176 834 1,836 10,327 821.7% Uzbekistan 0 0 35 122 98 n.a. Vanuatu 0 0 0 7 46 n.a. Venezuela 910 1,858 4,078 7,330 15,677 n.a. Vietnam 0 0 0 5 0 n.a. Virgin Islands (US) 133 331 726 1,242 3,252 122.6% Yemen 0 0 0 2 10 n.a. Yugoslavia 3 2 4 2,541 4,912 n.a. Zambia 0 69 69 173 181 37.9% Zimbabwe 0 19 93 177 600 214.9% Africa 11,639 29,042 53,175 109,929 144,379 87.7% Americas 1,467,062 3,080,753 5,913,686 9,918,654 19,394,554 90.7% Asia 87,009 193,440 473,678 1,149,020 2,032,630 119.8% Europe 614,721 1,195,352 2,569,217 4,283,025 7,213,688 85.1% Ocea n i a 99,024 201,319 302,863 632,757 915,355 74.4% WORLD 2,279,456 4,699,907 9,392,620 16,093,385 29,700,605 90.0% Note: An Internet host typically represents one computer connected to the Internet, although many users may be connected through a s~ngte host. Data d~stdbutes the generic Top Level Domain (gTLD) host registrations between countries according to registrations in September 1997. Source: Internatmnat Telecommunication Union, Challenges to the Network: Internet for Development, drawing upon Intemet host data published by Network Wizards and RIPE, and global Top Level Domain (gTLD) d~stribution data published by Imperative.com. © TeleGeography, Inc. 1998 149 TeleGeography 1999 © TeleGeography, Inc. 1998 Journey to the Center of the Web by Martin Dodge, University College London Introduction Where is the center of the World Wide Web? How do the tens of millions of pages connect to form the Web? How does the densi- ty of connections vary through web space? Are some parts tight- ly woven, while others remain loosely linked backwaters? This article explores how we might measure and map the structure of the Web to help answer these questions. To analyze the Web’s structure, I developed a metric of the cen- trality of web sites called WebX distance (Web ConneXions) based on the outgoing and incoming hyperlinks for any given site. [1] The structure of hyperlinks has been used by a number of academic researchers to analyze web structure. [2] Alternative methods use web traffic to determine the most popular and, therefore, central sites. [3] To illustrate the potential of the WebX distance metric, I analyzed a subset of the Web--122 sites of the universities and major col- leges in the United Kingdom. I visualize the results as Web Scans, using a style derived from astronomical charts. To the Center of the Web Data on the size of Web sites of the U.K. universities and num- ber of hyperlinks between them was gathered using the AltaVista search engine. A script was used to make the 14,884 separate queries to AltaVista, using the syntax +url:sitel>.ac.uk +link:site2>.ac.uk, necessary to count the interconnection between sites. Over 450,000 links were reported, although the vast majority were internal ones within sites. The connectivity data was analyzed to find the central web site, that is the node that has greatest connectivity to all the other 121 universities. This was the web site of the University of Oxford (http ://www.ox.ac. uk). To calculate the relative distance between web sites, hyperlinks were used to create a virtual distance measure--the WebX dis- tance~which was inversely proportional to the number of links between any two given sites, So the more links between two sites, the closer they are in WebX terms. The WebX distance from each university to every other one was calculated and stored as a graph. This graph was analyzed using a shortest path program to measure the distance from each web site to every other one. Figure 1. Long and Medium Range Web Scans Long Range Scan 10,000 Scale Key & Oxford Each web ring is 1,000 ¯ Web Site WebX Units wide. Source: Martin Dodge, Univers~ College London Medium Range Scan 1,000 Scale -- \\ Key ,̄ Oxford Each web ring is 100 ¯ Web Site WebX Units wide. ©TeleGeograph% Inc. 1998 150 © TeleGeography, Inc. 1998 TeleGeography 1999 Figure 2. Short Range Web Scan Key .:\ Oxford Web Site Each web ring is 10 WebX Units wide. Bham Birmingham Cam Cambridge Ed Edinburgh HW Heriot-Watt IC Imperial College KCL Kings College London Liv Liverpool Man Manchester NCL Newcastle (]MW (]ueen Mary & Westfieid College $oton Southampton Strath Strathciyde UCL University College LoNdon Source: Martin Dodge, University College London © TeleGeography, Inc. 1998 The results were normalized by web site size. Oxford’s web site had the lowest mean distance being the closest to all other sites. The centrality of other web sites can then be expressed in term of their distance from the Oxford center point (see Figures 1 and 2). Web Scans Taking an astronomic metaphor, the Web Scans have Oxford’s site as the sun. It is the central point of gravity and light for our small Web solar system, around which ]21 planetary web sites spin, their orbital distances being equal to their WebX distance from Oxford. Purple trace lines highlight this distance from the sun. The closer a site is to the Oxford sun, the more central it is. The orientation of the web sites is arbitrary, based on five degree intervals determined alphabetically from the name of the univer- sity, although in the future this position could be used to encode other data such as their approximate geographic location. Rings are overlaid on the scans, spaced at regular WebX distance inter- vals, to act as a visual guide to the scale of the system. Three Web Scans are shown in this article, mapping the system at dif- ferent scales. They were created using a GIS (geographical infor- mation system). Even though GIS technology is usually employed to analyze the real world, it is quite capable of mapping virtual worlds like the Web. The largest scale, "long range" Web Scan, shown in Figure 1 above, shows the whole system, including the most distant Web sites. The furthest away--the equivalent of Pluto in our Web solar system--is the University of Wales Institute Cardiff (http://www.uwic.ac.uk) way out beyond the 10,000 unit ring, by far the least well connected web site. The most striking feature is the dense cluster of web sites at the center of this system, forming a dense red nebula inside the smallest dng. There are a number of peripheral sites, such as the University of Westminster, Humberside University, the London Business School and Dartington College of Arts. To see more detail at the heart of the system we switch to a mid range Web Scan. The "medium range" scan in Figure 1 shows only the sites that are 1,000 units or closer to Oxford: At this scale, the web sites have become recognizable as planet-like disks rather than just red dots, their size proportional to the number of pages the site contains. At 874 WebX units from Oxford, Liverpool John Moores University is the furthest out in this scan, followed by University College Chester orbiting at 807--and because it is a small site, with just over 700 pages, it is represented by a very small red dot. The "short range" scan shows the 24 Web sites that are closest to the sun (see Figure 2). These are the core of the academic Web in the United Kingdom. Immediately striking are the two giant sites very close to Oxford, the universities of Cambridge and Edinburgh, which have large Web sites and are very closely interconnected. Cambridge has a WebX distance of ten and Edinburgh is only slightly further out at 13. Imperial College comes next with a WebX distance of 26, double that of the sec- ond place site. Further out from the top three, there is a cluster 151 TeleGeography 1999 © TeleGeography, Inc. 1,998 of sites around the 40 WebX mark. These are Heriot-Watt University, the universities of Leeds, Glasgow, Southampton, Queen Mary & Westfield College, and my own institution, University College London (UCL). All these universities are well connected and centrally placed in the academic Web. UCL has a WebX score of 42, putting it in seventh place away from Oxford, a respectable position given its historic role in the development of Cyberspace in the U.K. (it was the first organization in Britain connected to ARPANET, the lntemet’s forerunner, back in 1973). Finally, right on the edge of this scan is the University of Durham’s site with a WebX score of exactly 100. Why Measure WebX? Understanding the Web’s structure, using metrics like my WebX distance, is important for a number of reasons. On one level the Web is worthy of analysis and mapping because it is the terrain of the Information Age. Just as surveyors map the real world, so we should begin to chart virtual spaces. On a more prosaic, eco- nomic level, the Web has already become a significant source of wealth and employment, yet we know really very little about it. The WebX measure of centrality has potential practical benefit for users and content producers alike. Research has shown that measures of centrality, based on hyperiinks, can improve the accuracy of the results from search engines. [4] Hyperlinks can be thought of as virtual citations, which can aid in assessing the credibility and veracity of web sites. Knowledge of a page’s con- nectivity and centrality would prove useful in this regard. Perhaps what is required is a dynamic Web Scan map which is built into browser software so one can assess the location of the current web page in relation to a central site for the category of interest. It would enable users to see at a glance whether they are in the bustling downtown or lost out in the backwaters. The Web Scan could also help find potentially useful pages and sites that are nearby but are not immediately visible from your current location. The browser tool developed by Alexa Internet (http://www.alexa.com) offers some of the functionality described above. It can recommend sites similar to the one you are view- ing, but lacks any map presentation. Carl Malamud and Marshall Rose, at Invisible Worlds (http://invisible.net), plan to create interactive 3D web maps so that browsers can visualize the rela- tionship between different information resources on the Internet. Two demonstration maps are planned for 1999, one focusing on financial information and the other on music and media. Web Galaxies in the WWW Universe I analyzed one small web galaxy, the sites of U.K. academia, which is a homogenous and well connected part of the Web Universe. The U.K. academic web is part of larger stellar struc- tures, as it is an important component of the U.K. Web and is also connected to academic sites in other countries. There are really many different centers of the Web depending on what you are interested in. We might envision the Web as like the universe, with many different galaxies, clusters and nebulas. There would be massive bright galaxies of new information in one sector, and elsewhere the dark nebulas of long abandoned sites and the nascent clouds of newly forming web spaces. Different galaxies would be formed by web pages and sites that have sim- ilar content (e.g., all movie sites) or that are related by country or language. Each of these galaxies would have their own cen- tral points and would merge into larger cosmic structures. At the heart of the ever-expanding web universe are several extremely bright supernovae. They are known as portals in the current jargon and are the key points where many people begin their web explorations, There is currently a scramble by the largest sites to create web supernovae given the promise of finan- cial rewards. It will be interesting to see, as we approach the mil- lennium, which of these portals go the way of real supernovae and burn-out and which remain shining bdghtly to guide us through the web universe. Martin Dodge (m. dodge@ud, ac. uk) is a researcher in the Centre for Advanced Spatial Analysis (CASA), University College London, and the curator of "An Atlas of Cyberspaces," at www.cybergeography.org, which is mirrored in North America at www. cybergeograph y. com. Notes [1] I am very grateful to Naru Shiode for his assistance in calculating the WebX distance metric. [2] See for example, Tim Bray, "Measuring the Web" (http://www5conf.inda.fr/fich_html/papers/Pg/Overvlew.html) and Jon Kleinberg, "Authoritative sources in a hyperlinked environment" (http//sirnon.cs.comell.edu/home/kleinber/auth.pdf) [3] Several commercial companies survey Web site popularity, includ- ing Relevant Knowledge (http://www.relevantknowledge.com), Media Metrix (http://www.mediametrix.com), Web21 (www.100hot.corn). [4] See Mark Frauen Felder, "The Future of Search Engines," The Industry Standard, September 28, 1998, pp. 34-37; see also techni- cal papers by Massimo Marchiori, "The Quest for Correct Information on the Web: Hyper Search Engines," http://decweb.ethz.ch/WWW6/ Technical/Paper222/Paper222.html; Jeromy Carriere and Rick Kazman, "WebQuery: Searching and Visualizing the Web through Connectivity," http://www.cgl.uwaterloo.ca/Projects/Vamsh/webq uery- 1 .html. Also, see Google (google.stanford.edu), an experimental search engine cre- ated by researchers at Stanford University. 152 Traffic I TeleGeography 1999 © TeleGeography, Inc. 1998 © TeleGeography, Inc. 1998 TeleGeography 1999 Albania Largest Telecommunications Routes, 1997 Destination Percentage of O~going Traffic 6. 7. & 9. 10. 11. 12. 13. 14. 17. 20. Millions of MiTts 1. Italy .................. 15.8 2. Greece ............... 14.1 3. Germany .............. 2.0 4. United States .......... 1.8 France ................ 0.7 Switzerland ........... 0.7 Turkey ................ 0.6 Yugoslavia ............ 0.6 United Kingdom ........ 0.6 Austria ............... 0.5 Belgium ............... 0.4 Macedonia ............ 0.4 Bulgaria .............. 0.2 Croatia ............... 0.2 Canada ............... 0.2 Netherlands ........... 0.2 Egypt ................. 0.2 Romania .............. 0.2 Spain ................. 0.1 Hungary .............. 0.1 Other ................. 1.2 Total 40.8 ::~;!,,~,~,:~ 4.9% ~,: 4.4°/o ~i 1.8% ~ii! 1.6% 1.5% ~-~ 1.4% ~,~i 1.3% ii, o.9% 0.6% 0.4% 0.4% 0.4% o.4% o.4% 0.4% 0.3% 2. O/o © TeleGeogra phy, Inc. 1998 National Traffic Balance MiTT 1995 1996 1997 Incoming 37.3 46.0 n.a. Outgoing 23.1 20.6 40.8 Surplus (Deficit) 4.2 15.4 n.a. Total Volume 60.4 66.6 n.a. Note: M~’lq" ~s Minutes of Telecommumcatzons Traff=c Data are in mzlhons of minutes of public sw=tched traff=c Data for 1997 are based on bzlhng point of traffic. Route data may exclude some cross-border traff=c to Yugoslawa. © TeleGeography, Inc. t998 155 TeleGeography 1999 © TeleGeography, Inc. 1998 t Andorra Largest Telecommunications Routes, 1997 Destination Millions of MiTTs 1. Spain ................. 30.1 Z France ................ 7.4 3. Portugal .............. Z0 4. United Kingdom ........ 0.9 5. Germany .............. 0.3 6. Belgium ............... 0.2 7. Netherlands ........... 0.2 8. Switzerland ........... 0.2 9. Italy .................. 0.2 10. United States .......... 0.2 Other ................. 0.5 Total 42.2 Percentage of Outgoing Traffic ~~! 17.~% 4.7% ~i 2.2% o.6% i o.6% 1 o.5% Io.5% !o.5% 1o.4% 1.2% © TeleGeography, Inc. 1998 National Traffic Balance MiTT 1995 1996 1997 Incom!ng n.a. ¯ ¯ ¯ ¯ 27.3~ 30.1 Outgoing .... 36.0, 37.8 ...... 42.2 Surplus (Deficit) ,, n.a. (10.5) (12.ii Total Volume n.a. 65.1 72.3 Note: MiTT zs Minutes of Telecommunications Traffic. Data are m m~lhons of mznutes of pubhc switched traffic. 156 © TeleGeography, Inc. t998 © TeleGeography, Inc. 1998 TeleGeography 1999 Argentina Largest Telecommunications Routes, 1997 Destination Millions of MiTTs 1. United States ........... 42.3 2. Uruguay ............... 35.4 3. Brazil .................. 27.4 & Chile .................. 19.5 5. Spain .................. 14.2 6. Italy ................... 11.2 7. Peru ................... 10.7 8. Rolivia ................. 10.2 9. Paraguay .............. 10.2 10. France ................. 4.5 11. Mexico ................ 4.1 12. Germany ............... 4.0 13. United Kingdom ......... 3.8 14. Canada ................ 2,6 15. Venezuela .............. 2,3 16. Colombia ............... 2,3 17. Switzerland ............ 1,6 18. Israel .................. 1,3 19. China .................. 1,2 20. Cuba .................. 1.0 Other .................. 13.7 Total 223.4 © TeleGeography, Inc. 1998 National Traffic Balance MiTT 1995 1996 1997 Incoming 299.4 390.7 444.2 Outgoing 179.4 181.3 223.4 Surplus (Deficit) ¯ !,19.9 209.4 220.7 Total Volume 478.8 572.0 667.6 Note: Mi’i-r is Minutes of Telecommunications Traffzc. Data are in milhons of minutes of pubhc swJtched traffic Data for 1995 and 1996 are based on billing poznt of traffic Data for 1997 based on originating pmnt of traffic. © TeleGeography, Inc. t998 157 TeleGeography 1999 © TeleGeography. Inc. 1998 ( Armenia Largest Telecommunications Routes, 1997 Destination Millions of MiTTs 1. Russia ................ 29.9 2. Ukraine ............... 3.2 3. Georgia ............... 1.5 4. Beiarus ............... 0.6 5. Uzbekistan ............ 0.5 6. Kazakhstan ............ 0.4 7. Moldova .............. 0.1 8. Turkrnenistan .......... 0.1 Other ................ 12.5 Total 48.8 Percentage of Outgoing Traffic 6.6°/o 1.2% 0.8% i0.2% © TeleGeography, Inc, 1998 National Traffic Balance MiTT 1995 1996 1997 Incoming n.a. n.a. n.a. Outgoing n.a. 48.1 48.8 Surplus (Deficit) n.a. n.a. n.a. Total Volume n.a. n.a. n.a. Note: Mil-r is Mznutes of Telecommumcations Traffic. Data are in millions of mznutes of pubhc sw=tched traff=c. Totals for 1996 include only traffic to other member states of the Commonwealth of Independent States The "Other" cate- gory may include routes to non-CIS member states, including an estimated 1 0 milhon minutes to the U S., that rank among the top destmatzons for outgoing traffic. © TeleGeography, Inc. t998 © TeleGeography, Inc. 1998 TeleGeography 1999 Australia Largest Telecommunications Routes, 1997 Destination Millions of Mi’l’~s Percentage of Outgoing Traffic 1. United Kingdom ....... 230 ~.~,~:~.~!~-~’~.~’~’~ 15.2% Z New Zealand 175 ’~,~’~,~,~ 11.6% 3. United States ......... 170 ~:~,~,~:~,~,~,~ 11.3% 4. Singapore ............ 65 4.3% 5. Hong Kong ............ 60 4.0% 6. Japan ................ ~2.6% 7. Indonesia ............. 40 2.6% 8. China ................ 40 2.6% 9. Malaysia ............. 40 ~] 2.6% 10. Canada ............... 2.6% Other 610 ............ ’ .................. ............... ............ ,;-::~:;-,,~;~," ,~,,-’:~ .... 40.4% Total 1,510 © TeleBeography, tnc. 1998 National Traffic Balance MiTE 1995 FY 1996/97 FY 1997/98 Incoming n.a. n.a. 1,250 Outgoing 1,024 1,305 1,510 Surplus (Deficit) n.a. n.a. (260) Total Volume n.a. n.a. 2,760 Note: MiTT is Minutes of Telecommumcations Traffic. Data are in m~ll~ons of minutes of public switched traffzc. Route data are rounded to the nearest 5 m~lhon m~nutes. Because fiscal year reporting replaced calendar year reporting in 1996, totals for 1995, FY 1996/97, and FY 1997/98 are not dzrectly comparable. Rscal year ends 30 June. © TeleGeography, Inc, 1998 159 TeleGeography 1999 © TeleGeographyo Inc. 1998 ( Austria Largest Telecommunications Routes, 1997 Destination Millions of MiTTs Percentage of Outgoing Traffic 1. Germany .............. 427.6 2_ Italy.. 60.6 6.1% 3. Hungary .............. 39.3 3.9% 4. Yugoslavia ............ 38.1 ~3.8% United Kingdom ........ 33.6 3.4% Turkey ................ 29.1 2.9% Poland ................ 27.5 2.8% United States .......... 24.8 2.5% Netherlands ........... 24.6 2.5% Croatia ............... 24.2 2.4% France ................ 23.9 ~t 2.4% Czech Republic ........ 23.7 2.4% Slovenia .............. 14.9 1.5% Slovak Republic ........ 14.6 ~!~! 1.5% Romania .............. 13.7 i~ 1.4% Switzerland ........... 12.9 1.3% Bosnia-Herzegovina .... 12.3 1.2% Spain ................. 12.1 1.2% Belgium ............... 12.0 ~: 1.2% Sweden ............... 10.5 1.1% Other 115.9 -~:~,~:,~,~?~ 11.6% Total 9~.0 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. © TeleGeography, Inc. 1998 National Traffic Balance MiTT 1995 1996 1997 Incoming n.a. ~,, n.a. 957.7 Outgoing ,, 901 960 995.5 Surplus (Deficit), ¯ ¯ ¯n.a.’ ........ n.a. (37.8) Total Volume n.a. n.a. 1953.2 Note: MiTr is Mznutes of Telecommumcations Traffic. D ata are in mdlions of minutes of public sw=tched traffic. Data based on bilhng point of traffic Traff=c figures exclude most cross-border traffic to Sw=tzerland. 160 © TeleGeography, lnc. 1998 © TeleGeography, Inc. 1998 TeleGeography 1999 Bahamas Largest Telecommunications Routes, 1997 Destination Percentage of Outgoing Traffic 2. ~,; 5.1% 3. 2.5% 4. !i~ 2.0% 6. 7. 8. 9. 10. Millions of MiTTs United States ........... 53.6 Canada ................ 3.2 United Kingdom ......... 1.6 Jamaica ............... 1.2 Switzerland ............ 0.4 Germany ............... 0.3 France ................. 0.3 Haiti ................... 0.3 Turks & Caicos Islands.. 0.3 Brazil .................. 0.2 Other .................. 1.3 Total 62.7 0.7% O.5% 0.5% 0.5% 0.4% 0.4% ii~ 2.0% © TeleGeography, Inc. 1998 National Traffic Balance Mi-FI 1995 1996 1997 Incoming n.a. n.a. n.a. Outgoing ¯ ,n.a. 56.7, ¯ ¯ 62.7 Surplus (Deficit) ,, n.a. n.a. n.a. Total Volume n.a. n.a. n.a. Note: MzF] =s Minutes of Telecommumcations Traffzc. Data are in m~lhons of m~nutes of public switched traffic. © TeleGeography, Inc. 1998 161 TeleGeography 1999 © TeleGeography, Inc. 1998 Bahrain Largest Telecommunications Routes, 1997 Destination Millions of MiTTs Percentage of Outgoing Traffic Z Saud] Arabia 15.4 ~.~%~~.!.~;~i:;...~.~;~,~i.~.~,.~:.~ 14.4% 3. United Arab Emirates... 13.2 ~~~~~ 12.4% 4. United Kingdom. 9.0 ~,~,:~%~-,~,%~,%~, 8.4% 5. Kuwait 5.5 ~,~ ~;~,~ ~’,~,~ 5.2% ................ L 6, Pakistan .............. 5.1 ~~ 4.8% 7. Egypt ................. 4.8 ~,,~;~;~,~ 4.5% 8. United States. 4.8 ~~ 4.5% 9. Qatar. 4.3 ;~ 4.0% ................ ~, ~,, 10. Philippines ............. 2.3 ~;~:~;~~;~::~ 2.2% !1. Oman. 2.2 ~;~ 2.!% ................ ~,~,~;~L~ 12. Jordan. 1.8 ~;~’~ 1.7% 13. Bangladesh ............ 1.4 ~::~ 1.3% 1.2% 14. Sri Lanka .............. 1.3 ~,~,, 15. Morocco .............. 1.0 1.0% 16. France ................ 0.9 ~, 0.9% 0.9% 17. Svria .................. 0.9 ;?~ ~;~ 0.7% 18. Germany .............. 0.8 ~;~,:~ 19. Swi~erland ............ 0.7 ~;~ 0.7% 20. Yemen ................ 0.7 0.7% Other ................. 9.9 9.3% Total 106.6 © TeleGeography, Inc. t998 National Traffic Balance MiTT 1995 1996 1997 Incoming 62.6 69.4 85.4 Outgoing 88.7 92.2 106.6 Surplus (Deficit) (26.1) (22.8) (21.2) Total Volume 151.3 161.5 192.0 Note: MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes of public switched traffic. Data based on bilhng point of traff=c. 162 © TeleGeography, Inc. 1998 © TeleGeography, Inc. 1998 TeleGeography 1999 Bangladesh Largest Telecommunications Routes, 1997 Destination Millions of Mill’s 1. india .................. 5.5 2, United Kingdom ........ 5.1 3. United States .......... 4.5 4. Saudi Arabia ........... 2.8 5. Singapore ............. 2.7 6. Rep. of Korea .......... 2.0 7. Hong Kong ............. 2.0 8. United ArabEmirates... 1.8 9. Malaysia .............. 1.7 10. Pakistan ............... 1.4 11. China ................. (}.9 12. Germany .............. 0.8 13. Japan ................. 0.8 14. Italy ................... 0.8 15. Canada ................ 0.7 16. France ................ 0.6 17. Thailand ............... 0.6 18. Australia .............. 0.5 Other ................. 11.9 Total 46.9 Percentage of Outgoing Traffic ,~ ~,~-~,-~’ 5.9% ~:~;,~?~-~- 3.5% ~: 1.8% ;~:~,~:~: 1.7% ’::.’::~’.~’:~ 1.4% .3O/o .2O/o ~~,~ 1.1% © TeleGeography, Inc. 1998 National Traffic Balance MiTT 1995 1996 1997 Incoming 122.1 129.2 187.0 Outgoing 33.0 38.3 46.9 Surplus (Deficit) 89.1 90.9 140.2 Total Volume 55.1 167.5 233.9 Note: MiFr is Minutes of Telecommumcat=ons Traffic. Data are m millions of minutes of public sw=tched traffic. © TeleGeography, Inc. 1998 165 TeleGeography 1999 © TeleGeography, Inc. 1998 ( Belarus Largest Telecommunications Routes, 1997 Destination Percentage of Outgoing Traffic ~i~ 15.3% 6. 7. 8. 9. 10. 11. Millions of MiTts Russia ................ 84.9 Ukraine ............... 22.7 Moldova .............. 2.1 Kazakhstan ............ 2.0 Armenia .............. 1.3 Azerbaijan ............ 1.0 Uzbekistan ............ 0.9 Georgia ............... 0.6 Kyrgyzstan ............ 0.2 Turkrnenistan .......... 0.2 Tajikistan ............. 0.2 Other ................. 32.4 Total 148.6 0.9% 0.6% !o.4% 0.2% 0.1% o.1% © TeleGeography, Inc. 1998 National Traffic Balance MiTT 1995 1996 1997 Incoming ,,, n.a.~ ........... n.a. 185.2,, Outgoing,,, 106.6,, 104.9 Surplus (Deficit) n.a. n.a. 36.5 Total Volume n.a. n.a. 333.8 Note: Mi’Fr ~s Minutes of Telecommumcations Traff=c. Data are in mdlions of mznutes of pubhc switched traffic. Data based on bdling point of traffic. Data for 1995 and 1996 include only traffic to other member states of the Commonwealth of Independent States. The =Other" category may include routes to non-CIS member states, mclud- mg an estimated 3.0 million minutes to Germany, that rank among the top destinat=ons for outgoing traffic. © TeleGeography, Inc. t998 164 © TeleGeography, Inc. 1998 TeleGeography 1999 Belgium Largest Telecommunications Routes, 1996 Destination Millions of Mi’n’s 1. France ................. 285.0 2. Netherlands ............ 253.4 3. Germany ............... 148.9 4. United Kingdom ......... 108.3 5. Italy ................... 65.1 6. United States ........... 45.1 7. Luxembourg ............ 43.8 8. Spain .................. 40.9 9. Switzerland ............ 26.2 10. Sweden ................ 14.4 !1. Portugal ............... 13.2 12. Greece ................ 12.3 13. Denmark ............... 11.9 14. Turkey ................. 11.4 15. Austria ................. 10.8 16. Morocco ............... 9.9 17. Poland ................. 9.9 18. Ireland ................. 7.4 19. Canada ................ 6.2 20. Russia ................. 5.9 Other .................. 98.4 Total 1,228.4 Percentage of Outgoing Traffic ............................................................................................. ~, ................... 20.6% ,,~,~#~:,:~ 2.1% ~,:, 1.2% ~Y 1.1% ~;;,~, 1.0% {~:: 1.0% 0.9% ~t~- 0.9% :~: 0.8~ ~::~ 0.6% :" 0.5% © TeleGeography, lnc. t998 National Traffic Balauce MiTT 1995 1996 1997 Incoming 1,.172.0 1,289.1 ¯ 1,420 Outgoing 1,105.7 1,226.4 1,340 Surplus (Deficit) 66.3 60.6 80 Total Volume 2,277.7 2,517.5 2,760 Note: M~FF is M=nutes of Telecommunications Traffic. Data are in milhons of minutes of pubhc swJtched traffic. Data =s based on billing point of traffic. © TeleGeography, Inc. 1998 165 TeleGeography 1999 © TeleGeography, Inc. 1998 Bolivia Largest Telecommunications Routes, 1997 Destination Millions of MiTrs 1. United States ........... 5.1 2. Argentina .............. 3.3 3. Brazil .................. 2.8 4. Chile .................. 2.6 5. Peru ................... 2.0 6. Germany ............... 0.6 7. Colombia ............... 0.5 8. Spain .................. 0.5 9. Mexico ................ 0.5 10. Paraguay .............. 0.4 11. Canada ................ 0.3 12. Italy ................... 0.3 13. Venezuela .............. 0.3 14. Ecuador ................ 0.3 15. Japan ................. 0.3 16. United Kingdom ......... 0.2 17. France ................. 0.2 18. Uruguay ............... 0.2 19. Switzerland ............ 0.2 20. Panama ................ 0.2 Other .................. 1.7 Total ~.7 Percentage of Outgoing Traffic ~:~ 2.2% ~i~!~i 1.8% ~":"~ 1.3% 1.3% ~)~,:. 1.3% ~’. ~’~ 1.3% ~?;;; 0.9% IR.: 0.9% !i::"~:~ 0.9% 1~:]] 0.9% ii; 0.9% © TeleGeography, Inc. t998 National Traffic Balance Mi]-I 1995 1996 1997 Incoming _ 49.2 53.9++ 69.3,‘ Outgoing 20.8 21.4 22.7 Sur~plus !Deficit) 28.4 32.5 46.6 Total Volume 70.0 75.3 92.0 Note: M=TT is Minutes of Telecommumcat~ons Traffic. Data are +n millions of minutes of public switched traff=c. © TeleGeography, Inc. t998 © TeleGeography, Inc. 1998 TeleGeography 1999 Botswana Largest Telecommunications Routes, 1996 Destination Percentage of Outgoing Traffic 3. ~!i:ii~i’! 4,3% 4. ~ili 2.3% 6. 7. 8. 9. 11. 13. Millions of MiTl’s South Africa ........... 21.7 Zimbabwe ............. 2.9 United Kingdom ........ 1.3 United States .......... 0.7 Zambia ............... 0.5 India ................. 0.4 Namibia .............. 0.3 Germany .............. 0.2 Swaziland ............. 0.2 Kenya ................ 0.1 Lesotho ............... 0.1 Malawi ............... 0.1 Canada ............... 0.1 Australia .............. 0.1 China ................. 0.1 Tanzania .............. 0.1 France ................ 0.1 Belgium ............... 0.1 Sri Lanka .............. 0.1 Other ................. 1.4 Total 30.8 ~il 1.6% 1.3% 1.0% 0.5% 0.5% 0.5% 0.5% 0.4% 0.4% 0.4% 0.3% O.3% 0.3% 0.2% O.2% © TeleGeography, Inc. 1998 National Traffic Balance MiTT 1994 1995 1996 Incoming 21.9 21.8 22.7 Outgoing 28.4 29.8 30.8 Surplus (Deficit) (6.5) ,(8) (8.1) Total Volume 50.3 51.6 53.6 Note: MiTT zs Minutes of Telecommumcatzons Traffic. Data are ~n milhons of minutes of public switched traffic. © TeleGeography, Inc. 1998 167 TeleGeography 1999 © TeleGeography, Inc. 1998 ( Brazil Largest Telecommunications Routes, 1997 Destination Millions of Mi3"i’s Percentage of Outgoing Traffic 2. Argentina .............. 35.9 ~>........... ,~, o,~,~,~,~,,~,~,~~>~:~,~, 7.8% 4 7% 3. Italy ................... 21.5 :~ ¯ 4. Germany ............... 19.1 4.2% 5. Po~ugal 18.5 ~;~,~ 4.0% 6. United Kingdom ......... 17.7 ~.~?~’~’~ 3.9% 7. France ................. 14.9 ;~ 2.9% 8. Japan ................. 13.3 9. Spain .................. 12.7 ~,~ 2.8% 10. Uruguay ............... 10.1 2.2% 11. Canada ................ 9.4 ~: 2.1% 12. S~o Tom~ & Principe .... 8.8 ~:~ 1.9% 13. Paraguay .............. 8.8 ~:~, 1.9% 14. Chile ................... 8.6 ~:~. 1.9% 15. Bolivia ................. 8.5 ~;~ 1.8% 16. Moldova ............... 7.8 ~;~ 1.7% 17. Swi~erland ............ 6.6 1.4% :’~; 1.3% 18. Mexico ................ 6.0 19. Guyana ................ 4.5 1.0% 20. Venezuela .............. 4.5 1.0% Total ~9.1 © TeleGeography, Inc. 1998 National Traffic Balance MiTT 1995 1996 1997 Incoming, 495.5 624.4 761.3 Outg0!~,g ,, 319.4 366.9 459.1 Surplus (Deficit) 176.1 ..... 257;5 302.2 Total Volume 814.8 991.3 1,220.4 Note: MiTF is Minutes of Telecommumcations Traffic. Data are in millions of minutes of public swztched traffic. Data based on billing pmnt of traffic. 168 © TeleGeography, Inc, t998 © TeleGeography, Inc. 1998 TeleGeography 1999 Brunei Largest Telecommunications Routes, 1997 Destination Millions of Mi’l-rs Percentage of Outgoing Traffic 1. Malaysia 12.3 ........................................................................................................... 34.6% E Singapore 4.1 ......... .............................. 11.6% 3. Philippines 3.0 :’~,:~:~’~ 8.4% 4. U~ited Ei~9~o~ ........ 3.0 ~:~~ 8.4% 5. I~o~si~ 1.5 ~;~:%~ 4.2% ............. 6. Th~il~ .............. 1.2 3.4% 7. ~ustr~li~ .............. 1.1 3.1% 8. I~i~ ................. 1.0 2.8% 9. Unite~ Ststes .......... 0.6 1.7% 10. Hon9 Eong ............ 0.4 1.1% 11. ~ew Zeala~ .......... 0.3 ~: 0.8% 12. J~p~n ................ 0.3 0.7% 13. C~ ............... 0.2 0.6% 14. ~et~er]~s ........... 0.2 0.5% 15. ~e~ny .............. 0.2 0.5% 16. ~ngl~s~ ........... 0.2 0.4% 17. ~ri L~nE~ ............. 0.1 18. C~i~ ................. 0.1 = 0.4% 19. T~i~ ............... 0.1 0.4% 20. France ................ 0.1 0.3% Total 35.6 © TeleGeography, Inc. t998 National Traffic Balance MiTT 1995 1996 1997 Incoming n.a. n.a. 22.0 Outgoing ¯ 30.7 35.2 ,~ ¯ 35.6 SurPlus (Deficit) n.a. ,,~.a. (13.6) Total Volume n.a. n.a. 57.6 Note: MiTI" is Mznutes of Telecommunications Traffic. Data are in m~lhons of minutes of pubhc switched traffic. © TeleGeography, Inc. t998 169 TeleGeography 1999 © TeleGeography, Inc. 1998 Canada Largest Telecommunications Routes, 1997 Destination Millions of MiTrs Percentage of Outgoing Traffic 2. United Kingdom ........ 160 ~ii 3.7% 3. Hong Kong ............. 75 6! 1.7% 4. France ................ 55 1.3% 5. Germany .............. 55 1.3% 6. Italy ................... 50 1.2% 7. Philippines ............. 35 !! 0.8% 8. India .................. 35 0.8% 9. Australia .............. 35 ii 0.8% 10. Mexico ................ 35 il 0.8% 11. Jamaica ............... 25 0.6% 12. Netherlands ........... 20 i 0.5% Other .................. 531 ~,~;~;,,,,,~,~:~,~,~!~,~;~ 12.4% Total 4286 © TeleGeography, Inc. 1998 National Traffic Balance MiTT 1995 1996 1997 Incoming, 3,895.8 ,, 4,313.3 4,635.1 0 utg o in g ,2,667.1, 3,519.8 4,286.3 Surplus (Deficit) 1,228.7 793.5 348.7 Total Volume 6,562.9 7,833.1 8,921.4 Note: MiTT =s Minutes of Telecornmumcatmns Traffic. Data are m millions of minutes of pubhc switched traffic. Data based on billing point of traffic. Route data rounded to the nearest five m~lhon m=nutes. U.S. route data include pub- hc switched and ISR traffic, but ISR =s not zncluded for other routes (e.g., to the U.K.). © TeleGeography, Inc. 1998 170 © TeleGeography, Inc. 1998 TeleGeography 1999 Chile Largest Telecommunications Routes, 1997 Destination Millions of MiTrs Percentage of Outgoing Traffic 1. United States ........... 75 ~,’,’~’,,’.,~’~,~’,~,’-~.,~;,,~,,,~’~,~" "~,~,,’~’,’~,’~,~,,~,~,~,,,:~,~,’~;~,.,,~ .......... ;~,,;:~.~,’~",’,", ........... 31.0% 2. Argentina 35 ~‘‘:‘~‘‘‘~‘~.‘‘‘‘~‘:~‘‘‘‘‘:;‘‘~,~‘~‘‘,,‘‘,~,~;~!~,~‘~!~:‘~‘‘‘~‘,‘‘‘ 14.5% 3. Brazil .................. 15 ~,;~;,~#~..................... 6.2% 4. Spain 13 ~,~,~;,~%~:;~, 5.4% 5. Peru 11 ~’~:,~~- 4.5% 6. Canada 8 ~:,~,~:~;~:~ 3.3% 7. Mexico 6 ......... 2.5% ................ 8. 6ermany ............... 5 2.1% 9. France ................. 5 ~:,,~,~:~ 2.1% 10. United Kingdom ......... 5 ~~ 2.1% 11. Ecuador ............... 5 ~,,~;~ 2.1% 12. Italy ................... 5 13. Bolivia ................. 4 ~,~ 1.7% 14. Colombia ............... 4 ~,~:~ 1.7% 15. Venezuela .............. 3 ~,: 1.2% Total © TeleGeography, Inc. 1998 National Traffic Balance MiTT 1995 1996 1997 Incom!ng n.a. n.a. n.a. Outgoing 136.9 173.8 242 Surplus (Deficit) n.a. n.a. n.a. Total Volume n.a. n.a. n.a. Note: MiTT ~s M~nutes of Telecommumcatzons Traff=c. Data are m m=lhons of minutes of pubhc switched traffic. © TeleGeography, Inc. t998 TeleGeo~raphy 1999 © TeleGeography, Inc. 1998 ( China Largest Telecommunications Routes, 1997 Destination Millions of Mi~’~s Percentage of Outgoing Traffic Z Taiwan 150 ,~,:,~:~:~ 9.2% 3. Japan 105 ~:~;:~’~ 6.4% 4. United States .......... 60 ~;~ 3.7% 5. Rep. of Korea .......... 55 3.4% 6. Macau ................ 35 2.1% 7. Singapore ............. 30 1.8% 8. Australia .............. 15 0.9% 9. Germany .............. 15 0.9% 10. Canada ............... 15 0.9% 11. United Kingdom ........ 12 0.7% 12. France ................ 9 0.6% 13. Thailand .............. 7 0.4% 14. Russia ................ 8 ~0.5% 15. Malaysia .............. 8 0.5% 16. Italy .................. 8 ~0.5% 17. Indonesia ............. 5 ~0.3% Other ................. 240 ~.~;~;~;~:;~ 14.7% Total 1,~2 © TeleGeography, Inc. 1998 National Traffic Balance MiTT 1995 1996 1997 ,Incoming n.a. n.a. n.a. Outgoing 1,~339,~.1, i,433.2 i,631~.~8 Surplus (Deficit) n.a. n.a. n.a. Total Volume n.a. n.a. n.a. Note: MiTT is Minutes of Telecornrnunzcations Traffic. Data are in millions of minutes of pubhc switched traffic. © TeleGeography, Inc. 1898 © TeleGeography. Inc. 1998 TeleGeography 1999 Colombia Largest Telecommunications Routes, 1996 Destination Millions of MiT~s Percentage of Outgoing Traffic 1. United States. 60.2 ~,’,’,~;~"~,~,’~ ,,~,~,’~.~,’~:~’~ ~,~,,~’~,’,~,,-,’~,~,;~;.,~,~, ~’,~,-’~,-~’,,,,,,,’~::~,~,,~,,~-’~ ...... ~,’~,!~’~,,~-,,~,,~,,~ 44.4% 2. Venezuela .............. 14.1 °~:q~,;;%""&~:.:~;;?~:~ 10.4% 3. Ecuador ................ 6.3 ~’~’4 ~’-’-’’ 4.6% 4. Spain .................. 5.7 ::’,,,",;~:’,~,’,,~ 4.2% 5. Mexico ................. 4.8 ~!i~’! 3.5% 6. Panama ................ 4.8 ~!~,;~,:~ 3.5% 7. Brazil .................. 2.9 i~ 2.1% 8. Italy .................... 2.8 ,~ 9. Peru ................... 2.8 ~,’,,"~i 2.1% 10. United Kingdom ......... 2.5 !~!~ 1.8% 11.6ermany ............... 2.5 1.8% 12. Argentina ............... 2.3 1.7% 13. France ................. 2.3 ~i 1.7% 14. Canada ................. 2.3 @ 1.7% 15. Chile ................... 1.7 ~,i~ 1"3% 16. Costa Rica .............. 1.6 i~;!i 1.2% 17. Dominican Republic ..... 1.1 :~:" 1.2% 18. Switzerland ............. 0.9 ~ii 0.8% 19. Puerto Rico ............. 0.9 0.7% !~ 0.7% 20. Japan .................. 0.5 Other .................. 12.5 ,,,~,,,~, i~:, Z,.,, ;,~44&‘ 9.2% Total 135.5 © TeleGeography, Inc. 1998 National Traffic Balance MiTT 1994 1995 1996 Incoming 302.8 351.5 384.2 Outgoing 120.3 127.3 135.5 Surplus (Deficit) 182.5 224.2 248.7 Total Volume 423.1 478.8 519.7 Note: Mi’(-r is Minutes of Telecommumcatiens Traffic. Data are in millions of minutes of public switched traffic. Data based on bilhng point of traffic. © TeleGeography, Inc. 1998 173 TeleGeography 1999 © TeleGeography, Inc. 1998 0 d -~ E C~ C~ o 174 © TeleGeographyo Inc. 1998 TeleGeography t 999 Costa Rica Largest Telecommunications Routes, 1997 Destination Millions of Mil-rs 1. United States ........... 25.2 2. NicaraGua ............. 10.1 3. Panama ................ 4.4 4. Mexico ................ 3.9 5. Guatemala ............. 3.7 6. El Salvador ............. 3.1 7. Honduras .............. 2.4 8. Colombia ............... 1.7 9. Canada ................ 1.0 10. Italy ................... 0.9 11. Spain .................. 0.8 12. Germany ............... 0.8 13. Venezuela .............. 0.6 14. Peru ................... 0.6 15. Argentina .............. 0.5 0.8% 16. Cuba .................. 0.5 0.7% 17. Dominican Republic ..... 0.5 ~i 0.7% 18. Brazil .................. 0.5 "" 0.7% 19. Ecuador ............... 0.4 >" 0.7% 20. Chile .................. 0.4 Other .................. 4.9 Total 66.9 0.7% © TeleGeography, Inc, 1998 National Traffic Balance MiTT 1995 1996 1997 Incoming n.a. 87.8 111.6 Outgoing 52.8 55.0 66.9 Surplus (Deficit) n.a. 32.8 44.7 Total Volume n.a. 142.8 178.5 Note: MiFF ~s M~nutes of Telecommunicatzons Traffic, Data are in m~lhons of m=nutes of pubhc sw=tched traffic Data based on bilhng point of traffic © TeleGeography, Inc 1998 h’-7 ......... TeleGeography 1999 © TeleGeography, Inc. 1998 Cyprus Largest Telecommunications Routes, 1997 Destination Millions of MiTTs Percentage of Outgoing Traffic Z Greece 34.6 ~:~~.~?:~~~N~:~~~?~ 22.4% 3. Lebanon 13.0 ~,t~a~:~%~:~,~,~, 8.4% 4. Russia 7.8 ~:~ ~:~:~4~}~ 5.1% ................ 5. Germany .............. 5.7 ~:~:~,~,~a,,’~;~,,:~,,,;: 3.7% 6. United States 5.1 ~>~:~"-~:~,;~ 3.3% 7. Romania 3.8 ,-;~:~ 2.4% .............. 8. [~t ................. a.7 g~:2i~ 2.4N ]. Italg .................. 2.8a~:~,~ ~.~% ]2. ~kraine ............... 2.8 13. Yugoslavia ............ 2.2 ~:~i,, 1.4% 15. France ................ 2.0 ~,~:~: 1.3% ]E Swi~erland ........... ].9 ~:~57~,~: ~.2% ;~’~ 1.0% 19. Netherlands ........... t.6 20. Canada ............... ].2 Other ................. 19.4 .?:;~‘:~;:‘~‘?)~‘2~}:~}s~)~?~?i,‘~:~?~;~i:~‘i‘~‘;}~‘~ 12.6% Total 154.4 © TeleGeography, Inc. 1998 National Traffic Balance MiTT 1995 1996 1997 Inco,ming ’ 87.3 .... 92.0 Outgoing 11,7;4 128.6, 154.4 Surplus (Deficit) (30.2),, (36.6),, (39.2) Total Volume 204.7 220.5 269.6 Note: MiTT is Minutes of Telecommumcations Traffic. Data are =n millions of mznutes of pubhc switched traffic. Data exclude traffic to Turkey and some traff=c to Greece. 176 © TeleGeography, Inc, 1998 © TeleGeography, Inc. 1998 TeleGeography 1999 Czech Republic Largest Telecommunications Routes, 1997 Destination Millions of Mil-rs Percentage of Outgoing Traffic 3. Austria ............... 20.4 ~:~,~ 6.7% 4. United Kingdom ........ 15.9 ~,~,~’~ 5.2% 5. Italy .................. 12.0 ~................. 3.9% 6. Poland ................ 11.4 ~:~, 3.7% 7. France 9.3 ~,,~,~,~ 3.0% 8. Ukraine 8.8 ,~-~>~~::~ 2.9% 9. United States 8.7 ............. 2.8% .......... 10. Netherlands ........... 7.6 ,~:~::~;~ 2.5% 11. Russia 7.4 2.4% ................ 12. Swi~erland 6.9 ~;,,~:~,~,~ 2.3% 13. Belgium ............... 4.3 1.4% 14. Hunga~ .............. 4.0 1.3% 15. Spain ................. 3.5 ’~ 1.1% 16. Sweden ............... 3.0 ~:~; 1.0% 17. Croatia ............... 2.7 ~,~: 0.9% 18. Bulgaria .............. 2.2 ~., 0.7% 19. Greece ............... 2.2 0.7% 20. Yugoslavia ............ 2.1 :~:~ 0.7% Total 3~.1 © TeleGeegraphy, Inc. 1998 National Traffic Balance Mil-I" 1995 1996 1997 Incom!ng 223.7 324.4 355.0 Outgoing 186.8 281.2 306.1 Surplus (Defic!t), 36.9 43.2 48.9 Total Volume 410.5 605.6 661.1 Note: Mi’FI" ~s M~nutes of Telecommunications Traffic. Data are ~n millions of minutes of public switched traffzc. Data based on b~lhng pmnt of traffic. Data for 1995 exclude traffic to and from the Slovak Repubhc. © TeleGeography, Inc. 1998 177 TeleGeography 1999 © TeleGeography, Inc. 1998 t Denmark Largest Telecommunications Routes, 1997 Destination Millions of MiTTs Percentage of Outgoing Traffic 1. Germany .............. 115 ~:,~4~,~;,~~- :~:~;;~:~~;~,:, 18.3% 3. United Kingdom ........ 61 ~;;~:;,~:~;~:~’~, ~’~’ 9.7% 5. United States .......... 28 ~,~g~;:,~;~]} 4.5% 6. Netherlands ........... 24 ~7~ 3.8% ~97"~{~]~ 3.7% 7. France ................ 8. Italy .................. 14 2.2% 9. Finland 13 ........... 2.1% 10. Belgium ............... 12 ~: 1.9% 11. Spain ................. 12 @~ 1.9% 12. Poland ................ 12 ~:.C~ 1,9% 13. Swi~erland ........... 11 ~Tt:~ 1.8% 14. Faroe Islands .......... 10 ~7~ 1.6% 15. Turkey ................ 8 ~,~, 1.3% Other 1~ ~"~",:: ," ~"’,~’~: ..... :,z ~,~,~,~, >,,~-z~-~, ,,,~;,~,~,-;;,,~,~,~,~,~, 19.4% © TeleGeography, Inc. 1998 National Traffic Balance MiTT 1995 1996 1997 IncQming ,, 551.0 600.0 ,, 682 ,, Outgoing 532.6 573.2 628 Surplus (Deficit), 18.4 26.8 54 Total Volume 1,083.6 1,173.2 1,310 Note: MiTT is Minutes of Telecommumcatmns Traff=c. Data are in millions of minutes of public swztched traffic. Data based on billing po=nt of traffic. Totals for 1997 are for all carriers; route data are for Tele Danmark only. Totals for 1996 are for Tele Danmark only © TeleGeography, Inc. t898 178 © TeleGeography, Inc. 1998 TeleGeography 1999 Dominican Republic Largest Telecommunications Routes, 1997 Destination Millions of Mi’l-rs Percentage of Outgoing Traffic 2. Spain .................. 3.5 ~,:~: 2.4% 3. Italy ................... 2.7 !! 1.9% 4. Canada ................ 2.6 i!~ 1.8% 5. Germany ............... 2.3 ~’~ 1.8% 6. Venezuela .............. 1.3 0.9% 7. Mexico ................ 1.2 0.9% 8. Cuba .................. 1.0 0.7% 9. Switzerland ............ 1.0 0.7% 10. Colombia ............... 1.0 0.7% 11. Haiti ................... 1.0 ! 0.7% 12, United Kingdom ......... 0.9 0.6% 13. Argentina .............. 0.8 0.5% 14. France ................. 0.7 0.5% 15. Panama ................ 0.7 0.5% 16. Netherlands Antilles ..... 0.6 0.5% 17. Netherlands ............ 0.5 0.4% 18. Chile .................. 0.5 0.3% 19. Jamaica ............... 0.4 0.3% 20. Costa Rica ............. 0.4 0.3% Other .................. 6.6 "’,~i’~,. 4.6% Total 142.0 © TeleGeography, Inc. 1998 National Traffic Balance MiTT 1995 1996 1997 Incoming 424.1 450.9 476.9 Outgoing 85.4 126.6 142.0 Surplus (Deficit) 338.7 324.3 334.9 Total Volume 509.4 577.5 618.9 Note: Mz’13" ~s Minutes of Telecommunicatmns Traffic. Outgoing totals for 1995 are for Codetel only. Data are =n rnilhons of mznutes of public switched traff=c. © TeleGeography, Inc. 1998 179 TeleGeography 1999 © TeleGeography, Inc. 1998 Salvador Largest Telecommunications Routes, 1997 Destination Percentage of Ongoing Traffic Millions of MiTi’s 1. United States ........... 12.8 2. Guatemala ............. 7.6 3. Costa Rica ............. 2.8 4. Honduras .............. 2.7 5. Mexico ................ 2.4 6. Nicaragua ............. 1.6 7. Panama ................ 0.9 8. Canada ................ 0.5 9. Spain .................. 0.3 10. Colombia ............... 0.3 11. Germany ............... 0.2 12. Venezuela .............. 0.2 13. Italy ................... 0.2 14. Chile ................... 0.1 15. Rep. of Korea ........... 0.t t6. Argentina .............. 0.1 17. France ................. 0.1 18. Brazil .................. 0.1 19. Japan ................. 0.1 20. United Kingdom ......... 0.1 Other .................. 1.0 Total 34.3 ~N 6.9% 2.5% 1.4% 0.9% 0.9% 0.6%° 0.5% 0.5% o.4% o.4% ~0.4% ";0.4% ~0.3% 0.3% © TeleGeography, Inc. t998 National Traffic Balance MiTT 1995 1996 1997 Incoming, n.a. 160.5 168.,2 0utg#ing ..... 64.! 28.6 34.3 Surplus (Deficit) n.a. 131.9 133.9 Total Volume n.a. 189.1 202.5 Note: M=TT is M~nutes of Telecommunications Traffic. Data are in milhons of minutes of pubhc switched traffic. Data for 1996 and 1997 based on billing point of traffic. 180 © TeleGeography, Inc, t998 © TeleGeographyo Inc. 1998 TeleGeography 1999 Estonia Largest Telecommunications Routes, 1997 Destination Millions of MiTrs Percentage of Outgoing Traffic 2_ Russia ................ 15.2 3. Sweden 5.3 -,~-,,-~>~,,,,,,~,,~:,,~,,~,~,~,’. 8.1% ,~-~,4,~, ~!t,:t~,’,,~,,, 6.3% 4. Latvia ................. 4.2 5. Germany. 4.1 ,;;,’.t;~t~,,&~,’,:t,’,~, 6.3% 6. Ukraine ............... 2.8 i~,il;,~i 4.3% 7. Lithuania .............. 2.5 :,~j~;~ 3.8% 8. United Kingdom 1.8 ,;~,~,- 2.6% 9. Denmark. 1.5 ;;~"-",~ 2.3% 10. United States .......... 1.3 ~",",,i’~! 2.0% #~ 1.6% 11. Belarus ............... 1.1 12. Netherlands ........... 0.9 @ 1.4% 13. Norway ............... 0.8 ~! 1.3% 14. Italy .................. 0.6 ;i!i O.9% 15. France ................ 0.6 16. Poland ................ 0.5 ;,i!0.8% 17. Belgium ............... 0.4 ~!: 0.7% 18. Switzerland ........... 0.4 i! 0.6% 19. Austria ............... 0.3 i:i 0.4% 20. Spain ................. 0.2 ;: 0.4% Other ................. 2.3 ;",~;i:;i!iii~i 3.5°/0 Total 66.3 © TeleGeography, Inc. t998 National Traffic Balance MiTT 1995 1996 1997 inc,~mi~g ,, 56.0 60.! 67.0 ,Outgoing 53.0 58.5 ,, 66.3 Surplus (Deficit),,, 3.0 1.6 0.7 Total Volume i09.0 118.6 133.3 Note: Mi]-I =s Minutes of Telecommunicatmns Traffic. Data are =n milhons of m~nutes of public switched traffic. Data based on bdling point of traff=c © TeleGeography, Inc. 1998 181 TeleGeography 1999 © TeleGeography, Inc. 1998 Finland Largest Telecommunications Routes, 1997 Destination Millions of MiTrs Percentage of Outgoing Traffic 1. Sweden ............... 118.9 2. Germany. 33.5 ~’,"~~~ii 9.0% 3, United Kingdom ........ 26.0 4, Russia ................ 26.0 ~,~,~,~ 7.0% 5. Estonia ............... 22.3 6. United States .......... 14.9 7. Noway ............... 14.9 4.0% 8. Denmark .............. 11.2 3.0% 9. France ................ 11.2 3.0% 10. Netherlands ........... 7.4 ~:~:~ 2.0% Other ................. 81.8 Tetal 371.7 © TeleGeography, Inc. 1998 National Traffic Balance MiTT 1995 1996 1997 Inc£miln. g 345.0 n.a. n.a. Outgoing 315.4 332.0 371.7 Surplus (Deficit) 29.6 n.a. n.a. Total Volume 660.4 n.a. n.a. Note: MiFF zs M=nutes of Telecommun[cat=ons Traffic. Data are m mllhons of minutes of pubhc sw=tched traffic. 182 © TeleGeography, Inc, 1998 © TeleGeography, Inc. 1998 TeleGeography 1999 France Largest Telecommunications Routes, 1997 Destination Millions of Mi~-I’s Percentage of Outgoing Traffic 1. Germany. 412 ~,~‘~‘~‘‘~‘~‘:‘~‘~‘~‘~,~,~‘~:~‘~‘ 11.6% Z United Kingdom 386 +~,~:~:~~:~,.’;;~:~}:~~;>,~:~:~":-,,~;~;"~;~;~ 10.9% 3. Belgium 304 ~,4:;~’t-~&:~:~:;~,~:~;~;~ 8.6% 4. Italy .................. 298 ~,~:,~:~:~z~:~,,~,:~,,~ 8.4% 5. Swi~erland ........... 246 6. Spain ................. ~3 "~=~"~ ...................... ........ ,~,, ~7~,, 6.6% 7. United States .......... 201 8. Po~ugal .............. 153 ;:,,,~;~@:~::~,~,~ 4.3Y~ 9. Morocco .............. 125 !0. Netherlands ........... 122 11. Algeria ............... 100 ~,~4~;~ 28~ ~, ,::~:~ 4~;~ 1.8% 12. Tunisia ............... 66 13. Canada ............... 50 ~’:~ 1.4% 14. Turkey ................ 45 :,?::;~ 1.3% 15. Poland ................ 35 ~,,~’ 1.0% l& Sweden ............... 32 ~~’~; 0.9% 17. Denmark .............. 25 :~~ 0.7% 18. Austria ............... 25 ?~ 0.7% 19. Luxembourg ........... 25 ~,~-~ 0.7% 20. 6reeoe ............... 24 ;~~:~~ 0.7% Tetal 3,545 © TeleGeography, Inc. t998 National Traffic Balance MiFI" 1995 1996 1997 Incoming 2,958.9 3,283 3,609 0 utg o in g 2,804.6 3,273 3,545 Surplus (Deficit) 154.3 10 64 Total Volume 5,763.5 6,556 7,154 Note: M~’Fr =s Minutes of Telecommumcat~ons Traffic. Data are in mdhons of minutes of public switched traff=c The outgoing total for 1995 excludes an estimated 150 million minutes of cross-border traffic. © TeleGeogra phy, Inc. 1998 185 TeleGeography 1999 © TeleGeography, Inc. 1998 French Polynesia Largest Telecommunications Routes, 1997 Destination Millions of Mi’n’s Percentage of Outgoing Traffic 1. France 5.7 ,~,~,~~~;,,,~!~,,-,~,,,,~.~,,~,~-~.,-~:~:~, 63.1% 2. United States 0.9 ~t~ 9.5% 3. New Caledonia ........ 0.6 6.8% 4. New Zealand .......... 0.3 2.8% 5. Australia .............. 0.2 2.7% 6. Cooklslands .......... 0.2 2.0% 7. Japan ................ 0.1 1.1% 8. Italy .................. 0.1 1.0% 9. Reunion ............... 0.1 0.6% 10. United Kingdom ........ 0.1 0.6% Other ................. 0.9 ~,~,;~,~ ~.,~o Total 9.1 © TeleGeography, Inc. t998 National Traffic Balance MiTT 1995 1996 1997 Incomi,ng ,, n,a. ~, n.a." ¯ n.a. Outgo,!ng ,, 7.6 7.9 ...... 9.!, Surplus (Deficit) ..... n.a., ...... n.a. n.#.~ Total Volume n,a. n.a. n.a. Note: Mi’Fr is Minutes of Telecommunications Traffic. Data are in milhons of minutes of pubhc switched traffic. 184 © TeleGeography, Inc. 1998 © TeleGeography, Inc. 1998 TeleGeoBraphy 1999 Germany Largest Telecommunications Routes, 1997 Destination Millions of MiTrs 1. Austria ............... 384.9 2. Turkey ................ 377.4 3. Italy .................. 368.5 4. Poland ................ 353.7 5. Switzerland ........... 339,9 6. France ................ 331.8 7. Netherlands ........... 327.3 8. United Kingdom ........ 326.9 9. United States .......... 319.2 10. Spain ................. 209.4 11, Greece ............... 132.1 12. Belgium ............... 125.4 13. Yugoslavia ............ 101.4 14. Denmark .............. 97.1 15. Croatia ............... 94.8 16. Czech Republic ........ 94.2 17. Hungary .............. 72.3 18. Portugal .............. 67.3 19. Romania .............. 59.4 20. Sweden ............... 50.6 Other ................ 1,099.5 Total 5,333.1 Percentage of Outgoing Traffic i 6.11 I~ 2.4% ,,,~,~¢~,~ 1.9% 1.81 L,~,, 1.4% ~,¢i~, 1.3% i’:;~]i: 1.1% © TeleGeography, tnc, t998 National Traffic Balance MiTT 1995 1996 1997 Incoming 4,~2! 5 n.a. 5,206.0 Outgoing 5,~238 ,,5,!00 5,333.! Surplus (Deficit) (1,023) n.a. (127.1) Total Volume 9,453 n.a. 10,539.1 Note: MFn =s Minutes of Telecommumcations Traff=c. Data are zn milhons of minutes of public sw=tched traffic. Data for 1997 are based on billing point of traffic; data for 1995 and 1996 are based on originating point. Route data exclude some cross-border traffic to nelghbonng countries ©TeleGeography, Inc 1998 185 TeleGeography 1999 © TeleGeography, Inc. 1998 ( Greece Largest Telecommunications Routes, 1997 Destination Millions of MiTrs 1. Germany .............. 98.3 2_ United Kingdom ........ 82.3 3. Italy .................. 5t.4 4. United States .......... 35.5 5. France ................ 25.4 8. Albania ............... 23.5 7. Cyprus ................ 22.8 8. Bulgaria .............. 21.4 9. Romania .............. 15.7 10. Netherlands ........... 13.2 11. Belgium ............... 13.1 12. Canada ............... 12.4 13. Switzerland ........... 11.2 14. Sweden ............... 10.4 15. Russia ................ 9.4 16. Yugoslavia ............ 9.4 17. Ukraine ............... 9.0 18. Turkey ................ 8.7 19. Australia .............. 8.6 20. Austria ............... 8.2 Other ................. 103.9 Total 593.7 Percentage of Outgoing Traffic 4.3’/o 4.0% ~il 3-6% 2.6% -~,,--,>% ,, 2.2% ~Z~i\ 1.5% #ii 1.5% © TeleGeography, tnc, t998 National Traffic Balance MiTT 1995 1996 1997 Incoming 505.4 557.3 634.6 Outgoing 467.9 515.6 593.7 Surplus (Deficit) 37.5 41.7 40.9 Total Volume 973.3 1,072.8 1,228.3 Note: MiTT is M=nutes of Telecommunicatmns Traffic. Data are in milhons of minutes of pubhc sw=tched traff=c. © TeleGeography, Inc. t998 © TeleGeography, Inc. 1998 TeleGeography 1999 Guyana Largest Telecommunications Routes, 1997 Destination Millions of Mil’rs Percentage of Outgoing Traffic 1. United States ........... 12.5 2. Ca n a d a ................ 3.4 3. Trinidad &Tobago ...... 1.8 ~,~:,,:~, ,, 7.3~ 4. United Kingdom ......... 1.2 ~i~:i 4.9% 5. Barbados .............. 1.1 ~:,~:!",,,,’;:~ 4.7% ’-~"~ 1.6% 6. Suriname .............. 0.4 7. Antigua & Barbuda ...... 0.4 ~!i 1.5% 8. Jamaica ............... 0.3 1.3% 9. St. Lucia ............... 0.3 ~, 1.1% 10. French Guiana .......... 0.3 ii 1.1% 0th e r .................. 2.6 ~;~,~i~;~~,,,,;i#~,,,,?,~ 10.8% Total 24.1 © TeleGeography, Inc. t998 National Traffic Balance MiTT 1995 1996 1997 Incoming 139.7 162.8 142.4 Outgoing 20.6 29.8 24.1 Surplus (Deficit) 119.1 133.1 118.2 Total Volume 160.2 192.6 166.5 Note: MiTT is M~nutes of Telecommunicat=ons Traffic Data are in mdlions of m=nutes of pubhc switched traffic. TeleGeography, tnc, 1998 187 TeleGeography 1999 © TeleGeographyo Inc. 1998 Hong Kong Largest Telecommunications Routes, FY 1997/98 Destination Millions of Mi’l’rs Percentage of Outgoing Traffic Z United States .......... 68.7 4.0% 3. Philippines ............ 68.7 4.0% 4. Taiwan ............... 51.5 3.0% 5. Japan ................ 51.5 3.0% 6. United Kingdom ........ 51.5 3.0% 7. Macau ................ 51.5 ~i 3.0% 8. Singapore ............. 51.5 3.0% 9. Canada ............... 34.4 ~! 2.0% 10. Australia .............. 34.4 ~i 2.0% Other ................. 189.0 ~,,,~!~!~:~ 11.0% Total 1,718.0 © TeleGeography, Inc. t998 National Traffic Balance MiTT FY 1995/96 FY 1996/97 FY 1997/98 incoming 1,598.3 1,940.8 2,100.3 0 utg o i n g ¯ 1,691.8 1,738.6 1,718.0. Surplus (Deficit) (93.5) 202.2 382.2 Total Volume 3,290.2 3,679.4 3,818.2 Note: MiTF ~s Minutes of Telecommumcations Traffic. Data are in m~lhons of mznutes of public switched traffic. Data based on billing point of traffic. Fiscal year ends 31 March. 188 © TeleGeography, Inc. t998 © TeleGeography, Inc. 1998 TeleGeography 1999 Hungary Largest Telecommunications Routes, 1997 Destination Millions of MiTrs 1. Germany .............. 71.0 Z Austria ............... 35.8 3. Romania .............. 17.3 4, United Kingdom ........ 16.9 5. Italy .................. 16.0 6. United States .......... 14.8 7. France ................ 10.2 8. Netherlands ........... 8.0 9. Yugoslavia ............ 7.9 10. Switzerland ........... 7.8 11. Russia ................ 7.6 12. Slovak Republic ........ 7.2 13. Ukraine ............... 6.1 14. Sweden ............... 4.6 15. Poland ................ 4.6 16. Belgium ............... 4.6 17. Czech Republic ........ 4.1 18. Croatia ............... 4.0 19. Greece ............... 3.2 20. Canada ............... 3.0 Other ................. 32.3 Total 287.1 © TeleGeography, lnc. 1998 National Traffic Balance Mi-n 1995 1996 1997 Incoming 243.7 n.a. 324.6 O#tgoing ....... 247.5 265 287.1 Surplus (Deficit) (24-7) n:a.,, 37.5 Total Volume 448.5 n.a. 611.7 Note; MiTr ~s M~nutes of Telecommunications Traffic. Data are in mdlions of minutes of public swztched traffzc. Data based on b=llmg point of traffic. © TeleGeography, Inc. 1998 ......... TeleGeography 1999 © TeleGeography, Inc. 1998 ( Iceland Largest Telecommunications Routes, 1996 Destination Millions of MiTTs Percentage of Outgoing Traffic 1. United States ........... 7.2 2. Denmark ............... 5.5~~,~~:~:~,~’~......................................................... 16.9% 3. United Kingdom ......... 3.5~.~~~~~"~’~" ...... ~~,.,~. 10.8% 4. Sweden. 3.1 ............... ~;~;~;,~~~ 9.5% 5. Noway ................ 3.0 ,,~,,~~, ~%~" 9.2% 6. Germany ............... 2.5 2.5% 7. France 0.8 ................. 8. Netherlands 0.8 ~: 2.5% ............ 9. Spain .................. 0.6 ~) 1.8~o 10. Faroe Islands ........... 0.6 ~) 1.8% 11. Canada ................ 0.5 1.5% 12. Italy ................... 0.5 1.5% 13. Belgium ................ 0.4 1.2% 14. Finland ................. 0.4 ~) 1.2% 15. Swi~erland ............ 0.3 0.9% 16. Luxembourg ............ 0.2 0.6% 17. Austria ................. 0.2 0.6% 18. Russia ................. 0.2 0.6% 19. Po~ugal ............... 0.2 0.6% 20. Poland ................. 0.2 0.6% Other .................. 1.9 Total ~.5 © TeleGeography, Inc. 1998 National Traffic Balance Mill" 1994 1995 1996 Incoming 25.5 28.4 32.0 Outgoing, 26.0 28.9 .... 32.5 Surplus (Deficit). (~].4) (0.6) (0.5) Total Volume 51.5 57.3 64.5 Note: Mi’lT ~s Minutes of Telecommumcations Traffic. Data are in mdhons of minutes of public switched traffic. Data based on bilhng point of traffic. © TeleGeography, Inc. 1998 © TeleGeography, Inc. 1998 TeleGeography 1999 Largest Telecommunications Routes, FY 1997198 India Destination Millions of MiTrs 1. Saudi Arabia .......... 85.9 2_ United States .......... 52.1 3. United Kingdom ........ 37.6 4. United Arab Emirates... 34.1 5. Singapore ............. 18.6 6. Oman ................. 14.1 7. Germany .............. 14.0 8. Canada ............... 11.1 9. Hong Kong ............ 9.2 10. Kuwait ................ 9.2 11. Japan ................ 8.3 12. Australia .............. 7.2 13. France ................ 6.7 14. Italy .................. 6.t 15. Sri Lanka .............. 6.0 16. Bangladesh ........... 5.6 17. Malaysia .............. 5.5 18. Philippines ............ 4,8 19, Qatar ................. 4.2 20. Pakistan .............. 4.1 Other ................. 76.1 Total 420.5 Percentage of Outgoing Traffic 4.4 go ~:~;~,~:>~;~ 2.2% 2.2% ~~2%~ 1.7% ;;~}! 1.4% ~::;~; 1.4% :,~:~; 1.3% ~’~; 1.0% © TeleGeography, Inc. t998 National Traffic Balance MiTT FY 1995/96 FY 1996/97 FY 1997/98 Incoming 805.4 1000.0 1,256.6 Outgoing 341.4 384.2 420.5 Surplus (Deficit) 464.0 615.8 836.1 Total Volume 1,146.8 1384.2 1,677.1 Note: Mzl-r ~s Minutes of Telecommunications Traffic. Data are m mtlions of minutes of public switched traffic. Data based on billing point of traffzc, and route traff=c and totals exclude cross-border traffic to Bangladesh, Nepal and Pakistan. © TeleGeography, inc. 1998 191 TeleGeography 1999 © TeleGeographyo Inc. 1998 ( Indonesia Largest Telecommunications Routes, 1997 Destination Millions of MiTrs Percentage of Outgoing Traffic 1. Singapore ............. 70.8 2_ Australia. 28.0 ............. ~~ 9.4% 3. United States. 26.2 ~~,~ 8.8% 4. Japan ................. 24.5 ~~~ 8.2% Malaysia .............. 23.0 ~;4~~’~ -, ;~:~,~ 7.7% Hong Kong ............. 15.3 Taiwan ................ 10.7 @~t~,~,~ 3.6% Rep. of Korea .......... 10.4 ~,~,~,~-~ 3.5% United Kingdom. 9.4 ~;~ 3.1% Germany .............. 6.6 2.2% ............. e.4 =.=% China ................. 6.4 2.2% Saudi Arabia ........... 6.1 2.0% Netherlands ........... 5.4 1.8% Canada ................ 4.9 1.6% India .................. 4.5 @~ 1.5% Thailand ............... 4.4 ~, 1.5% France ................ 3.8 1.3% Italy ................... 2.7 0.9% Swi~erland ............ 1.6 0.5% Other. 26.8 ........................................... 9.0% Total ~98.1 6. 7. 8. 9, 11, 12. 13. 15. 16. 17, 19. 20. © TeleGeography, Inc, t998 National Traffic Balance MiTT 1995 1996 1997 Inc0mi~g 294:0 ~, 356.4 Outgoing 182.5 280.2 351.6 Surplus (De~ic!t) ,~, 77.4 ....... ¯ 76.2, ..... 104.4 Total Volume 510.6 636.6 807.6 Note: Mi~-I" is Mznutes of Telecommunications Traffic. Data are in millions of minutes of public switched traff=c Data based on billing point of traffic. Totals include traff=c for PT Indosat and Satehndo; route data are for PT Indosat only. 192 © TeleGeography, Inc, t998 © TeleGeography, Inc. 1998 TeleGeography 1999 lran Largest Telecommunications Routes, 1997 Destination Millions of Mi~’l’s Percentage of Outgoing Traffic 3. United States.. 14.6 ~;~~~:-~,~’~~’~,~,~~:~ 9.1% 4. Pakistan ............... 8.8 United Kingdom ........ 8.8 ~t~~t,:~:t~ 5.4% Turkey. 8.7 ~4~~#,~~:,~::~ 5.4% ~#%;~5 3.9% Kuwait ................ 6.2 Canada ................ 6.1 Saudi Arabia.. 5.1 ~;~,~,~ 3.2% Sweden ............... 5.0 &4~{~&~ 3.1% France ................ 4.1 ~:~ 2.5% Azerbaijan. 3.4 ............ 2.1% Italy.. 3.3 ~,~’~" 2.0% ................. Netherlands. 2.1 1.3% .......... Russia ................ 2.0 1.3% Japan.. 2.0 .... ~,~,,,, 1.2% Austria ................ 1.8 ~:: 1.1% Swi~erland ............ 1.7 ~~~ 1.1% Qatar. 1.6 ................ 1.0% Armenia. 1.5 ;,~4,~ 0.9% .............. Total 160.7 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16, 17. 18. 19. 20. © Te~eGeograph¥, Inc. t998 National Traffic Balance MiTT 1995 1996 1997 Incoming,iI ........ 199 n.a~, 130.2 Outgoing 210.4 183.2 .... 160.7 Surplus (Deficit) (11) n.a. ,(30.5) Total Volume 409 n.a. 290.9 Note: Mi’lq" is Minutes of Telecommumcat~ons Traffic. Data are in millions of minutes of pubhc switched traffic. © TeleGeography, Inc. 1998 195 TeleGeography 1999 © TeleGeographyo Inc. 1998 Ireland Largest Telecommunications Routes, FY 1997198 Destination Millions of Mi~rs Percentage of Outgoing Traffic 1. United Kingdom ........ 515 2_ United States .......... 55 3. Germany .............. 20 2.9% 4. France ................ 20 2.9% 5. Netherlands ........... 17 2.4% 6. Italy .................. 9 1.3% 7. Spain ................. 7 1.0% 8. Canada ............... 7 1.0% 9. Belgium ............... 6 0.9% 10. Australia .............. 5 0.7% Other ................ 4.9% Total ~5 © TeleGeography, Inc, 1998 National Traffic Balance MiTT FY 1995/96 FY 1996/97 FY 1997/98 incoming ¯ n.a. n.a. n.a. Outgoing 407 580 695 Surplus (Deficit) n.a. n.a. n.a. Total Volume n.a. n.a. n.a. Note: Mi]-I is Minutes of Telecommumcations Traffic. Data are ~n milhons of minutes of public swztched traffic. Data for the top four routes are rounded to the nearest five million. Totals for FY 1995/96 and 1996/97 exclude some cross- border traffic to Northern Ireland. Fiscal year ends March 31. © TeleGeography, Inc. 1998 194 © TeleGeography, Inc. 1998 TeleGeography 1999 Israel Largest Telecommunications Routes, 1996 Destination Millions of Mi’Frs Percentage of O~going Traffic 2. United Kingdom 29.0 ~,~:,,~A~,;,,,,’~,;,,,, 9.1% 3. Canada. 23.7 ~,-x~::,:s~,~:s~,,~~ 7.4% 4. France ................. ~8.4 ~:,,~;~:~ 5.8% 5. Jordan 17.6 ~,~;’;~:,,~,~,’~’~,,~’, 5.5% 6. Germany.. 16.5 ~A~;~::~,~,~ 5.2% 7. Italy ................... 12.0 ~a~;~’~,.~: 3.8% ~,: 3.5% & ~ussia ................. J.] 9. Ukraine 7.1 :~,~ 2.2% ................ 10. Netherlands ............ 6.5 ~,~; 2.0% 1]. Turke7 ................. &2 1.~% 1.8% 12. Australia ............... 5.9 ~a. S~i~erlan6 5.8 ~.8% ]4. ~l~ium ................ &8 ~5. S~ain .................. a.7 ~;~ ~-~"~ ~6. N~mania ............... a.6 ;% ~.~% ]7. Sauth Bfrica ............ a.4 ~a. [~t .................. ~.a ~. Sweden ................ 2.4 20. Bustria ................ 23 ;~;: &7% Other .................. 34.4 Total 319.7 © TeleGeography, Inc. 1998 National Traffic Balance MiTE 1994 1995 1996 Incoming n.a. 345.6 468.1 Outgoing 213.0 252.3 319.7 Sur~plus (Deficit) n.a. 93.3 ,!48.4 Total Volume n.a. 597.9 787.9 Note: MiTT is M~nutes of Telecommunications Traffic. Data are in millions of minutes of public switched traffic. Data based on billing point of traffic. © TeleGeography, Inc. t998 195 TeleGeo~raphy 1999 © TeleGeogrephy, Inc. 1998 ( Italy Largest Telecommunications Routes, 1997 Destination Millions of Mi’n’s Percentage of Outgoing Traffic 1. Germany. 340.3 ~~!,~~,~’~:~,,,~ 14.5% ............. Z France .266.3 ,~’ .............. ~~,~’ 11.3% 3. United States 251.4 :~,~, ~, ,:~?~~ 10.7% 4. Switzerland ........... 198.1 ~-~,,~:~i~,~~~,~:~ i~ 8.4% 5. United Kingdom ........ 196.8 ~~ 8.4% 3.9% 6. Spain ................. 91.2 7. Belgium ............... 64.7 2.8% 8. Austria ............... 60.8 2.6% 9. Netherlands ........... 54.4 :~;~ 2.3% 10. Romania .............. 46.0 2.0% 11. Greece ............... 42.6 ~i~i:::~ 1.8% 12. Morocco .............. 37.6 ~:: 1.6% 13. Croatia ............... 32.9 14. Tunisia ............... 28.0 ~i 1.2% 15. Albania ............... 27.2 1.2% 16. Canada ............... 26.8 ,~ 1.1% 17. Chile ................. 23.7 ,~,:~ 1.0% 18. Russia ................ 22.5 ~I 1.0% 19. Sweden ............... 20.5 ii~ii! 0.90/o 20. Australia .............. 19.9 ~!:~~ii 0.80/o Total 2~351.9 © TeleGeography, Inc. 1998 National Traffic Balance MiTT 1995 1996 1997 l~,€o~in9 1,999.8 2,253.5 2,475.1 0 utg o!ng 1,908.2, 2,124.0 ,2,35! .9 Surplus (Deficit) ..... 91.6 129.5 .... 123.2 Total Volume 3,908.1 4,377.4 4,827.0 Note: M~l-r is Minutes of Telecommunications Traff=c. Data are in millions of minutes of pubhc sw=tched traffic. Data exclude some traffzc to France, Poland, SIovenia, and Switzerland. 196 © TeleGeography, Inc. 1998 © TeleGeographyo Inc. 1998 TeleGeography 1999 Japan Largest Telecommunications Routes, FY 1997198 Destination Millions of MiTts Percentage of Outgoing Traffic Z China ................. 229.5 ~t~:~~~,~ 13.6% 3. Rep. of Korea. 156.0 !~#,,~,’~-’:,~,.-’,~,,"~,#;-~~,~4~;~,°~ 9.2% 4. Philippines. 154.3 .......... ,~,~,,~,~,-~ .......... ~’~,~, 9.1% Taiwan. 88.3 ............... ~:~5~a~:5~ 5.2% Thailand. 72.4 ~,~,:,~,~,~,~,~a~,,,,, 4.3% Hong Kong ............. 59.5;,r;,:~;~:~;~?:~":’~,?;?"~;~; 3.5% United Kingdom. 52.6 Brazil ................. 52.4 ............ Australia.. 38.4 ~:~5~;~ 2.3% ............ Indonesia .............. ~4.7 a~i~ ................ ~a.~ ~i~ ~.a% c~d~ ................ ~.~ ~i%~ ~.7% France ................ 22.5 ~}~: ~.~% ~u .................. ~s.~ ?~:~’~ ~.~% Italg ................... ~&7 0.8% India .................. ]2.4 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. © TeleGeography, Inc. 1998 National Traffic Balance MiTT FY 1995/96 FY 1996/97 FY 1997/98 Incoming 1,320.8 1,519.1 1,635.0 Outgoing 1,631.3 ,.1,710.6 1,771.7 Surplus (Deficit) (310.5) (191.5) .(136.7) Total Volume 2,952.1 3,229.7 3,406.7 Note: MiTr is Minutes of Telecommumcat=ons Traffic. Data are in m~ll~ons of minutes of public sw=tched traffic Route data =nclude only Internat=onal Direct D~al (IDD) calls, while totals include operator assisted calls as well. F~scal year ends 31 March. © TeleGeography, Inc. 1998 TeleGeography 1999 © TeleGeography, Inc. 1998 ( Jordan Largest Telecommunications Routes, 1997 Destination Millions of Mi’l-rs Percentage of Outgoing Traffic 1. Israel.. 13.6 ~~~~~ 14.8% ............... 2_ Saudi Arabia. 12.1 13.2% 3. Egypt. 9.3 ~::%%~!,’,~~~:~,~-~~’ 10.1% ................ 4. Syria .................. 8.1 ~~?i:’? 8.8% United Arab Emirates... 6.4 United States .......... 5.4~~~:~:~>~’>~ 5.9% Iraq ................... 4.8 5.3% Lebanon. 3.7 ~:;~:;~ :, 4.1% .............. United Kingdom ........ 3.g Kuwait. 3.4 ~.,~,~g~; 3.7% Germany .............. 1.8 1.9% ,,~,,,,#:::;~, 1.8% Qatar ................. 1.7 ~# 1.4% Italy ................... 1.3 ~:~’’ 1.3% France ................ 1.2 0man ................. 1.0 ~::~: 1.1% Yemen ................ 1.B 1.1% ~:~:;~g" 1.0% Canada ................ 0.9 Turkey ................ 0.9 }~ 1.0% Bahrain. 0.9 ........ 1.0% .............. Swi~erland ............ E5 Other ................. 10.3 Tetal 91.9 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. © TeleGeography, Inc. 1998 National Traffic Balance MiTT 1995 1996 1997 ,Incoming 118.0 133.1 145.0 Outgoing ......... 7!,:7 ..... 74.6 .... 9!.9 Surplus (Deficit) ........... 46:3 ,, 58.5 ........... 53.! Total Volume 189.7 207.7 236.9 Note: M~’FI" zs Minutes of Telecommumcations Traffic, Data are m milhons of mznutes of pubhc switched traffic. Jordan-Israel route data include traffic to the Occupied Territories (West Bank) 198 © TeleGeography, Inc. 1898 © TeleGeography, Inc. 1998 TeleGeography 1999 Kazakhstan Largest Telecommunications Routes, 1997 Destination Millions of MiTrs Percentage of Outgoing Traffic 1. Russia ................ 68.9 ~%!r!~!~‘:~1~~?~f~!~‘~‘~‘~ 60.1% Z Uzbekistan ............ 7.8 6.8% 3. Ukraine ............... 5.9 ~} 5.2% 4. Kyrgyzstan 5.8 ~:~:~ 5.1% 5. Belarus ............... 2.0 ~,~ 1.7% 6. Azerbaijan ............ 1.4 1.2% 7. Tajikistan ............. 1.2 1.1% 8. Turkmenistan .......... 1.1 }~ 1.0% 9. Armenia .............. 1.0 [~ 0.9% 10. Georgia ............... 0.5 0.5% 11. Moldova .............. 0.4 0.3% Other ................. 18.4 ’~,~,’:~:~’~,~’~’,,,,,~’ 16.1% Total 114.7 @ TeleGeography, Inc, 1998 National Traffic Balance MiTI" 1995 1996 1997 incoming n.a. n.a. n.a. Outgoing 111.1 i 02.5 114.7 SurPlus (Deficit) n.a. n.a. n.a. Total Volume n.a. n.a. n.a. Note: MiTF is Minutes of Telecornmumcatzons Traffic. Data are in mdbons of m~nutes of public switched traffzc. Totals for 1995 and 1996 include only traffic to other member states of the Commonwealth of Independent States The "Dther" category may include routes to non-CIS member states, including an estimated 3.5 rnilhon minutes to Germany and 1.6 million mznutes to the U.S., that rank among the top destinatzons for outgoing traffic. © TeleGeography, Inc. 1998 199 TeleGeography 1999 © TeleGeography, Inc. 1998 Republic of Korea Largest Telecommunications Routes, 1997 Destination Millions of MiTrs Percentage of Outgoing Traffic 1. United States .......... 205 ~~~~~ 23.2% ................. ~,~ ~~ 4. Hong Kong ............ 35 5. Australia .............. 20 2.3% 6. Indonesia ............. 17 1.9% 7. Canada ............... 16 1.8% 8. Germany .............. 15 1.7% 9. United Kingdom ........ 15 1.7% 10. Philippines ............ 14 1.6% 11. Taiwan ............... 12 1.4% 12. Thailand .............. 12 1.4% 13. Singapore ............. 11 1.2% 14. Vietnam ............... 10 1.1% 15. France ................ 7 ~,~ 0.8% Total © TeleGeography, Inc. 1998 National Traffic Balance MiTT 1995 1996 Inco_m,!ng ........ 6,72 740.6 Outgoing 557 699.3 Surplus (Deficit) 115 ~!.3 Total Volume 1,229 1,439.9 1997 n.a. 885 Note: Mil-r is Minutes of Telecommunications Traffic. Data are in milhons of minutes of public switched traff=c. Route data for top five routes are rounded to the nearest five million m=nutes. 200 © TeleGeography, Inc. 1998 © TeleGeography, Inc. 1998 TeleGeography 1999 Kuwait Largest Telecommunications Routes, 1997 Destination Millions of Mitts Percentage of O~going Traffic 3. Saudi Arabia ........... 16.2 ~~;~‘~‘‘‘~‘~‘‘~;‘‘:~;‘‘‘:~::~‘:~‘‘~‘‘~:~‘~:~‘;~,‘~,~.~+~‘~:~.~‘°~‘‘‘~:~‘‘‘~‘‘~>,~‘‘‘‘‘,~,~,°,,,:‘‘+‘‘‘~>:~;~::~, 10.1% 4. Syria. 10.9 ~-,.~4~,~!~,?~,,,,,,.;:,~:~,,,,,,:~,,,,,~:~,,,,,,,,,.,,,~ 6.8% 5. United Arab Emirates. 10.1 ~,’.-:~:~>~:~’~",~:~i-~;,,~,~,-"~k~;,4~;~r,-~ 6.3% 6. United States. 8.8 ~,,,~,~:~?~,’.t-,," 5.5% 7. Iran. 8.8 ~,,,, ,..-~o~,~s,~.~;!~.~., 5.5% ~’~N~%"’;’~;~ 5.3% 8. Pakistan ............... 8.5 9. United Kingdom. 8.2 a~’.:~.-’..~;~+’..,.~ 5.1% 10. Jordan ................ 5.4 11. Lebanon ............... 4.6 ~x~z~!,~,~;: 2.~%o ;,~,,a~’~,~ 2.7% 12. Bahrain ............... 4.3 13. Bangladesh. 2.4 ,,,~;,~ 1.5% 14. Philippines ............. 1.6 i~,:!,~?~: 1.0% 15. Canada ................ 1.6 {~)i~; 1-0°/o 16. eater ................. 1.5 17. ~rance ................ 13 18. Sermany .............. 13 l& Sri Lanka .............. 1.2 20. 0man ................. 1.2 i;:~,! 0.7% Other ................. 17.0 :~:il ~;:!::iit 2~ii!:,,-,i ~": ~7i,-7~::,;;, 2 ~: ~i!~i,",": )~i~!,,,,,’,,.it,,’:i,L 2; 10.6% Total 160.5 © TeleGeography, Inc. 1998 National Traffic Balance Mi’FI" 1995 1996 1997 Incoming 130.2 131.2 n.a. Outgoing 125.9 140.7 160.5 Surplus (Deficit) 4.3, (9.4) n.a. Total Volume 256.1 271.9 n.a. Note; MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes of public switched traffic. © TeleGeography, Inc. 1998 TeleGeography 1999 © TeleGeography, Inc. 1998 Luxembourg Largest Telecommunications Routes, 1996 Destination Millions of Mi~’l"s Percentage of Outgoing Traffic 2. Germany. 50.9 ~;~-~’~ ~g~,:,~,,%~:~-~,-~=~ 20.5% 3. France ................. 50.3 ~:~~~~~:~:~¢~5~ 20.2% 4. Po~ugal. 15.7 ~:~;~ 6.3% ............... 5. United Kingdom. 14.0 ........ ~..~;,~ 6. Italy.. 11.4 ~,~?~ 4.6% .................. 7. Netherlands. 8.7 ~;~;~ 3.5% 8. Swi~erland ............. 6.6 9. United States. 5.7 ~,~,~ 2.3% 10. Spain .................. 3.9 1.6% 11. Denmark ............... 3.2 1.3% 12. Austria ................. 2.3 0.9% 13. Sweden ................ 2.3 0.9% 14. Greece ................. 1.7 0.7% 15. Ireland ................. 1.1 ~? 0.4% 16. Finland ................. 1.0 0.4% 17. Poland ................. 0.9 0.4% 18. Russia .................. 0.8 0.3% 19. Japan .................. 0.7 ~: 0.3% 20. Canada ................. 0.7 ~, 0.3% Other .................. 10.7 ~,~5"~:~,~;::~-~:~ 4.3% Total 248.5 © TeleGeography, Inc. 1998 National Traffic Balance MiTT 1994 1995 1996 incom!ng 145.2 174.5 189.8 outgoing 213.5 232.2 248.5 Surplus (Deficit) ..... (68.3) (57.7) (58.8) Total Volume 358.7 406.7 438.3 Note: MiTT is Minutes of Telecommunications Traffic. Data are in millions of minutes of public switched traffic. © TeleGeography, Inc. 1998 © TeleGeography, Inc. 1998 TeleGeography 1999 Macau Largest Telecommunications Routes, 1997 Destination Millions of MH-rs 1. Hong Kong ............ 50,3 Z China ................. 50.2 3. Portugal .............. 3.8 4. Taiwan ............... 3.6 5. United States .......... 1.8 6. Philippines ............ 1.5 7. Canada ............... 1,4 8. Thailand .............. 1.1 9. Australia .............. 0.9 10. United Kingdom ........ 0.7 11. Singapore ............. 0.5 12. Japan ................ 0.4 13. Malaysia .............. 0.3 14. South Korea ........... 0.3 15. France ................ 0,3 16. Indonesia ............. 0.2 17. New Zealand .......... 0.1 18. Germany .............. 0.1 19. Vietnam ............... 0.1 Other ................. 1.3 Total 119.0 Percentage of Outgoing Traffic ~->~,,’~,’~’,~’;’; ........ ’,~,’;,,’.~,;;’; ................... ............................. ....... ........ ~-’" .... 42.2% i~>;~% 3.2°/° ~-3.0% 1.5% ~"h~ ~.3% ~ 0.9% 0.8% k: 0.6% o. O/o ;o.4% O.3% 0.3% 0.2% ,0.2% 0.1% 0.1% 1,1% ® TeleGeography, Inc. 1998 National Traffic Balance Mi-I-I- 1995 1996 1997 Incoming 90.4 92.1 92.2 Outgoing 108.1 112.5 119.0 Su rp! us (g efi cit) (17.7) (20.4) (26.7) Total Volume 198.5 204.6 211.2 Note: Mi’Fr is Minutes of Telecommumcat=ons Traffic. Data are in millions of minutes of public switched traffic. Data based on bdhng point of traffic. ® TeleGeography, Inc. 1998 203 TeleGeography 1999 © TeleGeography, Inc. 1998 Macedonia Largest Telecommunications Routes, 1997 Destination Millions of Mi3-1"s Percentage of Outgoing Traffic ~~ 9.6% Z Germany. 5.0 ~,~ ..... 3. Bulgaria 3.2 ~~ 6.2% .............. ’ ........... 5.1% 4. Swi~erland 2.6 :~ 5. Greece 2.6 4.9% ............... 6. Slovenia .............. 2.1 ~;~ 4.0% 7. Turkey 2.0 ~%~ 3.9% ................ 8. Croatia 2.0 ~:~: 3.9% ............... ~~ 9. United States .......... 1.7 3.2% 10. Italy .................. 1.6 ~;;~A~ 3.0% 11. Austria 1.3 ~~:~ 2.5% 12. Canada ............... 1.0 ~#~ 2.0% 13. Australia .............. 0.9 ~c;~:~ 1.8% 14. United Kingdom ........ 0.8 1.5% 15. Sweden ............... 0.6 1.1% 16. Albania ............... 0.6 ~C~ 1.1% 17. Nethede~ds ........... Q.6 ~:~ 1,1% 18. Russia ................ 0.5 0.9% 19. Bosnia-Herzegovina .... 0.5 0.9% 20. Belgium ............... 0.4 ;~ 0.7% Other ................. 4.0 ~,,,;,,,~,-; ~% ~:,,~; Total 51.7 © TeleGeography, Inc. 1898 National Traffic Balance MiTT 1995 1996 1997 Incoming ,, 82.0 8!,~9, 85.2 Outgoing 45.0 51.0 51.7 Surp,!us (Deficit) ,, 37.0 ........ 30,9 33.5 Total Volume 127.0 132,9 136.9 Note: Mi’lq- is M~nutes of Telecommunications Traffic. Data are in milhons of minutes of public swztched traffic. 204 © TeleGeography, Inc. 1998 © TeleGeography, Inc. 1998 TeleGeography 1999 Malaysia ~g,: 4.3 Yo ~{:%,~ 3.8% #~.~. 3.7% @ 2.6% 2.5% ~: 2.2% ;:.g~ 2.0% ~:~ 1.8% 1.0% 0.8% 0.7% 0.6% 0.6% 0.5% 0,5% 0.5% 6. 7. 8. 9. 10. 11. 13. 14. 15. 16. 17. 18. 19. 20. Millions of Mi’n’s 1. Singapore ............. 330.1 Z Indonesia .............. 37.3 3. Japan ................. 25.1 4. United Kingdom ........ 22.4 Australia .............. 21.5 United States .......... 18.0 Thailand ............... 15.2 Hong Kong ............. 15.0 India .................. 13.0 Taiwan ................ 11.5 Philippines ............. 10.4 China ................. 7.1 Bangladesh ............ 6.0 Germany .............. 5.0 South Korea ........... 4.2 Pakistan ............... 3.7 Canada ................ 3.4 Saudi Arabia ........... 3.2 Myanmar .............. 3.1 Brunei ................ 2.8 Other ................. 30.5 Total 588.5 © TeleGeography, Inc. 1998 National Traffic Balance MiTE FY 1995/96 FY 1996/97 FY 1997/98 Incoming 442.0 581.9 592.0 Outgoing 408.3 570.5 588.5 Surplus (Deficit) 33.7 11.4 3.5 Total Volume 850.3 1,152.4 1180.5 Note: Mi’lq" ~s Minutes of Telecommunications Traffic. Data are in m~llions of minutes of public switched traffic. Data based on bilhng point of traffJc Traffic figures for FY 1995/96 exclude some cross-border traffic to S~ngapore, Traffic is for Telekom Malays=a only. Other carriers originated an estimated 135 milhon minutes of addltmnal traffic in FY 1997/98. Fiscal year ends 31 March. © TeleGeography, Inc. 1998 205 TeleGeography 1999 © TeleGeographyo Inc. 1998 Malta Largest Telecommunications Routes, 1997 Destination Millions of Mi~’]’s Percentage of Outgoing Traffic 1. United Kingdom ........ 11.3 ~~;‘~‘~‘‘~:~‘~‘~‘~;~‘:~‘~:‘~‘~:!~‘~z~ 32.7% 3. Germany .............. 2.8 ~~ 8.1% ~,~ 4.2% 4. Libya ................. 1.5 5. France ................ 1.3 ~ii 3.8% 6. United States .......... 1.3 ~!! 3.8% 7. Netherlands ........... 1.0 2.9% 8. Australia .............. 0.9 ~i 2.6% 9. Russia ................ 0,8 2.5% 10. Switzerland ........... 0.6 ~i 1.9% 11. Belgium ............... 0.5 ~!i 1.6% 12. Canada ............... 0.5 1.3% 13. Tunisia ............... 0.4 ~i 1.1% 14. Austria ............... 0.4 ~’~ 1.1% 15. Egypt ................. 0.3 i~ 1.0% 16. Spain ................. 0.3 .ii~i 0.9% 17. Greece ............... 0.3 ii~ 0.9% 18. Turkey ................ 0.3 !~i 0.9% 19. Sweden ............... 0.3 ii 0.8% 20. Yugoslavia ............ 0.3 i!~! 0.8% Other. 3.8 ..... ....................... 11.0% Total 34.4 © Te~eGeograph¥, Inc. 1998 National Traffic Balance MiTT 1995 1996 1997 Incoming ’,, ¯3!,.0 ........ 34.0,. n.a. Outgoing 28.5 31.7 34.4 Surplus (Deficit,) ..... 2.~5 ,, ,2.3 n.a. Total Volume 69.5 65.7 n.a. Note: M~’FI" ~s Minutes of Telecommumcat=ons Traffic Data are in millions of minutes of public switched traffic. 206 © TeleGeography, Inc. 1998 © TeleGeography, Inc. 1998 TeleGeography 1999 Mexico Largest Telecommunications Routes, 1997 Destination Millions of MWI’s 1. United States .......... 89;3.5 2. Canada ............... 16.2 3. Spain ................. 9.4 4. Italy .................. 6.0 5. Germany .............. 5.9 6. France ................ 5.8 7, Guatemala ............ 5.6 8. United Kingdom ........ 4.4 9. Colombia .............. 4.2 10. Cuba ................. 3.9 11. Costa Rica ............ 3.8 12. Argentina ............. 3.6 13. Chile ................. 3.6 14. Brazil ................. 3.6 15. Peru .................. 3.0 16. El Salvador ............ 2.7 17, Japan ................ 2.6 18. Venezuela ............. 2.4 19. Puerto Rico ........... 2,2 20. Honduras ............. 2,0 Other ................. 24.4 Total 1,008.9 Percentage of Outgoing Traffic ill 1.6% 0.9% i 0,6% ! 0.6% 0.6% 0,6% o.4% 0.4% 0.4% 0.4% 0.4% 0.4% 0.4% 0.3% o.3% 0,3% 0.2% 0.2% 0.2% i~: 2.4% © TeleGeography, Inc. 1998 National Traffic Balance MiTT 1995 1996 1997 Incoming 2,114.0 2,489.7 n.a. Outgoing 950.0 1,070.7 ..... 1,200 Surplus (Deficit) ,, 1,164.0 1,419.~0 n.a. Total Volume 3,064.0 3,560.4 n.a. Note: MiTI zs Minutes of Telecommunications Traffic. Data are zn millions of minutes of public switched traffic. Data based on bilhng po=nt of traffic Totals for 1997 include all carriers; route data are for Telmex only. © TeleGeography, Inc. 1998 ......... TeleGeography 1999 © TeleGeography, Inc. 1998 Moldova Largest Telecommunications Routes, 1997 Destination Millions of Mi’l-rs 1. Russia ................ 24.0 2_ Ukraine ............... 17.0 3. Romania .............. 5.0 4. Belarus ............... 2.0 5. Germany .............. 1.5 6. United States .......... 0.7 7. Israel ................. 0.7 8. Greece ............... 0.7 9. Bulgaria .............. 0.8 10. Turkey ................ 0.5 11. Italy .................. 0.5 12. Poland ................ 0.3 13. United Kingdom ........ 0.2 14. Hungary .............. 0.2 15. France ................ 0.2 18. Kazakhstan ............ 0.2 17. Canada ............... 0.2 18. Lithuania .............. 0.2 19. Latvia ................. 0.2 20. Czech Republic ........ 0.2 Other ................. 1.8 Total 56.8 Percentage of Outgoing Traffic 3.5% ~ii~;~ 2.6% ~i:i 1.2% 1.2O/o t~i 1.2% ~i; 1.1% t~ 0.9%o 0.6% 0.5%o i o.4% 0.4% ;~ 0.4% 0.4% 0.4% 0.3% 0.3% 0.3% 3.1% © TeleGeography, Inc. t998 National Traffic Balance Mi-FI 1995 1996 1997 !nc,om,,ing ......... n,.a. n.a. 80.~ Outg~i,ng ......... 50.8 50.2 56:8., Surplus(Deficit) ,n-a., n.a. 23.3 Total Volume n.a. n.a. 137.0 Note: M~FF is M~nutes of Telecommunications Traffic. Data are =n millions of minutes of public swztched traffic. Data based on billing po=nt of traffic. Totals for 1995 and 1996 include only traffic to other member states of the Commonwealth of Independent States. See page 174 for a matrix of traffic from other CIS member states. © TeleGeography, Inc. 1998 2O8 © TeleGeography, Inc. 1998 TeleGeography 1999 Namibia Largest Telecommunications Routes, 1997 Destination Millions of MiTrs 1. South Africa ........... 42.5 2_ Germany .............. 1.7 3. United Kingdom ........ 0.8 4. United States .......... 0.6 5. Zimbabwe ............. 0.6 6, Botswana ............. 0.4 7. Angola ................ 0.4 8. Zambia ............... 0.3 9. Spain ................. 0.2 10. France ................ 0.2 11. Italy .................. 0.2 12. Netherlands ........... 0.2 13. Portugal .............. 0.1 14. Switzerland ........... 0.1 15. China ................. 0.1 16. Australia .............. 0.1 17. Norway ............... 0.1 18. Austria ............... 0.1 19. Moldova .............. 0.1 20. Russia ................ 0.1 Other ................. 3.3 Total 52.3 Percentage of Outgoing Traffic ii~i 3.3% }:il 1.6% ~, 1.1% ~i 0.8% O2% O2% 0.4% 0.4% O3% O3% 0.2% © TeleGeography, Inc. t998 National Traffic Balance MiTT 1995 1996 1997 Incoming n.a. n.a. 47:0 Outgoing 50.9 51.4 52.3 Surplus (Deficit) n.a. n.a. (5.3) Total Volume n.a. n.a. 99.3 Note: M~TT is Minutes of Telecommunications Traffic. Data are in milhons of minutes of pubhc switched traffic © TeleGeography, Inc. 1998 209 TeleGeography 1999 © TeleGeography, Inc. 1998 ( Netherlands Largest Telecommunications Routes, 1997 Destination Millions of MiTrs Percentage of Outgoing Traffic 3. United Kingdom 198.3 ~, ~,~,,~,,~-,~%~,;~,,;~,~,,,,,~,,?~,,~-~,~,,% 12.9% 4. France 111.2 ~,~!,~?, ,;~;~,~,~,~’~, 7 2% 5. United States. 96.1 ~,-~-~,~:~,~+,~%~#z#~ 6.3% 6. Italy.. 48.3 7. Spain ................. 46.8 ~#~ 3.0% & Swi~erland ........... 39.8 2.6% ~@~4 2.O% 9. Turkey ................ 30.5 4 10, Sweden ............... 24.6 ~2~ 1.6% ’~:~;~’~’~’~ 1.4% 11. De~mark .............. 21.2 ;~, 12. Austria ............... 20.0 ~;:, 1.3% ;~ 1,3% 13. Canada ............... 19.6 ~:~:~ 14. Poland 16,0 ~;~;~ 1.0% 15. Ireland ................ 12.5 ;~, 0.8% 16. Noway ............... 12.5 ~:~:~ 0.8% 17. PoKugal .............. 12.2 ;~0.8~o 0.7% 18. Greece ............... 11.5 ~:~ 19. Morocco .............. 10.7 ~.~ 07% 20. Suriname ............. 10.5 0.7% Total 1,~5.0 © TeleGeography, Inc. 1998 National Traffic Balance MiTT 1995 1996 1997 Incoming 1,453.0 1,584.6 1,712.2 Outgoing 1,458;7 _ 1,534.1 1,535.0 Surplus, (Deficit) (5.7) ,, 50.5 177.2 Total Volume 2,911.7 3,118.7 3,247.3 Note: Mi]-I" ~s Minutes of Telecommumcations Traffic Data are in milhons of minutes of pubhc switched traffic. Data based on bilhng po=nt of traffic Traffic is for P]-I" Telecom (KPN) only. 210 © TeleGeography, Inc. 1998 © TeleGeography, Inc. 1998 TeleGeography 1999 New Zealand Largest Telecommunications Routes, FY 1997/98 Destination Millions of MiTrs Percentage of Outgoing Traffic 1. Australia 185 ............... ~............... ~,’, ......... ~, ......... ," ............................................... 45.5% Z United Kingdom 50 ~~ ~-~’;,,,,~:~;,-~,,’,’.,~,,,,~’~,~,~4:~-~:,. 12.3% 3. United States 40 ~;~:>~’~t~":~:~f 9.8% 4. Japan ................ 9 2.2% 5. Canada ............... 9 ~:~ 2.2% 6. Hong Kong ............ 8 2.0% 7. Fiji ................... 8 ~] 2.0% 8. Singapore ............. 7 1.7% 9. Malaysia .............. 6 1.5% 10. Western Samoa ....... 5 ~,~ 1.2% 11. Germany .............. 5 1.2% 12. Taiwan ............... 4 1.0% Tetal © TeleGeography, Inc. 1998 Natioaal Traffie Balaaee MiTT FY 1995/96 FY 1996/97 FY 1997/98 Incoming 327 380 n.a. Outgoing 312 353 407 Surplus (Deficit) 15 27 n.a. Total Volume 639 733 n.a. Note: M~’FI" =s Minutes of Telecommumcations Traffic. Data are in mill=ons of minutes of pubhc switched traffic. F=scal Year ends 31 March. © TeleGeography, Inc. 1998 211 TeleGeography 1999 © TeleGeography, Inc. 1998 Nicaragua Largest Telecommunications Routes, 1997 Destination Millions of Mil"l’s Percentage of Outgoing Traffic 1. United States. 16.1 ............. "-’,~ ................................................................. ................. 39.7% 2. Costa Rica 9.2 {-,,~:,,,,,~,~,,~,~;:~,~-,,-~,,R~,,,~,~,~ 22.8% 3. Guatemala 3.3 ~,~"~,,,~,:~,,,,,~:~: 8.2% 4. El Salvador 3.0 ~’~’~;~4.~, 7,5% .............. ~; 7.os 5. Honduras 2.8 "~~:~;~£~"~::~:;’ 6. Mexico ................ 1.0 2.6% 7. Panama ................ 0.9 2.3% 8. Canada ................ 0.7 1.7% 9. Spain .................. 0.5 1.3% 10. Colombia ............... 0.3 0.6% 11. Brazil .................. 0.2 0.5% 12. Cuba .................. 0.2 0.5% 13. Germany ............... 0.2 0.4% 14. Argentina .............. 0.2 0.4% 15. Italy ................... 0.2 0.4% 16. Chile ................... 0.1 0.3% 17. Netherlands ............ 0.1 ; 0.3% 18. Peru ................... 0.1 0.3% 19. Venezuela .............. 0.1 0.2% 20. France ................. 0.1 ,0.2% Other .................. 1.1 ~,::,~ 2.8% Total ~.4 © TeleGeography, Inc. t998 National Traffic Balance Mi’l-I" 1995 1996 1997 !ncoming n.a. n.a. ,, 52.5 Outgoing, ¯ 29.4 ,~ n.a. 40.4 Surp!u,s (Deficit) n.#. n.a. 12.1 Total Volume n.a. n.a. 92.9 Note: MiFr is Minutes of Telecommumcat=ons Traffic Data are in millions of minutes of pubhc sw=tched traffic, Data based on b=lling point of traff=c 212 © TeleGeography, Inc. t998 © TeleGeography, Inc. 1998 TeleGeography 1999 Norway Total 481 © TeleGeography, Inc, 1998 National Traffic Balance MiTT 1995 1996 1997 Incoming 373.2 422.3 n.a. Outgoing 431.5 443.5 481 Surplus (Deficit) (58.3) (21.2) n.a. Total Volume 804.7 865.8 n.a. Note: MiFr is Mznutes of Telecommumcations Traffzc. Data are m mzlhons of minutes of pubhc switched traffzc Data for 1995 based on billing point of traffic. © TeleGeography, Inc. 1998 ......... TeleGeography 1999 © TeleGeography, Inc. 1998 t Oman Largest Telecommunications Routes, 1997 Destination Millions of Mil-I’s 1. United Arab Emirates... 19.7 2_ India .................. 19.6 3. United Kingdom ........ 4.7 4. Pakistan ............... 4.4 5. Egypt ................. 2.8 6. Saudi Arabia ........... 2.3 7. Bahrain ............... 2.1 8. Bangladesh ............ 1.6 9. Jordan ................ 1.3 10. Qatar ................. 1.2 11. Kuwait ................ 1.2 12. United States .......... 1.1 13. Sri Lanka .............. 0.8 14. Tanzania .............. 0.8 15. Philippines ............. 0.7 16. Germany .............. 0.6 17. France ................ 0.6 18. Netherlands ........... 0.6 19. Canada ................ 0.6 Other ................. 7.6 Total 74.3 Percentage of Outgoing Traffic ~" 3.8% 3.1% ~’.’," ~&’~:"~-~ 2.8% ~,.~ 2.2% ~,~" 1.7% ...... 1.6% ;~;~’~ 1.6% 1.5% 1.1% 1.1% 1.0% ~i~ 0.8~o %~ 0.7% © TeteGeography, tnc, 1998 National Traffic Balance MiTT 1995 1996 1997 Incoming, ’ ..... 53.3 ...... 58.0 70.4 Outgoing 54.4 62.6 74.3 Surplu,s (Deficit) (1.1) (4.6) (3.9), Total Volume 107.6 120.6 144.7 Note: MiTT =s M~nutes of Telecommunications Traffic. Data are in rndlions of rmnutes of public switched traffic. Data based on bilhng point of traffic. © TeleGeography, Inc, 1998 © TeleGeography, Inc. 1998 TeleGeography 1999 Pakistan Largest Telecommunications Routes, FY 1997/98 Destination Millions of Mi~-I’s Percentage of Outgoing Traffic 3. United Arab Emirates 9.9 ~‘;~‘~‘~:~:~:~‘~::~t~:~::‘~‘~‘~:~v~~ 11.8% 4. Saudi Arabia 7.1 ......... ,~,;~-~:,,-~,,~:~,:~:~;~z,~,~,~ 8.4% 5. Canada 4.2 ...... ,,’~’ ........................... 5.0% 6. Italy 3.2 ~;~:~;;;~;~;~?~" 3.9% 7. Germany ............... 2.8 ~,~;,~,,~,,,~,,,~,~,~:,,~;~,,~,~,: 3.3% 8. France.. 2.8 ¢~t,~,~,~,g,~ 3.1% *’,~’~’~~ 2.7% g. Singapore .............. 2.3 .................. 11. India ................... 2.0 12. Kuwait. 1.g ~!:~ 1.9% 13. Hong Kong .............. 1.4 ~1Z’~ 1.7% ~’~’~ 1.3% 14. Netherlands ............ 1.1 15. Ghina .................. 1.0 ~::~ 1.2% 1~. Bangladesh ............. 1.6 ~,,~:~’~ 1.2% 17. Turkey ................. 1.0 ~[~’~,: 1.2% 18. Iran .................... 1.0 ~’~:~ 1.2% lg. Oman .................. 0.8 ~:~:~: 1.0% 20, Swi~erland ............. 8.~ ~i}" 0.8% Other ................... 7.8 Tetal © TeleGeography, Inc. t998 National Traffic Balance Mi-I-I- FY 1995/96 FY 1996/97 FY 1997/98 Incoming 362.1 488.4 557.8 Outgoing 65.9 77.0 84.1 Surplus (Deficit) 296.2 411.5 473.7 Total Volume 428.0 565.4 641.9 Note: M=’FI" =s M=nutes of Telecommumcations Traff=c. Data are =n m~llions of minutes of public switched traffic. Fiscal year ends 30 June. Data exclude some cross-border traffic to Indza. © TeleGeography, Inc. 1998 215 TeleGeography 1999 © TeleGeography, Inc. 1998 Panama Largest Telecommunications Routes, 1997 Destination Millions of MiTrs Percentage of Outgoing Traffic 2. Colombia 4.7 ,-~,~:~s,~ 11.3% 3. Costa Rica 2.8 ~;~-~ 6.7% ............. ~#~Z~ 4. Mexico 1.8 4.3% ................ 5. Venezuela .............. 1.0 2.5% 6. Guatemala ............. 0,9 2.2% 7. Ecuador ................ 0.8 5~ 2.0~o ~:~ 2.0% 8. Dominican Republic ..... 0.8 ~:~ 9. Brazil .................. 0.7 ~;~ 1.8% 10. El Salvador ............. 0.7 ~;;~ 1.7% 11. Spain .................. 0.7 1.7% 12. Peru ................... 0.6 1.5% 13. Nicaragua ............. 0.6 1.4% 14. Cuba .................. 0.5 ~::~ 1.2% 15. Honduras .............. 0.5 1.2% 16. Canada ................ 0.5 :~:! 1.1% 17. Argentina .............. 0.5 1.1% 18. Chile ................... 0.4 ~:~ 1.1% 19. United Kingdom ......... 0.4 1.0% Other .................. 4.0 ~::~:~;~,, :~ ~: 9.5% Total 41.4 © TeleGeography, Inc. 1998 National Traffic Balance Mi]3 Incoming, Outgoing Surplus (Deficit) Total Volume 1995 1996 1997 94.2 97.7 95.1 39.5 41.2 41.4 54.7 56.5 53.7 133.7 138.9 136.5 Note: MzI-F =s Minutes of Telecommumcations Traffzc Data are =n m~llzons of m=nutes of public switched traff=c. Data based on bilhng pmnt of traff=c. © TeleGeography, Inc. 1998 © TeleGeography, Inc. 1998 TeleGeography 1999 Papua New Guinea Largest Telecommunications Routes, 1997 Destination Millions of Mi’l-i’s Percentage of Outgoing Traffic 1. Australia 16.2 ............................................. ¯ ................... ,,~ ...................... 69.0% 2. Philippines ............ 1.0 ~:i 4.4% 3. Malaysia .............. 1.0 ~i~i 4.1%o 4. NewZealand .......... 0,9 ~i~ 3.8% 5. United States .......... 0.7 ~i: 3.2% 6. Singapore ............. 0.6 ~i! 2.7% 7. United Kingdom ........ 0.6 ~!i 2-4% 8. Indonesia ............. 0.4 I~i,i 1.8% 9. French Polynesia ...... 0.4 Ii~~. 1.6% 10. Japan ................ 0.4 ~i 1.6% ~,"~ 5.1% Other ................. 1.2 Total 23.4 © TeleGeographyo Inc, t998 National Traffic Balance Mi-I-I 1995 1996 1997 Incoming, ,, n.a. n.a. 20.6 Outgoing 23.9 26.9 23.4 Surp, lus (Deficit) n.a. ¯ n.a. (2.8),~ Total Volume n.a. n.a. 44.0 Note: MiTT is Minutes of Telecommunicat=ons Traffic. Data are in mdhons of minutes of public switched traffic. © TeleGeography, Inc, t998 217 TeleGeography 1999 © TeleGeography, Inc. 1998 ( Paraguay Largest Telecommunications Routes, 1996 Destination Millions of MiTrs 1. ArBe~tina .............. 8.2 2. Brazil .................. 6.3 3. United States ........... 3.0 4. Chile ................... 1.2 5. Uruguay ............... 1.2 6. Germany ............... 0.5 7. Taiwan ................ 0.4 8. Spain .................. 0.4 9. Rep. of Korea ........... 0.4 10. Bolivia ................. 0.3 11. Peru ................... 0.3 12. Italy ................... 0.2 13, Japan ................. 0.2 14. France ................. 0.2 15. Mexico ................ 0.2 16. Colombia ............... 0.2 17. United Kingdom ......... 0.1 18. Panama ................ 0.1 19. Switzerland ............ 0.1 20. Hong Kong ............. 0.1 Other .................. 1.3 Total 24.9 Percentage of Outgoing Traffic ~;~:~;t,’ 4.8% 2.0% ~! 1.6% 1.6% ~!i:i 1.6% ~’.~: 1.2% !~: 0.8% :f~:~ 0.8% 0.8% o.e% :~’:: 0.8% 0.4% 0.4% ;, 0.4% o.,% © TeleGeography, Inc, 1998 National Traffic Balance Mi-I-F 1994 1995 1996 Incoming .... 30.6 n.a. 49.4 Outgo!ng 18.1 20.9 24.9 Surplus (Deficit) ,, 12.5, ¯n.a. 24.5 Total Volume 48.7 n.a. 74.3 Note: Mi’FI" ~s Minutes of Telecommunicatmns Traffic. Data are in millions of minutes of public switched traffic. 218 © TeleGeography, Inc. 1998 © TeleGeography, Inc. 1998 TeleGeography 1999 Largest Telecommunications Routes, 1997 Destination Millions of MiTTs 1. United States ........... 29.7 2. Chile ................... 6.0 3. Spain .................. 5.8 4. Argentina .............. 5.5 5. Japan ................. 3.5 6. Italy ................... 3.3 7. Colombia ............... 3.2 8. Brazil .................. 2.7 9. Venezuela .............. 2.6 10. Bolivia ................. 2.4 11, Mexico ................ 2.1 12. Canada ................ 1.9 13. Germany ............... 1.5 14. United Kingdom ......... 1.4 15. Ecuador ................ 1.3 16. France ................. 1.0 Other .................. 5.5 Total 79.4 Peru Percentage of Outgoing Traffic 7.4% ,~ 4.1% ,~ 3.3% ~.~ 3.3% ~:~£~ 3.0% 2.4% ~#~ ~, 1.9% ~@ 1.7% ;2~ 1.7% ,~ 1.3% © TeleGeography, Inc, t998 National Traffic Balance MiTT 1995 1996 1997 Incoming 195.4 226.5 256.9 Outgoing 62.6 66.7 79.4 Surplus (Deficit) 132.8 159.7 ,, 177.5 Total Volume 258.0 293.2 336.4 Note: M~l-r is Minutes of Telecommumcations Traffic. Data are in m=lhons of minutes of pubhc switched traffic Data based on billing point of traffic. © TeleGeography, Inc. 1998 ......... TeleGeography 1999 © TeleGeography, Inc. 1998 Philippines Largest Telecommunications Routes, FY 1997198 Destination Millions of MiTts Percentage of O~going Traffic 3. Hong Kong ............ 35 4. Canada ............... 15 ~:~ii~{ 5.1% 5. Singapore ............. I4 !~12~,:"2;’)~ 4.7% 6. Taiwan 12 ..... ~.z~: .... 4.1% 7. ~ep. of Eorea .......... 12 ~:~a~~: 4.1% & Australia .............. & Saudi Arabia .......... 7 Total ~5 © TeleGeography, Inc. t998 National Traffic Balance Mil-I" FY 1995/96 FY 1996/97 FY 1997/98 I ~ncoming 691 767 ,, 930 Outgoing 174 240 295 Surplus (Deficit) ¯ ,,5!7, 527 635 Total Volume 865 1,007 1,225 Note: Mi’Fr =s M=nutes of Telecommunications Traff=c Data are zn millions of minutes of pubhc switched traff=c. Data based on bdling point of traffzc. Route data for top four routes rounded to the nearest five million minutes. F=scal year ends 31 March © TeleGeography, Inc. 1998 22O © TeleGeography, Inc. 1998 TeleGeography 1999 Poland Largest Telecommunications Routes, 1996 Destination Millions of Mi’l-rs Percentage of Outgoing Traffic 2. United Kingdom ......... 25.0 ~,,~;I~ 5.7% 3. United States ........... 24.2 5.5% 4. France ................. 23.2 ~,~ 5.3% Italy .................... 22.3 ~>~’~:~’~’~,~ 5.1% Austria ................. 15,0 ~, 3.4% Netherlands. 13.9 ~;~ 3.2% Russia .................. 13.1 ~&~ 3.0% ~’~,~,~ Ukraine ................. 12.5 ~:,~: 2.9% Sweden ................ 12.0 Belgium ................ 9.0 ~;~f~ 2.1% Czech Republic .......... 8.7 2.0% Canada ................. 8.1 ~:~ 1.9% Denmark ............... 7.4 ’,::~:~ 1.7% Swi~erland ............. 6.2 ~’~ 1.4% Belarus ................. 6.0 ~" 1.4% Vietnam ................ 5.2 1.2% Spain .................. 4.4 ]~ 1.0% 0.8% Hunga~ ................ 3.4 Noway ................ 3.2 :~’~ 0.7% Other .................. 42.0 Total ~7.2 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. © TeleGeography, Inc. 1998 National Traffic Balance MiTT 1994 1995 1996 Incoming 643.8 649.3 725.5 Outgoing 356.6 381.4 437.2 Surplus (Deficit) 287.2 267.9 288.3 Total Volume 1,000.4 1,030.7 1,162.7 Note: MiTT is M~nutes of Telecommunications Traffic. Data are ~n millions of minutes of pubhc switched traffic. Data based on b=lhng pmnt of traff=c. ©TeleGeography, Inc 1998 221 TeleGeography 1999 © TeleGeography, Inc. 1998 ( Portugal Largest Telecommunications Routes, 1997 Destination Millions of MiTrs 1. France ................ 76.3 2_ Spain ................. 66.8 3. Germany .............. 42.5 4. United Kingdom ........ 41.9 5. Switzerland ........... 18.6 6. Brazil ................. 16.8 7. United States .......... 14.8 8. Italy .................. 14.6 9. Netherlands ........... 13.5 10. Belgium ............... 10.2 11. Angola ................ 8.9 12. Canada ............... 5.4 13. Luxembourg ........... 4.2 14. Cape Verde ........... 4.0 15. Guinea-Bissau ......... 3.9 16. Mozambique .......... 3.1 17, Sweden ............... 3.0 18, South Africa ........... 2.8 19. Denmark .............. 2.7 20. Venezuela ............. 2.7 Other ................. 36.5 Total 393.3 © TeleGeography, Inc. 1998 National Traffic Balance MiTT 1995 1996 1997 Incoming 525.0 571.4 628.8 Outgoing 283.9 340.0 ¯ 393.3 Surplus (Deficit) 241.1 231.4 235.5 Total Volume 808.9 911.4 1,022.1 Note: MzFr is Minutes of Telecommunications Traffic. Data are =n milbons of minutes of pubhc switched traffic. Data based on b~lling point of traffic. © TeleGeography, Inc. 1998 © TeleGeography, Inc. 1998 TeleGeography 1999 Romania Largest Telecommunications Routes, 1997 Destination Millions of MiTts Percentage of O~going Traffic 1.~ ynermanu. ,~n.~ .... ~,~;,’,~,~ ~;--~- ........... ,-,~,~,, "~’~,,~-~ ............. -’;,"~,’~’~,’,+~ ,-’~ ~"~’~ ~" ............ ...... 1£ ~o/~ & Italy .................. 15.6 ......... .......... ............................... "~’*~............................ ~:~’ ......... 14.1% 3. Hunga~ 7.7 ~",,’,~,,.~;’~," :,~,~:~’,~:,:~:’ ~;,, 6.9% France 7.4 Greece 5.9 ~’~~,~,~ ........... ~~,~ ...... 5.3% United Kingdom ........ 5.8 ~:~2~s~:’~-~&~s~:;~:~?~:~ 5.2% Turkey ................ &6 ~nite6 States .......... 4.B Austria ............... 4.8 ~:~~:: 2.0% Belgium ............... 2.2 Israel ................. 2.] ~::~:~:: ~.~% gu~oslavia ............ ].5 S~ain ................. ~.2 aussia ................ 0.s ~kraina ............... 0.8 Other ................. 14.8 ~,‘‘~‘~,‘‘~:‘‘~‘::;::~?:~‘~‘‘‘:~;~,‘‘;,~::~£~‘:£‘~s~‘~‘~,~‘~‘~‘c~‘f~‘‘ 13.4% Total 110.8 o 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. © TeleGeography, Inc. t998 National Traffic Balance MiFr 1995 1996 1997 Incoming 198.0 237.5,, 278.6 Outgoing 78.1 91.5 110.8 Surplus (Deficit), 119.9 146.0 167.8 Total Volume 276.1 429.0 389.4 Note: M=TT is M~nutes of Telecommunications Traffic. Data are ~n milhons of minutes of pubhc switched traffic. © TeleGeography, Inc. t998 ¯ ......... TeleGeography 1999 © TeleGeography, Inc. 1998 ( Russia Largest Telecommunications Routes, 1997 Destination Millions of MiTTs 1. Ukraine ............... 249.0 2_ Belarus ............... 98.9 3. Kazakhstan ............ 90.6 4. Germany .............. 48.9 5. Uzbekistan ............ 32.3 6. Georgia ............... 27.6 7. Azerbaijan ............ 27.5 8. Armenia .............. 25,7 9. Latvia ................. 22.2 10. United States .......... 21.6 11. Lithuania .............. 16.5 12. United Kingdom ........ 16.1 13. Moldova .............. 15.7 14. Kyrgyzstan ............ 14.4 15. Italy .................. 13.8 16. Estonia ............... 13.1 17. Finland ............... 12.4 18. Turkey ................ 11,2 19. France ................ 11.1 20. Israel ................. 9.6 Other ................. 161.1 Total 939.3 Percentage of Outgoing Traffic i:~ 2.9% 2.7% 2.4% ~%~ 1.7% ~;:~" 1.5% ;" 1.5% 1.4% ~,? 1.3% ’,~ 1.2% © TeleGeography, tnc, 1998 National Traffic Balance Mi]-I 1995 1996 1997 Incoming 448.1 1,037.6 1,035.6 Outgoing 287.4 851.3 939.3 Surplus (Deficit) 160:7 ¯ 186.3 96.3 Total Volume 735.5 1,888.9 1,974.3 Note: Mi’lq" ~s Minutes of Telecommunications Traffic. Data are in millions of minutes of public switched traffic. Data =s for Rostelecom only. Data for 1995 include traffic to other member states of the Commonwealth of Independent States only. See page 174 for a matrix of traffic from other ClS member states. © TeleGeography, Inc, 1998 224 © TeleGeographyo Inc. 1998 TeleGeography 1999 Saudi Arabia Largest Telecommunications Routes, 1997 Destination Millions of Mi~-I’s Percentage of Outgoing Traffic 1. Egypt. 164 ~’~";~ "~’:~ ~"~:~’~’~’~"~................... ~’~" .............. .......... ~’~ ......... ~" ...... ,,~,~- ~,,~,,, ,,~, ,,,,,, ,,~,:~:~.;~ ~,~,~,,~.,~: ,- ,~,,~,,~,~,~:,-,~--,~-,’,," ~%~t ,,~’,-,, 24.8% Z Pakistan 82 ~:~,,~’;,~ ~, ,~,;~-~,,~,~,~;~;,~,~,~:~,~ 12.4% 3. India ................. 68 & Syria ................. 36~,:,:;~;;~:~~t~ 5.5% 5. United States .......... 32 6. United Kingdom ........ 30 ~,4~:~,~,#,~;~,,,,~;~;~ 4.5% 7. Bahrain ............... 24 8. Yemen ................ 19 ~: 2.9% 9. Jordan ................ 19 2.9% 10. Philippines ............ 18 2.7°Io 11. United Arab Emirates... 17 2.6% 12. Kuwait ................ 15 Tetal © TeleGeography, Inc. 1998 National Traffic Balance MiTT 1995 1996 1997 Incoming n.a. n.a. n.a. Outgoing 537.3 584.4 660 Surplus (Deficit) n.a. n.a. n.a. Total Volume n.a. n.a. n.a. Note: MiTT is M=nutes of Telecommunications Traffic. Data are m milhons of m~nutes of public sw=tched traffic. © TeleGeography, Inc. 1998 ¯ ......... TeleGeography 1999 © TeleGeographyo Inc. 1998 ( Singapore Largest Telecommunications Routes, FY 1997198 Destination Millions of MiTts Percentage of Outgoing Traffic Z Indonesia ............. 90 .... 7.8% 3. Hong Kong ............ 85 ~}}’~,~’~ 5.6% 4. United States .......... 65 )~ .......... 4.7% 5. Australia .............. 55 6. China ................. 55~:~:~,,~;~x 4.7% 7. Japan ................ 50 & Thailand .............. S. ~nited ~in~dom ........ ~5 &0% Other. 208 :~ =,,,~:~s~}~,~::~;~ ~’~;~, 17.9% Total 1,161 © TeleGeography, Inc. t998 National Traffic Balance MiTT FY 1995/96 FY 1996/97 FY 1997/98 Incom!ng n.a. n.a. n.a. Outgoing 773 942 !,,161 Surplus (Deficit) ,, n.a. n.a. n.a. Total Volume n.a. n.a. n.a. Note: M=l-r ~s Minutes of Telecommunications Traffic. Data are =n millions of minutes of pubhc swztched traff=c. Data based on billing point of traffzc. Data for 1995 excludes some cross-border traff=c w=th Malaysia Rscal year ends 31 March © TeleGeography, Inc. 1998 226 © TeleGeogrephy, Inc. 1998 TeleGeography 1999 Slovak Republic Largest Telecommunications Routes, 1997 Destination Percentage of Outgoing Traffic 6. 7. 8. 9. 12. 13. 14. 15. 16. 17. 18. 19. 20. Millions of Mi’l-rs Czech Republic ........ 64.9 Germany .............. 16.0 Austria ............... 11.9 Hungary .............. 7.1 United Kingdom ........ 5.3 Italy .................. 4.7 Poland ................ 3.4 United States .......... 3.3 France ................ 3.1 Ukraine ............... 3.1 Russia ................ 2.6 Switzerland ........... 2.5 Netherlands ........... 1.9 Belgium ............... 1.3 Croatia ............... 1.2 Yugoslavia ............ 1.0 Vietnam ............... 0.8 Sweden ............... 0.7 Spain ................. 0.7 Greece ............... 0.6 Other ................. 8.5 Total 144.7 ~i~ 3.6% 3.2% !~,.-:;, 2.4% ~i~’: 2.3% ~i~’i 2.1% ~i",~ 2.1% 1.8O/o ~!ii 1,3% o.9% O2% O.6% 0.5% 0.5% 0.4% © TeleGeography, Inc. t998 National Traffic Balance MiFl" 1995 1996 1997 Incoming 81.6 159.0 174.4 Outgoing 58.8 134.1 144.7 Surplus (Deficit) 22.8 24.9 29.8 Total Volume 140.4 293.1 319.1 Note: MiFr =s M~nutes of Telecommumcat=ons Traffic. Data are in mdlions of minutes of pubhc sw=tched traffic. Data based on b~lhng pmnt of traff=c. Data for 1995 exclude cross-border traffic with the Czech Repubhc © TeleGeography, Inc. 1998 227 TeleGeography 1999 © TeleGeography, Inc. 1998 Slovenia Largest Telecommunications Routes, 1997 Destination Millions of Mi~’l’s Percentage of Outgoing Traffic 1. Croatia 28.4 ............................................................................................................ E Austria 13.5 ............ ’~ ............ ............. ......... ,".~ 3. Germany .............. 13.0 ~c,~%~~~............ ~-~’ 11.5% 5. Yugoslavia 11.8 ~,’~,~,,~ ~:;~;,,~,~,~’:;,~:,’~,~ 10.4% 6. Bosnia-Herzegovina .... 5.1 ~;~,,:4~~~’~ o 7. Macedonia ............ 2.7 2.3% 8. Swi~erland ........... 2.6 2.3% 9. United Kingdom ........ 2.6 ~:~ 2.3% 10. France ................ 2.4 ~,~ 2.1% 11. United States .......... 1.9 1.7% 12. Hunga~ .............. 1.8 ~:~ 1.6% 13. Russia ................ 1.6 ~,~ 1.4% 14. Netherlands ........... 1.4 1.2% 15. Czech Republic ........ 1.2 ~;~ 1.0% 16. Sweden ............... 0.9 ~~ 0.8% 17. Belgium ............... 0.9 :~,~ 0.8% 18. Poland ................ 0.7 0.6% 19. Canada ............... 0.6 ~,~ 0.6% 20. Ukraine ............... 0.6 0.5% Other ................. 7.0 ~,:~,~’:~ 6.1% Tetal 113.5 © TeleGeography, Inc. 1998 National Traffic Balance MiFr 1995 1996 1997 Incoming 121.2 113.9 118.9 Outgoin£ 100.6 105.3 113.5 Surplus, (Deficit) 20.6 8.6 5.4 Total Volume 221.8 219.2 232.4 Note: MiTT is M~nutes of Telecommumcatlons Traffic. Data are in m~lhons of minutes of public switched traffic. 228 © TeleGeography, lnc. 1998 © TeleGeography, Inc. 1998 TeleGeography 1999 South Africa Largest Telecommunications Routes, 1997 Destination Millions of Mirrs Percentage of Outgoing Traffic 5. Mozambique 15 .............. ...... 4.1% 6. Botswana ............. 15 ~~ 4.1% 7. Germany. 13 ~;~,~,~ 3.5% 8. Swaziland ............. 12 ~:, 3.3% 9. Australia .............. 10 ~:~ 2.7% 10. Lesotho ............... 9 ~:~: 2.4% 11, Netherlands ........... 7 ~::~::~~ ~:~:f 1,9% 12, France ................ 7 ~>~:~ 13, Canada ............... 7 ~:~’ 1,9% 14, Zambia 6 ~’ 1,6% 15, Italy .................. S ~]~(~ 1,6% ~. ~o~u~a~ .............. 5 ~:~ ~.~% 17, Swi~erland ........... 5 :~ 1,4% ~. ~,,~a ................. 5 ~t~ ~.~% 19, Malawi ............... 4 ~:~.~:~ 1,1% 20, NewZealand .......... 4 ~:~ 1,1% Other ................. 54 :;~;;~’;~,~,:~"~:~ ~, ~,~;~ ~;~;~;~ ~,;~’:?:~:;:’~;’~:~:’~ ’:,~, 14.6% Tetal ~B.8 © TeleGeography, Inc. 1998 National Traffic Balance Mi-l-r 1995 1996 1997 Incoming n.a. n.a. 343.2 Outgoing ¯ ~305.0 353.0 368.8 Surplus (Deficit) n.a. n.a. (25.6) Total Volume n.a. n.a. 712.0 Note: Mz’13" =s Minutes of Telecommunications Traffic. Data are m milhons of minutes of public switched traffic. Route data for top six routes rounded to the nearest five milhon minutes. © TeleGeography, Inc. 1998 2)_9 TeleGeography 999 © TeleGeography, Inc. 1998 t Spain Largest Telecommunications Routes, 1995 Destination Millions of Mi’l-rs Percentage of Outgoing Traffic 1. G e rm a ny 160.7 ;~’~ .......... ~""~ ~’’,~’,~’~’’~’~ ..................... ’ ........... "~’°~" ....... .......... "~"" ............. 4. Italy ................... 68.2 ~~;~I~~~~ 5.7% 5. United States ........... 48.3 ~t~:~;;~&~;~:~4:~;:~ 6. Po~ugal ~.6 ~4~’t~:~4~."~’~,>’~ 3.4% ~~% 3.0% 7. Swi~erland 36.1 .............. .... 8. Belgium ................ 35.4 ~~:~~ 3.0% 9. Netherlands.. 35.3 ~=~~ 3.0% 10. Morocco 22.8 :,~,,,>,~+ :4~, 1.9% 11. Andorra ................ 17.2 1.4% 12. Argentina .............. 17.1 ,:~~:~, ~3. s~a~, ................ ~4.~ 1.2s 14. Chile ................... 12.9 1.1% 15. Colombia ............... 10.5 0.9% 16. Benmark ............... 9.5 ~4~ 0.8% 17. Mexico ................ 8.9 0.7% 18. Brazil .................. 8.8 0.7% 19. Cuba ................... 8.7 0.7% 20. Austria ................. 8.4 :~t 0.7% Total 1,024.9 © TeleGeography, Inc. 1998 National Traffic Balance Mi-I-] 1994 1995 1996 Incoming 969.9 1,076.4 n.a. Outgoing 948.3 ,,~ !,024.6 ,1,189.,0, ¯ Surplus (Deficit) 21.6 51.8 n.a. Total Volume 1,918.2 2,101.0 n.a. Note: MiFr is Minutes of Telecornmunicat~ons Traffic. Data are ~n milhons of minutes of public switched traffic. Data based on bilhng pmnt of traffic. 230 © TeleGeography, Inc. 1998 © TeleGeogruphg, Inc. 1998 TeleGeography 1999 Sri Lanka Largest Telecommunications Routes, 1997 Destination Millions of Mi’n’s Percentage of Outgoing Traffic Z United Kingdom 3.4 ......... ...................... .............. ............ 10.2% 3. Singapore 2.5 ............. 5. United States .......... 1.8 ~;~:~;~? 5.5% 6. Australia 1.6 ~:;’~:~,,~x N~{~:~N,~ 4.8% .............. 7. Hong Kong ............ 1.4 ~{~-;%:5~;~;&~,,~: 4.3% 8, Rep, of Korea .......... 1,1 #~;~4,;:%~;~:?~ 3.4% 9. eermanv .............. 1.1 10. United Arab Emirates 1.1 ~-,~,~,,,:~.~,~- 3.2% 11. Saudi Arabia 0.9 12. Italy 0.8 .... ~,",,-~"~,, 2.5% .................. l& Maldivas .............. 0.7 ~g~;: 2.~% 14. Pakistan .............. 0.6 ~:?~:~f:~# 1.9% 15. ~ranca ................ 0.a la. Canada ............... 0.a ~i~ ~.~% 17. Kuwait ................ 0.6 :,~,~,;,~, 1.8% 18. Thailand .............. 0.5 ;~’~:;:,, 1.7% 19. Malaysia .............. 0.5 ~1~:, 1.6% 20. Swi~erland ........... 0.5 ....... 1.5% Other ................. 4.6 ~"~:::~:)~ ~i~ t ::~2~;~)~;~):~" ’)~,:)~;’~:,,;}2~}~}~:’~’~" 14.0go Total ~.2 © TeleGeography, Inc. t998 National Traffic Balance MiFr 1995 1996 1997 Incoming ¯ ¯ 92.0 96.0 124.3 Outgoing 27.5 29.3 ,33.2 Surplus (Deficit) 64.5 66.7 ,9,1.1, Total Volume 119.5 125.3 157.5 Note: M=Fr ~s M=nutes of Telecomrnunicat~ons Traffic. Data are ~n rnilhons of minutes of public sw=tched traffic. © TeleGeography, Inc. 1998 251 TeleGeol~raphy 1999 © TeleGeography, Inc, 1998 ( Sweden Largest Telecommunications Routes, 1997 Destination Millions of Mi’l-rs Percentage of Outgoing Traffic 1. United Kingdom ........ 160 ,~:::~:~2:~k~;~2N~;~?~,~ga~%~*~;~4: ~;,~,~:;i~;~’;: 14.0% 4. Denmark .............. 110 5. aermany .............. 100 ~.;~a~a,~a~:~,;~:~ 8.8% 6. United States 100 ~?~:~%,~4%,~v~,~,~ ",~::~?~ 8.8% 7. Netherlands ........... 35 8. France 30 g~~:~}, 2.6% ................ 9. Poland 30 ~:~ 2.6% ................ ~0. S~i~erland ........... 25 ~} 2.2% Total 1,1~ © TeleGeography, Inc. 1998 National Traffic Balance MiFI 1995 1996 1997 Incoming ,, n.a. n.a. n.a. Outgoing ........ 900 1,026 ,,1,!,40 Surplus, !Deficit) ...... n.a. ,,, n.a. ,, n.a. Total Volume n.a. n.a. n.a. Note: Mz’lq" is Minutes of Telecommumcat~ons Traffic. Data are ~n mdhons of mznutes of public switched traffic. Route data rounded to the nearest five milhon minutes Data based on billing point of traffic. 232 © TeleGeography, Inc. 1998 © TeleGeogrephy, Inc. 1998 TeleGeography 1999 Switzerland Largest Telecommunications Routes, 1997 Destination Millions of Mi~l’s Percentage of Outgoing Traffic Z France 3~ "~ ............................. ~" ........... ~-~,,~’~,+~*~’,~,,,~~, 15.9% 3. Italy .................. 292 ~’-,,,,,,,~;,,~ ......... ;,~,~-,~’-~’~’,-’,~’~-~’~’:~,,":~--~" ~,~,~, 4. United Kingdom 119 ,~’~,~;;~;~;,~:~; 5.5% 5. United States .......... 97 ~:;~;~;~ 6. Austria ............... 88 ~:~%~;~,~’,:,~ 4.1% 7. Po~ugal 72 "~:;~?~s~&, 3.3% .............. ~t L ~~-~z~%~L~, ~#~;’~" 3.3% 8. Spain ................. 71 :~’t~&~j~: 9. Netherlands 52 ~,:~,’~,,~’~ 2.4% ~;~ 2.3% 10. Yugoslavia ............ 50 :~t~: Tetal 2,164 © TeleGeography, Inc. t998 National Traffic Balance MiF[ 1995 1996 1997 Incoming 1439.3 1,562.8 1,723 Outgoing 1,778.4 1,935.5 2,164 Surplus (Deficit) (339.1) (372.7) (441) Total Volume 3,217.7 3,498.4 3,887 Note: Mz]q zs Minutes of Telecommunicatzons Traffzc. Data are in mi]hons of m~nutes of pubhc swztched traffic. ©TeleGeography, Inc t998 25,5 TeteGeography 1999 © TeleGeography, Inc. 1998 t Syria Largest Telecommunications Routes, 1997 Destination Millions of Mitts Perce~age of O~going Trattic Z Saudi Arabia. 13.9 ~,~,~,~,~,-;~}!~i~;~:,~?~i,~;~!:,,,,,,;~ 15.8% 3. United Arab Emirates. 5.9 ........ ~:~’~.......... --;~ 6.6% 4. Jordan. 6.4 :~ ,~,,,,,,,,~$~A~~,,,, 7.1% 5. Kuwait ................ 4.9 ~i~:~’~ 4.9% 6. Canada ................ 4.4 7. United States. 4.3 [~{A;,~:,~ 4 8% ................ 9. ~r~ce ................ 2.8 ~. Turkeg. 2.2 2.5N ............... 12. United Kingdom ........ 1.6 ~;~:~:~ 1.8% 13. Italy. 1.6 .................. ~:~: 1.8% ~4. ~ussia ................ ~.0 ~:?~?~ ~.~% 15. Greece. 0.7 ~’~ 0.8% 16. Cyprus ................ 0.7 ~7. Su6an ................. 0.7 ~:: Other. 8.9 © TeleGeography, Inc, 1998 National Traffic Balance Mil-I" 1995 1996 1997 Incoming, 100 156.0 173.2 Outgoing 60 78.9 89.3 Surplus(Deficit) 40 ¯ 77.1 83:9, Total Volume 160 234.9 262.5 Note: M~FF ~s M~nutes of Telecommumcations Traffic Data are ~n mdlions of minutes of public switched traffic. Data based on bilhng pmnt of traffzc. 234 © TeleGeography, Inc. 1998 © TeleGeography, Inc. 1998 TeleGeography 1999 Taiwan Largest Telecommunications Routes, P/’ 1997/98 Destination Millions of MiTrs Percentage of Outgoing Traffic 2. United States. 153.6 ,’,’~,, ~,;~‘~‘‘~,;‘~‘~‘‘~‘‘~;;‘~‘~#i‘~‘‘~:~‘~‘‘‘!~‘‘~‘~‘~‘‘:~‘*::‘‘‘‘~‘i,‘‘‘~:~*‘~,~:~ 19.5% 3. Japan 73.3 ~:,,~,~,,~z~,,,~&,~,~,~;,; 9.3% 4. Hong Kong 70.9 ................... ;~’~’~ ...... 9.0% 5. Philippines 45.1 ;:~,~4:~,~,:z4~,:,~,: 5.7% 6. Thailand .............. 33.5 ~:~,~ 4.2% 7. Canada 27.1 ~;#~:?;~; 3.4% & Singapore ............. 23.5 :~;~, 9. Indonesia ............. 17.1 2.2% 10. Australia .............. 16.0 2.0% 11. Vietnam ............... 15.2 1.9% 12. Malaysia .............. 14.5 1.8% 13. United Kingdom ........ 10.7 1.4% 14. 6ermany .............. 9.8 ;~ 1.2% 15. Rep. of Korea .......... 9.4 1.2% 16. Russia ................ 6.4 0.8% 17. NewZealand .......... 6.5 ~; 0.8% 18. France ................ 5.6 ~: 0.7% 19. Macau ................ 4.1 0.5% Other ................. 48.7 ~t~-;,z,:,,~,,,,,,,~,-- 6.2% Total 789.0 © TeleGeography, Inc. t998 National Traffic Balance M iTl FY 1995/96 FY 1996/97 FY 1997/98 Incoming 545.3 736.8 842.2 0utg0ing 592.8 674.0 789.0 Surplus (Deficit) (47.5) 62.8 53.2 Total Volume 1,138.1 1,410.8 1,631.2 Note: Mi]q" is Minutes of Telecommunicatzons Traffic. Data are ~n mllhons of minutes of pubhc switched traffic. Rscal year ends 31 March. ©TeleGeography, Inc 1998 255 TeleGeography 1999 © TeleGeography, Inc. 1998 Thailand Largest Telecommunications Routes, 1997 Destination Millions of Mil-rs 1. ,Japan ................ 44.9 2. Singapore ............. 31.2 3. United States .......... 30.0 4. Hong Kong ............ 19.6 5. Taiwan ............... 15.7 6. United Kingdom ........ 15.3 7. Australia .............. 12.7 8. China ................. 12.3 9. Germany .............. 10.9 10. Rep. of Korea .......... 7.2 11. Myanmar ............. 6.7 12. India ................. 6.6 13. France ................ 5.6 14. Philippines ............ 5.5 15. Indonesia ............. 5.1 16. Italy .................. 3.7 17. Switzerland ........... 3.5 18. Vietnam ............... 3.2 19. Canada ............... 2.7 20. Netherlands ........... 2.7 Other ................. 33.4 Total 278.4 © TeleGeography, Inc, 1998 National Traffic Balance MiTr 1995 1996 1997 coming 277.7 376.2 408.5 Outgoing 218.8 247.~ 278.4, Surplus (Deficit) 58.9 128.7 130.1 Total Volume 496.5 623.6 686.9 Note: Mi]-I" is M=nutes of Telecommunications Traffic. Data are m milhons of minutes of pubhc switched traffic. Data based on billing po=nt of traffic. Data does not include cross-border traff=c w=th Laos and Malays=a © TeleGeography, Inc. 1998 © TeleGeography, Inc. 1998 TeleGeography 1999 Turkey Largest Telecommunications Routes, 1997 Destination Millions of Mil"rs 1. Germany .............. 172.5 2_ United Kingdom ........ 38.5 3. France ................ 29.2 4. Netherlands ........... 24.1 5. United States .......... 24.1 6. Russia ................ 23.9 7. Italy .................. 18.1 8. Switzerland ........... 15.0 9. Austria ............... 14.4 10. Romania .............. 13.4 11. Belgium ............... 11.4 12. Bulgaria .............. 10.9 13. Greece ............... 8.5 14. Azerbaijan ............ 8.1 15. Ukraine ............... 8.1 16. Iran .................. 6.0 17. Sweden ............... 5.5 18. Saudi Arabia .......... 4.9 19. Israel ................. 4.9 20. Denmark .............. 4.4 Other ................. 111.7 Total 557.5 Percentage of Outgoing Traffic ~:~i!~J! 4.3% ~:~,,~:~ 4.3% :~, 4.3% 3.2% 2.7% ; 2.0% ~;~ 2.0% 1.5% 1.5% ;~: 1.o% ~:~ o.~% o. o/o 0.8% © TeleGeography, Inc. 1998 National Traffic Balance MiTT 1995 1996 1997 Incoming 705.0 755.0 836.0 373.6 473.4 557.5 Outgoing Surplus (Deficit) 331.5 281.6 278.5 Total Volume 1,078.6 1,228.4 1,393.5 Note: MiT[" zs Minutes of Telecommumcat~ons Trafftc. Data are m millions of minutes of publzc swztched traffzc © TeleGeography, Inc. 1998 ¯ ......... TeleGeography 1999 © TeleGeography, Inc. 1998 ( Ukraine Largest Telecommunications Routes, 1997 Destination Millions of MiTrs Percentage of Outgoing Traffic Z Belarus ............... 26.0 5.3% ~:~"~’;’:~ 4.1% 3. Moldova .............. 20.0 4. Germany .............. 15.9 i~i~i!i,,’ii 3.3% 5. Poland ................ 10.7 ii~!~ 2.2% 6. Armenia .............. 7.3 1.5% 7. United States .......... 6.7 ~i"~. 1.4% 8. Kazakhstan ............ 5.4 !~ 1.1% 9. Czech Republic ........ 4.9 ~i:. 1.0% 10. Latvia ................. 4.7 ii~ 1.0% 11. Uzbekistan ............ 4.6 ii 0.9% 12. United Kingdom ........ 4.5 ii~ 0.9% 13. Georgia ............... 4.1 iii 0.8% 14. Azerbaijan ............ 4.1 i~ 0.8% 15. Hungary .............. 4.1 ;; 0.8% 16. Turkey ................ 3.5 ~" 0.7% 17. Italy .................. 3.5 i~ 0.7% 18. Israel ................. 3.4 i:’ 0.7% 19. Lithuania .............. 3.3 !. 0.7% 20. Bulgaria .............. 3.0 ; 0.6% Other ................. 50,4 ii ,~;~,-’~,,~,;’~,~-,~ 10.4% Total 486.8 © TeleGeography, Inc. 1998 National Traffic Balance Mi-Fr 1995 1996 1997 !n, coming n.a. n.a. n.a. 0 utg oi~g ~301¯̄ .8 340.8 486.8 Surplus (Deficit) n.a. n.a. n.a. Total Volume n.a. n.a. n.a. Note: MiTr is M=nutes of Telecommunicatzons Traffic. Data are in milhons of m~nutes of pubhc sw=tched traffic. Data for 1995 and 1996 ~nclude traffic to other member states of the Commonwealth of Independent States only. See page 174 for a matrix of traff=c from other CIS member states. © TeJeGeograph~/, Inc. 1998 258 © TeleGeography, Inc. 1998 TeleGeography 1999 United Arab Emirates © TeleGeography, Inc. 1998 National Traffic Balance MiTT 1994 1995 1996 Incoming n.a. n.a. n.a. Outgoing 428.2 503.6 589.3 Surplus (Deficit) n.a. n.a. n.a. Total Volume n.a. n.a. n.a. Nete: M~" ~s Minutes of Telecommunications Traffic. Data are ~n millions of minutes of public switched traffic. ©~leGeography, lnc, 1998 259 TeleGeography 1999 © TeleGeographyo Inc. 1998 United Kingdom---Outgoing Largest Telecommunications Routes, FY 1997/98 Destination Millions of Mi’l’rs 1. United States .......... 735.8 2. Ireland ................ 658.6 3. Germany .............. 504.5 4. France ................ 502.0 5. Spain ................. 270.6 6. Italy .................. 270.1 7. Netherlands ........... 223.2 8. Australia .............. 175.2 9, Canada ............... 153,9 10. Belgium ............... 143.9 11. Switzerland ........... 125.5 12. Greece ............... 103.6 13. Hong Kong ............ 98.0 14. Turkey ................ 94.7 15. South Africa ........... 86.9 16. Sweden ............... 79.3 17. Poland ................ 78.7 18. India ................. 77.8 19. Denmark .............. 74.2 20. Portugal .............. 73.9 Other ............... 1,269.4 Total 5,799.5 Percentage of Outgoing Traffic ~i~,~~ 12.7% 8.7O/o 4.7 % ~.~% ~-~ 1.8% ;~;~ 1.4% ~t,~ 1.3% 1.3% ~;~: 1.3% © TeieGeograpb¥, inc. t998 Natioual Traffic Balance MiFr FY 1995/96 FY 1996/97 FY 1997/98 Incom!,ng 4,021 4,360.0 n.a. Outgoing 4,016 4,569.2, 6,600 Surplus(,Deficit) 5 ,(209.2) n.a. Total Volume 8,037 8,929.2 n.a. Note: Mrlqis Minutes of Telecommunications Traffic. Data are in millions of minutes of public switched traff=c. Data based on billing point of traffic. Outgoing total for FY 1997/98 includes pubhc switched and ISR traffic data from all carriers and may include some traffic refiled via the U K., thus overstating U.K.-origmated volumes FY 1997/98 route data and totals from previous years are for BT and C&W Communicatmns only. © TeleGeography, Inc. 1998 © TeleGeography, Inc. 1998 Te~eGeography 1999 United Kingdom--lncoming Largest Telecommunications Routes, FY 1997]98 Destination Millions of MiTts Percentage of O~going Traffic 1. United States. 1,229.0 ............ ~’,~-’,-’--~’-,’ ,~;’,"-"~,~ ......... ?~’,-’," ........... ~.,’~," ............ : -’,’~,,,,-~-",~ ~,’-,,~’,--,~ ....... 24.4°/o 2. Ireland ,479.3 ~,~,,.~,.~x~,;.-~-;:~ ,,;,~,~,:~,’,’,,~,:,:~:-’,,/, 9.5% 3. France ................ 405.1 ’,’?~’~’~!~;%;~,"~?!~}~#~ 8.0% 4. 6ermany .............. 376.1 ‘~.‘:~:~‘‘~‘‘~‘~‘~,‘‘~‘:~‘‘~‘,~,‘‘‘~,,:‘;‘~:~‘~‘‘‘‘‘‘~:~‘‘~,~,~4~‘~‘;~#~:~ 7.5% Spain 218.7 ~’~" "~’;~’~"~;"’ 4.3% ................. Australia .............. 201.6 ~-~:’~,,~:’~ 4.0% Italy .................. 182.7 ~:2~,,?,~,~& 3.6% Canada ............... 146.6 ~:~:~&~ 2.9% Sweden ............... ~2a.~ ~’~’*"-~ 2.2% Swi~erland 111.0 Belgium ............... 10&6 ~,~) 2.2% South Bfrica ........... 7a.2 ~enmark .............. 57.a ~,~ ~.~% da~an ................ 50.4 .............. 5.a Hon~ Kon~ ............ 423 )~~ 0.aN Other ................. 831.8 ~~;:}~,?~i~;,~?;~:;~’~;,~;"’5,2;;~,~£~;52~,)~/~~ ~i~::£-,~1) 16.5% Total 5,~.8 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17, 18. 19. 20. © TeleGeography, Inc. 1998 TeleGeography 1999 © TeleGeography, Inc. 1998 U.K. Top 100 Correspondents October December 1997 October December 1997 Country Outgoing MiTT Incoming MiTT Country Outgoing MiTT Incoming MiTT Algeria 4.8 2.0 Libya 1.6 1.2 Argentina 2.2 1.2 Lithuania 1.2 0.2 Armenia 1.4 0.5 Luxembourg 6.0 3.8 Australia 55.9 59.3 Malaysia 6.3 6.2 Austria 19.8 8.4 Malta 3.6 2.8 Bahrain 2.3 3.1 Mauritius 2.8 1.1 Bangladesh 5.6 0.9 Mexico 2.7 1.3 Barbados 2.2 1.1 Monaco 1.7 1.4 Belgium 43.8 32.5 Morocco 9.3 1.5 Bermuda 1.3 1.7 Myanmar 1.2 0.6 Bosnia 1.4 0.6 Netherlands Antilles 2.3 2.1 Brazil 5.8 4.1 Netherlands 61.9 49.0 Bulgaria 2.2 0.9 New Zealand 12.1 12.8 Canada 42.6 38.9 Nigeria 11.9 3.7 Chile 2.5 1.0 Norway 17.6 16.9 China 8.4 3.0 Oman 2.1 1.5 Colombia 2.8 1.0 Pakistan 22.5 7.1 Congo 2.1 0.5 Philippines 4.0 1.6 Cote d’lvoire 2.2 0.3 Poland 20.2 8.0 Croatia 3.6 1.5 Portugal 19.4 11.0 Cyprus 9.3 7.8 Qatar 2.2 1.6 Czech Republic 8.9 4.6 Romania 2.9 1.7 Denmark 22.1 15,2 Russia 19.8 8.5 Ecuador 1.1 0,2 Saudi Arabia 15.3 12.5 Egypt 7.2 2.6 Senegal 2.9 0.2 Ethiopia 2.2 0.4 Sierra Leone 1.5 0.2 Finland 9.2 7.4 Singapore 13.0 6.1 France 147,4 101,4 Slovak Republic 3,9 15,0 Germany 151.3 98.6 Slovenia 1.3 0.8 Ghana 5.7 2.4 South Africa 29.6 18.9 Gibraltar 2.1 1.5 Spain 73.4 52.0 Greece 29.2 21.7 Sri Lanka 4.3 1.0 Guyana 2.5 0.3 Sweden 26.8 41.2 Hong Kong 25.1 12.9 Switzerland 35.6 27.6 Hungary 8.2 4.1 Syria 2.1 0.5 Iceland 1.6 1.2 Taiwan 3.4 2.3 India 25.6 11.8 Tanzania 1.3 0.5 Indones=a 4.2 3.0 Thailand 4.3 3.8 Iran 4.0 2.2 Trinidad & Tobago 3.7 1.6 Ireland 190.3 128.5 Tunisia 3.9 0.7 Israel 10.1 4.3 Turkey 24.0 10.1 Italy 81.4 46.7 Tuvalu 1.5 0.0 Jamaica 6.7 1.3 UAE 9.5 11.5 Japan 19.4 12.8 Ukraine 3.2 1.1 Jordan 3.4 1.0 United States 264.0 332.8 Kenya 3.7 1.6 Uzbekistan 3.0 0.4 Korea, Rep. of 4.8 3.7 Vietnam 7.5 0.9 Kuwait 4.1 2.0 Yemen 2.3 0.3 Latvia 2.0 0.8 Yugoslavia 2.8 1.3 Lebanon 4.4 0.6 Zimbabwe 2.4 2.2 Note: Data are m=ll=ons of minutes of pubhc switched and International S=mple Resale (ISR) traff=c for the seventeen largest U.K. =nternat=onal carriers during the October- December quarter of 1997 Route data may include some calls refiled via the U.K., thus overstat=ng actual U.K.-origmated traffic. Source Office ofTelecommumcations (OFTEL) ©TeleGeography, Inc 1998 © TeleGeography, Inc. 1998 TeleGeography 1999 United States--Outgoing Largest Telecommunications Routes, 1997 Destination Millions of Mi’l-rs Percentage of Outgoing Traffic 1. Canada .............. 3,926.6 ~z~+:~w;~-~;~;::~?~:,~~ 17.3% Mexico .............. 2,757.6 ,~ .......... ~,~,,~,~ ........... ....... 3. United Kingdom ...... 1,532.6 ~;77:+’;~;::~77~:~:~7~D~;~:~,~{ &8% 4. Oerman7 .............. 994.3 ~’~i,~77~7~,f~ 4.470 5. Japan ................ 830.5 ~-~:~;~7~’77~S~ 3.7% 6. Hong Kong 673.2 ~,,~ 7. India ................. 579.4 ,~ 2.6% 8. France ................ 500.0 ~g~7,~ 2,2% 9. Brazil .497.3 ~:~ 2.2% ................ ~:~, ~’~ 2.1% 10. Italy .................. 475.6 :~ 11. Rep. of Korea .......... 421.9 1.9% 12. Philippines ............ 417.2 1.8% 13. Dominican Republic .... 392.1 1.7% 14. China ................. 388.1 1.7% 15. Australia .............. 383.5 ~,~ 1.7% 16. Taiwan ............... 379.1 ~,~ 1.7% 17. Jamaica .............. 262.6 ~,~ 1.2% 18. Colombia .............. 260.5 ~~ 1.1% 19. Argentina ............. 228.1 :~,, 1.0% 20. Swi~erland ........... 225.8 ~’, 1.0% -~ ~;;; "’~V~,~ ~: ~,;~’~~,," L "~ "’: ~;’~::" ,?~"L~’~"~: ~"~":’~?~’~’~ 29.0% Tetal ~,700.1 © TeleGeography, Inc. t998 National Traffic Balance Mi-FF 1995 1996 1997 In c omi n g 7,004.6 8,194.9 9,219.3 Outgoing 15,837.1 19,119.1 22,700.1 Surplus (Deficit) (8,832.5) (10,924.1) (13,480.8) Total Volume 22,841.7 27,314.0 31,919.4 Note: MiTT is Minutes of Telecommumcations Traffic. Data are mflhons of minutes of public switched and Internat=onal Szmple Resale (ISR) traffic. Data based on billing point of traffic. Carriers and traff=c from outside the U.S states and Puerto Rico (e.g., Virgin Islands, Guam) are excluded © TeleGeography, Inc. 1998 243 TeleGeography 1999 © TeleGeography, Inc. 1998 ( United States Incoming Largest Telecommunications Routes, 1997 Destination Millions of MiTTs Percentage of Outgoing Traffic 2_ United Kingdom ........ 950.2 ~ii1%:=::,’.,~,,%~i~; 10.3% 3. Mexico 946.5 ~",~’’":,~,~1>~:;;:~;-~:~%%~!~". 10.3% 4. Japan ................ 339.4 ~? 3.7% 5, Germany. 324.6 %~’~:~’ 3.5% ............. 6, France 216.6 ~:~ 2.3% 7. Rep. of Korea 203.5 ~~,, 2.2% 8. Taiwan 174,3 ............... ~: 1.9% 9. Israel ................. 161.7 ~% 1.8% 10. Brazil ................. 159.6 1.7% 11. Australia .............. 123.5 ~~?~ 1.3% 12. Italy .................. 121.9 1.3% 13. Swi~erland ........... 108.1 ~:~ 1.2% 14, Netherlands ........... 105.8 ~’~ 1.1% 15. Spain ................. 100.6 1.1% 16. Dominican Republic .... 95.6 1.0% 17. Spain ................. 88.1 1,0% 18. Hong Kong ............ 82.4 0.9% 19. Chile ................. 75.1 0.8% 20. Colombia .............. 68.7 0.7% Other ,1,599.7 ,’:~,,~,~-,,~,~ ~’-~:~,-~,,-,,,~,,~,’,~:,,~:~’~:, 17.4% Total 9,219.3 © TeleGeography, Inc. 1998 244 © TeleGeography, Inc, 1998 TeleGeography 1999 U.S. Top 1 O0 Correspondents Country Antigua & Barbuda 25.6 52.6 Argentina 221.4 228.1 Aruba 14.4 16.8 Australia 282.5 383.5 Austria 53.9 58.8 Bahamas 61.8 66.4 Bahrain 12.4 11.7 Bangladesh 46.3 59.0 Barbados 33.2 38.4 Belgium 113.5 120.7 Belize 14.9 14.5 Bermuda 35.8 38.7 Bolivia 29.6 34.5 Bosnia 23.0 20.3 Brazil 372.8 497.3 Bulgaria 12.0 12.2 Canada 3,463.4 3,926.6 Cayman Islands 19.3 21.9 Chile 79.0 115.5 China 298.4 388.1 Colombia 284.9 260.5 Costa Rica 67.4 75.3 Croatia 17.3 22.5 Cuba 103.0 118.9 Czech Rep. 25.8 26.9 Denmark 60.7 75.3 Dominica 10.5 16.4 Dominican Rep. 416.4 392.1 Ecuador 156.6 156.7 Egypt 89.7 111.2 El Salvador 141.3 154.6 Ethio pia 18.3 18.0 Finland 26.7 29.4 France 442.1 500.0 (3ermany 781.8 994.3 (3hana 32.8 50.0 (3reece 90.6 97.4 Grenada 14.0 16.9 Guatemala 127.7 126.1 Guyana 86.7 59.7 Haiti 78.7 92.2 Honduras 89.8 111.1 Hong Kong 539.5 673.2 Hungary 36.1 41.0 India 423.9 579.4 Indonesia 92.1 119.2 Iran 55.5 46.0 Ireland 121.0 134.8 Israel 238.6 216.5 Italy 334.1 475.6 Outgoing Mi’l’]" Incoming Mi’l’]" Outgoing Mi’l’]" Incoming Mi’l’]" 1996 1997 1996 1997 Country 1996 1997 1996 1997 6,7 6.7 Ivory Coast 12.8 14.8 2.6 4.4 22.4 38.8 Jamaica 219.4 262.6 50.1 50.2 6.1 6.6 Japan 703.1 830.5 346.5 339.4 155.8 123.5 Jordan 43.0 57.1 4.8 5.4 24.6 24.9 Kenya 25.4 24.2 2.8 3.5 46.0 46.5 Korea, Rep. of 382.5 421.9 159.1 203.5 4.3 4.7 Kuwait 40.8 48.3 8.5 8.8 1.5 4.5 Lebanon 54.0 38.6 3.5 4.4 12.0 12.8 Malaysia 67.8 83.4 19.4 26.4 43.2 49.7 Mexico 2,380.9 2,757.6 948.1 946.5 3.7 4.2 Montserrat 4.5 12.4 0.4 0.4 29.5 29.9 Morocco 13.3 13.8 4.8 5.8 4.8 3.6 Netherlands 207.1 222.3 85.6 105.8 1,3 2.6 N L Anti ll es 33.9 42.3 11.1 6.0 123.5 159.6 New Zealand 58.9 107.8 29.3 40.6 1.7 1.3 Nicaragua 45.2 50.3 5.2 6,1 2,845.9 3,173.3 Nigeria 75.0 101.8 9.2 8.7 12.8 12.1 Norway 50.9 66.5 34.2 52.5 52.0 75.1 Pakistan 133.6 162.9 10.0 10.5 58.3 61.7 Panama 69.5 68.2 17.9 19.3 61.4 68.7 Paraguay 14.4 10.4 2.6 3.3 24.0 26.4 Peru 144.2 164.9 25.5 24.5 5.0 6.0 Philippines 356.1 417.2 50.2 32.2 0.8 1.2 Poland 160.4 170.9 27.1 27.3 8.0 9.2 Portugal 44.6 54.8 14.1 15.1 28.2 27.9 Romania 30.6 28.9 5.1 5.2 2.2 1.9 Russia 99.9 102.9 31.0 32.9 125.1 95.6 St, Lucia 12.7 14.7 3.0 3.2 15.1 17.8 St, Vincent 22.9 16.7 1.7 1.7 9.6 10.7 Saudi Arabia 101.2 111.4 27.2 32.1 9.9 12.8 Senegal 17.0 18.9 1.6 1.7 1.7 1.6 Singapore 148.7 197.8 53.5 63.5 17.0 17.8 South Africa 97.4 110.6 29.3 31.2 203.6 216.6 Spain 147.3 190.8 61.5 88.1 313.0 324.6 Sri Lanka 14.3 16.2 1.9 2.0 4.0 5.2 Sweden 103.2 140.0 70.4 100.6 33.0 35.0 Switzerland 181.0 225.8 91.9 108.1 2.5 2.4 Syria 18.1 15.2 3.1 2.9 11.8 15.5 Taiwan 321.7 379.1 112.2 174.3 15.8 12.5 Thailand 114.5 117.7 29.7 26.2 7.1 5.1 Trinidad & Tobago 86.4 94.4 26.6 26.2 9.4 11.5 Turkey 58.4 62.0 25.9 28.3 96.3 82.4 Ukraine 36.2 44.4 6.5 6.9 14.5 15.2 UAE 53.8 52,3 26,2 28.4 50.1 49.7 United K~ngdom 1,226.4 1,532.6 737.7 950.2 24.8 28.1 Uruguay 24.9 26.2 5.6 7.5 15.3 16.0 Venezuela 154.2 219.8 58.5 63.9 53.4 49.0 Vietnam 114.2 157.0 5.2 4.0 78.9 161.7 Yemen 21.2 23.9 1.2 1,2 114.4 121.9 Yugoslavia 25,5 15.7 9.3 5.5 Note All data are rnillions of m=nutes of public switched and International Simple Resale (ISR) traffic. Because data are based on the billing point of the traffic, route data may not exactly correspond w=th traffic volumes as measured by the originating point of traffic (see Methodology on page 268). Carriers and traffic from outside the U.S. states and Puerto Rico (e.g., Virgin Islands, Guam) are excluded. © TeleGeography, Inc 1998 TeleGeography 1999 © TeleGeographg, Inc. 1998 Tra ffi ¢ by Ca rri e r Market Share of International Traffic by Route, 1997 U.S. Billed Traffic Foreign Billed Traffic AT&T MCI Sprint WorldCom AT&T MCI Sprint WorldCom 36.68 31.87 16.91 7.94 43.40 30.37 15.37 7.02 14.75 19.64 8.89 6.71 11.39 32.08 24.23 10.79 47.17 14.93 17.51 18.57 53.45 16.17 19.62 10.61 51.02 22.95 14.15 1.23 44.10 20.33 28.56 1.50 38.83 24.90 10.28 4.32 54.32 27.18 12.B0 4.63 48.03 19.81 12.16 15.50 48.68 20.77 9.33 20.26 18.27 21.45 10.93 18,22 26,61 42,81 6.85 14.51 53.60 17.49 11.57 14.48 15.20 70.82 9.15 4.82 42.19 27.42 12.73 12.66 39.18 28.56 16.22 15.93 43.73 28.32 9.32 10.64 45.63 30.71 9.93 9.95 20,12 46.80 22.64 5.77 27.88 38.86 27.64 3.97 34,81 51.27 8.12 0.21 46.60 43.16 9.03 0.00 52,50 23.35 15.21 5.03 59.70 23.48 13.48 3.34 44.91 27.26 15.60 10.43 54.06 19.33 14.11 12.22 48,11 29.21 9.75 9.12 61.16 29.54 8.13 1.18 36,32 42.80 10.25 6.72 38.68 39.44 14.14 5.66 42,69 27.84 13.95 7.88 56.26 28.91 11.00 3.75 58,12 24.71 9.63 7.40 48.05 18.53 14.78 9.13 41.58 22.83 11.76 6.82 36.22 41.54 14.52 2,22 56.34 16.48 9.45 2.32 72.67 18.46 8.86 0.00 70.29 17.91 6.40 0.98 46.28 29.29 16.08 5.70 36.51 21.13 9.26 8.11 44.96 14,53 18.31 14.96 34.77 35.69 15.78 5.50 48.44 31.49 13.50 4.07 49.66 22.08 11.38 7.54 53.27 21.02 14.30 7,21 37.25 36.90 11.48 5.88 48.50 19.00 19.71 6.99 Argentina Australia Brazil Canada China Colombia Dominican Republic Ecuador France Germany Hong Kong India Israel Italy Jamaica Japan Rep. of Korea Mexico Netherlands Philippines Poland Switzerland Taiwan United Kingdom Venezuela © TeteGeography, Inc. 1998 Traffic Carried by Second Tier U.S. Facilities-Based International Carriers, 1997 Outbound inbound Top Outbound Carrier Minutes (m) Minutes (m) Routes (Minutes) Pacific Gateway Exchange (PGE) 742.9 95.5 Australia (133.1) WorldxChange 430.1 -- Mexico (58.0) Fonorola 405.4 169.9 Canada (405.4) Facilicom 223.2 10.7 Germany (40.4) STAR Telecommunications 116.2 19.2 China (59.7) RSL Corn USA 94.4 -- Dominican Rep. (85.2) Viatel 58.5 -- Spain (11.4) Primus 50.4 4.2 Australia (32.9) Cable & Wireless, Inc. 44.0 -- U.K. (20.6) Tricom USA 29.2 -- Dominican Rep. (29.1) Total 2,194.3 299.5 Note: All data in millions of minutes based on billing point of call. Carriers and traffic from outside the U.S. states and Puerto Rico (e.g., Virgin Islands, Guam) are excluded. Data includes traffic carried on tmernatinnal Simple Resale (ISR) facilities. © TeleGeography, Inc. 1998 246 © TeleGeography, Inc. 1998 TeleGeography 1999 U.S. Resale Traffic Top 15 U,S. Switched Service Resale Carriers, 1997 Outbound Share of Rank Resale Carrier Minutes (m) Outbound Resale Minutes 1. Cable &Wireless 1,228.1 17.5% 2. PT-1 Communications 714.6 10.2% 3. STAR Telecommunications 636.8 9.1% 4. WorldCom 600.1 8.6% 5. RSL Corn 321.1 4.6% 6. Teleglobe USA 301.6 4.3% 7. IDT 300.5 4.3% 8. LCI International 297.1 4.2% 9. Telegroup 276.9 4.0% 10. Pacific 6ateway Exchange (P6E) 209.0 3.0% 11. Access Authority 189.3 2.7% 12. Primus 158.2 2.3% 13. ACC Long Distance 134.5 1.9% 14. Sprint 124.2 1.8% 15. Startec 120.9 1.7% Top 15 Total 71.0% 5,045.0 Note: All data in millions of minutes based on billing point of call. Carriers and traffic from outside the U.S. states and Puerto R~co (e.g., V~rgin Islands, Guam) are excluded. Switched service resale carriers are resellers of the internatzonal switched voice services that are actual- ly provided by other, facilities-based carriers. The 7.9 billion minutes of U.S. switched resale traffic are thus included in the U,S, outgo- ~ng traffic data on page 243. © Tele6eography, Inc. t998 Share of U.S. International Switched Traffic Resold, 1991-1997 1993 1994 1995 1996 1997 © TeleGeography, Inc. 1998 247 TeleGeography 1999 © TeleGeography, Inc. 1998 t Uruguay Largest Telecommunications Routes, 1997 Destination Percentage of Outgoing Traffic 1. ~,~ .......... ............. ,, ~,,, ,~,~ ,~,~, ~-,,,~,~,,~,~ ........ :~,~.,,,,~,, -~,- ,~,,~,~,~, ,~,,~ ~, ~,, 52.6% 2. ~-~ :: ~-:,~ ~"::’~ 13.2% 6. 7. 8. 9. 10. 11. 12. 14. 15. 20. Millions of MiTl’s Argentina .............. 35.9 Brazil .................. 9.0 United States ........... 7.0 Spain .................. 2.9 Chile ................... 1.6 Paraguay .............. 1.4 Italy ................... 1.2 France ................. 0.8 Germany ............... 0.7 Mexico ................ 0.6 Canada ................ 0.6 United Kingdom ......... 0.6 Israel .................. 0.5 Venezuela .............. 0.5 Switzerland ............ 0.4 Australia ............... 0.4 Peru ................... 0.4 Colombia ............... 0.3 Bolivia ................. 0.2 Sweden ................ 0.2 Other .................. 3.1 Total 68.4 ~i 2.4% ~;~’" 2.1% ~il 1.8% 1.1O/o 1.o% 0.9% 0.9% o.8o/o 0.7% 0.7% 0.5% 0.5% 0.5% 0.5% 0.3% 0.3% © TeleGeography, Inc. t998 National Traffic Balance MiTT 1995 1996 1997 Incoming ,, 73.9 80.1 93.7 Outgo!rig 49.9 54.5 68.4 Surplus (Deficit) 24.0 25.6 25.3 Total Volume 123.8 134.5 162.1 Note: Mi’Fi" zs M=nutes of Telecommumcations Traffic. Data are in millions of minutes of public switched traffic. Data based on bilhng point of traffic. 248 © TeleGeography, Inc. 1998 © TeleGeographg, Inc. 1998 TeleOeography 1999 Uzbekistan Largest Telecommunications Routes, 1997 Destination Percentage of Outgoing Traffic 6. 7. 8. 9, 10. 11. Millions of MiTrs 1. Russia ................ 29.6 2_ Kazakhstan ............ 7.9 3. Kyrgyzstan ............ 3.4 4. Ukraine ............... 3.1 Tajikistan ............. 2.1 Turkrnenistan .......... 1.2 Belarus ............... 0.8 Azerbaijan ............ 0.6 Armenia .............. 0.5 Georgia ............... 0.3 Moldova .............. 0.2 Other ................. 13.6 Total 63.1 5.3O/o ~;: 3.3% ~3.:..: 2.0% ~:~: 1.3% ~:~’,, ~;~ o.~% 0.4% 0.2% © TeleGeography, Inc. 1998 National Traffic Balance MiTT 1995 1996 1997 Incoming n.a. ,,, n.a. n.a. Outgoing n.a. 54.2 63.1 Surplus, (,Deficit) n.a. n.a. n.a. Total Volume n.a. n.a. n.a. Note: MtTT ts M~nutes of Telecommunications Traffic Data are in millions of minutes of public switched traffic. Data for 1996 include only traff=c to other member states of the Commonwealth of Independent States The "Other" cate- gory may =nclude routes to non-CIS member states, including an estimated 1.5 million minutes to the U.S. and 1.6 lion minutes to the U.K., that rank among the top destmatzons for outgoing traffic ©TeleGeography, Inc 1998 TeleGeography 1999 © TeleGeography, Inc. 1998 ( Venezuela Largest Telecommunications Routes, 1997 Destination Millions of MiTrs 1. United States ........... 74.8 2. Colombia ............... 16.0 3. Canada ................ 13.5 4. Spain .................. 8.0 5. Italy ................... 6.3 6. United Kingdom ......... 4.7 7. Peru ................... 3.4 8. Mexico ................ 2.7 9. Brazil .................. 2.6 10. Argentina .............. 2.5 11. Ecuador ............... 2.2 12. Portugal ............... 2.1 13. France ................. 1.9 14, Deminican Republic ..... 1,9 15. Chile ................... 1.9 ii"~ 1.2% 16. Germany ............... 1.5 ’:, 0.9% 17. Lebanon ............... 1.4 i,i, 0.9% 18. Cuba .................. 1.3 !!:~ 0.9% 19. Trinidad &Tobago ...... 1.3 ii’~ 0.8% 20. Panama ................ 1.0 !~: 0.8% Other .................. 8.3 i~!ii:ili~!,’~i"~:i~, 5.2% Total 159.2 Percentage of Outgoing Traffic ,: ~,,~;: ,~’,’-~,,,~ ,"~:~ ,’-~-,~ 10.0% 3.0% 2.1% 1.7% 1.7% 1.6% 1.4% 1.3% :;;,~ 1.2% 1,2% © TeleGeography, Inc. 1998 Natioaal Traffic Balance MiTT 1995 1996 1997 Incoming 186.6 228.8 286.9 Outgoing 129.1 139.0 159.2 Surplus (Deficit) ST.~ ¯ 89.8 12717 Total Volum e 315.7 367.8 446.1 Note: Mi’FI" is Minutes of Telecommunications Traffic. Data are m millions of m~nutes of pubhc switched traffic. Data based on bilhng point of traff=c. 250 ©TeleGeography, Inc 1998 © TeleGeography, Inc. 1998 TeleGeography 1999 Vietnam Largest Telecommunications Routes, 1995 Destination Millions of Mi~’s Percentage of Outgoing Traffic 1. China.. 8.0 ~!~,~,~" ~,~,,’-"~,’~’,"~ ......... ~,*,,~,,~;,~, ~,~ ........ 11.5% 2. Taiwan ................. 3.0 :~;,,ozll;;l££+ti;,l~,z~;~;~;~:~ 5.7% 3. United States ........... 3.0 ~-~’:~t~,;~:1-~-:?;’;~];)}{~:; 5.7% 4. Philippines. 2.0 ;’="~:+"’> .......... 3.8% 5. Hon9 Eon9 .............. 2.0 ~:::~,~:,~:~a~’~............... 3.8% 6. France ................. 2.0 ,:’:~,s~,,~5~:~ 3.8% 7. Rep. of Korea. 1.5 ....... ~,,~¢~,=,~+ 2.9% &~,~:~ 2.9% 8. Singapore .............. 1.5 9. Thailand ................ 1.5 ~0. aermang ............... 1.0 Total 35.1 © TeleGeography, tnc, 1998 National Traffic Balance MiTT 1995 1996 1997 Incoming n.a. n.a. 294.6 O~tgoing 35.1 52.4 ,, 50.2 Surplus (Deficit) n.a. n.a. 244.4 Total Volume n.a. n.a. 344.8 Note: M~I-F ~s M~nutes of Telecommunications Traffic. Data are in mdhons of m~nutes of public switched traffic © TeleGeography, Inc. 1998 251 TeleGeography 1999 © TeleGeography, Inc. 1998 Yugoslavia Largest Telecommunications Routes, 1997 Destination Millions of MiTTs Percentage of Outgoing Traffic ~~i110.5% 8.9% 2_ Croatia ............... 22.7 3. Austria ............... 20.5 4. Switzerland ........... 19.3 5. Italy .................. 11.4 6. Slovenia .............. 11.3 7. Hungary .............. 8.2 8. Macedonia ............ 8.1 9. France ................ 8.0 10. Canada ............... 8.0 11. Greece ............... 6.6 12. United States .......... 6.3 13. United Kingdom ........ 5.4 14. Russia ................ 5.3 15. Sweden ............... 4.6 16. Australia .............. 3.3 17. Netherlands ........... 3.1 18. Turkey ................ 2.9 19. Romania .............. 2.5 20. Bulgaria .............. 2.4 Other ................. 9.9 Total 217.0 © TeleGeography, Inc. 1998 National Traffic Balance MrFI 1995 1996 1997 Incoming 296.0 325.7 332.0 Outgoing 212.8 237.2 ,,, 217.0, Surplus (Deficit) 83.2 88.5 115.0 Total Volume 508.8 562.9 549.0 Note: MrFF =s Minutes of Telecommunications Traffic, Data are =n milhons of minutes of pubhc sw=tched traffic. Data exclude traffic to Bosma. 252 © TeleGeography, Inc, 1998 © TeleGeographyo Inc. 1998 TeleGeography 1999 Global Traffic Review 1500 143.2 150 projected 124.5 1250 ~4 Fixed main lines 125 ._ ¯ Total international traffic 108.2 1000 100 .£ ~81.8 I i Ii, IliIlli’ ~o .. "’ I~ .~ II 1 I 0 0 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Note: Deta inctude outbound international traffic on public networks only. Projections assume 15 aercent traffic growth and 6 percent main line growm, Source: ITU, TeleGeogreph¥, Inc. © TeteGeography, Inc, 1998 Figure 2. International Traffic, Revenue, and Subscriber Growth Historical Trend Slow Growth Same Growth Fast Growth CAGR CAGR CAGR CAGR Indicator 1988 1997 1988-97 2001 1997-2001 2001 1997-2001 2001 1997-2001 Calls (Bn) 5.1 24.5 19.2% 46.8 17.2% 48.2 18.1% 53.5 21.2% Estimated call length (rains) 4.4 3.3 -3.0% 3.0 -2.7% 3.0 -2.7% 3.0 -2.7% Minutes (Bn) 22.3 81.8 15.5% 128.7 12% 143.1 15% 158.6 18% Per main line subscriber 47.2 103.3 9.1% 132.0 6.3% 143.1 8.5% 154.7 10.6% Per main line plus mobile 46.8 81.6 6.4% 91.6 2.9% 98.7 4.9% 106,1 6.8% Revenue (US$bn) 26.8 65.9 10.6% 77.5 4,0% 79.0 4.5% 79.9 4.7% Assumptions Price per MiTT(US$) 1.20 0.81 -4.3% 0.60 -7.1% 0,55 -9.1% 0.50 -11,3% Main lines (M) 473 792 5.9% 975 5.3% 1’000 6.0% 1’025 6.7% Mobile subscribers (M) 4.2 211 54.4% 430 19.5% 450 20.8% 470 22,2% Note: 1988-97 based on reported data. 1998-2001 based on ITU forecasts. Scenarios are as follows: 1. Slow Growth: Traffic growth slows but network infrastructure continues on current growth trend. 2. Same Growth: Continuing traffic growth rate of last five years, assuming faster network growth rate and faster rates of price-cutting. 3. Fast Growth: Faster traffic growth rate than last five years, assuming a faster network growth rate and faster rates of price-cutting, plus a significant component of new demand created by international traffic generated from mobiles. Sourcei tTU World Telecommunication Indicators Database and ITU estimates: © TeleGeogra phY; In¢i i998 253 TeleGeosraphy 1999 © TeleGeography, Inc. 1998 Figure 3. Intercontinental Traffic Flows, 1995 & 1997 1995 1997 ,! Key Traffic Flows 2 500 1,000 500 100 (Mdhons of MiTTs) Each band is proportional to the total annual traffic on the public network in both directions between each pair of countries. These maps shows all inter-conti- nental routes with an annual volume of more than 100 million minutes. The total volume of these routes in 1997 was 20.8 billion minutes, approximately 25.8% of global inter- national traffic. © ]eleGeography, Inc. 1998 254 © TeleGeography, Inc. 1998 TeleGeography 1999 International Traffic by Region Figure 4. Interregional Traffic Flows, 1997 5 All figures are g~ven in billions of Each band ~s proportional to the Arrows indicate d~rection Numbers ~n a mrcle indicate the Minutes ofTelecommumcations Traf- total annual traffic from one oftraffic between regions, total amount of international traf- fic for the public telephone network, regmn to another, fic for countries w~th~n a regmn. o Note: Data for interreg~enal traffic based on top 20 international routes for 90 countries, accounting for approximately 25 of global international traffic. © TeleGeograph¥, Inc. 1998 Figure 5. International Traffic by Origin, 1997 Oceania Africa 1.9% Latin America & Caribbean 4.1%~ © TeleGeography, Inc. 1998 Figure 6. Traffic Growth by Region, 1996-1997 25% 20% 15% Global Average 10% 5% 0% La:~n Amenca & Caribbean Europe Africa Asia Oceama North America © TeleGeography, Inc. 1998 255 TeleGeography 1999 © TeleGeography, Inc. 1998 Figure 7. European Telecommunications Traffic Flows, 1997 SWEDEN FINLAND POLAND © TeleGeography, Inc. 1998 7~%~ To Turkey Key All figures are given in millions of Minutes of Telecommunications Traffic (mM~3-fs), for the public telephone network. The map shows all intra-European routes with a t897 volume of more than 120 mMil-rs. Traffic Flows 600 400 200 Each band is propo~onat to the total annual traffic on the public telephone network in both directions between each pair of countries. (~Total Outgoing Traffic 5,OOO The area of each circle is proportzonal to the volume of the total annual outgoing traffic from each country. ,~*~-Balance of Traffic On routes where traffic in one direction accounts for more than 60% of the total, an arrow shows the dzrect=on most of the traffic flows. © TeleGeography, Inc, 1998 © Tele~eography, Inc. 1998 Figure 8. Latin American Telecommunications Traffic Flows, 1997 TeleGeography 1999 ELSALVAOOR COSTA RICA MICAP, AGUA COLOMBIA IRUGUAY © TeleGeography, Inc. 1998 Key All figures are given in millions of Minutes of Tetecommumcat=ons Traffic (mMiTTs), for the public telephone network. The map shows all routes within Latin America w~th a 1997 volume of more than 10 mM~-rTs. J~Traffic Flows 100 50 10 Each band is proportional to the tota( annuat traffic on the pubtic telephone network in both directions between each pair of countries. (~)Total Outgoing Traffic oo 5OO 1,OOO The area of each circle is propo~onat to the volume of the total annual outgo=ng traffic from each country. ~,, Balance of Traffic On routes where traffic in one direction accounts for more than 60% of the total, an arrow shows the direct=on most of the traffic flows. © TeteGeography, Inc. 1998 257 TeleGeography 1999 © TeleGeography, Inc. 1998 Figure 9. Asian Telecommunications Traffic Flows, 1997 © TeleGeography, Inc, 1998 AUSTRALIA ,Ksy All figures are given in millions of Minutes of Telecommunications Traffic (mMiTTs), for the public telephone neWvork. The map shows all intra-Asian routes with a t997 volume of more than 50 mMiTls. © TeleGeography, Inc. 1998 Traffic Flows 500 200 50 Each band is proportional to the total annual traffic on the public telephone neWvork in both directions between each pair of countries. OTotal Outgoing Traffic (~ t00 5O0 t,500 The area of each circle is proportional to the volume of the total annual outgoing traffic from each country. ~=~ Balance of Traffic On routes where traffic in one direction accounts for more than 60% of the total, an arrow shows the direction most of the traffic flows. ......... 7 1," © TeleGeography, Inc. 1998 TeleGeography 1999 International Traffic by Country Figure 10. International Telephone Traffic Growth for Selected Countries, 1996-1997 3O% 25% 2O% 15% lO% 5% Global Average -5% -lO% © l’eleGeography, Inc, 1998 Figure 11. Telephone Traffic Balances for Selected Countries, 1997 Countries with a traffic surplus Phihpp,nes ~!i~; ;iil !~,!~:~" Hong Kong Brazil Oominican Republic Turkey Vietnam ~o~u~al !~b.~L~ Romania El Salvador Italy Russia South Africa Iran [] 6ermany [] Kuwait Countries witb a traffic deficit Australia Switzerland United Kingdom ’ United States -2,000 -1,000 0 1,000 International Telephone Traffic Balance (Millions of Minutes) 2,000 © TeleGeography, Inc. 1998 259 TeleGeography 1999 © TeleGeography, Inc. 1998 Figure 12. International Traffic Indicators, 1997 International Traffic Outgoing Incoming Surplus(Deficit) Population Mi’l’r(out) Main Lines Mi’rr(out) (mMi’l’r) (mMi’l’r) (mMi’l’r) (m) per Capita (thous.) per Main Line Albania (a,c) 40.8 n.a. n.a. 3.3 12.4 87 469.0 Andorra 42.2 30.1 -12.1 0.1 659.4 32 1,318.8 Arg e nti n a 223.4 444.2 220.7 35.7 6.3 6,750 33.1 Armenia 48.8 n.a. n.a. 3.4 14.2 n.a. n.a. Australia (b) 1,510.0 1,250.0 -260.0 18.4 81.9 9,350 161.5 Austria (a,c) 995.5 957.7 -37.8 8.1 122.4 3,969 250.8 Bahamas 62.7 n.a. n.a. 0.3 227.2 96 653.1 Bahrain (a) 106.6 85.4 -21.2 0.6 176.7 152 701.3 B a n g la d e sh 46.9 187.0 140.2 125.3 0.4 297 157.9 Belarus (a) 148.6 185.2 36.6 10.4 14.3 2,313 64.2 Belgium (a) 1,340.0 1,420.0 80.0 10.2 131.8 4,769 281.0 Bolivia 22.7 69.3 46.6 7.7 3.0 535 42.4 Botswana (d) 30.8 22.7 -8.1 1.5 20.5 84 366.7 Brazil (a) 459.1 761.3 302.2 164.5 2.8 15,106 30.4 Brunei 35.6 22.0 -13.6 0.3 115.7 79 450.6 Canada (a) 4,286.3 4,635.1 348.7 30.3 141.3 18,460 232.2 Chile 242.0 n.a. n.a. 14.5 16.7 2,600 93.1 Chin a 1,631.8 n.a. n.a. 1,221.6 1.3 70,310 23.2 Colombia (a,d) 135.5 384.2 248.7 37.4 3.6 5,334 25.4 Costa Rica (a) 66.9 111.6 44.7 3.5 18.9 584 114.6 Cyprus (a,c) 154.4 115.2 -39.2 0.8 205.1 385 401.0 Czech Republic (a) 306.1 355.0 48.9 10.3 29.7 3,275 93.5 Denmark (a) 628.0 682.0 54.0 5.3 118.4 3,339 188.1 Dominican Republic 142.0 476.9 334.9 7.9 18.0 709 200.3 El Salvador (a) 34.3 168.2 133.9 5.7 6.1 325 105.5 Estoma (a) 66.3 67.0 0.7 1.4 46.2 469 141.4 Rnland 371.7 n.a. n.a. 5.1 72.4 2,866 129.7 France 3,545.0 3,609.0 64.0 58.6 60.5 33,700 105.2 French Polynesia 9.1 n.a. n.a. 0.2 39.0 52 175.0 G e rm a ny ( a, c) 5,333.1 5,206.0 127.1 82.1 65.0 45,200 118.0 Greece 593.7 634.6 40.9 10.6 55.9 5,328 111.4 Guyana 24.1 142.4 118.2 0.7 34.1 50 482.0 Hong Kong (a,b) 1,718.0 2,100.3 382.3 6.5 262.4 3,647 471.1 Hungary (a) 287.1 324.6 37.5 10.2 28.1 n.a. n.a. Iceland (a,d) 32.5 32.0 -0.5 0.3 120.5 168 193.5 India (a,c) 420.5 1,256.6 836.1 966.8 0.4 17,802 23.6 Indonesia (a) 351.6 456.0 104.4 209.8 1.7 4,982 70.6 I ran 160.7 130.2 -30.5 67.5 2.4 6,513 24.7 Ireland (a,b) 695.0 n.a. n.a. 3.6 192.7 1,500 463.3 Israel (a,d) 319.7 468.1 148.4 5.5 57.8 2,656 120.4 Italy (c) 2,351.9 2,475.1 123.2 56.8 41.4 25,698 91.5 Japan (b) 1,771.7 1,635.0 -136.7 125.7 14.1 60,381 29.3 Jordan (b) 91.9 145.0 53.1 4.3 21.3 403 228.0 Kazakhstan 114.7 n.a. n.a. 16.9 6.8 1,917 59.8 Korea, Rep. of 885.0 n.a. n.a. 45.9 19.3 20,422 43.3 Kuwait (a) 160.5 n.a. n.a. 1.8 87.5 412 389.6 Notes: Mi~-I" ~s Minutes of Telecommunications Traffic. Data are in millions of minutes of pubhc switched traffic. a. Internatzonal Ml~-I" based on billing point of traffic. b. Internationaltraff~c foryear endzng 31 March. Austraha and Pakzstan ends30 June. c. Traffic data exclude some carriers or routes (See country table for details.) d. Traffic data is for 1996. © TeleGeography, Inc. 1998 260 © TeleGeography, Inc. 1998 TeleGeography 1999 Figure 12. International Traffic Indicators, 1997 (continued) International Traffic Outgoing Incoming Surplus(Deficit) Population Mi’I’I’(Out) Main Lines Mi’l’r(out) (mMi’n’) (mMi’n’) (mMi’n’) (m) per Capita (thous.) per Main Line Luxembourg (d) 248.5 189.8 -58.8 0.4 591.1 245 1,014.3 Macau (a) 119.0 92.2 -26.7 0.5 236.9 170 700.0 Macedoma 51.7 85.2 33.5 2.0 25.9 n.a. n.a. Malaysia (a,b,c) 588.5 592.0 3.5 20.5 28.7 4,223 139.4 Malta 34.4 n.a. n.a. 0.4 91.2 187 184.0 Mexico (a) 1,200.0 n.a. n.a. 97.6 12.3 9,264 129.5 Motdova (a) 56.8 80.2 23.3 4.5 12.7 627 90.6 Namibia 52.3 47.0 -5.3 1.7 30.8 101 517.8 Netherlands (a,c) 1,535.0 1,712.2 177.2 15.6 98.1 8,860 173.3 New Zealand (b) 407.0 n.a. n.a. 3.6 113.5 1,840 221.2 Nicaragua (a) 40.4 52.5 12.1 4.4 9.2 128 315.6 Norway 481.0 n.a. n.a. 4.4 109.3 2,325 206.9 0man (a) 74.3 70.4 -3.9 2.3 32.8 201 369.7 Pakistan (b,c) 84.1 557.8 473.7 132.2 0.6 2,557 32.9 Panama (a) 41.4 95.1 53.7 2.7 15.4 325 127.4 Papua New Guinea 23.4 20.6 -2.8 4.4 5.3 47 497.9 Paraguay (d) 24.9 49.4 24.5 5.7 4.4 176 141.5 Peru (a) 79.4 256.9 177.5 25.6 3.1 1,640 48.2 Philippines (a,b) 295.0 930.0 635.0 76.1 3.9 2,078 142.0 Poland (a,d) 437.2 725.5 288.3 38.6 11.3 7,510 58.2 Portugal (a) 393.3 628.8 235.5 9.9 39.6 3,819 103.0 Romania 110.8 278.6 167.8 22.5 4.9 n.a. n.a. Russia (c) 939.3 1,035.6 96.3 147.3 6.4 26,875 35.0 Saudi Arabia 660.0 n.a. n.a. 20.1 32.9 2,285 288.8 Singapore (a,b) 1,161.0 n.a. n.a. 3.4 337.4 1,685 689.0 Slovak Republic (a) 144.7 174.4 29.8 5.4 26.9 1,392 104.0 Slovenia 113.5 118.9 5.4 2.0 57.5 722 157.2 South Africa 368.8 343.2 -25.6 42.3 8.7 4,648 79.4 Spain (a,d) 1,189.0 n.a. n.a. 39.1 30.4 15,854 75.0 Sri Lanka 33.2 124.3 91.1 18.7 1.8 315 105.4 Sweden (a) 1,140.0 n.a. n.a. 8.9 128.6 6,010 189.7 Switzerland 2,164.0 1,723.0 -441.0 7.2 298.9 4,688 461.6 Syria (a) 89.3 173.3 84.0 16.1 5.5 1,312 68.1 Taiwan (b) 789.0 842.2 53.2 21.7 36.4 10,862 72.6 Thailand (a,c) 278.4 408.5 130.1 59.5 4.7 4,815 57.8 Turkey 557.5 836.0 278.5 63.5 8.8 15,744 35.4 Ukraine 486.8 n.a. n.a. 50.4 9.6 9,410 51.7 United Arab Emirates (d) 589.3 n.a. n.a. 2.3 260.5 835 705.7 United Kingdom (a,b) 6,600.0 n.a. n.a. 57.6 115.6 14,300 465.7 United States (a) 22,700.1 9,219.3 -13,480.8 268.0 84.7 170,568 133.1 Uruguay (a) 68.4 93.7 25.3 3.3 20.9 761 89.9 Uzbekistan 63.1 n.a. n.a. 23.5 2.7 1,531 41.2 Venezuela (a) 159.2 286.9 127.7 22.4 7.1 2,804 56.8 Vietnam (a) 50.2 294.6 244.4 75.1 0.7 1,587 31.6 Yugoslavia (c) 217.0 332.0 115.0 11.2 19.3 2,082 104.2 Notes: Mil-r ~s M~nutes of Telecommunicatzons Traffic. Data are ~n millions of minutes of public switched traffic. a. lnternatzonal Mil-r based on bzlling point of traffzc. b International traffic for year ending 3t March. Austraha and Pakistan ends 30 June. c Traffic data exclude some carrzers or routes. (See country table for detazls.) d Traffic data is for 1996. © TeleGeography, Inc. 1998 261 TeleGeography 1999 © TeleGeographyo Inc. 1998 International Traffic by Route Figure 13. The Top 50 International Routes, 1997 Countries Mi’r]" each way Total Mi’r]" 1. U.S. Canada 3,926.6 3,173.3 7,099.9 2. U.S. Mexico 2,757.6 950.2 3,707.8 3. U.S. U.K. 1,532.6 946.5 2,479.0 4. Hong Kong China 1,065.2 855.0 1,920.2 5. U.S. Germany 994.3 324.6 1,318.9 6. U.S. Japan 830.5 339.4 1,169.9 7. U.K. Ireland 658.6 504.0 1,162.6 8. U.K. France 502.0 386.0 888.0 9. U.K. Germany 504.5 326.9 831.4 10. Switzerland Germany 477.0 339.9 816.9 11. Austria Germany 427.6 384.9 812.5 12. U.S. Hong Kong 673.2 82.4 755.6 13. France Germany 412.0- 331.8 743.8 14. Singapore Malaysia 408.0 330.1 738.1 15. U.S. France 500.0 216.6 716.5 16. Germany Italy 368.5 340.3 708.8 17. Netherlands Germany 346.7 327.3 674.0 18. U.S. Brazil 497.3 159.6 656.9 19. U.S. India 579.4- 49.7 629.1 20. U.S. Rep. of Korea 421.9 203.5 625.4 21. France Belgium 304.0- 299.3 603.3 22. U.S. Italy 475.6 121.9 597.5 23. Switzerland France 344.0- 246.0 590.0 24. France Italy 298.0- 266.3 564.3 25. U.S.-Taiwan 379.1 174.3 553.3 26. Germany Turkey 377.4 172.5 549.9 27. Ukraine Russia 296.7 249.0 545.7 28. Germany Poland 353.7 190.0 543.7 29. Belgium Netherlands 266.1 254.5 520.6 30. U.S. Australia 383.5 123.5 507.0 31. Switzerland Italy 292.0 198.1 490.1 32. U.S. Dominican Republic 392.1 95.6 487.7 33. U.K. Spain 270.6 208.0 478.6 34. U.K. Italy 270.1 196.8 466.9 35. Australia U.K. 230.0- 223.6 453.6 36. U.S. China 388.1 61.7 449.8 37. U.S. Philippines 417.2 32.2 449.4 38. France Spain 233.0 209.2 442.2 39. Spain Germany 212.5 209.4 421.9 40. U.K. Netherlands 223.2- 198.3 421.4 41. U.S. Israel 216.5 161.7 378.2 42. New Zealand Australia 185.0 175.0 360.0 43. Taiwan China 198.1 150.0 348.1 44. Japan China 229.5- 105.0 334.5 45. U.S. Switzerland 225.8 108.1 333.8 46. U.K. Canada 170.4- 160.0 330.4 47. U.S. Colombia 260.5 68.7 329.2 48. U.S. Netherlands 222.3 105.8 328.1 49. U.S. Jamaica 262.6 50.2 312.8 50. Japan Rep. of Korea 156.0 150.0 306.0 Note’ All data ~n mzllions of minutes of telecommumcations traffic (Mi’Tl’). The country which generates more traffic on each route zs listed first. The routes listed above total 41.5 bdlion mznutes, 51 percent of all ~nternatzonal traffic. For routes to and from the United States, calls are measured by pmnt of b~lhng in both directions for calendar year 1997 On non-U.S, routes, data for Australia, Hong Kong, Ireland, Malaysza, New Zealand, Singapore, Talwan, and the U.K. are for fiscal year 1997/98. International S~mple Resale (]SR) traffzc on non-U.S, routes is excluded. © TeleGeography, Inc. 1998 262 © TeleGeography, Inc. 1998 TeleGeography 1999 Figure 14. Traffic Imbalances on U.S. Routes, 1997 U.S.-Canada U.S -U K U.S.-Mex~co U.S.-Germany U S.-Dominican Rep. U.S.-Saud~ Arab=a U.S.-South Africa U.S.-Italy U.S,-Guyana U.S.-Argentma U.S.-China U,S.-Hong Kong U.S.-Honduras U.S.-Ind~a U.S.-EI Salvador U.S.-Phihppines U.S.-Pakistan U.S.-Ha~t= U.S,-Wetnam 1:1 5:1 10:1 15:1 20:1 25:1 30:1 35:1 Ratio of Outgoing to Incoming Traffic 40:1 Note: Country with traffic deficit on route listed first. A ratio of 1:1 would indicate a perfect balance on a route. U.S. data based on billing point of call, and may not reflect actual call ratios due to call-back. © TeleGeography, Inc. 1998 Figure 15. Traffic Imbalances on Non-U.S. Routes, 1997 Slngapore-Phlhppines U.K.-Chma UAE-Jordan Rep. of Korea-China Australia-China Saudi Arabia-Phihppmes U.K.-Phllippmes Japan-Philippines Taiwan-Phil=pp~nes Switzerland-Portugal Canada-Pakistan Canada-lnd=a Italy-Chile 12:1 Ratio of Outgoing to Incoming Traffic Note: Country with traffic deficit on route listed first. A ra+ao of 1:1 would indicate a perfect balance on a route. Data for some countries based on billing point of call, and may not reflect actual call ratios due to call-back. © TeleGeography, Inc. t998 263 TeleGeography 1999 © TeleGeography, Inc. 1998 Figure 16, International Outbound Routes with Rapidly Growing Traffic, 1996-1997 Note: Country originating ~raffic listed first; coun~ termina~ng t~affic listed second. Some da~a based on billing point of call, and may not reflect actual route growt~ rates due to call-back. ©TeleGeography, Inc. 1998 Figure 17. International Outbound Routes with Declining Traffic, 1996-1997 o%° o -10% -15% r’~ ~-- -20% -25% -30% Note: Country originating traffic listed f~rst; countryterm~natmg t~affic I~sted second. Some data based on billing point of call, and may not reflect actual rates at which traffic flows are changing due to call-back. ©TeleGeography, Inc. 1998 264 © TeleGeography, Inc. 1998 TeleGeography 1999 Alternative Traffic Arrangements Figure 18. Alternative Traffic Arrangements as a Share of Global International Traffic, 1996-1997 1996 Total International Traffic 1997 Total International Traffic Note: Area of circles proportional to total international traffic, which equaled 71.7 billion minutes in 1996 and 81.8 b~llion minutes in 1997. © TeleGeography, Inc. 1998 Figure 19. Selected Origin Countries of Call-Back Traffic, 1997 600 500 ._~ 200 100 Argentina Philippines Taiwan NetherLands China France Hong Kong Germany Source: M.J. Scheele & Assomates (wvwv.caltbackinfo.com); Tel. +1 415 824 2400; Fax +1 415 824 2429 265 TeleGeography 1999 © TeleGeography, Inc. 1998 Figure 20. Reported U.S.-Hubbed Refile and Calling Card Traffic Top Destinations for Calls Reoriginated via the U.S., 1996 Destination Millions of Mi’rrs 1. United Kingdom 159.1 2. Germany 86.2 3. Japan 60.8 4. Italy 60.3 5. Mexico 52.5 6. France 50.0 7. Canada 35.9 8. Rep. of Korea 35.5 9. Belgium 26.2 10. Spain 25.5 11. China 24.2 12. Brazil 22.9 13. Philippines 21.2 14. Australia 20.9 15. Hong Kong 20.8 16. Israel 20.0 17. Netherlands 17.9 18. Guatemala 17.2 19. Jamaica 16.2 20. Switzerland 15.6 Total Traffic Reoriginated via the U.S., 1995-1997 2,000 1,000 500 754 1,181 1,700 1995 1996 1997 Note: U.S. cell reoriginatJon includes calls that are placed in non-U.S, countries, re-routed via the U.S., and terminated in a third countn/. The route totals include refile and calling card calls, but exclude call-back traffic. Data are based on U.S. carrier filings with the FCC, and may understate actual volume of reoriginated calls. Source: FCC ©TeleGeography, Inc. 1998 Figure 21. Illicit Bypass Traffic, 1997 Hong Kong China Japan Taiwan Italy Brazil Germany India Russia Philippines Argentina South Africa Estimated Traffic Bypassing Settlement Rate System to Selected Destinations, 1997 100 200 300 400 Incoming Traffic Bypassing Settlement Rate System (millions of minutes) 5OO Note: Illicrt bypass includes international calls that, through the use of extra-legal means by originating carr}ers, are not covered by trad~onat settlement payments to foreign Pl-rs. For example, an internatmnal carrier may send calls over an interna~Jonat private line directly to a foreign cellular serv=ce provider, which then switches the traffic ontO the local wireline network. Source: M.J. Scheele & Assomates (www.mjscheele.com); Tel. +I 415 824 2400; Fax +1 415 824 2429 266 © TeleGeography, Inc. 1998 TeleGeography 1999 Figure 22. International IP Telephony Forecasts International Call Minutes Transmitted via IP 50% 3O% 20% Delta Three Whrte Paper O 10% Tarifica O Delta Three White Paper O Analysys O Yankee Group o Yankee Group 1996 1998 2Q00 2002 2004 20Q6 Total Revenues from International ]P Telephony "~ Analysys == $7 o $3 Forrester E Research = $2 O = Delta Three $1 White Paper cc O pulver.com 1996 1998 2OQQ 2002 2(~04 3O06 Note: These e~mates of lntemet Protocol (IP) telephony call traffic and revenue include all voice traffic transmitted via IP, including calls carrzed on the public lnternet, carrier-owned pdvate lines, and the Public Switched Telephone Network (PSTN). Sources: Analysys Ltd, (wwvv.analysysocom), Delta Three Inc. (wwvv.deltathree.com); Forrester Research Inc. (www.forrester.com); pulver.com (wwvv, pulverocom); Tarffica, Ltd. (ww~v.tarifica.com); The Yankee Group (vwvw.yankeegroup.com) © TeleGeography, Inc. 1998 267 TeleGeography 1999 © TeleGeography, Inc. 1998 Methodology The traffic statistics in TeleGeography 1999 were compiled pri- marily from an independent survey of telecommunications ser- vice providers by TeleGeography, Inc. (TGI). For some countries and carriers, traffic data have been estimated based upon annual reports, government publications and industry inter- views. See the footnotes to each table for further information. The Direction of Traffic database, jointly compiled by TGI and the International Telecommunication Union (l’l’U), was also con- sulted. To enable comparisons of countries’ international traffic statis- tics, TGI has endeavored to apply a consistent methodology. When reviewing the traffic statistics in TeleGeography 1999, however, readers should keep in mind the following issues which may cause traffic data to appear inconsistent: Public Switched Network vs. Private Line Traffic Traffic volumes in TeleGeography 1999 are generally reported in minutes or M~TT (Minutes of Telecommunications Traffic). In most cases M~TT refer to paid minutes of traffic on public switched voice circuits and thus include voice as well as non- voice (facsimile or data) traffic. Unless otherwise stated, traffic carried by International Simple Resale (ISR) carriers is excluded. ISR carriers resell the capac- ity of international private lines (IPLs) for switched services by interconnecting their IPLs to the public switched network at one or both ends. Traffic carried by "pure" resellers of international switched voice services is included in this report. These resellers do not own or lease their own international transmission facilities. Instead, they resell the traffic of other carriers; thus, pure resale traffic is counted as part of the Mi-i=[ for the facilities-based carrier whose services are resold. Cross-Border Traffic Neighboring countries may not classify local cross border traffic in the same way. That is, one country may treat some cross- border traffic as local, while its neighbor counts all such traffic as international. Billing Point vs. Originating Point of Traffic Unless otherwise stated in the notes to a table, the outbound MiTT reported for countries in TeleGeography 1999 refers to outbound traffic originated in the reporting country even if it is billed in another country. In the past, most international calls were billed at the point of origination. The number of billed minutes thus coincided with the volume of outgoing traffic. Billed minutes also included col- lect or reverse charge calls because the calls were set up by an operator in the originating country. However, the recent use of calling card and call-back services has shifted the billing point for many international calls. For example, calls from Italy to the United States (or a third country, such as Argentina) may now be set up and billed in the U.S. Some countries, including the U.S., report international traffic data based solely on the location where the traffic is billed. Consequently, "outbound" traffic data for these countries can include traffic actually originating in another. Thus, incoming Mi-FF reported for one country may not match the outgoing Mi-I-F on the same route by the correspondent country. Some double counting may also occur For example, a call from Thailand to the U.S. which is billed to a U.S. calling card is reported by the U.S. carrier as outbound U.S. M~ the same call also is reported as outbound M~TI- by Thailand. Accordingly, in countries where calling card and call-back ser- vices are widely used, a year-to-year comparison of national Mi-I-F also requires examining the statistics of countries, such as the U.S. and the U.K., where the calls are hubbed. Transit Traffic Unlike calling card and call-back traffic, TeleGeography I999 excludes from country totals transit traffic--that is, traffic which merely passes through a given country, but is not refited via the switched network in the reporting country. Rounding Rounding may cause the figures on total national incoming and outgoing traffic to appear inconsistent with other national data. Revised Data Some differences exist between the historical statistics (1996 or earlier) reported in TeleGeography 1999 and data stated in prior TGI reports or Direction of Traffic. The variations reflect corrections and/or revised data subsequently provided to TGI. 268 Global Reference: Blue Pages TeleOeo6raphy 1999 © TeleGeography, Inc. 1998 National Telecommunications Indicators (A-K) GDP 1997 Population Main Lines Lines Per Cellular Users PCs Internet Users Countries (USSbillions) 1997 (millions) 1997 (thous.) 100 (thous.) 1997 (thous.) (thous.) 1997 (thous.) Albania (a,c) n.a. 3.3 87 2.6 n.a. n.a. n.a. Andorra n.a. 0.1 32 50.0 10 n.a. 1.0" Argentina (d) 323.2 35.7 6,750 18.9 2,013 1,400 170.0 Armenia (d) 1.6 3.4 n.a. n.a. n.a. n.a. n.a. Australia (b) 346.3 18.4 9,350 50.7 4,893 6,700 1,600.0 Austria (a,c) 206.2 8.1 3,969 48.8 1,160 1,700 650.0 Bahamas 2.5 0.3 96 34.9 6,152 n.a. 5.0 Bahrain (a) 5.4* 0.6 152 25.2 59 40 2.0" Bangladesh 30.9 125.3 297* 0.2* n.a. n.a. n.a. Belarus (a,d) 13.5 10.4 2,313 22.2 8 n.a. 0.7* Belgium (a) 242.4 10.2 4,769 46.9 974 2,400 300.0* Bolivia 7.8 7.7 535 7.0 116 n.a. 3.0* Botswana 4.9 1.5 84 n.a. n.a. n.a. n.a. Brazil (a) 688.1" 164.5 15,106" 10.0" 4,400 4,200 1,310.0 Brunei n.a. 0.3 79* 25.7* n.a. n.a. 3.4* Canada (a) 617.6 30.3 18,460 60.8 3,420* 8,200 4,500.0 Chile 77.1 14.5 2,600 17.9 410 790 2,000.0* China 917.7 1,221.6 70,310 5.8 13,233 7,500 400.0 Colombia (a) 76.1" 37.4 5,334 14.3 1,265 1,214 130.0" Costa Rica (a) 9.5 3.5 584 16.5 64 n.a. 30.0* Cyprus (a,c) 8.9 0.8 385 51.1 92 n.a. 5.0* Czech Republic (a) 52.0 10.3 3,275 31.8 526 850 200.0* Denmark (a,d) 169.7 5.3 3,339 62.9 1,450 1,900 300.0* Dominican Republic 14.9 7.9 709 9.0 130 n.a. 6.2 El Salvador (a) 11.3 5.7 325 5.7 40 n.a. 2.6 Estonia (a) 4.7 1.4 489 32.6 144 n.a. n.a. Finland 119.8 5.1 2,866 55.8 2,147 1,600 1,000.0" France 1,392.9 58.6 33,700 57.5 5,817 10,200 500.0" French Polynesia 3.1" 0.2 52 22.4 5 n.a. 0.1 Germany (a,c) 2,102.7 82.1 45,200 55.1 8,170 21,000 2,500.0" Greece 120.9 10.6 5,328* 51.0" 938 470 150.0" Guyana 0.7 0.7 50* 6.0* 1" n.a. 0.5 Hong Kong (a,b) 172.3 6.5 3,647 55.7 2,230 1,500 675.0 Hungary (a) 43.7 10.2 2,660 26.6 265* 500 107.7" Iceland (a) 7.4 0.3 168 62.1 66 n.a. 40.0" India (a,c) 324.1 966.8 17,802 1.8 882 2,000 300.0 Indonesia (a) 214.6 209.8 4,982 2.4 916 1,600 250.0 lran 132.9 67.5 6,513 9.6 239 2,000* 30.0 Ireland (a,b,d) 73.2 3.6 1,500 41.6 533 520 80.0 Israel (a) 97.9 5.5 2,656 48.0 1,672 1,100 300.0 Italy (c) 1,145.4 56.8 25,698 45.2 11,738 6,500 585.0 Japan (b) 4,192.7 125.7 60,381 48.0 38,254 25,500 8,500.0 Jordan (b) 3.6 4.3 403 9.3 n.a. 40* 2.0 Kazakhstan (d) 2.1 16.9 1,917" 11.4" 10" n.a. 6.0 Korea, Rep. of 442.5 45.9 20,422 44.4 6,910 6,931 800.0 Kuwait (a) 26.7* 1.8 412 22.4 210 125" 15.0" Source: International Telecommunication Union, International Monetary Fund, Central Intelligence Agency, and TeleGeogrophy, Inc. 270 © TeleGeography, Inc., 1998 © TeleGeographg, Inc. 1998 TeleGeography 1999 International Telephone Traffic Outgoing MiTr (millions) Incoming MiTI" (millions) Traffic Balance 1996 1997 % Change 1996 1997 % Change 1996 1997 Countries 20.6 40.8 98.2% 46.0 ~.a. n.a. 25.4 n.a. Albania (a,c) 37.8 42.2 11.6% 27.3 30.1 10.3% -10.5 -12.1 Andorra 181.3 223.4 n.a. 390.7 444.2 n.a. 209.4 220.7 Argentina (d) 48.1 48.8 n.a. n.a. n.a. n.a. n.a. n.a. Armenia (d) 1,305.0 1,510.0 15.7% n.a. 1,250.0 n.a. n.a. -260.0 Australia (b) 960.0 995.5 3.8% n.a. 957.7 n.a. n.a. -37.8 Austria (a,c) 56.7 62.7 10.5% n.a. n.a. n.a. n.a. n.a. Bahamas 92.2 106.6 15.6% 69.4 85.4 23.1% -22.8 -21.2 Bahrain (a) 38.3 46.9 22.5% 129.2 187.0 44.7% 90.9 140.2 Bangladesh 104.9 148.6 n.a. n.a. 185.2 n.a. n.a. 36.6 Belarus (a,d) 1,228.4 1,340.0 9.1% 1,289.1 1,420.0 10.2% 60.7 80.0 Belgium (a) 21.4 22.7 6.1% 53.9 69.3 28.6 % 32.5 40.6 B olivi a 30.8 n.a. n.a. 22.7 n.a. n.a. -8.1 n.a. Botswana 366.9 459.1 25.1% 624.4 761.3 21.9% 257.5 302.2 Brazil (a) 35.2 35.6 1.1% n.a. 22.0 n.a. n.a. -13.6 Brunei 3,519.8 4,286.3 21.8% 4,313.3 4,635.1 7.5% 793.5 340.7 Canada (a) 173.8 242.0 39.2% n.a. n.a. n.a. n.a. n.a. Chile 1.433.2 1,631.8 13.9% n.a. n.a. n.a. n.a. n.a. China 135.5 n.a. n.a. 384.2 n.a. n.a. 248.7 n.a. Colombia (a) 55.0 66.9 21.6% 87.8 111.6 27.1% 32.8 44.7 Costa Rica (a) 128.6 154.4 20.1% 92.0 115.2 25.2% -36.6 -39.2 Cyprus (a.c) 281.2 306.1 8.9% 324.4 355.0 9.4% 43.2 48.9 Czech Republic (a) 573.2 628.0 n.a. 600.0 682.0 n.a. 26.8 54.0 Denmark (a,d) 126.6 142.0 12.2% 4503 476.9 5.8% 324.3 3343 Dominican Republic 28.6 34.3 20.0% 160.5 168.2 4.8% 131.9 133.9 El Salvador (a) 58.5 66.3 13.4% 60.1 67.0 11.5% 1.6 0.7 Estonia (a) 332.0 371.7 12.0% n.a. n.a. n.a. n.a. n.a. Finland 3.273.0 3.545.0 8.3% 3.283.0 3,609.0 9.9% 10.0 64.0 France 7.9 9.1 15.2% n.a. n.a. n.a. n.a. n.a. French Polynesia 5,100.0 5,333.1 4.6% n.a. 5,206.0 n.a. n.a. -127.1 Germany (a,c) 515.6 593.7 15.2% 557.3 634.6 13.9% 41.7 40.9 Greece 29.8 24.1 19.1% 162.8 142.4 12.6 % 133.0 118.2 G uya n a 1,738.6 1,718.0 -1.2% 1,940.8 2,100.3 8.2% 202.2 382.3 Hong Kong (a,b) 265.0 287.1 8.4% n.a. 324.6 n.a. n.a. 37.5 Hungary (a) 32.5 n.a. n.a. 32.0 n.a. n.a. -0.5 n.a. Iceland (a) 384.2 420.5 9.5% 1,000.0 1,256.6 25.7% 615.8 836.1 India (a,c) 280.2 351.6 25.5% 356.4 456.0 27.9% 76.2 104.4 Indonesia (a) 183.2 160.7 -12.3% n.a. 130.2 n.a. n.a. -30.5 Iran 580.0 695.0 n.a. n.a. n.a. n.a. n.a. n.a. Ireland (a,b,d) 319.7 n.a. n.a. 468.1 n.a. n.a. 148.4 n.a. Israel (a) 2,124.0 2,351.9 10.7% 2,253.5 2,475.1 9.8% 129.5 123.2 italy (c) 1,710.6 1,771.7 3.6% 1,519.1 1,635.0 7.6% -191.5 -136.7 Japan (b) 74.6 91.9 23.2% 133.1 145.0 8.9% 58.5 53.1 Jordan (b) 102.5 114.7 n.a. n.a. n.a. n.a. n.a. n.a. Kazakhstan (d) 699.3 885.0 26.6% 740.6 n.a. n.a. 41.3 n.a. Korea, Rep. of 140.7 160.5 14.1% 131.2 n.a. n.a. -9.4 n.a. Kuwait (a) Notes: a. International Mi]-r based on billing point of traffic. b. International traffic for year ending 31 March (30 June for Australia, Pakistan) c. Traffic data exclude some carriers or routes. d. 1996 and 1997 traffic data are not directly comparable. See countw pages for details. Data for 1996. © TeleGeography, Inc. 1998 271 TeleGeo{~raphy 1999 © TeleGeograph~l, Inc. 1998 National Telecommunications Indicators (L-Z) GDP 1997 Population Main Lines Lines Per Cellular Users PCs Internet Users Countries (US$billions) 1997 (millions) 1997 (thous.) 100 (thous.) 1997 (thous.) (thous.) 1997 (thous.) Luxembour9 12.9" 0.4 245 58.4 67 n.a. 23.0* Macau (a) n.a. 0.5 170 33.8 51 n.a. 3.0* Macedonia n.a. 2.0 n.a. n.a. n.a. n.a. n.a. Malaysia (a,b,c) 98.5 20.5 4,223 20.6 2,461 1,000 600.0 Malta 0.5 0.4 187 49.6 18 n.a. n.a. Mexico (a) 402.8 97.6 9,264 9.5 1,745 3,600 520.0 Moldova (a,d) 1.9 4.5 627 14.1 2 11" 0.2 Namibia 3.2 1.7 101 5.9 13 n.a. 1.0 Netherlands (a,c,d) 360.5 15.6 8,860 56.6 1,717 4,400 900.0* New Zealand (b) 57.1 3.6 1,840 51.3 565 1,000 550.0 Nicaragua (a) 2.0 4.4 128 2.9 8 n.a. n.a. Norway 153.4 4.4 2,325 52.8 1,677 950* 500.0* Oman (a) 16.2 2.3 201 8.9 60 25* n.a. Pakistan (b,c) 60.9 132.2 2,557 1.9 110 n.a. 4.0 Panama (a) 8.1 2.7 320* 12.0" 0 n.a. 6.0 Papua New Guinea n.a. 4.4 47* 1.1" 3* n.a. 0.1 Paraguay 7.7 5.7 181" 4.0* 33* n.a. 1.0" Peru (a) 65.2 25.6 1,640 6.4 436 n.a. 60.0* Philippines (a,b) 82.2 76.1 2,078 2.7 1,302 970 100.0 Poland (a) 135.6 38.6 7,510 19.4 857 1,400 800.0 Portugal (a) 102.3" 9.9 3,819 38.5 1,507 740 500.0 Romania 12.8 22.5 n.a. n.a. n.a. n.a. n.a. Russia (c) 449.8 147.3 26,875 18.2 485 4,700 600.0* Saudi Arabia 125.5" 20.1 2,285 11.4 332 700* 20.0 Singapore (a,b) 83.7* 3.4 1,685 49.0 849 1,270 500.0 Slovak Republic (a) 19.5 5.4 1,392 25.8 200 1,000" 100.0" Slovenia 18.6" 2.0 722 36.6 93 n.a. 100.0" South Africa 129.1 42.3 4,646 11.0 1,600 1,800 800.0 Spain (a) 531.3 39.1 15,854 40.5 4,338 4,800 525.0 Sri Lanka 15.1 18.7 315 1.7 112 60* 2.5 Sweden (a) 227.8 8.9 6,010 67.8 3,169 3,100 800.0* Switzerland 255.0 7.2 4,688 64.7 1,044 2,800 370.0* Syria (a) 59.9* 16.1 1,312 8.1 0 20* n.a. Taiwan (b) n.a. 21.7 10,862 50.1 1,492 2,570 1,500.0 Thailand (a,c) 152.6 59.5 4,815 8.1 2,003 1,200 150.0 Turkey 191.9 63.5 15,744 24.8 1,610 1,300 120.0" Ukraine (d) 49.7 50.4 9,410 18.7 57 n.a. 50.0" United Arab Emirates 39.1" 2.3 835 36.9 309 160 9.7* United Kingdom (a,b,d) 1,288.2 57.6 30,292* 51.8 8,993 11,200" 2,500.0* United States (a) 8,079.9 268.0 170,568" 64.0" 55,312 109,000 40,000.0 Uruguay (a) 20.0 3.3 761 23.3 150 n.a. 75.0* Uzbekistan (d) 21.6 23.5 1,814" 7.8* 4 n.a. 1.0 Venezuela (a) 87.5 22.4 2,804 12.5 1,072 850 35.0 Vietnam (a) 20.4* 75.1 1,587 2.1 134 250* 0.1 Yugoslavia (c) n.a. 11.2 2,082 20.0 87 n.a. 20.0* Source: International Telecommunication Union, International Monetary Fund, Central Intelligence Agency, and TeleGeography, Inc. © TeleGeography, Inc., 1998 272 © TeleGeography, Inc. 1998 TeleGeography 1999 International Telephone Traffic (L-Z) Outgoing MiTr (millions) incoming Mil"r (millions) Traffic Balance 1996 1997 % Change 1996 1997 % Change 1996 1997 Countries 240.5 n.a. n.a. 189.8 n.a. n.a. -58.8 n.a. Luxembourg 112.5 119.0 5.8% 92.1 92.2 0.2% -20.4 -28.7 Macau (a) 51.0 51.7 1.4% 81.9 85.2 4.1% 30.9 33.5 Macedonia 570.5 588.5 3.1% 581.9 592.0 1.7% 11.4 3.5 Malaysia (a,b,c) 31.7 34.4 8.6% 34.0 n.a. n.a. 2.3 n.a. Malta 1,070.7 1,200.0 12.1% 2,489.7 n.a. 5.0% 1,419.0 n.a. Mexico (a) 50.2 56.8 n.a. n.a. 80.2 n.a. n.a. 23.3 Moldova (a,d) 51.4 52.3 1.8% n.a. 47.0 n.a. n.a. -5.3 Namibia 1,534.1 1,535.0 n.a. 1,584.6 1,712.2 n.a. 50.5 177.2 Netherlands (a,c,d) 353.0 407.0 15.3% 380.0 n.a. n.a. 27.0 n.a. New Zealand (b) n.a. 40.4 n.a. n.a. 52.5 n.a. n.a. 12.1 Nicaragua (a) 443.5 401.0 8.5% 422.3 n.a. n.a. -21.2 n.a. Norway 62.6 74.3 18.7% 58.0 70.4 21.4% -4.6 -3.9 Oman (a) 77.0 84.1 9.2% 408.4 557.8 14.2% 411.4 473.7 Pakistan (b,c) 41.2 41.4 0.5% 97.7 95.1 -2.6% 56.5 53.7 Panama (a) 28.9 23.4 -13.0% n.a. 20.6 n.a. n.a. -2.8 Papua New Guinea 24.9 n.a. n.a. 49.4 n.a. n.a. 24.5 n.a. Paraguay 66.7 79.4 19.1% 226.5 258.9 13.4% 159.7 177.5 Peru (a) 240.0 295.0 22.9% 767.0 930.0 21.3% 527.0 635.0 Philippines (a,b) 407.2 n.a. n.a. 725.5 n.a. n.a. 288.3 n.a. Poland (a) 340.0 393.3 15.7% 571.4 628.8 10.1% 231.4 235.5 Portugal (a) 91.5 110.8 21.1% 237.5 278.6 17.3% 146.0 167.8 Ro m an ia 851.3 939.3 10.3% 1,037.6 1,035.6 -0.2% 186.3 96.3 Russia (c) 584.4 660.0 12.9% n.a. n.a. n.a. n.a. n.a. Saudi Arabia 942.0 1,161.0 23.3% n.a. n.a. n.a. n.a. n.a. Singapore (a,b) 134.1 144.7 7.9% 159.0 174.4 9.7% 24.9 29.8 Slovak Republic (a) 105.3 113.5 7.8% 113.9 118.9 4.4% 8.6 5.4 Slovenia 353.0 368.8 4.5% n.a. 343.2 n.a. n.a. -25.6 South Africa 1,189.0 n.a. n.a. n.a. n.a. n.a. n.a. n.a. Spain (a) 29.3 33.2 13.2% 96.0 124.3 29.5% 66.7 91.1 Sri Lanka 1,026.0 1,140.0 11.1% n.a. n.a. n.a. n.a. n.a. Sweden (a) 1,935.5 2,164.0 11.8% 1,562.8 1,723.0 10.3% -372.7 -441.0 Switzerland 78.9 89.3 13.2% 156.0 173.3 11.1 % 77.1 84.0 Syria (a) 674.0 789.0 17.1% 736.8 842.2 14.3% 62.8 53.2 Taiwan (b) 247.4 278.4 12.5% 376.2 408.5 8.6% 128.7 130.1 Thailand (a,c) 473.4 557.5 17.8% 755.0 836.0 10.7% 281.6 278.5 Turkey 340.8 486.8 n.a. n.a. n.a. n.a. n.a. n.a. Ukraine (d) 589.3 n.a. n.a. n.a. n.a. n.a. n.a. n.a. United Arab Emirates 4,569.2 6,600.0 n.a 4,380.0 n.a. n.a. -209.2 n.a. United Kingdom (a,b,d) 19,119.1 22,700.1 18.7% 8,194.9 9,219.3 12.5% -10,924.1 -13,400.8 United States (a) 54.5 68.4 25.5% 80.1 93.7 17.0% 25.6 25.3 Uruguay (a) 54.2 63.1 n.a. n.a. n.a. n.a. n.a. n.a. Uzbekistan (d) 139.0 159.2 14.5% 228.8 288.9 25.3% 89.8 127.7 Venezuela (a) 52.4 50.2 -4.2% n.a. 294.6 n.a. n.a. 244.4 Vietnam (a) 237.2 217.0 -8.5% 325.7 332.0 1.9% 88.5 115.0 Yugoslavia (c) Notes: a. International Mil-r based on billing point of traffic. b. International traffic for year ending 31 March (30 June for Australia, Pakistan) c, Traffic data exclude some carriers or routes. d. 1996 and 1997 traffic data are not directly comparable. See countn/pages for details. Data for 1996. © TeleGeography, Inc. 1998 273 © TeleGeography, Inc. 1998 TeleGeography 1999 lnternati 1 Canada Guam Northern Marianas United States Caribbean L~ Egypt Z12 Morocco 213 Algeria 216 Tunisia 216 Libya Gambia ~.~1 Senegal Mauritania Mall ~4 Guinea ~.5 Ivory Coast Burkina Faso ~.7 Niger Togo Z~9 Benin Z38 Mauritius ~11 Liberia ¯ 1~ Sierra Leone Ghana Nigeria ¯ 15 Chad Z38 Central African Bepublic ~l? Cameroon CapeVerde Islands Sao Tome and Principe 2~ Equatorial Guinea 241 Gabon 242 Congo (Brazzaville) Z~3 Congo (Kinshasa) 2~ Angola 245 Guinea-Bissau 2~ Diego Garcia 24? Ascension Island 246 Seychelles 249 Sudan Z96 Rwanda ~1 Ethiopia Somalia Djibouti ~.~4 Kenya 255 Tanzania 2r~6 Uganda ~.~7 Burundi Mozambique 7̄~ Zambia ~1 Madagascar Reunion Island Zimbabwe ~64 Namibia ~5 Malawi onal Dialing Codes, by ZB8 Lesotho Falkland Islands ~7 Botswana 961 Belize 298 Swaziland 962 Guatemala ~’9 Comoros & Mayotte El Salvador 27 South Africa 964 Honduras St. Helena 965 Nicaragua ~91 Eritrea Costa Rica Z97 Aruba 967 Panama Faroe Islands St. Pierre & Miquelon ~98 Greenland ~9 Haiti 38 Greece 51 Peru 31 Netherlands Mexico 32 Belgium Cuba 3̄ France Argentina :34 Spain 5~ Brazil :396 Gibraltar 96 Chile 381 Portugal; Azores 57 Colombia 3~ Luxembourg 96 Venezuela Ireland 96~ Guadeloupe :3~ Iceland 961 Bolivia :1~ Albania Guyana 388 Malta Ecuador 357 Cyprus 594 French Guiana 396 Finland 595 Paraguay 359 Bulgaria 596 Martinique 38 Hungary 597 Suriname 370 Lithuania 598 Uruguay 371 Latvia 299 Netherlands Antilles 372 Estonia 68 Malaysia 373 Moldova 61 Australia 374 Armenia I~ Indonesia 375 Belarus 93 Philippines 376 Andorra 64 New Zealand 377 Monaco 65 Singapore 378 San Marino 68 Thailand 379 Vatican City 672 Australian Territories Ukraine 673 Brunei 3111 Yugoslavia 674 Nauru 3115 Croatia 675 Papua NewGuinea ~6 Slovenia 676 Tonga lslands 367 Bosnia-Hercegovina 677 Sotomonlslands 3~9 Macedonia 678 Vanuatu ~9 Italy 679 Fiji Romania Palau 41 Switzerland 6111 Wallis & Futuna 41-75Liechtenstein 68Z Cook Islands 4~ Czech Republic Niue ~.1 Slovak Republic 684 American Samoa Austria 965 Western Samoa ~, United Kingdom 986 Kiribati 45 Denmark 987 New Caledonia Sweden 988 Tuvalu 47 Norway 969 French Polynesia 411 Poland 698 Tokelau 49 Germany 691 Micronesia Number 692 Marshall Islands Kazakhstan Russia Tajikistan Uzbekistan ~68 International Freephone 81 Japan 112 South Korea 64 Vietnam 696 North Korea 6~. Hong Kong 6~ Macau II~ Cambodia 696 Laos 68 China 6?6 Inmarsat Special I171 Inmarsat East Atlantic 1172 Inmarsat Pacific 6~J Inmarsat Indian 674 Inmarsat West Atlantic ~68 Bangladesh 981x Global Mobile Satellite Systems 688 Taiwan 98 Turkey 91 India ~!2 Pakistan 93 Afghanistan 94 Sri Lanka 9,~ Myanmar 968 Maldives 961 Lebanon 98Z Jordan 9t~ Syria 984 Iraq 985 Kuwait 968 Saudi Arabia 967 Yemen Oman 971 United Arab Emirates 97Z Israel 9?3 Bahrain 974 Qatar 975 Bhutan 9?6 Mongolia 9?? Nepal 96 Iran 992 Tajikstan 993 Turkmenistan 994 Azerbaijan 99~ Georgia 996 Kyrgyzstan 275 TeleGeography 1999 © TeleGeographg, Inc. 1998 International Dialing Codes, by Country Afghanistan ................. 93 Albania ................... ,355 ~rana ................. 42 Algeria ................... 213 Ngiers ................. 2 American Samoa ............ 684 Andorra .................. 376 Angola ................... 2.44 Luanda ................. 2 Anguilla ................. 1-264 Antigua & Barbuda ........ 1-250 Argentina ................... 54 Buenos Aires ............. 1 Armenia .................. .374 Yerevan ................. 2 Aruba .................... 297 Ascension Island .......... 2.47 Australia ................... 61 Melbourne ............... 3 Sydney .................. 2 Australian Territories ....... .672 Austria ..................... 43 Vienne .................. 1 Azerbaijan ................. 994 Baku .................. 12 Bahamas ................. 1-242 Bahrain .................... 973 Bangladesh ............... ~95 Dhaka .................. 2 Barbados ................ 1-240 Belarus ................... "175 Minsk ................ 172 Belgium .................... 32 Brussels ................ 2 Belize .................... 501 Belmepan ............... 8 Benin .................... 229 Bermuda ................. 1-441 Bhutan .................... 975 Bolivia .................... 591 La Paz .................. 2 Busnia .................... :387 Sarajevo ................ 71 Botswana ................. 267 Brazil ...................... 55 Brasilia ................ 61 Rio de Janeiro ........... 21 S~o Paulo ............... 11 British Virgin Islands ...... 1-284 Brunei ..................... 67:3 Bandar Seri Begawan ....... 2 Bulgaria ................... 359 Sofia ................... 2 Burkina Paso .............. 226 Burma ...................... 95 Burundi ................... 2.57 Cambodia .................. 855 Cameroon ................. 2,37 Canada ...................... 1 Montreal ........... 514/450 Ottawa ................ 613 Toronto ................ 418 Vancouver .............. 604 Cape Verde ............... 3.38 Cayman Islands ........... 1-345 Central African Republic .... 236 Bangui ................. 61 Chad ..................... 3.35 Chile ....................... ~6 Santiago ................ 2 China, People’s Republic of .... 95 Beijing .................. 1 Guangzhou .............. 20 Shanghai ............... 21 Colombia ................... 57 Bogota ................. 1 Cocos Islands; Norfolk & Christmas Islands ......... .672 Comoros .................. 2,69 Congo .................... 2.42 Brazzavilie ......... 81/82/83 Congo .................... 243 Kinsbasa ............... 12 Costa Rica ................. 506 Croatia ................... .385 7agreb .................. 1 Cuba ....................... 53 Havana ................. 7 Cyprus .................... .157 Nicosia ................. 2 Czech Republic ............. 420 Prague .................. 2 Denmark .................... 45 Diego Garcia .............. 2.45 Djibouti ................... 2.53 Dominica ................ 1-767 Dominican Republic ....... 1-B09 Ecuador ................... 593 Quite ................... 2 Egypt ...................... 2.0 Cairo ................... 2 El Salvador ................. 503 Equatorial Guinea ........... 240 Eritrea .................... 2,91 Estonia .................... 372 Tallinn .................. 2 Ethiopia .................. 251 Addis Ababa ............. 1 Falkland Islands ............ 500 Faroe Islands .............. 298 Fiji ...................... 679 Finland .................... 350 Helsinki ................. 9 France ..................... 33 Paris ................... I Marseille ................ 4 Fronch Antilles ............. 596 French Guiaae .............. 594 Freooh Polynesie ........... Gabon .................... 241 Gambia ................... 220 Georgia .................... 995 Tbilisi ............... 8832 Germany .................... 49 Berlin .................. 30 Bonn ................. 228 Frankfurt ................ 69 Munich ................. 89 Ghana .................... Accra .................. 21 Gibraltar ................... 350 Global Mobile Satellite Systems Globalstar ........ 8818/8819 IC0 Global Comm..8810/8811 Iridium .......... 8816/8817 Greece ..................... ~0 Athens ................. 1 Greenland ................ .299 Grenada ................. 1-473 Guadeloupe ................ 550 Guam .................... 1-671 Guatemala ................. Guinea .................... 224 Guinea-Bissau ............. 245 Guyana .................... 592 Georgetown ............. 2 Haiti ...................... 509 Honduras .................. 504 Hong Kong ................. 852 Hungary .................... 36 Budapest ................ 1 Iceland .................... 354 India ....................... 91 Bombay ................ 22 Calcutta ................ 33 New Delhi .............. 11 Indonesia .................. 62 Jakarta ................ 21 Inmarsat Special ................ 870 East Atlantic ............ 871 Pacific ................ 872 Indian ................. 873 West Atlantic ........... 874 International Freephone ...... 800 Iran ....................... 96 Tehran ................. 21 Iraq ...................... 964 Baghdad ................ 1 Ireland .................... 353 Dublin .................. 1 Israel ..................... 972 Jerusalem ............... 2 Tel Aviv ................. 3 Italy ....................... 39 Rome .................. 06 Milan .................. 02 Ivory Coast ................ 2.25 Jamaica ................. 1-876 Japan ...................... 81 Osaka ................... 6 Tokyo ................... 3 Jordan .................... 962 Amman ................. 6 Kazakhstan .................. 7 Almaty ............... 3272 Kenya .................... .254 Nairobi ................. 2 Kiribati .................... 686 Kuwait .................... 965 Kyrgyzstan ................. 996 Bishkek .............. 3312 Lags ...................... 856 Latvia ..................... 371 Riga ................... 2 Lebanon ................... 961 Beirut .................. 1 Lesotho ................... 2,66 Liberia .................... 231 Libya ..................... 2,18 Tripoli ................. 21 Liechtenstein ............ .41-75 Lithuania .................. 370 Vilnius .................. 2 276 © TeleGeography, Inc. 1998 TeleGeography 1999 Luxembourg ................ 352 Macau .................... 853 Macedonia ................ .389 Skopje ................. 91 Madagascar ............... 2.61 Malawi ................... 265 Malaysia ................... 60 Kuala Lumpur ............ 3 Maldives .................. 960 Mall ..................... 223 Malta .................... .356 Marshall Islands ........... .692 Martinique ................. 596 Mauritania ................ 222 Mauritius ................. 230 Mayotte .................. 2.69 Mexico ..................... 92 Guadalajara ............. 36 Mexico Cit7 .............. 5 Monterrey .............. 83 Micronesia ................ 691 Moldova .................. .373 Chisinau ................ 41 Monaco .................. .377 Mongolia .................. 976 Ulaanbaatar ............. 1 Montserrat ............... 1-664 Morocco .................. 2.12 Casablanca ............... 2 Rabat .................. 7 Mozambique .............. 258 Maputo ................. 1 Namibia .................. 2.64 Windhoek .............. 61 Nauru ..................... 674 Nepal ..................... 977 Kathmandu ............... 1 Netherlands ................. 31 Amsterdam ............. 20 Netherlands Antilles ........ 599 New Caledonia ............. 687 New Zealand ................ 64 Auckland ................ 9 Wellington .............. 4 Nicaragua ................. 505 Managua ................ 2 Niger ..................... 227 Nigeria ................... 234 Lagos ................... 1 Niue ...................... 683 North Korea ................ 850 Pyongyang ............... 2 Northern Marianas .......... 670 Saipan ................ 322 NonNay ..................... 47 0slo ................... 2 Oman ..................... 968 Pakistan .................... 92 lslamabad .............. 51 Palau ..................... 680 Panama ................... 507 Papua New Guinea .......... 675 Paraguay ................. .595 Asuncion ............... 21 Peru ....................... 51 Lima .................. 14 Philippines .................. 63 Manila ................. 2 Poland ..................... 48 Warsaw ............... 22 Portugal ................... 351 Lisbon .................. 1 Puerto Rico ............... 1-787 Qatar ...................... 974 Reunion Island ............ 2.62 Romania .................... 48 Bucharest ............... 1 Russia ....................... 7 Moscow .............. 095 St. Petersburg ........... 812 Rwanda .................. 2.50 St. Kitts & Nevis ........... 1-869 St. Lucia ................. 1-758 SL Pierre & Miquelon ........ 508 St. Vincent & the Grenadines .............. 1-809 San Marino ................ 378 S~o Tome and Principe ..... 239 Saudi Arabia ............... 966 Riyadh .................. 1 Senegal .................. 221 Seychelles ................ 248 Sierra Leone .............. 2.32 Freetown ............... 22 Singapore ................... 65 Slovak Republic ............ 421 Bratislava ............... 7 Slovenia ................... 386 Ljubljana ............... 61 Solomon Islands ............ 677 Somalia .................. .292 Mogadishu .............. 1 South Africa ................. 27 Johannesburg ............ 11 Pretoria ................ 12 South Korea ................. 82 Seoul ................... 2 Spain ...................... 34 rvladdd ................ 91 Barcelona ............... 93 Sri Lanka ................... 94 Colombo ................ 1 Sudan .................... 2.49 Khartoum .............. 11 Suriname .................. 597 Swazilaod ................ ,268 Sweden .................... 46 Stockholm ............... 8 Switzerland ................. 41 Berne .................. 31 Zurich ................... 1 Syria ...................... 963 Damascus .............. 11 Tahiti ...................... 689 Taiwan .................... 686 Taipei .................. 2 fajikistan .................... 7 Dushanbe ............ 3772 Tanzania .................. .255 Dar Es Salaam ........... 51 Thailand .................... 56 Bangkok ................ 2 Togo ...................... 228 Tokelau .................... 690 Tonga ..................... 676 Trinidad & Tobago ......... 1-868 Tunisia .................... 216 Tunis ................... I Turkey ...................... 90 Ankara ................ 312 Istanbul ............... 212 Turkmenistan ............... 993 Ashkhabad ............. 12 Turks & Caicos ........... 1-649 Tuvalu ..................... 688 Uganda ................... 2.56 Kampala ............... 41 Ukraine .................... 380 Kiev ................... 44 United Arab Emirates ........ 971 Abu Dhabi ............... 2 Dubai ................... 4 United Kingdom .............. 44 Cardiff ............... 1222 Glasgow .............. 141 London ............ 171/181 Manchester ............ 161 United States ................. 1 Chicago ............ 312/773 Houston ............ 713/281 Los Angeles ......... 213/323 Miami ............. 305/786 New York .......... 212/917 Washington ............ 202 U.S. Virgin Islands ......... 1-340 Uruguay ................... 598 Montevideo .............. 2 gzbekistan ................... 7 Tashkent ............. 3712 Vanuatu ................... 678 Vatican City ................ 379 Venezuela .................. 58 Caracas ................. 2 Vietnam ................... :94 Wallis & Futuna ............ 681 Western Samoa ............. 685 Yemen ..................... 967 Sanaa .................. 2 Yugoslavia ................. 381 Belgrade ................ 11 Zambia ................... .268 Lusaka .................. 1 Zanzibar (Tanzania) .......... 255 Zimbabwe ................ 2.63 Harare .................. 4 277 TeleGeography 1999 © TeleGeogrophy, Inc. 1998 World Dialing Codes ! M~J PERU ,; laP= AR6ENTINA BRAZIL Br~ilia 278 © TeleGeography, Inc. 1998 TeleGeography 1999 7 © TeleGeography, Inc. 1998 For map licenses and custom mapping services, contact: TeleGeography, Inc. 1730 Rhode Island Avenue, NW; Suite 1205; Washington DC 20036 USA Tel. +I 202-467-0017; fax +I 202-467-0851 e-mail: info@telegeography.com Unauthorized reproduction is prohibited. Aft rights reserved. AUSTRALIA 279 TeleGeography 1999 © TeleGeographg, Inc. 1998 North American Area Codes, by Jurisdiction Alabama Birmingham .............. 205 Huntsville ................. 256 Montgomery ............... 334 Alaska ....................... .907 Alberta Calgary ................. .403 Edmonton ................. 780 A~gnilla ..................... 264 Antigna & Bad~nda ............ 268 Arizona Phoenix .................. 602 Tucson .................. .520 Arkansas Jonesboro ............... .870 Little Rock ................ 501 Bahamaa ..................... 242 Barhad~ .................... 246 Garmnda ...................... 441 British Columbia Victoria .................. .250 Vancouver ................ .604 British Virgia Islands .......... 284 California Anaheim ................. .714 Bakersfield ............... .805 Burbank ................. .626 Concord ................. .925 Fresno .................. 209 Irvine .................... 949 Long Beach ............... .562 Los Angeles ........... .213/323 Monterey ................ .831 Oakland .................. 510 Palm Springs ............. .760 Redding ................. .530 Riverside ................. 909 Sacramento .............. .916 Pale Alto ................ .650 San Diego ................ .619 San Femando ............. .818 San Francisco ............ .415 San Jose ............... .408 Santa Ana .............. .714 Santa Monica ............ .310 Santa Rosa .............. .707 Visalia ................. .559 Coymnn Islaodn ................ 345 Colorado Springs .......... .719 Denver .............. .303/720 Ft. Collins ................. 970 Connecticut Bridgeport .............. .203 Hartford ................ .860 Delaware ................... .302 District of Columbia Washington .............. 202 Dominica ..................... 767 Dominican Republic ........... .809 Florida Ft. Lauderdale ............. .954 Ft. Myers ................ .941 Gainesville ................ 352 Jacksonville ............... 904 Miami ............... 305/786 Orlando ................. .407 Tallahassee .............. .850 Tampa ............... .813/727 W. Palm Beach ............. 561 Georgia Athens ................... 706 Atlanta .............. .404/678 Madetta .............. 770/678 Savannah ................. 912 Grenada ..................... .473 Guam ....................... .671 Hawaii ...................... .808 Idaho ....................... .208 Illinois Aurora .................. .630 Cairo .................... 618 Chicago .............. .312/773 Evanston ................. .847 Oak Brook ................ .708 Peoda ................... .399 Rockford ................. .815 Springfield ................ 217 Indiana Evansville ................ .812 Gary ..................... 219 Indianapolis ............... 317 Lafayette ................. 7B5 Council Bluffs ............. .712 Des Moines .............. .515 Dubuque .................. 319 Jamaica ..................... .876 Kansas Kansas City .............. .913 Topeka .................. .785 Wichita .................. 316 Kentucky Dade Park ................ .812 Lexington ................ .606 Louisville ................ .502 Louisiana Baton Rouge ............... 225 New Odeans .............. 504 Shreveport ............... .318 Maine ....................... .207 Manitaba .................... .204 Maryland Baltimore ............ .410/443 Rockville ............. .301/240 Masaachusatta Boston .................. .617 Cambridge ................ 781 Lowell .................. .978 New Bedford .............. 508 Springfield ............... .413 Ann Arbor ................ 234 Detroit .................. 313 Flint ..................... 810 Grand Rapids .............. 616 Lansing ................... 517 Pontiac .................. 248 Sault Ste. Made ........... .906 Minnesota Duluth ................... 218 Minneapolis ............... 612 Rochester ................. 507 St. Cloud ................ .320 St. Paul ................... 651 Biloxi .................... 228 Jackson .................. 601 Misaouri Jefferson City ............. .573 Kansas City .............. .816 St. Joseph ............... .660 St. Louis ................. .314 Springfield ............... .417 Monserret .................... .664 Montana ..................... .406 Nebraska North Platte : ............. .308 Omaha .................. .402 Nevada Las Vegas ............... .702 Reno ................... .775 New Brunswick ................ 506 New Hampshire ............... .603 New Jersey Elizabeth ................. 908 Jersey City ............... 201 Newark ................. .973 New Brunswick ........... .732 Trenton .................. .609 New Mexico ................. 505 New York Albany ................... 518 Queens .................. .718 Buffalo .................. .716 Long Island ............... .516 Ithaca ................... .607 Manhattan ........... 212/917 Syracuse ................. 315 White Plains .............. .914 Newfouadland ................ .709 North Carolina Asheville ................ .828 Charlotte ................ .704 Fayetteville ............... .910 Greensboro ............... .336 Raleigh .................. .919 Rock,/Mountain ........... 252 North Dakota ................. .701 Northern Marianas ............ ,670 NW Territories/Yukon .......... .867 Nova Scotia & Prince Edward Island ........... .902 Ohio Canton .................. .330 Cincinnati ................ .513 Cleveland ................ .216 Columbus ................ .614 Oayton .................. 937 Marietta ................. 240 Oberlin .................. .440 Toledo .................. .419 Oklahoma Enid .................... .580 Oklahoma City ............ A05 Tulsa .................... .918 Ontario Hamilton ................. .905 London .................. .519 North Bay ................ .705 Ottawa .................. 613 Thunder Bay .............. .807 Toronto .................. .416 Oregon Eugene ................... 541 Portland ................. .503 Pennsylvania Allentown ................. 610 Altoona .................. 814 Harrisburg ............... .717 New Castle .............. .724 Philadelphia .............. 215 Pittsburgh ................ .412 Puerto Rico ................... 287 Quebec Montreal ................. 514 Quebec .................. .418 Sherbrooke ................ 819 Trois Rivi~,res ............. .450 Rhode Inland ................. .401 St. Kitts ...................... .869 SL Lucia ...................... 758 St. Vincent ................... .809 Saskatchewan ................ .306 Soulh Carolina Charleston ................ 843 Columbia ................. 803 Greenville ................ .864 Santh Quknta ................. .605 Tenaassaa Columbia ................. 931 Memphis ................. 901 Nashville ................ .615 Knoxville ................. .423 Texas Amarillo ................. .806 Austin .................... 512 Brownsville ............... 956 Dallas ................... 214 Deer Park ................ .281 Del Rio .................. .830 El Paso ................... 915 Fort Worth ................ 817 Galveston ................ .409 Houston .............. .713/281 Irving ................... .972 San Antonio .............. .210 Tyler .................... .903 Waco ................... 254 Wichita Falls ............. .940 Trinidad & Tobago ............. .868 Turks & Coicos ............... .649 U.S. Virgin Islands ............ .340 Utah Salt Lake City ............ .801 Prove .................. .435 Vermont ...................... .802 Virginia Alexandria ................ 703 Richmond ................. 804 Boanoke .................. 540 Norfolk .................. 257 Washington Bellevue ................. .425 Olympia .................. 360 Seattle ................... 206 Spokane .................. 509 Tacoma .................. 253 West Virginia ................. .304 Wisconsin Green Bay ................. 820 Madison .................. 608 Milwaukee ............... A14 Eau Claire ................ .715 Wyaming ..................... .307 Note: Two codes separated by a slash (e.g., in Houston, Texas) indicate an overlay; multiple codes are used for the same geographic area. ..............280 © TeleGeographu, Inc. 1998 TeleGeography 1999 North American Area 201 New Jersey 202 District of Columbia 203 Connecticut 204 Manitoba 205 Alabama 206 Washington 207 Maine 208 Idaho 209 California 210 Texas ZtZ New York Z13 California 214 Texas 215 Pennsylvania 216 Ohio 217 Illinois 218 Minnesota 219 Indiana 225 Louisiana 228 Mississippi 240 Maryland 242 Bahamas 246 Barbados 248 Michigan 250 British Columbia 252 North Carolina 253 Washington 254 Texas 256 Alabama 264 Anguilla 267 Pennsylvania 268 Antigua & Barbuda 281 Texas 284 British Virgin Islands 301 Maryland 302 Delaware 303 Colorado 304 West Virginia 305 Rorida 306 Saskatchewan 307 Wyoming 308 Nebraska 309 Illinois 310 California 312 Illinois 313 Michigan 314 Missouri 315 New York 316 Kansas 317 Indiana 318 Louisiana 319 Iowa 320 Minnesota 323 California 330 Ohio 334 Alabama 336 North Carolina 340 U.S. Virgin Islands 345 Cayman Islands 352 Florida 360 Washington 401 Rhode Island 402 Nebraska 403 Alberta 404 Georgia 405 Oklahoma 406 Montana 407 Florida 408 California 409 Texas 410 Maryland 411 Directory Assistance 412 Pennsylvania 413 Massachusetts 414 Wisconsin 415 California 416 Ontario 417 Missouri 418 Quebec 419 Ohio 423 Tennessee 425 Washington 435 Utah 440 Ohio 441 Bermuda 443 Maryland 450 Quebec 473 Grenada 484 Pennsylvania 500 Personal Comm.Serv.(PCS) 501 Arkansas 502 Kentucky 503 Oregon 504 Louisiana 505 New Mexico 506 New Brunswick 507 Minnesota 508 Massachusetts 509 Washington 510 California 512 Texas 513 Ohio 514 Quebec 515 Iowa 516 New York 517 Michigan 518 New York 519 Ontario 520 Arizona 530 California 540 Virginia 541 Oregon 559 California 501 Rorida 562 California 570 Pennsylvania 573 Missouri 580 Oklahoma 601 Mississippi 602 Arizona 603 New Hampshire 604 British Columbia 505 South Dakota 606 Kentucky 607 New York 608 Wisconsin 609 New Jersey 610 Pennsylvania 611 Repair Service 612 Minnesota Codes, by 613 Ontario 614 Ohio 615 Tennessee 616 Michigan 617 Massachusetts 618 Illinois 619 California 626 California 630 Illinois 640 Turks & Caicos 650 California 551 Minnesota 660 Missouri 661 California 664 Montserrat 670 Northern Marianas 671 Guam 670 Georgia 701 North Dakota 702 Nevada 703 Virginia 704 North Carolina 755 Ontario 706 Georgia 707 California 708 Illinois 708 Newfoundland 710 U.S, Government Emergency 712 Iowa 713 Texas 714 California 715 Wisconsin 716 New York 717 Pennsylvania 718 New York 719 Colorado 720 Colorado 724 Pennsylvania 727 Rorida 732 New Jersey 734 Michigan 740 Ohio 757 Virginia 758 St, Lucia 760 California 765 Indiana 767 Dominica 770 Georgia 773 Illinois 775 Nevada 780 Alberta 781 Massachusetts 784 St. Vincent/Grenadines 785 Kansas 786 Florida 787 Puerto Rico 800 Toll-free services 801 Utah 802 Vermont 803 South Carolina 804 Virginia 805 California 806 Texas 807 Ontario 808 Hawaii Number 809 Dominican Republic 809 St Vincent 810 Michigan 812 Indiana 813 Florida 814 Pennsylvania 815 Illinois 816 Missouri 817 Texas 818 California 819 Quebec 828 North Carolina 830 Texas 831 California 843 South Carolina 847 Illinois 850 Florida 860 Connecticut 864 South Carolina 867 Northern Territories 868 Trinidad &Tobago 869 St. Kitts/Nevis 870 Arkansas 876 Jamaica 877 Toll-free services 880 Toll-free services 881 Toll-free services 882 Toll-free services 888 Toll-free services 500 Information Services 501 Tennessee 902 Nova Scotia & PEI 903 Texas 904 Florida 505 Ontario 906 Michigan 907 Alaska 908 New Jersey 909 California 810 North Carolina 911 Emergency Services 912 Georgia 913 Kansas 914 New York 915 Texas 916 California 917 New York 918 Oklahoma 919 North Carolina 920 Wisconsin 925 California 931 Tennessee 937 Ohio 940 Texas 941 Florida 949 California 954 Florida 936 Texas 970 Colorado 972 Texas 073 New Jersey 978 Massachusetts 281 TeleGeography 1999 © TeleGeography, Inc. 998 282 North American Area Codes ~~ Edmonton ¯ BRITISH COLUMBIA ~’~\~ l~ 2.,~ % SASKATCHEWAN "~h~uver 3~ ~BERTA WAShiNGTON MONTANA OREGON 54x IDAHO 20B NEVADA CALIFORNIA UTAH ARIZONA ,/ !, ¯ Phoenix WYOMING 970 Regina ¯ ¯ Uet~ver COLORADO 719 NEW MEXICO ,~0’ Personal services ~/~I Toll-free Toll-free to U.S, from Canada and Caribbean Winnipeg ¯ NORTH DAKOTA 218 MINNESOTA 701 SOUTH DAKOTA \~EBBASKA IOWA ~x5 712 "\ 316 OKLAHOMA i ! 918 -~ .............. ..~’~ ,, TEXAS y ,,~...:? .............. 915 254 ./":~ -.. © TeleGeographyo Inc. 1998 TeleGeography 1999 807 ONFABIO 705 715 MICHIGAN INDIANA 314 = 573 ARKANSAS 937 6/8 K~NTUCKY ~.~ 615./ TEr’~JESSEE 931 931 z 256 7O6 205 ALABAMA SIPPI 318 519 Quebec 418 ~614 3O4 i804t GEORGIA 912 MAINE For map licenses, send email to: info@telegeogra phy.com © TeleGeography, Inc. 1998 283 End Pages TeleGeo~raphy 1999 © TeleGeographg, Inc. 1998 Key Maximum Data Rate (Kbps) 0-1 1-10 10-40 40-150 150-300 300-1,000 Note: Background net represents solar plane. Distances to planets from Earth are notto scale. Average actual distance in millions of Km are: 78 Earth-Asteroid Belt: 256 629 1,277 Map designed by Michael Slage and The Society for International Space Cooperation http://www.sisc.org~4~! Source: NASA and the Institute of Space and Astronautical Science, Japan. All rights reserved. © TeleGeography, Inc. 1998 286 © TeleCeographg, Inc. 1998 TeleGeography 1999 Mission Histories and Bandwidth Mission Year Launched Transmitter Power Band Type Hz Max Kbps Min Kbps Cassini 1997 20 Watt X-Band 8GHz 249.0 20.0 Deep Space 1 1998 15 Watt X-Band 8GHz 8.0 2.0 Deep Space 2 1999 7.0 1.0 Galileo** 1989 20 Watt S-Band 2294 MHz 0.16 0.04 Hubbla Space Telescope 1990 1" S-Band 2294 MHz 1,000.0 0.5 Huygens 1997 10 Watt S-Band 2295 MHz 8.0 0.001 Mars Climate Orbiter 1998 15 Watt X-Band 8GHz 249.0 20.0 Mars Global Surveyor 1996 25 Watt X-Band 8GHz 85.0 21.33 Mars Pathfinder 1996 13 Watt X-Band 8GHz 25.0 2.0 Mars Polar Lander 1999 tt tt tt 128.0 2.0 NEAR 1996 5 Watt X-Band 8GHz 27.0 6.0 Nozomi(Planet-B) 1998 2.5 Watt S-Band (X-Band) 2294 MHz (8 GHz) 32.0 0.064 Stardust 1999 15 Watt X-Band 8GHz 8.0 2.0 Voyager 1 $ 1977 23 Watt S-Band (X-Band) 2294 MHz (8 GHz) 0.16 0.16 Voyager 2 $ 1977 23 Watt S-Band (X-Band) 2294 MHz (8 GHz) 0.16 0.16 Data will be stored on-board the NM-DS-2 probe and then transmitted to the orbiting Mars Global Surveyor spacecraft for relay back to the Deep Space Network (DSN) on Earth. ** Originally 8.415 GHz X-Band at 134 Kbps. t The Hubble Space Telescope was the first scientific spacecraft designed to utilize the full capabilities of the Earth-orbiting Tracking and Data Relay Satellite System (TDRSS), communicating over either multiple-access or single-access channels at any of the supported transmission rates. ft The orbiter overflies the lander for typically 5 8 minutes about 10 times a Martian day. During this brief time the lander communicates with the orbiter at high speed: 128,000 bits/second vs. 2,000 bits/second with conventional X-band telecommunications systems. $ Maximum data return during active mission phase was 115.2 Kbps. The Deep Space Network continues to receive daily Ultra Violet and fields/particles data. The imaging cameras, photopolarimeter and IR spectrometer were turned off on both spacecraft after Voyager 2’s 1989 Neptune flyby. The remaining seven instruments (UV spectrometer, cosmic ray telescope, low energy charged particle expt., magnetometer, plasma and plasma wave subsystems and the plane- tary radio astronomy unit) continue to return data at a combined rate of 100 bps. Their data return and communications could continue to 2017 when power levels go critical. © TeleGeography, Inc. 1998 287 TeleGeo~aphy 1999 © TeleGeographg, Inc. 1998 288 Over 1,000 international telephone companies are now in business. How many can you name? New International Carriers 1998 The market for international telephone services is global. And so is our directory. Because we know that it’s hard to stay ahead of the compe- tition if you don’t know who it is. TeleGeography’s unique three volume directory has the market intelli- gence you need to stay on top of the market. The books provide cru- cial information on over 1,000 international carders in the world’s top 110 economies. Each carder listing contains: ¯ Address, telephone, fax, web site ¯ Names of key contacts and executives ¯ Status of international service license ¯ Details on shareholders and affiliates ¯ Stock symbols and exchange listings Buy all three volumes for $I 195. The price for single volumes is: $495 (Americas; Europe) and $385 (Asia). To order, contact us at: TeleGeography, Inc. tel: +1 202 467 0017 fax: 1 202 467 0851 http://www.telegeography.com info@telegeography.com About TeleGeography, Inc. telegeography \t~l "a-je-Og’ra-f~ n (1990) abbrv, of telecommunications geography [fr. Gk tele, far off, at a distance and L. communicatus, pp. of communicare to impart ft. Gk geo (earth) graphein, (to write)] 1. a new branch of geography that maps the pattern of tele- phone traffic and other electronic communication flows; 2. places created by or perceived solely via telecommunications (e.g., a computer network address); 3. the telecommunications artifacts (radio antennae, terminals, signs) on a site; 4. the balance of telecommunications power in one country or region vis-a-vis another (cf. geopolitics, archaic). The old geography of countries and coast lines is giving way to a new geography marked by tele- phone codes, satellite "foot prints" and lnternet addresses. Electronic networks have made the world smaller. But they also have created countless new places where people work and play. This expanding electronic terrain--call it telegeography~emands a new cartography. That is the raison d’etre of TeleGeography, Inc. (TGI). Founded by Washington, D.C. lawyer Gregory C. Staple, TGI is the world’s leading publisher of reports on international telecommunications flows. It also is an authoritative source of data on the ownership of information-communication compa- nies worldwide. In addition, the company provides customized mapping and consulting services. TGI’s flagship report--TeleGeographg--has been published annually since 1989 in cooperation with the London-based International Institute of Communications (IIC). Other TGI publications include New International Carriers, a three volume directory of competing international telephone companies, and the Telecommunications Map of the World, a poster-sized map of world telecommunications. TGI’s directories are used by leading communication companies, consultancies, governments and financial institutions in over 60 countries. To learn more, please visit us at http://www.telegeography.com. TELEGEOGRAPHY, INC. 1730 Rhode Island Avenue, NW ¯ Suite 1205 Washington, DC 20036 USA Tel. 1 (202) 467-0017 ¯ Fax 1 (202) 467-0851 ¯ E-mail: info@telegeography.com http://www.telegeography.com THIS REPORT IS PUBLISHED ANNUALLY BY TELEGEOGRAPHY, INC. SUITE 1205 1730 RHODE ISLAND AVENUE, NW WASHINGTON, DC 20036 USA TEL. I (202) 467-0017 ¯ FAX I (202) 467-0851 E-MAIL: info@telegeography.com ¯ http://www.telegeography.com ISBN 1-886142-15-7