THE INTERNET, ECONOMIC GROWTH, AND TELECOMMUNICATIONS POLICY By Charles H. Ferguson Visiting Scholar, Center for International Studies, M.I.T., and Haas School of Business, University of California, Berkeley April 14, 1997 Email: chf@cferguson.com; Mail: 65 Arden Road, Berkeley CA 94704 SYNOPSIS This essay analyzes policy issues related to future local telecommunications services, particularly the infrastructure needed for Internet services, electronic commerce, open-systems HDTV, videoconferencing, and future voice telephony. I conclude that neither the Telecommunications Act of 1996 nor Federal policy has yet resolved these issues. The 1996 Act and FCC policy fail to give Internet and/or Enhanced Service Providers (ISPs/ESPs) sufficient interconnection and collocation rights with respect to monopoly local exchange telephone carriers (LECs), which creates entry barriers, retards technical progress, and may cause network congestion. U.S. policy has also allowed LEC underinvestment in technology, monopolistic practices, mergers, and anticompetitive conduct. Local services could thus remain dominated by LECs whose technological performance, strategic conduct, and lobbying are inappropriate for a modern information economy. Evidence suggests this could cause major damage to the telecommunications industry itself, to high technology industries, and to U.S. economic growth. Several reforms are therefore recommended. ISPs, ESPs, and long distance firms should be granted full rights to interconnect with LEC networks, to collocate data networking equipment on LEC premises, and to offer data services without the any requirement to provide conventional telephone service. Second, the CATV industry should also be unbundled. Third, the LECs' anticompetitive behavior should be curbed via antitrust and/or other sanctions. Fourth, corporate spending on regulatory lobbying and academic policy research should be curbed or at least fully disclosed. Finally, the efficiency, high technology expertise, and possibly independence of the FCC and the Department of Justice Antitrust Division require improvement. C Copyright 1997 by Charles H. Ferguson. Fair use, including downloading from the Internet, is permitted if (a) no further copying occurs, (b) this copyright notice is included in all copies, and (c) no commercial use occurs. Absolutely no other copying without permission. URL for this paper beginning 5/1/97: http://www-eecs.mit.edu/people/ferguson. The author is extremely grateful for the extensive research and administrative assistance of Simone Ross, for the comments of Charles Morris on early drafts, and for information provided by many others. However, the author alone is responsible for the contents of this document, including all remaining errors. TABLE OF CONTENTS EXECUTIVE SUMMARY - p. 4 1. INTRODUCTION AND OVERVIEW - p. 8 1.1 Local Telecommunications: The Unfinished Business of U.S. High Technology 1.2 Local Telecommunications: A High Technology Industry Not Behaving As One 1.3 Structure of the Essay 2. TELECOMMUNICATIONS IN THE INTERNET AGE: VERY HIGH STAKES - p.21 2.1 An Overview of the Revolution 2.2 Economic Effects of Telecommunications Modernization Upon the Telecommunications Industry Itself 2.3 Economic Effects of Telecommunications Services on U.S. High Technology Industry 2.4 Economywide Effects of the Internet and Telecommunications Modernization 3. THE CURRENT LOCAL SYSTEM: ASSESSMENT AND ANALYSIS - p. 43 3.1 Introduction 3.2 LEC Financial and Operational Performance 3.3 LEC Technological Behavior and the Price/Performance of LEC Digital Services: An Astonishing Situation 3.4 LEC Capital Investment and Financial Priorities 3.5 LEC Strategic Behavior, Lobbying, and Funding of Policy Researchers 3.6 The LEC "Internet Congestion" Allegations 3.7 LEC Governance and Management 3.8 Another Sobering Benchmark: AT&T Since Divestiture 3.9 The CATV Industry: No Savior, and Often Similar 3.10 In Summary: There Is a Problem Here CONTENTS (Continued) 4. CURRENT CONDITIONS: POLICY & STRUCTURAL IMPEDIMENTS TO INTERNET GROWTH AND INFRASTRUCTURE MODERNIZATION - p. 122 4.1 Technology, Market, and Economic Problems 4.2 The Current Policy Regime: The 1996 Act, The FCC, The Antitrust System, State PUCs, and the Courts 4.3 Long Term Pressures In Favor of Progress 4.4. In Conclusion: Action Still Required 5. POLICY IMPLICATIONS - p. 139 5.1 Overview; General Policy Objectives 5.2 Specific Policy Issues and Recommendations 5.3 The Structure and Conduct of Federal Regulatory Institutions 5.4 Wider Implications: Electric Power, Foreign Telecommunications Markets 5.5 Conclusion: An Unusual Opportunity AUTHOR INFORMATION AND DISCLOSURE - p. 155 NOTES - p. 156 STATISTICAL GRAPHS - p. 164 EXECUTIVE SUMMARY The Internet, Economic Growth, and Telecommunications Policy by Charles H. Ferguson Contact: chf@cferguson.com This paper analyzes policy issues related to future local telecommunications services, particularly the infrastructure and digital services needed for advanced Internet services, electronic commerce, open-systems HDTV, videoconferencing, and future voice telephony. I conclude that the Telecommunications Act of 1996 does not sufficiently address these issues. Even more importantly, current FCC and Department of Justice policies have created major barriers to new entrants and to the development of a technologically progressive, competitive industry. As a result, local services remain dominated by monopoly telephone companies whose technological performance is unacceptably poor for a modern information economy. It is unlikely that competition from the cable television industry will compensate for these problems. Changes in Federal policy are therefore required if the local loop is not to become a significant bottleneck to U.S. economic growth. Background. It is now clear, and generally accepted, that the rise of the Internet represents a major technological, economic, and educational revolution. It has also created one of the most important economic policy issues in U.S. history, one involving the contending interests of a half dozen industries that directly account for over one-tenth of U.S. GNP. Underlying this policy issue is a technological revolution that is inverting the previous relationships between voice telephone service, on the one hand, and digital services on the other. The traditional networks that still dominate U.S. public telecommunications were designed primarily to provide circuit-switched voice services, with data services a secondary function added onto the voice network. Indeed for the last decade, providers of data services have been termed Enhanced Service Providers, or ESPs, by the FCC, and are still so termed in the 1996 Telecommunications Act and the first FCC Order implementing it1. This is one symptom of the slowness of FCC and other Federal policymaking, which remains excessively focused on traditional voice services to the neglect of digital systems and data services. Moreover, the direction of modern networking technology, together with the rise of Internet services, is to reverse the traditional relationship between voice and data services. Already, Internet-driven data networks are being constructed alongside the telephone network and interconnected with it (albeit suboptimally so, due to collocation restrictions). Eventually, voice telephony will become one application among many on a high speed, general purpose, digital, packet-switched (as opposed to circuit-switched) network. In fact, technology and market trends suggest that within a decade, the total bandwidth used for Internet services will be higher than that used for conventional voice traffic. New networking technology encourages deployment of digital and/or Internet services that will be able to provide voicemail, fax, broadcast audio, and even video in addition to electronic mail and Web services. These new technologies also favor the conversion of the communications industry from a centralized hierarchically managed monopoly industry to a modular, decentralized, competitive industry following the "Silicon Valley Model,"2 e.g. similar to the personal computer industry. Economic Importance of Reform. Much evidence suggests that this transition from monopoly to decentralized competition is critical for U.S. economic growth. There is reason to believe that the transition from the current industry to "Silicon Valley" behavior could accelerate the industry's technological progress by 20% per year or more (see chapters 2 and 3), and that this could raise productivity growth in the entire U.S. economy by 1% per year or more, i.e. a doubling of recent productivity growth. There is also reason to believe that this transition could increase global productivity as well. This is because the price-performance of U.S. local telecommunications services (particularly digital services such as ISDN, HDSL, HDSL2, SDSL, T1, T3, ADSL, and VDSL3) is now becoming the critical bottleneck to the future progress of Internet access, Web services, electronic commerce, and even voice telephony. Improvement in these Internet-related and other digital communications services is becoming the most critical driver and/or bottleneck for growth in the U.S. information technology sector, i.e. industries such as semiconductors, personal computers, networking equipment, and software. And, in turn, progress in telecommunications and high technology is critical to realizing major economywide productivity gains, including the transformation of financial services, publishing, payments and billing processes, telecommuting, collaborative work systems, and many retailing, distribution, and customer service functions. Regulated Industry Performance. Unfortunately, the conduct and performance of the local telephone industry is unacceptable by high technology standards. This industry is dominated by the eight local exchange carriers, or "LECs," i.e. the seven Regional Bell Operating Companies and GTE. The behavior of the LECs (and also, often, the cable TV industry) is the precise opposite of competitive high technology patterns. These companies largely operate the U.S. local network as a closed system, one which is treated as a cash cow used to fund diversifications into cable television, wireless businesses, and foreign markets. The LECs' capital investments in the U.S. network are generally flat or even declining; they perform little R&D and commercialize new technologies slowly; and their services display very low, or (astonishingly) sometimes negative progress in price/performance, even in areas such as data communications whose underlying technologies are improving over 50% per year. Moreover, the LECs have many cooperative and/or anticompetitive relationships with each other, the wireless industry, and even some with the CATV industry. They routinely engage in strategic and political actions, often cooperatively, to oppose structural reforms, open systems architectures, competition, and modernization. The LEC spend hundreds of millions of dollars per year on litigation, regulatory filings, lobbying, political contributions, industry associations, and funding of academic experts' policy research and consulting. Finally, the LECs appear to have management and corporate governance problems. LEC managements generally lack skills and experience appropriate for a competitive, Internet-driven industry. In some cases the personal interests of LEC executives and directors may also be inconsistent with competition and reform. LEC operational efficiency, technological performance, strategic behavior, and governance are considered in detail in Chapter 3, which also briefly assesses the CATV industry. Inadequacy of Current Federal Policy. Federal policy has not sufficiently addressed these problems. In principle, the 1996 Act ended the regulated monopoly regime and established the basis for a competitive local telecommunications industry in the United States. Yet 15 months after its enactment, little competition and/or technical progress is visible in either conventional voice service or digital services such as high speed Internet access. This is not an aberration, but rather the combined result of flaws in the 1996 Act, defects in FCC policy, passive antitrust policy, LEC resistance and delaying tactics, and political pressures from a variety of sources. The 1996 Act. The 1996 Act requires incumbent local telecommunications carriers to unbundle their networks and to allow competitors to interconnect with them. However, the Act is primarily oriented towards traditional voice telephone systems, with apparently little consideration given to future requirements, modern digital technologies, and/or the Internet. It only requires LECs to interconnect with "telecommunications carriers," not ISPs or ESPs, and only requires that LECs permit collocation of equipment required for telecommunications services. However, the Act is also quite flexible, perhaps in part because Congress and the Administration were loathe to take sides. The Act permits and even in some areas requires the FCC to take independent actions to promote competition and the deployment of advanced services, and even allows the FCC to override other provisions of the 1996 Act in order to do so. Furthermore the FCC possesses other regulatory powers which provide it with substantial latitude to demonopolize and/or restructure the industry, and which are independent of the 1996 Act.4 However, the FCC and the Justice Department have not acted to promote the development of a technologically progressive, competitive industry. Both organizations appear to have little institutional understanding of high technology, and to have slow, highly bureaucratic procedures poorly suited to rapidly changing industries. In fact the FCC's interpretation and implementation of the 1996 Act has been extremely restrictive. FCC Policy. In the first of three projected Orders pursuant to the 1996 Act, the FCC explicitly declined to give the Internet industry and all other ESPs the right to compete, interconnect, and/or collocate equipment with the incumbent local exchange telephone carriers (or LECs, an industry dominated by the Regional Bell Operating Companies plus GTE). The FCC has also declined to require subloop unbundling, which is increasingly important for competition in data services, and declined to require the LECs to permit collocation of any equipment except that required to provide traditional telecommunications services. Thus, for example, advanced networking equipment necessary for high speed Internet access cannot be collocated on LEC premises without LEC permission. No LEC has voluntarily granted such permission. Current Antitrust Policy. Federal antitrust policy towards the LECs, and potentially the CATV industry, has been extremely passive. The refusal of the LECs to offer HDSL and SDSL services, the LECs' failure to improve price-performance in monopoly ISDN and near-monopoly T1 services, the lack of competition in local service more than a year after signature of the 1996 Act, the absence of significant competition between the LECs in any of their core businesses, the LECs' structural and financial relationships with each other, the clear patterns of LEC cooperative behavior, the LECs' entry into businesses such as Internet access on terms unavailable to ESPs, the merger plans of 4 of the 8 major LECs, and cooperative LEC litigation to delay implementation of the 1996 Act, are all extremely disturbing signs. The inaction of the FCC and more particularly the Department of Justice in the face of these conditions is disturbing, although perhaps not surprising given the political sensitivity of the issue and the low technological sophistication of these organizations. In fairness to the FCC, it has done several things well, including resisting LEC pressure to levy access charges on Internet providers. Its policymaking has also been subjected to intense lobbying and political pressures, and has sometimes been blocked by industry-sponsored litigation. However, critical Internet policy issues remain largely unresolved, the incumbents' power remains, and entry by innovative new firms is being impeded. As a result, there is now a major risk that development of digital and/or Internet services will be retarded by the private interests of the LECs. There is also some risk that both the public voice network and the Internet could suffer increasing congestion as a combined result of LEC underinvestment, slow LEC deployment of new technologies, and the absence of competitive interconnection from innovative data services firms. While increasing pressure from consumers and high technology industry will eventually force change, the difference between prompt action and delay could have major consequences for U.S. economic growth and social welfare - comparable, in fact, to U.S. fiscal and budgetary policies. These matters are discussed in chapters 2 and 4. Policy Recommendations. Several reforms are recommended. First, Internet providers and all other ESPs should be granted full interconnection and collocation rights for LEC networks, including the right to purchase unbundled local loops and to collocate data networking equipment. The current requirement that interconnecting firms be "telecommunications carriers" should be eliminated or redefined. Second, unbundling should increasingly be required of the cable television (CATV) industry as well. Third, the anticompetitive behavior of the incumbent local carriers should be addressed via antitrust and/or other sanctions. Fourth, industry regulatory lobbying and funding of university policy researchers should be curbed or at least publicly disclosed. And finally, measures should be taken to increase the efficiency, technological expertise, and general high technology awareness of the FCC and DOJ Antitrust Division. Note to readers: There is some duplication between the Executive Summary and the introduction and overview presented in chapter 1. Those readers familiar with the Internet and/or telecommunications industries, and who have read the Executive Summary, can proceed directly to chapter 2. 1. INTRODUCTION AND OVERVIEW 1.1 Local Telecommunications: The Unfinished Business of U.S. High Technology It is now clear that the Internet is driving a major technological, economic, and educational revolution. It has also created one of the most important economic policy issues in U.S. history, and generated a struggle over political and market influence involving a half dozen industries that directly account for over one-tenth of U.S. GNP. The affected sectors include local telephone carriers, cable television, long distance telecommunications, the computer and software industries, the networking and telecommunications equipment industries, consumer electronics, the print publishing and broadcast media sectors, online information services, and the embryonic Internet industry itself. As we shall see shortly, U.S. policy towards these industries and the Internet could easily be comparable in importance to US. macroeconomic or fiscal policy in promoting, or retarding, U.S. economic growth and social welfare in coming decades. Underlying these questions is a technological revolution that is inverting the previous relationships between voice telephony and data services. The traditional networks that still dominate U.S. public telecommunications were designed primarily for circuit-switched voice services, with data services a secondary function. Indeed even in the Telecommunications Act of 1996 and the first FCC Order implementing it, providers of data services are still termed Enhanced Service Providers, or ESPs. However, the direction of modern networking technology, together with the rise of the Internet, is to reverse the traditional relationship such that voice telephony becomes one application among many on a high speed, general purpose, digital, packet-switched network. In fact, technology and market trends suggest that within a decade, the bandwidth used for Internet services will be far higher than that used for pure voice traffic. Increasingly, modern networking technology permits digital and/or Internet services that handle voice, data, fax, and even video. This new technology also favors the conversion of the communications industry from centralized hierarchically managed monopoly systems to a modular, decentralized competitive industry similar in structure to the personal computer industry. However, as a result of legal and regulatory deficiencies, Federal policy has yet to address these trends, despite signature of the Telecommunications Act of 1996. In part this failures derives from deficiencies in the Act itself, which is primarily oriented towards traditional voice telephone systems, with apparently little thought given to modern digital technologies. However, the 1996 Act is also quite flexible, perhaps because Congress and the Administration were loathe to take sides. It permits and even requires the FCC to promote competition and the deployment of advanced services, and even permits the FCC to override provisions of the Act in order to do so. The greater problem is that the FCC's policies, including its interpretation and implementation of the 1996 Act, have been restrictive and even reactionary. FCC policy has explicitly declined to give the Internet industry and all other ESPs the right to compete, interconnect, and/or collocate equipment with the incumbent local exchange telephone carriers (or LECs, an industry dominated by the Regional Bell Operating Companies plus GTE). FCC policy has also declined to require LECs to permit the collocation of any equipment except that required to provide basic telecommunications services. Recent policy decisions by the FCC and also by the Department of Justice Antitrust Division have also permitted increased concentration in the industry. The antitrust division has already approved one huge merger between LEC monopolies (SBC and Pacific Telesis; another merger, between Nynex and Bell Atlantic, is not yet finally resolved). Antitrust policy has also simply failed to challenge or curb a wide array of LEC monopolistic, cooperative, and anticompetitive practices. The combined result of these policies is effectively to ratify the LECs' power, reduce technological progress, and restrict participation in the industry to a small number of large incumbent firms. There is also some possibility that current conditions - i.e., continued LEC dominance, insufficient LEC capital investment, and slow LEC deployment of new technology - could give rise to serious congestion of the public telephone network. While the LECs have warned of network congestion caused by Internet growth, continuation of their own behavior is a far more likely cause of such problems. Thus, some of the most important policy issues related to the Internet revolution remain unaddressed. In fairness to the FCC, part of the difficulty is that FCC policymaking has been subjected to great political pressures and sometimes blocked by industry-sponsored litigation. However, in large measure the FCC and other Federal agencies do possess the power to make good policy, yet have failed to do so. There is now a major risk that development of digital and/or Internet services will be obstructed by the private interests of the telecommunications industry, particularly the LECs. This issue has created a conflict which primarily involves the LECs, cable, and media industries on one side and competitive information technology industries on the other, with the long distance communications industry somewhere in between. Internet growth clearly benefits, and is embraced by, most of the information technology sector. However, its continued growth requires an acceleration of technical and economic progress in the delivery of Internet services and digital communications generally. This threatens to destabilize a wide array of long-established, extremely comfortable, institutional and personal arrangements, particularly in the local telecommunications and cable television (CATV) industries. As the Internet develops further, it will also begin to transform and destabilize other industries, including the entertainment media, print publishing, retailing, distribution, and financial services sectors. We can therefore anticipate a series of debates over the impact of Internet technology in coming years. This situation has many historical precedents. Technological revolutions often undermine previously dominant institutions, as well as the leaders and beneficiaries of these institutions. This is particularly true when the incumbents' power is based upon an obsolete technology, when firms have become too rigid or inefficient to respond effectively to change, and/or when such change threatens the personal interests of leaders. Unfortunately, threatened incumbents are often much more effective at blocking change than embracing it. Even if the new technology wins in the end, the drag caused by the resistance of entrenched interests can be extremely damaging to a nation's economic growth and international position. The comparative openness of the U.S. economic system - to innovation, to the creation and financing of new companies, and to competitive entry against existing firms - is a major reason for America's strength in most new industries, including those driving the information technology revolution. The United States contains very few state-owned industries or legally protected monopolies, while conversely U.S. law and policy generally restrain monopolies and incent the commercialization of new technology and the creation of new firms. Silicon Valley and the Internet revolution are testaments to the success of this approach, while the pervasive decline of European high technology industries over the past decade illustrates the dangers of long-term state-protected monopolies. While the U.S. high technology system generally performs well, it has a number of weaknesses, at least in this case. The first and most important is that U.S. high technology contains two large regulated monopoly industries, the LEC and CATV industries. While these industries and their regulators deserve credit for the historical provision of universal telephone service, their conduct also combines some of the worst features of bureaucratic regulation and monopoly power. The inefficiency, monopoly power, and political influence of these industries, particularly the LECs, are now becoming the largest bottleneck to the further progress of the Internet, the computer industry, and even of voice services. Absent change, they will soon become a significant limiting factor on U.S. economic growth. The second, somewhat related problem, concerns regulation. Like the LECs themselves, regulators such as the FCC, the DOJ Antitrust Division, and state PUCs are not well adapted to dealing with competitive, rapidly changing, high technology environments, and their policymaking shows it. The third problem is corporate governance. The U.S. system contains few mechanisms for sanctioning mismanagement by entrenched top executives and boards of directors of large firms. Such failures of accountability can result in major damage to companies and industries, and there is some reason to believe that this is a problem in the U.S. communications industry. Finally, the U.S. system sometimes has difficulty financing competitive entry in markets with large capital requirements. The U.S. venture capital and startup systems are superb at generating new, innovative competitors when initial capital requirements are comparatively small. However, this system is less well equipped for situations in which large up-front investments are required. As a result, the development of competition in such industries sometimes takes a very long time, and/or falls to foreign competitors (as occurred for example in the automobile and steel industries). In the case of the Internet, both LEC strategic behavior and, unfortunately, Federal regulatory policy have raised major barriers against the entry of innovative small firms into advanced digital communications services, including high speed Internet service. These problems have creates a major impediment to future U.S. economic growth. While telecommunications technology and regulation are extremely complex, the principal issue facing policymakers can be stated quite simply: how to transform local telecommunications from a slow moving, inefficient monopoly sector into the modern, open, competitive high technology industry it should already be. (In the longer term, there may also be a need to forestall the formation of future monopolies in the Internet sector; however, at present local telecommunications is the only major problem.) America's (and most of the world's) politically powerful regulated telephone monopolies - and their executives - are not interested in being submitted to intense competition. They resist competitive discipline through huge expenditures on litigation, lobbying, and academic experts; they also treat their core services as cash cows, perform comparatively little R&D of value, adopt new technology slowly, and deliver astoundingly slow progress in price/performance to their customers. 1.2 Local Telecommunications: A High Technology Industry Not Behaving As One Current digital communications and Internet technology are such that local services should now become a modern information technology industry. Its technologies are now essentially those of large scale networking, which is currently the fastest moving sector in all of information technology, exhibiting growth rates in excess of 75% per year. The constituent parts and functions of local telecommunications - computers, software, wiring, modems and other terminal equipment, computerized network administration systems, switches and routers, system engineering, maintenance, customer support, etc. - are generally similar to those of information technology companies, and/or the internal networks of large information-intensive corporations. As we shall see, most claims made by the LECs that their unique characteristics justify high costs and monopoly status turn out to be of secondary importance, and/or simply untrue. Modern information technology (IT) industries - e.g. semiconductors, computers, software, systems integration, online services, digital consumer electronics, networking equipment - generally have a decentralized, modular "Silicon Valley" industry structure. Their products and services have modular designs and open system interfaces that both reflect and contribute to this industry structure. These industries are highly innovative, entrepreneurial, and competitive; they are characterized by rapidly evolving industrywide technical standards, high levels of R&D spending and capital investment, high rates of technological change, and 30-50% per year improvement in the price/performance of their goods and services. Firms rarely pay large dividends, preferring to reinvest heavily in innovation and growth. And when a leading IT firm becomes lazy, within at most a decade it is brutally punished by new entrants with superior technology - as IBM was by Intel, Microsoft, and Compaq, as Novell was by Microsoft, and as Microsoft perhaps will be by Netscape. Yet, the behavior of both the local telephone and cable TV industries is the precise opposite of this competitive high technology pattern. These companies have centralized management and closed system architectures; their capital investment in modernizing their core businesses is flat or declining; they perform little R&D and commercialize new technologies slowly; and they display very low, zero, or (astonishingly) sometimes even negative progress in price/performance.5 They often act strategically and politically to oppose structural reforms, open systems, competition, and modernization. In principle, the Telecommunications Act of 1996 ended the regulated monopoly regime and established the legal basis for a decentralized, competitive local telecommunications industry in the United States. The 1996 Act requires incumbent local telecommunications carriers to unbundle their networks and to allow competitors to interconnect with them. Yet 15 months later, little competition and/or technical progress is visible in either conventional voice service or digital services such as high speed Internet access. This is not an accident. It is the combined result of flaws in the 1996 Act, mistakes in FCC policy, other Federal policy errors, LEC delaying tactics, and political pressures, including those mentioned above. In fact, I argue in chapter 4 that little progress can be anticipated without major shifts in FCC policy and perhaps DOJ antitrust actions. Under current policy and industry conditions it is probably rational corporate strategy for the LECs to retard technical change and competition, and given current Federal policy they can probably succeed. They have extensive lobbying, litigation, and political capacities; they still possess effective monopolies; they have common interests and close, interlocking relationships with each other and even with the CATV industry. But these companies also resist reform in part precisely because even if it is in the long term interest of the company to prepare for progress and competition, it is not necessarily in the personal interest of their top managers and boards of directors. Like IBM in the 1980s, the LECs appear to have major corporate governance problems resulting in decision-making that can be contrary to the long run interest of the firm's employees, customers, and/or shareholders. Many of the LECs' executives, managers and board members would not succeed in and/or enjoy a modern, highly competitive industry, and therefore resist changes that would lead to such an industry. The executives of U.S. telephone companies know how to run stable, slowly evolving monopoly telephone networks; they also know how to file lawsuits, contribute to PACs, fund academic policy studies, hire professors as expert witnesses, recruit politically powerful board members and lobbyists, prepare complex regulatory filings, and lobby Congressmen and regulators. Conversely they have much less experience with rapidly growing, highly competitive digital systems businesses based on technologies that continuously improve 40% per year. As a result, telecommunications is one of the largest pieces of unfinished business in the American economy. While the 1980s and early 1990s saw the crisis, reform, and/or competitive replacement of many large, bureaucratic Fortune 500 firms (GM, IBM, etc.), the LECs and CATV vendors have remained legally protected monopolies. At present the LECs have about $120 billion in combined annual revenues, as against less than $5 billion for all their competitors in local services combined. Although by world standards the U.S. does have very high quality basic telephone service, the longstanding monopoly power of the LECs is now the major impediment to necessary future progress in telecommunications, Internet services, and U.S. information infrastructure. The CATV industry's record is no better and often similar. LEC and CATV performance and strategy are assessed and analyzed in considerable detail in chapter 3. An Analogy: IBM Mainframes Versus PCs and the Cost of Computing The resemblance between the position of the local telecommunications industry as it faces the Internet revolution, on the one hand, and that of IBM when personal computing arose in the 1980s, on the other, is quite striking. As is all too often the case, both are resisting the cannibalization of a comfortable, profitable monopoly by new and superior technology. By the mid-1980s IBM had long dominated the computer industry, had locked its customers into its mainframe products, and enjoyed high margins. IBM's products were carefully priced to preserve high margins and to prevent arbitrage between low-end and high-end products - which, as we shall see, is also a major issue for the LECs. But in the absence of real competition, price-performance ceased improving rapidly and IBM became lazy. Management deteriorated and the firm gradually became bureaucratic, politicized, and inefficient. Then the advent of microprocessors and personal computing dramatically improved price/performance and brought into being a new industry that was able to undercut IBM and provide rapid technical progress. This development threatened not only IBM but the longstanding personal positions of IBM's executives and its complacent, largely ornamental board of directors. As competitive pressure mounted IBM's top management stonewalled, the company became ever more paralyzed by internal politics, and its board of directors did nothing - until it was plunged into crisis. IBM's decade-long resistance to modern technology and open systems not only caused the company itself $30 billion in losses and 200,000 layoffs, but also substantially slowed technical progress in the computer industry for many years, and probably cost the U.S. economy several hundred billion dollars. The reason for this is that for roughly a decade, the rate of progress in computer price / performance was perhaps 20% per year less than it should have been, even in mainframe computers themselves. Now the LECs and CATV vendors are facing similar writing on the wall, and are reacting largely as IBM did. To the extent they succeed, this will impose huge costs on the economy, and probably also cause them to undergo layoffs, losses, and restructuring when serious competition finally arrives. Furthermore their continued resistance will probably also, not coincidentally, keep the price of traditional telephone, data, and cable services higher than they need to be. In a high technology industry, the best way to reduce costs and prices is to maximize technological progress rather than reducing investment or profits. Hence competitive reform would not only provide vital improvements in data communications for Internet services, but would also probably be superior to the old regulated monopoly system even for the goal of providing inexpensive basic voice service (often referred to as plain old telephone service, or POTS). The general pattern in a competitive information technology sector is that if the average price of a service or product (whether a personal computer, an Internet router, or a local loop) remains constant, the number and quality of functions or services increases and the cost of each one declines rapidly. In a dynamic, modern, local telecommunications industry, POTS would rather quickly become just one, rather undemanding, service provided on all lines; so while the total cost of service over each copper local loop might remain roughly constant, the cost of voice service would decline sharply. If local telecommunications followed the pattern of personal computers, software, networking equipment, or corporate networks, then by the time every U.S. home and business has high speed Internet access, the price of ordinary voice telephone service should probably decline to less than $5 per month. Thus a decade from now, the average household's "phone bill" would be dominated by high speed data services for which the LECs would face serious competition, and which might even provide voice service along with other applications. Thus one reason the LECs resist competitive interconnection and the deployment of modern open systems digital technology is that competitors using digital systems could rapidly and sharply underprice the LECs in both voice and data services. This is discussed in chapter 3. Sizing the Problem The price-performance of local telecommunications bandwidth, e.g. of digital services such as ISDN, HDSL, SDSL, T1, T3, ADSL, and soon VDSL, is now the critical bottleneck to the future progress of communications services, ranging from Internet access to voice telephony. The usage, quality, and price of all Internet based services - including Web services, electronic commerce, electronic mail, Internet radio, electronic process control, video conferencing, and open architecture digital video - is increasingly limited by local communications costs and performance. Technical progress in these services, in turn, is critical to the future health of other U.S. high technology sectors such as computers and software, and to the general growth and competitiveness of the U.S. economy. The economic stakes associated with "putting telecommunications on the technology curve," in the words of one MIT professor,6 are therefore extraordinarily high - far higher even than many participants in the industry realize. Internet-driven local telecommunications policy is arguably the most important, and clearest, economic policy issue of the late 20th century. The difference between continuation of the present situation versus a rapid, effective transition to a modern industry could easily have a larger effect on U.S. productivity, growth, and economic welfare than any feasible or likely change in U.S. macroeconomic or fiscal policy. The calculations described in chapter 2 suggest that development of a modern local communications industry could easily increase U.S. GNP growth by 1% per year or more. The present telecommunications situation thus raises both government policy and corporate governance questions even more serious than those raised by the decline of IBM in the face of personal computing, or by the decline of other large U.S. industries such as the automobile, clothing, and steel companies in the 1980s. Indeed the major differences between the earlier experience of the mainframe computer industry and the current situation in telecommunications is that now, the issues are even more important and the incumbents possess not only size and deep pockets but also legal, regulatory, and political tools derived from their long history of regulated monopoly status. The Regulatory and Antitrust Systems In the case of telecommunications, the industry's history of regulation, and the regulators themselves, also bear significant responsibility for the problem. The LECs and the CATV industry are in part (though only in part) the creations of their regulators, with whom they have co-evolved. In many respects the two sides are disturbingly similar, and can be regarded as two interrelated portions of an outdated system. To be sure, the FCC has improved somewhat in recent years. But the FCC, most state PUCs, and the DOJ Antitrust Division do not yet possess nearly enough of the technological skills, understanding of high technology industries, political independence, or even simple efficiency required to negotiate the Internet revolution. For example the FCC remains almost exclusively dependent upon lawyers and theoretical economists, who are severely burdened by legalistic procedural requirements. The FCC thus remains poorly informed about the technological fundamentals and business realities of the industry, which are often critical to accurate policy judgments. Some recent FCC policies, e.g. provisions of the first of the FCC Orders implementing the 1996 Telecommunications Act, have actively impeded innovation, technological progress, and new entry into digital services. The FCC and many state regulatory agencies are also poorly computerized, collect insufficient industry data, and consequently possess insufficient information about the industries they regulate.7 The DOJ Antitrust Division shares many of these problems as well, and has been remarkably passive with regard to local telecommunications. Moreover, overlaid on the regulatory system are courts, political processes, and lobbying. Many regulatory decisions and arbitrations can be appealed to state and Federal courts, causing further delay and impediments to policymaking. Major portions of the FCC interconnection order concerning pricing of unbundled services pursuant to the 1996 Act are now being challenged by the LECs via Federal lawsuits. The LECs spend several hundred million dollars per year on regulatory lobbying, litigation, political contributions, industry associations, academic consultants and expert witnesses, antitrust consulting, and the like. The long distance, CATV, and broadcasting industries also lobby heavily. All of these factors reduce the regulators' ability to accelerate industry reform. These issues are discussed in chapter 4. Policy: Reform, But Not Yet Laissez-Faire Despite the LECs' political spending, litigation, and other factors inhibiting effective policymaking, the regulators clearly have substantial powers. However, these powers do no good if they are unused. The current situation is such that in the short term U.S. policy must paradoxically become more interventionist in order to effectively deregulate the industry in the long run. Aggressive and sophisticated Federal policy intervention, and continuation of a major regulatory role during the transition to a sustainably competitive and open digital communications services industry are imperative for the development of adequate information infrastructure for the United States. This cannot be achieved, however, without strengthening Federal regulatory agencies and improving their high technology expertise. Thus, recent events demonstrate that the need to reform local telecommunications also implies the need to reform the FCC, antitrust policy, and the regulatory system. 1.3 Structure of the Essay Following this introduction, chapter 2 surveys the status and economic importance of the Internet and open digital communications technology, the nature of modern information technology industries, the rise of Internet-based services, and the macroeconomic importance of bringing the rate of technical progress in telecommunications to competitive levels. Then, chapter 3 assesses the conduct and performance of the LECs and the CATV industry. First I assess their performance in R&D, capital investment, technology deployment, price/performance improvement, customer service functions, and investment priorities. LECs and CATV vendors are benchmarked against their own past behavior, the performance of AT&T, available commercial technology, proxies such as large information intensive corporations, and the performance of the competitive information technology sector (semiconductors, computers, networking equipment, digital consumer electronics, software, online services, outsourcing management, and systems integration). Then I assess the industry's strategic, managerial, and political conduct, particularly its corporate governance, anticompetitive behavior, the use of POTS and digital services as cash cows, and its defense of its monopoly position through politics, lobbying, litigation, and delay. This is followed in chapter 4 by an assessment of current Federal policy and future trends facing private industry, regulators, and policymakers. Chapter 5 assesses policy issues, and provides some policy recommendations. At the end of the document are an author background description including a financial interest disclosure statement, and notes. 2. TELECOMMUNICATIONS IN THE INTERNET AGE: VERY HIGH STAKES 2.1 An Overview of the Revolution The events of the past three years, together with the rate at which the Internet continues to grow and new Internet applications continue to appear, clearly demonstrate that something very unusual is underway. The industrial, technological, and educational revolution driven by the Internet is probably rivaled only by the invention of movable type printing by Gutenberg or the Industrial Revolution of steam powered manufacturing in 18th century England. About 15% of U.S. households already use the Internet, versus essentially zero in 1993. Internet usage is growing about 10-15% per month, Web service usage is growing even faster, and private "Intranets" are rapidly becoming the standard platform for information sharing within organizations. What follows is a brief survey of the near-term characteristics of this revolution, of the role played by telecommunications within it, and of the resultant importance of a dynamic, open-architecture, competitive telecommunications industry. There is now a consensus within U.S. high technology that the Internet will evolve into the dominant platform not only for electronic mail, file transfer, and current Web services (its current primary uses), but also for many other information functions. For example, Internet-based systems are increasingly used for voicemail, fax services, electronic and multimedia publishing, videotelephony and videoconferencing, distributed and collaborative workflow systems within and between organizations, remote information processing services, industrial process control, electronic banking and payment services, catalog shopping, library services, home information systems, and even audio and video entertainment, ranging from music to multiplayer games. Eventually, the Internet will quite possibly absorb the two other largest telecommunications services, namely voice telephony and entertainment broadcasting, which currently use networks optimized for their special characteristics. These two services will over the next decade certainly come to be transmitted digitally, and will often use the same physical networks as Internet services. Telephony, radio broadcasting, and television (soon to be digital HDTV) probably will gradually become directly Internet based as the Internet and U.S. telecommunications infrastructure are improved to offer sufficient bandwidth and real-time quality of service (QOS) guarantees.8 There are already a considerable number of Internet radio stations, Web services that contain audio and video clips, and users of low-quality, but very inexpensive, Internet telephony. Thus as the Internet revolution is overlaid on existing communications systems, it will place increasing pressure on incumbent providers to digitize and modernize their networks. In fact, there is already under development a standard, VDSL, for transmitting enough bandwidth over local loop copper wire to provide live, full HDTV, which requires about 20 million bits per second. Efforts are also underway to equip the Internet with the capability to supply the quality of service guarantees needed for real-time services such as voice telephony, live videoconferencing, and live television. But all of this requires radically increased bandwidth and reduced cost. For example, current Internet service generally depends either upon analog modems running at 28.8 - 56 kilobits per second, or expensive digital services primarily used by businesses. But high quality web services, remote videoconferencing, CD-quality audio, and TV-quality video require 500 kilobits to 5 megabits per second. High quality multimedia services, advanced graphics applications, and live HDTV require 10-50 megabits per second. Thus for the full promise of the Internet to be realized, a 1,000-fold improvement in communications bandwidth is required over the next ten to twenty years. If telecommunications becomes the dynamic and competitive industry it should be, this will not be a problem. The technology will be available; but rates of technological progress and price/performance in the services actually delivered by the industry would need to accelerate dramatically, from their current very low rate (perhaps 5-10% per year) to perhaps 50-100% per year. Internet traffic, growing about 15% per month now, would probably grow even faster if telecommunications services improved at competitive rates. Within five years, at least half of all U.S. households and workers will be connected, and there is every reason to assume that Internet connections will eventually become at least as ubiquitous as telephones. With a properly dynamic industry, this growth could be accommodated at the same time that service quality and price-performance improve enormously. It is worth remembering that the entire computer industry has already improved its price / performance by a factor of about 10,000 since the early 1960s, and that its technical progress accelerated dramatically after the IBM-dominated mainframe industry was superseded by the open-systems, competitive industry of today. The equivalent transition will change the telecommunications industry dramatically. First, the rise of the Internet, if accompanied by sufficient bandwidth growth, offers the prospect of converting many current telecommunications services to modern, open digital technologies and/or to the Internet. Telecommunications services already enabled by digital technology include voice service itself, call waiting, call forwarding, voicemail, caller ID, call return, and fax. Internet telephony, currently rather a joke because of its poor quality, becomes quite acceptable over faster lines, a possibility which endangers both the LECs and the major long-distance companies. Internet voicemail already works quite well, because voicemail does not require true real-time data delivery, and is already being used by many web sites. Much of the traffic currently carried by the telephone companies' circuit switched, still partially analog networks can and eventually will be carried more effectively over modern packet-switched systems than over the current network. Such systems will involve the Internet, private Intranets using the basic Internet architecture, and/or hybrid systems involving various combinations of the Internet, and technologies such as xDSL, advanced routers, and frame relay and/or ATM switches.9 There are also already inefficiencies in LEC system use, because an increasing fraction of local and long distance traffic consists of fax and modem-based PC usage, including Internet access, for which analog voice lines are poorly suited. Furthermore, as we shall see in chapter 3, the LECs deliberately make it difficult for customers to use digital technology effectively, because this would reduce their profits and market power. Secondly, over the next quarter century the volume of Internet traffic - mostly for web services, email, and other non-telephony applications - will simply overwhelm traditional voice traffic. Internet service will thus gradually come to dominate telecommunications services, in both bandwidth and revenue. This is because Internet services can absorb huge amounts of bandwidth, and will deliver so many applications (including voice services) that users will tend to use Internet service for up to several hours per day. Conversely POTS requires a maximum of 64 kilobits of bandwidth per line, and people use their telephones on average half an hour per day, or perhaps 1-2 hours for heavy business users. Thus, if the average user a decade from now possessed 1.5 megabit Internet service and used it 2 hours per day, then Internet service would consume 25-50 times as much bandwidth as voice telephony. Within two or three years, it will be technologically feasible to deliver this much bandwidth to most U.S. homes and residences at costs not much higher than current local telephone service. Over time, therefore, digital services and Internet usage will drive the industry and traditional circuit-switched POTS, i.e. the core competence of the LECs, will become less important, even for voice service. In fact voice service in general (whether delivered via the Internet or as a separate service using the same local loop) will cease to be the principal service delivered by the telecommunications industry; rather it will become one comparatively inexpensive service delivered on a common high bandwidth line to every home and office. As we shall see, some of the LECs' resistance to progress and competition is related to this fact. In addition, overlaid on this situation is the embryonic revolution in digital video systems, including the newly approved "Grand Alliance" HDTV standard. The most innovative opportunities opened by this new standard do not derive from conventional broadcasting of digital HDTV signals, although that will probably be a large business. More interestingly, the use of this standard by emerging digital video hardware (cameras, editing systems, and players), software, and video server computers will revolutionize film creation and distribution. When sufficient Internet bandwidth is available, the HDTV standard will permit very inexpensive, decentralized, digital filmmaking and server-based distribution analogous to (and maybe using) Web services. Inexpensive digital still and video cameras are improving rapidly, and in a few years will permit almost anyone to make high quality images and films that could be broadcast or posted on the Internet. Delivering live and/or full HDTV in this manner would require bandwidth of 20-50 megabits per second. Using open architecture packet-switched networks, video content could be placed on and retrieved from any video server on the network. Such technology, which is already well under development, will not only reduce the cost barriers for making films and dramatically broaden choices in entertainment, but will also enable network collaboration in digital filmmaking, personal film archives, large-scale videoconferencing, documentary and educational applications, and telecommuting applications such as remote medical diagnosis requiring high-resolution imaging such as X-rays. Thus despite the popularity the Internet has already attained, its economic and social impact remains in its infancy. It will remake not only telecommunications itself but the entire information technology sector and, eventually, every activity related to information, which is to say most economic, educational, research, and governmental activities. Therefore, as with many technological revolutions, the rise of the Internet also entails a wave of creative destruction. Internet technologies enable new applications, but also threaten entrenched interests by substituting less expensive, higher quality Internet-based systems for traditional technologies in existing applications. This facilitates new competitive entry and places pressure on slow-moving industries to keep up with the Internet's pace. Many are doing better than telecommunications, which is more the exception than the rule. Thus the gap between the conventional telephone industry's practices and America's future needs is growing ever larger. The Ferment in the Competitive Industry Improvement in the technology and equipment needed to modernize the LEC and/or CATV networks has not been, and will not be, the bottleneck to progress. The computer, networking, and telecommunications equipment industries, led by firms such as Cisco, Fore, Cascade, Pairgain, Amati, Ascend, ADC, 3Com, Cabletron, Bay Networks, Lucent, IBM, Northern Telecom, Alcatel, Motorola, Rockwell, and U.S. Robotics are moving rapidly to provide better equipment every year, including routers, switches, modems, ADSL and HDSL terminal equipment, and products that allow TCP/IP, the Internet's protocol, to interoperate with frame relay and ATM systems.10 These and other firms are also working on related technologies such as network security systems, DSL multiplexers (so called DSLAMs), network management systems, and so forth. In addition, innumerable startups are active in these areas. The achievements of this industry will be discussed further in chapter 3, in conjunction with assessing the LECs' technological performance. Many other companies, mostly newly formed startups, are developing and offering a wide array of Internet based services. These include electronic payment systems, Internet telephony, fax and image transfer, voicemail management, collaborative groupware and messaging systems, transaction processing services, multiplayer game systems, search and retrieval systems, audio broadcasting, and multimedia document transfer services, in addition to innumerable "conventional" web sites. Often these vendors and their customers offer products and services far beyond those developed by the LECs. For example, NetCentric and other startups are delivering fax services over the Internet; dozens of companies offer local, U.S., and global Internet directory services, classified advertising, and Internet Yellow Pages services. Many companies, including online bookstores and newspapers, offer Internet based consumer transaction services such as purchasing and bill paying; as we shall see, the LECs do not. (In fact, most LECs don't even accept credit cards.) In addition, many univiersities, nonprofit organizations, and government agencies have developed highly innovative Internet-based systems including web sites. Hundreds of thousands of Internet companies, products, and services have appeared since this Internet revolution began in 1993-4.11 For this transformation to continue, however, requires a modern telecommunications services industry that delivers high rates of technological progress and innovation. Simply put, the industry should be run on the "Silicon Valley Model." Its technologies, companies, and industry structure should be open, modular, decentralized, and fast-moving, united by open industry standards. Its rate of technical progress, structural change, and growth should be similar to, and should probably be even higher than the computer industry's, for two reasons. First, the (unregulated, competitive) networking equipment and data communications industry now has the fastest rates of growth and technological change in all of U.S. information technology - even faster than the PC and software industries, for example. And second, as it modernizes, the telecommunications system should make up for its last decade of slow technical progress under a regulated monopoly structure. With this background, consider the economic stakes associated with future Internet services, and with developing the modern, competitive telecommunications industry required to provide them. I will analyze three successively broader domains: the U.S. telecommunications industry itself; the U.S. information technology sector, and the U.S. economy as a whole. 2.2 Economic Effects of Telecommunications Modernization Upon the Telecommunications Industry Itself Even if one considers solely the telecommunications industry itself, leaving aside all effects upon other industries, the difference between a "legacy mainframe" outcome and a "Silicon Valley" outcome for the U.S. telecommunications industry is of major importance for the U.S. economy. Provision of public telecommunications services, not counting internal corporate or other nonpublic (e.g. defense) systems, is already one of the largest industries in the world. The combined revenues of U.S. local exchange carriers are now about $120 billion; the long distance industry (which relies very heavily on the local network) adds $50 billion or more; related industries supplying network equipment, services, and CPE (switches, control systems, optical fiber, PBXs, telephones, modems, fax machines, answering machines) add perhaps another $25-50 billion. Thus even by a very narrow definition, e.g. not counting subcontractors and resellers, the telecommunications industry generates over $200 billion per year, or 3% of U.S. GNP. U.S. telecommunications carriers (the RBOCs, GTE, long distance carriers) employ nearly 1 million people directly, many more indirectly, and pay high wages. AT&T and the RBOC stocks are the most widely held common stocks in the U.S. If the broadcast, wireless, and cable TV industries were included, the industry's size and impact would be even higher. Moreover, the industry's impact upon U.S. productivity is larger than others of comparable size. Its functions are essential and nearly all of them take place within the United States; you can't import local services. For the same reason, the U.S. could not rely on import substitution to discipline inefficiency, but rather must reform the domestic industry. Consequently U.S. economic activity depends heavily upon a large monopolistic industry, demand for whose services is extremely elastic with respect to both price and technological quality. For these reasons, even a modest (by high technology standards) acceleration in the industry's technological progress and/or revenue growth would have significant economic effects. For example, if the difference between a monopoly industry and a highly competitive one caused technological change to accelerate by only 7% per year, say from 10% to 17%, the industry's real technology level and productivity ten years from now would be double the level generated by the slower rate. This change alone would raise the productivity of the entire U.S. economy by 1.5%, i.e. would contribute an incremental 10% increase in U.S. productivity growth at its recent trend rate. In fact, the economic impact of switching to a dynamic, efficient industry would probably be much larger, because the difference between current and potential performance is probably 20% per year or more. Thus, putting telecommunications on the same technology curve as every other digital systems industry, which is where it largely belongs, could potentially yield productivity and/or GNP gains of 2.5-5% over the ten to twenty years. If this seems farfetched, consider the computer industry over the last decade. Modern, open systems computer technology (personal computers, servers, etc.) improves roughly 40-50% per year in price/performance, about three times as fast as the IBM 360-derived mainframe technology which still dominated the computer industry only a decade ago. In addition, newer computer technologies also allow for much more efficient use of computer professionals (of whom there are over 1 million in the U.S.), not to mention non-computer professionals who comprise the vast majority of PC users. Once Again, the Computer Industry Analogy IBM's share of the U.S. computer systems industry has been halved in the past decade, and IBM's own products and revenues have been forced by competition to shift towards modern, microprocessor-based PCs and servers. The industry has been driven by the spectacular innovation and growth of companies such as Compaq, Dell, Gateway, Intel, Apple, 3Com, Cisco, Microsoft, Sun, Seagate, EMC, Oracle, EDS, U.S. Robotics, Hewlett Packard, and many others; these firms now account for over $150 billion per year in annual revenues. The Internet revolution has displayed even faster innovation and growth; Netscape, an Internet software company founded in 1994, is already a major institution, with sales doubling every quarter and projected revenues of over half a billion dollars in 1997. (Internet service providers such as UUNet, PSI, America Online, the Microsoft Network, and Netcom have also grown very rapidly.) Under competitive pressures such as these, even IBM and the other "legacy system" incumbents (DEC, Amdahl, Data General, Unisys, NCR) have shifted towards open systems and greatly improved the price/performance ratios of their remaining mainframe and minicomputer products. Thus over the past decade, a $100 billion closed mainframe industry with roughly 15%/year technological change and 10% revenue growth has been transformed into a $250 billion open systems industry that delivers roughly 40%/year technical change and 20-30% per year revenue growth. If computing had remained dominated by IBM and proprietary mainframe systems, the U.S. (and global) economy would have suffered considerably. Indeed it did suffer substantial productivity losses during the decade required to replace the incumbent industry, a transition which is still underway. But because competition, new entry, and innovation were permitted to discipline IBM and reshape the industry, the competitive information technology sector has renewed itself and emerged as the engine of U.S. economic growth. Now, it is the telecommunications industry's turn. General information technology trends have applied to telecommunications for at least five years, and probably longer. However, rates of technological change, price / performance improvement, and revenue growth in local telecommunications services lag far behind both their potential and the performance of all other information technology sectors. This gap seems to be increasing. For example, the number of T1 (1.5 megabit) lines is growing only 40% per year,12 versus general Internet growth of 200-300% per year, in large part because T1 prices have remained astoundingly high, declining only slowly despite rapid technological progress in T1 technology and equipment. As will become clear in chapter 3 below, it is very difficult - in fact, virtually impossible - to see how this situation derives from anything other than the regulated monopoly structure and inefficiency of the industry, which prevents the competitive renewal and technical progress witnessed in computer systems, semiconductors, computer networking, and software. When this observation is made, the LECs and/or their defenders tend to offer some combination of four defenses. The first is that regulators impose financial burdens which make modernization difficult. There is some limited truth to this argument, but much less than previously; in most jurisdictions, price cap regulation replaced rate-base regulation in the early 1990s and in 1994 regulatory accounting was abandoned and depreciation rates substantially liberalized.13 The second LEC argument is that the regulatory mandate to provide universal, highly reliable residential voice service imposes costs and problems not seen in the unsubsidized, primarily business markets of other information technology sectors. There is once again some limited truth in this, but much less than the LECs claim; the reliability levels of many services subject to competition are very high, and most LEC capital equipment comes from normal commercial sources. Their third defense is that voice service generally is a mature technology not susceptible to the same rate of technical progress as computer networking and other IT sectors. And finally, the LECs argue that because local service requires running long copper wire loops to every residence and business, their cost structures are fundamentally different from those of high technology companies. In fact, all of these arguments turn out to be largely erroneous. For example, the cost of laying copper wire for local loops is about 40% of the LECs' telephone network capital investment, which in turn is less than 20% of their revenues.14 Even if the entire cost of the copper local loop was in excess of and different from the cost structures of other industries, which is false, this would represent only about 7% of the LECs' total cost structure. The overwhelming majority of the LECs' cost structure and technology base is highly similar to high technology vendors and large computer networks. Similarly, it would be quite odd if their service quality requirements reduced technical progress, particularly because the LECs are extremely heavy purchasers of normal commercial computers, upon which their mission-critical services are all based. And as many of us know from personal experience, LEC service quality is not perfect. Furthermore, there is clear evidence that even the provision of voice service is in fact highly susceptible to technological improvement, through the use of some of the same digital communications technologies needed to improve high speed data and Internet services. In fact even under the regulated regime, the LECs sometimes use modern digital technologies to lower their own cost of providing POTS, but without passing these savings on to users. Nor do the LECs usually permit users of these technologies to configure their own systems (as would often be technically optimal). This issue is discussed detail in chapter 3. Thus, while for various reasons - the local loop, subsidized residential service, costs of regulation - the LECs' absolute costs might well be higher than those of unregulated technology sectors, there is every reason to think that their rate of technological progress should be just as high. It also turns out that very similar arguments to those now advanced by the LECs were previously made by the mainframe computer industry and by the old AT&T before divestiture. In sum, the rate of technological progress in local telecommunications should probably be at or near high technology levels, and can be dramatically accelerated. The experience of the computer industry suggests that even the direct effects of telecommunications modernization, i.e. those generated only by the industry itself, would be of major economic importance. Furthermore, the telecommunications industry itself is globalizing; following the WTO agreement, national markets will in principle be deregulated starting in 1998. The competitiveness of the U.S. industry in global markets will depend heavily upon its technological performance. Worldwide, telecommunications services is roughly a $600 billion industry with high growth rates. The LECs have thusfar had rather mixed success internationally, and face significant competition from foreign firms. Hence, the supposition that direct benefits from U.S. telecommunications reform could be the order of 5% of GNP is not unreasonable. Clearly, however, the largest effects of improved telecommunications and Internet services will be their catalytic effects upon the rest of the private economy, as well upon nonbusiness activities (education, research, defense). The economic evidence suggests that the payoff to improved communications services, even conventional ones, is very high. It is widely recognized that the United States benefitted greatly, throughout the 20th century, from having the best and most universal telephone service in the world. Extremely high levels of telecommunications investment (for both voice and data applications) is characteristic of all rapidly growing economies, e.g. China and Poland, as well of rapidly growing industries in advanced economies. With the rise of the Internet, ubiquitous personal computing, and inexpensive computer servers, the economic importance of telecommunications is increasing further. In fact, the United States has a unique requirement in this regard because it is the location of one-third to one-half of the world's information technology production and consumption. And the continued growth of U.S. high technology industries depend heavily upon continuous, rapid improvement in telecommunications bandwidth and price-performance and Internet services. 2.3 Economic Effects of Telecommunications Services on U.S. High Technology Industry Technological progress in telecommunications and Internet services is becoming critical to the growth and global competitiveness of U.S. information technology sectors such as the semiconductor, personal computer, and software industries. This is true for three reasons. First, Internet services are becoming essential to their operational efficiency and competitive advantage. These industries are so dependent upon speed and responsiveness that high speed electronic communications is already considered a basic necessity. Second, Internet performance increasingly drives - and limits - demand for these industries' products. And third, the telecommunications industry itself is, and/or should be, an extremely large market for advanced information technology systems; if this industry stagnates, its supplier industries will suffer. Any variable significantly affecting the productivity and growth of U.S. information technology would be macroeconomically significant. In total, these industries now generate about 10% of U.S. GNP. They also account for at least one third, and perhaps over half, of all U.S. economic growth. The major sectors of U.S. information technology exhibit technological progress and growth rates of 25%-75% per year, with networking equipment at the high end of this range. These industries are also heavily export dependent and their competitive advantage in global competition derives primarily from technological superiority, which in turn derives in part from their own access to information technology, including networking and Internet service, and to leading edge domestic markets. For the U.S. computer, networking, and telecommunications equipment industries to remain globally competitive, they must have a growing, technologically advanced domestic customer base. These industries' growth is now becoming heavily and directly dependent upon the Internet and even more specifically upon local telecommunications bandwidth. As personal computing spreads from business to small offices / home offices ("SOHOs"), residences, and schools, the bandwidth and price-performance of local loop services are becoming the critical bottleneck for all information technology industries, and soon for all information-intensive activities. The fundamental reason for this is that in information technology, total system price / performance is generally limited by the slowest and/or costliest system component, which, in Internet applications, is now the local loop. There is little reason to purchase an extremely fast computer if the communications channel upon which it depends is too slow. This issue will soon be the largest problem facing the U.S. personal computer industry. As of early 1997 approximately 40% of all U.S. households had personal computers, and about 15% of all households used the Internet. Both percentages are up from effectively zero only four years ago, and both are still increasing rapidly. (PC and Internet usage are also increasing very rapidly in SOHOs and schools.) These percentages will soon grow to become the majority of the population, and demands for new services and higher speeds will accelerate. In fact, growth of home and/or noncorporate usage is already the largest source of growth for the U.S. personal computer industry. Home PCs are now estimated to account for roughly one-third of U.S. personal computer demand, and U.S. consumers now spend more money on personal computers and software than on televisions. In addition to personal computer systems, a wide array of other Internet-related computer peripherals and digital systems are also gaining in power and entering SOHO and home use. These include fax machines, high resolution color printers, scanners, multifunction peripherals, microphones, digital still and video cameras, CD and DVD systems, speakers, digital tape recorders, game systems, and even digital filmmaking and editing equipment. Other potentially huge markets, such as videotelephony, customized entertainment, and home computer networks, are still embryonic but expected to grow as technology improves. Virtually all of the vendors of these products and services agree that the Internet will be critical to the growth of their markets. Many companies in these sectors have designed and in some cases are already selling Internet enabled products and services. It is also clear, however, that future growth in these areas depends upon faster and less expensive communications bandwidth to residences, SOHOs, and even to large businesses. This is even true of pure video entertainment. As noted above, the new Grand Alliance HDTV standard will allow development of new systems that unite television, video services, and computer functions. But over-the-air HDTV broadcasting permits only one-way mass transmission of programs. Richer services including open video servers, interactive entertainment, and videoconferencing will become available only when local telecommunications bandwidth improves enough to permit fast, interactive, open architecture services. The importance of local telecommunications bandwidth and Internet services to the information technology sector can be judged by several indicators. For business services, one indicator is the stunning growth rate of the computer networking industry; another is the fact that LEC T1 lines and revenues are growing 35-50% per year despite extremely high prices and little improvement in price/peformance (see chapter 3). There can be no doubt whatsoever that if T1 price / performance improved at a competitive rate, say 50% per year, growth rates and usage would be much higher. The networking equipment industry is growing over 70% per year, and the price/performance of its technologies and products improves at a similar rate. The industry's leading firm, Cisco, has been growing about 90% per year for several years even though it is already a $4 billion company. The total bandwidth handled by the Internet backbone is approximately tripling every year. Similarly, in the home and SOHO markets, the importance of improved bandwidth is demonstrated in the growth of the modem industry and in second telephone lines for residences. Home PC and Internet demand is such that the U.S. modem industry, led by U.S. Robotics and Rockwell, is growing at least 50% per year. Modem speeds have increased from about 9.6 kbps five years ago to 56 kbps by mid-1997. U.S. Robotics, the leading firm, is now a $2 billion company whose revenues are nearly doubling every year.15 Furthermore, it is clear from the purchasing patterns of both the home / SOHO and corporate markets that improved bandwidth is extremely important. Product life cycles in the modem industry are rarely longer than one year, and products rapidly decline in price when a newer, faster generation appears. Modems generally become effectively unsaleable less than 3 years after their introduction. Unfortunately, however, analog modem technology is reaching its technological limits. For speeds starting at ISDN levels (128 kbps and above), digital lines and terminal equipment must be used. Delivery and use of these faster services depends upon infrastructure improvements and/or policy changes which can only be made by local telecommunications services providers - the LEC monopolies. The problem is not technological. Technologies such as T1, ISDN, SDSL and HDSL, already in commercial use, permit up to 1.5 megabits per second to be transmitted over copper wire. But these technologies remain either unavailable from the LECs or far too expensive. The computer and modem industries are approaching this brick wall quite fast. For many business users and for some individual professionals, such as graphic designers and computer programmers working from home or small offices, this is already a serious problem. For the rest of us, it's coming soon. The combined effects of continued progress in PC technology, home PC penetration rates, and the limits of analog modem technology are such that local telecommunications speeds will become a major bottleneck to the PC industry, and to total U.S. computer and Internet usage, beginning about 3 years from now. By then, the majority of U.S. residences and SOHOs will have Internet service, fast PCs, and the fastest modems permitted by traditional telephone lines. At that point, control over bandwidth will have shifted completely from the modem industry to the LECs. This is particularly ironic because the direction of the technology is to reduce the importance of centralized switching, and to place more intelligence in terminal equipment and distributed routers or packet switches, as the Internet does. But unless the telecommunications industry changes its behavior dramatically, the trend line in communications bandwidth price-performance will drag the computer industry down with it. It is impossible to estimate precisely how large this drag is now, or would become. But it is clearly large. The T1 market will represent nearly $10 billion in LEC revenues and nearly 2 million lines, a high fraction of them used for business Internet access, by the end of 1997. Suppose prices are three times as high as they should be (as it seems they are: see chapter 3). Depending on the exact price elasticity of demand, which is clearly high, this problem alone could represent a direct economic loss of $5 billion per year or more. The high cost, complexity, and limited availability of ISDN probably causes economic losses of similar size. These problems also represent a drag on the development of the independent Internet industry, the network equipment industry, and on the computer sector, because of reduced demand for their products. Suppose that beginning in another year or two, this drag effect reduces the average annual growth rate of the U.S. technology sector by 5% per year (e.g. to 20% per year versus 25% per year) relative to growth levels that could be achieved in the presence of rapidly improving Internet services and a progressive telecommunications sector. If this effect lasted a decade, it would cause a 5% loss to U.S. annual GNP, because the technology sector would only be two thirds as large as its potential. This effect would be in addition to the direct losses caused by the telecommunications sector's own inefficiency and foregone growth over roughly the same period. The two combined effects could represent over 10% of foregone GNP, not a trivial matter. However, the largest effects of the Internet revolution will almost certainly be those generated in the rest of the economy. 2.4 Economywide Effects of the Internet and Telecommunications Modernization The economywide importance of Internet development is by itself more than enough reason for Federal policy to ensure that the U.S. possesses a modern, technically progressive, competitive, open systems local communications infrastructure. It should be noted, however, that there is also another reason for such policy - i.e., that the cost, price, and quality of traditional voice services would probably also improve greatly. In fact, policy reforms driven by the demands of the Internet may well be the best way to reduce the average cost of POTS. The following analysis, however, concentrates on the Internet revolution. A wide array of Internet-based activities and services are beginning to change work processes throughout the entire economy - within and between organizations, between households and firms, and between households and government. External Web sites continue to grow nearly 20% per month, and number an estimated half million as of early 1997. Internal Web-based applications are being used to manage workflow, discussion forums, broadcast announcements, customer support services, and investor relations. To support telecommuting, remote access from branch offices, vendor relationships, and independent home offices, Internet based systems are being developed and marketed for many applications including electronic commerce, videoconferencing, PC application sharing, electronic whiteboard systems, remote processing services (such as running computer programs, preparation of presentation graphics, and printing jobs from Internet-submitted files), and remote database access. Over the next decade or two, these Internet based systems will fundamentally reshape business activity and competition, both in high technology and in other, information-intensive sectors such as financial services, publishing, broadcasting, retailing and distribution, mail and delivery services, the professions, education, basic research, health care, and even manufacturing. And although the United States currently enjoys a large lead in Internet-related technologies,`the long term ability of the rest of the U.S. economy to apply these technologies, and to compete globally, will depend critically upon rapid improvements in U.S. telecommunications infrastructure and services. As information technology in general and Internet technology in particular becomes a higher fraction of U.S. capital investment and an increasingly important determinant of competitive advantage, the structures and strategies of user industries are increasingly shaped by its dynamics. In major industries as diverse as financial services, military systems, publishing, and telecommunications services, information technology now accounts for the majority of total capital investment and is an increasingly critical driver of strategic decisions. As a result these industries' structures, strategies, operations, and products are changing faster, and are increasingly taking on the modular, open systems designs characteristic of modern information technology. With the rise of the Internet these effects will become even more intense. It is impossible to estimate precisely how large these effects will be, how rapidly they will arrive, or to what degree they are dependent upon technical progress in local telecommunications. They also depend upon policy measures in other areas, for example basic research, encryption policy, and the provision of technology, bandwidth, and training to the educational system, research institutions, and libraries. It is very clear, however, that these effects are potentially enormous, comparable to the gains from the introduction of assembly line manufacturing early in this century, and that they are highly dependent upon a huge improvement in the structure, conduct, performance, and technological clock rate of the local telecommunications services industry. On reason for believing this is that Internet-based services will force major productivity improvements in a number of other major industries and economic functions which have stagnated for a long time. Internet-based startups, and/or established firms that lead their industries in use of Internet technology, will force productivity improvements industrywide - just as personal computers forced IBM to improve the price/performance of all of its computers, even mainframes. Industries and functions subject to such change include mail and package delivery, the sales and marketing functions in many industries, wholesale distribution, retail and catalog sales of many products and services, print publishing, advertising, retail financial services, and customer service functions such as billing, problem-handling, and payments processing. Early evidence suggests that in some large industries, these effects will be extremely significant. I will mention four examples simply to illustrate the point: retail bookselling, retail financial services, electronic software distribution, and consumer billing/payments processing. 1. Amazon (www.amazon.com) is a Seattle-based company that runs a Web-based bookselling service. It has a fully searchable online catalog of 2.5 million (primarily English-language) books, many of them heavily discounted from retail price, and virtually none of which Amazon needs to stock itself. There are also online reviews plus lists of books reviewed by publications such as the New York Times Book Review. Customers order over the Internet by credit card, receive billing and fulfillment information automatically by email, and can request automatic email notification of new publications by subject or author. Customers can also select gift wrapping styles, type personalized messages which will be printed on the delivery envelope, and ask for overnight delivery at extra cost. Amazon's revenues are growing 30% per month, and its growth has placed major competitive pressure on the $4 billion U.S. book retailing industry; Amazon also plans to expand internationally. The two largest U.S. bookstore chains (Borders and Barnes & Noble) have now been forced to announce web-based services of their own. 2. Although banks and other financial services firms have been steadily automating, many retail financial services are still very expensive, paper-dependent, and slow. In some areas, such as stock trading, discount brokers and no-load mutual funds have greatly improved costs and productivity; but even in these firms, Internet based systems apparently represent a major advance. For example Charles Schwab, the largest U.S. discount broker, announced in early 1997 that of its approximately $270 billion in assets under management, over $50 billion already derived from Internet-based services offered for the first time less than one year earlier. Online trades now account for about 30% of Schwab's total trades.16 Other major firms such as Fidelity Investments also report extensive use of their Internet services, and several rapidly growing startups such as E*Trade (which already has nearly $4 billion in assets under management) specialize solely in Internet-based transactions.17 3. Netscape was the first company to offer Internet-based distribution of its product. Internet based (and/or internal "Intranet" based distribution still represents a very small fraction of all software sales, but it is increasing rapidly. It is also very important; the software industry is large, its products change frequently, and inefficient physical distribution typically represents more than one-half of the price paid by retail customers. Like much of electronic publishing, with which it shares many characteristics, electronic software distribution is also extremely sensitive to Internet bandwidth because software products are essentially very large files. Residential and SOHO users with 28.8 or even 56 kilobit analog modems must now literally wait hours to download even relatively small programs such as the Netscape browser. Internet based distribution of large software systems, and electronic publishing generally, will become feasible only when high bandwidth Internet access is inexpensive and widely available. 4. Billing, account information, and payments generate several billion pieces of mail per year in the United States, much of which is processed manually. Most of this represents billions of dollars of pure cost to both business and consumers. Internet-based billing, account information, and payment systems could and surely will substantially reduce these costs while improving service, e.g. through faster, more accurate processing and by permitting customers to review and pay bills even while traveling. Given enough Internet bandwidth, it should also be possible to post voicemail questions, service requests, and announcements with emailed bills, statements, and payments, offering substantial productivity improvements in these functions. There are many more examples of a similar sort, in which Internet-based systems permit and/or force industrywide productivity improvements, and many more will arise in coming decades. Industries and internal operational functions in which use of Internet technology could generate major productivity improvements could easily represent over half of U.S. GNP. The quality of local service will be a major factor determining the rate of growth of these new products and services, and likewise the growth and productivity of the firms offering them. And in many information-driven industries, global markets tend to be dominated by the first firms or national industry to develop a large new market and the standards associated with it. For the United States to remain the leading center of information based activity, and for these new U.S. industries to contribute fully to U.S. economic growth, requires continuous modernization of the infrastructure systems upon which these entrepreneurs depend. Thus, rather paradoxically, the quality of local and domestic communications infrastructure is becoming more important to the geographical location of high technology activities and to competitive success in increasingly global markets. Finally, intangibles can be important too. In situations such as the Internet revolution one often observes a virtuous (or vicious) circle phenomenon, in which the performance of users reacts back upon the supplier industries, to the benefit of both. Users' experience with leading edge technology enables them to develop better and leading edge applications. This stimulates their domestic competitors to improve, and also gives them ideas about how to push the technology further, which incents the development of more leading edge technology. The result is that both suppliers and users become more dynamic and competitive in world markets, and more valuable to consumers. Conversely stagnation in services breeds stagnation in usage; indeed this was a problem seen in some U.S. industries during the era of mainframe-dominated computing. The difference between dynamism and stagnation in local telecommunications and in Internet services could well be extremely large, even if the productivity acceleration is only temporary. Suppose the total effect of creating a fully competitive, dynamic local telecommunications industry (through its combined effects upon the industry itself, the IT sector, and the rest of the economy) causes a 2% annual increase in U.S. GNP and/or productivity growth for just one decade beginning in the year 2000. This would be sufficient to raise U.S. productivity, GNP, and living standards by about 25%. A more gradual transition, say a 1-1.5% annual improvement lasting two decades, brings a roughly similar result. The experience of other industrial revolutions suggests that these are not unreasonable estimates, and might even be quite on the low side. As the United States continues to shift to being an information-driven, high technology society in an increasingly global economy, the quality of its communications infrastructure will in fact play a major role in its growth. Conversely the negative consequences of delaying the transformation of local telecommunications could be quite large, without even counting adjustment costs and effects upon U.S. international competitiveness. In fact, the last decade of restructuring and downsizing suggests that these adjustment costs can be quite large, and worsen when productivity improvements are postponed. Altogether, then, it would seem that the Internet revolution is worth doing, and doing well, and that major impediments to it should be removed. We will now consider how well the LECs and CATV industry are performing. 3. THE CURRENT LOCAL SYSTEM: ASSESSMENT AND ANALYSIS 3.1 Introduction This chapter assesses LEC and CATV industry performance using a variety of operational, technological, financial, strategic, and managerial indicators. Many benchmarks are drawn from high technology sectors and/or users - e.g. computer systems, networking equipment, long distance communications, digital consumer electronics, and software sectors, AT&T, systems integration vendors, the Internet services industry, and large, IT-intensive services firms (such as FedEx, financial services firms, etc.). I also consider the price-performance of LEC services, LEC technology practices, and the ways in which the LECs do, or do not, make their technology and cost improvements available to users. The picture that emerges is of a group of powerful but slow-moving firms endeavoring to perpetuate their monopoly power (e.g. via lobbying, mergers with each other, litigation, and cooperative behavior), their current industry environment, and their incumbent management. The LECs have a poor record in innovation, R&D, the standardization and deployment of new technologies, investment in network modernization, delivery of price-performance improvements to customers, customer service, management of open systems architectures, success in real competition, and even in internal use of their own technologies and services. Perhaps most seriously, the price/performance of LEC services, including both digitally implemented voice services and data services such as ISDN and T1, has improved little and sometimes even deteriorated over the past decade. This is an astonishing situation given that the most of the underlying technologies for these services improve 40%-60% per year. This behavior is unique among U.S. information technology industries, and very difficult to comprehend as anything other than a result of some combination of monopoly power and extreme inefficiency. Similarly, despite many LEC announcements of plans to invest in advanced digital network infrastructures, the LECs are actually reducing their investment in the U.S. local network. LECs are moving away from U.S. local services and diversifying into others (cable, entertainment content, cellular, PCS, foreign markets), primarily via acquisition rather than internal development. The cash for these investments has been made available by price cap regulation combined with declining workforces, high prices, and reduced investment in the LECs' core telephone and data networks. Consequently over the past five years, as LEC profits have grown, the Internet industry emerged, and the local network emerged as the bottleneck in U.S. information infrastructure, both R&D spending and capital investment per telephone line have actually been flat or declining in the United States. In fact, capital investment as a percentage of data revenues and traffic has declined sharply. The capital stock of the U.S. telecommunications industry has been flat since the late 1980s, after having grown steadily for the previous 20 years.18 Inflation-adjusted LEC capital investment peaked in 1985 at $20 billion (in 1987 dollars), declining to $16.8 billion by 1994; during this period, the LECs' capital investment declined from 70% to 60% of total telecommunications investment, because the long distance industry invested heavily due to increasing competition.19 This situation, too, is astonishing, and once again sets the LECs apart from all other information technology sectors. The LECs' poor technological and operational performance seems to be accompanied by strategic, financial, and political conduct intended to preserve monopoly power rather than improve technological performance. It is a highly systematic, consistent pattern including R&D strategy, capital investment, diversification, pricing, new market entry, mergers and acquisitions, alliances, dealings with ISPs, reactions to the 1996 Act, and interconnection behavior with competitors, as well as highly cooperative strategic behavior towards each other, spending on academic expert witnesses and policy research, litigation, regulatory strategy, PAC contributions, and lobbying. In all of these domains the LECs' behavior is almost the precise opposite of that exhibited by successful, dynamic high technology firms, and of what optimal Internet and information infrastructure growth seem to demand. Thus while the LECs and their regulators deserve credit for their past achievements in providing high quality universal voice services, their conduct over the past decade, and particularly since the advent of the Internet revolution, is not impressive. The LECs' skills and preferences seem to lie in running closed, slow-moving monopoly systems and in strategic, regulatory, legal, and political behavior intended to forestall competition. The CATV industry's situation is in many respects similar, and probably no better. There are fundamental reasons to doubt that the CATV industry is either desirous or capable of providing advanced, open Internet services and/or effective competition for the LECs on a large-scale basis within the next five years. Finally, the LECs show signs of having serious corporate governance problems, and to have highly entrenched top managements. There is little evidence that LEC boards of directors have pressured management to accelerate technical progress or internal reform, and examination of LEC boards does not suggest that they are incented, disposed, or well suited to do so. This suggests that LEC executives and boards will continue to resist competition and the delivery of rapid price / performance improvements to consumers. The LECs seem inclined and incented to do this both as a matter of rational strategy given their industry's circumstances, i.e. its continued monopoly power, and also because it is in the personal interest of LEC executives. The top managements of the LECs are generally lifetime Bell system employees, with little or no technical training and/or experience in modern information technology. They consequently appear to be poorly equipped to manage highly competitive, fast moving environments exhibiting high rates of technical progress. The overall pattern of LEC conduct seems generally similar to that of other monopolies or oligopolies in long term decline, whether or not regulated. While regulatory policy certainly has shaped and often distorted the LECs' behavior, it does not seem to be the primary cause of poor LEC conduct, performance or governance - except, of course, in that the regulators have allowed the LECs to get away with it. The principal cause of LEC poor technological performance and resistance to change is simply their monopoly status. This in turn suggests that major changes in policy, industry structure, and conduct will be required if the U.S. system is to exhibit the high rate of price / performance improvement, service quality, and innovation required for development of an information economy. The LECs' Activities and Environment Since Divestiture I assess the LECs' performance in part by comparing them to other industries. While it is not entirely appropriate to compare the LECs to unregulated firms, many LEC activities are now largely free of regulation, and/or are reasonably similar to those found in the competitive IT sector or in technology intensive services industries. LEC activities with counterparts in competitive sectors include data services, Internet access service, customer service functions such as sales and support, installation, repairs, billing, payments processing, and information services on and off the Web (yellow pages, white pages, customer and investor relations services, etc.). The RBOCs have been free to enter most businesses outside their regions, both domestically and internationally, since 1984. They and GTE have also operated under price cap regulation for about five years. While price cap arrangements vary by jurisdiction, most give the LECs considerable freedom to price services, and allow them to retain a substantial fraction of their profit growth. They have been given this freedom largely at their own request, and/or on the theory that it frees them from onerous regulatory obligations and incents them to operate efficiently. They can also enter most Internet-related businesses, and can of course freely use Internet technology, in their own operations. Furthermore, in 1994 regulatory policies towards LEC depreciation rates were liberalized, the LECs discontinued regulatory accounting, and actual depreciation rates were changed, resulting in much higher rates (i.e. shorter accounting lives) for most categories of plant and equipment. With the signature of the 1996 Act, the LECs gained further freedoms, including the ability to enter cellular long distance and CATV, even in their own jurisdictions, and either by new investment or acquisition. The LECs also have the right to enter general long distance markets once their local markets are certified to be competitive and/or they pass a checklist of items certifying that they have opened themselves to competition. And, last but definitely not least, since signature of the 1996 Act the LECs have had the right to enter local telephone markets outside of their own areas, i.e. to compete with other LECs. Thus, although no single indicator is definitive, the LECs by now have sufficient resemblances to nonregulated firms to render comparisons to them at least interesting, if not by themselves conclusive. We can assess their customer service, adoption of widely used managerial and technical practices, and their broad financial performance, e.g. trends in revenues, earnings, market capitalization, and conventional productivity indicators such as revenue or value added per employee. We can also to some degree assess the LECs' R&D and innovation record by looking at the recent history of innovation in communications and the R&D spending of innovative and/or successful firms. 3.2 LEC Financial and Operational Performance General Financial Performance The LECs' market capitalizations and total returns have roughly tracked the S&P 500 since the RBOCs became independent firms following the breakup of the AT&T monopoly in 1984. The LECs have generally displayed considerable pride in this fact. (See for example the annual reports of SWBT, BellSouth, and NYNEX; GTE has also made similar statements). However, this performance is less impressive than it might seem. First, the LECs' profit growth derives not from technological or competitive dynamism but on the contrary from three other sources. Most important by far is their partial price deregulation (via price cap regulation) in combination with their continued monopoly status in providing both basic voice and data services. In general, these price cap arrangements incent the RBOCs to reduce costs and/or increase prices, but not to improve service quality, invest in R&D, or rapidly deploy new technology. Monopoly LEC businesses have particularly benefitted from growth in the Internet and corporate networking, because LEC data revenues are increasing rapidly while prices have decreased little if at all. Secondarily, and rather interestingly, they have largely avoided competing either with cable companies or with each other. In fact they cooperate with and assist each other in many domains, as we shall see shortly. Third, the LECs have made large investments and acquisitions in other heavily regulated, concentrated markets - basic telephone service in developing economies, regulated cellular services, paging, cable TV, and PCS. It is actually not clear that the LECs have fared well in these investments on a net basis, however, and they generally do not disclose their results. Furthermore, LEC financial performance is poor compared to similar nonregulated companies. Since becoming independent in 1984, the RBOCs' total market capitalization has slightly underperformed the S&P 500; GTE's performance has been simiar. RBOCs generally pay much higher dividends than the S&P 500 average of 2-3%, but their revenue and earnings per share have grown more slowly than S&P averages.20 But the RBOCs and GTE are not the S&P 500. They are not in the automobile, oil, gas, steel, glass, mining, concrete, clothing, retailing, food, or electric utility businesses. They are in the information technology sector, which is the fastest growing large industry in economic history; and in fact they are now in the fastest growing portion of it - communications. Although a majority of their revenues derive from local services, even these services are growing; and much of their revenues derive from cellular, voicemail, data, Internet access, and foreign markets that are experiencing extremely rapid growth. Moreover, although they are indeed regulated, over the past five years their pricing and behavioral freedom has increased much more than any competitive pressures upon their core businesses. In this light their performance is not impressive, and it becomes even less so in comparison to information technology firms. For example, the LECs' total returns since 1990 have been extremely poor relative to those of four of the largest firms in information technology - Intel, Microsoft, Compaq, and Hewlett-Packard. (The total returns of networking companies such as Cisco, Cabletron, Fore, Ascend, Bay Networks, Cascade, and 3Com have been even higher, but are sometimes difficult to compute due to factors such as mergers, acquisitions, recent IPOs, etc.) Some LEC versus high technology total return comparisons are shown in graphs at the end of this paper. Thus the LECs' performance, perhaps reasonable for a low technology, low growth regulated utility such as an electric power producer, is dramatically inferior to all other information technology sectors. By virtually any measure - revenues, market capitalization, earnings per share growth, total return - the LECs are the worst performing group in the entire information technology sector. This performance is remarkable given that it occurred in a highly favorable environment - continued monopoly combined with rising demand, a fiercely competitive and technologically progressive supplier industry, substantial deregulation of prices, and sharply increasing freedom to enter new, rapidly growing markets. Rising demand from digital services carried over normal telephone lines - fax and modem data traffic - has also helped the LECs. Then, beginning in 1994, large scale use of Internet services began to affect the industry. And although in the long term the Internet is a threat to the LECs, its immediate effect is to raise their revenues and profits substantially, particularly through increased demand for business data services, business and residential ISDN, and residential second lines (for modem-based Internet access). However, where the LECs have had to face real competition, they generally appear to have fared poorly in both financial and market share terms. In their investments outside of their core network businesses, the LECs' record appears to be poor, or at least substantially inferior to the growth rates of the markets they enter. In a number of foreign markets including Mexico and Eastern Europe, the LECs have suffered major and embarrassing failures. Where they have faced competition within their core markets, which is rarely, they have not done well either. The LECs face effective competition in their regulated wireline businesses only in the market for large business users of high speed digital services in a few urban centers. The competition comes from AT&T and MCI bypass services and/or from smaller recent entrants such as MFS and Teleport (so-called Competitive Access Providers, or CAPs). In this sole competitive market segment, accounting for perhaps 5% of their total local markets, the RBOCs have improved their technology and lowered their prices far more than elsewhere, and yet have still lost market share rapidly. In lower Manhattan, for example, NYNEX now holds less than a third of the market for high speed data links to financial services firms, despite very steep price cuts.21 General Productivity Indicators The LECs also seem to fare poorly by general productivity measures such as revenue growth and revenue or value-added per employee. The LECs have sharply reduced their workforces over the last decade, particularly in their core network businesses, while adding employment in the new businesses they have entered. LEC average revenue per employee has doubled since the 1984 AT&T divestiture to about $200,000 per person, another fact which the LECs frequently cite as evidence of their progress and efficiency. However, for several reasons this fact is less impressive that at first it may seem. First, while the LECs have surely have increased their conventional productivity (i.e. uncorrected for technical progress or quality) to some extent, it does not seem that these increases have been very rapid. Much of the increase in LEC revenue per employee results from outsourcing, not real or internal productivity improvements. All of the LECs have sharply increased their dependence upon outside sales personnel, particularly for small business services and ISDN. Pacific Bell and other LECs have terminated considerable numbers of field employees and then rehired them as contractors, rather than "employees." Bell Atlantic has outsourced at least some directory assistance operations to low-wage locations in the Southwest U.S.; anecdotal evidence suggests this has caused a decline in service quality, because databases are less current and operators are unfamiliar with Bell Atlantic's geographical service area.22 But secondly, $200,000 per employee is low by high technology standards. Firms such as Compaq, Intel, Hewlett-Packard, and Microsoft generally have revenue per employee ratios ranging from $300,000 to $500,000. In fact, revenue per employee for the LECs is still only twice as high as that of FedEx (about $200,000 per person for the LECs versus $100,000 for FedEx). Given that FedEx is essentially forced to run a nationwide daily trucking service linked to an airmail and package sorting system, and given that FedEx's service levels are so clearly superior to those of the LECs, it is somewhat perplexing that the LECs need as many employees as they do. The LECs' businesses are perhaps more subject to automation and computerization than any other sector in information technology - certainly far more so than the software industry, which remains extremely labor-intensive. Thus, while the LECs' monopoly status makes direct comparisons impossible, there is some reason to believe that even their conventional productivity levels (again, unadjusted for quality or technology levels) are poor. When faced with this and similar comparisons or criticisms, the LECs often respond by arguing that they must serve everyone, not just profitable customers, and that such service requires the LECs to construct and maintain very large networks of long-distance copper loops, which constitute an enormous low-technology construction activity. Actually, however, the LECs' total expenditures on all local wiring combined are only about 8% of their revenues. Most other categories of LEC costs - switches, fiberoptic cable, office buildings, general purpose computers, programmers, management, etc. - are highly similar both technologically and economically to those of information technology firms and highly IT-intensive sectors such as financial services. Thus the LECs' productivity behavior is hard to explain through any uniqueness associated with their need to maintain local loops. For various reasons, the LECs' revenue growth rates are equally unimpressive. The LECs have had unregulated access to world telecommunications markets and most new businesses for over a decade, and have entered many new areas including domestic and foreign cellular service, foreign local service, Internet access, data services, domestic and foreign paging, and domestic and foreign cable television. Yet LEC total revenues have grown quite slowly, and while the LECs generally do not discuss the profitability of their new businesses, it appears that the on average LECs have not been highly successful in these ventures. Furthermore the LECs' record in revenue growth for data services, to the extent we have evidence about it, appears to be poor, particularly in T1, ISDN, and Internet services. Although total U.S. T1 lines and revenues are growing about 40% per year, this is far less than the Internet is growing, and less than revenues would grow if the price / performance of services were improving rapidly. Network equipment vendors and Internet access providers have experienced compound revenue growth rates of 75%-300% per year over the last five years. Similar statements apply to the LECs' ISDN businesses, which still lag far behind several countries (France, Germany, Singapore) where ISDN is available for the same price as POTS, and now far more widely used. Most importantly, however, LEC productivity is abysmal when adjusted for technology and quality. The conventional productivity measures fail to capture the extent of LEC inefficiency and/or monopolistic behavior in this case because of the high technology nature of the industry. In such industries, conventional indices are easily overwhelmed by three other factors: (a) whether new technologies and functions are introduced rapidly or only after long delays, (b) the rate at which new technology is reflected in price/performance improvements, which should generally average 30-50% per year, and (c) service quality, including high technology service functions such as use of email and electronic payments. In fact, it may be that part of the apparent improvement in LEC conventional productivity measures has been achieved by slow technology adoption and reduced service quality, which the LECs (unlike most firms) can impose on customers because of their monopoly position. I consider the quality issue first. LEC Operational Performance: Customer Service Functions and Service Quality LEC customer service performance can be imperfectly but usefully assessed by considering LEC behavior relative to similar functions in companies which, like the RBOCs and GTE, must serve very large, price-sensitive, geographically dispersed, network-like and/or residential markets. Several companies and industries, including some with spectacular financial and productivity performance, have customers, services, and problems somewhat similar to those of RBOC/GTE residential markets. For example, UPS and Federal Express have to deal with broadly similar universal service issues. They offer service to both individuals and businesses, including suburban and rural areas, and generally do not charge rural customers higher prices than urban ones. Their markets are mature, clearly growing less rapidly than communications, and depend upon expensive physical transportation systems whose underlying technologies change slowly. UPS also has a unionized workforce. Yet both firms offer highly reliable same-day pickup (usually on 2 hours' notice in fact), next-day delivery guarantees for essentially the entire United States. Other sectors with at least some similar problems include large consumer banks such as Bank of America, Wells Fargo Bank, and Citicorp, and/or consumer financial services firms such as Fidelity Investments and Charles Schwab. Yet all of these firms, despite being subject to intense competition and sometimes also significant regulation, seem to do just fine. Perhaps even more importantly, however, service quality comparisons suggest that the RBOCs are behind all of these benchmark firms, particularly though not exclusively in use of information technology. For example, Fedex and many services firms (including several Internet access providers) accept payment over the Internet, will upon request automatically bill your credit card, and also take credit cards for individual purchases or services. Most RBOCs do not accept either payment over the Internet or even credit card payments at all; they require a paper check to clear before they will give you a phone line. If you are moving across RBOC boundaries, this can be quite a problem. Similarly, Fedex, UPS, and many financial services firms allow customers to schedule and track transactions and accounts via the Web; Fedex offered this service by late 1994. A number of major financial services firms, including Fidelity, Schwab, and Wells Fargo, as well as startups such E*Trade, allow customers to check their accounts, make payments, order products and services, and engage in transactions over the Web. Citicorp has offered PC-based electronic banking in some form since 1992. Internet-derived transactions now account for over 20% of Charles Schwab's assets. These firms also have extensive links to related web sites. Schwab, for example, provides extensive equity research information online, with live links to the websites of many companies. Cisco, the largest vendor of networking equipment, recently announced that it had sold $75 million over the Web this year, with Web business increasing sharply. Dell and several other PC vendors also allow shopping via the Web, and have extremely comprehensive product information on their websites. All of these firms allow and routinely use email for communications between customers and the firm. Many of them now allow electronic payment over the Web, and almost all of them post job openings on the Web and accept resumes and job applications by email. Most LECs offer few or none of these options, and are far behind most other service industries in using the Internet even for routine internal functions. The RBOCs do not yet use email, or permit their customers to use it, for communications between the company and customers. For example, and in stark contrast to Internet service providers and online services, no RBOC will routinely email you an account statement or bill, or permit customers to send maintenance requests via email. No LEC accepts electronic payment except through pre-arranged monthly debiting of checking accounts; none accept payment over the Internet. Problems with telephone service can be reported only by telephone, which can be hard if your problem is that your telephone doesn't work; no LEC allows service requests, or offers service status reports by email or on the web. No LEC appears to post job openings or accept job applications over the Internet. Indeed, the LECs were very late to have web sites and internal email at all. Given that the LECs are now offering Internet access including email as a commercial service, it is rather peculiar that they cannot use it to communicate with their own customers. However, this is not an isolated example; in several other ways, service firms such as FedEx adopt communications technologies faster than the LECs who provide them. For example, in many regions where Signaling System 7 services such as caller ID are available, FedEx has already integrated SS7 information with its own databases and service functions. When I call FedEx in Berkeley, within ten seconds a FedEx operator answers, addresses me by name, and asks if I am calling from my home address. Customers calling their LEC in regard to their telephone services do not typically have that experience. And when the LEC service organization calls, it is sometimes far from welcome. The last time I reported a problem to Nynex, a Nynex autodialer woke me at 7 am the following day to notify me that the problem should have been fixed. The use of information technology in RBOC residential field service is similarly uneven, as is service quality generally. When a Nynex installer or repair person goes on a call, Nynex provides only a mainframe-derived printout of the latest service request. No information concerning service history or the systems in place at the site is provided or available to the installer or service person, which causes many delays and repeat visits to fix the same problem. Once a service call is scheduled, it is frequently necessary for the customer to wait half a day or even an entire day for the visit; neither NYNEX nor Pacific Bell will be any more specific than the date when they will arrive. Furthermore, after service interruptions it is now fairly common at least for NYNEX to force customers to wait several days without service before a repair visit. Fedex will generally pick up any package, anywhere, within two hours on the same day, often until mid-evening, and will also pick up on Saturday. And if FedEx fails to deliver on-time, it refunds your money. Anecdotal evidence (I know of no systematic study) suggests that the quality of LEC telephone service may be declining, or at least failing to keep up with the standards of other (competitive) service industries. While it is now quite rare for banks with large numbers of depositors and large ATM networks to be unable to complete customer requests for days on end, it is not infrequent for local telephone service to fail for substantial periods, and for customers to wait several days for service calls. My service in Cambridge has failed several times in the last three years, once for several days; voicemail has failed several times as well. Friends of mine in Seattle and Washington, DC tell me that with increasing frequency they cannot obtain a dialtone; none of them has ever reported to me that their bank could not cash a check, that they could not use their ATM cards, or that Fedex could not pick up their packages. The LECs claim that failure to obtain dialtone is due to Internet-driven congestion, which they assert imposes unfair costs upon them, and/or impairs their ability to provide service. This claim does not hold up on examination; I consider this issue below. In the approximately one year I have had cellular service through Bell Atlantic-NYNEX, my service has been erroneously cut off twice by the provider; in both cases it was restored only after I called them multiple times, and not because they detected the problem themselves. Indeed, NYNEX has recently been fined by the state of New York for persistently low service quality.23 Perhaps even more seriously, it seems clear that the LECs' performance in selling, explaining, installing and maintaining digital services such as basic ISDN, primary rate (PRI) ISDN, and T1 is very poor. Once again I have a personal experience; when I was still with Vermeer Technologies (the Internet software company I founded in 1994 with Randy Forgaard), a NYNEX worker cut our T1 line, disabling it for two days during the Internet World trade show. Most of the high technology professionals and FCC employees with whom I have spoken concur that getting new ISDN service is often a nightmare, though they have noted some recent improvement. Service options are extremely complex, with equally complex price structures, and LEC employees are often unable to explain services, technology, or prices competently. ISDN installation and equipment configuration procedures are complex, highly error-prone, time-consuming, and require far more technical expertise than should be necessary. Several LECs have begun using reseller networks and paid external expediters as a result, and a consortium of three LECs has recently developed a simplified menu of services. Maintenance and repairs appear to be problematic as well. Anecdotal reports also suggest that the quality of PRI-level (i.e. 1.5 megabit, 24 channel) ISDN installation, service and maintenance is also poor, even in urban areas with large concentrations of digital services customers. PRI ISDN is often used by Internet access providers, several of whom have privately told me that their installation and maintenance experiences have constituted serious problems for their business. One senior FCC official told me that businesses in New York sometimes pay competent NYNEX field personnel to moonlight so that they can maintain their ISDN services privately, avoiding the need to deal with NYNEX directly. Substantial numbers of complaints have been made to the LECs, posted on the Internet, and filed with state PUCs regarding customer service for both voice and data services. For example, one particularly detailed (and to this reader, convincing) case history of the painful experiences of an Internet access provider can be found in "ISP Adventures: Dealing with Pacific Bell," on the Web at www.catch22.com. My personal conversations indicate that many users involved with high technology are deterred from purchasing ISDN service (e.g. for their home offices) simply because it seems to be too complicated, time-consuming, and expensive. Yet the LECs are considered to have comparatively high service call costs as compared to other information technology service industries. But much more importantly for the U.S. economy, the RBOCs' financial and customer service performance, however mediocre, are very good compared to their technological, productivity, and competitive performance, at least in digital services. Even if the LECs' costs were high, this would be comparatively unimportant if they were improving the price and quality of their services at a high and/or competitive rate. But they are not. 3.3 LEC Technological Behavior and the Price/Performance of LEC Digital Services: An Astonishing Situation Overview In this section we assess LEC technological and price performance relative to the networking equipment industry, other portions of the information technology sector, and occasionally to the performance of corporate networks, foreign PTTs, competitive access providers (CAPs), long-distance carriers (IXCs), and the LECs' own internal behavior. Our principal foci of interest will be (a) the long-term rate at which LEC price/performance improves, and (b) the conditions and timing of the LECs' offering new technologies and services in the market, relative to their availability and/or adoption by the competitive private sector. For example, we shall see that the LECs do not offer some digital services, e.g. HDSL, because they might cannibalize highly profitable businesses, such as T1 and business voice, whose prices remain extremely high. This analysis will concentrate on digital technologies and services, which however are both rapidly growing and are increasingly used to implement voice as well as data applications. So the analysis covers a substantial and growing fraction of the LECs' total business, including most or all future Internet services. Digital technologies for transmission and routing/switching will eventually dominate total U.S. communications traffic and equipment for most applications, including voice, computer data, and video entertainment. Very generally speaking, the economics of these technologies and businesses have substantial initial and fixed costs for laying wiring, administration, maintenance, and so forth, but also have very high rates of technological progress. Thus while the cost of service may never decline below some baseline level associated with average costs of physical wiring, buildings, and overhead, very rapid technological change should increase service performance and bandwidth rapidly. The most advanced and newest services would initially be expensive, but technical progress would bring their cost down rapidly towards the baseline level. In a competitive environment, prices should similarly decline sharply until they approach average basic network costs, i.e. some modest premium over the price of current POTS. The fraction of the service required for basic voice will decline, and voice service will thus become very inexpensive for the majority of users who will also want an average package of services, which five to ten years from now will surely include high speed Internet access. This pattern is similar, in many ways, to the behavior of many other information technology sectors such as PCs and software. To design, manufacture, test, package, ship, support, and inventory a full-featured PC inevitably costs several hundred dollars, and no new PCs sell for less than about $750. Thus, the price of each new technology generation starts far higher (generally $5,000-$8,000) but declines rapidly, reaching this minimum level after about 4 years. And while one can dispute the price rates of progress and cost levels involved, there is no question that the costs and prices of LEC digital services should follow this general behavior. However, they don't. And the LECs' divergences from technology industry norms are so enormous that subsidies to basic service cannot account for them. The evidence thus suggests that LEC performance is quite poor. This appears to be the result both of poor technological performance and the use of monopoly power to avoid passing technological improvements through to consumers via new services and/or reduced prices. For example, in their (usually still monopoly) T1 and ISDN markets, the LECs' prices and price-performance have been roughly flat for the last five years. T1 prices have declined only gradually, while ISDN prices have remained constant or even increased.24 By comparison, during the five years most information technology sectors, including network equipment vendors and services firms such as Internet access providers, have improved their price/performance by approximately a factor of 10. In most digital services markets, where the LECs face no effective competition, they have kept prices flat or even sometimes raised them, e.g. with ISDN.25 (Conversely, in the small portion of digital services markets subject to real competition, e.g. lower Manhattan, the LECs have cut prices while still losing market share.26) The T1 market, in particular, represents a situation unprecedented in modern information technology, and which I discuss in considerable detail. Finally, we shall see that even in analog POTS, particularly for businesses and apartment complexes, the LECs' performance leaves something to be desired. The Critical Role of xDSL and Future Networks: HDSL, SDSL, ADSL, VDSL Finally, it will be important to consider the LECs' behavior with respect to the technologies collectively referred to as DSL or xDSL (for Digital Subscriber Line) services. These services, together with advanced routers, and packet switches, are emerging as the likely dominant technology for future mass-market digital services to both residences and homes. I will not describe xDSL technology in detail here; an excellent overview can be found at www.telechoice.com, a consulting firm that maintains informational web pages on digital telecommunications industries. Briefly, however, there are currently four DSL technologies, which offer various forms of high-speed digital transmission over copper wire local loops. They are in different stages of commercial development, use, and public deployment which partially reflect their relative technological maturity, but also in part the LECs' monopoly power and economic interests. HDSL offers 1.5 megabit, symmetrical two digital service over two twisted pairs (i.e. two local loops) of copper wire. SDSL offers half the speed of HDSL service over one twisted pair. ADSL offers asymmetrical digital service weighted towards the downstream direction (i.e. towards the home, away from the central office), plus an analog POTS channel carried on the same line. Finally, VDSL, still under development, offers extremely high speed service over shorter loops, and will probably be used for videoconferencing and live HDTV. None of these services are yet being publicly offered. The LECs' current efforts towards public deployment - such as they are - are concentrated on ADSL, which is optimized for use by single residences. This is somewhat curious, because HDSL has characteristics highly attractive for businesses and apartment buildings, and is actually a fairly mature technology already widely used to implement T1 service by the LECs themselves.27 However, it seems that the LECs are not moving aggressively to offer public HDSL service, probably because it would undercut the high prices of their existing services, particularly T1, Primary Rate ISDN, and business voice.28 The LECs would rationally be particularly worried about the use of HDSL, in combination with new switching/routing technologies, by competitors taking advantage of the loop unbundling and interconnection provisions of the 1996 Act. I discuss this question below in connection with T1 services and also in connection with potential competition in the local loop. LEC Performance in Basic Services: Even Some Question Here While U.S. basic residential analog voice service (POTS) remains of high quality and admirably inexpensive by international standards, a number of nations now have basic infrastructures, and offer basic residential and business services, more technologically advanced than those offered by the LECs. Even more surprising, certain newer services, such as residential ISDN, are under some usage conditions already less expensive in some other nations, e.g. France and Germany, where 64 kilobit ISDN is now universally available at the same price as analog POTS. Even for analog residential voice services, the LECs' performance may be poor, since an increasing fraction of voice traffic and services are, or should be, completely digitized. This statement should even apply to the installation, diagnostics, repair, and maintenance functions, although substantial amounts of physical construction will always be required. In fact, the LECs do frequently supply groups of business lines by using 1.5 megabits per second digital service - HDSL, T1, or PRI ISDN - carried over a few copper loops.29 One FCC document suggests, furthermore, that 1.5 megabit PRI ISDN service only costs the LECs about the same amount as ten analog POTS lines.30 However, the LECs still charge the standard price for each business line. The LECs also, however, seem to be implementing some digital technologies more slowly than other nations, and certainly far more slowly than they could. France and Germany have had 100% digital switching for over five years; several RBOCs, such as SWBT, still use analog switching for 20% or more of their lines.31 In several LECs, over 20% of total access lines still do not have access to ISDN.32 Nonetheless, most RBOC voice functions and/or services such as switching, trunk transmission, caller ID, voicemail, service activation/deactivation, and directory services either are or certainly should be very highly computerized. These functions and services should thus be subject to the rapid, continuous technical progress that characterizes all information technology. Only local loop transmission to homes and telephones themselves remain predominantly analog, and as we noted earlier the cost of constructing the entire copper local loop for both homes and businesses is less than 10% of LEC revenues. Thus even in voice services it is somewhat puzzling that local rates, and rates for fully digital services such as voicemail, have not declined more substantially. However, in the case of residential POTS, one can give the LECs the benefit of the doubt: these are mature markets, still partially dependent upon analog technologies, much of whose market is composed of residential consumers, still very heavily regulated, and burdened with various large cross-subsidizations and political pressures. Furthermore, the line quality of U.S. POTS, which is critical for modem usage (e.g. dialup residential Internet access), unquestionably remains far superior to analog line quality in most of the world, including most of Europe, particularly France and southern Europe. So in residential POTS, the verdict on LEC performance can be debated, although the lack of rapid long term improvement does suggest a problem. But digital services including large-volume business voice, ISDN, T1, and HDSL, are a different story. Digital Services: An Overview of ISDN, T1, and xDSL Services Both ISDN and T1 are digital services used predominantly by businesses and delivered primarily over existing copper wire loops. T1 service, which provides basic digital transport of 1.5 megabits per second, has been available for over a decade. ISDN, which although first developed 20 years ago has been commercially available for only about 5 years, is available in two principal forms: Basic Rate service, which provides 64-128 kilobits per second, and Primary Rate (PRI) service, which provides 1.5 megabits per second divisible into 24 voice and/or data channels. Both BRI and PRI ISDN also provide signaling and control information that allow for flexible use of various services including alternating between voice and data, use of multiple telephone numbers on a single line, and others. Demand for both ISDN and T1 is growing sharply, spurred by the growth of SOHOs, commercial Web sites, high speed fax and data transmission, corporate networking, and high speed Internet access. In these inherently digital services, the LECs have little excuse for slow technical change or price-performance improvement. Yet for the last five years, despite increasing traffic and rapid technical progress in equipment, T1 and ISDN prices have at best declined slowly, have often remained flat, and have sometimes increased. (DSL services cannot be measured because they are not yet available, even though HDSL is already in internal use by the LECs.) The price-performance of the LECs' ISDN and T1 services have therefore deteriorated sharply relative to corporate networks, services subject to competition, and even relative to some foreign government telephone monopolies. To be sure, complete price histories for these services are hard to obtain; most LECs declined to provide them to me despite repeated requests, including requests made personally to senior executives of two RBOCs. Obtaining information from regulators is possible but difficult. Most are not computerized and in general quite bureaucratic. They also often discard old (paper) tariff documents when new ones take effect; they keep poor records; some don't even have public photocopiers. Most state PUCs and the FCC therefore do not even have systematic knowledge of the costs, price histories, or revenue levels of digital services under their own jurisdictions. However, from PUC tariffs, information provided by some LECs, interviews with users, interviews with resellers, various published sources, and other interviews, I have been able to obtain reasonable confidence concerning the price evolution of digital services. ISDN As of year-end 1996, there were roughly 1 million ISDN users in the United States, and the number is growing rapidly. However, ISDN prices in non-competitive markets (i.e., most markets) have remained flat for the last 5 years. Some however, have even increased, and would have increased more if LEC efforts had not been resisted by consumer groups and computer firms. ISDN is also extremely expensive, particularly for heavy users. Although basic services charges are typically $40-80 per month for 2-channel BRI (i.e. 128 kilobits from two 64 bit channels), usage charges generally average 1-2 cents per minute per channel, even for local connections. For heavily used lines, this can bring bills to hundreds of dollars per month. This is very far in excess of cost estimates, even those of the LECs themselves. Even more surprisingly, several RBOCs have raised (or tried to raise) ISDN prices substantially. In 1996 U.S. West even tried to triple ISDN prices (in Washington state) to over $150 per month; opposition led by Intel forced the company to back down.33 Pacific Telesis also has raised prices recently. The only exception to this pattern has been Bell Atlantic, which in 1995 reduced ISDN prices modestly to attract new customers, with apparent success. But even Bell Atlantic's price-performance has not improved rapidly over the last five years, and it is the sole LEC to behave in this way. Furthermore, ISDN is still not easy to obtain. First, it is not even available in all areas because the LECs have not invested sufficiently in modernizing their switches and in some cases their loops. For example, ISDN was still available to less than 80% of SBC's lines as of the end of 1996. No RBOC has yet provided 100% availability, and only two are even close.34 But second, ISDN ordering, installation and service packages have also until very recently been complex and time-consuming, and in many areas remain so. This has largely been the result of poor LEC/Bellcore standardization, service definition, and marketing efforts, which failed to consider ease of use as a goal of standards development and product marketing.35 Partially as a result of this complexity, as well as high prices, ISDN usage and even availability in the U.S. trail far behind several other nations including France, Germany, and Singapore. ISDN is available to every French and German user, residential or business, for roughly the same price as analog POTS. There are now about 1.5 million German ISDN users - a larger number than in the U.S., for a nation whose population is one-fourth that of the U.S. and which has only one-fifth as many telephones. Furthermore, German ISDN usage continues to increase rapidly. T1 and HDSL T1 was developed more than a decade ago.36 It was initially intended primarily for internal use by telephone companies, e.g. for trunk communications between central offices, and secondarily for use by very large corporate users. When originally defined, T1 was implemented with repeaters and required complex, expensive electronics technologies. However, T1 usage began to grow rapidly as corporate networking and online services became more widespread; and as usage volumes grew, technology improved. T1 is now used for a variety of purposes including high volume voice traffic, general corporate networking (e.g. connecting LANs located across a city from each other), and - increasingly - for high speed business Internet access. T1 lines are generally not switched, but rather used as dedicated lines from one location to another, e.g. from a business to its Internet service provider. Moreover, in contrast to residences, which tend to send small amounts of data but receive large amounts (e.g. while surfing the web), businesses tend to need symmetrical service, for example because they are operating web sites that are being read by residential users. Thus, because T1 provides symmetric transmission (1.5 megabits in each direction), it has become the LECs' principal business data service offering. Although the LECs refuse to release their corporate statistics, it is estimated with reasonable reliability that as of year-end 1996, there were more than 1.3 million T1 lines in use in the United States. T1 usage, in both lines and revenue, is growing approximately 40% per year, driven primarily by Internet use and by the related growth of corporate "Intranets." It is not clear what percentage of T1 lines are still used internally by LECs as opposed to paying customers, but it is clear that the external, revenue-generating fraction is increasing. T1 service is very expensive. Although once again the LECs refuse to release revenue data, and there is substantial variation in prices by LEC and jurisdiction, tariffs and price schedules can be obtained and average real prices estimated.37 Installation costs range from $300 to $2,500. Monthly charges are generally mileage-dependent, and there are often additional charges for other conditions, e.g. if the line goes through more than one central office. The consensus is that the national average real price of a T1 line seems to be somewhere between $300 and $1,000 per month, probably around $400-600.38 In addition to examining tariffs, these estimated prices were confirmed by a number of private interviews with users, Internet access providers, and by having my assistant request personal quotes for T1 service to my home offices. Only one LEC supplied any estimate of its average T1 revenues per line publicly (Pacific Telesis, which estimated its T1 lines averaged approximately $500 per month). This suggests that in 1997, T1 service will be roughly a $5-10 billion business for the LECs, or roughly 4-8% of their total revenues. T1 prices, like ISDN prices, have often remained flat for the last five years, although there have been selective, gradual price reductions in many areas. There have been some major price reductions, not primarily in tariffed rates but rather in connection with service packages for very large customers - in order to match competitors in the small number of dense major urban business markets that are subject to true competition. However, for the past five years T1 and ISDN prices for the overwhelming majority of users - small business, most cities, all suburbs, and nearly all apartment buildings and residences - have been roughly flat or declining very slowly in both listed tariffs and actual prices paid. They clearly remain extraordinarily high and have declined less than technology has reduced costs. Senior RBOC executives have in fact admitted to me privately that costs have declined, margins have increased, and that prices are being maintained simply because there is no real competition. T1 and HDSL: An Interesting Window on T1 Profitability and LEC Strategy In fact, the T1 situation is even more extraordinary, and even more interesting, than the above discussion suggests. T1 service was originally very expensive to implement, because it required repeaters and sometimes fiber. However, most new T1 lines are no longer implemented using this technology. For the last several years, the LECs have implemented an increasing fraction of new T1 lines using HDSL. With one very minor exception (an unregulated subsidiary of U.S. West which offers SDSL, not HDSL), the LECs do not ever exhibit an HDSL interface to users, nor do they ever offer HDSL service. But they use HDSL technology extensively to implement the T1 service that they do sell, and that their customers must use. HDSL, which stands for High speed Digital Subscriber Line, is one member of the family of DSL (or xDSL) technologies discussed above. HDSL requires two copper loops (i.e. two twisted pairs of copper wire). Current HDSL (a faster version is under development) provides symmetrical 1.5 megabits per second service in a fashion which is highly compatible with both T1 and Primary Rate ISDN, so it is comparatively easy for LEC networks to handle HDSL traffic. HDSL has several other interesting characteristics. Unlike POTS and the earlier T1 technology, HDSL requires neither a powered line nor line conditioning such as filtering; for most copper loops, HDSL terminal equipment needs only a bare copper wire. Terminal equipment costs $300-$600 per end (i.e. $600-$1,200 per circuit) and is provided by an extremely competitive, entrepreneurial equipment industry including firms such as Pairgain, ADC, Wiltel, large firms such as Alcatel, and many others. Unit costs are declining rapidly as the industry follows the typical electronics technology curve. It would appear that well over half a million HDSL lines are already in service. Pairgain Technologies (www.pairgain.com), which has the highest market share, is now a $200 million company growing 90% per year. Although copper loops sometimes present problems and HDSL cannot yet be used on very long loops, for the majority of loops the operational costs of HDSL service are not much higher than two to four POTS lines. Often HDSL lines might even be less expensive than that. An HDSL-based local T1 line does require 1.5 megabits of dedicated bandwidth between central offices, which POTS does not, but conversely (and unlike POTS) it requires no switching resources, no power supply, no usage monitoring, and no line filtering. In fact a single HDSL link is sometimes also used by LECs to provide up to 24 business voice lines to a single location, with a box on the customer site decoding the HDSL line into separate voice lines. Furthermore, HDSL continues to undergo rapid technological progress, which is producing less expensive terminal equipment and dramatically improved capabilities. First, a recent variant of HDSL called SDSL (Single line DSL), already provides one-half a HDSL channel (i.e. 768 kilobits per second) over a single copper twisted pair, making it suitable for many residences and home offices.39 Second, a new upward compatible generation of technology, HDSL-2, is being developed which at least doubles the speed of HDSL and extends its range (i.e. the length of the copper loops over which it can function). HDSL-2 will provide 1.5 megabits per second, i.e. full T1, over a single twisted pair, or 3 megabit service over two twisted pairs.40 The third development is in many ways the most interesting. Current HDSL can only be used for dedicated lines; a user cannot connect to different numbers at will, because current central office switches cannot handle HDSL traffic. However, the HDSL equipment industry is now developing DSL Access Multiplexers (DSLAMs), which aggregate DSL lines so they can be presented to an ATM switch or advanced router. One DSLAM, from Alcatel, is already being marketed; another dozen will be commercially available within one year. The level of entrepreneurial activity in this area is very, very intense, and it is clear that at most 18 months from now fully switched HDSL service will be technologically and commercially feasible (though perhaps not yet available if the LECs continue to refuse to deploy it).41 At that point, HDSL (and other DSL services) could become a highly attractive service usable by both residences and businesses for a wide array of voice and data applications. Interestingly, the LECs have not yet fully standardized HDSL, a situation which if not dealt with could pose problems for large-scale public HDSL service.42 Once again, this might be in their interest, since competitors offering public service would benefit from standardization, while the LECs do not need standardization to offer dedicated T1 lines, and the LECs probably want to delay the availability of public HDSL service as long as possible. At this point, we might ask two questions. First: if a high fraction of new T1 lines are implemented using HDSL, why does T1 service still cost $500 per month on average and why do installation costs sometimes exceed $3,000? And second: why don't the LECs rush to offer public and soon switched HDSL service as a tariffed service, just as they offer ISDN and T1? There does not appear to be a very nice answer to either question, at least from the viewpoint of economic welfare. Analysis of LEC Conduct in Digital Services It is clear that the transition from repeater-based T1 to HDSL, and the continuous development of HDSL technology by a highly competitive industry, has resulted in dramatic cost reductions. It is also likely that major cost reductions have been achieved, and/or are achievable, in ISDN. It is certainly understandable, however, why the LECs might not wish to reduce T1 prices; they are making huge amounts of money from T1, just as IBM once did from mainframes. However, the direct threat to the T1 revenues and profits is only one reason, and perhaps not the largest one, that the LECs do not reduce T1 prices and/or offer public HDSL service. Assuming that the LECs wish to maximize their profits and/or the personal positions of their executives, they would wish to hold their monopoly position, charge high prices, obtain high profits, and reduce the rate of technical change in the industry. To do this, however, the LECs need to engage in price discrimination and prevent arbitrage between differently priced services. The problem they increasingly face is that long run technology trends are bringing the cost of high performance services down to a common level not much higher than POTS. But if either the LECs or competitors begin offering inexpensive high performance services such as switched or even dedicated line HDSL, this would allow cannibalization of T1, ISDN, and business voice revenues. (This is similar to what happened in computing: microprocessors brought the performance of inexpensive machines close to that of mainframes, destroying IBM's mainframe pricing.) A single SDSL channel over one twisted pair provides six times the bandwidth of basic ISDN, or up to twelve voice channels. HDSL over two twisted pairs provides double this bandwidth, and HDSL-2 will provide yet another doubling of the capacity of either one or two twisted pairs. This would make voice service to businesses and apartment buildings, and/or high speed data services, very inexpensive. The LECs have definitely noticed this, and do not like it.43 At present, use of HDSL in this manner is prevented by the LECs because they simply do not offer HDSL service, and do not permit competitors to offer it either. Thus, the prospect of widely available HDSL service in combination with unbundling, competition, and increased collocation rights for ISPs must surely seem very unappetizing for the LECs. This also may suggest why the LECs seem less antagonistic to another digital technology, ADSL, despite the fact that HDSL is more mature. HDSL is easier for competitors to implement, addresses highly profitable business markets, and requires less dependence upon the LEC. The LECs' Preference for ADSL In contrast to HDSL, some of the LECs are moving - not with astonishing speed, to be sure - to provide ADSL service to residences. The LECs' preference for ADSL is interesting. ADSL (for Asymmetrical DSL) provides 1-6 megabits per second in the downstream direction (i.e. away from the central office), but only 350-700 kilobits per second upstream. It also can operate over somewhat longer loops than current HDSL; however, HDSL-2 will eliminate this difference. Unlike HDSL, ADSL also provides analog POTS - so ADSL lines require power and filtering. For residential use, ADSL has desirable characteristics, since in the short term most home users want to keep their current analog phones, and home users also generally want to receive data more quickly than they need to send it, e.g. while surfing the Web. However, for business use, ADSL is a poor choice. Businesses require symmetrical or even outward-weighted service to operate high speed web sites, and symmetrical service to replace business voice lines. ADSL could also be more complicated for competitors to offer or interact with, because ADSL lines must be powered and filtered, and the analog POTS signal must be split off from the data signal. Competitors would either need to do this themselves, or rely upon the LECs to do it for them. No new entrant, at least no ISP or ESP offering modern digital services, would want to handle analog POTS. This would raise substantially higher barriers to entry relative to HDSL, which requires only a bare copper wire and which is highly suitable for a wide array of applications, including arbitrary combinations of voice and data. In fact, with HDSL or SDSL, what a competitor basically wants (once it has access to copper loops, collocation rights, and sufficient LEC operational information) is simply for the phone company to get out of the way. It is easy to understand why the LECs might not enjoy this position. The LECs' defenses It is difficult to find any legitimate basis for the LECs' conduct and performance in regard to ISDN, T1, and the various xDSL services. The LECs have offered three reasons for their high prices and service quality problems: slow standardization; low volume; and the difficulties of dealing with wide area, outdoor networks. None hold up on examination. Although costs may have been high when ISDN was new and had few customers, that stage is now clearly past (or should be if the LECs are even minimally efficient). Demand and usage are growing fast, high volume production of standardized ISDN equipment is now underway by multiple suppliers, the underlying technologies are improving 40% per year or more, and a global market has arisen. One senior RBOC executive told me that ISDN costs had remained high because ISDN technology had only recently became standardized in a manner permitting efficient deployment. This, however, is a very poor excuse. In the first place, if it is true, it is the LECs' fault. ISDN standards were largely developed by, and under the control of, the LECs themselves (and Bellcore, their research arm). ISDN has been in deployment for six years, and was under development for many years previously. Moreover, its underlying technologies and components, e.g. digital switches, have been improving on the general information technology curve. It is true, however, that many official telecommunications standards, including ISDN, take longer to develop than they should - on average, probably two to five times as long as computer networking standards. They are also, often, developed in comparative isolation from market needs, which does affect their market acceptance and cost (e.g. via the complexity and installation problems mentioned above). Thus Bellcore-LEC standards such as ISDN take longer to develop, and yet are sometimes less successful, than standards developed quickly in industries such as computer networking. During the period from initial ISDN development to the present, the personal computer and corporate networking industries went through at least three generations of standardization and far greater increases in performance. The LECs second justification for high and stagnant ISDN costs and/or prices is that although U.S. ISDN usage is growing fast, it is still very low. Both U.S. growth rates and absolute usage levels remain lower than for several other nations. Some LECs still have less than 100,000 ISDN lines installed. The U.S. total is about 1 million, i.e. less than one-fifth the penetration rate of Germany, and less than one-seventh the rate of Western Germany. This is despite the fact that penetration rates of business and home personal computer ownership, business computer networking, and business and home Internet usage are far higher in the United States than in Germany or anywhere else. And once again, the LECs' performance in digital services contrasts starkly with behavior in similar, competitive, U.S. industries. In fairness to the LECs, voice service quality remains better than the quality of Internet service, which is often poor due to the Internet's 15% to 25% per month growth rate. However, the corporate networking sector grows rapidly - less rapidly than the Internet, to be sure - with fairly high general quality. And the Internet services industry has been coping with explosive technical change and demand growth, as well as the emergence and rapid evolution of new standards, for the last four years - but with very different strategies, and different results, than the LECs. U.S. residential Internet usage has consequently risen from effectively zero to 15 million households in 3 years; yet prices have declined, and access speeds have greatly improved (thanks to better modems, routers, and computers, not to the LECs). Moreover, many multinational banks and high technology firms operate networks with hundreds of thousands of nodes, including high speed LANs and WANs, remote access facilities, Intranets, Internet access to desktop PCs, and large internal telephone, voicemail, and paging networks. The size, bandwidth, and price-performance of many of these networks have been growing and improving for many years. While the LECs' cost structures may be higher and/or different, it is difficult to see why LEC services do not exhibit comparable rates of improvement. Hence some LECs offer a third explanation for high ISDN and T1 costs, namely that wide-area residential networks with long copper loops have inherently higher costs. They do, but those costs and prices should still decline over time, particularly for digital services. Furthermore, as observed above, the entire copper loop infrastructure, for all homes and businesses, accounts for 40% of LEC capital costs, and LEC capital spending is less than 20% of revenues. Much of the rest should be subject to substantial technological progress, including the LECs' own MIS organizations as well as their switching, trunk lines, network control systems and OSSs, and so forth. While it is reasonable, and for some services even inevitable, that the LECs' absolute cost levels and price-performance ratios would lag those of the technology sector generally and/or of concentrated indoor business networking markets, LEC rates of price-performance improvement and technological progress should still be substantial. In fact, because the LECs have been slow to modernize, installation of modern technology should now bring them greater productivity gains than corporations that continuously replace five-year old systems. Of course, that won't happen if LEC network capital investment levels are declining as a percentage of revenues, access lines, and access minutes. And finally, one can wryly observe that the RBOCs and AT&T have long argued (especially when seeking to prevent antitrust actions and/or competition) that economies of scale and integration bring huge benefits in operating networks.44 The RBOCs are five to ten times larger than even the largest global corporate networks and Internet providers, and the scale of their data services business is growing rapidly. Where are scale economies when you need them? A General Observation: LEC Slowness in Technology Development and Adoption In addition to rational monopolistic conduct, the LECs' behavior is almost certainly attributable in part to the slowness and management entrenchment that often arise in long term monopolies or oligopolies. There is much evidence that even where it would reduce costs and/or increase revenues, the LECs are sometimes quite slow to implement new technologies. We noted above that the RBOCs have been slow to use Internet technology in traditional functions, and that LEC implementation of digital switching and ISDN have even lagged other government-run PTTs. The LECs have been equally slow to enter Internet access service, Web hosting, commercial Web services, and Internet/Intranet systems integration; Nynex still does not offer Internet access service at all. Indeed, the RBOCs have generally avoided entering rapidly moving, highly competitive arenas in which their primary competitors would be startups and/or aggresssive firms from the computer, software, or networking industries. Where they do enter new digital services markets such as Internet access, they enter late and have captured low market shares despite their huge advantages in financial power, organizational resources, and experience. For example, all of the LECs plus AT&T combined still hold less than 10% of the Internet access market, in part because they entered years later than other firms such as America Online, UUNet, PSI, CompuServe, Netcom, and Microsoft (through the Microsoft Network).45 Indeed, several RBOCs still have not made their white pages and yellow pages services fully available on the World Wide Web; all were late to do so. Nor do they permit remote electronic updating of yellow pages entries by customer-advertisers, a capability routinely offered by many web hosting firms. Conversely there are now literally hundreds of startup firms offering both specialized and general purpose yellow pages, white pages, classified advertising, and location/identification services, and general web hosting. In fairness, AT&T and RBOC Internet service offerings do seem to get high marks for reliability and line quality. Similarly, the RBOCs have generally been slower than CAPs, IXCs, and/or corporate networks to introduce other popular advanced digital services such as ATM services. Frame relay service is probably the only widely used digital service the LECs have been relatively quick to implement and offer. The only advanced general purpose digital service the RBOCs and GTE have been consistently earliest to offer is the least popular - Switched Multimegabit Data Service (SMDS). SMDS was developed and standardized by the RBOCs, GTE, and Bellcore themselves. It offers advanced services, but assumes that users will be willing to install expensive new equipment on their premises. Moreover, SMDS is not available outside the U.S., is often expensive, must compete with increasingly popular frame relay and ATM services, and offers incomplete interoperability despite the LECs' control over its standardization. This issue brings us to the final area in which we will consider LEC technological performance, namely generating and standardizing new technologies. LEC Performance in Innovation, Standardization, and R&D Innovation and R&D The RBOCs and GTE have a very mixed record in R&D and innovation. On the one hand, the LECs, AT&T, Bell Labs and/or Bellcore have been responsible for many of the architectures now being used to provide digital network services, including SONET, ATM, xDSL, and others. However, the ponderous, extremely slow, sometimes user-insensitive nature of official Bellcore standards, including ISDN and SMDS, has caused them to develop slowly and sometimes, as with SMDS, to be rejected or superseded by private industry. And while some standards such as ATM appear to be quite successful, in other cases including ISDN and xDSL, standardization has remained incomplete for long periods, which has caused incompatible variants to enter use, even within the LECs. Moreover, successes such as ATM and xDSL seem to be exceptional in the overall context of the development of communications and networking. Ranging from the Internet to the Web to local area and wide area business networks, virtually every other major, broadly used innovation in digital communications and/or information services technology over the last 15 years was developed outside of the LECs and adopted by them later than by others. Moreover, the pace of standards development and embodied technical change in the private sector is much, much faster than that produced by monopoly and/or official processes. The RBOCs' main R&D vehicle has been Bellcore, which has a budget of roughly $1 billion per year, about four-fifths of which is contributed by the RBOCs. It is rather arguable as to whether Bellcore has produced good value for the money. Moreover, the RBOCs' combined contributions to Bellcore have been declining,46 and they have decided to dispose of Bellcore - they have sold it to SAIC for a nominal amount. And aside from Bellcore efforts, it is exceedingly difficult to identify any major architecture, innovation, or even R&D effort in digital communications for which the LECs are responsible. Most LEC annual reports do not even discuss R&D as a subject, itemize it in cost discussions, or provide any information concerning the size of their independent R&D budgets. Their Bellcore contributions average about 0.6% of revenues. I have however been able to obtain information on some LEC R&D efforts (see graphs at the end of this document). It appears that LEC R&D is roughly flat, i.e. declining as a percentage of revenues. Most LEC R&D is conducted through Bellcore; the LECs' independent R&D spending is small. GTE, which is not a member of Bellcore, does itemize its total R&D spending. GTE's R&D expenditures are low and declining. They have declined in absolute terms since 1990, and are now less than 2/3 of 1% of revenues, i.e. about the same as the RBOCs' Bellcore contributions. Two other RBOCs provided R&D spending information; in addition a Bell Atlantic executive told me privately that his company's R&D budget was about $150 million, and declining slowly. For both the RBOCs and GTE this is astounding, both in level and direction. In no other portion of the entire information technology sector, ranging from consumer electronics to semiconductors to computer networking, is R&D spending either so low, or a declining percentage of revenues. Even the personal computer industry, which has very low R&D spending because it basically assembles components from other industries, has R&D spending of 3-5% of revenues.47 On the other hand, the LECs don't seem to generate much with their R&D spending, so in a bizarre way this strategy may be rational given current LEC management. But both the low level of spending and the paucity of results indicate a serious problem. Furthermore, no other segments of the networking or communications industries have had any problem finding things to invent recently. Over the past decade networking and data communications has been the fastest-moving technology area in information technology, and quite possibly in all of economic history. The ferment in this industry, both from large firms and startups, has been nothing short of stunning. Much of it, ranging from the evolution of the Internet and the Web to xDSL and ATM, will have a profound effect upon the LECs' future. Yet the RBOCs and GTE have been essentially absent from most of it, despite the fact that for every year in the past decade, RBOC and GTE R&D spending (on Bellcore plus internal efforts) has exceeded total U.S. venture capital investment in the telecommunications sector. Over the past decade academia, government labs, and private industry have generated a spectacular array of innovations in digital telecommunications. These include but are not limited to the World Wide Web itself, graphical browsers for the Web, new generations of the Internet protocol, the RSVP Internet resource reservation protocol, the CUSeeMee Internet videoconferencing standard, copper wire LAN architectures such as 10BaseT, switched Ethernet and gigabit Ethernet, a wide range of digital audio, image, multimedia, and video transmission standards (such as MIDI, Real Audio, JPEG, the VRML virtual reality standard, streaming technologies by Narrative and others, MPEG, MPEG2, the Grand Alliance digital HDTV standard, and many more), rapidly improving digital still and film cameras, many wireless and satellite technologies, encryption and security technologies, and a huge array of commercial products and services based on these and other innovations. The LECs have been almost entirely absent from these efforts. It is striking that the LECs are absent from the most important standardization efforts in networking, for example those related to the Internet. There is not a single LEC representative on the board of the Internet Society, of the Internet's governing standards body (the Internet Engineering Task Force, or IETF), nor on any of the principal IETF working groups, including those working on Internet telephony.48 The IETF, incidentally, has functioned very effectively as an informal, volunteer body with very limited funding. This group has upgraded Internet standards fast enough and well enough to handle traffic levels that have been doubling every year for nearly two decades, and that by most measures now exceed the traffic volume handled by any individual LEC. By comparison, therefore, it is not clear that the American people have received very much in return for the $10 billion or so the RBOCs have spent on R&D since divestiture, or the $1 billion GTE has spent. Once again, this situation parallels that of other declining dominant firms such as IBM in the decade prior to John Akers' termination in 1993. Between the early 1980s and 1993, i.e. during the period of its worst managerial decline, IBM spent over $50 billion on R&D. By the early 1990s its annual R&D budget was over $6 billion - more than all other computer firms combined, and also more than total annual U.S. venture capital investments in all industries. Yet IBM was defeated by other firms with only a fraction of its resources, because they were more effective in using them. The LECs show signs of a similar condition. In this connection, it should be noted that there is a major, global problem with the nature and pace of standardization processes in the regulated telecommunications industry. Both domestic and global telecommunications standardization is often far slower (often by five years or more) than standardization efforts in the personal computer, software, business networking, and Internet sectors. Standards in these competitive industries are often developed by consortia and voluntarily adopted by an entire industry, or in other cases de facto standards are developed by individual firms and established through competition. But in either case, the process is fast, generally market-sensitive, and permits rapid, continuous evolution of standards as technology improves. This is an essential characteristic of standards in rapidly changing industries, because it is impossible to specify how technology and demand will look a decade in the future. Decade-long standardization efforts are an affordable luxury only when technological change is glacially slow. In modern information technology, slow and/or long-term standardization efforts which try to anticipate all future possibilities in advance are doomed to fail. Such standards are excessively complex and obsolete by the time they are implemented. This problem, too, was a characteristic and extremely debilitating weakness of the old IBM, and one the LECs seem to share. In fairness to the LECs, part of the problem is beyond their control. Standardization in the public telecommunications industry continues to involve governments and multiple, often international committee approvals, and is further slowed because many of the operating organizations or committee members are governments and/or PTT monopolies. The Europeans and Japanese often worsen the problem, as do organizations such as the ISO and ITU. These entrenched practices are contrary to virtually every trend in information technology. However, the LECs and Bellcore do have substantial control over U.S. standards, and they have not adequately responded to change. There is no excuse for their slowness in using the Internet or offering Internet service. While they are not alone, and telecommunications ministries and PTTs are not the world's most flexible, responsive industrial organizations, there this little reason for the United States industry (or for U.S. government policy) to perpetuate this behavior. 3.4 LEC Capital Investment and Financial Priorities Examination of LEC financial and investment behavior suggests strongly that they have been treating their core telecommunications networks (in both voice and data) as cash cows.49 They invest less in them than a decade ago as a fraction of revenues, cash flow, profits, and total capital stock. Their margins have increased, but instead of investing to modernize their networks, the LECs have used the cash thereby generated for diversification, dividend payments, stock buybacks, debt retirement, and/or retained earnings. This is surely one reason that the LECs' rates of technology improvement are consequently inferior to competitive high technology firms and/or the internal networks run by large information-based organizations. While the LECs are inherently capital intensive businesses and their capital investment remains high by the standards of most businesses (including many high technology businesses), the trend is clearly in the wrong direction. Total capital investment is flat, and the proportion of capital investment directed at the telephone network is declining. Furthermore, network usage is growing. On average, in the 1990s the RBOCs and GTE have been growing roughly 5% per year in revenues and access lines served; cash flow and profits have grown faster. Yet the LECs' capital expenditures on their core telephony businesses are at best flat and often declining. Since new access lines are being added at a reasonably high rate, and the physical laying of these additional lines constitutes a substantial part of the LECs' reported capital investment, real capital investment supporting each line is actually decreasing. Capital investment relative to usage, measured for example by access minutes and data traffic levels, is therefore decreasing substantially. And the LECs do have the money. Their cash flows, net income, and dividend payments have increased substantially throughout the 1990s. An increasing fraction of total LEC capital expenditures and cash flow are being devoted to diversification, primarily by acquisition or financial investment rather than real system development. The LECs are all investing heavily in cellular, paging, and cable TV systems, and/or buying their way into foreign markets. For example total capital expenditures by Ameritech, considered one of the more progressive RBOCs, were flat at $2.2 billion between 1990 through 1996. But Ameritech's annual capital expenditures on its own telephone network actually declined from $1.9 billion in 1990 to $1.6 billion in 1995, rising again in 1996 but still remaining below the 1990 level. The remainder went to investments in cable TV and cellular systems. Even U.S. West, the LEC widely agreed to have the most backward network and also the LEC with the most aggressive capital spending in recent years, has increased its capital investment in its core telecommunications network less than 8% per year over the last 5 years. Since U.S. West's access line growth has averaged over 4% per year, its capital spending on new technology has actually been rather modest. Similarly, NYNEX's total capital expenditures have increased modestly, e.g. from $2.7 billion in 1993 to $3.2 billion in 1995. However, NYNEX's capital expenditures on its local network have declined. The difference has been invested in cable systems in England and other diversifications, including a $1.2 billion investment in Viacom. Given that NYNEX's access lines grew 3.4% and access usage grew 8.6% in 1995, this is hardly impressive. Bell Atlantic, supposedly the most technically progressive of the RBOCs, shows a surprisingly similar or even worse picture, with flat or declining capital spending in the face of increasing demand. This is interesting, given that Bell Atlantic's CEO, Ray Smith, has made numerous public statements concerning grand visions for advanced networks. In fact, some of Bell Atlantic's announced plans (e.g. an expensive proposed video-on-demand network) were abandoned early because they proved ill-advised, and actual investment in advanced technology has been low. Bell Atlantic's behavior is in some ways worse than some of the other LECs. For example Bell Atlantic's R&D spending is declining, capital investment is flat, but dividend payments are now over $1.2 billion and growing. If its recent behavior continues, Bell Atlantic's dividends will exceed its capital expenditures in another seven years. In fact over the past five years, Bell Atlantic's net income has increased 40% (from $1.34 billion in 1992 to $1.89 billion in 1996), and its cash flow increased substantially. Access lines increased from 18.2 million to 20.6 million during the same period. But additions to plant, property and equipment stayed almost exactly flat at about $2.5 billion. The LECs' capital investment behavior is shown graphically at the end of this document; all data come from LEC annual reports, 10Ks, and proxy statements. I show their capital spending levels in current dollars and relative to revenues, access lines, and access minutes. For four LECs, I was able to obtain specific data on their telephone network capital expenditures, and for these firms I show network capital expenditures relative to these variables. As the graphs show, network capital expenditures are a decreasing fraction of total capital investment, and have therefore been declining substantially relative to revenues, cash flow, and network traffic levels. As suggested above, LEC R&D spending follows a similar and even more disturbing pattern. LEC R&D expenditures are not only declining as a percentage of revenues, but are also among the lowest of any large U.S. industry - lower, even, than many low technology sectors. This holds even counting the LECs' contributions to Bellcore. The LECs' capital investment and R&D behavior is in contrast, once again, with the behavior of competitive and successful high technology firms, where high growth rates in R&D, capital investment, and technical change are understood to be absolutely critical. Conversely, superior high technology companies pay low or zero dividends, preferring to reinvest profits for future growth, whereas the LECs pay high dividends. For example Microsoft, a $10 billion software company, pays no dividends on its common stock but now spends more on R&D than all the LECs combined, and also more than Ameritech does on capital investment. Microsoft also recently announced that it would triple its basic research spending to $300 million per year. R&D spending by firms such as Intel, Microsoft, Cisco, Hewlett-Packard, DSC, and Motorola tends to average over 10% of revenues and to grow 20%-75% per year (i.e., roughly as fast as revenues). Capital spending varies widely among high technology companies (e.g. semiconductor companies are far more capital intensive than PC or software companies) but, as with R&D, capital investment universally at least keeps pace with revenue growth. For example Intel's capital spending has for many years averaged over 20% of total revenues, i.e. has grown over 30% per year. To take another example, between 1993 and 1995 Compaq's capital spending more than doubled (from $145 million to $391 million), NOT counting acquisitions or technology licensing, which were also substantial and growing. During the same two year period Compaq's R&D spending rose from $169 million to $511 million (which includes $241 million worth of purchased technology in 1995). Both R&D and capital spending grew faster than revenues, despite the fact that Compaq's revenues doubled in this period. Finally, it is worth commenting on what the LECs have invested in. Their diversification efforts have primarily been directed at entering other heavily regulated, highly concentrated, and/or comparatively low technology markets: cable television content and distribution, foreign telephone markets, cellular service, and paging. Conversely they have not invested in advanced information technology, even though they are now largely permitted to do so. Thus, Bell Atlantic nearly bought TCI, and invested both in New Zealand's national PTT and in a Mexican telephone company (which it has been forced to write down); U.S. West purchased Continental Cablevision for $5 billion, plus 25% of Time Warner; NYNEX purchased an equity interest in Viacom for $1.2 billion. The LECs have also invested large sums in a number of joint ventures with each other, such as BANM, Airtouch, PCS Primeco, and Americast. Moreover, the LECs' investments, particularly in conventional CATV, also say something about their technological vision and/or priorities. For a fraction of the price of their CATV acquisitions or of their annual cash flow, the RBOCs could have purchased outright the entire Internet industry, including all Internet access providers and software firms, as recently as last year.50 However, these investment patterns also raise another very interesting issue, to which I now turn, namely that of LEC strategic behavior, which is often highly cooperative. 3.5 LEC Strategic Behavior, Lobbying, and Funding of Policy Researchers The LECs display a consistent pattern of large-scale strategic behavior that reaches far beyond, and complements, their technological, pricing, and investment behavior. This pattern includes highly cooperative business relationships with each other; avoidance of competition with each other; huge and often cooperative investments in lobbying, political contributions, regulatory affairs, and litigation, coordination of public policy positions and statements; extremely heavy funding of academic researchers as policy analysts, consultants, and expert witnesses. For these and other reasons, the LECs turn out to have strong common interests with each other, and even in some areas with the long distance (IXC) and CATV industries. These patterns are so broad and consistent as to raise extremely serious questions concerning potential collusion and anticompetitive behavior among the LECs, and the absence of U.S. government response thereto. With minor exceptions, the LECs essentially do not compete with each other, even where logic suggests they would, while conversely they cooperate very closely and extensively through a wide array of joint ventures, industry associations, political activities, and investments. Despite some cracks in the LECs' unified front, many of these arrangements appear to be successful and to be continuing even in the supposedly competitive regime of the present. This suggests that real, technologically progressive competition in the local loop will probably not come from the LECs or even the IXCs, but rather from the Internet industry, CAPs, and/or new entrepreneurship. This situation also raises a question as to why the FCC and the Antitrust Division of the Department of Justice have been as passive as they have. These issues will be discussed in chapters 4 and 5. LEC Cooperative, Noncompetitive, and/or Collusive Behavior In general, the LECs have avoided competing with each other in their core monopoly markets, have jointly resisted initiatives to open the local loop to real competition, have acted cooperatively rather than competitively in emerging markets, have focused their new business efforts on highly regulated, concentrated markets, and are acting to increase the concentration of the U.S. industry by merging with each other among other means. They are also investing, sometimes individually and sometimes together, in several proprietary and/or monopolistic CATV and video entertainment ventures. The following is a summary of the LECs' behavior with respect to each other. 1. Mergers. Bell Atlantic is planning to merge with NYNEX, and SBC is planning to merge with Pacific Bell. The latter merger has already received initial approval from the Justice Department. This will reduce the number of major U.S. LECs from 8 to 6. In addition, all four of the companies planning to merge already have significant cooperative relationships and joint activities, as we shall see shortly. 2. Reciprocity in Non-Competition. First, the LECs have avoided competing with each other in local service. There are currently 8 major LECs and over 1,500 minor ones in the United States. For more than one year, i.e. since enactment of the 1996 Act, all of the LECs have been legally permitted to compete with each other in basic local telephone service, to purchase unbundled elements from each other and to collocate equipment in each others' central offices with which, by law, the incumbent must agree to interconnect. None of the LECs has even tried, despite the fact that most of the LECs are negotiating, or in some cases have reached arbitrated agreements, with dozens of other companies trying to enter local service (primarily via resale of LEC services as opposed to real competition). The absence of LEC attempts to compete with each other is particularly striking because they have collectively acted to block via litigation the pricing scheme specified by the FCC's Interconnection Order last year, alleging that FCC-formula prices for loops would be too low. If this were true, it would be in the interest of any single LEC to purchase inexpensive loops from the others and offer nationwide local service. None of them seem to have seen it that way, however.51 Furthermore, the LECs have avoided competing with each other directly in most of the other markets they have entered. They generally enter new markets either via joint ventures with other LECs, or in markets structured such that LECs do not directly compete with each other. In some other cases, they are beginning to enter markets with some degree of competition, but even these tend to be highly concentrated and/or regulated oligopolies, e.g. cellular and PCS services. Conversely in major new areas in which nationwide competition would seem logical, the LECs have avoided it. For example, most of the LECs have (belatedly) entered the market for Internet service (although Nynex still doesn't offer Internet access). However, instead of purchasing or building national networks, they have chosen to offer Internet service only within their geographical monopoly telephone service areas. The only exception to this is that in a few cases, CATV companies owned by LECs are beginning to offer Internet service to small numbers of customers; interestingly, these CATV vendors are generally not offering telephone service, and none are offering it on a large scale. The same holds, as we shall see, for the LECs' new efforts to enter both regular and cellular long distance services; their entry is confined to their own operating regions. Thus in most cases the effect of allowing the LECs to enter long distance markets will not be to increase the number of long distance competitors by 8 (or by 6, if pending mergers are completed), but rather only by one. Moreover, the LECs are lobbying to obtain the right to enter long distance markets by satisfying checklist requirements even in the absence of actual competition in their local markets. 3. Major Cooperative Relationships. The LECs have a wide array of major joint ventures and/or coinvestments with each other, and many of them also, incidentally, have such relationships with the wireless and CATV industries. These relationships have existed for many years, are already extensive, and continue to be formed. For example, Tele-TV is a joint venture formed in 1994 by Bell Atlantic, Nynex, and Pacific Telesis with the stated intention of providing interactive television services. After estimated investments of over $500 million, the venture is being abandoned as a failure. Americast is a newer joint venture of Ameritech, BellSouth, SBC, GTE, and The Walt Disney Company, whose goal is to provide interactive video services using proprietary Disney content as well as other content sources. Even more interesting, in some ways, is the structure of LEC cooperation in wireless markets. In 1995, Bell Atlantic and Nynex pooled their cellular operations and formed a joint venture, Bell Atlantic Nynex Mobile (BANM), to run them. Around the same time, Pacific Telesis spun off its cellular operations to form Airtouch, an independent company. However, Airtouch then pooled its cellular operations with those of U.S. West, with U.S. West acquiring a minority equity position and various governance rights in the combined company. Furthermore Airtouch has formed extensive business relationships with the other LECs, through which the LECs cooperate both with Airtouch and with each other.52 In 1995 Airtouch, U.S. West, Bell Atlantic, and Nynex formed Primeco Personal Communications, which spent over $2 billion to purchase large quantities of PCS spectrum through the FCC auction process. Simultaneously, the same four companies formed TomCom, a joint venture whose goal is creation of common standards and marketing strategies' for all four companies' cellular and PCS services. Furthermore, Airtouch and LECs sometimes coinvest. For example, Airtouch and Bell Atlantic are coinvestors in Omnitel, an Italian cellular provider. In addition, the LECs coinvest with each other and sometimes with CATV vendors in a variety of other ventures. For example, New Zealand privatized its national telephone system several years ago, forming Telecom Corporation of New Zealand, which is now essentially wholly owned by a consortium consisting of Ameritech, Bell Atlantic, TCI, and Time Warner. Similar structures exist in various joint ventures and equity investments in other overseas telephone and cable operations in Britain, Eastern Europe, and elsewhere; for example Britain's largest cable system, TeleWest, is a joint venture between U.S. West and TCI. Various LECs have also formed joint ventures for the collective purchasing of telecommunications equipment, for the creation of simplified and similar ISDN service packages, and other purposes. The LECs have also acquired, and/or have considered acquiring, and/or may in the future acquire, some of the largest CATV operators in the U.S. Two years ago Bell Atlantic nearly purchased TCI, the largest CATV producer. US West owns a substantial piece of Time-Warner, the second largest, and acquired Continental Cablevision, which is the third largest. Similarly, in 1993 NYNEX invested $1.2 billion in Viacom, which owns HBO and the Blockbuster video store chain. Given the LECs' cooperation with each other, their shared interests, and their strong incentives to avoid direct competition (whether from each other or any other source), one must ask how strongly CATV systems linked to one LEC will be inclined to compete with another LEC by entering markets for phone service or Internet access. In fact, domestic LEC-owned CATV companies have been extremely slow to offer even Internet access, and none offer voice service in any significant way. In contrast, at least two foreign CATV providers in which the LECs have invested are already providing voice telephone service in competition to the incumbent PTT. The reason for this difference appears to be that in foreign markets, the LECs are undercutting other people's monopolies, not their own, and in foreign markets the risks of retaliation and the rewards of cooperation are small. The LECs' avoidance of competition with each other, preference for monopoly, structural linkages, and cooperative behavior forms a pattern is so widespread and consistent that it suggests the serious possibility of major antitrust violations. LEC cooperative behavior and resistance to technologically progressive competition extends into several domains other than the marketplace. The LECs cooperate politically, in regulatory affairs, in lawsuits related to regulatory affairs and consumer rights activities, in antitrust litigation, and in purchasing equipment (including ISDN and ADSL equipment). Thus while no LEC has attempted to compete with any of the others by seeking to exploit interconnection rights under the 1996 law, most have joined the lawsuit initially filed by GTE seeking to overturn the FCC's proposed pricing policies in the Interconnection Order implementing that law. The LECs also continue to cooperate in lobbying with respect to the alleged Internet congestion and universal access issues being considered by the FCC, and in resisting attempts by ISPs and ESPs to acquire regulatory rights to unbundled network elements and collocation privileges. They also refer expert witnesses and consultants to each other for use in litigation and in state and FCC rate proceedings.53 In fact, the LECs' political activities and cooperation therein are one of their largest and most effective business investments. In fairness, there are some tensions, and even some embryonic potential competition, among the LECs. In regulatory strategy, the LECs have partially divergent interests in regard to access charge reform. Due to demographics and access charge formulas, LECs receive widely varying access charge payments, and are affected quite differently by alternative reform plans. There have also been a few signs of competition between LECs and the CATV industry, and even between the wireless and LEC industries. Airtouch and SBC are in a dispute concerning interconnection rights; and Americast (a joint venture of several LECs) is litigating with several CATV providers concerning transmission of cable channels. There is some possibility that there could arise indirect competition between the LECs themselves, because several LEC-owned CATV companies are beginning to provide Internet access (though not telephone service) in other LECs' territories. However, the level of this competition is trivial - probably a few thousand households in the entire country - and there is no evidence of competition between the LECs in their core businesses, despite opportunities which have in fact been exploited by other firms such as the major CAPs (MFS and Teleport). The overwhelming picture of LEC relationships is one of reciprocity, non-competition, and even major cooperation in both business and politics. LEC Political and Regulatory Activities: Their Core Competence The LECs have invested enormous resources in lobbying, singly and together. (The same is true of the IXCs, although at a slightly lesser scale.) The primary goals of this lobbying have been to obtain increasing pricing, financial, and operational freedom while maintaining their regulated monopoly status, to slow down rates of technical and market change, and generally to retain the comforts associated with old business models and a regulated utility mindset. In fact the LECs are probably more skilled at regulatory and political lobbying than they are at R&D or deploying new technology, and they spend huge sums at it. Over the past decade, furthermore, they have devoted massive resources including major litigation efforts to avoiding precisely the kind of open architecture industry that future digital services require. At the Federal level alone, the RBOCs maintain over 200 full time government relations employees in Washington DC; GTE has another 50.54 These numbers do not include the hundreds of persons employed by the LECs indirectly, through their law firms, lobbying firms, political consultants, industry associations, PACs, and public relations firms in Washington. In addition the LECs retain many economic and/or antitrust consulting firms, such as Charles River Associates in Cambridge and the Law and Economics Consulting Group in Berkeley, plus large numbers of academic policy analysts hired as consultants and for expert testimony in lawsuits and regulatory proceedings, and, of course, large numbers of former government officials. Amusingly, in 1994 and 1995 both Pacific Telesis and Sprint hired Webster Hubbell, the former Clinton Administration assistant attorney general now in prison, to lobby on opposite sides of the access charges issue; apparently Mr. Hubbell neglected to tell either firm that he had also been retained by the other.55 In addition, the LECs make part-time use of hundreds more LEC employees in lobbying, regulatory, public relations, and litigation efforts, including numerous visits to the FCC.56 The industry also contributes millions of dollars to industry associations, including the U.S. Telephone Association, and to a variety of PACs in every election cycle; several of the LECs have their own PACs. The LECs, sometimes through their charitable foundations, also subsidize academic policy work (see below).57 Finally, the LECs have roughly another hundred government relations employees operating at the state level, plus comparable use of law firms, lobbyists, etc. Indeed, the LECs' state-level political and regulatory spending is probably at least one-quarter the size of their efforts at the Federal level. It is impossible to accurately account for this activity, in part because both the FCC and state PUCs are not computerized and keep poor records, a point to which I shall return in discussing policy. However, a rough estimate would suggest that combining state and Federal lobbying, the LECs probably spend over $200 million annually, and perhaps as much as $500 million, on lobbying, political contributions, and regulatory efforts in all forms. That's more than GTE, a $20 billion company, spends on R&D. The LECs are powerful enough that senior FCC and White House officials have rejected internal FCC requests to subpoena LEC cost and profit information, on the grounds that such subpoenas would be politically unacceptable.58 The ability of the LECs to reduce the degree of scrutiny to which they are subjected, and to thereby constrain FCC action, of course worsens the policy analysis problem since it makes the regulators heavily dependent upon the LECs for data. The LECs have, not surprisingly, become highly skilled at this game, which is one reason they retain such large numbers of academic expert witnesses and consulting firms. The LECs have used a wide array of rate proceedings, FCC regulatory dockets, and lawsuits in order to restrict and delay the advent of an open systems industry. In the 1980s, the LECs litigated and filed FCC documents to prevent the CATV industry from using telephone poles for cable wiring. In the mid-1980s, the LECs sued successfully in D.C. district court to restrict collocation rights of competitive access providers such as MFS, arguing that collocation would be confiscatory and would compromise network security. These are arguments that AT&T used for decades prior to its divestiture to prevent competition in terminal equipment and long distance services. In the proceedings prior to the FCC interconnection order implementing the 1996 Act, the LECs argued against subloop unbundling for the same network security reasons, and prevailed - the FCC interconnection order states the LECs are not required to provide it. The LECs are now challenging the pricing mechanisms of the interconnection order in Federal court. Rough estimates would suggest that the LECs have probably spent over $100 million on anticompetitive litigation since the 1984 divestiture. LEC Funding of Academic Policy Research, Consulting, and Expert Witnesses Over the past decade, and paralleling the growth of corporate lobbying, PACs, and campaign spending, there has been enormous growth in corporate (and specifically LEC) funding of academic researchers and policy research, particularly economists. A very high fraction of eminent academic economists seem to be available for hire to support LEC regulatory claims. Furthermore, university regulations limiting consulting to 20% of faculty time and/or 1 day per week are now widely evaded and/or violated.59 Consulting rates have risen to $300-$1,500 per hour, professors often consult for up to 1,000 hours per year, and companies often pay for their research assistants and support staff in addition to consulting time. As a result, the size of consulting payments now often dwarfs these professors' academic salaries.60 Frequently these professors' policy-oriented research publications do not disclose these consulting relationships, even when they are relevant to the research in question. A number of prominent economists who publish research in the areas of regulatory, antitrust, industrial organization, and/or telecommunications policy now receive consulting incomes of half a million dollars per year or more from LECs and/or other telecommunications companies. This statement applies to former chief economists of the Antitrust Division of the Justice Department, former FCC economists, economists who have served as expert witnesses in major LEC regulatory proceedings, economists who have served as expert witnesses in telecommunications antitrust cases, and economists who have published major studies of telecommunications policy.61 Examples include Jerry Hausman at MIT, who has worked for Pacific Telesis; Peter Temin, who worked extensively for AT&T (and for whom this author worked for one summer as a graduate student); William Baumol at NYU, who appears to be willing to work as an expert witness for just about anybody; economists at Rand, Brookings, and other think tanks; economists employed by Charles River Associates (founded by MIT economic professors); and a number of professors affiliated with the Law and Economics Consulting Group (LECG), including one former DOJ chief antitrust economist and former NEC Director Laura Tyson, who recently both LECG and Ameritech's board of directors.62 All of them provide antitrust and/or regulatory policy consulting, and there are many others. It is extremely lucrative, particularly for economists specializing in regulation and/or who are former FCC, White House, or Department of Justice officials. The consulting version of William Baumol's CV contains a 50 page supplement solely devoted to his expert witness engagements, for which he charges over $1,000 per hour.63 In addition, the LECs and other telecommunications companies such as AT&T are heavy contributors to academic research projects on telecommunications policy at Columbia, Berkeley, Brookings, Michigan, MIT, Stanford, and other institutions. One telecommunications policy research project at UC Berkeley and Michigan is currently funded in its entirety by the Ameritech Foundation, the (supposedly) charitable foundation created and funded by Ameritech; the director of the program also consults for BellSouth on regulatory policy.64 This is not an isolated situation; many if not most of the principal researchers in these projects also receive large personal consulting incomes from LECs and/or other telecommunications companies.65 While many of these people are generally ethical, there can be little doubt that the LECs' spending on academic research and consulting have an effect on academic policy research in this area, and constitute a major conflict of interest problem. Few prominent economists have recently criticized LEC monopoly power or technological performance publicly. Nor have many advocated greater interconnection and collocation rights for new entrants, ISPs, and/or ESPs. There has been little complaint from academic economists on the peculiar absence of competition among LECs, their extensive cooperative relationships, or the desirability of stronger antitrust enforcement. Nor has there been much comment on the very low rates of price/performance improvement in LEC digital services, and apparently no academic economic study of the subject has ever been conducted.66 There certainly has been very little public complaint about the conflicts of interest posed by LEC funding of academic research and professors' consulting. Conversely a number of academic economists have made statements in favor of instituting access charges and/or usage-based pricing on ISPs, ESPs, and/or for Internet services generally. LEC Strategy in Local, Long Distance, and Internet Competition Under the 1996 Act Through a variety of regulatory, legal, and competitive means, the LECs are seeking to maximize their ability to enter long distance competition while minimizing the degree to which their businesses, both voice and data, are subjected to real local competition from technologically progressive entrants. To be sure, they have received some major assistance from the 1996 Act itself and especially from the FCC's interconnection order, which I discuss shortly. First, the LECs and/or their wireless affiliates are already entering cellular long distance, which is permitted by the 1996 Act. Ameritech already has 2 million cellular long distance customers.67 In addition, the LECs are trying to enter long distance markets by satisfying the FCC's so-called "checklist" requirements for entry, but without actually submitting themselves to major competition. BellSouth recently issued an extraordinary press release stating that in its view, the presence of real competition is not and should not be a requirement for its entry into the long distance business.68 BellSouth also stated that it connected its first cellular long distance call "within seconds" of the signature of the 1996 Act. Ameritech petitioned the FCC to enter terrestrial long distance but was apparently forced to withdraw as a result of erroneous claims; recently SBC has petitioned the FCC to offer long distance service in Oklahoma, and BellSouth has stated that it shortly intends to petition the FCC to offer long distance service in Georgia. Furthermore, the LECs have available to them a variety of (currently legally permissible) strategies for advantaging themselves relative to ISPs in providing Internet and other data services. First, as a consequence of an extraordinary situation resulting from the FCC Interconnection Order and discussed in chapter 4, ISPs do not currently have the right to purchase either unbundled loops or to collocate facilities in LEC offices. Even other telephone companies, who do have those rights, are restricted in their collocation rights. Furthermore, the LECs can legally discriminate against competing ISP traffic relative to their own. For example, LECs can use the new Nortel switch to offload Internet traffic when its destination is the LECs' own Internet service business, but are not obligated to use the same technology in handling traffic to other ISPs; at least one LEC, SBC, is apparently already behaving in this manner.69 To whatever extent that the Internet congestion problem discussed below is or becomes real, this would significantly advantage the LECs' own Internet access businesses. Indeed as mentioned above, under current FCC regulations, and to some extent because of the wording of and/or FCC interpretation of the 1996 Act (see chapter 4), ISPs and ESPs currently do not possess the interconnection and collocation rights which are technologically required for provision of optimal high speed digital services. In addition, and also as mentioned above, the LECs are challenging the pricing mechanisms of the FCC interconnection order through litigation. For a combination of these and other reasons, not a single company, either from the traditional telephone industry or the ISP sector, has yet obtained the right to offer either HDSL or Internet access service using unbundled loops or collocation rights under the 1996 Act. Under the current FCC rules, which I discuss below and argue to be severely dysfunctional, the only way for any firm - an ISP or even another telecommunications carrier (a LEC or IXC) to offer data services requiring collocation of networking equipment would be to partner with a LEC in order to negotiate loop access and collocation rights which they are not legally required to provide. Not surprisingly, no LEC has done this. 3.6 The LEC "Internet Congestion" Allegations Given the foregoing assessments, consider now an example of current LEC strategy, namely the supposed "Internet congestion" problem and related ISP access charges controversy. Beginning in late 1996, and supported now by a number of academic consultants, Bellcore and the LECs began to put forward the claim that growth in Internet usage, particularly by dialup users, was causing congestion in LEC networks that could threaten the quality of public voice service.70 The LECs argued that Internet usage places a substantial and unfair financial penalty upon them to handle this problem, and that the only reasonable solution was to establish charges that reflected actual burden placed upon the network. The LECs also argue that this problem is rapidly growing more serious, and should be corrected by levying access charges on ISPs and/or usage charges on Internet users. Although it is absolutely correct that in the long run Internet usage will become the dominant driver of telecommunications bandwidth demand, it turns out that the LECs' claims are generally incorrect. While deferring policy discussions until chapter 5, I will consider here the empirical issues: the quantity of congestion currently generated by the Internet, the Internet's contribution to whatever total network congestion does exist, the net financial costs and benefits of Internet usage to the LECs, the cost of addressing whatever problems do exist, and the potential future trajectory of these variables. I will also consider the question of LEC strategy in putting forth these claims and policy proposals. The LECs' claims have focused on traffic generated by residential and SOHO users dialing in to ISPs, using analog modems over the switched network. Their claim is that this generates congestion in central office switches, particularly those which serve the local POPs of the ISPs, because Internet calls last far longer than voice calls, generating disproportionate switch usage. To be sure, Internet usage is growing extremely rapidly. Currently about 15% of U.S. households use the Internet, although fewer households are regular or intensive users. Surveys indicate that Internet calls are several times longer on average than voice calls, and that average household Internet dialup usage is 10-15 hours per week, versus perhaps 3-5 hours per week for residential voice calls. These Internet calls are made on flat-rate local lines, so users have no incentive from telephone prices to restrict their usage. Upon first inspection, and given continued rapid growth in Internet usage, all this would seem to indicate that the Internet does indeed represent a substantial and rapidly growing burden on the LECs and their switching infrastructure. However, the reality is quite different, at least for now and probably for several years to come. In fact, Internet usage causes little current congestion; the local network congestion that does exist is probably more attributable to LEC inefficiency and/or underinvestment than Internet usage; and, perhaps most importantly and interestingly, on a net basis the growth of the Internet is not a financial drain on the LECs, but on the contrary is in fact enormously profitable for them, and will remain so for at least the immediate future. Recent LEC analyses (such as the Pacific Bell white paper posted on its web site) claiming to demonstrate losses from growing Internet use generally omit major revenue and profit categories. For example, the Pacific Bell analysis entirely omits business usage of ISDN, frame relay, and T1 services for Internet applications such as operating web sites and corporate Internet gateways. These Internet-derived business revenues will probably exceed $5 billion nationwide in 1997, are highly profitable, and are growing more rapidly than total LEC costs or revenues (over 40% per year). Furthermore whatever Internet-caused congestion does exist could be relieved without great difficulty, at least for now. In fact, by using only a fraction of their Internet-derived profits, the LECs could greatly increase total switching capacity in the United States. In all probability, the LECs' attempt to levy access charges on ISPs derives not from current financial burdens, but rather from two other sources - the LECs' own aspirations in the ISP business, and the threat that ISPs and ESPs could pose to LEC businesses by offering competitive, modern technologies and services in an unbundled, competitive world. Internet and Non-Internet Contributions to Network Congestion Despite the undoubted fact that Internet usage will dominate telecommunications traffic over time, residential Internet usage is not yet a large-scale contributor to switching or other network congestion in the United States.71 First, network usage is uneven across time and users. Total network capacity is determined by peak usage of the entire network, of which residences are a minority part - business usage is greater and generates most peak usage periods (except for a few holidays, e.g. Christmas). Peak usage generally occurs during the day and is dominated by business-to-business and business-to-residence calls, not by residence to residence calls. In these peak usage periods, Internet usage still plays a minority role and does not constitute a net financial or technical drag. There are several reasons for this, including the fact that most business connections to the Internet use high speed digital and/or dedicated (non-switched) lines. Residential Internet usage peaks in the evening and on weekends, and for now remains well below business-dominated daytime usage. Pacific Bell may be an exception as a result of the concentration of high technology industry in California; but this also generates higher revenues, as we shall see. There are some areas in which network congestion does seem to be a problem - I have experienced it myself in Washington DC, and friends of mine report occasionally quite serious problems in the Pacific Northwest, particularly around Seattle. These problems occur during the day, not the evening, and are probably not driven primarily by Internet use, but rather by inadequate capacity, outdated equipment, and/or inefficient management of switching capacity. Simple underinvestment is probably the largest single factor; as we have seen, the LECs have been reducing their R&D and capital investment as a percentage of revenues and/or network demand for the past five years. Furthermore, the growth of Internet usage over the past four years has hardly been a secret. Every newspaper in the United States has been discussing it, and there have for several years existed publicly available statistics on Internet usage. In fact, the LECs themselves have for several years spoken publicly, and favorably, of the rising use of second lines for data purposes including fax machines and online services.72 Yet, as we have noted, the LECs generally declined to increase their capital investment levels, and even sometimes reduced them. With this in mind, consider now the economics of Internet growth from the LECs' point of view. The Financial Impact of Internet Growth Upon the LECs In their congestion allegations and policy proposals, the LECs have focused on flat rate residential POTS pricing as contrasted with the supposedly rapidly growing burdens on their switching fabric. However, residential POTS is only one of three sources of additional LEC revenues derived from Internet usage. The other two are business lines and data services, particularly ISDN and T1. The LECs don't discuss them very much for the simple reason that they are hugely profitable. Revenues from frame relay and even higher speed services such as T3 are also growing rapidly. Consequently, a fairly straightforward calculation suggests that far from being an increasing burden, total Internet-related growth in telecommunications services usage may now contribute up to a quarter of the LECs' net profits in their core network business. Consider revenues. First, if we assume that two thirds of home Internet usage generates a second line, and each such line generates two thirds as much revenue as a first line primarily used for voice, then Internet second lines have increased residential POTS revenues by about 7% over the last 3 years. Because Internet usage is growing rapidly, Internet second lines will generate another 5% or so increase in residential POTS revenues this year. Second, consider business lines. Small business users of the Internet purchase substantial numbers of "nth" lines in order to use the Internet, and business lines are substantially more expensive than residential lines. In addition, Internet access providers must purchase large numbers of business lines to receive the Internet calls placed by users, and the number of lines they must purchase is directly proportional to Internet usage levels. So although the business rate per line may be a flat rate, the total sum paid by ISPs and received by LECs is very directly proportional to usage levels. (So is the amount of equipment the ISPs must purchase, e.g. modems and routers.) Thus it is not the case, as the LECs claim, that they derive no revenues, and that ISPs pay no fees, related to real usage costs. And because residential Internet usage peaks sharply in the evening, ISPs must probably purchase many more lines per customer than most businesses. But third and finally, major business Internet usage, such as operating a heavily used web site, requires businesses to purchase ISDN lines, frame relay services, T1 lines, and in some cases even higher speed T3 lines. As we have seen, T1 usage is growing 40% per year; T3 usage is growing even faster, probably over 100% per year. In 1997 T1 and T3 will probably constitute a $10 billion business for the LECs. While there is no publicly available data on this issue, it is clear that Internet usage now accounts for a major fraction of all T1 lines, and is the major driver of T1 growth. Combining these three revenue sources - residential second lines, business lines, and data services including ISDN and T1 - it would appear by this year (1997) the LECs will derive over $5 billion in annual revenues directly from Internet use, that the LECs' Internet related revenues are probably increasing 50-75% per year, and that this revenue segment is highly profitable. Now consider the LECs' costs and their relation to the supposed Internet congestion problem, beginning with dialup usage. Until Internet usage begins to reach peak capacities in central offices, the LECs' incremental costs are very low; this appears largely still to be the case. When peaks are reached, the LECs will indeed need to add capacity to some extent. However, for several reasons this can be done fairly inexpensively. For example, Nortel now sells a switch that differentiates ISP calls from others and handles them outside the normal switching fabric. Internet peaks will not be marbled throughout the entire network, but rather are concentrated where ISPs have POPs, so additional capacity can be targeted fairly easily. Furthermore, most T1 lines are not switched at all; they are quite inexpensively dedicated from the customer to a single end-point - such as an ISP's office or an Internet NAP. Given the huge margins enjoyed by LECs in the T1 business, this must be counted as heavily profitable, with few incremental costs and no burden on the switching fabric. Thus for now, the LECs' Internet revenues are growing sharply and consist in the majority of high margin business and data services revenues. Incremental costs still remain fairly low. Long Run Issues and Solutions Obviously at some point, continued growth in Internet usage would indeed strain current peak capacity, begin to cause serious congestion, and require major capital investment to increase switching capacity. However, it is important to make three observations about this situation. First, the LECs' Internet revenues give them more than enough money to enable them to increase capacity greatly if required, at least for the next several years. As we have seen, on average the LECs' network capital investment is less than 20% of revenues and has been declining relative to cash flow and also relative to network usage as measured by access minutes and data traffic. Switching equipment is currently perhaps one third of capital investment. Thus switching investments represent less than 7% of LEC revenues. Furthermore, the price/performance of new switches is much greater than that of many obsolete switches still in use as a consequence of the LECs' low investment levels over the last five years. Combining these factors, it would appear that by increasing capital investment levels by 2% of revenues, the LECs could rapidly double the total switching capacity of the entire U.S. network. Since this investment would be necessary only under conditions of substantial revenue growth, it is quite possible that the LECs would continue to find Internet growth profitable even using conventional technology. Moreover, less expensive options than general capacity increases are available, because Internet-generated congestion is more focused, and Internet-specific solutions such as the Nortel switch and others are increasingly becoming available. Thus, the LECs could handle this problem for the next several years with a minority of their incremental Internet revenues and profits. The LECs' underinvestment over the past decade does not come from financial hardship, but from the fact that they have chosen instead to take higher current profits, pay high dividends, and/or to invest in cellular systems, foreign ventures, entertainment content, and CATV. Second, whatever Internet-driven problems do arise, now or in the future, may be the LECs' own fault. The LECs are not handling Internet growth, and/or growing demand for data services generally, in an efficient, responsive, or farsighted manner. They have been slow to embrace new technology (including the Internet itself), are widely viewed and reported to be stonewalling most attempts by ESPs and ISPs to interconnect with them, have not increased capital investment sufficiently, and are completely absent from Internet technology and standards development. LEC reaction to the Internet has been slow, anticompetitive, and reactive, despite the fact that the problems posed thusfar by the Internet for LECs are small compared to those faced by competitive high technology firms. In fact Internet service providers, semiconductor producers, network equipment vendors, and/or independent software firms routinely face far more sudden, powerful challenges and higher rates of change. They just respond differently - they increase R&D and capital spending rather than asking for regulatory relief - and if they fail to respond, competitors arise to fill the void. When the rise of the Internet and Netscape began to fundamentally threaten Microsoft's position in the software industry, it did not ask Washington for help; it doubled R&D spending within one year, tripled basic research, and started a ferocious competitive counterattack against Netscape. Netscape itself has grown over 500% per year, while firms such as Hewlett-Packard, Microsoft, and Intel (which are now larger than most LECs) routinely manage growth of 25-50% per year. The Preferred Alternative: Forward-Looking Behavior Third and finally, the proper long-term technological response to Internet growth is not to continue increasing conventional circuit-switching capacity, but rather to begin constructing a modern, open, primarily packet-switched local network alongside the existing voice network, and interconnected with it where appropriate. In particular, if they are concerned about network congestion, the LECs would logically want ISPs to collocate in LEC central offices in order to offload Internet traffic optimally. Much of the new packet-switched infrastructure, in fact, should be built and managed by ISPs, ESPs and/or direct competitors.73 The LECs could also offer innovative services that incent ISPs to interoperate more effectively with them. In fact, by using the Internet, and integrating Internet services with voice telephony, the LECs could in some cases actually reduce the burden on their current infrastructure as well as offer new services. For example voicemail and fax could be carried by the Internet rather than delivered via the conventional switched network.74 As the Internet grows, the LECs will become progressively less suited to handle it alone. Indeed one of the major advantages of decentralized, modular, open systems - from networks to industries - is that they relieve managerial bottlenecks by offloading management of growth to a decentralized industry requiring very little central coordination. Required coordination is achieved through technical standards, strategic alliances, and market processes, as occurs in the personal computer industry, rather than centralized control. Despite the fact that no one firm controls the PC industry at the system level, the personal computer industry still displays excellent compatibility characteristics and high rates of growth and technical progress. This also permits many niche suppliers to serve specialized markets. However, the LECs show few signs of embracing this model, and in fact specialized high end users such as the film industry have begun to complain that the services they need are simply unavailable.75 The growth in LEC Internet revenues and profits raises the question of why the LECs have made the congestion allegations and policy proposals that they have. There appear to be several reasons. First, they are in the Internet access business, and it would be convenient to impose differential costs on their more successful rivals. Second, the LECs surely fear even more serious competition from low cost data and voice services delivered by ESPs and/or over the Internet, which would cannibalize their highest-margin (and currently still monopoly) businesses. Third, the LECs are faced with the prospect of declining access charge revenue through pending FCC policy decisions, and ISP access charges could replace this lost revenue. Indeed if Internet traffic comes to dominate total bandwidth, as it probably will over the next two decades, even modest access charges levied on ISPs would bring very substantial revenue. And fourth, the LECs may indeed fear that as a consequence of their inefficiency and inability to manage rapid growth in competitive systems environments, they themselves could cause a serious congestion problem unless Internet growth rates are reduced. 3.7 LEC Governance and Management Familiar Warning Signs The LECs' top management, personnel policies, and governance are distressingly reminiscent of the automobile, steel, mainframe computer, and other mature industries during the 1970s and 1980s. During this period, the Fortune 500 lost over 2 million net jobs, and a number of the largest firms in the United States experienced severe declines in market share, profits, and valuation. The poor governance of firms such as IBM and General Motors seems to have been implicated in their long term decline, which generally persisted for long periods of time without sanctions or removal of responsible top management. By the 1980s, the dominant firms in several large U.S. industries were widely considered to be poorly managed, with highly politicized (in the sense of corporate politics) management and complacent boards. Top managers in these firms were generally lifetime employees promoted through marketing, sales, and finance. Few had technical training, international backgrounds, and/or outside experience, even in high technology industries such as computers. Often these firms, like the LECs, devoted huge resources to resisting the advent of real competition and/or technical progress, e.g. via lobbying for import restrictions. These firms' boards were generally composed of insiders, retired insiders, retired CEOs, former high government officials, presidents of local universities, Washington lawyers, lobbyists, CEOs of large local customers, etc. Outside directors generally knew little about the business and/or the technology underlying it. They were often in some way beholden to the firm and/or CEO, e.g. through compensation, pension plans, social relationships, common memberships on other boards, and/or interlocking board memberships (such as existed between DEC and Ford). Some directors were on many boards simultaneously, and often these directors regularly failed to attend meetings. Compensation of top managers and directors was often not strongly linked to the long term success of the firm. Poor CEO performance was not penalized, CEOs were rarely fired, and golden parachutes and generous pensions were commonly awarded even in the wake of disastrous performance, as for example with W. Michael Blumenthal of Unisys. When time finally ran out and crisis arrived, the top managements of auto, steel, and mainframe computer firms then imposed many of the costs of their own failure on lower level employees through huge layoffs and wage cuts.76 Shareholders often suffered as well.77 Yet management was rarely disciplined by their boards until an obvious, major crisis appeared,78 and the boards certainly didn't discipline themselves. Share ownership was generally too diffuse to permit a serious challenge to board membership, although the rise of large mutual funds is beginning to change this to some extent. Histories of lax governance and dysfunctional incentives thus often end badly, particularly in the information technology sector, which displays a striking pattern of generational succession. The companies and managers that lead one generation of technology, products, and/or architectures are often unable to retain leadership positions when new generations of technology arise. IBM was a member of the elite group of exceptional, self-renewing firms for over 20 years, then deteriorated, and has now partially recovered. IBM's board was passive and essentially beholden to management; it included such persons as the president of Occidental College, the then-president of Wellesley, Nan Keohane, and Cyrus Vance. Only one IBM outside director, Harold Brown, knew anything about technology. When IBM's major crisis finally came in 1992-3, IBM lost $25 billion virtually overnight. The long-inert board was finally force to replace the CEO. The new CEO replaced much of IBM's top management, and then laid off over 200,000 employees - half its workforce - within three years, reducing costs by almost $10 billion per year. Only a relatively few truly excellent companies have been able to avoid this fate, and to remain industry leaders across multiple generations of technology, products, and management. Current examples of successful and well-governed high technology firms include Hewlett-Packard, Motorola, Microsoft, Oracle, Compaq, Intel, and Cisco. These firms seem to share certain characteristics. They have fairly young, technically knowledgeable top management; decentralized structures with highly accountable business unit managers; flexible personnel and M&A policies allowing for acquisition of new technologies and skills through outside investment; rapid promotion based on merit; and broad-based financial incentives for their workforces, linked both to individual performance and long term corporate growth. A number of top managers and/or board members have large stock positions, and ESOPs and stock option plans are widespread. Microsoft, for example, offers stock options to all employees, over 4,000 of whom have thus become millionaires. Intel and Hewlett-Packard also have broad stock option plans and profit sharing plans that include all full time employees and involve large sums of money. Inevitably, however, it is also sometimes critical - as it has been with Compaq, IBM, Unisys, DEC, Apple, and others - to remove a CEO. In this event it is vital to have a board able to move forcefully and competently to replace a CEO or founder before a problem becomes unmanageable, and to choose a qualified replacement. In such cases, it is important to have outside directors who are both competent and strongly incented to care about the long term success of the firm. This was the case at Compaq, whose chairman was Ben Rosen, a venture capitalist with large Compaq holdings. In fact, the boards of most high technology startups consist of the founders, major shareholders, plus one or two senior industry executives or academic specialists. Boards of young high technology companies are therefore generally both fairly knowledgeable and financially incented to improve the performance of the firm. However, the quality of governance in U.S. high technology companies often declines after they have been successful and publicly traded for a substantial number of years, original founders have retired, and so forth. It is generally at this point that long term problems begin to arise, and that generate major crises later, upon the arrival of a new technology or competitor. Absent effective board reaction to organizational decline, a firm's institutional capital can be permanently damaged, leading to its being marginalized or even bankrupted by a new generation of technologies and companies, as has occurred with Apple, DEC, Unisys, Borland, Lotus, WordPerfect, Novell, and many other companies (even IBM). Thus, the general record suggests, first, that competition and new entry are extremely important to the information technology sector's long term growth and global competitiveness; and second, that good corporate governance is critical to the ability of a major company to remain successful across multiple technology generations. Without competition and new entry, the importance of governance increases further. Once again, the RBOCs and the CATV industry are nearly the opposite of the behavior exhibited by successful high technology firms. LEC Top Management and Corporate Governance I have examined the top managements and boards of most of the LECs and the largest CATV vendors.79 Their executives and directors are far older than those in competitive high technology, they have much less technical training, their financial stakes in the success of their company are much smaller. The financial risks they bear are generally smaller as well, as a result of high salaries, cash bonuses, stock options as opposed to ownership, and liberal pensions and severance agreements. They tend to have incentive structures such as salaries and pensions that are structured to reward longevity relative to innovation or growth, although several LECs are improving in this regard. Where used, which is very little by high technology standards, stock options are reserved for senior management and granted on favorable terms rather than being used to incent long term performance, and/or made broadly available for employee motivation. Most RBOC executives have never worked anywhere except the old AT&T and their current RBOC successor firm. It appears that no CEO has ever been fired by a LEC board. Consider management first. RBOC top managements are uniformly dominated by a rather old cohort composed entirely of lifetime employees who have spent 25 to 35 years inside the old AT&T and then the successor RBOC, usually with no other professional experience. Very few have any technical training, and where they do it is obsolete (because of their age). Nynex and Bell Atlantic are very partial exceptions. Although they share the usual AT&T lifetime employment pattern, the CEO of Bell Atlantic, Ray Smith, and several Nynex officers, including its CEO Ivan Seidenberg, do at least have technical bachelor's degrees. However, the most recent was in the late 1960s, none of these firms' top managers have advanced degrees in any technical field, no NYNEX or Bell Atlantic officer has ever worked in a fast-growing high technology firm, and none has ever run a major R&D organization.80 The top managements of LECs are heavily weighted towards law, public relations, government relations, and finance as opposed to research, development, or high technology. Unlike nearly all unregulated information technology firms, few LECs list a chief technology officer, VP of Engineering, and/or VP of R&D among the corporate officers and/or senior executive team. Conversely all LECs list a VP of Public Relations and a VP of Government Affairs or Government Relations as officers, and often have several other officers in related functions, such as law, regulatory affairs, communications, and public policy. Research and technology usually have zero representatives among top management, and the experience of the highest ranking RBOC technologists tends to be quite weak. For example, Bell Atlantic lists about 40 senior executives, with biographies, on its web site (www.ba.com). None is responsible for research, R&D, or technology at the corporate level. The sole senior executive with any corporate technology responsibilities is John Seazholtz, Chief Technology Officer of Bell Atlantic Network Services. He has a technical bachelor's degree dating from the 1950s, no advanced degree, and has never been employed anywhere except AT&T and Bell Atlantic. Most of the RBOCs are even worse. For example, none of the officers of Ameritech has a technical degree of any sort.81 LEC executives are also far older than their counterparts in competitive high technology, and have much less experience with modern information technology, research environments, and/or open systems technologies. On average, LEC senior executives are about 10-20 years older than the senior management teams of firms such as Microsoft, Oracle, Netscape, Cisco, Compaq, or the Internet services industry. No officer of any LEC has ever held a research position; and with one possible exception, none has been responsible for developing any major new technology; none has ever managed a high growth, high technology business even in their own firm. The one possible exception is Dave Dornan, President of Pacific Bell, the network subsidiary of Pacific Telesis; his previous experience was in traditional (pre-Internet) data communications. In contrast the founders and management teams of successful high technology firms tend to be highly technical, well educated, and also comparatively young. They often contain persons with PhDs, sometimes previously distinguished faculty members of research universities. Jim Clark taught at Stanford before founding Silicon Graphics and then Netscape, which he co-founded with 23-year old graduate student Marc Andreeson. Charlie Bass taught computer science at Berkeley before founding Ungermann-Bass; the four founders of Intel, two of whom have also been CEO, all had PhDs in physics from MIT; Bill Joy, one of the four founders of Sun Microsystems, developed much of Berkeley UNIX himself; Cisco was founded in 1984 by Stanford professors; 3Com was founded by Bob Metcalfe, who invented Ethernet while a researcher at Xerox PARC; and Bill Gates, although he never finished college, was famous as a hacker even in high school and developed some of Microsoft's early products himself. Thus even when high technology founders and CEOs have little formal training, they often had highly technical work in their past, and are often heavily involved in developing products. Moreover, these firms emphasize technical skills, roles, and experience in top management. About half of Microsoft's Office of the President have technical backgrounds, including technical PhDs. All of these firms have a VP for research, Chief Technical Officer, Chief Architect, Chief Scientist, Chief Engineer, and/or VP of Engineering at top levels of management and reporting directly to the CEO. In many cases even operational and marketing executives have technical backgrounds. Often there are several technical executives and/or university engineering professors on the board of directors. Compensation in high technology firms, while neither uniform nor without abuses, tends to be far more consistently based on incentives and on long term corporate performance than is the case with the LECs. Cash salaries tend to be fairly low, while stock options and stock ownership tend to be large. Executives, CEOs, and board members tend to have major financial interests in the firm; directors' pensions based on length of service are extremely rare. In fact, many high technology firms offer no pensions even to their employees. The LECs, however, are different. LEC Boards of Directors LEC boards exhibit several patterns similar to firms such as GM or mainframe vendors in decline. They also exhibit some of the same problems as LEC top management, such as lack of modern technical expertise. There are many insiders, past and present; many retired executives; there are interlocking directorships and common memberships on third boards; compensation is not strongly tied to long term performance; all LECs offer directors' pensions based on longevity of service; and the outsiders generally don't know anything about modern networking, computers, high technology, or even traditional local telecommunications. Where there are any high technology representatives on a LEC board there is only one or two, and they are usually from a large, politically powerful local employer, customer, and/or supplier. For example the sole high technology representatives on the Pacific Telesis board are the CEOs of Hewlett-Packard and Apple. Both are California firms, and although HP is a superb firm, it is also a major supplier of large systems to RBOCs. Bell Atlantic does have one good director from the PC industry, Eckhard Pfeiffer of Compaq; but otherwise its board is the usual list of insiders (4), retired CEOs, university presidents, lawyers, etc. Furthermore, the PacTel and Bell Atlantic boards, though hardly ideal, are unusually good in this regard. Most LEC boards do not contain any high technology, networking, software, or computer representation at all. They are more likely to contain politically influential lawyers or former government officials. One LEC director is Frank Carlucci, a former secretary of defense who apparently has never seen a board he doesn't like; he sits on two dozen of them, while also running his own merchant bank. Ameritech's newest director is Laura Tyson, former chairwoman of the Council of Economic Advisors and director of the National Economic Council in the first Clinton Administration. She is also affiliated with the Law and Economics Consulting Group, a corporate antitrust consulting firm founded by law and economics professors, 25% of whose revenues derive from the LECs. Similarly, the NYNEX board includes nobody with a high technology background. There are two insiders and two interlocking board memberships. NYNEX's CEO is on the board and compensation committees of the Melville Corporation, whose CEO is on the NYNEX board; and a similar relationship exists with Chemical Banking. There are also several shared directorships, where multiple NYNEX directors sit together on other boards including Allied-Signal, Metropolitan Life, Chemical Banking, and Viacom. In the cases of Allied-Signal and Melville, they sit alongside with the NYNEX CEO. Thus any board attempt to sanction or remove a CEO would make for a very messy social life, and could put other directorships at risk. Similarly, the outside directors of SouthWestern Bell include one person with any high technology experience at all - Admiral Bobby Inman. Conversely there are four directors from financial services firms, two from energy, a professor of history, the CEO of Emerson Electric, the CEO of Anheuser-Busch (August Busch), a perfume company CEO, and a Mexican billionaire. All are from SWBT's region. Similarly, GTE's board contains 4 current or former insiders, no representatives from high technology, and various present or retired large company CEOs, bankers, and lawyers. No LEC board contains anyone from the software, Internet, or consumer electronics industries. Where LEC boards contain university officials, they are innocuous and have no background in telecommunications or information technology. Nor have I been able to find any LEC outside director with a major stock position in the firm, comparable to those held by high technology venture capitalists or strategic investors. By contrast, Cisco's directors include Edward Kozel, the Chief Technical Officer; John Gibbons, professor of electrical engineering and dean of the school of engineering at Stanford; Robert Puette, CEO of Netframe, a server company; Don Valentine, a prominent high technology venture capitalist and investor in Cisco; Carol Bartz, CEO of Autodesk; Masayoshi Son, president of Softbank; and Steven West, president of Hitachi Data Systems. Sun Microsystems' board of directors includes John Doerr of Kleiner Perkins, the principal venture investor in Sun; Michael Spence, dean of Stanford Business School; Judith Estrin, a prominent high technology entrepreneur and currently CEO of Precept, her latest company; and Ken Oshman, founder of ROLM and currently CEO of Echelon, a networking company. LEC Executive and Director Compensation Financial compensation and incentives are also different from successful information technology companies. In general LEC directors and officers have fairly high effective cash salaries small, low-risk stakeholdings. In no LEC do all of the directors and officers combined hold even 1% of the stock of the company. In the case of Nynex, for example, the total combined percentage of director and officer ownership is 0.3%. Nynex directors receive $30,000 salaries, $1,500 board and committee meeting fees, and substantial lifetime cash pensions that vest over ten years of board service. Conversely they receive only 100 shares of stock (although some have purchased modest additional amounts). Thus their incentives are slanted towards staying on the board rather than ensuring the company's future success. This is in fact the general pattern. All LECs offer large lifetime pensions to CEOs and directors. In all of the LECs, the value of directors' pensions is indexed solely to length of service as a director. This is hardly an incentive to ask hard questions at board meetings. Such arrangements have come under increasing criticism from the corporate governance movement and from organizations such as the Council for Institutional Investors. Several LEC shareholders have introduced shareholder resolutions intended to shift CEO and director compensation towards stock and away from longevity; in every case LEC management has recommended against them and they have been voted down.82 NYNEX has increased the use of stock options for executives, but not for directors. Moreover, many of the executive options are structured as low-risk options that vest over two years, an exceptionally short period, yet can be exercised at any time over the following ten years, a very long period. By contrast in most high technology firms options vest over four or five years, but can be exercised only while the option holder is still an employee of the firm. Furthermore, in most well-run high technology firms, even large ones such as Hewlett-Packard or Microsoft, every employee receives stock options and/or profit sharing opportunities. For example, every newly hired Microsoft employee generally receives at least 1,000 stock options vesting over 4.5 years. Conversely, LEC stock options are largely restricted to top management. In short, the top management, boards, and management practices of the LECs may have been suitable for a bygone era of stable regulated monopoly, but are likely to cause problems in the future. Firms with such compensation structures, executive teams, and boards will probably prove far more adept at regulatory and political rearguard actions as opposed to investment in modern technology and adjustment to open systems competition. This is likely to continue to impede technical change and competitive behavior, and may contribute to serious organizational and financial problems once the effects of competition and rapid change begin to manifest themselves. This in turn could generate major future problems for LEC shareholders, LEC employees, and perhaps the American economy. The result could be a painful contraction of the LECs later in the decade, and possibly problems with the operation of the national voice network as a result of LEC inability to adapt to high technology realities. This would seem to suggest that the application of pressure now for improved governance and management would be desirable. 3.8 Another Sobering Benchmark: AT&T Since Divestiture The generally very poor performance of AT&T since divestiture, despite its freedom and extraordinary technical assets, provides another benchmark for the RBOCs - not because AT&T is in the same businesses or regulatory position, but rather because until a decade ago it shared the same management, skills, history, and culture (and in many cases still does). AT&T's performance over the past decade thus sheds some light on the quality of RBOC management and how the RBOCs might be expected to fare under conditions of similarly increased competition. When the "new" AT&T was created in 1984, it was composed of the technically strongest portions of the Bell system, including Bell Labs, Western Electric, and AT&T Long Lines. It also, however, had basically the same management and culture as the RBOCs, since they were once the same company. Recently, AT&T hired its first outside CEO after having divested itself both of Western Electric (now Lucent) and NCR, the computer vendor it had acquired several years previously. Since 1984, AT&T's overall financial performance has been among the poorest in information technology and even in American business generally (including regulated telecommunications) by virtually every measure - revenue growth, market share, earnings, market capitalization. Over the past five years, AT&T's earnings per share have declined at an average compound rate of 3% per year. The company's strategic and operational performance has been similarly uninspiring. Its core long distance business has lost market share dramatically, now holdingly only about 50% of the U.S. market. It has made several enormous strategic errors costing billions of dollars each, in some cases errors highly similar to those made by the LECs. Both AT&T and the LECs, for example, ignored opportunities in the Internet while making huge acquisitions of companies in mature businesses. In fairness, AT&T initially faced a difficult situation as a result of divestiture, and has been reasonably effective in lowering costs and prices for traditional, domestic long distance service. But AT&T has also been very late to enter new markets, and it has blundered stupendously in information technology, particularly in the computer, software, and online services industries. It lost several billion dollars on internal efforts to develop a computer business, then lost further billions following a hostile acquisition of NCR, which AT&T later divested for $3 billion less than the original purchase price. Despite the fact that by the mid-1980s UNIX (an AT&T invention) had become the standard operating system used by most midrange and high performance computers including the product lines of Sun, DEC, and Hewlett-Packard, AT&T consistently lost money on UNIX and mismanaged it badly. AT&T failed to develop a unitary UNIX standard, and eventually sold it for a trivial sum. AT&T also invested and lost hundreds of millions of dollars in worthless online services and game companies, including several non-Internet companies that it acquired even after the rise of the Internet had obviously made them obsolete. It was also quite late to offer Internet access service. AT&T may also have governance problems. (Amusingly, I was unable to obtain governance information from AT&T's web site because it is severely out of date. There are two separate pages purporting to list AT&T's board, both of which are inaccurate and which are inconsistent with each other. The older of the two lists Louis Gerstner as CEO of RJR Nabisco, and contains a link to AT&T's 1991 Annual Report; the newer one still lists Robert Allen as CEO of AT&T. Not so good for an ambitious Internet company. The "newer" board listing contains no high technology members.) Recently AT&T paid its retiring CEO, Robert Allen, huge bonuses (tens of millions of dollars) at the same time as it has been announcing poor financial performance and large layoffs. AT&T has also lost large amounts of managerial and technical talent to rivals and startups. AT&T's board did not, apparently, see fit to intervene in this situation, although upon Allen's retirement it did hire a CEO from outside the company. It should be noted that AT&T is now moving, albeit slowly, to compete with the LECs by re-entering the local telephone market. It initially announced the intention of offering local service in all 50 states, and entered negotiations with the LECs, but then scaled back its ambitions. The LECs have been unable to reach agreement with AT&T concerning pricing and service offerings, hardly a surprise, and as a result AT&T and the LECs are now in arbitration proceedings, as specified by the 1996 Act, in 30 states. Several lawsuits have also been filed. However, for reasons discussed in section 4, this does not guarantee that the LECs will face serious competition from AT&T in most markets and/or in advanced digital services, or that such competition would result in rapid improvement in price-performance. AT&T's record suggests how inefficient it had become prior to divestiture, and how inefficient the RBOCs may still be now. AT&T had largely the same management pool that is still running the RBOCs. Furthermore the technological and competitive change generated by the Internet revolution will be far more intense than competition in conventional long distance service over the past decade. AT&T's experience thus does not bode well for the LECs' ability or desire to compete through technological vision and rapid price-performance improvements. Rather, the AT&T experience once again indicates what can happen to shareholders and employees when a monopolist goes too long without being subjected to competition and/or managerial reform. 3.9 The CATV Industry: No Savior, and Often Similar Analysts of telecommunications and/or Internet markets often suggest that the CATV industry will discipline the LECs' behavior and/or take up their slack by providing advanced digital services. For a number of powerful reasons, it is unlikely that this will occur on a large scale soon, say within the next five or even ten years. Indeed, the structure, behavior and current problems of the CATV industry at least equal and often parallel those of the LECs, while adding some unique problems as well. First, CATV systems in general pass only residences, not businesses, and therefore cannot offer business Internet service. And in fact, even residential coverage is incomplete; less than 70% of U.S. households subscribe to cable, versus 90-95% penetration for telephone service. With the advent of digital satellite broadcasting and over-the-air digital HDTV in 1998, residential CATV penetration may even decline. Second, the architectures and technology of U.S. CATV systems are not currently well suited for high speed two way digital services. In this respect, U.S. CATV systems are inferior to those of some other nations, including Canada and some European nations. Hence large investments would be required - to upgrade U.S. systems, and to distribute cable modems and other PC peripheral equipment not currently in wide use. Even if the CATV industry had the money and the desire, upgrading these systems would take several years. Furthermore, almost all CATV industry plans involve asymmetric service, with higher bandwidth downstream than upstream. This makes CATV Internet service inappropriate for business use, even if businesses somehow became connected to the CATV network. In fact, some CATV plans involve using CATV infrastructure only for downstream delivery, with upstream signals carried by the LECs, which once again will require improved LEC digital infrastructure. Third, the CATV industry's structure and financial incentives are by no means completely aligned with the deployment of high bandwidth open systems, which could dramatically reduce barriers to entry in the creation and distribution of video entertainment content over the next decade. While low speed Internet access does not threaten the LECs, high speed Internet access with real time quality of service - i.e. suitable for carrying video - threatens them profoundly, for reasons described shortly. Fourth, many of the independent CATV firms simply do not have the money, for a combination of several reasons including high debt loads, competitive pressure, and possibly poor management. For example TCI, often considered the most technically sophisticated, is now in considerable financial trouble, and has pulled back from many of its announced technology modernization plans. TCI's market capitalization has remained flat for the last five years, it recently announced that it would lay off 6% of its total workforce, and the company has a very high debt load. Furthermore, the initial investment required to modernize a network for Internet service is fairly high, but would be shared across only 10-20% of CATV subscribers. This is due to the fact that for the next several years less than half of U.S. households will be Internet users, and the CATV industry will be competing with the LECs and ISPs who currently dominate the residential market. Most CATV vendors also, apparently, have little expertise with advanced digital networks. For monopoly U.S. CATV firms whose debt service sometimes exceeds 25% of revenues, and whose technological expertise is limited, entering the Internet fray this is thus not always a highly attractive proposition. Not entirely unlike the LECs, the CATV industry's principal real asset is a distribution bottleneck and a highly concentrated structure. (Satellite systems offering hundreds of channels may change this, if they gain sufficient market share, but they have not done so yet, and they cannot carry local television channels.) Current cable systems support about 50-100 channel slots, of which up to two dozen must by law and/or regulation be ceded to local broadcast channels and noncommercial purposes. (In New York City, for example, five channels are designated for public purposes and 16 local channels are carried.) CATV firms' bargaining position relative to independent channels is further strengthened because they own equity interests in many of the major channels they carry. Of the channel slots remaining after public access and local station allotments, a substantial fraction are typically given to channels linked to the CATV vendor, leaving only 25-50 "discretionary" slots. Over 200 channels producing content compete for these 25-50 channel slots, which are controlled by a small number of large firms. Additionally, the CATV industry is actually quite concentrated. The major independent firms (TCI, etc.) have interdependencies with each other, with the non-independent firms owned by LECs or media conglomerates, and strong equity and strategic linkages with other proprietary media industries, even those with which they compete, a condition typical of the media sector in general. Most large CATV providers own or control large quantities of proprietary content, which they use on their own systems and sell to each other. Furthermore, the larger non-independent CATV providers are owned by large media conglomerates that also own cable channels, broadcast networks, video chains, and/or other proprietary media properties such as movie studios, and newspaper chains. There are also significant cross-linkages between the largest cable providers, both independents and those with media parents. For example TCI's CEO, John Malone, owns a substantial stake in his firm, which is the largest CATV provider in the U.S., with 14 million subscribers (roughly a 25% share of the cable market). However, Malone was also a major shareholder of Turner Broadcasting, which included CNN. Turner has now been sold to Time-Warner, which in addition to being publisher of Time, Fortune, People, and Life is also the second largest CATV operator in the U.S., with about 10 million subscribers. With this sale, Malone became the second largest shareholder of Time-Warner, supposedly his principal competitor. Time-Warner's largest shareholder is U.S. West, one of the LECs which also owns Continental, the third largest CATV provider. In addition, TCI holds a 20% position in Fox News, one of Rupert Murdoch's cable channels, as well as equity positions in a number Liberty Media and other cable channels and content providers. Such arrangements are common in the CATV industry. Moreover, even the ownership structure of the CATV industry's competitors, i.e. the network broadcasters and satellite carriers, is closely intertwined with that of content providers and cable channels. ABC is owned by Disney, which also owns cable channels and has a major joint venture with four LECs. NBC, which is owned by General Electric, owns several cable channels including CNBC; Murdoch owns satellite distributors that carry Time-Warner-owned channels as well as owning two cable channels himself. In addition, as mentioned above, several of the LECs, including NYNEX and U.S. West, have strong financial connections with CATV vendors including TCI, Time Warner, and Viacom. For example, TCI and Ameritech are coinvestors in New Zealand's telephone network. TeleWest PlC, the largest cable television provider in England, is a joint venture of US West and TCI. Perhaps most importantly, all major CATV vendors and content providers are interested in keeping video-capable bandwidth scarce, retaining maximum possible control over content creation and/or distribution, and maximizing barriers to new entry in these areas. Although each firm is interested in maximizing its own share of the programming and distribution systems that are the industry's main assets, none of them is very interested in having these proprietary bottlenecks and entry barriers destroyed. And that is what high bandwidth local service, video servers, and open architecture systems (such as 20 megabit HDTV-capable residential Internet access) would do. In fact such open systems, together with the progress of inexpensive digital filmmaking and video server technologies, would dangerously undercut the proprietary positions of movie studios, magazines, television studios, broadcast networks, satellite broadcasters, AND the CATV industry. This risk is exactly analogous, in the domain of video entertainment, to how the current World Wide Web challenges proprietary online services, newspapers, magazines, retailers, yellow pages services, and so forth. If it becomes possible for develop video content inexpensively and independently, to post it on any video server, and then for all consumers to have access to such servers over open networks such as the Internet, the proprietary position of the current incumbents decline significantly. In fairness the CATV industry does see some potentially lucrative business opportunities in digital and/or Internet services, and efforts are being made by some firms to provide Internet service. A few CATV firms, including TCI, have made small investments in providing Internet access. TCI, for example, is an investor in @Home, a startup which is trying (thusfar without much success) to provide Internet access via cable systems. If the CATV industry assumes that eventually open services will inevitably appear, and that the industry will be forced into real competition, any individual CATV vendor might rationally calculate that is better to make money by offering new services than to be entirely shut out. This is the calculation, for example, made by America Online and The Microsoft Network in the context of the current Web; both have decided to offer Internet access alongside their own proprietary services. However, despite many public statements to the contrary, the U.S. CATV industry has not, in general, shown a major shift towards advanced technology or competitive behavior. Incumbents in such positions are rarely leaders of self-cannibalizing technology revolutions; in fact even Microsoft and America Online initially resisted the Internet revolution. The CATV industry hardly shows any indication of being an exception to this rule; indeed quite the contrary. None of the CATV vendors has moved aggressively, despite occasional public pronouncements. All have resisted open architecture systems in their own core businesses, e.g. with regard to standards for interconnection between CATV systems and set-top boxes. None has entered into significant competition with the LECs for provision of local telephone service under the 1996 Act. Nor are any CATV vendors currently offering Internet access or high speed digital services to more than a few thousand people. This constitutes a powerful reason why none of the incumbent actors - CATV vendors, video content providers such as movie studios, broadcasters, theater chains - are likely soon to be aggressive and/or completely open Internet access providers at bandwidths capable of accommodating video. In fact, because (unlike LECs) the CATV industry does not have any current obligation to open its networks at all, CATV providers could even offer Internet access in ways that would advantage their own content. This is because the technologically optimal location for video servers would be in the CATV vendor's equivalent of a LEC central office, i.e. a CATV network "head end." Because the 1996 Act does not require the CATV industry to offer unbundling or collocation, CATV vendors could offer very high speed Internet delivery of their own video content (or that of favored affiliates) while offering lower quality general Internet access to all others. This possibility is already under active discussion within TCI and @Home.83 As a result, large U.S. CATV vendors have at best mixed incentives to invest aggressively in providing truly open high speed Internet access. Even if they did begin offering large-scale residential Internet access, it might be restricted, overly expensive, and/or obligatorily bundled with their proprietary content, and it will not serve the business market at all. CATV industry management also seems poorly equipped to deal with the advent of competition driven by radically faster technical change. This is true both of the independents and of CATV vendors who have been acquired, since the purchasers are generally LECs, broadcasters, movie studios, publishers, or newspaper chains. No major CATV vendor is owned by a high technology firm. Among cable channels and video content providers, the same is true, with only one comparatively minor exception - MSNBC, which is a joint venture of Microsoft and NBC. Thus, the U.S. CATV industry often resembles the LECs, and in some ways is even more resistant to change. CATV vendors have low capital investment levels and closed systems with proprietary control over interconnection and much of their programming. They have resisted efforts to provide open system architectures, e.g. by loosening their control over set-top converters or their control over which programming and channels they carry. And, as with the LECs, most CATV vendors have low R&D budgets and limited technological expertise in top management. In the few cases of their entry into such arenas, CATV companies often use joint ventures and partnerships requiring little capital expenditure or risk on their part - TCI's investments in @Home and in British cable-based telephone competition being examples. Altogether, it does not appear that the CATV industry will be an adequate substitute for, or spur to, a technologically progressive telecommunications system. 3.10 In Summary: There Is a Problem Here The foregoing discussion provides many reasons for believing that, left to themselves, the LECs will not supply the United States with adequate information infrastructure, and that the CATV industry is not likely to make up for the LECs' deficiencies. The LECs' problems include their poor record in improving price-performance, their low capital investment levels and use of their core services as cash cows, their financial and investment priorities, their cooperative and anticompetitive behavior, their poor R&D record and slow deployment of new technology, their rational economic incentives to resist competition, the personal incentives of their executives and directors, governance issues, their history lobbying and litigation, their funding of academic policy experts and former government policy officials, their general avoidance of competitive businesses, their apparent inefficiency and inexperience with high technology, and their regulatory and legal environment (of which more shortly). Analysis of the CATV industry similarly suggests that for a number of reasons, it is unlikely to compensate for and/or discipline the LECs' behavior sufficiently to supply the digital infrastructure necessary for U.S. economic growth. Many of these facts and observations can of course be challenged to some degree. Comparisons between regulated common carriers and competitive activities may not be entirely fair; the LECs' technological performance or governance may not be quite as bad as portrayed; there may indeed be some legitimate differences in the technological characteristics of phone networks and other information technology industries; some of the fault for LEC behavior may lie with earlier regulatory regimes which no longer obtain; and so forth. However, the evidence of serious problems in LEC and CATV behavior is so broad, comes from so many different sources, and is so consistent that one cannot avoid the conclusion that the current situation in local telecommunications constitutes a major impediment to the future progress of U.S. high technology industries and the U.S. economy. In short, there is a problem here. This industry now represents a major economic policy problem for the United States. In the next section, I consider the current policy situation and structural trends acting upon this problem, in order to assess whether it will correct itself naturally or whether further policy intervention is required. The answer, it turns out, is clearly the latter. The LECs' behavior, together with major defects in current Federal policy, imply that the LECs could easily continue to hold and use monopoly power for a long time to come. While technological, economic, and political forces will surely force change in the long run, it will not come quickly or optimally without more forceful and effective Federal policy intervention. 4. CURRENT CONDITIONS: POLICY & STRUCTURAL IMPEDIMENTS TO INTERNET GROWTH AND INFRASTRUCTURE MODERNIZATION 4.1 Technology, Market, and Economic Problems Given that the LECs' performance is inadequate, and that the CATV industry is unlikely to make up for it, prospects for improved Internet and/or digital services depend upon other competitors, interest groups, and regulators. However, under current policy conditions other competitors and new entrants cannot develop modern infrastructure systems effectively. Thus, without more forceful and progressive Federal (FCC and DOJ) policy intervention, current trends offer a mixture of pessimism for the next five years and optimism for the longer term concerning the development of information infrastructure in the United States. The need for improved digital services and lower prices will soon become serious. Modem technology is nearing its limits. If one extrapolates forward based upon the experience of the last five years, then in 2-4 years the majority of U.S. residences will need more bandwidth than modems can provide. Businesses will be in the same condition, and in some cases already are. For example, technology executives in the California and New York film and television industries have already complained privately that the metropolitan area bandwidths they need are simply unavailable from the LECs, at any price. For the economic reasons described in chapter 2, even if sufficient bandwidth is available, excessive prices and slow technology deployment will engender serious economic losses for the United States. Moreover, the current situation is already causing substantial damage even before absolute limits on modem speeds are reached. ISDN as currently offered is seriously inadequate - its current prices, complexity, service quality, and even availability are inadequate. The same is true of current T1, because of its excessive prices, HDSL / SDSL services (because they are not even available), and very high bandwidth services (generally not commercially available either). At some point, quite possibly as soon as 3-4 years from now, the same will also probably be true of T3 service, HDSL-2, 1.5-6 megabit ADSL, VDSL, and/or Internet based HDTV, as well as a wide array of applications and value-added services that require modern digital transmission. Similarly U.S. businesses will soon need ISDN / HDSL / T1 speeds almost universally, and even faster services on a fairly broad basis, at prices far below those currently charged by the LECs. Many businesses will also want levels of integration between voice telephony, Internet telephony, and general Internet service that the LECs are unlikely to provide cost-effectively, or at all, themselves. Thus the LECs' behavior will therefore become, as suggested above, an increasingly serious limit on U.S. economic growth. These problems come not only from the LEC and CATV industry behavior discussed above, but also from a policy regime which impedes progress, partially continues the LECs' entrenched position, and raises large barriers to new entry and/or innovation, particularly in Internet service and digital communications. And, unfortunately, this is true not only of the pre-1996 regime, but of the current Federal policy regime as well, including the FCC's regulations issued pursuant to the 1996 Act. 4.2 The Current Policy Regime: The 1996 Act, The FCC, The Antitrust System, State PUCs, and the Courts The 1996 Act: Primarily Looking Backward, But Probably Flexible Enough The 1996 Act both represents substantial progress and generates some impediments to progress towards an open, Internet-friendly, technologically progressive U.S. system. Its limits derive from its focus on traditional voice telephony, the burdens and rights that the Act specifies for "telecommunications carriers," and its ambiguous and sometimes restrictive language with respect to Enhanced Service Providers (ESPs), including ISPs. The Act distinguishes "telecommunications carriers" from ESPs, and provides the former with substantial rights to interconnect with LECs, while saying comparatively little about the latter. The Act also requires LECs to provide for collocation of telecommunications equipment, but says nothing about data networking equipment or computer servers. In defense of the FCC (whose timidity and/or slowness is seriously criticized below), there can be little doubt that the comparative treatment of telecommunications incumbents (LECs and IXCs alike) relative to ESPs derives in part from the comparative lobbying power of the two industries. At the same time, however, the 1996 Act also explicitly provides at least two mechanisms permitting the FCC to create broader rights for ISPs, ESPs, and/or other new entrants, and leaves intact pre-existing FCC powers which give it substantial control over industry structure. I will now summarize these problems and possibilities. The Problems Apparently, the 1996 Act was not written with a digital and/or Internet-driven future in mind, perhaps as a result of the political power of the incumbent LECs and IXCs. In any event, the 1996 Act as written is structured primarily to create a reasonably open, competitive environment in traditional voice telecommunications services, and in the context of the structure of the traditional telecommunications industry. In several critical areas, it disadvantages or simply ignores the Internet, ISPs, ESPs, and/or data communications providers. Another possible deficit of the Act, somewhat unique to this industry, is that it says little about FCC or other enforcement powers in the event of LEC noncompliance. Given the history of slow FCC movement, LEC regulatory lobbying, and LEC litigation, it might have been quite helpful, for example, to have provided for fast-track adjudication of problems, and/or an explicit legal provision that LEC abuse of monopoly power would lead to treble damages liability, as is the case with violations of antitrust law. The Act generally ignores the fact that traditional distinctions between voice service, digital services, and data services are being rendered obsolete by the Internet and other digital technologies, services, and applications. It also seems insensitive to the need for increased flexibility, technological change, and competition in all telecommunications services (including but not limited to voice) as a result of the spread of decentralized, modular, digital technologies. Not surprisingly, therefore, the Act places little emphasis upon the potential role of digital services, data services providers, and/or the Internet in providing competitive discipline for the LECs. This is a shame, because the most effective way to provide competition and reduced rates for POTS may quite possibly be through increased technical progress in voice-capable digital services provided by ESPs, e.g. Internet service (Internet telephony, Internet voicemail) and/or use of HDSL to provide voice services. These problems are manifested in section 251 of the 1996 Act, which specifies the rights and duties of "telecommunications carriers" and "local exchange carriers," including "incumbent local exchange carriers." LECs and ILECs must provide nondiscriminatory service, access to unbundled elements and collocation rights, and interconnection with each other. ILECs are specifically required to permit collocation in their central offices of equipment required for interconnection and provision of telecommunications services. While section 251 does not explicitly deny such rights to ESPs, it does not explicitly provide them, either. Nor does section 251 explicitly require the LECs to provide collocation rights for any class of equipment other than conventional telecommunications equipment, such as DSLAMs, ATM or frame relay switches, routers, Internet servers, or video servers. And finally, the Act imposes no interconnection or collocation requirements on CATV vendors whatsoever, unless they offer voice service and thereby become "telecommunications carriers." But even in this case, they would be subject only the same requirements as the LECs. Furthermore, if CATV vendors offer Internet service but not traditional voice service, they are subject to no open systems requirements at all. Some have suggested that one way to circumvent some of these problems might be for ISPs and ESPs to become classified as "telecommunications carriers" by virtue of their capability to carry Internet telephony. In this case they could at least purchase unbundled loops from the LECs. However, this reclassification could have the disadvantage of forcing them to uphold various regulatory, financial, and legal obligations which ISPs might reasonably find quite distasteful, and which are inconsistent with the behavior required of a highly competitive, rapidly changing industry. Furthermore, even if they reclassified themselves, under current FCC rules ISPs still could not collocate networking equipment. The fact that the Act does not explicitly grant ESPs section 251 rights thus poses a problem. On the other hand, the Act does not at all seem to prohibit increased ESP rights, and/or FCC actions to provide such rights. In fact, in several places the Act give the FCC the right and even the responsibility to promote competition and the deployment of advanced services, even if FCC actions for this purpose override or violate other provisions of the Act.84 Flexibility And Possible Solutions Within the 1996 Act In addition to its ambiguities, the 1996 Act contains several provisions suggesting that the FCC has, and is intended to have, wide latitude in prescribing the conditions of interconnection, collocation, and competition generally. First, the Act explicitly states that other FCC regulatory decisions in these areas are not invalidated by the Act. Second, the Act contains "forbearance" provisions which permit or even require the FCC to forbear from enforcing provisions of the Act which might interfere with the promotion of competition. And third, the Act contains provisions mandating the FCC to promote the development and deployment of "advanced services," including specifically their provision to schools and educational institutions. And finally, the Act explicitly delegates to the FCC the responsibility for interpretation and implementation of the Act, including the requirement to publish Orders containing detailed implementation language. Thus, while the 1996 Act is certainly flawed, incomplete, and oriented towards the former world of regulated, voice-dominated, circuit-switched networks, it is not so flawed or rigid that the FCC's hands are tied. The Act appears to give the FCC the power to do the right thing, both through language in the Act itself and through leaving intact the FCC's other regulatory powers. As it turns out, however, in some critical areas the FCC has not done the right thing. The FCC Interconnection Order, the first of a projected three FCC Orders implementing the Act, adopts extremely conservative interpretations of the purposes and language of the Act. It explicitly declines to provide ISPs and ESPs with the rights they would need for construction of a modern Internet infrastructure in the United States. The FCC Interconnection Order In some ways, the FCC Interconnection Order effectively furthers the procompetitive goals of the 1996 Act. In others, however, it does quite the reverse, to a degree that seems to reflect a combination of technological ignorance, political timidity, and/or general bureaucratic resistance to change. In several places where the Act is silent or ambiguous, the FCC Interconnection Order is all too explicit in restricting the rights of competitors and the corresponding obligations of incumbent LECs. The FCC Order specifically states that ESPs do not have section 251 interconnection or collocation rights. Furthermore, the Order states that the LECs have no requirement to permit collocation of equipment whose purpose is to provide enhanced services, even when the equipment is provided by a telecommunications carrier possessing these section 251 rights. In fact, the Order restricts equipment for which LECs are obligated to grant collocation to equipment which is required for interconnection with the LEC and used solely for the provision of telecommunications services. The Order also explicitly declines to require the LECs to provide subloop unbundling, even to telecommunications carriers. Subloop unbundling, which permits access to LEC infrastructure at points closer to users than central offices, is already technologically important and will become increasingly so for high speed Internet access and/or high bandwidth services such as VDSL and/or HDTV. Even worse, after noting that new technology was "blurring" the FCC's distinctions regarding equipment categories, the same paragraph of the Order states that where there may be any dispute concerning whether LECs must permit collocation of a specific kind of technology, such disputes will be resolved by individual state PUCs, which could effectively render consistent national services impossible whenever opposed by one or more LECs. The Order contains some provisions requiring the LECs to accord competitors the same collocation rights that they accord their own unregulated subsidiaries, but there are serious loopholes in this requirement. And of course, if the LECs simply fail to deploy a given advanced service at all, they can thereby avoid all obligations to give others the collocation rights required to do so. These provisions of the Order have three negative consequences. The first is to ratify the market power of the large telecommunications incumbents - certainly the LECs, but probably also the three large IXCs (AT&T, MCI, and Sprint), plus perhaps a few of the larger CAPs such as MFS. At present, because independent ISPs and ESPs do not have section 251 rights and their equipment is not eligible for collocation, they must depend upon partnerships with LECs and/or IXCs, all of whom are (a) their competitors and (b) slower moving than the independent ISP industry. Most new entrants will essentially be shut out of direct usage of the 1996 Act by the financial, technological, regulatory, and litigation barriers associated with offering traditional telecommunications services, fulfilling the legal requirements of telecommunications carriers, dealing with the FCC and state PUCs, and handling the LECs' legal and regulatory maneuvering. Third and relatedly, even those who surmount these barriers will be forced to have inefficient technological arrangements and cost structures by virtue of the requirement to offer conventional telephone service and their inability to collocate data networking equipment. The Order thus impedes development of technologically innovative, competitive, and important digital services and industries. If ESPs cannot collocate and the technology required for enhanced services cannot be collocated by anybody (except, probably, the LECs), the development of high bandwidth Internet access and other advanced services will be damaged quite substantially. The most innovative services and technology in these areas consistently come from entrepreneurs and new entrants, not from incumbents. Under these conditions, network congestion could indeed become a very real problem several years from now. Needless to say, however, while the LECs have legally challenged the pricing provisions of the Order, they have not challenged the provisions limiting collocation and ESP rights. Thankfully in the same paragraph of the Interconnection Order in which it restricts LEC collocation obligations and ESP rights, the FCC also says that it reserves the right to revisit these questions in light of its procompetitive goals. As the saying goes, there is no time like the present. In fact, the Internet Access Coalition has filed a document with the FCC (via its Notice of Inquiry concerning access provisions of the Act and implementation thereof) protesting this situation and arguing for requiring LECs to provide full collocation and interconnection rights, including extending those rights to ISPs and other ESPs.85 The LECs naturally oppose this request. As mentioned above, the FCC recently published a long essay on the Internet by Kevin Werbach, of the FCC's Office of Plans and Policy, which describes these issues but provides no recommendations or indications of the FCC's future direction. The FCC will apparently decide some of these matters in late 1997. However, the defects of the Interconnection Order are only the most recent symptoms of the FCC's longstanding problems, and which are at least partly shared by other relevant Federal regulatory authorities, especially the antitrust authorities - the FTC and the DOJ Antitrust Division. A General Observation: The Procedural Morass of FCC Policymaking The FCC Interconnection Order, complex and flawed though it is, is only the latest of many FCC notices of inquiry, proceedings, dockets, and orders which it has produced over the last 20 years related to interconnection and/or competition in the telephone system. Others include Computer III (which required years and thousands of pages, and which followed the equally lengthy Computer I and Computer II) and the FCC Extended Interconnection proceedings and order of the early and mid-1980s. These three sets of FCC activity - Computer III, Extended Interconnection, and the Interconnection Order of last year - now represent three independent, sometimes apparently contradictory, legal and policy bases for FCC decisions in regard to competition in local telecommunications.86 In addition these and other FCC decisions have in several cases been modified or struck down by courts after LEC or IXC litigation. This does not appear to be a highly unusual situation in regulatory policymaking. Like much commercial and antitrust litigation, FCC rulemaking is extremely legalistic, very expensive, procedurally baroque, slow, and therefore ill-suited to rapidly moving, high technology industries. While this situation is understandable for historical reasons and due to the FCC's legalistic culture, it represents a major drag on productive activity and, in the case of the Internet, is no longer an affordable luxury. It is regrettable, but understandable, that Silicon Valley firms avoid contact with such processes whenever possible. In private, computer and Internet executives have repeatedly expressed to me opinions about FCC procedures that could not be printed in any family newspaper. Implementation and Enforcement Problems In addition to the intrinsic defects of the Act and the FCC Interconnection Order, there are several practical reasons to question their effectiveness in developing real competition. First, as a practical matter the LECs are resisting competition through regulatory, political, legal, and operational delay tactics. They are doing so not only through litigation and lobbying but also by delaying negotiation and implementation of agreements with competitors. More than one year after signature of the Act, there is still very little competition even in traditional POTS, even from large and powerful companies experienced in regulatory politics, such as AT&T and MCI. Second, there is even some question as to how seriously the major IXCs will enter local service, at least in the short term. Because the LECs' ability to enter long distance markets are at least to some extent tied to the level of competition in local services, there is a significant incentive for both the IXCs not to reach agreements and/or high market penetration levels very quickly unless they can obtain highly favorable terms from the LECs. Third, the IXCs and other potential competitors trying to interconnect with LECs under the 1996 Act will initially need to rely heavily on LEC infrastructure, such as Operations Support Systems (OSSs). This infrastructure is complicated. The LECs understand it well, while competitors do not. But efficient operations will require cooperation with the LECs, including detailed technical information. AT&T has even taken the unusual step of posting complaints about access to LEC OSSs on its web site. In addition, competitors's services will in some ways be limited by the LEC systems which they must use. Recall, for example, that even ISDN and digital switching remain unavailable in about 20% of the United States. The LECs, not competitors, will decide when and/or how to provide new capabilities in their networks. To be fair to the LECs, current price-cap systems and depreciation regulation may worsen this situation because they incent cost reduction but not investment or technical progress. There is also a possibility that the 1996 Act and/or FCC policy could also create new disincentives for the LECs to invest in new technology. For example, if the LECs are required to offer new technologies on an unbundled basis at low prices, they may have a legitimate complaint that they have no incentive to invest because they cannot profit from those investments. Third, neither the LECs, the likely early competitors such as IXCs, nor the state PUCs seem inclined to move very aggressively in providing broad access to high speed Internet services. The early competitors under the FCC's current interpretation of the 1996 Act will be a small number of large firms, particularly the "big three" of AT&T, MCI, and Sprint. Any ESP, including any Internet service provider, will have to arrange loop access through them, since ESPs currently do not have their own section 251 interconnection rights. Not only are these large long distance companies competitors in Internet access, and therefore not ideal partners for ISPs, but they are also not the most rapidly moving providers of novel or advanced digital services. Furthermore, high speed Internet service threatens the IXCs as well, because the advent of effective Internet telephony would undercut their interstate, international long distance, and voicemail revenues. State PUCs, sometimes heavily influenced by the LECs, may also retard implementation of the Act if they fear it would increase local and/or rural residential rates. Altogether, therefore, one cannot rely on the 1996 Act, the FCC's current implementation efforts, or the incumbent industry to generate the level of technical progress and digital services deployment required by the U.S. economy. And while the LECs are the largest cause of this condition, the FCC and other regulatory bodies also bear a not inconsequential share of the blame for it. The FCC, in fact, bears a disturbing resemblance to the LECs in some ways, and is very far from being the kind of organization that should hold major regulatory powers over a rapidly changing, high technology industry. The FCC, The Antitrust System, State PUCs, and the Courts The FCC While the FCC has improved recently, its actions are considerably behind its public rhetoric. It is too much like the industry it regulates, perhaps because like the LECs it has been accustomed to slow-moving monopoly conditions, and seems quite sensitive to political pressure. (To be sure, Congress is one source of that pressure, and Congress doesn't win many prizes here either.) The FCC contains few if any people, its commissioners included, who have any serious experience with high technology. FCC commissioners are nearly always lawyers or broadcasters by trade. Senior staff are almost entirely lawyers and/or career civil servants, plus a few academic economists on short term leave. Although there are a considerable number of very able and dedicated people at the FCC, there also appears to be a fair amount of bureaucratic dead wood. For various reasons including Civil Service regulations, inertia, and political pressures, the FCC has not created a high quality corps of analysts with relevant technical, economic, and/or business experience. Limited skills and knowledge, combined with overwhelming procedural requirements, mean that even the best FCC staff inevitably fall behind current developments in technology, research, and business. This condition is worsened by the severe shortage of information and FCC data collection (especially in suitably computerized form) concerning the industry, its technology, and its economics. Both the FCC and most state PUCs are poorly computerized, although the FCC web site is now quite useful. For all these reasons, major issues receive less, and lower quality, substantive analysis than they require. Despite the fact that the FCC has a budget of nearly $200 million, the resources it devotes to Internet and/or advanced technology issues are minimal at best.87 As a consequence, there is within the FCC a general ignorance of modern information technology that is astounding, even at senior policymaking levels. This condition cannot be blamed entirely on universal problems of Federal bureaucracies, real though they are. Several Federal organizations consistently recruit high quality technical personnel, deal with interfaces between high technology and business, and are capable of responding quickly to change. Examples include the NIH, the NSF, DARPA, the Office of Naval Research, and portions of the intelligence community. Thus the fact of being a Federal agency per se does not completely explain the FCC's condition. The FCC's and state regulators' knowledge is further reduced by political and/or industry pressures that limit their acquisition of information from the LECs. The FCC has improved substantially under Reed Hundt, and has sometimes taken positions antagonistic to the large incumbents, including declining for the present to levy access charges on ISPs. However it is also clear, both from the FCC's general characteristics and from its recent actions concerning the Internet and the 1996 Act, that more progress is required. The Antitrust System The FTC and the Department of Justice Antitrust Division seem to have problems similar to those of the FCC. Their high technology expertise has traditionally been poor and remains limited; their procedures are slow and dominated by lawyers and economists with little technical or business experience. Economic analyses of high technology industries, in the absence of substantial industrial or technical understanding, tend to be highly simplistic and to miss important questions. While this problem is endemic to economics and not specific to antitrust policy, the DOJ has apparently done little to address it by augmenting theoretical economics with sophisticated technology assessment. On occasion, DOJ economic analysis may also be affected by conflicts of interest and/or simple ignorance of real industrial behavior. Many former economists of the DOJ Antitrust Division, including former DOJ chief economists, now engage in corporate antitrust consulting, and some did so even prior to their taking their DOJ positions.88 Analogous statements can be made of some DOJ attorneys now in private law practice. The antitrust system also has coordination problems, because for historical reasons the FTC handles the CATV and software industries, while the DOJ handles telecommunications and computers. And finally, antitrust lawsuits once filed often take many years, and hundreds of millions of dollars to litigate; the LECs would surely outspend DOJ by at least ten to one in any major antitrust case. For example, the lawsuit leading to the 1982 AT&T consent decree and 1984 divestiture was filed in 1976, and cost AT&T literally hundreds of millions of dollars. Nonetheless, antitrust powers are quite substantial. It is comparatively easy to block mergers and acquisitions that exceed threshold sizes established by the Hart-Scott-Rodino Act, and the prospect of treble damages is taken seriously by private firms. It is unclear why the Justice Department has recently been so passive in regard to the LECs and the telecommunications industry. Over the past three years of the Internet revolution, there has been not a single antitrust lawsuit filed by DOJ against the LECs. While the 1996 Act liberalized many previous restrictions pertaining to the structure of the industry, the Act did not provide any antitrust exemptions or repeal any antitrust laws. Indeed, under the 1996 Act the FCC receives opinions on the state of competition in local markets from the DOJ, and the DOJ retains general authority with respect to antitrust enforcement. It is therefore somewhat curious that while the Federal Trade Commission recently blocked the merger of two $4 billion office supply retailers, the Justice Department recently approved the merger of SBC and Pacific Telesis, two of the largest monopolies in the United States. Another huge merger, between Nynex and Bell Atlantic, is still pending at this writing, although it is known that, to its credit, the Antitrust Division is worried about it. However, the analysis of the Nynex-Bell Atlantic merger is a symptom of how peculiar the situation is. Rather than noting wide, longstanding patterns of monopolistic and cooperative behavior among the two firms, the Justice Department tried to find evidence that the Bell Atlantic and Nynex would have started to compete if they remained independent. As a result, in this case, the absence of evidence that they planned to compete became evidence in favor of permitting the merger.89 First, as a tactical matter the DOJ may not find sufficient evidence. But second, it is a bizarre and unfortunate illustration of how antitrust analysis can miss the forest for the trees. The fact that these firms don't compete with each other is precisely the point, and is a major element of their overall pattern of monopolistic and anticompetitive behavior. Conversely, to my knowledge there has been no DOJ challenge to, nor any major investigation of, the LECs' monopolistic data services pricing, the cooperative behavior described above, the LECs' avoidance of competition with each other via the provisions of the 1996 Act, or their practice of restricting their service offerings (such as high speed data services and Internet access) to their operating geographies even when they could be offered nationally. The need for improved antitrust analysis is particularly acute in high technology, where technological, qualitative, and dynamic factors are more important than in traditional industries. DOJ analyses have generally tended to be focused upon static, measurable criteria used by economists with no technical or business experience. Such analyses emphasize market shares and industry concentration ratios but generally neglect strategic business interdependencies, interlocking board memberships, political strategy and alliances, technological progress or the lack thereof, benchmarking against other industries, and international behavior among other factors. These analyses are also prone to error if the lawyers and economists performing them simply don't know much about the industries and technologies in question. But these problems are tractable, and other portions of the Federal government possess very substantial high technology expertise. Thus there seems to be little excuse for not strengthening of antitrust policy, skills, and enforcement in the current situation. State PUCs and Courts State PUCs have long been considered to be highly influenced by LEC lobbying, as are many municipalities in regard to CATV licensing. While a few (California, Illinois, New York) tend to act independently and have significant staffs, most are no match for the LECs - or for the complexity of new technologies. State PUCs generally have even less understanding of high technology than the FCC, sometimes make inconsistent decisions, and are sometimes even slower to act than the FCC. The same holds for the courts. As a result LECs often employ delaying tactics in both forums, often with considerable success. Moreover, even without any LEC activity, the need to deal with these institutions increases delays and entry barriers that are severe by high technology standards. The risk that FCC and/or court decisions could progressively force the Internet industry into significant portions of the conventional regulatory regime is therefore a potentially major problem. 4.3 Long Term Pressures In Favor of Progress Despite the present impasse, it is clear that time and political trends are against the incumbents and in favor of technical progress. The conditions described above are already beginning to generate powerful countervailing forces in favor of modernization (at least in the United States, which is unique in this respect). These pressures will soon come not only from the U.S. information technology industry, but also from business users of communications technology and consumers. Already, the high technology sector has reacted to the telecommunications problem by forming the Internet Access Coalition, and stronger actions will surely be taken if no progress is forthcoming. Traditionally, the LECs have counted on their size, financial strength, and investments in lobbying to guarantee their superior influence in political and regulatory affairs. But their power is already beginning to wane, and will surely be much reduced over the next five years. There are two fundamental reasons to predict the decline of LEC influence. First the United States, unlike most nations burdened with stagnant telecommunications monopolies, is the world's primary user of advanced information technology. And second, the U.S. is also the largest producer of high technology, an industry that is rapidly becoming more influential. Ten years ago Microsoft, Intel, Compaq, Dell, and Cisco were politically insignificant, with combined revenues of less than $5 billion. Today their combined revenues are about $75 billion and growing fast. In addition a growing collection of other information technology firms including Hewlett-Packard, Oracle, Computer Associates, EDS, Netscape, Bay Networks, 3Com, U.S. Robotics, Dell, Gateway, Lucent, DEC, and even IBM are becoming increasingly dependent upon the Internet for their growth. This sector has larger revenues than the LECs; and it is growing about 30% per year. Its employment is also substantial; Intel already has almost 50,000 employees worldwide, Microsoft about 25,000, H-P over 100,000, IBM about 250,00. Within five years, local communications bandwidth will probably be the single largest driver - or bottleneck - of high technology production and market growth in the United States, so the issue will be on everyone's mind. And firms like Intel, Microsoft, Compaq, Oracle, and Cisco are not sleepy regulated utilities. As those who have dealt with them know, when they want something badly they become quite vigorous indeed, and they absolutely hate to lose. They may not be experts at regulatory politics yet, but they can learn, and money will not be an object. Microsoft has $9 billion in cash, Intel only slightly less. And they will have the growing resentment of both business and home Internet users on their side. Unlike the LECs, the modem, computer networking, and PC industries have been doubling the price performance of their products every two years. Users will expect local loop bandwidth to keep up. If one simply extrapolates the trends of the past five years, by the turn of the century over 50 million home Internet users will expect 128 kilobit Internet access for the price of POTS; if they don't get it, they will be angry. Business users won't be any happier. The formation of the Internet Access Coalition is one early sign of this growing discontent with the state of local loop technology and regulation. Thus I would predict that within five years, the political equation will be very different than it is now. For politicians and policymakers with sufficiently long time horizons, the coming political decline of the LECs should soon be apparent. However, these trends notwithstanding, it must be emphasized that for now the LECs remain dominant in the local loop, and that this fact is a major problem. 4.4 In Conclusion: Action Still Required The industrial, political, and regulatory systems related to telecommunications have great inertia. The LECs have long experience, $120 billion in revenues, 250 employees in Washington, and thousands of attorneys, lobbyists, expert witnesses, public relations experts, industry association employees, PAC employees, and so forth. By contrast Intel has six persons in its Washington government relations office, Microsoft has four, and Cisco has zero - because it has no government relations function at all. Furthermore the LECs are also large purchasers of computers and networking equipment, a fact which makes many high technology firms hesitant to anger them. Consequently, U.S. policy cannot rely solely on "self-correcting" forces, either political or competitive, to ensure suitably rapid development of information infrastructure for the United States. Without further policy actions, U.S. economic growth and welfare will probably suffer a decade of deadweight losses as a result of inadequate, overpriced local services. These economic losses could easily run to hundreds of billions of dollars. Furthermore, these losses will not just be abstractions. As the decline of IBM, AT&T, and the U.S. automobile and steel companies shows, monopolistic conditions and failure to remain competitive eventually result in extremely painful adjustment costs. The future competitiveness, productivity, technical progress, and growth of the local telecommunications industry will have very concrete effects on many human beings, as well upon economic statistics. If we can improve the situation, the economic stakes are high enough that it is worth the trouble to do so. 5. POLICY IMPLICATIONS 5.1 Overview; General Policy Objectives Federal support for the development of a universally accessible, advanced digital information infrastructure is remarkably consistent with the historical traditions of U.S. policy; so is the promotion of competition and innovation. Although the technological particulars of the Internet are unique, the social and economic objectives associated with U.S. information infrastructure policy have parallels in earlier Federal efforts such as development of canals, postal service, and railroads in the 19th century, universal telephone service earlier in this century, the National Defense Student Loan program and the Interstate highway system in the postwar period, DARPA and NIH support for new technologies, the divestiture of the AT&T monopoly in the 1980s, and the development of the original Internet itself thirty years ago. The main novel feature of the current situation is the greater importance, and much higher rate, of technological progress. Static price levels or profit margins are comparatively unimportant; rather, what matters most is the rate of improvement in technology, quality, and price-performance ratios. This fact affects policy objectives and requirements, by making R&D, capital investment and price-performance improvements primary goals for local services policy, while reducing the importance of regulating static costs or imposing price controls over existing services. The primacy of technological progress also increases the importance of competition, not just because monopolies might charge excessive prices, but more importantly because monopolists tend over time to become technologically stagnant and inefficient.90 However, the underlying values of economic progress, fairness, and access to education that make information infrastructure policy important are quite similar to the goals that have motivated U.S. policy in other situations, including both FCC and DOJ policy towards traditional telecommunications. Thus in some ways, the Internet is simply the dominant public infrastructure of the next generation, and does have parallels infrastructure industries. U.S. policy objectives for local telecommunications might therefore be summarized as follows: (1) to ensure the continued universal availability of voice service; (2) to foster creation of a modern, dynamic, competitive, largely unregulated, open architecture industry for provision of digital services, Internet services, and the equipment upon which they are based; (3) to continue the high levels of standardization, compatibility, and interoperability that have thusfar characterized both traditional POTS and the Internet; and (4) to improve the governance and performance of the LECs, so that this transition may be effected with less financial and human cost than often accompanies technological revolutions. 5.2 Specific Policy Issues and Recommendations ISP/ESP Access to Loops, Subloops, Interconnection, and Full Collocation Rights The FCC should grant independent ESPs the same rights (to unbundled elements, interconnection, and collocation) as competing telecommunications carriers are granted under section 251 of the 1996 Act. Furthermore, subloop unbundling should be mandated as soon as possible, and access to LEC central offices, including to equipment collocated by competitors, should be extended to newly constructed non-LEC copper and fiber lines. And finally, the equipment categories covered by all firms' collocation rights should be expanded to include equipment used to provide general digital services, data services, and Internet access.91 These policies should be strongly enforced, e.g. through prohibition of LEC long distance entry and/or DOJ antitrust actions until suitable levels of real competition, interconnection, and collocation are achieved. These measures are required to remedy the most important deficit of the current policy regime, which is its failure to foster technological progress and open architecture competition in innovative digital services. Over the long run, from the viewpoint of economic and social welfare, fostering competition among the large incumbents in traditional circuit-switched, analog POTS alone is less important than new entry, innovation, and competition in digital services. This is not to say that competition in traditional local service is a bad idea. It makes good sense, provided that it is done so as to incent capital investment in the shared network and so as not to disincent entry by independent ESPs and new entrants (which is what current FCC policy does). Thus the slow appearance of competition even in basic voice service is both a problem in itself, and even more importantly a symptom of the more general problems associated with depending upon the large incumbents. In fact, the most effective way to reduce the cost of POTS over the long run is simply to improve the price/performance of basic digital services, with voice gradually becoming simply one application (along with Internet access and others) using these services. ESP / ISP access to network elements and collocation rights for their equipment are the best way to ensure this, and collocation of advanced networking equipment will in fact eventually be necessary to avoid the network congestion problems about which the LECs complain so loudly. While the LECs have greatly exaggerated the current congestion problem, they are correct that without major changes the problem will eventually occur. The fundamental solution to the problem is fully open interconnection and collocation, rather than repressing demand through access charges, as the LECs recommend. Over the long term, only a suitably advanced network can both reduce costs and avoid congestion, because only via collocated equipment is it possible to efficiently separate circuit-switched POTS from digital traffic, while also gaining the benefit of spreading the fixed costs of wiring, buildings, and so forth across all packet-switched digital services including digital voice. Residential service a decade from now probably could include bandwidth of 5-10 megabits per second and support for multimedia Web services, open architecture video, and/or videoconferencing on demand. In this environment, voice service would be quite inexpensive; but to achieve it, large-scale collocation of networking systems is required. A major practical question is whether increased ESP rights should be accomplished by classifying them as "telecommunications carriers" under the 1996 Act, as opposed to using other mechanisms independent of the 1996 Act, such as the FCC Expanded Interconnection order. I have no specific recommendation in this regard. There is some risk that classifying Internet providers as "carriers" under the 1996 Act may create bureaucratic and regulatory burdens on the Internet industry, which would be a serious mistake. While there are some Internet issues which may justify regulatory intervention, the Internet industry should not, under any circumstances, be subject to anything remotely approaching the regulatory treatment accorded the LECs. Access Charges and Network Congestion The policy question of how to reform the access charges regime governing conventional local and long distance services is not addressed here. However, the structure of the future access charges regime is, in some ways, a relevant question. A number of economists have argued in favor of usage-based pricing and/or access charges on ISPs and/or ESPs, and some have also advocated usage-based pricing for all Internet services. However, at least one FCC policymaker has pointed out that allowing LECs to charge others for causing congestion could give the LECs (and/or other carriers) a disincentive to remove that congestion. In fact, nations with high levels of Internet usage generally do not have metered local service, while those with expensive, metered local service have experienced less Internet growth. More generally, the current discussion of proposals for usage based Internet access charges seems to assume that local bandwidth is a scarce resource that must be rationed. While in theory all resources are scarce to some degree, this notion, which seems to unite the LECs and some theoretical economists, is rather odd in the context of high technology, where rapid and continuous improvement is the industry's fuel. It is even more peculiar in this situation, which combines rapid growth with the availability of technology capable of greatly expanding capacity and reducing costs. Any burden currently placed by ISPs on a LEC network is far more likely to be an artifact of FCC and/or LEC policies impeding investment and optimal technology deployment, as opposed to real scarcity. In fact, the LECs' high prices and low rates of price-performance improvement for digital services such as T1 probably constitute "access charges" for Internet access that are proportionately as large as current real access charges for long distance. The calculations described in chapter 3 suggest that under the current monopoly regime, ISP traffic provides more incremental revenue and profits than it costs. Furthermore, using technologies that can be deployed on a large scale within several years, copper wire can support another 100-fold increase in speed relative to current modem technology. And by using collocated packet-switching equipment, ISPs could avoid causing usage-sensitive costs for the LECs or burden LEC switching systems. In supplying HDSL or SDSL service, for example, an ISP would simply want the LEC to lease it bare copper and collocation space, and then to get out of the way. As we have seen, the LECs may be more afraid of this condition than of congestion. Of course, ISP usage of LEC loops, switching, and trunk systems does, and should, cost something. But charges paid by ISPs/ESPs should be reasonable prices for the portions of LEC infrastructure they actually use under reasonable interconnection and collocation conditions, not charges for excessive and inefficient usage patterns imposed upon them by LEC conduct and current FCC policy. The current situation is extremely artificial, with the major "congestion" problem (to the extent that there is or will soon be one) being generated by a combination of LEC underinvestment in monopoly local networks, FCC and PUC depreciation policy, the FCC's refusal to require that LECs allow collocation of digital networking equipment, FCC and LEC policy preventing ESPs from having access to the local loop, LEC inefficiency, and the LECs' desire to provide competitive advantage to their own Internet access businesses. The FCC should attack some of these problems, not to mention reforming the current access charges regime for voice calls, before charging for congestion and/or imposing access charges on ISPs or ESPs. Real Network Congestion Problems Although the LECs' current claims of congestion and financial distress are exaggerated, there is a potential for two real congestion problems that merit FCC attention: first, congestion of the telephone network caused by insufficient LEC capital investment, slow LEC deployment of new technology, and the inability of ESPs to interconnect with LECs efficiently; and second, congestion of the Internet caused by the same problems. The FCC should monitor this situation closely, and it may prove necessary to require that the LECs provide minimum guaranteed service levels - even if this requires them to invest more than they would like. However, the possibility of truly serious congestion problems in LEC networks is the strongest argument for ensuring real competition and completely open interconnection for voice and data carriers alike. Promoting Open Architecture Digital Infrastructure: Standards, Quality of Service, Competitive Loop Providers, OSS APIs, CATV Unbundling Apart from LEC/ESP interconnection and collocation issues, there are two other domains in which Federal intervention may be required to ensure the openness of the U.S. system. The first is standardization and technical compatibility. The need to maintain (in some cases, create for the first time) a decentralized and open, but still interoperable, national system raises the issue of the FCC's role in standardization. The other problem is the creation of open systems where they are currently closed, both in the LEC networks and, equally importantly, by unbundling the CATV system and requiring CATV networks to offer interconnection and collocation rights similar to those that should be required of the LECs with respect to ESPs. Both technical incompatibility across major infrastructure systems and the persistence of monopolistic closed systems (e.g. CATV networks) could seriously compromise policy goals. Future conditions may sometimes require regulatory intervention to guarantee interoperability even as the local services industry becomes more decentralized and competitive. The history of the computer industry suggests that competitive market pressures usually generate high levels of industrywide interoperability, but do not always do so; for example, competition between hardware companies has caused a proliferation of dialects of the UNIX operating system. In fact, these problems even arise in regulated telecommunications; the LECs have tolerated incompatible versions of HDSL, which could in fact become a serious problem. For reasons similar to those motivating the 1996 Act, Federal policy should probably force the CATV industry towards an open architecture structure, particularly with regard to placement of content on Internet and/or video servers. For similar reasons, there may be a policy argument in favor of requiring the LECs (and perhaps the CATV industry) to publish and make technically available the APIs (i.e. software interfaces) of their control systems so that any communications vendor, including ESPs, can routinely develop products that request data from LEC and CATV infrastructures. Over the long term, Federal intervention may be required if the nature of Internet competition and standardization change for the worse. Thusfar, voluntary and/or market processes have generally worked well for the Internet; but future risks do exist. One is that one firm or a cartel may acquire and then abuse proprietary control over a standard or architecture necessary to a broad array of infrastructure systems and applications, somewhat analogously to the LECs' control over the local loop. This situation could arise through some future patent position and/or through the marketplace dominance of a firm, such as Microsoft or closed system CATV vendors. Another risk is that due to private competitive interests, the market will fail to produce a standard in some critical area, such as videotelephony or Internet Quality of Service standards. This has occurred in the UNIX market, for example, and has cause major problems in the computer industry. In fact, even the regulated telecommunications industry, despite its centralization of standards development in Bellcore, seems to have had difficulty with these issues. These are difficult areas for government regulation, and the FCC should tread lightly where healthy and unregulated market competition already exists. Especially given its current severe lack of high technology expertise, the FCC has absolutely no business developing, choosing among, or mandating technical standards; but if it gains enough technical expertise and a sufficiently broad spectrum of industrial advice, it maybe possible for the FCC to influence standards development processes, and/or prescribe interoperability requirements for services. Future Internet and digital communications standards should still be developed and managed by the IETF, private architectural leaders, and/or industry consortia, perhaps in some cases their adoption mandated for common carriers once they have been clearly adopted by the private sector. Dependence upon glacially slow standards organizations such as the ITU, ANSI, the LECs and Bellcore for future standard-setting should be avoided, as should direct Federal participation in standards development. Federal policy could more safely become activist, however, in increasing the openness of closed systems now dominated by monopolies, e.g. the LEC and CATV systems. And in special cases in which (analogously to UNIX) the market fails to produce interoperability in a critical public infrastructure system, Federal intervention on behalf of standardization may be justified. Antitrust Policy Towards LECs, the CATV Industry, and Others There is a strong case to be made for stronger Federal antitrust action against the LECs, and potentially the CATV industry as well. The refusal of the LECs to offer HDSL and SDSL services, the LECs' failure to improve price-performance in monopoly ISDN and T1 services, the lack of competition in local service more than a year after signature of the 1996 Act, the absence of competition between the LECs in their core businesses, the LECs' strong tendency to offer Internet and long distance services only within their monopoly operating areas, the LECs' structural and financial relationships with each other, the LECs' linkages with the CATV industry, the structure of the CATV industry itself, the absence of major competition in voice and/or Internet services from LEC-owned CATV companies, the clear patterns of LEC cooperative behavior, and the merger plans of 4 of the 8 major LECs, are all extremely disturbing signs. The apparent passivity of the Department of Justice in the face of these conditions is disturbing as well, although perhaps not surprising given the political sensitivity of the issue and the low technological sophistication of the antitrust establishment. There would seem, for example, to be abundant reason to challenge the LEC mergers that have been announced, one of which DOJ has already approved. There would also seem to be good reason to initiate investigations into possible collusion in restraint of trade. However, as with intervention in standardization and open interface issues, successful Federal antitrust action would require greatly improved quality and technological skill within the antitrust system, and particularly within the Department of Justice. Universal Service Rights to Voice and Internet Access There are strong social policy and economic arguments for encouraging broad access to both voice telephony and Internet service, particularly to lower-income groups and the educational system. The mechanisms established by the 1996 Act seem primarily intended to provide universal voice service for residences, but also provides for subsidized Internet access for schools and libraries. There is also an argument for including some level of Internet access in the universal service subsidies provided to residences. The social and educational benefits of doing so could be substantial, as could be the economic benefit of stimulating broader usage of advanced services. However, there is also a major risk in doing so, namely that it will subject the Internet industry to the highly artificial and even bizarre regulatory constraints, cross-subsidizations and political-economic pressures that have characterized the regulated telephone system. The best long term guarantee of broad access to these services is rapid technological progress, which only a largely unregulated, competitive industry is likely to provide. On the other hand, there is no doubt that without some policy intervention, many poor individuals, libraries, and schools would go without Internet access. Thus as FCC policymakers contemplate how to provide for universal access subsidies for residential and educational Internet service, they should take care not to cause the Internet industry any significant regulatory burdens. Whatever mechanism is adopted should be simple, clean and spare. Purely as an example, one possibility might be a low percentage tax on all telecommunications services revenues, transferred to a fund that subsidizes usage by low income persons - say, by giving credits for 50% of the first $25 of residential telecommunications services, in whatever combination of voice and Internet access the user desires. As bandwidth increases, some people might prefer to have Internet access alone, because Internet telephony provides for their needs, and they should be permitted to make that tradeoff as they choose. Most importantly, however, any policy burdening digital services and/or Internet providers with the kind of regulatory accounting demanded of the LECs should be avoided. Soft Money and Campaign Finance Reform in Academia as Well as Politics The extraordinary lobbying investments of the LECs and IXCs, in combination with the policy problems discussed above, suggest that this industry shares fully in the difficulties posed by the use of interest group money in Washington politics. This problem has already been widely discussed elsewhere, and this study has nothing to add. Less widely known, however, is that over the past decade, the academic system (particularly departments of economics and the economics divisions of public policy and business schools) has developed a similar and now very serious problem. As with actual campaign finance, the corporate and/or interest group funding of academic economic and regulatory policy research, of regulatory consulting and expert testimony by prominent economists, and other similar activities proliferating in a manner that threatens to compromise the objectivity of academic work and the availability of expert opinion for policymaking and public service. The resulting problem takes two forms in regulatory issues. First, a great deal of supposedly objective academic research and policy analysis is less objective than it purports or seems to be. And second, it is increasingly difficult to find high quality personnel to serve in government or to study policy problems, because conflicts of interest are so pervasive and the income differential between government and industry is so enormous. Much of this problem is beyond the scope of this document, and it potentially affects many policy domains other than telecommunications. However, it does strikingly affect regulatory issues, and there are some things the government could do about it. Disclosure requirements and limited sanctions, e.g. for flagrant conflicts of interest and/or violations of the university 20% consulting rule, might be in order. The major funding agencies such as the NSF could impose disclosure requirements, conflict of interest standards, and/or limits on consulting activities for grant recipients in policy areas related to the grant. Perhaps even more usefully, the Federal government could establish a small number of long term grants, for discretionary use for any combination of income supplements and research costs. These grants would be made available only to researchers who were not principals in any regulatory consulting firm, who agreed to perform some quota of public sector consulting during the grant period, and/or who agreed not to accept private sector funding in policy-related areas during the grant period.92 Even a fairly small program (by Federal research funding standards) might have the major beneficial effect of supplying a number of high quality, unbiased researchers to agencies such as the FCC. A program of twenty such grants per year, each providing $250,000 per year above normal academic salaries for five years, would cost $25 million per year when at full force. These grant levels may seem high, but this author knows of many professors of economics, including several former chief economists of the DOJ antitrust division and FCC, who derive over half a million dollars per year from corporate antitrust and regulatory consulting.93 In fact, if $25 million annually could have a substantial beneficial effect upon Federal policy analysis in critical areas such as regulation of telecommunications, electric power, environmental policies, and the like, it would be a bargain relative to many Federal programs. Alternatively to using the NSF, the budgets of the FCC and similar agencies could be supplemented with funds for this purpose, or the regulatory agencies could request slots for specific purposes within programs administered by agencies such as NSF, CEA, or OSTP. It would be most valuable if some of these grants were reserved for technology assessment, and/or for applied and empirical economics, as opposed to economic theorists with no factual knowledge of the problem domain. Legitimate LEC Complaints: Depreciation, Investment Incentives, and Constraints The LECs have several legitimate complaints concerning the regulatory constraints under which they have been forced to live. For example depreciation rates, although substantially liberalized several years ago, remain partially regulated. It is very difficult to see how the FCC and state PUCs are better at determining the appropriate depreciation rates of DSL access multiplexers and frame relay switches than the companies that use them. Similarly, the various states, state PUCs and the FCC have an extraordinarily bewildering and, at times, rather absurd menagerie of required accounting categories, regulations, tariffs, and other bureaucratic impediments to effective behavior. FCC preemption of state PUCs and/or deregulation in many of these areas would be helpful to all concerned, although it would also be politically difficult in many cases. In addition to political pressures from interest groups such as rural telephone users, state PUCs have over 20,000 employees and many LECs are partially unionized. Another legitimate LEC complaint concerns incentives to invest. If LECs are forced by the FCC to unbundle their entire networks at prices based on marginal cost, and to offer network elements to competitors as soon as they have constructed them, their incentives to modernize the network are severely reduced. While the LECs have been underinvesting for many years even in the absence of unbundling requirements and competition, it is nonetheless fair and appropriate to give the LECs reasonable investment incentives if and when competition is forced upon them. Price cap arrangements may also sometimes incent the LECs to reduce current costs rather than investing for future growth. A variety of policies could deal with this problem, including pricing schemes for unbundled elements that take into account the vintage of their capital stock. The FCC is apparently studying this problem now, although it has said little publicly about it. At the same time, it should be noted that actions to liberate the LECs may help them less than they claim, and the LECs in fact may not always really want to be freed. The reason for this is that the LECs are highly adapted to a bureaucratic and regulated world, while ISPs and high technology firms are not. Arcane and burdensome regulations create competitive advantages for LECs in major markets, because they raise major barriers to entry against smaller firms with less cash and regulatory experience. Furthermore, the LECs appear to have been underinvesting in their networks recently, even after liberalization of depreciation schedules in 1994. Competition will probably do more to spur investment than will any further deregulation without it. 5.3 The Structure and Conduct of Federal Regulatory Institutions The flexibility and technological sophistication of the FCC and the Antitrust Division compare quite unfavorably several other Federal agencies including DARPA, NSF, and NIH. While the FCC has improved its economic expertise over the past several years, it has not comparably improved its expertise in high technology. In part, this may reflect opposition from the FCC's lawyers, economists, and incumbent technology specialists, all of whose bureaucratic interests would be threatened by the presence of people whose knowledge of high technology would exceed theirs. It is also amusing that the FCC's sole currently designated Internet policy specialist (who is a very intelligent and capable person) is a lawyer. It doesn't work that way at Intel, Microsoft, or Cisco - or at DARPA, the NSF, or NIH. Indeed the governance of the FCC, including the backgrounds of most commissioners and senior personnel, is not entirely dissimilar to that of the LECs, and is equally ill-suited to the future direction of the industry. This is particularly troubling given that the FCC is now beginning to address Internet issues, with potentially enormous consequences for the Internet and computer industries. Actions to improve this situation might include creation of an office of science and technology within the FCC, appointment of commissioners with high technology backgrounds, and introduction of rotations of researchers and technologists such as DARPA and some other agencies have used with great effectiveness. Some exemption from Civil Service requirements, restrictions on the duration of litigation following FCC rulemaking, and/or provision of funds for technical consulting (at higher than normal government pay rates) might also be useful. 5.4 Wider Implications: Electric Power, Foreign Telecommunications Markets The U.S. telecommunications situation seems to have some parallels in other industries, particularly electric power. The Administration and Congress are now considering policy changes intended to introduce competition and regulatory changes into this sector. The electric power sector, like the local telecommunications system, is comprised of longstanding regulated monopolies. Some analysts have argued that these firms and their regulators also display poor technological performance, and that competition should be introduced into the industry through legal and regulatory change. Lobbying in favor of such changes is apparently being led by large business users of electricity, who would benefit from lower rates. The telecommunications experience suggests that this might be highly desirable, while at the same time requiring great care. For example, the introduction of telecommunications competition is forcing a restructuring of complex subsidies that have benefitted low income consumers; the 1996 Act addresses this via the universal service fund. In electric power, deregulation could raise concerns including nuclear safety, winter heating for low income persons, pollution, greenhouse emissions, energy efficiency, and foreign energy dependence. Declining electric rates could cause increased consumption, and it not clear that greater energy use is beneficial for the United States. Nor is it clear that the electric power sector could experience the high rate of technological progress that characterizes telecommunications. Finally, there is no equivalent of the FCC in the electric power industry. With all the FCC's problems, it has been more progressive and independent of industry pressure than state PUCs, and it has imposed some degree of nationally uniform policy on the industry. On the other hand, if the LECs are any guide, the persistence of a regulated, but poorly governed and monopolistic, industry engenders serious problems. Thus it might turn out that under proper Federal regulatory conditions, a competitive industry would pollute less than the current one. This tradeoff, of course, depends crucially on the economic and technological facts of this specific industry. Furthermore, industry expenditures on lobbying and academic experts will again be a problem. For example, the politically powerful large electricity users lobbying in favor of deregulation probably would oppose the increased taxes on electricity that might be required to prevent increased consumption and pollution. Foreign Telecommunications Markets and Global Growth The LECs' problems are shared by many national telecommunications systems, which for historical reasons remain dominated by government-owned monopolies. U.S. development of open architecture competition and a technically progressive digital infrastructure will therefore establish an important precedent, watched and sometimes emulated by foreign systems. An open, competitive domestic industry will also give the U.S. government a credible position from which to pressure foreign governments to open their own monopoly systems to competition. As with the United States, such policies will have positive economic effects on these nations. They will also, however, benefit the U.S. information technology sector, which is the world's leading provider of personal computers, servers, Internet technology, software, Internet services, and digital content. Under suitable conditions, the United States might also become a leading global provider of future digital telecommunications systems and services. Thus, this appears to be a case in which nearly everyone will gain from open markets, competition, and trade. Conversely, it would be difficult and hypocritical for the U.S. government to attempt to force open foreign markets, either through the WTO or through bilateral pressure, if the U.S. industry remains dominated by LEC monopolies, or even by LECs and IXCs. Thus the interests of other nations and of the U.S. high technology sector in global trade provide further indirect but substantial arguments in favor of effective, open competition in the United States. 5.5 Conclusion: An Unusual Opportunity It is rare for a national government to have the opportunity to improve economic growth and welfare in a major way with a comparatively simple policy change. Technological revolutions comparable in importance to the Internet are unusual, and often involve economic, social, and/or military costs as well as benefits. While the rise of digital communications and the Internet has produced some complexities and problems, e.g. encryption and privacy issues, the Internet generally seems to be a good thing. Moreover, technological and economic forces are providing strong incentives for its commercial development. The extent of required government intrusion seems principally confined to ensuring a transition to open systems competition in the LEC (and secondarily CATV) systems, i.e. the problems addressed here. Prompt Federal policy intervention would have a major effect - upon Internet growth, future telecommunications services, U.S. high technology industry, and economic growth in the United States and throughout the world. That is something worth doing. END AUTHOR INFORMATION AND DISCLOSURE Dr. Charles H. Ferguson Visiting Scholar, Haas School of Business, University of California, Berkeley and Center for International Studies, MIT 21B Lee Street Cambridge, MA 02139 Email: chf@cferguson.com Fax: 617-497-2585 Assistant: Simone Ross, email scr@cferguson.com or tel: 617-262-6935 Author Biography and Financial Disclosure The author is currently a visiting scholar at the Center for International Studies at MIT and the Haas School of Business at the University of California, Berkeley. He holds a BA in mathematics from the University of California, a Ph.D. in political science from MIT, and is a member of the Council on Foreign Relations. The author owns a substantial quantity of Microsoft stock which constitutes the majority of his net worth and which he is legally obligated to hold until January 1998, as a result of Microsoft' acquisition in January 1996 of Vermeer Technologies, the Internet software company founded by the author and Randy Forgaard in early 1994. The author also holds a comparatively small quantity of Hewlett-Packard stock, is an investor in venture capital funds, and is a director of the Alliance Technology Fund (a large technology sector mutual fund managed by Alliance Capital), but has no other financial interest relevant to this document. The author has no professional or financial relationship of any kind with any computer vendor, software vendor networking equipment vendor, Internet service provider or other ESP, LEC, IXC, CATV provider, telecommunications company, lobbying firm, industry association, or any other entity financially affected by the subjects discussed in this essay. NOTES 1. The text of the 1996 Act, the FCC Interconnection Order, and many other useful documents can be found on the FCC's web site at www.fcc.gov. 2. The concept of the "Silicon Valley Model," i.e. a modular industry structured by systems architecture, was developed independently by Andrew Grove, CEO of Intel, and by myself and Charles Morris. See either Morris & Ferguson, "How Architecture Wins Technology Wars," Harvard Business Review, March-April 1993, or the longer exposition in Ferguson & Morris, "Computer Wars: The Fall of IBM and the Future of Global Technology," Random House/Times Books, 1993. 3. The so-called xDSL (for Digital Subscriber Line) technologies provide various high speed data services over copper wire. See www.telechoice.com and www.pairgain.com for descriptions of these technologies and information about their standardization and commercial development. 4. There is a useful discussion of these provisions in FCC analyst Kevin Werbach's paper "Digital Tornado," which is the only public FCC analysis of the Internet in relation to the 1996 Act. The paper can be found at www.fcc.gov. 5. For example, several LECs have been raising prices for digital services, particularly ISDN. See chapter 3. 6. This phrase is the title of David Tennenhouse's course. 7. Confidential interviews, plus, unfortunately, the personal experience of myself and my assistant Simone Ross in attempting to obtain data such as price histories and usage levels for digital services. 8. Some sense of the pace of Inernet and web development can be obtained by looking at: www.newslinx.com, a daily list of links to news stories about the web; www.netscape.com, the leading vendor of Internet software; and www.isoc.org and www.ietf.org, the web sites of the Internet Society and the Internet Engineering Task Force. 9. Frame relay and ATM are both modern data communications technologies than can transmit both voice and data traffic. 10. A good sense of the pace of this industry can be obtained by looking at these companies' websites. The comparison with LEC websites is quite clear. 11. One of the largest Internet / web directory and search services, www.yahoo.com, now registers several thousand new web sites per day. 12. For information on the T1 market see www.pairgain.com. 13. For a survey of price cap regulation across jurisdictions, see Robert Crandall and Leonard Waverman, "Talk Is Cheap," Brookings Press, 1996, ch. 2. 14. LEC Annual Reports, 10Ks, and proxy statements. See also graphs in this paper. 15. See www.usrobotics.com. For fiscal 1996, U.S. Robotics' revenues grew 86%. 16. See www.schwab.com. 17. See www.etrade.com. 18. See Crandall & Waverman, p. 27. 19. Ibid, p. 269. 20. readers can examine LEC performance relative to market indexes and high technology firms online. See www.pcquote.com and www.secapl.com. The second, Security APL, provides 5 year growth rates in earnings per share as well as stock price data. Revenue and total return information can also be found on the LECs' web sites and in their annual reports and 10Ks. 21. Personal interviews, LEC resellers and FCC officials. 22. Personal interviews with both LEC employees and users. 23. For those with an appropriate sense of humor, there is a website devoted to Nynex horror stories. Its url uses an obscene word. 24. Price data comes from the authors' surveys of users, LEC resellers, PUC tariffs, and in a few cases LEC-supplied data. Astoundingly, the FCC does not have any systematic price data, not even computerized data on the prices and tariffs under FCC control. 25. For information on attempted and actual LEC price increases for ISDN, see www.essential.org, which maintains a web based bulletin board on ISDN developments, plus the web sites of LECs, especially Pacific Telesis and U.S. West, both of which have recently sought and/or obtained ISDN price increases. 26. Personal interviews, LEC resellers. 27. Personal interviews, network equipment vendors and LEC employees. 28. Personal interviews, network equipment vendors. 29. Personal interviews, LEC executives, LEC resellers, and networking equipment vendors. In addition, see also Lee Selwyn and Joseph Laszlo, "The Effect of Internet Use on the Nation's Telephone Network," January 22, 1997, at www.iti.org. This document was prepared for an industry association composed of Internet companies and therefore may be biased. However, the document contains a great deal of useful and interesting information. 30. See Selwyn & Laszlo, who cite from the FCC Access Charge NPRM, CC Docket 96-262, 12/24/96, paragraph 70. 31. See FCC Statistics of Common Carriers; also LEC annual reports, 10Ks, and proxy statements. 32. Ibid. 33. See www.essential.org. I also conducted interviews with Intel personnel. 34. FCC Statistics of Common Carriers, which is available on the FCC's web site, www.fcc.gov. 35. Confidential interviews, LEC executives and resellers. 36. For a history and description of T1 and the replacement of its original technology by HDSL, see www.pairgain.com. 37. T1 price data was obtained by examining PUC tariffs, FCC tariffs, by interviewing users, by requesting T1 quotes for my own home office, by interviewing ISPs, and by interviewing LEC employees and resellers. Only one LEC, Pacific Telesis, provided average revenue data. The others refused or ignored my requests. 38. As it turns out this number, which I estimated independently, is also Pacific Telesis' estimate of its average T1 revenue per month per line. 39. See www.telechoice.com. 40. Personal interviews, networking equipment executives and technologists. 41. Confidential interviews, networking equipment executives, university engineering professors. 42. Confidential interviews, networking equipment executives. 43. Confidential interviews, LEC and networking equipment executives. 44. During the U.S. v. AT&T antitrust case, AT&T hired literally dozens of academic economists to testify that scale economies in telecommunications were so large that an entire national monopoly was technologically required to provide efficient service. 45. A recent survey by Jupiter Communications, a consulting firm, indicated that AOL has 8 million Internet customers, followed by Compuserve, MSN, and Prodigy in the low millions. The largest number of any telecommunications company was held by AT&T, with 800,000. The LECs often have less than a quarter million, and some have less than 50,000. Nynex, of course, has zero. 46. See Bellcore annual reports, and the graph of LEC R&D spending at the end of this document. 47. See for example web sites and annual reports of Compaq, Dell, and Hewlett-Packard. 48. See www.isoc.org and www.ietf.org, the web sites of the Internet Society and the IETF respectively. 49. All data in this section are derived from LEC annual reports, 10Ks, and proxy statements. All data are also shown in graphs at the end of this document. 50. This is a particularly sobering commentary on LEC behavior. When Netscape went public its valuation was about $1.5 billion. In late 1995 or early 1996 the total valuation of Netscape, Open Market, my own company Vermeer Technologies, Cybercash, Verisign, and the three largest Internet access providers COMBINED was less than $5 billion. U.S. West alone paid over $5 billion for one cable television company, and the LECs' combined investments in and purchases of CATV companies now clearly exceed $10 billion. 51. To his credit, Reed Hundt has made this point publicly. 52. See the Airtouch web site, www.airtouch.com. 53. Confidential personal interviews with LEC expert witnesses. 54. Interviews with LEC employees in Washington, DC offices. 55. Personal interview. 56. Personal interviews. 57. Personal interview, university professor. 58. Confidential interviews, senior government official. 59. Personal interviews, senior professors. 60. Personal interviews, senior professors, and confidential documents. 61. Personal interviews, senior professors. 62. Personal interviews, senior professors. 63. Personal interview, LEC expert witness. 64. Personal interviews, senior professors. 65. Personal interviews, senior professors. 66. Personal interviews, senior professors of economics, and author's literature search. 67. Ameritech press release, available at www.ameritech.com. 68. BellSouth press release, "A Message to Members of the Press Regarding a Miscommunication," February 27, 1997, available at www.bellsouth.com. 69. Confidential interview, telecommunications attorney. 70. Such claims and studies have now been produced by Bellcore, Bell Atlantic, and Pacific Telesis. The Pacific Telesis study can be found at www.pactel.com, link labelled "The White Paper." 71. For a detailed analysis of this issue, see Selwyn & Laszlo. 72. Virtually every LEC annual report prior to this year mentions the rise of second lines for data applications as a FAVORABLE development. 73. The Pacific Telesis White Paper actually states that access charges on ISPs are required so that the LECs will have enough money to create packet switched networks, which the Pacific Telesis study acknowledges are the correct technical solution. But the study does not seem to contemplate the possibility that such networks can or should be constructed by anyone other than the LECs. 74. This is not a minor detail. A very substantial fraction of current traffic and switching activity consists in calls with no answer, leaving voicemail, or fax messages. 75. Confidential interviews. 76. The numbers involved are sometimes very large. In a single year, GM announced 71,000 layoffs. DEC's workforce peaked at 110,000, and is now 60,000. IBM's workforce peaked at 400,000 and is now about 250,000. 77. The shareholder returns to declining incumbents have generally been very poor. In this sense, the stock market has treated them rationally and even with some foresight. For example, IBM's stock price peaked in 1987, had declined over 70% by its trough in 1993, and has still not returned to its 1987 level ten years later. The performance of major automobile, steel, machine tool, minicomputer, and textile companies has been similar. 78. GM fired its CEO only once, after announcing huge losses and 71,000 layoffs. IBM behaved similarly, firing John Akers only after announcing a $25 billion loss; the same was true for DEC's replacement of Ken Olsen by Robert Palmer. Many declining firms have NEVER fired a CEO, including most large steel companies. 79. The information in this section is taken from annual reports, 10Ks, proxy statements, and LEC web sites, several of which have biographies of officers, directors, and key executives. 80. See the two companies' web sites. 81. See www.ameritech.com plus Ameritech annual reports. 82. LEC proxy statements; sometimes available on web sites. 83. Confidential interviews. 84. For a discussion of this subject, see Kevin Werbach's "Digital Tornado," the FCC paper referred to previously, at www.fcc.gov. 85. For a copy of the full document contact the Internet Access Coalition, on the web at www.iac.org. For a summary of the IAC document, see www.itic.org/fcc_iac.html. 86. I am grateful to several telecommunications attorneys for explaining this situation. Any remaining errors are mine. 87. Personal interviews, FCC personnel. 88. Personal interviews. 89. See the Wall Street Journal, April 17, 1997, page B8. 90. This subject has been generally neglected by economic theory. However, for one excellent economic study of this effect in the context of pre-divestiture telecommunications, see Gerald Brock, "The Telecommunications Industry: The Dynamics of Industry Structure," Harvard University Press. 91. The document submitted to the FCC by the Internet Access Coalition makes essentially these recommendations. 92. A friend suggested this idea to me. 93. Personal interviews. STATISTICAL GRAPHS Following this page are the graphs referred to in the text concerning LEC total returns, R&D spending, and capital investment. 165