NET Institute* www.NETinst.org Working Paper #03-04 Revised February 2004 Assessing The Impact Of Internet Telephony On The Deployment Of Telecommunications Infrastructure by Martha Garcia-Murillo School of Information Studies Syracuse University * The Networks, Electronic Commerce, and Telecommunications (?NET?) Institute, http://www.NETinst.org, is a non-profit institution devoted to research on network industries, electronic commerce, telecommunications, the Internet, ?virtual networks? comprised of computers that share the same technical standard or operating system, and on network issues in general. Research Supported by The Net Institute September 2003 Garcia-Murillo, M. 2003, Assessing The Impact Of Internet Telephony On The Deployment Of Telecommunications Infrastructure, paper presented to NET Institute Conference at Stern/NYU, New York, NY Dec.12. Available at: http ://www.netinst.org/2003_conference.htm Assessing the impact of Intern et Telephony on the deployment of telecommunications infrastructure Martha Garcia-Murillo, Ph.D.1 Syracuse University School of Information Studies Revised version - February 2004 Abstract The purpose of this research is to determine whether Internet telephony has had a negative impact on a country?s teleco mmunications infrastructure. Using panel data the statistical analysis shows that call-back did not have a negative effect on infrastructure and high income countries benefited from that technology. Internet telephony shows a negative impact in high and lower middle income countries but at such a small scale that governments should not be concerned. Additionally the year variables show that for all income levels infrastructure has increased which means that if this technology will show any negative effects it will be in a gradual manner and should give carriers enough time to adjust their practices. Keywords: Regulation, Internet telephony, call-back, infrastructure, teledensity JEL Clasification: O14, O33, O38 1 Corresponding author Martha Garc?a-Murillo, Syracuse University, School of Information Studies, 4-206 Center for Science and Technology, Syracuse, NY 13244-4100,Tel: 315-443-4101, Fax: 315-443-5806, mgarciam@syr.edu. 3 Assessing the impact of Intern et telephony on the deployment of telecommunications infrastructure 1. Introduction The purpose of this paper is to show the impact of Internet telephony on telecommunications infrastructure. It is based on an empirical investigation of panel data with approximately 180 nations over a pe riod of four years. The focus is on Internet telephony because there are governments that have banned the provision of these services. There is some justification for their concerns and this is related to the low level of telecommunications infrastructure that they currently have. While in developed countries on average there are 50 phone lines per 100 inhabitants, the average in developing countries is 10 (IT U, 2002). Infrastructure in developing countries thus lags substantially behind that of the developed world and requires further development to foster economic activity. In the absence of basic and affordable services, the emergence of the Internet has allowed people without access to household or mobile phones to make international calls at prices 30% to 50% cheaper through Internet caf?s. Regulator s in countries with low levels of infrastructure are concerned that permitting Internet telephony will destroy the telecommunications operators? incentive for infrastructure investment. This is because in many less developed countries (L DCs) a large percentage of revenue is still generated from incoming and outgoing international calls, which subsidize local services. In the presence of such competition there is a fear that the carrier will not be able to cover its investment in the network and will thus have no desire to expand it. There are, nonetheless, studies (Lam , 1997; Frieden, 1998) that have argued that voice over the Internet (VOI) does not affect infrastructure because previous technologies such as call-back and international resale as well as political pressure from the United States and the World Trade Organization (W TO) have already reduced revenues from international traffic close to real costs. If this is the case, why are there still so many nations that prohibit the use of VOI? 4 Governments of developed countries decided in the late 1990s to allow the provision of VOI services. They may need to revise these decisions because the technology has evolved to offer quality comparab le to that of traditional carriers. The U.S. is currently reviewing this issue because VOI carriers do not have to pay international settlement or local access charges or contribute to universal service funds that traditional carriers do. At the same time the technology is becoming more prevalent and quality has significantly improved. Because there are many factors that a ffect infrastructure investment, the empirical model takes into consideration factors such as privatization of the incumbent carrier, level of competition, outgoing international telecommunications traffic, price of international calls, and control variables such as population density and income per capita. This research is aimed at providing policy guidelines. It is timely and relevant because it will help regulators concerned about the regulatory implications of this technology for the expansion of telecommuni cations infrastructure. This research will also contribute to the academic literature on creative destruction. New technologies have the potential of displacing outdated ones and thus bringing with them numerous benefits. 2 . Understanding the problem The use of the Internet to make telephone calls originated in 1994 with computer enthusiasts who, using special software, were able to send voice messages (Kelly et al., 2001). The technology rapi dly attracted attention more for its innovativeness than for its practical use. In developing countries it seemed to be more a curiosity than a real threat to the established traditional telecommunications operators. Internet telephony is a progression from the technologies that form the Internet. At first, making calls with computers required special software, a microphone, and a headset. VOI service provi ders now generally offer the software free of charge. As well, microphones are often integrated in laptop computers and basic headphones can be used without having echo interference from voices when both parties talk simultaneously. 5 In nearly ten years since the technology first emerged important developments have taken place. Internet telephony has been adopted differently in developed and less developed countries. In the developed world personal computers are hardly ever used to make phone calls. Young people and international students have used them to save money when calling family and friends. For the most part, the technology has aimed to optimize the tr aditional circuit switched networks that made an inefficient use of the infrastructure by having only two people use an entire line. Using IP packets, bandwidth is released for additional voice conversations and data services can also be added. In LDCs there have been limited effort s to upgrade infrastructure to use IP technology. Instead companies providing Internet telephony have perfected their systems to allow people to use the public Internet to make telephone calls. Equipment has been developed to add cards to computers so that telephone sets can be connected to make the calling experi ence more natural and similar to the experience of using tradit ional switching technology. Companies such as Net2Phone and Voca ltech have developed sophisticated packages aimed at Internet caf?s, pub lic calling centers, calling card providers, hotels, and marketing centers among other businesses. The solutions are intended to generate revenue for both the VOI facilita tor and these organizat ions. For Internet caf?s, for example, the marketing promoti on of Net2Phone highlights the ability of the new equipment to generate additional revenue for the establishment because users who want to make calls no longer need to be tied to the computer. They can use the handset and then allow other users to browse the Internet while the call is taking place. The caf? can thus obtain reve nue from two sources at the same time, effectively multiplying the use of a computer and connection to the Internet. Companies that offer these services advertise 90% to 95% discounts from the retail cost of a call through the traditional telecommunications carrier. Table 1 lists the prices for Net2Phone and traditional communications from a random sample of countries. 6 Table 1 Price comparison of a 3 minute call to the U.S. ($US) Country Traditional 2001* Extrapolated 2003** Internet 2003 Percentage difference Hong Kong, China 2.6 2.62 0.117 2,235 Laos P.D.R. 6.3 5.44 0.147 4,330 Azerbaijan 7.8 7.4 0.147 4,733 Niger 10.4 10.4 0.147 5,963 Mali 14.6 14.6 0.147 8,352 Paraguay 1.08 0.72 0.15 549 Portugal 1.02 1.02 0.15 576 Hungary 1.18 1.63 0.147 668 Belize 2.83 2.83 0.15 1,890 Jordan 2.67 0.46 0.147 1,820 Source: ITU World Telecommunications Development Indicators and Net2Phone *Latest year available ** Calc ulated from previous years 7 With the release of bandwidth that is possible with this technology companies are able to offer competitive rates. 2.1. The market In analyzing the effects of Inte rnet Telephony on the deployment of telecommunications infrastructure one needs to consider countries where ICT infrastructure is limited. In developed countries, where teledensity is greater than 50 lines per 100 inhabitants, there is little concern and incumbents do not consider it a threat. In these markets communication access is almost universal and rates are relatively low. Similarly the quality of the service provided through Internet telephony has been low enough that the price difference does not justify the inconvenience. In developing countries with limited infrastructure the situation is quite the reverse. There are places where people have to wait for several years before they can obtain a telephone line. In low income countries the average waiting time is 1.5 years with some countries like Kenya or Uk raine where people wait 8.1 and 7.4 years respectively (ITU World Telecommunications Indicators, 2002). It is thus not surprising that the Pakistani government made Internet telephony illegal, enforcing this through restrictive contracts that explicitly prohi bit companies from providing the service. In LDCs we encounter two types of problems. One is that phone lines are widely available but income is so low that few can afford them. In many countries the problem is not obtaining a line but rather the funds people have to keep it. In these countries the problem is not the lack of infrastructure but poverty. Another situation is where the infrastructure does not exist and people want access to these services. Of the 211 countries that are li sted in the ITU?s World Telecommunications Indicators, 103 have wa iting lists. The longest list was that of Russia with more than 6 million people wait ing for a line, followed by Syria, India, and Ukraine, which had between two and three million people waiting in 2000. Countries with waiting lists also tend to be poor, as 37% of those are in the low or low middle income category. When people are unable to obtain a phone line and 8 rates to call abroad are artificially high there is a strong incentive for them to try inexpensive alternatives. In the mid 1980s i ndividuals filled this need by setting up clandestine operations that allowed people to make illegal telephone calls at a lower expense. Soon afterward call-back services became popular even though many nations prohibited them. Internet telephony is thus another alternative where, in spite of the inconvenience and lower quality, people are willing to use it. The alternatives are either not available or not affordable. This is why the technology has become much more popular in LDCs than in developed countries. 2.2. Incentives for carriers It is not only the governments of some countries that have prohibited the use of the public Internet for voice. Communication carriers themselves may have justifiable incentives to pr ohibit its use. The technology can affect these carriers? revenues significantly. While in the 1950s international telephone services amounted to only 20% of a carrier?s revenue, in 2000 this corresponded to 95% (Sharifi, 2001, p. 315). International incomi ng and outgoing telephone traffic is highly profitable, particularly in countries where artificially high accounting and settlement rates lead to substantial hard currency remittances from foreign carriers. Revenue from data is lower. ISP carriers that lease the infrastructure are not obliged to pay per minute charges to the carrier for interconnection. They also do not have to pay settlement rates for international traffic, as recommended by the World Trade Organization. This means that an ISP providing this type of service can generate additional revenue from voice traffic without having to pay extra to the carrier that leases the infrastructure to it. For people who have Internet access at home they can place international calls through their dial-up connection to the Internet paying only for a local call and the much lower rate that Internet telephony carriers charge, for which the telephone company does not obtain any additional revenue. In most countries ISPs are not obligated to contribute to universal service initiatives. This means that Internet telephony may have a negative effect on traditional carriers? revenue. The scenario is different in developed than in developing countries. When VOI first emerged, governments had to decide if they would allow it or not. If they 9 were to do so they would need to determine if they could regulate it. In developed countries regulators considered that VOI ha d such primitive features that it could not considered voice telephony (McKnight and Le ida, 2001, p. 213) and that it posed no threat to current carriers. In spite of these assertions there was concern among some observers about the technology. In the United States some members of Congress felt that diversion of traffic to the Internet and lack of contribution from ISPs could endanger universal service (Weinberg, 2001, p. 336). Some Federal Communications Commission (FCC) officials recognized that the exemption could create an incentive to shift traffic to IP networks (Weinberg, 2001, p. 338). When the technology first appeared a group of operators created an association called ACTA to lobby the government for the regulation of Internet telephony. In spite of ACTA pressure, the FCC decided not to intervene. In Canada the concern was about contributions for the continued investment in communication networks. At the time the government decided not to require contributions from ISPs but left the door open for future consideration. In the European Union, the European Commission?s Directorate General for Competition provided a series of guidelines to determine if Internet telephony should be regulated or not. The decision was to not regulate it. The main factor in the set of guidelines was the specification regarding quality. For a service to be regarded as voice telephony it had to be real time. As Internet telephony at that time experienced substantial delay, it did not meet this criterion. In Japan th e provision of these services was allowed but required appr oval of the Ministry of Post and Telecommunication (MPT). In spite of their initial opposition, carriers in developed countries determined that the technology had benefits and they adopted IP protocols in their traditional networks. The objective has been to optimize the use of these lines. In LDCs the situation is different. L ack of lines has made voice over the Internet quite attractive. In Tegucigalpa an Internet caf? in the main mall is even considered a tourist attraction because of the number of people who are waiting in line to make an Internet call in spite of the fact that the service is prohibited. Another factor contributing to the wider use of this technology are the much higher prices for 10 international calls that still exist in many of these countries. The more extensive use of Internet telephony together with the reduced revenue that traditional carriers are able to obtain from this service have prompted many carriers to adopt measures that block this activity. Some have lobbied governments to regulate or prohibit Internet telephony. Monopoly carriers are unwilling to negotiate agreements with ISPs that provide Internet telephony and they can also blocks ports that are used to route voice traffic. One could argue that it is foolish of these organizations to oppose the provision of these services considering that voice traffic is growing at a considerably slower rate than data traffic. Carriers in these countries do not see it this way because they do not experience much data traffic ye t and would like to delay for as long as possible the reduction in their revenues resulting from lower voice traffic. Additionally, the existing revenues of carri ers are insufficient to upgrade their obsolete and limited infrastructure. These companies require not only the revenues that they may be obtaining from artificially high rates but also direct support from governments, if they are still state owned, to provide more advanced communication services to the population. These carriers, unlike their developed world counterparts, do not have the resources to upgrade their infrastructure to install equipment that can take advantage of IP technologies. These countries see Internet telephony as a direct competitor and threat. We will explore be low if the fears of these carriers and governments are justified. 2.3 Why prohibit Internet telephony? The previous section described the reasons why carriers in LDCs would not like ISPs to offer Internet telephony. There are also legal constraints. A government, for example, might want a new technology to compete directly with the incumbent carrier as a way of introducing competition to the market and to motivate the traditional operator to upgrade its network. The problem is that many of these governments made concession agreements during the privatization process under which they promised companies a certain number of exclusive services to make the offer more attractive. They thus were able to sell the operator but are now unable to allow voice telephony through the Internet because this would violate the agreement 11 with the privatized operator. There are othe r countries, such as Costa Rica, that have constitutional limitations to opening markets for Internet telephony. 3 . Previous research contributions The main concern over the impact of Internet telephony is with the potentially negative impact that the technology can have on operators? revenues, which are derived primarily from international traffic. This section reviews previous research that has analyzed the impact that technologies and political pressure from abroad have on international revenues. 3.1. The impact of information technologies on international revenues Internet telephony is not the first technology that has posed a threat to revenue for carriers. There have been others such as call-back and international resale that were expected by many to negatively affect flows of capital for a country?s investment in infrastructure. Scholars have argued that call-back, international resale, and whipsawing2 have put pressure on accounting rates. They expect that ar bitrage opportunities would thus disappear (Lam, 1997; Friede n, 1998; Clark, 2001). One could thus expect that these services would have al ready reduced accounting rates to the point that Internet telephony should thus have no impact. This means that if there were a negative impact on infrastructure it would have happened in the early 1990s when call-back services were entering international telecommunication markets. Similarly, statistics from the FCC about international payments to foreign carriers indicate that settlement payments have been reduced from 1995 to 2001 by more than two thirds. 3 3.2. The impact of international pressure In 1997 there was also pressure from both the United States and the WTO to reduce accounting rates. The U.S. was concerned about the large payments that its 2 This allows a person in Zimbabwe, for example, to make an international phone call to South Africa, but make it seem like the call originated in the U.S. 3 Statistics of Communications Common Carriers, FCC, http://www.fcc.gov/Bureaus/Common_Carrier/Reports/FCC-State_Link/socc.html , accessed May 4, 2003. 12 national operators were making to foreign carriers, thus leading to a large trade imbalance. This prompted the FCC to develop a series of benchmarks regarding the maximum amounts that U.S. carriers were a llowed to pay foreign operators. At that time there was strong criticism on the part of foreign governments who complained to the U.S. on two grounds: that there was a lack of U.S. jurisdiction, and that such measures could severely cripple investment in infrastructure in poor countries. The FCC argued that they have the right to determine rates for their carriers under both domestic and international law (Cowhey, 1998, p. 905). Aware of the potential impact on investment, the FCC set its benchmarks for LDCs at a higher level and they were also given a longer time to comply (Cowhey, 1998, p. 906). The International Bureau of the FCC conducte d a study to determine the impact of international settlement and network build out. It concluded that there was not a statistically significant relationship (Cowhey, 1998, p. 907). The imposition of benchmarks by the U.S. in 1997 should also ha ve negatively affected foreign carrier revenues and thus investment in infrastructure. Pressure also came from the WTO where, in 1997, 69 countries developed guidelines for the liberaliza tion of telecommunications markets. Among the series of commitments was the introduction of competition during a specified period. These measures were thus expected to reduce the trade imbalances that were generated by monopolized markets and artificially high international accounting rates. Of the studies reviewed on this subj ect only the FCC study mentioned by Cowhey (1998) provided empirical evidence. We thus cannot assert with certainty that these services indeed reduced international accounting rates and thus revenues for LDC carriers. 3.3. The impact of the Internet on information infrastructure In spite of the work from scholars that theorized that call-back, resale, and international pressure would result in lower revenues for foreign carriers, there was considerable debate about the impact of Internet telephony when the technology first emerged in 1994 (ITU, 2001). Scholars that ha ve written on the subject fall on both sides of the issue. Cawley (1997) predicte d that Internet telephony does not lead to lower investment but instead motivates the upgrade and greater adoption of other 13 telecommunication technologies. Mason (19 98) argued that In ternet telephony should have no effects on accounting revenues because this technology is so inferior to alternative technologies that he finds it unlikely that people would prefer that medium over others that are more convenient and of higher quality than an Internet connection. He argues that phone based se rvices such as call-back were better alternatives and that VOI would not have an impact on investment. Similarly the International Telecommunications Union (ITU) in its report about Internet Telephony (2001) concluded that other tec hnologies have already put substantial pressure on the revenues that operators could generate. The authors did not believe that VOI alone could have a great impact on investment either. In the United States carriers were concerned about a decline in their revenues. The ACTA association of carriers argued that this would lead to increased traffic on a network that was only designed to carry short voice conversations. In their lobbying efforts they stated that the upgrade of their networks to carry this amount of traffic would cost them millions of dollars (Moore, 1997). They wanted to be compensated in the form of access charges. Some scholars believe that Internet telephony threatens the viability of the international accounting system (McGarty and McKnight, 2001). We now know that in the U.S. the growth of the Internet led to greater investment in telecommunications and increased revenues. While this success story has been repeated in other developed countries this has not been the case in less developed ones. There are several factors that inhibit progress in LDCs: (1) lack of infrastructure to carry even basic voice; (2) low personal income, which limits the amount that households spend on communication services; and (3) obsolete communications equipment. It is thus in LDCs where Internet telephony could potentially have a negative impact on telecommunications investment. Previous technologies such as call-back may have already had a negative impact. Although none of the studies reviewed provide empirical evidence, the general consensus appears to be that Internet telephony would not have an effect on telecommunications infrastructure. In spite of this belief, there are approximately 40 countries that have reported that they prohibit VOI. Most countries that forbid VOI 14 have either state monopolies or privatized enterprises with continued state participation. Only five of the countries that prohibit VOI have a privatized telecommunications carrier. Why is this? If revenues are already low as a result of the implementation of other technologies as well as international pressure, governments should not be as concerned about the impact of VOI. The purpose of this research is to determine if these concerns have merit by testing empirically if Internet telephony is inhibiting infrastructure deployment. As the technology has improved substantially, some now believe that regulation is necessary. Kiser and Collins (2003) argue that Internet telephony is now a technology that is more widely used and, in spite of this it still faces no regulatory burden. Operators providing Internet telephony are not obliged to pay settlement charges for international voice communication and they do not contribute to universal service funds. They thus suspected that regulators around the world would revisit this issue given that they perceive the advantage of these carriers as unfair. 4 . Hypothesis development The metric that is used in this study to measure telecommunications infrastructure is total number of telephone lines per 100 inhabitants, which includes both wired and wireless lines. Multiple factors affect the deployment of a telecommunications infrastructure and these must be considered in addition to Internet telephony. Scholars have identified factors within several categories. Table 1 shows those that have been presented by authors of other studies and how they were operationalized in this study. The table is divi ded in three categories. Their origin is from Bernt and Weiss (1993) who iden tified four categories: regulatory, organizational, economic, and technical. A more recent study by Mbarika et al. (2002) added four more categories: financial, managerial, political, and geographical. Table 2 includes only those categories and variables for which data is available for the period from 1999 to 2002. 15 Table 2 Included Variables SOURCE FACTOR VARIABLE Economic Mbarika (1999) GDP GDP per capita (1995 $U.S.) Yatrakis (1992) Volume of trad e Trade a as percentage of GDP CCITT (1965,1995) as reported by Mbarika Mbarika (1999) Tertiary sectors % employment in agriculture % employment in manufacturing % employment in services ITU (1994) Inadequate private sector involvement Level of Competition (Local, Int?l and ISPs) 1=monopoly 2=duopoly 3=partial competition 4=full competition Maitland Commission (1984) Wellenius (1989) ITU (1994) Foreign capital Financing from abroad % GDP Organizational Wellenius (1989) ITU (1994) Ramamurti (1996) Ros (1999) Autonomy from government/Privatization Dummies 1=state owned, 2= semi-privatized, 3=privatized Regulatory/Policy/Political ITU (1994) Lack of a Universal Service Policy Dummies: Universal Service Policy 1=Yes Kirunda-Kkivenjinja (1995) Roads, Sewa ge, Water Roads (network size in km) ITU (2001) Bernard (1994) Lam (1997) Frieden (1998) Clark 2001). Call-back, international resale, and whipsawing Dummy: Call-back allowed by the government 1=yes Internet Telephony Dummy: Internet telephony allowed by government 1=yes Based on the research that other schola rs have done on Internet telephony, we know that there is still disagreement with respect to the effect of voice on the Internet on the further development of the telecommunications infrastructure. Scholars tend to believe that this technology should have no effect on further investment in infrastructure but private sector officials as well as regulators in many countries believe the contrary. This study developed hypotheses based on arguments made in the literature with respect to previous technology, international pressure, and economic factors. Scholars that have studied the effect of voice over the Internet believe that the call- back, international resale, and whipsawing effects that emerged in the late 1980s would have already reduced the revenues of carriers. Even though this is likely to be the case, there were countries that also banned call-back and whipsawing. One could 16 thus potentially argue that in those countries revenue may not have fallen as much as these scholars expected. These types of services may, however, have been so difficult to monitor and enforce that even with government opposition, they were provided, thus negatively affecting revenues of carriers. This study includes dummy variables for both call-back services and Internet telephony. H1: After controlling for call-back, telecommunications infrastructure (teledensity) does not decrease as a result of Internet telephony. As outlined above, there was domestic pressure to reduce settlement rates for international traffic. Unfortunately there is no publicly available data about these rates and thus it is not possible to evaluate the impact that international pressure had on the costs of transborder calls, the revenues of carriers, and infrastructure expansion or improvement. There are, however, other more subtle international factors that can equally affect the prices of telecommunications services and a carriers? telecommunications revenue. These include the amount of trade activity, the level of employment in the service and manufacturing sectors, the amount of foreign capital that governments receive, as well as the amount in minutes of telecommunications traffic from other countries. The following hypothesis is based on assuming these factors as proxies for international pressure. H2: After controlling for the sizes of the service sector , trade, foreign aid, and international traffic, the presence of Internet telephony has no impact on infrastructure. The economic circumstances of a country can affect investment in telecommunication networks. There is a widely held belief that potentially low levels of demand in the poorest countries will mean that investment in these services is unprofitable. Under these circumstances many governments are unable to attract enough private sector capital. In Guatemala where a Universal Service fund was established, the organizati on responsible for the disbursement of subsidies was unable to attract any providers of services. Because this was a competitive process 17 no company was willing to do the market research and then fail to obtain the subsidy during the bidding process. Lack of income influences both privat e and public sector decisions. Without this type of information investors will be unwilling to take risks. Taking into consideration that economic factors play a role in a company?s wi llingness to invest we would expect: H3: After controlling for the income level of the population, Internet telephony has no impact on the telecommunications infrastructure of a country. 5 . Data analysis This section examines the impact that technological, regulatory, and economic factors have on the deployment of telecommunications infrastructure. The first part presents descriptive statistics to help understand the general circumstances of the market, the carriers, and the regulatory restrictions that were put on these technologies. It will help identify the countries where call-back was available. It also identifies the market conditions of those countries with respect to the ownership status of the main telecommunications operator as well as the level of competition to determine if these factors were related to the decision to allow or prohibit the use of technologies such as call-back and Internet telephony. The second part presents a regression analysis of the variables of interest. 5.1 Descriptive statistics Technologies are likely to have different effects on countries depending on level of income. Table 3 shows descriptive statistics for each of the factors included in the statistical analysis by level of income. 18 Table 3 Descriptive statistics: means Variable Low income Medium low income Medium high income High income Telecom revenue 58,016,035 166,575,630 566,881,021 1,557,223,088 GDP per capita 379 1,409 4,782 19,437 Trade 68 90 93 105 Employment in services 46 54 55 63 Employment in agriculture 43 26 15 5 Employment in industry 16 22 26 27 Financing from abroad 2.8 2.2 2.0 1.9 International incoming telephone traffic 28,234,918 75,575,295 142,508,186 338,407,044 Cost of 3 min call to USA 4.45 5.3 3.82 5.44 Population 9,480,238 4,143,411 4,095,379 2,995,644 Number of countries 61 50 33 43 We can see from this table that higher income economies also have higher levels of trade, greater levels of service and industry workforce, and a smaller agricultural sector. There is more traffic flowing into higher income countries and a call to the U.S. is cheaper as income increases. Scholars have argued that technologies such as call-back, which came before Internet telephony, had a negative impact on the revenues of the telecommunications operators. In 1999 and 2000 the ITU asked te lecommunications regulators if they allowed or prohibited call-back services. This information is summarized in Tables 5 and 5. The majority of countries did not allo w it. Call-back was less likely to affect the revenue of operators in places where it was not permitted. The large number of countries that did not allow call-back services in 2000 is surprising considering that carriers had many years to lower prices to meet this competition. It is also surprising because in 2000 many countries had already privatized and liberalized their telecommunications industry and thus imposing restrictions on these services would not reduce rates. Restrictions on call-back services and subsequently on Internet telephony may have been related to the terms of privatizat ion that were granted to 19 the company. In many of these cases private operators were given several years of monopoly status. Tables 4 and 5 present the percentage of countries that prohibited call-back services and compares these based on the ownership of the incumbent and the level of competition. Table 4 shows that in those countries where the market is monopolized 69.5% of regulators did not al low call-back services while 67.4% of countries that experienced either full or partial competition permitted call-back services. Similarly in 54.3% of the count ries that did not allow call-back, the telecommunications operator remained under state ownership while 84% of those countries that allowed the services where either privatized or partially privatized. This could be an indication of government concerns about having a decline in international revenues as a result of this service. It could also be an indication of the concession agreements that were done when privatization took place. Tables 4 and 5 Percentage of countries that allow call-back services according to the ownership of the incumbent and level of local competition Call-back Call-back Not allowed Allowed Not allowed Allowed Monopoly 69.5 30.2 State owned 54.3 16.3 Duopoly 5.5 2.3 Privatized 5.4 32.6 Partial competition 10.2 55.8 Partially privatized 40.3 51.2 Competition 14.1 11.6 N = 172 N = 171 Pearson chi2(4) = 40.5013 Pr = 0.000 Pearson chi2(2) = 30.7213 Pr = 0.000 Table 9 lists the countries that proh ibit Internet telephony by income level. Countries of the European Union are not included in this table because of a directive that permits the service because it does not meet the standards of voice telephony. Tables 6 and 7 present the percentage of countries that prohibited this service according to the level of competition and the ownership of the incumbent. The results are similar to those of call-back services. 78% of the countries that decided to prohibit Internet telephony also have monopolized markets and the carriers are state owned. In contrast, 73.5% of the countri es that allowed these services also experience either full or partial compe tition. Similarly, 53% of countries where Internet telephony was banned still maintained a state carrier under state ownership 20 while of those countries where the service was allowed 73% had been privatized or partially privatized. In both cases the ?2 is significant. Tables 6 and 7 Percentage of countries that allow Internet telephony services according to the ownership of the incumbent and level of local competition Internet Telephony Internet Telephony Prohibited Allowed Prohibited Allowed Monopoly 78.18 26.53 State owned 52.73 26.53 Competition 5.45 48.98 Privatized 7.27 22.45 Partial competition 16.36 24.49 Partially privatized 40 51.02 N = 104 N = 104 Pearson chi2(2) = 32.5957 Pr = 0.000 Pearson chi2(2) = 9.2380 Pr = 0.010 Table 9 Countries that prohibited Internet telephony services by income level (2000) Low income countries Lower middle income countries Upper-middle income countries High income countries Eritrea Liberia Kenya Ethiopia Benin Nigeria Cote d?Ivoire Senegal Mozambique Guinea Cameroon Central African Rep. Nicaragua Azerbaijan Armenia Nepal Myanmar Indonesia India Mongolia Cambodia Pakistan Comoros Yemen Namibia Swaziland Cuba Ecuador Belize Paraguay Peru Romania Latvia Lithuania Albania Philippines Thailand Tunisia Jordan Syria Morocco Botswana Gabon Seychelles Argentina Trinidad and Tobago Panama Turkey Croatia Estonia Cyprus Israel Kuwait Qatar Total = 24 Total = 17 Total = 9 Total = 4 Source: Telecommunications Regulation Database, ITU 2002 Table 8 Countries that prohibited call-back services by income level (2000) Low income countries Lower middle income countries Upper-middle income countries High income countries Haiti Georgia Moldova Tajikistan Kyrgyzstan Uzbekistan Armenia Azerbaijan Ukraine Indonesia Nepal Lao P.D.R. Viet Nam Mongolia India Bangladesh Afghanistan Cambodia Pakistan Myanmar Comoros Yemen Sudan Mauritania Eritrea Cameroon Burundi Burkina Faso Cote d?Ivoire Senegal Benin Congo Chad Zambia Mozambique Central African Rep. Congo (Democratic Republic) Uganda Guinea Tanzania Ghana Gambia Niger Malawi Kenya Sierra Leone Nigeria Zimbabwe Ethiopia Rwanda Mali Madagascar Cape Verde Swaziland Equatorial Guinea Jamaica Colombia Ecuador Belize Honduras Bolivia Suriname Peru Cuba Bosnia and Herzegovina T.F.Y.R. Macedonia Lithuania Kazakhstan Bulgaria Latvia Belarus Romania Albania Fiji China Maldives Tonga Iran (Islamic Rep. of) Sri Lanka Thailand Papua New Guinea Philippines Marshall Islands Jordan Morocco Syria Egypt Tunisia Algeria Djibouti Mauritius Seychelles South Africa Gabon Botswana Antigua and Barbuda Dominica Uruguay Mexico Panama Grenada Costa Rica Venezuela Brazil Hungary Croatia Slovak Republic Turkey Czech Republic Poland Malaysia Oman Saudi Arabia Libya Bahrain Lebanon Bahamas Slovenia Cyprus Greece Malta Israel Brunei Darussalam Kuwait Qatar United Arab Emirates Total = 52 Total = 38 Total = 26 Total = 10 One of the main obstacles in doing empirical international research is in many cases the lack of complete data. This study is not the exception. Although institutions like the World Bank, the ITU, and th e IMF collect statistics from more than 200 countries, much of the information is missing. It has been well documented in the statistics field that doing analysis of only those cases that have complete data can lead to biased results. In this study the initial number of countries was 213. The list was reduced to 188 be cause many of those countries had data on only two or three of the variables that were included in the analysis. The number of countries eliminated should not cause bias in these results because they are either very small economies or are going through major trans itions. Examples of countries that were eliminated include Afghanist an, Andorra, Liechtenstein, Iraq, Mayotte, and Andorra. This does n ot mean that there was complete data for the rest of them. There were some variables for which many data points were missing. Table 10 presents the list of variables and the number of observations that were available. The total number of observations was thus 11 28 because data was collected for 6 y ears for each of these countries. Table 10 Percentage of missing data per variable Variable No. of complete observations % Missing GDP per capita 1041 8% Trade 979 13% Employment in services 394 65% Employment in agriculture 414 63% Employment in industry 416 63% Competition in international telecom service 916 19% Intlfinance 488 57% Number of Faults 666 41% Privatizatization 885 22% Universal Service Program 621 45% Roads 889 21% 23 VOI regulation 617 45% Level of ISP competition 258 77% Call-back regulation 987 13% Telecommunication revenue 1037 8% Incoming international traffic 913 19% Outgoing international traffic 1070 5% Price of a 3 min call to the U.S. 772 32% Teledensity 1116 1% Population 1100 2% While the models could not have included all the variables there would have been some for which less than 300 observations would have been included. Because missing variables is a common problem, scholars have devised techniques to calculate the miss ing variables from existing ones. Some methods that have been found to be inadequate (L ittle & Rubin, 1987; Graham, Hofer & Piccini n, 1994) include ad hoc ones, such as using the mean from the observ ed values, extrapolating from th e latest value available, also known as last-observation-carried-forward (LOCF), and the use of regres sion analysis to estimate values. For this study these were particularly inappropriate because for some variables there was missing data for two or three consecutive years. This meant that the same value would have been given for the three years in the case of any of the ad hoc or regression methods. The approach used in this project was that of multiple imputation where missing data is generated simultaneously from all the available variables for all the observations and years. The mathematical algorithms that are necessary for multiple imputation are now easier to use thanks to advances in computing. In this project I used the Am elia program developed by Honaker, Joseph, and King (2000). Infrastructure is measured by the number of telephone lines per 100 inhab itants. Table 11 presents the results per income level of the countries in the regression analysis. It uses a fixed effects model that includes the year dummies to capture the changes in infrastructure deployment. The fixed model was selected becau se of the recognition that the way technology affects each country will 24 depend on its specific circumstances. We are thus interested in the differences within each country and how these technologies affect the deployment of infrastructure. The analysis used the following model. )()()()(lnln 43210 ltemplagricuesemplservictradegdppcteledesity ????? ++++= )(ln)(ln)int(ln)(int)( 98765 usacalltrafinlfinancecompdryemplindust ????? +++++ )()()()()(ln 1413121110 voipunivservdfullprivparcprivapopulation ????? +++++ )00()99()98()97()96()( 201918171615 yyyyycallback ?????? ++++++ (1) Table 11 Effect of market, organizational, and regulatory conditions on telecommunications infrastructure as measured by teledensity Teledensity All countries Low income Low middle income Upper middle income High income Telecom revenue 1.046* 0.948 1.033 1.012 1.099 GDP per capita 1.328*** 1.613*** 0.949 1.094 1.132** Trade -0.001** -0.001 -0.001* 0.000 -0.002*** Employment in services -0.001 -0.002 0.001 0.000 -0.002 Employment in industry 0.007*** 0.010*** 0.005** 0.006*** 0.002 International Long distance competition 0.028 0.069 0.097** 0.052 -0.026 Financing from abroad 1.026* 1.051 0.995 1.043*** 1.006*** International incoming telephone traffic 1.162*** 1.182*** 1.115*** 1.194 1.155 Cost of 3 min call to USA 1.006 1.001 0.997 0.984 0.992 25 Ln Population 0.897*** 1.199 5.812*** 0.801*** 0.835*** Incumbent privatized 0.062* 0.116* 0.058* 0.017 0.006 Universal Service Fund 0.041* -0.038 0.005 0.069** 0.057 Internet Telephony Policy -0.018 0.031 -0.057 -0.026 -0.112** Call-back policy 0.065 0.152 0.035 -0.084 0.102* Year 96 0.075*** 0.056 0.079** 0.097*** 0.055 Year 97 0.157*** 0.125* 0.174*** 0.229*** 0.110** Year 98 0.253*** 0.162* 0.237*** 0.363*** 0.243*** Year 99 0.383*** 0.224** 0.337*** 0.523*** 0.434*** Year 00 0.535*** 0.340*** 0.491*** 0.720*** 0.598*** _cons -2.132*** -7.486 -27.225*** 1.806 0.839 N 1128 366 300 198 258 Within R-sq 0.62 0.55 0.74 0.89 0.69 *Significant at .10 ** Significant at .05 One interesting finding is the way economic factors affect countries in the four income levels. The richer the country is the less that economic factors matter. Telecommunications revenues do not seem to have much effect on infrastructure. For high income and lower middle income countries revenues were non-significant at the 10% le vel but significant at higher levels. This means that revenues are invested in infrastructure in some countries. GDP per capita is a significant variable and, as expected, higher in comes have a positive effect on infrastructure. Dividing the countries by income levels makes this factor non-significant because there is little variability among them. Employment in industry has a positive impact on teledensity. As stated above, scholars have found a positive relationship between the service economy and teledensity. Employment in services is not significant with the exception of low i ncome countries. All other income levels were significant at a lower confidence level of 15%. 26 Having large incoming intern ational traffic also positively affects the deployment of infrastructure. This is truer in lower income countries where it is clear that hard currency revenue has an impact. Higher income countries generally originate more international calls than they receive. Ha ving a privatized carrier also has a positiv e impact on a country?s telecom infrastruc ture particularly in lower income countries, where privatizations have been shown to have a positive effect. With respect to the variables of interest in this research, specifically call-back and Internet telephony, neither was significant. This means that they do not have either a positive or a negative effect on the telecom infrastructure of a country. The variable corresponding to the level of competition is significant only for low middle income countries. This indicates that competition has a positive effect on infrastructure. Because devel oping countries have liberalized their markets more slowly th an rich economies, the effects of competition are being felt as they have increased teledensity over the years of the study. In many high income countries, however, competition was introduced earlier and teledensity is quite high. Thus, any additional competition m ay have an effect on prices but not necessarily on the further development of basic telecommunications infrastructure. The amount of finance received from abroad appears to have an effect in upper income countries only. It is possible that financing from abroad is not used for telecommunication infrastructure projects in th e lowest income countries. The price of a three minute call to the U.S. does not seem to have any effect on infrastructure. In this model universal service was coded as one if the country had a universal service fund and zero if it did not. Universal service funds are still relatively new and thus there are only a few countries that have adopted them. The variable is significant for upper middle income countries, which are also those that have more broadly adopted these programs. The two variables of greatest interest in this study are call-back and Internet telephony. Neither is significant for all countries except the higher income ones. It is not surp rising that call-back has no effect as the technology is no longer new and any negative effects would have been felt in the years before the ones included here. The effects of Internet telephony nonetheless should have been 27 felt in the years of the study because the technology was first implemented in the mid-1990s. It is interesting to see how the effect of Internet telephony is negative while the effect of call-back is positive in high income countries. The negative effect of Internet telephony is not surprising and could be related to the use of these technologies by people who want to call their home countries without using traditional carriers, which have to pay high settlement charges. Call-back services were favorable to countries such as the U.S. that have lower settlement rates. This thus led to people in high rate countries to use these services, which generated greater traffic for carriers that offered lower rates. The fact that the coefficient is not significant also does not suggest that it has a positive effect. This is important because some people argue that allowing these technologies in the market will force the incumbent carrier to become more efficient because it is a form of competition. If this had been the case then the coefficient would have been significant with a positive sign. The most likely event is that the level of use is too low to have any impact on infrastructure. Both call-back services and Internet telephony are expected to have a negative impact on the revenues of carriers. This is represented in the model below. Table 12 show s the regression results for that variable. It is also organized by income level. )()int(ln)(int)(lnln 43210 privatelfinancecompdgdppctelrev ????? ++++= )(ln)(ln)()()( 98765 usacalltrafincallbackvoipunivservd ????? +++++ )00()99()98()97()96()(ln 151413121110 yyyyypopulation ?????? ++++++ (2) 28 Table 12 Effect of market, organizational, and regulatory conditions on telecommunications revenue Telecommunications Revenue All countries Low income Low middle income Upper middle income High income GDP per capita 2.260*** 2.703*** 0.725 3.025*** 2.078*** International Long distance competition 0.052 0.144 -0.024 0.044 0.017 Financing from abroad 0.970* 0.939** 0.946 0.970 0.988 International incoming telephone traffic 1.327*** 1.232*** 1.376*** 1.368*** 1.234*** Cost of 3 min call to USA 1.001 0.997 1.010 1.010 0.985 Ln Population 1.744*** 0.832 0.556 1.677*** 1.829*** Privatized carrier 0.016 -0.083 0.132 -0.058 0.011 Universal Service Fund -0.027 -0.130** 0.071 0.017 -0.037 Internet Telephony Policy 0.047 0.046 0.086 -0.027 0.023 Call-back policy 0.110* -0.091 -0.151 -0.047 0.322*** Year 96 0.046** 0.058 0.102 0.100 0.024 Year 97 0.033** 0.053 0.089 0.090 0.088 Year 98 0.083* 0.137 0.140 0.219*** 0.091 Year 99 0.081* 0.002 0.202* 0.257*** 0.198*** Year 00 0.096* 0.134 0.244* 0.287*** 0.103 _cons -0.809 11.425 24.343 -3.074 0.628 N 1128 366 300 198 258 Within R2 0.57 0.74 0.22 0.50 0.75 29 *Significant at .10 ** Significant at .05 The revenue of carriers seems to be affected only by the amount of traffic that goes through the network and the income level of the population. The implementation of universal service policies does not seem to have an effect on revenues. When it does, as is the case in low income countries, the impact is negative. The level of revenues from one year to the next seems to vary as some are able to predict revenue for that year and some are not. This may be a reflection of the volatility of the industry after privatization and liberalization. The two important variables for this study, call-back and Internet telephony, are not significant either except for high income countries where this technology appears to have a positive effect. These two technologies may be used at such low levels compared to the use of traditional communications that it simply has not had an impact on these carriers? revenues. The final model used in this analysis is a simultaneous equation estimat ion. This was necessary because the impact that any of these technologies can have on infrastructure is related to how they affect revenues. Thus I used a two stage least square proc edure for panel data. )()()()(lnln 43210 ltemplagricuesemplservictradetelrevtden ????? ++++= )()()(ln)int(ln)( 98765 voipunivservdpopulationlfinanceryemplindust ????? +++++ )00()99()98()97()96()( 151413121110 yyyyycallback ?????? ++++++ (3) Where: 30 )()()(int)(lnln 43210 fullprivparcprivacompdgdppctelrev ????? ++++= )(ln)(ln 65 usacalltrafin ?? ++ (4) Table 13 Simultaneous equation model of the effects of market, organizational, and regulatory conditions on telecommunications infrastructure with telecom revenue as a nested equation Teledensity All countries Low income Low middle income Upper middle income High income Telecom revenue 1.888*** 1.596*** 1.471*** 1.181** 1.389*** Trade 0.000 0.001 -0.001 0.000 -0.002*** Employment in services 0.000 -0.002 0.003** 0.000 -0.001 Employment in industry 0.009*** 0.011*** 0.006** 0.006*** 0.002 Financing from abroad1.044*** 1.065* 1.016 1.044*** 1.004 Ln Population 0.583*** 1.554 7.079*** 0.765*** 0.764*** Universal Service Fund 0.048 0.028 -0.008 0.079*** 0.067* Internet Telephony Policy -0.083** 0.011 -0.091** -0.036 -0.108* Call-back policy -0.045 0.195 0.072 -0.060 0.044 Year 96 0.036 0.033 0.046 0.103*** 0.053 Year 97 0.117*** 0.100 0.148*** 0.243*** 0.099** Year 98 0.195*** 0.140 0.201*** 0.374*** 0.231*** Year 99 0.309*** 0.260** 0.287*** 0.541*** 0.404*** Year 00 0.435*** 0.311** 0.433*** 0.756*** 0.616*** _cons -2.014*** -15.440 -35.394*** 3.388 1.190 Obs 1128 366 300 198 258 R-sqr 0.45 0.41 0.56 0.87 0.66 *Significant at .10 ** Significant at .05 31 The results of this last model are similar to those obtained in the separate models. Call-back is not significant and Internet telephony when allowed appear to have an negative impact on infrastructure for high and low middle income countries but the impact is too small to be of concern to governments. Infrastructure increases every year and all of the coefficients are significant. If any of the two technologies had had a negative impact it would have showed some reduction or perhaps even a negative number. There were several problems that needed to be overcome in this research. He teroskedasticity was present, although the model was corrected by the use of a robust regression. Similarly multicollinearity problems were solved by corrections in the model and the elimination of variables because the original design was overspecified. Although the problem of missing data was solved by the use of multiple imputation techniques, the resulting numbers can still result in inaccuracies, even though this method is the best available in statistical research. The results, although consistent with previous research, should be taken with caution. 6 . Implications and conclusions Governments are challenged when new technologies emerge. They are problematic because in many cases these innovations do not fit within the existing policy or re gulatory frameworks. Call-back services in the 1980s and Intern et telephony in the 1 990s are two examples of the many technologies that have had an aff ect on policy. More recently digital trunking challenged regulators on spectrum allocation. Regulators will face a similar challenge when wireless fidelity (WiFi) becomes more prevalent. In these cases there are concerns about the impact on existing companies and on the regulators themselves, as they do not know how to regulate these new technologies, particularly when existing policies may be difficult to change . There are also cases where the regulator may be sympathetic to the new technology but is pressured by companies to prevent its introduction. In the telecommunications sector this 32 has been a natural response. Many of these companies have had monopolized mark ets for many years. They benefited from an unchallenged inflow of revenues and, because of their highly regulated status, they learned to operate and relate to regulators. In a situation where their traditional business models are being threatened, the natural response is to impede change. Regulators should, however, be concerned about frequently cha nging policies because laws and regulations are the basis for functioning markets and thus they cannot be changed at will. There are contracts and commitments in place that may prevent regulators from changing policy. In many cases, nonetheless, there simply is no framework that accommodates these new technologies. When a regulator decides to change a rule to accommodate innovation, it may be flooded with lawsuits. What approach, then, should regulators take? For developing countries the results of this analysis show that at least two technologies, call-back and Internet telephony, have not had negative impact on infrastructure. This shows that the effect that these technologies may have on the existing carriers is, if anything, gradual. It would thus be premature to impede innovation, particularly when benefits to the public are clear. Operators are aware of these technologies and the gradual improvement of these new services should give them time to respond in a manner that strengthens their competitive position. Regulators should thus approach the matter cautiously, simply observing the evolution of the technology as it affects the market and making agreements with operators to review policies within specific periods. If there are signs of a negative impact policies could be revised. Regulators should be prepared to answer carrier?s challenges. They may simp ly follow the current law, which is often broad enough to embrace new technologies. Banning the technologies without evidence of negative effects on society, as opposed to carriers alone, can prevent these countries from incorporating services that would benefit the public. 33 References Bernt, P., and Weiss, M., 1993. International telecomm unications, 1st edn (Carmel, IN, Sams Publications). Cawley, R. A., 1997. Internet, lies and tele phony, Telecommunications Policy 21, 513-532. Clark, D., 2001. A taxonomy of internet tele phony applications, in: L. W. McKnight, W. Lehr, and D. D. Clark (Eds.) Internet te lephony, (Cambridge, MA, MIT Press). Cowhey, P., 1998. FCC benchmarks and the reform of the interna tional telecommunications market, Telecommunications Policy 22, 8 99-911. Frieden, R., 1998. Falling through the crack s - international accounting rate reform at the ITU and WTO, Telecommunications Policy 22, 963-975. Galal, A., Jones, L., Tandon, P., and V ogelsang, I., 1994. Welfare consequences of selling public enterprises: An emprical analysis (Washington, DC, The World Bank). Graham, J. W., Hofer, S. M., and Piccinin, A. M., 1994. Analys is with missing data in drug prevention research, in: L. Collins, and L. Seitz (Eds.) National institute on drug abuse research monograph series, (Washington, DC, National Institute on Drug Abuse), pp. 13-62. Honaker, J., Joseph, A., King, G., Scheve , K., and Singh, N., 2003. Amelia: A progr am for missing data, Windows Version 2.1. ed n (Cambridge, MA., Harvard University,). ITU Internet reports: IP telephony, 2001. (Eds.), (Geneva, International Telecommunications Union). Ke, B., 1994. New global network arrangement s - regulatory and trade considerations, Telecommunications Policy 18, 378-396. Kelly, T., McTaggart, C., Petrazzini, B., Sh aw, R., and Woodall, M., 2001. ITU Internet reports: IP telephony (Geneva, International Telecommunications Union). King, G., Honaker, J., Joseph, A., and Sc heve, K., 2001. Analyzing incomplete political science data: An alternative algorithm for multiple imputation, American Political Science Review 95, 49-69. 34 King, G., Tomz, M., and Wittenberg, J., 2000. Making the most of statistical analysis: Improving interpretation and presentation, American Journal of Political Science 44, 347-361. Kiser, C. R., and Collins, A. F., 2003. Regulation on the horizon: Are regulators poised to address the status of ip telephony? , Communications Law Conspectus 19, 19-44. Lam, P.-L., 1997. Erosion of monopoly power by call-back - l essons from Hong Kong, Telecommuni cations Policy 21, 693-695. Little, R. J. A., and Rubin, D. B., 2002. Statistical analys is with missing data, 2nd edn (New York, Wiley & Sons). Mason, R., 1998. Internet telephony and the international accounting rate system, Telecommunications Policy 22, 931-944. Mbarika, V., 1999. Factors that affect growth of teledensity in least developed countries, Paper presented at: 22nd Conference on Information Systems Research (IRIS 22) (Jyvaskyla, Findland). Mbarika, V., Byrd, T., and Raymond, J., 2002. Growth of teledens ity in least developed countries: Need for a mitigated euphoria, Journal of Global Information Management 10, 14-28. McGarty, T., and McKnight, L. W., 2001. Vi rtually global telcos: International internet telephony architectures, in: L. W. McKn ight, W. Lehr, and D. D. Clark (Eds.) Internet telephony, (Cambridge, MA, MIT Press). McKnight, L., and Leida, B., 2001. Internet telephony service provide rs, in: L. W. McKnight, W. Lehr, and D. Clark (Eds.) Inter net telephony, (Cambridge, MA, MIT Press), pp. 193-213. Megginson, W. L., Nash, R. C., and Van Rande rborgh, M., 1994. The financial and opera ting performance of newly privatized firms : An international empirical analysis, Journal of Finance 49, 403-452. The missing link, 1984. (Eds.), (Geneva, Independent Commission for World Wide Telecommunications Development (The Maitland Commission)). Moore, D. W., 1997. Regulation of the internet and internet telephony through the imposition of access charges, Texas Law Revie w 76, 183-214. 35 Ramamurti, R., 1996. Privatizing monopolies: Lessons from the te lecommunications and transport sectors in latin america (Baltim ore, MD, The Johns Hopkins University Press). Ros, A., 1999. Does ownership or competition matter? The effects of telecommunications reform on network expansion and efficien cy, Journal of Regulatory Economics 15, 65-92. Sharifi, H., 2001. Internet telephony carrier strategies, in: L. W. McKnight, W. Lehr, and D. Clark (Eds.) Internet telephony, (Cambridge, MA, MIT Press), pp. 301-333. Silber, S. C., 1996. The fcc's call-back order: Proper respect for international comity?, George Washington Journal of Intern ational Law & Economics 30, 96-195. Statistics of communications common carriers, (Eds.), (Federal Communications Commission). Telecommunication regulatory database, 2002. (Eds.), (Geneva, International Telecommunications Union). Vickers, J., and Yarrow, G., 1988. Privatization: An economic analysis (Cambridge, MA, MIT Press). Weinberg, J., Internet telephony regulation, in: L. MCKnight, W. Lehr, and D. Clark (Eds.) Internet telephony, (Cambridge, MA, MIT Press). Wellenius, B., 1989. The impact of modern telecommunicati ons (Eds.), IEEE Technology and Society Magazine, pp. 3-6. World telecommunication indicators, 2002. (Eds.), (Geneva, International Telecommunications Union). World telecommunications development report, 1994. (Eds .), (Geneva, International Telecommunications Union). Yatrakis, P., 1992. Determinants of the demand for international telecommunications, Telecommunications Journal 39, 732-746.