POLICY RESEARCH WORKING PAPER 2979 Regulation ancl Internet tJse in Developing Countries Scott Wallsten The World Bank Development Research Group W Investment Climate March 2003 POLICY RESEARCH WORK[NG PAPER 2979 Abstract Policymnakers are simultaneously concerned about the with lower Internet penetration and higher Internet consequences of a worsening "digital divide" between access charges. More specifically, controlling for factors rich and poor countries and hopeful that information such as income, development of the telecommunications and computing technologies could increase economic infrastructure, ubiquilty of personal computers, and time growth in developing countries. But very little research trends, couLntries that require formal regulatory approval has explored the reasons for the digital divide beyond for Internet service providers (ISPs) to begin operations noting that it is strongly correlated with standard have fewer Internet users and Internet hosts than development indicators, and no empirical research has countries that do not require such approval. Moreover, explored the role of regulation. countries that regulate ISP final-user prices have higher Wallsten uses data from a unique new survey of Internet access prices than countries that do not have telecommniulcations regulators and other sources to such regulations. These results suggest that developing measure the effects of regulation in Internet countries' own regulatory policies can have large impacts development. He finds regulation strongly correlated on1 the digital divide. This paper-a product of Investment Climate, Developmiienit Research Group-is part of a larger effort in the group to understanid regulatory and infrastructure sector reforms. Copies of the paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Paulina Sintin-Aboagye, room MC3-422, telephone 202-473-7644, fax 202-522-1 155, email address psintimilaboagye @aworldbank.org. Policy Research Working Papers are also posted on the Web at http://ccon.worldbank.org. The author may be contacted at swallsteiln(@worldbank.org. March 2003. (24 pages) The Policv Research Working Paper Series disse,nilnztcs the fmidntgs of ivork ti progress to encourage the exchange of ideas abouit development issutes. An objective of the series is to get the findings out qtickly, eveni if the presenitations are less than filly polished. The papers carry the niamies of the authors an7d sbozld be cited accordingly. Tbe findings, interpretatiois, and conclusions expressed in this paper are entirely those of the aujthors. They do not necessardiy represent the view of the W/orld Bank. its Executive Directors, or the countries they represent. Produced by the Research Advisory Staff Regulation and Internet Use in Developing Countries Scott Wallsten swallsten(worldbank.org DRAFT COMMENTS WELCOME * I am grateful for helpful comments from George Clarke and Greg Rosston. I. Introduction As information and communication technologies (ICTs) became ever rnore ubiquitous in the late 1990s, many began to worry about an increasing "digital divide" between rich nations and poor.' That is, those in rich nations had increasingly better access to ICTs and were using them more intensively while poor nations continued to fall further behind. At the same time, many believed that the Internet and ICTs potentially offered developing countries an opportunity to increase productivity and economic growth. These two beliefs have led to many initiatives by industrialized countries, NGOs, and intemational organizations such as the World Bank and the UN intended to increase Internet usage and, presumably, bridge the digital divide in developing countries. While many of these initiatives mention the importance of competition and regulatory structure, few have given it much attention, and almost no research has explored the effects of regulation on Internet development. In this paper I combine data from a new survey of regulators, the International Telecommunications Union (ITU), and World Bank statistics to estimate the effects of regulation on the Internet in developing countries. Controlling for factors including income, development of the telecommunications network, and number of computers in the country, regulations targeted at the Internet were correlated with lower Intemet usage, fewer Internet hosts, and higher prices for Internet access. More specifically, countries that require formal approval for Internet Service Providers (ISPs) to begin operations had fewer Intemet users and Internet hosts, while countries that regulate ISP prices have higher prices for Internet access. II. The Digital Divide People in richer countries have better access to ICTs and use them more intensively than do people in poorer countries. Figure 1 shows Internet use over time by countries' average income. It is clear that Internet usage is far more cormmon in richer countries than in poorer countries, and also that the number of users has grown more quickly in richer countries. The torrid pace of technological change-and the ensuing technology bubble-in industrialized 1 There is also concem about so-called "digital divides" within countries, but this paper is concemed with the gap between countries. 2 countries led to the simultaneous concerns that this divide would continue to grow with grave consequences, and also that the Internet and new technologies could help stimulate economic development (see, for example, Hammond 2001; Ishaq 2001; Norris 2001; Steinmueller 2001). Concerns about consequences of the widening digital divide permeated discussions of economic development. Kofi Anan, for example, said that "People lack many things: jobs, shelter, food, health care, and drinkable water. Today, being cut off from telecommunications services is a hardship almost as acute as these other deprivations, and may indeed reduce the chances of finding remedies to them" (as quoted in Norris 2001). The concerns about the digital divide have probably been overstated and the potential benefits from ICTs over-hyped. Kenny (2002), for example, notes that the likely impact of any particular technology is likely to be small,2 and Pohjola (2002) finds little correlation between ICT investments and economic growth in 42 developing countries from 1985-1999. Nonetheless, there is evidence that ICTs can improve productivity and economic growth in developing countries. Clarke (2002), for example, using enterprise-level data in Eastern European transition economies, finds that even controlling for endogeneity, firms that have Internet access are more likely to export than firms that do not.3 Bhavani (2002) finds that use of technology is beneficial for firms in the Indian auto components industry. Moreover, ICTs-or, more accurately, involvement in ICT industries-have also been important in spurring regional economic growth in places such as Taiwan and Bangalore (Arora, et al. 2001; Athreye 2002; Saxenian and Hsu 2000). In response to the concems and hopes, a wide range of organizations established initiatives to "bridge the digital divide." As Heeks (2002) put it, "donors, attracted by a combination of the hype and hope generated by ICTs have altered their funding priorities and pushed ICTs up the development agenda."4 These initiatives include, but are not limited to, the United Nations Information and Communication Technologies Task-Force, the Digital 2 While any given technology is unlikely to have a major impact on economic growth, some innovations become "general purpose technologies" that do generate a burst of productivity and economic growth (e.g., Bresnahan and Trajtenberg 1992; Helpman 1998; Rosenberg and Trajtenberg 2001). Whether certain ICTs are general purpose technologies is still a matter of debate. 3 In a complementary paper, Clarke (2001) finds that foreign-owned firms are more likely to have Intemet access. Moreover, he found evidence of spillovers from this access, with FDI increasing Internet access among domestic firms other than firm receiving the FDI. 4 Heeks (2002) also notes that this is not necessarily good: "Like sharks drawn to blood in the ocean, a whole host of consultants, academics, vendors, and development organization staff have been drawn into the e-development arena by the scent of money." 3 Opportunities Task Force created by the G8 and administered by the World Bank and UNDP, the World Economic Forum Global Digital Divide Initiative, and agreements between the ITU and vendors such as Cisco and Oracle to bring training centers to poor countries (Campbell 2001). These initiatives typically share common goals and methods, consistent with the view that the key to building ICT use in developing countries was some kind of direct provision, investment, or demand-push from govemments (see, for example, Kapur 2002). InfoDev at the World Bank, for example, describes itself as "a multi-donor program supporting, through a mompetitive grant mechanism, innovative activities demonstrating the opportunities offered by new technologies in information and communications. Its mission is to utilize these technologies for economic and social development, with a special emphasis on the needs of the poor in developing economies" (InfoDev 2002). Funded projects and programs generally focus on iirectly providing the technology or training necessary to bring ICTs to the pooir.5 Even if targeted projects achieve their goals, truly addressing the problem requires mderstanding factors underlying the divide. Given the potential of ICTs in developing countries to improve growth and the resources already committed to addressing the issue, it is important to Lnvestigate whether certain obstacles make the divide worse than one would expect even given a -ountry's general economic situation. A few papers have empirically explored ICT use in leveloping countries. This work is generally based on data collected by the International Felecommunications Union (ITU), and tends to find digital technologies correlated with the Yeneral level of development such as trade openness and income (Baliamoune 2002; Onyeiwu 2002). The results from this literature are perhaps not surprising, but suggest that technology liffusion in developing countries may not be so different from other goods and services. These apers do suffer from some shortcomings, however. First, they include Internet users, Internet losts, number of computers, telephone lines, and mobile telephone customers all as dependent iariables.6 While this choice is not necessarily incorrect, the resulting empirical specifications ;annot control for many relevant necessary factors. For example, as discussed below, the lumber of Intemet users is likely to be (partly) a function of the development of' the elecommunications infrastructure and the number of computers in the country. Also, the See htto://www.infodev.org/index.html for lists of proposals and funded projects. Indeed, Onyeiwu (2002) combines these four variables into a single, "digitalization" index. 4 analyses typically do not incorporate or acknowledge the collected knowledge of telecommunications reforms, which suggests that some measure of market structure and liberalization should be taken into account. One notable exception is a study by Dasgupta, et al. (2001), who find that the digital divide is better characterized as a lack of telecommunications access than a lack of access to ICTs, per se. As a result, competition policy proves to be important-countries with higher general scores on competition policy have higher Intemet intensity (Internet subscriptions per telephone mainline). Their conclusions are sensible given other work on telecommunications reforms in developing countries. By now a great deal of research has demonstrated the benefits of competition in telecommunications (e.g., Ambrose, et al. 1990; Fink, et al. 2002; Li and Xu 2001; Ros 1999; Wallsten 2001). Nonetheless, their analysis uses a very general measure of overall "competitiveness" within a country-an index constructed in 1995 that measures on a scale of I to 6 whether "the state inhibits a competitive private sector, either through direct regulation or by reserving significant economic activities for state-controlled entities." In this paper I use data from a new survey of telecommunications regulators to assess the effects of targeted regulations on the development of the Internet. Specifically, the data allows me to explore the effect of ISP regulations on Intemet usage and access pricing, while controlling for other factors that one would expect to affect these variables. In the sections below I discuss regulation and the Intemet and then proceed to explain the data, empirical methods, and results. III. Regulation and the Internet Regulation is an important component of competition policy in the context of telecommunications. A key problem in telecommunications reforms is that while competition is technically possible, new entrants require the cooperation of the incumbent firm, which often has nearly a complete monopoly. In particular, a new entrant's customers must be able to reach the incumbent's customers if the entrant's network is to have any value. In order to be viable, therefore, the entrant must be able to interconnect with the incumbent's network. Because a monopolist has no incentive to interconnect, and every incentive to exercise its market power at the expense of new entrants and consumers, most countries established regulatory agencies as 5 part of their telecommunications reform process in large part to curb the incumbent's market power and promote competition in the industry. But the definition of "telecommunications" is not clear-cut, and regulatory discretion is rarely limited to interconnection issues. In particular, in addition to telephony, some countries regulate aspects of the Internet, especially Internet Service Providers (ISPs). Regulatory discretion is not the only ambiguous feature of these agencies. Whether regulatory agencies effectively correct market failures has been a hotly debated issue for several decades, as discussed below. Helping Hand, Capture, and the Grabbing Hand In an ideal world, economic regulation is intended to correct market failures (Pigou's (1938) "helping hand" theory). For example, in industries considered to be natural monopolies-where increasing returns to scale mean that one firm can supply the market at lowest cost-a laissez-faire approach would ultimately result in a single firm dominating the market free to charge monopoly prices and providing lower quantities of the service. The regulator was supposed to ensure that the firm kept costs down while continuing to innovate and improve service as if it operated in a competitive inclustry. Regulations, however, can end up primarily benefit the regulated firm. As Stigler (1971) pointed out, it is easy for the regulated firm to "capture" the regulator. In this case, regulations can become a vehicle not for maximizing total or consumer welfare, but for protecting the firm's monopoly and profits. It is easy to understand how such regulatory failure can happen, even assuming a benevolent regulator. The firm always has more information about its cost structure than does the regulator, making it difficult for the regulator to independently evaluate the firm and the industry. The firm can strategically control the information available to the regulator to increase the probability that decisions are in the firm's favor. Likewise, the regulator may impose entry barriers to the industry, either out of a belief that entry would be harmful or, if truly captured, simply to protect the incumbent. In either case, the incumbent firms benefits from the regulation. Developing countries face an additional regulatory pitfall. Regulatory agencies create more opportunities for governments to extend a "grabbing hand"-that is, they increase the 6 number of interacfion points between government and industry where bribes may be required for firms to begin or continue operations. This pattern of administrative barriers imposed on firms in developing countries is becoming well-documented (e.g., de Soto 1989; Djankov, et al. 2002; Emery, et al. 2000; Friedman, et al. 2000). These barriers impose real costs on developing economies. Emery, et al. (2000) discovered that "when added together, this whole maze of often duplicative, complex, and non-transparent procedures can mean delays of up to two years to get investments approved and operational." Moreover, Djankov, et al. (2002) finds entry regulation in developing countries to be associated with lower quality public and private goods. The net effect of telecommunications regulation in developing countries is unclear. On the one hand, regulatory agencies may be crucial in encouraging entry and investment in the face of a dominant incumbent. On the other hand, the track record is not good. Developing country regulators face the dual problems of avoiding capture and becoming another grabbing hand. Regulation and the Internet The effects of regulation on telecommunications reforms in general have only begun to be extensively investigated.7 Research, however, has almost completely ignored the effects of regulation on Internet development. Many of the "digital divide" reports note that promoting competition and Internet-friendly regulatory policies are an important component of addressing the issue (e.g., International Labour Office 2001). Nonetheless, to my knowledge no empirical study has explored the effects of regulation on the Internet in developing countries. This omission is especially unfortunate given that regulatory agencies are likely to be the primary way public policy affects Internet development. One case study on the subject highlights the potential importance of regulation on Internet development. Petrazzini and Guerrero (2000) investigate development of the Internet in Argentina. In particular, they try to explain Argentina's progression from having among the lowest Internet penetration in Latin America in 1993 to among the highest in 1999. Essentially, they find that regulatory intervention was key in that non-competitive market, with the government mandating large reductions in prices of leased lines and the creation of a special 7 Many researchers have discussed telecommunications regulation (e.g., Galal and Nauriyal 1995; Gasmi, et al. 2000; Guasch and Hahn 1997; Hill and Abdala 1996; Noll 2000; Stiglitz 1999; Wallsten 2001; Wellenius and Stem 1994). To date, however, most analysis consists of case studies; there is very little empirical work on the issue. 7 local dialing scheme for calls to connect to the Internet. Even with those mandated reductions, however, the authors note that high prices still seemed to be a major obstacle to higher Internet penetration. While the Argentina case study highlights the potential influence regulatory policy can have on Intemet development, this paper aims to address this question empirically by exploiting a new, detailed dataset based on a survey of telecommrunications regulators in 44 developing countries. As discussed below, the survey contained several questions on regulation of ISPs, which, combined with standard data from other sources allow me to test the effects of regulation on Intemet development. The following sections describe the data, empirical methods, and results. IV. Data and Empirical Analysis The point of this paper is to measure the effects of regulation on the Internet in developing countries. I do the analysis in the context of a simple regression framework, estimating several versions of equation (1). (1) y = 1% + 3, *(regulation) + 832Z + e In this equation, the dependent variable y is the relevant Intemet measure, regulation is the relevant regulation variable, and Z is a vector of control variables. These are described in more detail below. The regulation variables-the heart of this paper-are derived from a survey of regulators in developing countries conducted from March 2001 to August 2001. We contacted 60 regulatory agencies in developing countries around the world and received responses from 45 (see Wallsten, et al. 2002 for a complete description of the data).8 The survey asked detailed questions about the agency's independence, composition, appointment and removal processes, transparency, accountability, and discretion. Several questions about the regulator's discretion focused on ISPs, resulting in two useful regulation variables for this analysis. ]First, the survey asked "What approval do Intemet Service Providers have to get before they can start operating," We actually received 46 responses, but one retumed the survey unanswered. 8 where possible answers were "none", "notification", and "formal approval." For the analysis I coded regulatory approval as a dummy variable that equals one if the operator requires formal approval to start operating, and zero otherwise. Second, the survey asked whether "final-user prices [are] regulated" for ISPs. Again, price regulation becomes a dummy variable that equals one if ISP prices are regulated. Table 1 shows the results of these questions by country. The table shows that of 38 countries that answered the question, 23 require formal approval for ISPs to start operation. Of 41 countries that responded to the second question, six regulate final-user ISP prices. The two Intemet measures come from the ITU. The two measures of Intemet penetration include the number of Intemet users and the number of Internet hosts (Table 2). The number of Intemet users is "based on reported estimates, derivations based on reported Intemet Access Provider subscriber counts, or calculated by multiplying the number of hosts by an estimated multiplier" (Intemational Telecommunications Union 2002c). While imperfect, this estimate seems to be the best estimate of Internet use available. The second variable, the number of Intemet hosts, "refers to the number of computers in an economy that are directly linked to the worldwide Intemet network. This statistic is based on the country code in the host address and thus may not correspond with the actual physical location" (Intemational Telecommunications Union 2002b). The qualifier on the definition is likely the explanation for large variation in the number of hosts per country. The problem is essentially that while each country is assigned an Intemet identifier (e.g., .uk for Britain), there is no obligation for everyone in the country to use it, resulting in ambiguity regarding the physical location of a given host.9 These issues result in strange anomalies in the data. For example, while it is conceivable that Brazil had nearly 663,000 Intemet hosts in the year 2000, it is strains credulity to believe that Malawi-despite being one of the poorest countries in the world-had only one.' ' Data on Internet pricing also comes from the ITU. The pricing information is the ISP charge, and is defined as "the costs associated with 30 off-peak hours of dial-up Intemet time per month. It is the monthly Intemet subscription rate plus extra charges once free hours have been used up" (Intemational Telecommunications Union 2002a). While the measures of Intemet use 9 Any intemet address registered as .com or .org, for example, is counted as being located in the United States. '° The reason for Malawi's poor showing appears to be a dispute over Malawi's Intemet identifier, .mw. Until recently Malawi was unable to use it, because it was being held by a "British-South African entrepreneur" (Blantyre 2000). 9 are available from the early 1990s for many countries, the price data are available only for 2001. The last column of Table 2 shows ISP charges that year for the countries in the sample, revealing large variation across countries. The vector Z attempts to control for other factors affecting Intemet usage and pricing. In particular, it includes per capita income, development of the telecommunications industry (number of telephone mainlines), number of personal computers, whether the incumbent telecommunications provider has been privatized, and the share of trade in the economy. The combination of income, telecommunications development, and ubiquity of personal computers is likely to be a good indicator of potential demand for Intemet services-or at least the number of users as measured here. The privatization variable provides some measure of whether the industry has been liberalized or is still run by the state. Trade is included as a measure of international integration, under the assumption that countries that are more connected to the rest of the world are more likely to tolerate or promote Intemet access. Moreover, such openness measures are the one factor consistently significant in econometric analyses discussed above. One potential problem here is the likely correlation between the control variables. That is, the number of personal computers per capita in Et country is likely to be determined in large part by a country's per capita income. Likewise, an enormous literature has explored the relationship between trade and income. However, ][ntemet use is likely to be a function of all these variables, meaning it would be a mistake to exclude them, and they are unlikely to be correlated with the regulatory variable of interest. The implication of this collinearity is that interpreting the coefficients on the control variables may not be easy, but it should not affect the variables of interest and excluding them is likely to bias the coefficient on the regulation variable. Nonetheless, I add control variables to each regression slowly both to see how including one variable affects the others, and to test the robustness of the results on the policy variables. Internet users and hosts The first part of the analysis explores the relationship between regulations and Internet penetration by estimating equation (2), which is similar to equation (l) except that the data exist over time and across countries. 10 (2) y = /3o + /, *(formal approval requiredfor ISP to operate?)i + /32Zit + y. + ei, The first set of regressions uses the number of Intemet users per capita as the dependent variable, and the second set of regressions uses the number of Internet hosts per capita as the dependent variable. The control variables are the ones described above, though the panel nature of the data means that I can also control for year fixed effects. This time trend is an especially important control with a new technology such as the Intemet, which is growing rapidly over time. Note that the regulation variable does not vary over time. The survey was conducted in 2001, and I assume that this regulatory function was in place from the time the regulatory agency was founded. In one sense this is a bad assumption. The Chilean regulatory agency, for example, was founded in 1977 and is unlikely to have considered regulations regarding a technology that, from the point of view of a telecommunications authority, did not yet exist. In a practical sense, however, this assumption is reasonable for two reasons. First, the first year any country in the sample reports Internet users is 1991 (in Brazil, which had 5,000), and more commonly not until 1995 or 1996, meaning that the analysis never includes an observation that assumes Intemet regulation to exist absurdly early. Second, most regulatory agencies were formed around the same time the Intemet was becoming popular and countries were already considering its implications, making it likely that these regulations were part of the regulatory agency's purview from its inception. The implication of this discussion is that the dataset becomes an unbalanced panel, where an observation is a country-year, and the first year a country appears in the regression is the first year the regulatory agency was operating. Table 3 reports the results of estimating equation (2) with the number of Intemet users per capita as the dependent variable. Interpreting the coefficients on the control variables is not easy, as discussed above, though broadly speaking the results are consistent with expectations. The number of telephone mainlines per capita is positively and significantly associated with Intemet users, though it becomes insignificant as other variables are added. The coefficient on per capita income is negative, though significant only in some specifications. It is well-known that the most important determinant of telecommunications development, however, is income, so it is not surprising that including both of those variables together makes interpreting them difficult. The number of personal computers per capita is robustly positively correlated with Internet use, and having a privatized telecommunications incumbent is also positively and significantly associated with Intemet use. The regulatory variable is negatively and significantly associated with per capita Intemet use across all specifications. That is, countries that require ISPs to get formal regulatory approval to start operations have fewer Internet users per capita than countries that allow ISPs to begin operations without such formal approval. Moreover, the coefficient is not only negative, but also fairly large in magnitude at just about half the mean of the dependent variable. Table 4 shows the results of estimating equation (2) when the dependent variable is Internet hosts per capita. The results on the control variables are basically the same as described above. The coefficient on the regulation variable, however, while always negative, ranges from being significant at the one percent level when in the regression by itself and also controlling for income (and year dummies), to not significant at all when including mainlines per capita, privatized incumbent, and personal computers per capita. A slightly different specification, however, yields more robust results. Table 5 shows the results of estimating equation (2) using log-levels. In this specification, the regulation variable is robustly negative across specifications. Even including any combination of income, GDP, population, number of telephone lines, privatized iincumbent, number of personal computers, and trade variables, requiring formal approval for ISPs to start operations is correlated with fewer Internet hosts per capita.' l Using logs is arguably more sensible in this case given the very large variation in reported number of Internet hosts. Nonetheless, the large error inherent in the measurement of this indicator discussed above suggests that the results in general may be less reliable than for other variables. ISP Final-User Prices Equation (3) presents the equation to be estimated using the price of monthly ISP access as the dependent variable, and whether or not ISP final-user prices are regulated as the policy variable of interest. (3) y = /3o + /3j *(ISPfinal-user prices regulated?)i +82Zi + Ei " The results are, not surprisingly, equally robust when using log-levels without normalizing by population (that is, not in per capita temns). 12 While the Internet penetration variables used above are available for several years, ISP prices are reported only for the year 2001. Unfortunately, not all of the control variables are yet available for that year for all countries, so I use year 2000 figures for those variables. The result is a cross- section of 29 countries using variables from 2000 and 2001. Table 6 shows the results of this estimation. None of the control variables are significant across specifications. There is some evidence that a larger number of personal computers is correlated with lower ISP prices, but the coefficient is barely significant, at best. The regulatory variable, however, is positive and statistically significant across specifications. In other words, Intemet access is more expensive in countries that regulate final-user ISP prices. V. Discussion and Conclusion The prospect of a growing "digital divide" between poor and rich countries and hope that ICTs may present an opportunity to improve productivity and economic growth has led to a number of initiatives designed to stimulate ICT use in developing countries. While many organizations note that competition and a friendly regulatory environment may be important to ICT development, few of these initiatives and very little research has actually focused on the role of regulation in promoting Internet use. This paper is an attempt to fill that gap. Using data from a new survey of telecommunications regulators in developing countries combined with publicly-available information from the ITU and the World Bank, I find regulation of Internet Service Providers to be correlated with worse outcomes. In particular, countries that require ISPs to get formal approval before beginning operations have fewer Internet users and Internet hosts, while ISP price regulation is correlated with higher ISP final-user prices. One potential concem with the analysis is whether the regulatory variables are endogenously correlated with the dependent variables. While this is a concern in a great deal of similar work on telecommunications reforms, reverse causality is unlikely to be the case here. First, these regulations were set largely in the early days of the technology's introduction into the country. In other words, the regulations probably came before the Internet was so ubiquitous, not the other way around. Second, the empirical results are consistent with many govemments' initial reaction to the Intemet, which was apprehension. More repressive govemments worried 13 about the effects of free access to information, and nearly all governments worried about the effects of Intemet telephony on their incumbent telecommunications provider."2 In this light, ISP regulations were likely promulgated precisely in order to suppress Internet access, and in that light, seems to have been successful. These results suggest that regulations on the Internet are not intended to correct market failures, and are more consistent with a capture or "grabbing hand" view of regulation. Price regulation correlated with higher prices is consistent with capture, and requiring formal approval for ISP operations consistent with both capture and grabbing hands. Requiring formal regulatory approval presents a barrier to entry to new ISPs. The finding that this entry barrier is correlated with lower development of the industry is also consistent with Djankov's (2002) finding that countries with heavier entry regulations have lower quality public and private goods. The results also suggest that a country's regulatory approach to the Internet and ICTs can have a large impact on its ubiquity throughout the country. While the sample contains a relatively small number of countries, the results suggest that reducing entry barriers and promoting competition is likely to yield large increases in the share of developing countries' populations with access to the Internet and any potential benefits that flow from such access. In other words, removing entry barriers and promoting ISP competition may present a low-cost and non-distortionary way to boost Internet use in developing countries. 12 I tried including various measures of civil liberties (e.g., from Freedom House) and corruption in the empirical analyses, but they were generally not significant. Moreover, because such indicators are not available for all countries over all years their inclusion caused the sample size to shrink substantially. 14 References Ambrose, William, Paul Hennemeyer, and Jean-Paul Chapon. 1990. "Privatizing Telecommunications Systems: Business Opportunities in Developing Countries." IFC Discussion Paper: Washington, DC. Arora, Ashish, Alfonso Gambardella, and Salvadore Torrisi. 2001. "In the Footsteps of Silicon Valley? Indian and Irish Software in the International Division of Labour." SIEPR Discussion Paper: Stanford. Athreye, Suma. 2002. "The Indian Software Industry and Its Evolving Service Capability." The Open University Working Paper: UK. Baliamoune, Mina N. 2002. "The New Economy and Developing Countries." WIDER Discussion Paper. Bhavani, T.A. 2002. "Impact of Technology on the Competitiveness of the Indian Small Manufacturing Sector: A Case Study of the Automotive Component Industry." WIDER Discussion Paper. Blantyre. 2000. "Wrangle Over Malawi's Main Internet Domain." Panafrican News Agency. Bresnahan, Timothy F. and Manuel Trajtenberg. 1992. "General purpose technologies: "engines of growth?"" National Bureau of Economic Research. Working Paper Series, No. 4148, pp. 1- [41]. Campbell, Duncan. 2001. "Can the digital divide be contained?" International Labour Review, 140:2, pp. 119-41. Clarke, George. 2001. "Bridging the Digital Divide: How Enterprise Ownership and Foreign Competition Affect Intemet Access in Eastern Europe and Central Asia." World Bank Policy Research Working Paper: Washington. Clarke, George. 2002. "Does Intemet Connectivity Affect Export Performance? Evidence from Transition Countries." WIDER Discussion Paper. Dasgupta, Susmita, Somik Lall, and David Wheeler. 2001. "Policy Reform, Economic Growth, and the Digital Divide: An Econometric Analysis." World Bank Policy Research Paper: Washington. de Soto, Hernando. 1989. The Other Path. New York: Harpers & Row. Djankov, Simeon, Rafael La Porta, Florencio Lopez-de-Silanes, and Andrei Shleifer. 2002. "The Regulation of Entry." Quarterly Journal of Economics, 117: 1, pp. 1-37. 15 Emery, James, Melvin T. Spence, Louis T. Wells, and Timothy Buehrer. 2000. "Administrative Barriers to Foreign Investment: Reducing Red Tape in Africa." The International Finance Corporation: Washington. Fink, Carsten, Aaditya Mattoo, and Randeep Rathindran. 2002. "Liberalizing Basic Telecommunications: Evidence from Developing Countries." World Bank Working Paper: Washington. Friedman, Eric, Simon Johnson, Daniel Kaufmann, and Pablo Zoido-Lobaton. 2000. "Dodging the Grabbing Hand: the Determinants of Unofficial Activity in 69 Countries." Journal of Public Economics, 76:3, pp. 459-93. Galal, Ahmed and Bharat Nauriyal. 1995. "Regulating Telecommunications in Developing Countries." World Bank Policy Research Working Paper: Washington, DC. Gasmi, F., Jean-Jacque Laffont, and W.W. Sharkey. 2000. "Competition, Universal Service and Telecommunications Policy in Developing Countries." Information Economics and Policy, 12:3, pp. 221-48. Guasch, J. Luis and Robert W. Hahn. 1997. "The Costs and Benefits of Regulation: Some Implications for Developing Countries." World Bank Policy Research Working Paper: Washington. Hammond, Allen L. 2001. "Digitally Empowered Development." Foreign Af airs, 80:2, pp. 96- 106. Heeks, Richard. 2002. "i-Development not e-Development: Special Issue on ICTs and Development." Journal of International Development, 14, pp. 1-11. Helpman, Elhanan. 1998. General purpose technologies and economic growth. Cambridge, Mass.: MIT Press. Hill, Alice and Manuel Abdala. 1996. "Argentina: The Sequencing of Privatization and Regulation," in Regulation, Institutions, and Commitment: Comparative Studies of Telecommunications. Pablo Spiller ed. Cambridge: Cambridge University Press, pp. 202-49. InfoDev. 2002. "Quarterly Report: Second Quarter 2002." The World Bank: Washington. International Labour Office. 2001. "World Employment Report 2001: Life at Work in the Information Economy." International Labour Organization: Geneva. International Telecommunications Union. 2002a. "Challenges to the Network: Internet for Development." Accessed through the World Bank Statistical Information Management and Analysis (SIMA). 16 International Telecommunications Union. 2002b. "International Telecommunications Database." Accessed through The World Bank's Statistical and Information Management & Analysis System. International Telecommunications Union. 2002c. "World Indicators: Information Technology." ITU News Magazine. Ishaq, Ashfaq. 2001. "On the Global Digital Divide." Finance and Development, 38:3. Kapur, Sandeep. 2002. "Developing Countries and the New Economy." WIDER Discussion Paper. Kenny, Charles. 2002. "The Internet and Economic Growth in Least Developed Countries: A Case of Managing Expectations?" WIDER Discussion Paper. Li, Wei and Lixin Colin Xu. 2001. "Liberalization and Performance in Telecommunications Sector around the World." Washington, DC. Noll, Roger. 2000. "Telecommunications Reform in Developing Countries." SIEPR Policy Paper: Stanford. Norris, Pippa. 2001. Digital Divide: Civic Engagement, Information Poverty, and The Internet Worldwide. Cambridge: Cambridge University Press. Onyeiwu, Steve. 2002. "Inter-Country Variations in Digital Technology in Africa: Evidence, Determinants, and Policy Implications." WIDER Discussion Paper. Petrazzini, Ben A. and Agustina Guerrero. 2000. "Promoting Internet Development: the Case of Argentina." Telecommunications Policy, 24, pp. 89-112. Pigou, Arthur. 1938. The Economics of Welfare. London: Macmillan and Co. Pohjola, Matti. 2002. "New Economy in Growth and Development." WIDER Discussion Paper. Ros, Agustin J. 1999. "Does Ownership or Competition Matter? The Effects of Telecommunications Reform on Network Expansion and Efficiency." Journal of Regulatory Economics, 15: 1, pp. 65-92. Rosenberg, Nathan and Manuel Trajtenberg. 2001. "A general purpose technology at work: the Corliss steam engine in the late 19th century US." National Bureau of Economic Research. Working Paper Series, No. 8485, pp. 1-67. Saxenian, AnnaLee and Jinn-Yuh Hsu. 2000. "The Silicon Valley-Hsinchu Connection: Technical Communities and Industrial Upgrading." 17 Steinmueller, W. Edward. 2001. "ICTs and the Possibilities for Leapfrogging by Developing Countries." International Labour Review, 140:2, pp. 193-210. Stigler, George. 1971. "The Theory of Economic Regulation." Bell Journal oJfEconomics and Management Science, 2, pp. 3-21. Stiglitz, Joseph E. 1999. "Promoting Competition in Telecommunications." Centro de Estudios Economicos de lat Regulacion Working Paper Series: Buenos Aires. Wallsten, Scott. 2001. "An Econometric Analysis of Telecom Competition, Plivatization, and Regulation in Africa and Latin America." Journal of Industrial Economics, 49:1, pp. 1-20. Wallsten, Scott, George Clarke, Luke Haggarty, Rosario Kaneshiro, and Lixin Colin Xu. 2002. "A Database of Telecommunications Regulators." Wtorld Bank Working Paper: Washington. Wellenius, Bjorn and Peter Stem. 1994. "Implementing Reforms in the Telecommunications Sector: Lessons From Experience," in Implementing Reforns in the Telecommunications Sector: Lessons From Experience. Peter Stern ed. Washington, DC: The World Bank. 18 Figure 1 Internet Users per 10,000 People 3000 - High income countries 2500 - - Middle income countries + Lower-middle income countries - Low income countries 2000 1500 1000 500 2 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Table I Regulators and Internet Regulations Year Approval required ISP rices Country regulator for ISP to start regulation? established opertions Argentina 1990 Formal No Barbados 2001 Formal No Belize 1987 n/a Yes Bolivia 1995 Notification No Bosnia and Herzegovina 2001 Notification n/a Brazil 1997 None No Bulgaria 1998 None No Chile 1977 None No Colombia 1994 Formal No Costa Rica 1996 Formal n/a Cote d'lvoire 1995 Formal Yes Czech Republic 2000 Notification No Dominican Republic 1998 Formal No Ecuador 1995 Formal No El Salvador 1996 None No Estonia 1998 Notification No Ghana 1996 Formal No Guatemala 1996 Formal No Honduras 1995 Formal No Hungary 1990 Formal No India 1997 Formal No Jamaica 1997 Formal No Jordan 1995 Formal Yes Kenya 1999 Formal No Latvia 1992 n/a No Malawi 1998 Formal Yes Malaysia 1998 None No Mexico 1996 Notification No Moldova 0 None No Mongolia 1995 Formal No Morocco 1998 n/a No Pakistan 1996 Notification No Panama 1996 Formal No Peru 1991 Formal No Poland 2000 Notification No Romania 2001 Formal No Slovakia 1993 Notification No South Africa 2000 Formal No Sri Lanka 1991 Formal Yes Tanzania 1993 None Yes Thailand 2001 n/a n/a Turkey 2000 n/a No Uganda 1997 n/a No Venezuela 1991 Formal No Table 2 Intemet Users, Hosts, and Access Prices (users and hosts in 2000, prices in 2001) Intemet users Intemet Hosts 1 Intemet Per 1000 Per 1000 access Country number Number people people charge* Argentina 2,500,000 67.51 175,303 4.73 77.90 Barbados 10,000 37.45 79 0.30 n/a Belize 15,000 62.50 291 1.21 n/a Bolivia 120,000 14.41 1,438 0.17 n/a Bosnia and Herzegovina 40,000 10.06 722 0.18 19.34 Brazil 5,000,000 29.34 662,910 3.89 n/a Bulgaria 430,000 52.65 15,353 1.88 7.75 Chile 2,537,308 166.80 51,380 3.38 n/a Colombia 878,000 20.76 42,927 1.01 0.01 Costa Rica 228,000 59.83 8,882 2.33 16.00 Cote d'lvoire 40,000 2.50 594 0.04 183.29 Czech Republic 1,000,000 97.34 138,060 13.44 0.00 Dominican Republic 159,000 18.99 7,841 0.94 17.97 Ecuador 180,000 14.23 2,106 0.17 n/a El Salvador 50,000 7.97 1,018 0.16 25.96 Estonia 391,600 286.05 35,773 26.13 n/a Ghana 30,000 1.55 119 0.01 36.00 Guatemala 80,000 7.03 2,894 0.25 n/a Honduras 40,000 6.23 123 0.02 15.00 Hungary 715,000 70.04 129,587 12.69 12.74 India 5,500,000 5.41 32,991 0.03 10.01 Jamaica 80,000 30.38 592 0.22 49.25 Jordan 127,317 26.05 709 0.15 23.94 Kenya 200,000 6.65 949 0.03 65.56 Latvia 150,000 63.24 15,773 6.65 28.52 Malawi 15,000 1.45 1 0.00 n/a Malaysia 3,700,000 159.00 64,081 2.75 5.26 Mexico 2,712,375 27.69 495,747 5.06 10.69 Moldova 52,600 12.28 1,713 0.40 33.31 Mongolia 30,000 12.51 168 0.07 51.50 Morocco 200,000 6.97 951 0.03 26.34 Pakistan 133,875 0.97 5,487 0.04 12.61 Panama 90,000 31.51 2,915 1.02 n/a Peru 2,500,000 97.42 9,967 0.39 n/a Poland 2,800,000 72.45 259,511 6.71 0.00 Romania 800,000 35.66 29,662 1.32 15.00 Slovakia 650,000 120.33 31,753 5.88 8.51 South Africa 2,400,000 56.07 184,547 4.31 8.50 Sri Lanka 121,500 6.28 1,754 0.09 6.49 Tanzania 115,000 3.41 536 0.02 69.00 Thailand 2,300,000 37.87 53,683 0.88 8.98 Turkey 2,000,000 30.63 108,410 1.66 1.35 Uganda 40,000 1.80 159 0.01 30.00 Venezuela 950,000 39.30 15,658 0.65 n/a * Access charge is number of US dollars for 30 off-peak hours of Intemet time per month Source: Intemational Telecommunications Union Table 3 Regulation and Intemet Users Dependent Varable: Internet Users per million people Mean of dependent variable: 17857 19232 Formal approval required by ISP? -8,261.859 -8,737.642 -6,961.854 -6,608.000 -6,611.632 (1.99)* (2.07)* (1.98)* (1.90)+ (1.95)+ Telephone mainlines per 100 people 2,065.406 1,9:20.932 -188.177 -14.794 -376.557 (9.45)** (6,34)** (0.60) (0.05) (1.12) GDP per capita 0.980 -2.448 -3.706 -1.254 (0.69) (2.06)* (2,87)** (0.83) Personal computers per million people 1.013 1.015 0.932 (10.64)** (10.82)** (9.73)" Incumbent privatized? 8,827.408 6,775.936 (2.33)* (1.81)+ Trade as share of GDP 145.625 (2.94)** Constant -14,196.594 -16,347.968 7,795.502 2,094.340 -11,140.177 (0.54) (0.61) (0.37) (0.10) (0.53) Observations 181 181 165 165 165 R-squared 0.52 0.52 0.73 0.74 0.75 Absolute value of t statistics in parentheses + significant at 10%; * significant at 5%; ** significant at 1% year dummies included Table 4 Regulation and Intemet Hosts Per Capita Dependent variable: Intemet hosts per million people Mean of dependent variable: 1190.56 1295.427 Formal approval required by ISP? -1,386.972 -1,222.210 -404.048 -196.327 -361.857 -129.819 (2.90)** (2.80)** (1.04) (0.50) (0.95) (0.35) gdpcap 0.648 -0.076 -0.290 -0.259 -0.506 (5.90)** (0.57) (2.18)* (1.78)+ (3.58)** lines 214.209 84.213 240.092 112.005 (7.61)** (2.44)* (8.30)** (3.29)** Private? 1,193.846 1,443.576 (2.89)** (3.59)** computercap 0.064 0.065 (6.16)** (6.51)* Constant 956.625 -1,020.359 -1,530.912 -950.265 -1,881.465 -1,374.777 (0.96) (1.06) (1.84)+ (0.98) (2.28)* (1.47) Observations 172 172 172 155 172 155 R-squared 0.14 0.29 0.48 0.59 0.50 0.62 Absolute value of t statistics in parentheses + significant at 10%; * significant at 5%; ** significant at 1% Year dummies Included Table 5 Regulation and log of Intemet hosts per capita Dependent variable: log(hosts per capita + 1) Mean of dep var 4.95 4.99 Formal approval required by ISP? -0.941 -0.686 -0.460 -0.475 -0.630 (2.47)* (4.08)** (2.82)** (2.88)** (4.05)** lngdpcap 2.171 1.494 1.574 0.968 (25.57)** (9.40)** (8.45)** (4.32)** Inlinecap 0.852 0.780 0.233 (4.92)** (4.03)*^ (1.14) Private? -0.157 -0.010 (0.83) (0.06) Incomputercap 0.840 (4.39)** Constant 3.861 -12.819 -9.638 -9.992 -11.807 (5.48)** (1 7.75)** (10.32)** (9.72)** (11.25)** Observations 165 165 165 165 152 R-squared 0.14 0.83 0.86 0.86 0.88 Absolute value of t statistics in parentheses + significant at 10%; * significant at 5%; ** significant at 1% Year dummies included Table 6 Price Regulation and Intemet Access Prices In 2000 Dependent variable: 30 off-peak hours of dial-up Intemet Ume per month Mean of dependent variable 28.7055 29.08 ISP prices regulated? 48.690 46.448 40.453 42.992 42.336 40.003 39.954 36.385 37.166 (2.74)' (2.46)' (2.14)' (2.24r (2.15)- (2.03)+ (2.02)+ (1.88)+ (1.79)+ GDP per capita -0.001 0.003 -0.001 -0.001 -0.000 -0.002 1-0.002 0.001 (0.41) (0.67) (0.27) (0.34) (0.07) (0.49) (0.55) (0.18) Telephone mainlines per 100 people -1.065 (1.44) Number telephone mainlines (millions) -.9409402 -.8668842 -1.337605 -3.238662 -3.488897 -2.737031 (0.99) (0.86) (1.24) (1.33) (1.47) (1.05) Incumbent privatized 3.778 3.197 0.001 2.849 0.983 (0.26) (0.22) (0.00) (0.19) (0.06) Trade as a share of GDP -0.175 -0.123 0.000 -0.077 (1.16) (0.75) (0.00) (0.41) GDP (milions USD) 134.3397 438.5419 93.95396 (.87) ( L93r (0.56) Number personal computers per thousand people -.2069728 (0.72) Number personal computers (millions) -24.59824 (1.82)+ Constant 21.990 25.413 31.846 28.891 26.981 42.710 42.107 33.964 40.407 (3.33)" (2.38)' (2.80)** (2.57)' (1.97)+ (2.22)* (2.18)* (1.73)+ (1.97)+ Observations 29 29 29 29 29 29 29 28 28 R-squared 0.22 0.22 0.28 0.25 0.25 0.30 0.32 0.42 0.34 Absolute value of t statistics in parentheses + significant at 10no; * signifiant at 5%; ** signifcant at 1 % Policy Research Working Paper Series Contact Title Author Dale for paper WPS2952 The Effects of a Fee-Waiver Program Nazmul Chaudhury January 2003 N. Chaudhury on Health Care Utilization among the Jeffrey Hammer 84230 Poor: Evidence from Armenia Edmundo Murrugarra WPS2953 Health Facility Surveys: An Magnus Lindelow January 2003 H. Sladovich Introduction Adam Wagstaff 37698 WPS2954 Never Too Late to Get Together Baitlomiej Kaminski January 2003 P. Flewitt Again: Turning the Czech and Slovak Beata Smarzynska 32724 Customs Union into a Stepping Stone to EU Integration WPS2955 The Perversity of Preferences: The Qaglar Ozden January 2003 P. Flewitt Generalized System of Preferences Eric Reinhardt 32724 and Developing Country Trade Policies, 1976-2000 WPS2956 Survey Compliance and the Johan A. Mistiaen January 2003 P. Sader Distribution of Income Martin Ravallion 33902 WPS2957 Mexico: In-Firm Training for the Hong Tan January 2003 H. Tan Knowledge Economy Gladys Lopez-Acevedo 33206 WPS2958 Globalization and Workers in Martin Rama January 2003 H. Sladovich Developing Countries 37698 WPS2959 Wage Differentials and State- Michael M. Lokshin January 2003 P. Sader Private Sector Employment Choice Branko Jovanovic 33902 in the Federal Republic of Yugoslavia WPS2960 The Poverty/Environment Nexus in Susmita Dasgupta January 2003 Y. D'Souza Cambodia and Lao People's Uwe Deichmann 31449 Democratic Republic Craig Meisner David Wheeler WPS2961 Strategic Planning for Poverty Rob Swinkels January 2003 N. Lopez Reduction in Vietnam: Progress and Carrie Turk 88032 Challenges for Meeting the Localized Millennium Development Goals WPS2962 High Consumption Volatility: Philippe Auff ret January 2003 K. Tomlinson The Impact of Natural Disasters? 39763 WPS2963 Catastrophe Insurance Market in the Philippe Auffret January 2003 K. Tomlinson Caribbean Region: Market Failures 39763 and Recommendations for Public Sector Interventions WPS2964 Wages and Productivity in Mexican Gladys L6pez-Acevedo January 2003 M. Geller Manufacturing 85155 WPS2965 Informality Revisited William F. Maloney January 2003 P. Soto 37892 WPS2966 Health and Poverty in Guaternala Michele Gragnolati January 2003 M. Gragnolati Alessandra Marini 85287 WPS2967 Malnutrition and Poverty in Alessandra Marini January 2003 M. Gragnolati Guatemala Michele Gragnolati 85287 Policy Research Working Paper Series Contact Title Author Date for paper WPS2968 Refining Policy with the Poor: Local Edwin Shanks January 2003 N Lopez Consultations on the Draft Carrie Turk 88032 Comprehensive Poverty Reduction and Growth Strategy in Vietnam WPS2969 Fostering Community-Driven Monica Das Gupta January 2003 M. Das Gupta Development: What Role for the Helene Grandvoinnet 31983 State? Mattia Romani WPS2970 The Social Impact of Social Funds Vijayendra Rao February 2003 P. Sader in Jamaica: A Mixed-Methods Ana Maria lbanez 33902 Analysis of Participation, Targeting, and Collective Action in Community- Driven Development WPS2971 Short but not Sweet: New Evidence Jishnu Das February 2003 H. Sladovich on Short Duration Morbidities Carolina SAnchez-Paramo 37698 from India WPS2972 Economic Growth, Inequality, and Richard H. Adams, Jr February 2003 N. Obias Poverty: Findings from a New Data Set 31986 WPS2973 Intellectual Property Rights, Guifang Yang February 2003 P. Flewitt Licensing, and Innovation Keith E. Maskus 32724 WPS2974 From Knowledge to Wealth: Alfred Watkins February 2003 A. Watkins Transforming Russian Science and 37277 Technology for a Modern Knowledge Economy WPS2975 Policy Options for Meeting the Francisco H. G. Ferreira February 2003 P. Sader Millennium Development Goals in Phillippe G. Leite 33902 Brazil: Can Micro-Simulations Help? WPS2976 Rural Extension Services Jock R. Anderson February 2003 P. Kokila Gershon Feder 33716 WPS2977 The Strategic Use and Potential Christopher Desmond February 2003 H. Sladovich for an HIV Vaccine in Southern Africa Robert Greener 37698 WPS2978 The Epidemiological Impact of an HIV Nico J. D. Nagelkerke February 2003 H. Sladovich Vaccine on the HIV/AIDS Epidemic Sake J. De Vlas 37698 in Southern India