POLICY RESEARCH WORKING PAPER 28 63 Getting Connected Competition and Diffusion in African Mobile Telecommunications Markets Frew Amare Gebreab The World Bank Development Research Group Regulation and Competition Policy June 2002 Poiicy RESEARCH-I WORKING PAP ER 2863 Abstract Gehreab studies the determinants of the diffusion of sequentia l entries into the m arket, but the malor effect of mobile telecommuiiuilcationis in Africa in a fixed effects competitioni on diffusionl ocCurs after the actual year of model. He uses data from 1987-2000 oil 4 1 African entry. The introductioll of digital techlinology has a countries that have adopted ceilular telecommunlications positive and significanit effect on the diffusion of iiiobile techiologies. He fmiids that competitioni is the driving phonies. The presence of an incubnient-owned cellilar force behind the mobile telecommuiluicationis explosioil in operator hlas a negative effect on the diffusion of Africa. mobiles, SuggeStiig an abuse of a dominlanit position hy Dluopoly and triopoly markets grow sigilificailtly faster the incuLImbIelnt fixed-line operator. However, thail moniopoly markets, although growthi does not privatization of thIe iicutiiheit fixed-lilne cellular appear to differ betweeii the first two markets. Evidence operator accelerates iilohile growth anld mitigates that of preemptive behavior is found in coillpetitive negative effect. TIhis paper-a product of Regulationi aind Comlpetition Policy, Development Researcl Group-is part of a larger effort in the group to proilmote telecomillmunications coulpetition, liberalizatioii, aidl privatizatioi in Africa. Copics of the paper are azvailable free from the World l3ank, 18 I cS H Strcet NW, Washingtoun D C 204.33. Pleasc cointact 'auhtina Siititnl-Aboagye, room MC3-422, telephone 202-473-8526, fax 202-522-1155, email address psintiilahoagye(i)worldhaik.org. Policy Research Workinig Papers are also posted on the Well at http://econ.worldhank.org. The authlor may be cotitacted at fgebreab@iworldbank.org. JuLic 2002. (33 pages) The Policy Research Working Paper Series disseminates the fnidi,ngs of Ivork ini progress to encouraige thise exc-hange of ideas abou0t dcuelopilientt issuies. An objective ofthe series is toget the findings out quickly, even if the/spresentations are less than foily polished. The papers carry the names of the autthors and should be cifed accordingly. 7The findings, interpretations, and conclisionis expresscd in this paper are entirely those of the auithors. Thsey do nlot necessarily represent the view of the W/orld/ Bantk Its Executtive Directors, or thte co(zntries they representt. Produced by the Researcll Advisory Staff GETTING CONNECTED: Competition and Diffusion in African Mobile Telecommunications Markets Frew Amare Gebreab* Competition is critical for rapid mobile expansion while participation in the mobile market by a state-owned incumbent operator has a negative impact on expansion. However, privatization of the incumbent operator that is involved in cellular mitigates that negative effect. Finally, network digitalization has a positive and significant impact on the diffusion of mobiles. * Development Research Group - Regulation and Competition Policy, World Bank. All findings, interpretations, and conclusions expressed in this paper are entirely those of the author and do not necessarily represent the views of the World Bank, its Executive Directors, or the countries they represent. I am grateful to Luke Haggarty, Scott Wallsten, Mary Shirley, Tilahun Temesgen, and L. Colin Xu for their excellent comments on earlier drafts. Contact information: Frew A. Gebreab, The World Bank, 1818 H Street NW, MSN 300, Washington, D.C. 20433. Phone: (202) 473-2903, Fax: (202) 522-1155, Email: fgebreab(.wor1dbank.org Table of Contents I. INTRODUCTION ................................................................ 4 II. TELECOMMUNICATIONS IN AFRICA ......................................................7 CELLULAR IN AFRICA ....................... ........................................................8 CELLULAR MARKET STRUCTURE & COMPETITION . ........................................................ 10 CELLULAR PENETRATION ................................... ........................................... I I CELLULAR TECHNOLOGIES IN AFRICA ........................................................................... 12 III. THE DATA ............................................................................. 13 IV. THE METHODOLOGY ................................................................ 14 V. THE RESULTS ................................................................ 18 THE EFFECT OF COMPETITION AND REGULATION ............................................................ 19 THE EFFECT OF MODE OF COMPETITIVE ENTRY ............................ .................................. 21 THE EFFECT OF TECHNOLOGIES .............................................................................. 22 THE EFFECT OF PRIVATIZATION AND INCUMBENT-OWNED CELLULAR .......... .................. 22 THE EFFECT OF COUNTRY CHARACTERISTICS .................................. ............................... 24 VI. CONCLUSION .......................................................................... 25 REFERENCES: ............................................................................. 29 Figures Figure 1: Main Lines Per 100 Population ............................ .....................................8 Figure 2: Ownership Structure of African Cellular Operators (2001) ........... ....................9 Figure 3: Cellular Subscribers per 100 Population in Afriica.11 Tables Table 1: Cellular Networks in Africa [2001] .......................................... .......................... 12 Table 2: The Estimation Results of Equation One ............................................................ 31 Table 3: The Estimation Results of Equation Two ........................................................... 32 Table 4: Cellular Entry and Technologies in Africa (countries selected for the study) .. 33 3 I. Introduction Technological changes have weakened the long-held argument of economies of scale and scope that favor a natural monopoly for the telecommunications industry. Therefore, competitive provision of telecommunications services has become a common phenomenon throughout the world, including Africa. Telecommunications competition in Africa has been seen primarily in the cellular segment of the industry, and rapid service expansions have been observed in many countries. For example, the average annual cellular expansion rate in Africa increased from 60 percent in the mid-90s to more than 100 percent in,2000 (ITU).' What determines the growth of mobile connections in Africa? It is'important because mobile technology growth has outpaced fixed wire growth in many countries in Africa, and yet is not the main source of telecommunications service for many people. A number of factors can explain this growth, including the traditional ones, such as long waiting-time for fixed line connections and overall dismal performance of the fixed telecommunications network. However, we have not seen any systematic explanation of this phenomenon in a developing country context although various case studies and anecdotal evidence have recently appeared. This study, therefore, tries to fill this gap by econometrically identifying the deterninants of mobile telecommunications diffusion/expansion using African data. Despite the lack of empirical evidence in a developing country context, there is some evidence of mobile diffusion from emerging and developed economies. Gruber (2001) analyzed the diffusion of mobiles in Central and Eastern Europe using a logistic fixed-effects model, and found that the speed of mobile diffusion increases with the number of firms, the size of the fixed telecommunications network and the length of the waiting list. In the same study, he also found that simultaneous entry is more effective than sequential entry in accelerating the speed of diffusion of mobiles. Gruber & Verboven (2000) also analyzed the evolution of the global mobile communications industry using the same logistic fixed-effects model, and found that timing of the first entry, competition and mode of the second cellular entry are the major determinants of 'Growth calculated by the author using ITU data. 4 the speed of mobile diffusions. They also found that income (as measured by GDP per capita), main lines and the waiting list have significant positive impact on the diffusion of mobiles. Furthermore, Gruber & Verboven (1999) studied the diffusion of mobile telecommunications in the EU using the same model, and reported that digitalization and competition had a significant impact on the diffusion of mobiles in the EU with the former having a lot more impact than the latter. They also reported that main line penetration has a negative impact on the speed of mobile diffusion, suggesting that mobiles are substitutes for fixed lines. While digitalization is an important explanatory factor in the EU study, it is not significant in either the CEE or the global mobile industry studies. The three studies also report divergent results on mode of entry and the size of the fixed network; however, they all agree on the fact that competition has a substantial impact on the growth of mobiles. This study aims, therefore, to shed some light on the issue and help resolve the observed anomalies. This study is of interest to policy makers because of the important link between telecommunications penetration and economic growth. Growth in telecommunications penetration is positively associated with economic development. Norton (1992), using data from 47 countries for the period 1947 to 1977 and controlling for the initial stage of the level of telecom development and a number of macroeconomic variables, finds that telecommunications has positive and significant impact on economic growth, and concludes that the existence of a telecommunications infrastructure reduces transaction costs since output rises when the infrastructure is present. Moreover, Roller and Waverman (2001) found a causal relationship between telecommunications infrastructure and aggregate output using OECD data for the period 1970 to 1990 after accounting for simultaneity and country-specific fixed effects. Appreciating this important link, multinational institutions, including the World Bank, have been showing increasing interest in issues of finding better ways of providing telecommunications services and increasing penetration (World Bank, 1994). 5 The paper is structured as follows: section 2 discusses telecommunications in Africa, and section 3 describes the data and their sources. Section 4 explains the model and section 5 presents the results. Finally, a concluding remark will follow. 6 II. Telecommunications in Africa Telecommunications in Africa, mainly fixed line services, can be characterized by low penetration, poor quality, and unreliability. For example, Africa, excluding South Africa, had only 1.82 main lines per 100 population in 1999 while East Asia had 8.23 main lines per 100 population in the same year and Latin America and the Caribbean (LAC) had 13.21 main lines per 100 population that year (Fig. 1). The situation was even worse in sub-Saharan Africa where telecommunications penetration was about 0.64 main lines per 100 population in the same year. The International Telecommunications Union (ITU) identified the following as the main reasons for the underdevelopment of telecommunications in Africa: (1) lack of investment, (2) investment inefficiencies, (3) inadequate private sector involvement, (4) foreign exchange scarcity, (5) poor management incentives and (6) insufficient regional development (ITU 1994). Recognizing this reality and pressured by technological changes and donors, most African countries started reforming their government-dominated and monopoly-based telecommunications sector beginning the mid-90s. Therefore, by 2001, about 17 African countries had already privatized their incumbent fixed-line operator and more were in the process of doing so, two countries had allowed competition in basic services, and more than 36 countries had created a separate regulatory body for the telecommunications sector. Moreover, about 45 countries have licensed private cellular operators and effective cellular competition has now emerged in many countries in the region. Almost all African countries have now introduced some kind of reforms in their telecommunications sector, and are at different stages of the reform process. 7 Figure 1: Main Lines Per 100 Population ---- L A C - -- E a s t A s ia _ A fifr ic a 1 2 I10 0- 1 9 9 0 1 9 9 1 9 9 2 1 9 9 3 1 9 9 4 1 9 9 5 1 9 9 6 1 9 9 7 1 9 9 8 1 9 9 9 Source. ITU, * Note Afnca doesn't include South Afnca Cellular in Africa Although the adoption of wireless telecommunications has been feasible since the late 1970s, it reached the African public mainly during the 1 990s.2 The introduction of cellular telecommunications in Africa disrupted the monopoly then enjoyed by African incumbent operators and for the first time opened access for private entrepreneurs to enter the telecommunications sector in Africa. By 2001, more than 90% of African countries had already adopted cellular telecommunications technologies compared to just 18% in 1993. In Africa, private operators have pioneered the development of cellular networks, and the incumbent fixed-line operators seem to have a relatively modest role. For example, of the total of 110 cellular operators in the year 2001, 60 percent were fully owned by private investors and 28 percent were joint ventures between private investors and the incumbent, and only 12 percent were full subsidiaries of the incumbent fixed-line operator (Figure 2). 2 This is except South Africa, Algeria, Egypt, Morocco, Tunisia, Congo (DR) and Mauritius which have had cellular since the second half of the 1980s. Figure 2: Ownership Structure of African Cellular Operators (2001) 12% 28/. 0 Fully Private Source: African Telecommunicanons Research Project, World Bank The leaders of the mobile miracle in Africa not only are private in their ownership, but also are pan-African and indigenous despite the presence of a number of foreign multinational carriers operating in different parts of the region. Mobile Systems International (MSI), MTN of South Africa and Orascom of Egypt are some of the indigenous pan-African operators. MSI was founded in 1998 by a Sudanese mobile communications expert, and it now holds mobile licenses in more than 14 African countries serving a population of over 250 million. Similarly, Orascom of Egypt, which was founded by an Egyptian entrepreneur, operates in more than 14 African countries and it is aggressively expanding in the region.3 Likewise, Mobile Telecommunications Networks (MTN), which is based in South Africa and owned by South Africans, has already expanded its service coverage to other five African countries, including Uganda, Swaziland, Rwanda, Cameroon and Nigeria. Other indigenous operators are also becoming pan-African and are increasing their market presence and level of operations 4 taking advantage of the changing telecom environment in the region. 3 See www.msi-cellular.com and www.orascomtelecom.com for further information. 4 Econet Wireless of Zimbabwe, which is owned by a Zimbabwean businessman, can be a good example. 9 Cellular Market Structure & Competition Most of the cellular markets in Africa are competitive although the level of competition differs from country to country. About 44% of the African cellular markets have two operators, 30% have three or more operators, and only 26% of the markets are monopoly.5 Africa has seen cellular competition in markets as small as Seychelles, which has a population of only 76,000. Competition in these small markets has helped disprove the idea that market size is a serious constraint to competition. The tendency in Africa is to issue nationwide licenses although some countries initially tried geographic-based licensing. Generally, cellular operators compete head-to- head against each other and the incumbent for local and long-distance traffic. As a result of the growing competition, the cellular market in Africa is being rapidly transformed. For example, cellular services which were limited to capital cities until recently can now be found in areas hundreds of miles away from the capital as operators compete for coverage. Almost all operators that started services using analog technologies have now migrated to digital technologies which allow for more customers to be served with the same amount of spectrum. The increasing shift from postpaid to prepaid services, which has led to rapid mobile use in the region, can also be attributed to the growing cellular competition. Cellular competition in most parts of Africa came later than the cellular services. For instance, more than 80% of the competitive entries in the region were introduced sequentially over an average period of 5.44 years (Table 4). This is not a surprising phenomenon as most African governments and several investors doubted the feasibility of telecommunications competition until recently. However, simultaneous entry has made headway in the region as governments increasingly appreciate the value of cellular competition. 5 Source: African Cellular Operators Database, African Telecommunications Research Project, World Bank 10 Cellular Penetration The growth of mobile penetration in Africa has been impressive. Cellular penetration had already surpassed fixed line penetration over half the continent by the end of 2001. Some countries, such as Gabon and Uganda, have seen their cellular system grow to more than three times of the size of their fixed line network. Almost all countries with more cellular than fixed, except Swaziland, have competitive cellular markets. Cellular penetration in Africa varies not only by market structure, but also by region. In 2000, for example, South Africa had 19.79 cellular phones per 100 population while North Africa had 2.79 cellular phones per 100 population and sub-Saharan Africa had only 0.56 cellular phones per 100 population (Figure 3). Smaller countries, such as Reunion and Seychelles, have already achieved as much as 50% penetration. Figure 3: Cellular Subscribers per 100 Population in Africa (South Africa right scale) 3.00 25 2.79 2.50 . - -/20 2.00~ ~ ~ ~~~~~~~~~~~~~~~~~~2 1.50 10 1.00 0.50 05 0.000 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Source: ITU _'AJI Africa SSA* -U- North Africa - South Africa (country) 11 Cellular Technologies in Africa A number of analog and digital systems have been tried in Africa. The Global System of Mobile Communications (GSM) standard is the leading technology, and has seen wider application in the region, followed by AMPS and TACS. Almost all entrants during the last five or six years have been deploying the GSM standard, and earlier entrants which started with other technologies have also been migrating to this digital system. The penetration of the US digital systems, such as DAMPS and CDMA that have better spectral efficiency than GSM, is still very low. The GSM standard is emerging as a default digital standard in Africa. This standard is about 4 times more efficient in its spectrum usage than most efficient analog systems, and hence has that much higher connection capacity (see Gruber and Verboven 2000). 6 GSM networks currently account for more than 95% of the total mobile subscribers in the region, which is high compared to 70% in the world.7 The region is likely to be fully digitalized in the very near future if the current trend of digitalization continues. Table 1: Cellular Networks in Africa 120011 Digital Analog Type Developer Number Type Developer Number GSM900 Europe 85 AMPS USA II DECT/WiLL Various 7 TACS UK 4 GSM1 800 Europe 5 E-TACS800 UK 2 CDMA USA 3 NMT-450 Scandinavian 2 D-AMPS USA I NMT-900 Scandinavian 1 TDMA IS-136 I RC2000 France/Italy I C-NETZ I E-TACS UK MATS I N-AMPS USA I Total 102 Total 2 Source: www.cellular.co.za/african-standards.htm, Access date 01/23/2002 6 Digital systems in general are 3 to 6 times more efficient than analog systems depending on the type of technologies (see Gruber and Verboven, 2000). 7See www.cellular.co.za, access date 01/23/2002 12 III. The Data The sample contains information on 41 African countries for the period 1987 to 2000. Four of the sample countries are from North Africa and the rest are from sub- Saharan Africa, including South Africa. Cellular services are available in all of our sample countries although service was introduced at different times. The dataset includes information on entry, competition, technology, regulation, the number of cellular subscribers, main lines, GDP per capita, population and % of urbanization. The data comes from different sources. The cellular and fixed line subscriber data comes from International Telecommunications Union (ITU) database, and the GDP, population and urbanization figures come from the World Bank SIMA database. The qualitative variables come from a combination of sources, such as EMC publications, Economist Intelligence Unit (EIU) publications, and http://www.cellular.co.za. Almost complete data, with very few exceptions, is available for all periods and countries. 13 IV. The Methodology I estimate the deterninants of cellular diffusion using a fixed-effects model controlling for different country characteristics. I also control for the effects of time using a non-linear time trend. First, I estimate equation (1) which simply defines the competition variable as the number of cellular operators to see the overall effects of competition, technological change, regulation and incumbent-owned cellular on mobile diffusion. yia = ai+ /31(ncel l) + /32(regi) + /33(ioci) + /4(prva) + S(dig ,) (1) + 9 (X,,) + &, Equation (1) uses a very general definition of competition. This specification doesn't capture the growth rate differences among the different market structures, nor does it capture the effects of having a separate sector regulator across the different markets. Therefore, I further refine the competition variable in equation (2) by segregating the markets into different market structures, such as monopoly, duopoly, and triopoly markets. Also, I introduce interaction terms between regulation and the market- structure dummies to see how different markets perform when interacted with regulation. Therefore, I estimate equation (2) with this refinement which will enable us to test various hypotheses, including whether there are significant growth variations among different market structures, and whether the effectiveness of regulation ( or having a separate telecom regulator) varies across different market structures. I will run equation (2) by dropping one of the market structure dummies alternatively to identify the relative significance of the included market structures as compared to the excluded one. y= a, + /3i(som n) + 132(dom ,) + /33(tom ,) + /34(som It* reg i) + /3s(dom it * rega) + /3 6(tom it * reg s) + 3 7(reg,,) + ,8(i C( i,,) + /39( pr i) + /3 io(prv , * ioc a ) (2) + ¢(digit) + y(cme ,) + 9(x,t) + ie Yjt is the logarithm of total cellular subscribers in country i at time t. It includes both prepaid and postpaid subscribers. This variable is the main explainable variable for which we are trying to unravel the determinants of its diffusion. 14 Ncelit is the nunber of cellular operators in country i at time t. As the number of cellular operators increases, available capacity and the intensity of competition is expected to rise, resulting in different competitive behaviors, such as price cuts, quality improvements, expansion of coverage. This leads to an increase in the number of people adopting cellular, and hence the variable is expected to be positively correlated with mobile growth. Regit is a dummy variable that captures the presence of a separate regulator in country i at time t. Therefore, this variable becomes one as soon as the country introduces a separate regulator for the telecommunications sector, and zero otherwise. A separate regulator is a specialized body that oversees the activities of all telecommunications sector players, and seems to be different, at least in form, from previous regulatory regimes, which were dominated by sector ministry and the incumbent. This variable could be positively or negatively correlated with mobile growth depending on whether the regulator is facilitating competition or is captured by players in the market. Iocit is a dummy variable that becomes one if country i at time t has a fully or partially incumbent-owned cellular operator, and zero otherwise. This variable can be either positively or negatively correlated with mobile growth depending on whether the incumbent is able to stifle competition. The presence of an incumbent-owned cellular in a competitive market could slow down the growth of mobiles if the incumbent abuses its dominant position. Prvit is a dummy variable that becomes one when the incumbent operator in country i at time t is privatized, and zero otherwise. This variable is expected to be either positively or negatively correlated with mobile growth depending on whether the privatization of the incumbent creates a more competitive or stifling environment. Somi, is a dummy variable that becomes one when the cellular market structure in country i at time t is a monopoly, and zero otherwise. 15 Domi, is a dummy variable that becomes one when the cellular market structure in country i at time t is a duopoly, and zero otherwise. As market structures change from a monopoly to a duopoly, mobile growth rates are expected to rise because of competition. Therefore, this will lead us to the following testable hypothesis: fl2 - f1 > 0 If the above hypothesis holds true, it implies that the coefficient of the duopoly dummy is positive and significant when the excluded category is the monopoly dummy only. Tom,, is a dummy variable that becomes one when the cellular market in country i at time t has three or more operators, and zero otherwise. Also, as the number of cellular operators rises to three or more, competition is expected to increase and hence mobile growth. Therefore, this will also lead us to the following testable hypothesis: P3 - 82 > 0 If the data supports the above hypothesis, it implies that the coefficient of the triopoly market dummy is positive and significant when the excluded category is the duopoly dummy. Alternatively, it also implies that the coefficient of the triopoly dummy is significantly higher than that of the duopoly dummy when the excluded category is the monopoly dummy. Cme1, is a vector of dummy variables for competitive modes of entries, such as sequential, simultaneous, and preemptive. (1) Sim,, is a dummy variable that becomes one as soon as country i at year t introduces a simultaneous competition, (2) Seqi, is a dummy variable that becomes one as soon as country i at year t introduces a sequential competition, (3) Pre1, is a dummy variable that becomes one only in the year prior to the introduction of the first sequential entry, and zero otherwise. This variable is expected to be positively correlated with mobile growth as incumbent mobile operators normally react to the potential threat of competition from a new entrant, (4) Seq1, (0) is a dummy variable that becomes one only in the year of the introduction of the first sequential entry, 16 and zero otherwise, (5) Seqit (-1) is a one-year lagged sequential competitive entry mode variable, and (6) Simit (-1) is a one-year lagged simultaneous competitive entry mode variable a and &£ are country specific fixed-effects and error terms, respectively. Xt is a vector of control variables, including main lines, population, per capita income, and degree of urbanization. These control variables are defined as follows: (a) Lnmli, is the logarithm of total main lines in country i at time t. A main line, as per the ITU's definition, is a telephone line connecting the subscriber's terminal equipment to the PSTN and has a dedicated port in the telephone exchange equipment. This variable can be positively or negatively correlated with the growth of cellular depending on whether cellular complements or substitutes fixed line telephony. (b) Lnpoit is the logarithm of total population in country i at time t, and this variable is expected to be positively correlated with mobile growth . (c)Lngdp1t is the logarithm of per capita real gross domestic product in US dollars of country i at time t, and it is expected to be positively correlated with mobile growth. (d) Urbit is the percentage of population living in urban areas in country i at time t, and it is expected to measure the level of urbanization of a country. People living in urban areas are usually considered to have a better income, standard of living, and awareness than those living in rural areas, and hence this variable is expected to be positively correlated with mobile growth. (e) Trdi, is the square of the time trend variable to control for time trend taking into account the non-linearity of the mobile diffusion over time. 17 V. The Results I present the estimation results of the above fixed-effects mobile growth model in this section. Table 2 presents the results of equation (1) and Table 3 presents the results of equation (2). Most of the results are consistent with our expectations. The competition variables in both equations are positive and significant, and all competitive markets grow faster than monopoly markets; however, no evidence is found to support the hypothesis that triopoly markets grow faster than duopoly markets. Triopoly markets, in fact, grow slower than duopoly markets when a separate regulator is introduced in the sector, suggesting the presence of a regulatory capture as markets become more competitive. A separate regulator also tends to decelerate mobile diffusion in both monopoly and duopoly markets, though the effects are not statistically significant. Evidence of pre- emptive behaviors in competitive sequential entries is also found. However, no evidence is found to support significant differences between simultaneous and sequential entries. In sequential entries, the major effect of competition on mobile growth occurs after the actual year of entry. Moreover, privatization of the incumbent fixed line operator significantly accelerates mobile growth while the presence of an incumbent-owned cellular significantly slows down the growth of mobiles. The latter result is consistent with the theory of the abuse of a dominant position by the dominant operator. As far as technology is concerned, digitalization is positive and significant. Furthermore, urbanization and main lines are positive and significant from among the country characteristics variables selected for the study. Income ( measured by real GDP per capita in US$) and population, which are traditionally perceived to be correlated with mobile growth, are not significant. The insignificance of the income variable could be partly because of its high and positive correlation with the main lines variable. All of the empirical results remain robust with different specifications of the model. I will thoroughly discuss the estimation results in the following sub-sections. 18 The effect of competition and regulation Table 2 presents the results of equation (1) which defines the competition variable as number-of-cellular-operators. The variable is significantly correlated with mobile growth, and the result suggests that adding one more mobile operator increases the total mobile subscribers by about 57 percent on average. Table 3 presents the results of equation (2) which defines the competition variable slightly differently by introducing separate dummy variables for monopoly, duopoly, and triopoly markets to test for any significant differences in mobile growth among the different market structures.8 The coefficients of the market structure dummies, which are reported in column V of Table 3, suggest that duopoly and triopoly markets grow significantly faster than monopoly markets; however, there is no sufficient evidence to support any significant differences in the speed of mobile growth between duopoly and triopoly markets. Although the speed of mobile growth in duopoly markets appears to be higher than that of triopoly markets, a Wald test conducted to test for differences between these coefficients did not find any significant differences. When splitting markets merely into competitive and monopoly, however, the coefficient of the competitive market dummy suggests that mobile growth in a competitive market in general is about 194 percent higher than that of a monopoly market. This effectively means that if a monopoly market grows by X percent, then a competitive market grows by 2.94X percent {=X+1.94X} (see column I of Table 3). 9 This strongly supports the effectiveness of competition to speed-up the diffusion of mobiles. However, the introduction of a separate sector regulator in triopoly markets significantly slows down mobile diffusion.10 As presented in column III of Table 3, the ' Although the triopoly market dummy covers markets with three operators by definition, it also includes five data points of four operators markets in Ghana, Madagascar, and Tanzania in our case. This is because the number of observations with four operators markets is so low that it does not justify a separate dumnmy variable. 9 Competitive in this case is defined as the presence of two or more mobile operators in the market. exp(l.077)- 1 = 1.94 '° The introduction of a separate regulator also tends to slow down mobile diffusion in monopoly and duopoly markets, but the effect is not statistically significant. 19 coefficient of the triopoly dummy is not significant when the duopoly dummy and the interaction terms are excluded, suggesting that mobile growth in triopoly markets is not different from that in duopoly markets. Nevertheless, when the interaction term between regulation and the market structure dummy is included in the same specification, the triopoly coefficient becomes negative and significant, regulation becomes negative and significant, and the interaction term between regulation and the triopoly dummy becomes positive and significant (see column IV of Table 3). Therefore, I conduct the following joint-coefficient test in order to determine the joint significance of the three variables when the interaction term is included: /,3 + ,!6 + 87 = O Where /33 is the coefficient of the triopoly dummy, /16 is the coefficient of the interaction term, and 817 is the coefficient of the regulation variable. The Chi-square value of the above restriction using a Wald test is significant at 7 percent significance level, indicating that the coefficients, which add up to negative 0.575 (= -1.098 + 1.057 - 0.534), are jointly significant and are different from zero. Therefore, this suggests that mobile growth in triopoly markets slows down when a separate regulator is introduced, further suggesting either the presence of a regulatory capture as markets become more competitive or the failure of the regulatory bodies to manage the increasing competition perhaps because of organizational weaknesses. The result is consistent with the numerous anecdotal evidence on the presence of a regulatory capture and the failure of many regulators to manage the increasing telecommunications competition in Africa partly because of organizational weaknesses. This finding is also theoretically plausible because the chances of competition- related intricacies that could slow down mobile growth increase as the number of operators increases.11 The increase in the incidence of these intricacies in turn makes the regulator more susceptible to regulatory capture by various interest groups, particularly I I The competition-related intricacies include, but not limited to, access to bottleneck facilities, interconnection, equipment co-location, numbering plans, spectrum allocations, and restricted services. These intricacies can slow down growth by themselves unless properly handled by an impartial regulator. 20 by the incumbent operator that may be interested to abuse its dominant position. Alternatively, the regulators may not have the required resources and skills to deal with the rising competition-related intricacies as the number of operators increases, resulting in the slow down of mobile growth as markets become more competitive, i.e. for triopoly or more markets. The effect of mode of competitive entry Equation (2) incorporates various entry variables to identify the effects, if any, of different modes of entries, such as simultaneous or sequential entries, and the presence of pre-emptive behavior in sequential entries. The result, as shown in column VIII of Table 3, suggests the presence of pre-emptive behavior in competitive sequential entries; however, no significant differences are found between simultaneous and sequential entries. The coefficient of the pre-emptive dummy suggests that the incumbent mobile operator increases its mobile subscriber size by about 38% in the year just prior to the actual entry of a second new entrant because of the threat of competition. This could be accomplished through substantial price cuts and/or expansion of coverage. Gruber and Verboven (2000) also reported the presence of pre-emptive behaviors in sequential entries in their global mobile communications study. Furthermore, the absence of significant differences between simultaneous and sequential entries seems to be consistent with other findings. Gruber and Verboven (2000) reported that sequential entry is more effective in accelerating growth than simultaneous entry in their global mobile communications study; however, Gruber (2001) later found that simultaneous entry is more effective to speed up mobile diffusion in his CEE study. Therefore, our finding of the absence of significant differences between sequential and simultaneous entries is in a way consistent with the prior inconsistent findings. In sequential entries, the major effect of competition occurs after the actual year of entry. The coefficients of the sequential entry variables suggest that a mobile market grows, ceteris paribus, by 0.997 in the year of competitive sequential entry, but it grows by 1.423 after the year of the actual entry (see column VIII of Table 3). 21 The effect of technologies In the technology discussion in section one, it is stated that digital systems are more efficient than analog ones by a factor of 3 to 6. This suggests that digital systems have less capacity constraint than analogs, and hence can bring, ceterius paribus, faster mobile growth. Digitalization in some countries in Africa started well before the introduction of the GSM technology, and these early digitalizers started with DAMPS and CDMA. Therefore, digitalization here refers to these three technologies although it is highly dominated by the GSM standard. As presented in both Tables 2 and 3, digitalization is positive and significant in all specifications. The coefficient of the digitalization variable suggests that when countries adopt a digital technology, mobile growth increases by about 66% on average. This suggests that countries that adopt a digital technology grow faster than those with other technologies. However, digitalization has mixed effects in other studies. Gruber and Verboven (1999) found digitalization to be positive and significant in their EU study, but same authors (2000) found digitalization to be insignificant in their Global Mobile Telecommunications study. Consistent with the latter finding, Gruber (2001) found digitalization to be insignificant in his Central and Eastern European study. Interestingly, our finding on digitalization is consistent with that found in the EU study which is also dominated by the GSM standard. The effect ofprivatization and incumbent-owned cellular Incumbent fixed line operators are the dominant players in African telecommunications markets. Thus, the involvement of the incumbent in cellular operations and changes in the management and ownership of the incumbent, for instance privatization, that are likely to improve its competitive edge theoretically affect the growth of mobiles. Some argue that allowing the incumbent to have its own cellular in a competitive market would defeat the very purpose of competition because (1) competition between un-equals is not likely to bring the expected results, (2) the 22 incunbent can easily cross-subsidize between its monopoly and competitive services to force cellular competitors out of the market, and (3) the incumbent can discriminate between its cellular subsidiary and competitors in providing access to bottleneck facilities. The discrimination in the third argument can take different forms, such as discrimination in the quality of the point of interconnections, in the prices of interconnections, and refusal to co-locate equipment that could potentially drive the investment and operational costs of competitors. The weakness or absence of telecommunications regulators in Africa further strengthens the validity of this argument. On the other hand, privatization gives incumbent operators better incentives to improve performance, but it could also further strengthen the incumbent to better exploit available opportunities through the above potential discriminatory practices. Therefore, the effect of privatization on mobile growth could be either positive or negative while the presence of an incumbent-owned cellular is expected to slow down mobile growth. Consistent with our expectation, the presence of an incumbent-owned cellular is negatively correlated with mobile growth in both equations (1) and (2) while privatization of the incumbent is positively correlated with mobile growth in both equations. This suggests that the presence of an incumbent-owned cellular in a mobile market significantly slows down mobile growth while the privatization of an incumbent operator accelerates it. The negative effect of an incumbent-owned cellular on mobile growth suggests the presence of an abuse of a dominant position by incumbent fixed-line operators. In competitive markets, the abuse of a dominant position could take the form of subtle denials of bottleneck facilities and subsidization of competitive services to stifle competition as discussed above. In monopoly markets, however, the abuse could take the forn of investment inefficiency and under investment. It should be remembered here that incumbent fixed-line operators in Africa in general have a very dismal record in the provision of telecommunications services, and most still suffer from rampant inefficiencies. 23 Furthermore, I introduce an interaction term between the incumbent-owned cellular dummy and the privatization dummy in order to better understand the effect of these variables when interacted. When the interaction term is included, the coefficient of the incumbent-owned cellular dummy remains negative and significant, the interaction term becomes positive and significant while the privatization dummy becomes insignificant (see column VIII of Table 3). This suggests that the positive effect of privatization comes not just simply from privatization of the incumbent fixed-line operator, but rather from privatization of the incumbent fixed-line operator that is involved in cellular. Overall, the results suggest that the negative effect of having an incumbent-owned cellular in the market can be mitigated through privatization. This is also consistent with the conventional wisdom that privatization improves performance. The effect of country characteristics Country characteristics variables, such as income, population, urbanization and main lines, are included in both equations and all specifications as presented in Tables 2 and 3. From among these country characteristics variables, urbanization and main lines are positive and significant while income (as measured by GDP per capita) and population are not significant. 12 The fact that urbanization is positive and significant suggests that the diffusion of mobiles is mainly concentrated in urban centers, and hence highly urbanized countries grow faster. This looks fairly reasonable as people living in urban centers are believed to have a better income, standard of living, and a higher propensity to adopt new technologies. Regarding main lines, its positive significance suggests that mobiles complement fixed line networks, further suggesting that mobiles are mainly getting into the hands of people who already have some form of access to the fixed network. The positive significance of main lines could also be due to positive network externalities from the fixed network that is that mobile becomes more attractive 12 The coefficients of the main lines and GDP per capita variables should be interpreted with some caution as the two are highly positively correlated. The correlation of these two variables in our data is about 0.94; therefore, the insignificance of income in our study could be because of the mnain lines variable. It is interesting to note that both become significant when they are run separately (see Columns IX and X of Table 3). 24 than before with the expansion of the fixed network because of the possibility of calling more people. In general, the coefficients of the urbanization and main lines variables are consistent with each other and both indicate that mobiles are following the footsteps of the incumbent fixed-line operators which don't have much presence in rural areas. Other authors reported mixed findings on the effect of main lines and income on mobile growth. Gruber and Verboven (2000) reported positive and significant coefficient for both income and main lines in their global mobile communications study; however, same authors (1999) found main lines to be negative and significant and income to be positive and significant in their EU study. In another study, Gruber (2001) found main lines to be positive and significant and income to be insignificant in his study of the diffusion of mobiles in Central and Eastern Europe. These findings suggest that mobiles are perceived as substitutes for fixed lines in developed markets, such as the EU, while they are perceived as complements to fixed lines in relatively less developed markets, such as Central and Eastern Europe. The coefficient of the main lines parameter in our study is also consistent with this categorization. VI. Conclusion This paper has analyzed the determinants of mobile telecommunications diffusion in Africa in a fixed-effects model. The determinants of mobile diffusion that this study has identified can be categorized into six major themes, including competition, regulation, technological change, privatization, and the presence of an incumbent-owned cellular. Policy differences in relation to these themes affected the diffusion of mobile telecommunications in Africa quite differently, explaining the mobile growth variations among countries in the region. First, mobile competition (as measured by the number of cellular operators) is positive and significant, and the result suggests that mobile subscribers increase by about 57 percent as an additional operator enters the market. Both triopoly and duopoly markets grow faster than monopoly markets; however, no evidence is found to support significant mobile growth differences between duopoly and triopoly markets. Triopoly 25 markets, in fact, grow slower than duopoly markets when a separate regulator is introduced, suggesting either the presence of a regulatory capture as the number of operators increases or the failure of the regulators to manage the increasing competition as markets become more competitive. The latter could be because of lack of the required resources and skills by the so-called separate regulators to deal with the rising competition-related intricacies as markets become more competitive. The presence of a separate regulator in monopoly and duopoly markets also seems to have some negative effect on the growth of mobiles, but the effect is not statistically significant. This could be because the incidence of competition-related intricacies and the chances of a regulatory capture are low in such markets. Second, evidence of pre-emptive behavior is found in competitive sequential entries. The coefficient of the pre-emptive dummy suggests that the incumbent mobile operator increases its subscriber size by 38 percent just in the year prior to the actual entry of a second mobile operator. This result is consistent with the threat of competition argument. On the other hand, no significant evidence is found to support differences between simultaneous and sequential entries. In sequential competitive entries, the major effect of competition on mobile growth occurs after the actual year of entry. Other studies reported inconsistent results on competitive mode of entries: Gruber and Verboven (2000) found sequential entry to be more effective in their global mobile communications study while Gruber (2001) found simultaneous entry to be more effective in his CEE study. Third, digitalization is positive and significant. The coefficient of the digitalization dummy suggests that the speed of mobile diffusion accelerates as countries adopt a digital technology. Gruber and Verboven (2000) also found digitalization to be positive and significant in their EU study. However, Gruber and Verboven (1999) and Gruber (2001) did not find digitalization to have an appreciable impact on the diffusion of mobiles . Our finding is consistent with the finding of the EU market study which is also dominated by the GSM standard. 26 Fourth, the presence of an incumbent-owned cellular is negative and significant, suggesting that the involvement of the incumbent fixed-line operator in mobile operations slows down the growth of mobile telecommunications. This is consistent with the theory of an abuse of a dominant position, and with the dismal performance record of African incumbent fixed-line operators. Privatization of the incumbent fixed-line operator is; however, positive and significant, suggesting that privatization accelerates mobile growth. Further analysis indicates that it is the privatization of the incumbent fixed-line operator that is involved in cellular, not just the privatization of the incumbent per se, that accelerates the growth of mobiles. This is also consistent with the notion that privatization improvers performance. Therefore, while the presence of an incumbent- owned cellular decelerates mobile growth, the privatization of an incumbent fixed-line operator that is involved in cellular mitigates that negative effect. Finally, urbanization (as measured by the % of urban population) and main lines are positive and significant; the rest of the country characteristics variables, such as income and population, are not significant. The positive significance of the urbanization variable suggests that mobile services in Africa are mainly concentrated in urban centers. Similarly, the positive significance of the main lines variable suggests that mobiles are perceived as complements to fixed lines; alternatively, it also suggests the presence of positive network externalities from the fixed network. The fact that urbanization and main lines are positive and significant are consistent with each other, and both indicate that mobiles are following the footsteps of the incumbent fixed-line operators which don't have much presence in rural areas. Furthermore, as far as the relationship between mobiles and fixed lines is concerned, findings from other studies suggest that mobiles are perceived as substitutes for fixed lines in developed markets, such as the EU, while they are perceived as complements to fixed lines in relatively less developed markets, such as Central and Eastern Europe. Interestingly, the coefficient of the main lines parameter in our study is consistent with these findings, i.e. mobiles in Africa are perceived as complements, at least at this stage, to fixed lines. 27 All of the above findings have relevant policy implications, and it is my hope that telecommunications policy makers in Africa can learn important lessons from this study. In the future, it is interesting to explore how competition, regulatory qualities, investor qualities, and other policy variables affect prices, access to the poor and qualities of services. It will also be interesting to consider regulatory and investor quality measures, which we have not included in this study because of the paucity of data, in future mobile diffusion studies. Also, one may want to model the mobile diffusion problem using the logistic model, which is widely used in diffusion studies, as a sensitivity analysis to the results found in this study. The author of this paper has plans to explore these issues. 28 References: EMC Publication (2001), "Africa & Middle East Mobile Communications Report", Detailed overview of the wireless markets of Africa, Middle East and Arab World, April, August and October. EMC Publication (2002), "Africa & Middle East Mobile Communications Report", Detailed overview of the wireless markets of Africa, Middle East and Arab World, February. Gruber, H. (2001), "Competition and Innovation: The Diffusion of Mobile Telecommunications in Central and Eastern Europe", Information Economics and Policy, 13, No. 1:19-34, March. Gruber, H. and Verboven, F. (1999), "The Diffusion of Mobile Telecommunications Services in European Union", Discussion Paper Series No. 2054, Center for Economic Policy Research, January. Gruber, H. and Verboven, F. (2000), "The Evolution of Markets Under Entry and Standards Regulation - The Case of Global Mobile Telecommunications", Discussion Paper Series No. 2440, Center for Economic Policy Research, April. International Telecommunications Union (1994), "World Telecommunications Development Report." International Telecommunications Union, (various years), "World Telecommunications Indicators Database." International Telecommunications Union (2001), "African Telecommunications Indicators", November. Norton, S. (1992), "Transaction Costs, Telecommunications and the Microeconomics of Macroeconomic Growth", Economic Development and Cultural Change, Volume 41, Number 1, October. Pyramid Research (1999), "Wireless Markets and Strategies in Africa", The Economist Intelligence Unit (EIU), Ltd., March. Roller, L. and Waverman, L. (2001), "Telecommunications Infrastructure and Economic Development: A Simultaneous Approach", The American Economic Review, Volume 91, Number 4, September. Wallsten, S. (1999), "An empirical Analysis of Competition, Privatization, and Regulation in Telecommunications Markets in Africa and Latin America", Policy Research Working Paper # 2136, World Bank. 29 World Bank (1994), "World Development Report: Infrastructure for Development", World Development Indicators, Oxford University Press. World Bank, (various years), "World Development Indicators Database", SIMA Statistical Information Management & Analysis. WWW.cellular.co.za/african-standards.htm. Access date 01/23/2002 WWW.msi-cellular.com, Access date 04/05/2002 WWW.Orascomtelecom.com, Access date 04/05/2002 30 Table 2: The Estimation Results of Equation One Dependent variable: LN (# of Mobile Subscribers) I II Number of operators 0.474 * 0.424 (4.12) (3.64) Separate regulator 0.103 0.098 (0.59) (0.57) Incumbent-Owned Cellular -1.426 * -1.390 (-3.65) (-3.58) Privatization 0.604 * 0.621 * (2.59) (2.68) Digitalization 0.563 * (3.13) GSM 0.681 * (3.73) % of urban population 0.087 ** 0.085 (2.17) (2.14) LN (Main lines) 1.007 * 0.941 (3.24) (3.04) LN (Population) 0.661 1.372 (0.20) (0.42) LN (GDP per capita) 0.780 0.876 (0.73) (0.82) Time trend 0.016 * 0.015 (4.12) (3.96) Country fixed-effects (average) Adj. R 0.89 0.89 F-stat 257 262 Number of countries 41 41 Total observations 279 279 Note: t-stats in parentheses. *Significant at 1% significance level. *- Significant at 2% significance level. **Significant at 5% significance level. ***Significant at 10% significance level. 31 Table 3: The Estimation Results of Equation Two Dependent variable: LN (# of Mobile Subscribers) I II III IV V VI VII VIII IX X Monopoly market -1.132 * -1.558 (-6.42) (-6.20) Duopoly market 1.132 ^ 1.558 * 1.203 - (6.42) (6.20) (6.90) Triopoly market -O 281 -1.098 '- 0.851 0 460 0.928 (-I 237) (-2.56) (3.41) (1.20) (3.77) Competitive 1.077 1.254 0.952 1.079 (6.31) (5.64) (5.52) (6.35) Separate Regulator 0 006 0.138 0030 -0.534"' 0 030 0.165 0001 0052 0.125 -0.007 (0.039) (O 70) (O 18) (-1.86) (O 18) (O 84) (O 1) (0.31) (O 74) (-O 04) Separate Regulator 0.699 Monopoly (2.18) Separate Regulator -0.699 Duopoly (-2.18) _ Separate Regulator * 1.057 0 358 Triopoly (2.29) (0.89) Separate Regulator -0 343 Competitive (-1.24) Incumnbent-Ownied Cellular -1.441 -1.318 -1.380 -1.148 -1.380 -1.148 * -1.847 * -1.464 * -0.832'- -1.443 (IOC) (-3.92) (-3.47) (-3.73) (-3.00) (-3.73) (-3.00) (-4.71) (-3.95) (-2.43) (-3.93) Incumbent OC * 1.209- Pnvatization (3.17) Privatization 0.670 0.696 * 0700 * 0.705 * 0.700 ' 0.705 * -0079 0.508 0.773 0.665' (3.02) (3.13) (3.14) (3.18) (3.14) (318)_ (-0,24) (2 23) (3.41) (3.01) Digitalization 0.474 * 0.498 0.513 * 0.547 * 0.513 * 0.547 * 0.502 * 0.531 * 0.604 * 0.471- (2.78) (2.91) (2.97) (3.18) (2.97) (3.18) (2.96) (3.13) (3.51) (2.78) Simultaneous entry (-I) 1.509 _9_(3.sn Pre-emptive 0.324 '' (1.95) Sequential entry (0) 0.997 * (4.85) Sequential entry (-1) 1.423 (5.82)_ _ _ _ _ % of urban population 0.112' 0.112- 0.116* 0.113 0.116 0.113 0.113- 0.133 0.129- 0.111 (2.89) 2.8 (297) (2.92) (2.97) (2.92) (2.96) (3.38) (3.24) (2.88) LN (Main lines) 1.159 * 1.123 1.164 * 1.035 1.164 * 1.035 * 1.216 1.136 * 1.198 (3.88) 5 (3.90) 3.45) (3.90) (3-45) (4.15) (3.84) (4.57) LN (Population) 1.004 1 100 1.320 1 990 1 320 1 990 0 886 1.375 5.058 02854 (O 32) (O 35) (O 42) (O 64) (O 42) (O 64) (O 29) (O 44) (1.67) ( ) LN (GDP per capita) 0 288 0 449 0 107 0.630 0.107 0 630 0 489 0 372 2.182 (0.28) (043) (0.10) (O 60) (0. 10) (O 60) (o 40) (O 36) (2.34) Time trend 0.014 * 0013 * 0.013 * 0.012 0.013 0.012 * 0.013 0.011 * 0.012 0.014 - (3.62) (3.57) (3.45) (3.31) (3.51) (2.89) (3 23) (3.81) Country fixed-effects (average) Adj. R- 0 90 0.90 0 90 0 90 0 90 0.90 0 90 0.90 0.91 0 92 F-stat 284 256 256 218 256 218 243 219 299 320 Numberofcountries 41 41 41 441 1 41 41 41 41 41 Total observations 279 279 279 279 279 279 279 279 279 279 Note: t-stats in parentheses *Significant at 1% significance level. *-Significant at 2% significance level. **Significant at 5% significance level. ***Significant at 10% significance level. 32 . .1 , . . . . __ . . _ . ': rMD ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ 0 cc A ., pX i : (|: "°''° Dj t r -- -;-E - --o°- N ; ---- If I iTITE 1; iIIt j . . . . . . * . . . . r n _XXL X e _ - - - GJ *0i4 1i.lL L--.t- Policy Research Working Paper Series Contact Title Author Date for paper WPS2837 Reform, Growth, and Poverty David Dollar May 2002 E. Khine in Vietnam 37471 WPS2838 Economic Mobility in Vietnam Paul Glewwe May 2002 E. Khine in the 1990s Phong Nguyen 37471 WPS2839 Marketing Externalities and M. Shahe Emran May 2002 F. Shilpi Market Development Forhad Shilpi 87476 WPS2840 Public Spending and Outcomes: Andrew Sunil Rajkumar May 2002 H. Sladovich Does Governance Matter? Vinaya Swaroop 37698 WPS2841 Contractual Savings in Countries Gregorio Impavido May 2002 P. Braxton with a Small Financial Sector Alberto R. Musalem 32720 Dimitri Vittas WPS2842 Financial Sector Inefficiencies Pierre-Richard Agenor May 2002 M. Gosiengfiao and the Debt Laffer Curve Joshua Aizenman 33363 WPS2843 A Practical Guide to Managing David Scott May 2002 L Yeargin Systemic Financial Crises. A Review 81553 of Approaches Taken in Indonesia, the Republic of Korea, and Thailand WPS2844 Money Demand in Venezuela: Mario A. Cuevas May 2002 M. Geller Multiple Cycle Extraction in a 85155 Cointegration Framework WPS2845 The Spatial Division of Labor Marcel Fafchamps May 2002 F. Shilpi in Nepal Forhad Shilpi 87476 WPS2846 Is India's Economic Growth Leaving Gaurav Datn May 2002 C. Cunanan the Poor Behind? Martin Ravallion 32301 WPS2847 The Nature and Dynamics of Poverty Hippolyte Fofack May 2002 P White Determinants in Burkina Faso 81131 in the 1990s WPS2848 Administrative Barriers to Foreign Jacques Morisset May 2002 M. Feghali Investment in Developing Countries Olivier Lumenga Neso 36177 WPS2849 Pooling, Savings, and Prevention: Truman G. Packard May 2002 T. Packard Mitigating the Risk of Old Age 75841 Poverty in Chile WPS2850 Determinants of Commercial Bank David A. Grigorian June 2002 S. Torres Performance in Transition: An Vlad Manole 39012 Application of Data Envelopment Analysis WPS2851 Economic Development and the Bernard Hoekman June 2002 P. Flewitt World Trade Organization After Doha 32724 WPS2852 Regional Agreements and Trade in Aaditya Mattoo June 2002 P. Flewitt Services: Policy Issues Carsten Fink 32724 WPS2853 Private Interhousehold Transfers in Donald Cox June 2002 E. Khine Vietnam in the Early and Late 1990s 37471 Policy Research Working Paper Series Contact Title Author Date for paper WPS2854 Rich andPowerful? Subjective Michael Lokshin June 2002 C. Cunanan Power and Welfare in Russia Martin Ravallion 32301 WPS2855 Financial Crises, Financial Luc Laeven June 2002 R. Vo Dependence, and Industry Growth Daniela Klingebiel 33722 Randy Kroszner WPS2856 Banking Policy and Macroeconomic Gerard Caprio, Jr. June 2002 A. Yaptenco Stability: An Exploration Patrick Honohan 31823 WPS2857 Markups, Returns to Scale, and Hiau Looi Kee June 2002 M Kasilag Productivity: A Case Study of 39081 Singapore's Manufacturing Sector WPS2858 The State of Corporate Governance: Olivier Fremond June 2002 G. Gorospe Experience from Country Mierta Capaul 32623 Assessments WPS2859 Ethnic and Gender Wage Mohamed Ihsan Ajwad June 2002 Z. Jetha Disparities in Sri Lanka Pradeep Kurukulasuriya 84321 WPS2860 Privatization in Competitive Sectors. Sunita Kikern June 2002 R. Bartolome The Record to Date John Nellis 35703 WPS2861 Trade-Related Technology Diffusion Maurice Schiff June 2002 M. Kasilag and the Dynamics of North-South Yanling Wang 39081 and South-South Integration Marcelo Olarreaga WPS2862 Tenure, Diversity, and Commitment: Somik V. Lall June 2002 Y. D'Souza Community Participation for Urban Uwe Deichmann 31449 Service Provision Mattias K. A. Lundberg Nazmul Chaudhury