Document(s) 16 of 21

Telecom operators around the world have long considered serving low-income markets as risky. Achieving profitability on the back of low-average revenue per user (ARPU) customers was considered nearly impossible, resulting in the gap between urban and rural areas widening progressively. With the reluctance of established players to roll-out rural networks, policy makers were compelled to encourage such roll-out through subsidies packaged in various forms. Critical analysis and debate on subsidy schemes seek continued refinement of such schemes, so that the intended beneficiaries can be served, not only efficiently, but also effectively and transparently.

The world's poor have been benefiting from micro-financing schemes for more than three decades, in very diverse cultures and economies. Quite naturally, telecom related micro-finance schemes aimed at providing access and services to the rural masses have evolved, especially in Africa, the Indian subcontinent, and Latin America. The Grameen Village Phone in Bangladesh and its adaptations in Uganda, the Communication for Empowerment (CoE) project in Brazil (Pontes, 2003), community owned networks in Peru, and the n-Logue business model in India are some examples for such operations. Most of these operations are based on franchise business models and multi-stakeholder partnerships (MSPs), with access to services facilitated through micro-finance. These operations served as an eye opener to some telecom companies (henceforth telcos) in realizing that there is money to be made at the bottom of the pyramid (BOP).

Today, ARPU is a widely accepted metric for measuring the success of a telco. Over the past decade, the industry has expressed dismay at declining ARPUs. Part of this dismay stems from the assumption that declining ARPU will lead to declining profits. However, the successes achieved by Grameenphone type franchise operations and some telcos in low-ARPU climates have punctured this correlation.

The focus has now shifted from ARPU to AMPU (average margin per user). AMPU is the difference between the revenue generated by a user and the cost of serving that user. AMPU can be either negative or positive and, therefore, low-ARPU need not preclude a positive AMPU. Greater the AMPU, the greater the profit and the enhancement of shareholder value.

Analysts and strategists who seek to develop high-AMPU business models that work in low-ARPU climates need a sophisticated understanding of the essential drivers and enablers that can make their models successful in tapping the riches at the BOP, including:

• A deeper understanding of customer behavior in low-income markets; to meet demand-side expectations and economic considerations, including affordability and ownership issues.
  
  
• Major cost factors that affect AMPU, in particular, those that operators can control.
  
  
• Key technological enablers and cost-efficient technology solutions.
  
  
• Transferable lessons from proven 'low-ARPU' business models.
  
  
• Regulatory reforms and preconditions conducive to growth and economic benefits to the society, including taxation and subsidization.

The research presented in this book sheds much light on the drivers and enablers from a regional perspective, and will be useful to analysts and strategists designing business models that work in low-ARPU climates. Lack of research findings based on regional data and applications have always posed a challenge to regional telcos in the past, as their extrapolated business models, adopted from the developed world, have often fallen short.

DEMAND-SIDE EXPECTATIONS IN LOW-INCOME MARKETS

In developing low-ARPU business models, a sophisticated understanding of customer behavior in low-income markets is essential. For telcos the basic unit of analysis is the customer. Telcos need to determine whether or not current and prospective customers add value. The key to understanding customer economics, especially in low-ARPU climates, is to disaggregate customer behavior to identify the leading indicators of customer value that ultimately drive shareholder value. Christensen and Raynor (2003, pp. 74–75) argue that 'the functional, emotional, and social dimensions of the jobs that customers need to get done constitute the circumstances in which they buy'. Unfortunately, many telcos use aggregated metrics to measure customer value, although such measures do not always get to the real reason a customer 'hires' the service.

What Do Users at the BOP Want?

The 'micro' analytical approach adopted in the study reported in Section 1 could serve as a further step towards disaggregating and understanding customer behavior in low-ARPU markets. The researchers seek to shed more light on 'functional, emotional, and social dimensions' associated with 'hiring' telecommunication services by users at the BOP.

The study represents people who earn less than US dollars (USD) 100 per month in 11 localities in India and Sri Lanka. Random purposeful sampling has been used to select a limited number of households within larger purposeful samples of priori heterogeneous localities, and Kish grid techniques have been used to select respondents to ensure against age and gender biases. As explained by Patton (1990), while random purposeful sampling adds credibility to a sample, when the potential purposeful sample is larger than one can handle, the technique is not recommended for generalizations or representativeness. In fact, the researchers state 'that admitted that the data does not represent India and Sri Lanka as wholes, only the "financially constrained" within the enumerated localities'.

The researchers have further stated that the issues of consistency and replicability will be addressed in the new study while a control group consisting of the financially 'unconstrained' will also be studied, in order to make more robust comparisons. Accordingly, while awaiting more representative and replicable findings from the new study, analysts and strategists could draw inferences from the present findings to develop business models for low-ARPU environments.

The research yielded a number of thought provoking findings. The finding that 99.7 percent of Sri Lankans and 88 percent of the Indians at the BOP approached for the research had used a telephone during the previous three months is startling and challenges the conventional inferences drawn from the metric 'teledensity' (the number of telephones per one hundred inhabitants). Amongst the BOP users that have been sampled, almost two thirds (58 percent) do not own the phone that they use. Additionally, 31 percent of fixed owners and 7 percent of mobile users allow non-family members to use their phones. These findings challenge the conventional metrics for telecom access.

The above set of findings has implications for telcos, given that one connection serves more than one user's needs. The inadequacy of the number of telephones per hundred inhabitants as the sole basis for forecasting potential service usage in a low-ARPU climate is amply demonstrated here. Other metrics like average MOU (minutes of use) per subscriber require attention.

The finding that phone ownership doubles as income moves from the 'below USD 50' to 'USD 50 to 100' segment seems to corroborate the contention of Nokia (2003) that halving the total cost of ownership (TCO) is the key to doubling the penetration and quite naturally opens up for discussion the issues related to TCO, and specifically, the cost of handsets—considered by many mobile operators to be the major barrier in serving low income groups.

Mobile handsets for low-income segments should satisfy three major expectations: affordability, usability, and widespread availability. Handset makers and the GSMA have been working on getting the cost down below USD 30 to help enough potential users overcome the 'handset barrier'. To that end the GSMA, through their Emerging Markets Handset (EMH) initiative, contracted Motorola to develop an 'ultra-low cost' mobile handset. Motorola has got the price down to less than USD 30 during 2006.

The EMH initiative rests on generating economies of scale for the low-cost handset supplier, through assured high volume purchases. In the initial phase, nine operators joined the collective tender process: AIS (Thailand), Bharti (India), Maxis (Malaysia), Globe and Smart (the Philippines), Telenor (Bangladesh, Pakistan, Kazakhstan, Russia, Ukraine), Turkcell (Turkey), Orascom (Africa, Middle East, Bangladesh, Pakistan), and GSMA board sponsor SingTel (Singapore). Each of the operators committed to purchase at least 500,000 handsets, thereby assuring sale of 6 million handsets in the first six months. In the second phase of the EMH initiative, GSMA allowed any GSM operator in an emerging market to participate. Mobile operators in developing countries have already bought or ordered more than 12 million mobile phones through the initiative. The operators involved in the second phase are: AIS, Bharti, BPL, Globe Telecom, Hutchison Essar, IDEA Cellular, MTN Group, Orascom Telecom, Telenor, and Vodacom. This ultra-low cost segment will potentially include 1.5 billion people (GSMA, 2005).

The usability expectations of such ultra-low cost handsets include functionality, power consumption and device longevity. Trade-offs in terms of handset design and functionality are unavoidable. Some argue that features such as color screens, built in cameras, and additional applications like WAP and MP3 are superfluous in emerging markets. On the other hand, others doubt whether such no-frills phones would be accepted even by 'financially constrained' users. They argue that a handset is an 'aspirational' item for the target user. Somewhat relatedly, the present study has established that 'symbolic' factors (such as 'fashion' or 'improved social status') feature in the user's choice, especially among newer adaptors. However, all agree that features such as a rugged exterior, strong radio frequency reception, and extended battery life are essential in such handsets.

The above requirements compel manufacturers to take a different approach to issues such as battery life and alternative power sources, especially in areas with unreliable electricity supply. Motorola's C115—the handset designed for phase 1 of GSMA's EMH initiative, for instance, is said to have a two and a half week standby time and eight hours of talk time—the longest in the world at the moment.

Widespread availability is the other requirement to lower the 'handset barrier'. Issues like operator-led distribution versus grey markets, as well as availability of recycled, refurbished or second-hand handsets, are important. Though the present study does not venture deeply into these aspects, the finding that 33 percent of Indian respondents used second-hand handsets indicates the potential of recycled, refurbished or second-hand handsets. The comparatively low use of second-hand handsets among the Sri Lankan respondents suggests further investigation of regulatory and/or trade barriers and logistic capabilities.

Supplying ultra-low cost handsets is not enough. Users must be able to afford the service. What is important is the TCO—which includes both handset and service costs. PMN Publications (2005), a UK-based research and consultancy firm specializing in mobile telecoms industry, states that 'realistic' monthly ARPU may be as low as USD 5 to 6 in emerging markets but even at these ARPU levels, the cost of buying a USD 40 mobile telephone and paying for wireless service in the first year would represent about 23.6 percent of per capita GNI in the 'low income' countries (as designated by the World Bank), 7.2 percent in 'lower middle income', and 5.5 percent in 'middle income' markets. However, there is a danger of viewing these figures from a first-world viewpoint, where the mobile handset is a very personal device, and to argue that it may not be economic to provide individual mobile services at the BOP. A wide range of alternative options for making mobile telephony available through group purchasing and micro-loans have been tried out successfully in countries like Bangladesh (discussed in Chapter 5), India, the Philippines, and Uganda.

Furthermore, while fixed and public-access users perceived the cost of telecom services to be affordable, the majority of mobile users perceived the cost to be either 'high', 'very high', or 'extremely high', especially amongst respondents in the Sri Lankan samples. Only 23 percent of mobile users have found mobiles to be affordable, compared to 56 percent and 59 percent of fixed and public access users, respectively. This sends a clear signal, especially to mobile operators, that they should develop calling plans responsive to these wishes. In India, for example, rates have plummeted to as low as Indian Rupee 1 (US cents 2.2) per minute for long-distance calls ('OneIndia' Tariff plan).

Wireless, especially cellular mobile, is the preferred technology in deploying network infrastructure in new growth markets because of the low level of initial investment, scalability, relatively simple technical deployment, low-cost, open standards, and the adaptability to both voice and data. However, a majority of mobile users at the BOP perceive the cost of mobile access to be high. Nevertheless, users are still willing to spend a considerable amount of their incomes on telecom services, even when they find the cost of these services to be high (64 percent of 'financially constrained' mobile users at the BOP spent at least USD 4 per month on mobile communication, which is at least 4 percent of heir monthly income).

The willingness to spend considerable amounts on mobile communications, in spite of the perception that mobile communication is costly, stems from convenience factors rather than limitations in access to other modes. Of the mobile users surveyed, only 24 percent have stated that the reason for using mobile communication was because they had no other choice. In fact they have cited convenience factors such as 'ability to access at any time' (71 percent), mobility (63 percent), and ease of use (61 percent) as the reasons for choosing mobile communications.

Knowledge that the 'financially constrained' are not driven solely by price but give considerable importance to convenience indicates that there could be a market for some premium services in such segments, provided the services add value.

In tailoring calling plans, whether basic or premium, telcos need to understand the applications that those at the BOP look for. The research shows that relationship maintenance is a priority. Such use includes keeping in touch with friends and family as well as sending news and messages. What is striking is that instrumental use is low, except for a small percentage who make business transactions and enquiries on the phone. However, the reasoning of the researchers that 'what the respondent broadly categorizes as "keeping in touch" might well have instrumental aspects also' due to 'a significant barter economy in existence at this level of society' may not be the only contributory factor towards the 'close to non-existent' instrumental use. Non-availability of applications and content and the lack of consumer education could be another major reason. Therefore, telcos, while promoting packages aimed at social usage, should also develop applications and content aimed at instrumental uses aspired to by the 'financially constrained'.

The research has estimated that the share of income spent on telecoms by the 'financially constrained' is in the range of 1–4 percent for fixed access users and 4–8 percent for mobile access users. The share of income spent on telecom services by the 'financially constrained' is far higher than the 2–3 percent rule-of-thumb generally used in developed countries and compares well with the 10 percent figure suggested in other contemporary research studies. This is good news for telcos as ARPUs amongst the 'financially constrained' may even reach USD 8 to 10 per month. The ARPU may increase further, should the telecom services provided to the 'financially constrained' result in improving their well-being and uplifting their spending capacities over time—as recent research (Ureta, 2005) has shown that expenditure on communications has a positive elasticity when a country moves from 'least developed' to the 'medium developed' status.

Strategies on a Shoestring

The 2005 study reveals that teleusers at the BOP are constrained in strategic use of telecom services. A majority of the 'financially constrained' have access to a phone for a limited period of time during the day and at specific places, thereby denying them the freedom to choose the kind of a phone/access mode to use and the freedom to 'mix and match' modes.

Table 11.1 recaps on the strategies available for users to manage their telecom expenditure and adoptability of such strategies by 'financially constrained' users, together with some observations of the author. This summary table would be a useful guide to telcos in the region, for targeting the low-ARPU market of 'financially constrained' users.

Apart from the accessibility limitation, the other reason that has been given for low use of 'strategies' by the 'financially constrained' is their low level of use of telecom services. The researchers point out that many, or all, of the few calls that the 'financially constrained' make are non-discretionary or unavoidable, thereby preventing them from being 'strategic' in their use. However, even among the 'financially constrained', the 'heavier' users are more 'strategic' than the 'lighter' users. Therefore, since expenditure on communications has a positive elasticity as income gradually increases up to certain levels (Ureta, 2005), one could expect to see the 'financially constrained' users becoming more 'strategic' as their income increases over time.

ACCESS FOR ALL: KEY ENABLERS, EFFECTIVE SOLUTIONS

Connecting the 'financially constrained' and keeping them connected largely depends on the ability of telcos to develop and implement high-AMPU business models that work in low-ARPU environments. In this context, it is necessary to identify the major cost factors, especially those which operators can control.

Table 11.1
Adoptability of Available Strategies at the BOP

Key Technological Enablers and Cost-effective Solutions

Demand density and distance from the center are the two key parameters that need to be considered in determining the access technology suitable for serving a particular locality or community, as shown by Figure 11.1, adapted from a presentation by Intelecon Research & Consultancy Ltd. (2002).

Figure 11.1
Access Technologies

Source: Author, adapted from Intelecon Research & Consultancy, 2002.

Wireless, especially cellular mobile, is the preferred technology in new growth markets. The key to keeping the BOP connected profitably (or achieving positive AMPU) is to reduce network costs.

Achieving the right balance between customer needs and network costs is essential in developing high-AMPU business models. As correctly highlighted by the Boston Consulting Group (2001):

Network design requires weighing the cost of functionality and new features against the value they provide to customers. Network engineers typically push for the most advanced technology available, while marketing departments want the best possible service. All too often, this dynamic leads to expensive solutions with far too many bells and whistles. Telcos need to make sure that the features they invest in are the ones that customers see as important—and thus worth paying for.

Technology at Work in Developing Countries: Patchwork Adoptions, Innovative Adaptations

Although there are effective technological solutions to control costs, the economic, social, technological, and regulatory circumstances in many developing countries hinder their deployment.

The result has been uneven adoption of technologies, leading to the coexistence of obsolete and state-of-the-art technologies, as against the normal smooth S-curve. This phenomenon of 'patchwork adoption of technology' (Garcia-Murillo, 2003) is evident not only in Latin America but in Africa and Asia. The resulting inefficiencies preclude achievement of least cost solutions.

Wi-Fi

Indonesia perhaps presents the most intriguing example of technology deployment in the context of patchwork adoption and innovative adaptation. The adoption of the relatively new technology Wi-Fi in conjunction with the conventional PSTN and leased line based services along with the innovative 'local' adaptations employed to condition the Wi-Fi technology, make the Indonesian experience unique. The findings on Wi-Fi 'innovations' in Indonesia in Chapter 6 make interesting reading not only for policy makers but also for telcos who strive to serve the BOP profitably.

As much as Mauritius and Macedonia, the first two countries in the world to offer nationwide wireless Internet access (with Singapore likely to join in the near future), the Indonesian case justifies recognition—not only for the extent of Wi-Fi coverage but also for the manner in which it has overcome numerous barriers.

The explanation that scarcity of infrastructure (the last mile and backbone) triggered the 'innovations' in Indonesia is acceptable. However, the argument that absence of regulations to unbundle the local loop and reluctance of incumbent and/or monopoly operators to invest (in order to safeguard their monopolistic profits) could cause such scarcities may not necessarily hold in many a developing country. In fact, this is a common misconception among some policy makers, and is conceived through undue reliance on solutions designed to address specific scenarios in developed countries. Unlike in the developed world where last mile infrastructure is abundantly available for unbundling, the problem in the developing world is the scarcity of last mile infrastructure and inability (not reluctance) of incumbent operators to make adequate investments for infrastructure development.

It is common knowledge that many government-owned incumbent monopolies have to contribute all their earnings to the government and receive only a fraction of it back for reinvestment. Of course, one could argue that the government-controlled monopoly operator and the government are one and the same. However, in reality, the investment priorities and objectives of the government and the operator are distinct and separate. Even when such markets have been opened for competition with government-owned incumbents partially or fully privatized, 'blanket' application of asymmetrical regulation by regulators in some developing countries (supposedly 'to level the playing field' for new entrants) have effectively inhibited the incumbents from investing in infrastructure in a commercially prudent manner. For instance, attempts by Sri Lanka Telecom (the partially privatized incumbent of Sri Lanka) to deploy CDMA technology (to lower its access costs) were not allowed by the regulator for seven years, ensuring that its competitors entered the CDMA market first. Such policies may have been adopted to offer competitive advantage to the new entrants, as it is known that emerging technologies offer competitive advantages to new entrants over the incumbents (Clemons, Croson and Weber, 1996). On the other end of the scale, some policies have prevented new entrants from investing in infrastructure, as shown in Chapter 5.

Another question that arises is whether Local Loop Unbundling (LLU) can be 'the' solution to promote competition in emerging telecommunication markets? The problem in the developing world is not unbundling the local loop but 'building' more local loops. In fact, Spiwak (1999) states that 'if telecoms regulators truly want restructuring to work, then they must view unbundling in the correct context. Remember, what we don't need are more "competitors" or "choices" per se—what we need are more loops'. He further points out that 'as long as regulators do absolutely nothing to expand the already extremely constrained market for local loop facilities, then consumers will suffer in the long-run with the current "static, incumbent-centric perpetual resale model", because neither the incumbents nor new entrants will have any incentive to invest in new plant'. Another practical reality in the developing world is the status of their copper networks on the ground. A fair share of these are not in the condition to carry higher bandwidth technologies as ADSL2+ or a multiplicity of services over realistic distances, making LLU irrelevant to broadband applications.

Chapter 6 sends out three forceful messages to policy makers wishing to promote cost effective Information and Communication Technologh (ICT) roll-out in developing countries: liberate 2.4 and 5.8 GHz bands to facilitate Wi-Fi deployment, authorize infrastructure development by new entrants through facility-based licensing and reduce international-bandwidth and domestic leased-line costs.

Perhaps the most important message is that Wi-Fi deployment can be commercially viable and sustainable with private investment. This is good news for telcos developing innovative and profitable business models to serve the BOP. Of course, the Indonesian experience is not directly applicable in other developing countries as violations of regulations like the use of unauthorized frequency bands and unauthorized resale, etc., could be risky. Furthermore, the role played by the civil society in insulating such activists against crackdowns by authorities may be unique to Indonesia. However, if policy makers could remove the bottlenecks identified in Chapter 6, telcos could 'fine tune' the Indonesian business model to serve the BOP profitably.

Increasing Internet Adoption in Indonesia

Chapters 4 and 6 show that the progress made in the ICT sector in Indonesia is 'largely due to the untiring efforts of the Indonesian civil society'. The dynamics of the civil society's influence on ICT sector development is captured in the following schematic (Figure 11.2). This figure has been prepared by combining the schematic flows presented in Chapter 4 and a presentation by a co-author of the chapter.

Figure 11.2 depicts a structured model that could be used for ICT development elsewhere.

Figure 11.2
ICT Sector Development in Indonesia—the Dynamics of Civil Society Influence

Source: Adapted from Chapter 4 and http://sandbox.bellanet.org/~onno/ (Retrieved on September 4, 2006).

The key question that would then arise in relation to the Indonesian model is, 'Why does the Indonesian civil society appear to be more influential and effective than its counterparts in other developing countries?' The answer, as seen in Figure 11.2, is the effectiveness of ICT-related knowledge sharing processes in Indonesia. In fact, Purbo, in an interview with IDRC, insists that the major barrier is the education process:

The barrier [is] actually not the tool. The barrier would be the education process: To educate the society to share the knowledge within the society, to encourage the society to produce their knowledge in local languages. That's a major barrier. So, it's not the tool, it's not the money, it's not the funding…. Of course, the government also creates some sort of barrier in the public regulation. [Governments] need to liberalize the regulation to enable…low cost Internet access.¹ Dr. Purbo's method [of educating the community] is twofold; communicate with people through public speaking, articles, and books; and educate others to do the same.²

Lessons from Proven 'Low-ARPU' Business Models

Grameenphone type franchise operators as well as some 'regular' telcos have succeeded in achieving profitability with low-ARPU customers. At the 3GSM World Congress (February 2005), Bharti Tele-Ventures of India revealed that it was running profitable mobile operations on a monthly prepaid ARPU of USD 5. (During the quarter ended December 2004, Bharti Cellular had achieved Earnings Before Interest, Taxes, Depreciation, and Amortization [EBITDA] margins of 36.4 percent.) It was also reported that their business model would work with a monthly ARPU as low as USD ³.

For the first six months of 2004, Smart, a mobile operator in the Philippines, reported an EBITDA margin of 69 percent on an average prepaid monthly ARPU of USD 7. In Sri Lanka, the mobile operator Dialog Telekom posted a 54 percent EBITDA margin for the first nine months of 2005 on a blended monthly ARPU of USD 7 and an average prepaid monthly ARPU of USD 4 (nearly 80 percent of its customer base is prepaid).

The saturation of demand in urban areas has fueled interest in developing business models that could work in low-ARPU environments and tap the riches at the BOP. Suppliers and vendors of network equipment and handsets too have joined the bandwagon, seeking to keep their long-term business opportunities alive.

Learning from Grameen

Chapter 5 uses the conceptual framework of Transaction Cost Economics (TCE) and argues that it is 'a model of transaction cost minimization that has tremendous appeal in the context of the low-trust/credibility conditions that are often prevalent in countries with weak laws and governance capacities'. While attributing the success of the Grameen solution to its 'in-house' transaction cost minimization model (where all parts of the process remain within the Grameen 'family'), the chapter argues that technology plays a lesser role compared to the ability to understand the market and the associated transaction costs. These conclusions give rise to several queries regarding the Grameen solution and its replicability.

¹ Can production cost be ignored in critically analyzing and learning from the Grameen model?

Though the total cost of a firm is made up of both production and transaction costs, in the classical microeconomic modeling it was assumed that transaction cost was zero. However, empirical evidence in the US has shown that the transaction cost could even exceed 50 percent of the total cost (especially in the case of private sector).³ With the emergence of TCE, both production and transaction costs are considered equally important. Therefore, viewing the Grameen model only in the context of transaction cost minimization, while ignoring the inherent production costs, would not ensure that any solution modeled on Grameen would deliver the optimum AMPU. In fact, a major criticism of the Grameenphone (GP) service is its high tariffs (Amin, Amin and Hussain, 2004) and inefficiencies in the utilization of technology. GP's use of cellular technology for fixed phone centers is neither efficient nor competitive over the long run (if not for the cross subsidy from urban users) and is more expensive than the fixed wireless local loop (WLL) systems used by its competitors (Cohen, 2001).

2. Can the success of the Grameen solution be attributed solely to 'in-house' cost minimization?

In devising its solution, Grameen Bank (GB) formed two entities: Grameen Telecom (GTC), a wholly-owned non-profit organization to provide telephone services in rural areas, and Grameenphone Limited, a for-profit entity that provided a national GSM cellular service (in partnership with US, Norwegian, and Japanese companies) (Cohen, 2001). The 'in-house' model has enabled GP to minimize several transaction-cost components related to billing, collection and credit control (including creditworthiness assessment), but it is doubtful whether the resulting transactional cost savings could allow the rate discount of 50 percent currently enjoyed by GTC. Village Phone Operators (VPOs) account for only 3.85 percent of the subscriber base of GP, whereas the rest of the subscribers are mostly in urban areas. GP can discount the rate charged to GTC for a phone call, through cross-subsidy. In fact, analysts⁴ have pointed out that 'Grameen Telecom's village phone venture as structured in Bangladesh would not be feasible without access to the credit and bill collection services provided by Grameen Bank and the infrastructure and urban network provided by Grameenphone.' They further point out that:

Village phones would be far less successful if GP were not able to discount by 50 percent the rate charged to GTC for a phone call, an underlying subsidy made possible by a transfer of profits from the more profitable urban part of the business to the rural sector—and a significant advantage unavailable to rural-only competitors. (Cohen, 2001)

This is a contention that is not rejected by the author of Chapter 5.

Regulation: To Stifle or Enable?

The challenge of expanding telecom networks to reach the whole population needs to overcome two separate 'gaps': the 'market efficiency gap' and 'the real access gap'.

High-AMPU business models seek to bridge the above 'market efficiency gap' by pushing the limits of commercial feasibility of conventional business models towards the 'affordability frontier'. However, success will greatly depend on the regulatory environment. If the environment is conducive, such models may be enhanced to bridge the 'real access gap' as well.

Chapter 8 investigates the conditions necessary to make 'smart' subsidies successful in bridging the real access gap in rural telecom services, taking Nepal's Eastern Development Region as the case.

The analysis defines 'smart' subsidy as 'the process used to provide the minimum required subsidy to bridge a defined access gap using a competitive bidding process, known as least-cost subsidy (LCS) auctions'. It is argued that in countries with poor governance frameworks, it is 'safer' to use the smart subsidy approach where only one number is evaluated, than a 'beauty contest', where the award is determined by an assessment of the applicant's ability to fulfill a given set of requirements. However, it appears that the LCS auction in Nepal has not provided the anticipated 'safeguards' against 'undesirable outcomes' of poor governance, as, (a) the decision to prevent the incumbent from bidding, (b) the use of rather 'liberal' eligibility conditions for potential bidders, (c) the decision to go ahead with the LCS auction, with only one eligible bidder on board (though it is reported that there were other potential bidders who had shown sufficient interest to participate in the process), (d) selection of a party with limited experience as a telecom operator, and (e) the strong 'technology bias' of the selected party (who actually was a large VSAT manufacturer). Chapter 8 makes a point that is normally not highlighted, but is of crucial importance, regarding the competence, impartiality and leadership of the government officials and consultants who drive the process.

Even without 'the poor governance issues', the Nepalese LCS attempt would have 'failed' due to reasons ranging from the unfavorable regulatory and politico-security environment, critical flaws in the design and implementation of the LCS process, failure to address sustainability of the project, and the lack of mechanisms for mid-course correction. However, the lessons learned from the Nepalese experience, together with the lessons from more successful Latin American and Indian experiences, indicate that three key areas are essential for the success of 'smart' subsidies: a conducive regulatory environment, a proper design and implementation process, and measures to ensure sustainability.

Regulatory Preconditions

The Latin American cases showed that sound regulatory framework is a precondition to make 'smart' subsidies work. A key to the failure in Nepal was the unfavorable regulatory environment. A favorable regulatory regime is a key ingredient not only in designing and implementing workable 'smart' subsidy programs but also in ensuring sustainability. Chapter 8 describes the good regulatory practices needed to make a 'smart' subsidy process work. The focus here is on the underlying objective of regulation.

The objective of regulation is consumer welfare, not limited to the short-run. In trying to achieve this goal, regulators devise various solutions, depending on the nature of the problem and the environment. The solutions may necessitate achievement of several sub-objectives. Therefore, achieving the same objective could necessitate achieving different sub-objectives in different environments. It is, therefore, futile to measure the regulatory success in terms of achieving sub-objectives such as implementing LCS projects, implementing LLU, facilitating a number of competitors or choices in the market, etc.

Design and Implementation of 'Smart' Subsidies

For any auction to be successful, there must be competition among bidders. As highlighted in Chapter 8, the use of lowest-cost technology is the key among the factors that could facilitate competition among bidders. High-AMPU business models that work in low-ARPU climates depend on the mobilization of cost-effective technology. In this context, 'technology neutrality' is essential to the success of a 'smart' subsidy. Other factors are: ways of attracting bidders with operational experience, assuring service-quality and availability obligations, and securing roll-out obligations and sustainability. The discussion here is confined to whether incumbent telcos should be prevented from bidding for LCS.

There appears to be reservations about the participation of incumbents in LCS auctions. For instance, in Nepal the incumbent was not allowed to participate. However, in some of the Latin American countries and in India, the incumbents were allowed to bid.

Allowing incumbents to compete in LCS auctions could enrich the outcome to the intended beneficiaries (the aspiring rural telecom user) in several ways. As presented in Chapter 8, incumbents' entry into the LCS process creates competition for the market; they will strive to defend their territories, while the new entrants will fight to gain a foothold in the demarcated license territories. The Latin American experience has shown that this competition (for the market) forced all the bidders to consider the most cost effective technology for the given geographic and socio-economic conditions, resulting in a reduced subsidy structure at the outset.

One reason for the failure of the business plan of the successful bidder in Nepal was the overestimation of demand and revenue. They have complained that accurate data was not made available to them. However, such problems would not occur if incumbents are allowed to participate. Incumbents are already in the business of providing services in the country, or more specifically in or around the intended areas for LCS. In most cases, this would allow the incumbents to utilize the subsidy more efficiently in rolling out networks, than an inexperienced, vendor-centric 'operator' experimenting with a business model based on inaccurate projections.

The usual argument against incumbents being allowed to participate in LCS bidding is that notwithstanding the monopoly profits and cross subsidies the incumbents enjoyed in the past, they have found it difficult to justify the extension of their networks to remote areas and if allowed to participate in an LCS program, they could resort to anti-competitive practices. However, it is up to the regulator to ensure that the process is so designed that it does not allow anti-competitive practices.

Sustainability of LCS Projects

Some literature⁵ suggests that the role of a 'smart' subsidy is to kick-start rural telecom roll-out, without creating subsidy dependency. The obvious concern that arises out of this contention is the sustainability of a project. While the overall sustainability of the project may depend upon a number of factors, the bottom line is that the operation has to be viable at least in the medium-run. Successful rural LCS projects show that the key to viability is cost-based asymmetric interconnection arrangements.

The logic for implementing asymmetric interconnection is two fold; the rural telecom operator should be adequately compensated for the high operating costs and also for the low revenue yield (de Silva and Tuladhar, 2006, p. 45). Though many would agree that the cost of rural operations is high, and, therefore, the termination costs should reflect this reality, the need to compensate for low revenue yield could be debated. However, as pointed out by Samarajiva (2002), the usual perception that the economic viability of connections depends solely on the originating revenues—when call termination is offered as a bundled 'free' service—is wrong; each 'free' call reception is accompanied by a revenue-generating call origination elsewhere on the network.

Therefore, it is obvious that 'smart' subsidies coupled with cost-based asymmetric interconnection regimes could facilitate the development of high-AMPU business models that are sustainable—-to bridge the market efficiency and access gaps observed in rural telecom markets.

INDIA'S UNIVERSAL SERVICE FOR TELECOM: POLICY AND REGULATORY GAPS

India used a smart subsidy program to disburse funds for universal access through its Universal Service Obligation Fund (USOF). Between 2003 and 2005, the USOF ran five LCS auctions, handing out a total of USD 500 million in subsidies. The fixed wire-line incumbent operator Bharat Sanchar Nigam Limited (BSNL) managed to win the majority of LCS auctions. Chapter 9 analyzes the design and implementation of the Indian universal service obligation (USO) program.

The Indian USO regime has produced mixed results as compared to the positive experiences in Latin America and the negative experience in Nepal. On the positive side, the costs of projects have declined over time; from bids matching exactly the benchmark value in the initial implementations, to 60–75 percent less from the benchmark value by March 2005.⁶ On the other hand, the Indian process is criticized for restricting participation, as the design gave an advantage to the incumbent BSNL.

Technology Restrictions

The importance of maintaining technological neutrality in universal-service provision is undisputed. However, this 'theory' should be applied objectively and intelligently, without ignoring the available knowledge on the capabilities, limitations and applications of extant technologies. Chapter 9 states that until recently, only fixed wire lines and wireless local loop technology qualified for USO subsidies. A correction to this policy is now said to be in the pipeline: to support cellular mobile services (both GSM and CDMA) as well. The chapter recommends that the proposed amendment should also allow emerging technologies such as WiMAX.

However, the Indian approach of restricting USO fund support initially to fixed wire and wireless technologies, with GSM and CDMA services being brought in after three to four years and no support for emerging technologies such as WiMAX, is justified.

• The necessity for USO funding arises from the failure of incumbents to supply rural demand, despite their past monopoly profits and cross-subsidies. However, incumbents often have considerable infrastructure and capacity in place to support rural network roll-out, though the sharing of this infrastructure has become a subject of debate. Therefore, it would be beneficial to entice the incumbent to participate in rural roll-out, by providing USO funding support for wired lines, as the advantages of incumbents' participation would offset any disadvantages. The approach of the Indian policy makers seems to have been based on ground realities, rather than on theories. This approach seems to have produced results; costs of projects have declined over time. Apart from including wire-lines, the other technology that was allowed initially was fixed WLL. In fact, fixed WLL was the most cost-effective solution at the time, as CDMA deployment was complicated for regulatory reasons.⁷ It should be noted that WLL technology is more efficient than GSM technology, especially for fixed services. GSM towers can provide service up to 15–20 km, while WLL towers provide coverage up to 50 km. WLL provides better bandwidth for data transmission at lower cost and is more cost effective than VSAT technology (de Silva and Tuladhar, 2006, p. 38, footnote 37).

• The decision of late to fund cellular mobile (both GSM and CDMA) is a logical development, rather than a correction. With the completion of the unified licensing regime in January 2005,⁸ there is no need to impose restrictions on technologies any further, justifying the application of the unified licensing regime to subordinate policies like USO. Furthermore, since 2003, several low-cost GSM solutions have been developed by suppliers.
  
  
• Certified WiMAX equipment appeared in the market only in late 2005. Therefore, WiMAX was not an option in the previous USO subsidy auctions.

Infrastructure Sharing

In the USO funded projects, the bidders were responsible for setting up infrastructure, despite the incumbent having a fair amount of 'shareable' infrastructure in place. This made it difficult for operators without a large subscriber base in rural areas to enter those markets in a cost-effective way; Chapters 7 and 9 point out that the incumbent refused to lease out its infrastructure on cost-oriented and non-discriminatory terms to the other bidders and leveraged its advantageous vertically integrated status to win a majority of the auctions. Therefore, it is stated that an important precondition to these auctions would have been an effective access regime.

In discussing solutions to the above impasse, Chapter 9 implies that the only concern is to facilitate provisioning of services through new entrants; the rights of incumbents to function as commercial entities, develop business strategies and implement long-term business plans are ignored. As much as the new entrants, the incumbents need to have strategies, plans, technology migration road maps and viable business models. Costs involved in laying telecom cables, be it copper or optical fiber, are high and escalating by the day. Optical fiber cables are deployed to cater for projected demand of up to 25 years. Whether the incumbent is government owned or not, the huge investments for such projects have to be secured through presentation of proper business plans, based on projected demand and technology migration plans. Because the world is moving towards IP based Next Generation Networks (NGN) and India has its own agenda for NGN migration,⁹ BSNL would have had definitive plans for loading its fiber optic backbone over the next 25 years. In such circumstances, the disruption of BSNL's long-term business plans by mandating the sharing of its optical fiber backbone at 'reasonable' terms would discourage further investments in backbone infrastructure by the incumbent and by new entrants.

It is interesting to note the contention that 'there are no private property rights issues involved, as BSNL is a public entity and its infrastructure properly belongs to all citizens'. If that is the case, the question is whether the citizens would consent to hire 'their' infrastructure to private entities. By implication, the public has entrusted 'their' infrastructure to be managed by the incumbent BSNL. Whether the public approves of such disruption of long term business plans of the incumbent is an open question.

Chapters 7, 9, and 10 suggest that if the industry structure is changed to favor alternative ubiquitous distribution companies which sell infrastructure elements in direct competition with the incumbent, incentives will be created for the incumbent to offer its own infrastructure on competitive terms. The Indian regulator is seeking to resolve the infrastructure sharing impasse by recommending subsidies for infrastructure and mandating the sharing of infrastructure.

INDIA'S UNIVERSAL SERVICE FOR TELECOM: ACCESS DEFICIT CHARGE

Chapter 10 critically analyses the Indian Access Deficit Charge (ADC) experience to extract the key generalizable principles. The definition in Chapter 10 is slightly different from the standard definition¹⁰ of ADC; while the discussion here is based on the former as does not affect the arguments made.

Even though ADC is considered one of the mechanisms available for implementing universality policies, Intven (2000) consider it inefficient and akin to an anti-competitive cross-subsidy. They point out that while some regulators have rejected reform of ADC regimes, others are reviewing them. Since its inception, the ADC regime in India has been criticized by many quarters. Several stakeholders, including consumer organizations, have questioned its logic, pointing out that the objective of promoting universal service can be achieved through the USO Fund. Some even suggest that the ADC regime should be wound up, as was done in some other countries, rather than debate eligibility and other details (Agarwal, May 20, 2005).

The opponents point out that ADC was initiated when fixed subscribers outnumbered mobile subscribers and, therefore, any increases in their rental/tariff would have made basic telecom services un-affordable to many. However, they point out that circumstances have changed, with mobile subscribers now outnumbering fixed-line subscribers. Therefore, they argue that the rationale of ensuring affordable fixed services no longer applies (Agarwal, May 20, 2005).

ADC is a useful mechanism to protect incumbents during the transition to competition. As argued by Clemons, Croson and Weber (1996), dominant players enjoying near-monopoly positions (but often required to provide universal access) can be seriously damaged by 'cream-skimming' new entrants if cross-subsidy has to continue. On the other hand, if tariffs are rebalanced, the less attractive segments will face increased tariffs. Therefore, during the transition period to full competition, incumbents need ADCs to maintain below-cost tariffs in the less attractive segments while offering cost-based tariff in the attractive segments. Incumbents have been and continue to be the main providers of telecom services to the vast rural masses. The need to sustain those services without compromising affordability and quality of service cannot be denied. This is where the policy makers must balance the protection of the incumbent and the new entrant. After all, protection is extended to a new entrant not as an end, but as the means to safeguard the interests of the end-beneficiary—the general public. Similarly, during the transition period the incumbents too need protection, especially in view of the 'baggage' they carry. Of course, ADC mechanisms should be viewed as transitional and terminated once the desired objectives are achieved.

Chapter 10 traces the many changes in the ADC regime in India. It is possible to identify a trend to achieve simplicity and uniformity and to reduce the ADC progressively, taking advantage of the rapid growth in the number of minutes due to mobile growth. There appears to be convergence between the positions of Telecom Regulatory Authority of India (TRAI) and those advocated for in the chapter. ADC calculations have been made simple progressively and revenue sharing has been introduced for domestic traffic, though there are still some concerns on the methodology of reckoning revenue.¹¹ The need to merge ADC and USO has been accepted and a definitive time frame for abolishing ADC has been announced.

CONCLUSION

The theme of this chapter is the creation of business models that work in low-ARPU markets. The key to creating sustainable business models in such markets is to strive for positive AMPU. The sustainability of such models depends on continuous refinement. The BOP research can yield the inputs for such refinement.

Connecting the 'financially constrained' and keeping them connected depends on the ability of telcos to make a profitable business out of serving such segments. High-AMPU business models that can bridge the 'market efficiency gap' can be further developed to serve markets beyond the 'affordability frontier' if the regulatory environment is conducive.

NOTES

¹ Interview with Dr Onno W. Purbo, Barriers to ICT Diffusion to Poor People. http://web.idrc.ca/ev_en.php?ID=46217_201&ID2=DO_TOPIC

² IDRC Man on a Mission: Researcher Profile, Dr Onno Purbo. http://web.idrc.ca/ev_en.php?ID=45872_201&ID2=DO_TOPIC

³ ENCYCOGOV.COM Statistics: Production costs v. transaction costs in the US GDP. http://www.encycogov.com/

⁴ Ibid.

⁵. Dymond, Andrew. Telecom & Information Services for the Poor: Towards a Strategy for Universal Access. Focus: The Two Gaps—Market and Access. http://www.InteleconResearch.com

⁶ Ibid., p. 11, Table 2.2.

⁷ CDMA is essentially a mobile technology, and therefore the cost efficiency drops when the technology is deployed to provide fixed or limited mobile services. Though the Unified Licensing regime in India came into force towards the end of 2003, it was not mandatory for operators to seek a unified license. Given this scenario, the division of voice operators as fixed and mobile continued and 'fixed only' operators was permitted to deploy CDMA—only to provide fixed and/or limited-mobility services.

⁸ In India, the Unified Licensing Regime was implemented in two phases. The first phase of implementation was in November 2003, and covered both fixed and mobile services. In the second phase, Unified Licensing covered all telecom services (commencing from January 2005). Under the Unified Licensing scheme, a licensee shall be able to provide any or all telecom services by acquiring a single license. As at present, migration to Unified Licensing is optional for service providers. However, this will become mandatory after five years. Further, from 2007, all new service providers will be licensed under the Unified Licensing Regime. (Further information on the Unified Licensing regime in India could be obtained from the TRAI website.)

⁹ TRAI has already published a comprehensive consultative paper on NGN deployment. The paper is available for downloading at the TRAI website.

¹⁰ An access deficit arise when an operator's average access charges (line rental and connection) are not set high enough to cover the long-run average incremental cost (LRAIC) of providing an access service (http://itu-coe.ofta.gov.hk/vtm/universal/faq/q10.htm). Note that in the strict definition of Access Deficit, only the line rental and connection charges (that is, traffic insensitive component) comes in to play whereas the traffic sensitive component, which is the usage charge, is left out of ADC.

¹¹ See the report filed at http://www.tribuneindia.com/2006/20060224/biz.htm#1. Under the revenue sharing arrangement the ADC is being levied at the rate of 1.5 percent on the total earnings, excluding the revenue from rural operations. This means that the revenue from ILD traffic is not excluded for the reckoning of the revenue share. However, a 'per minute' ADC is still applicable for ILD traffic, creating a situation of 'double taxation'.

REFERENCES

Agarwal, Manish (2005, 20 May). Call off the ADC. The Hindu Business Line.

Amin, Aasha, Amin, Mehreen and Hussain, Ahmede (December 10, 2004). Interview: A Viking's Success Story. Star Weekend Magazine. Retrieved November 8, 2006, from http://www.thedailystar.net/magazine/2004/12/02/interview.htm

Christensen, C.M. and Raynor, M.E. (2003). The Innovator's Solution. Boston, MA: Harvard Business School Press.

Clemons, Eric K., Croson, David C. and Weber, Bruce W. (1996). Market Dominance as a Precursor of Market Failure: Emerging Technologies and the Competitive Advantage of New Entrants. Organization Systems and Technology, 4, p. 262.

Cohen, Nevin (2001). What Works: Grameen Telecom's Village Phones rural telephony; shared-access; franchise via local entrepreneurs. World Resources Institute. Retrieved November 8, 2006, from http://www.d2igitaldividend.org/pdf/grameen.pdf

de Silva, H. and Tuladhar, R.K. (2006). Smart Subsidies: Getting the Conditions Right. The experience of expanding rural telecoms in Nepal. WDR Discussion Paper WDR0606. Retrieved November 12, 2006, from http://www.regulateonline.org

Garcia-Murillo, Martha (2003). Patchwork Adoption of ICTs in Latin America. The Electronic Journal on Information Systems in Developing Countries. Retrieved November 8, 2006, from http://www.ejisdc.org

GSMA (2005). Emerging Market Handset Programme. Retrieved November 6, 2006, from http://www.gsmworld.com/emh/

Intelecon Research & Consultancy Ltd. (2002). Universal Access & Rural Connectivity. Retrieved November 12, 2006, from http://www.InteleconResearch.com

Intven, Hank (2000). Telecommunications Regulation Handbook. Washington DC: World Bank.

Nokia (2003). Mobile Entry Solution for New Growth Markets. Retrieved November 5, 2006, from http://www.nokia.com/link?cid=EDITORIAL_1607

Patton, Michael Q. (1990). Qualitative evaluation and research methods (2nd ed.). Newbury Park, CA: Sage Publications.

PMN Publications (2005). Low-cost handset initiative promotes agenda of global inclusion. Retrieved November 8, 2006, from http://www.pmn.co.uk

Pontes, Daniella (2003). Communications for Empowerment (CoE)—Brazil Case Study. Retrieved September 4, 2006, from http://www.daniella.rdvp.org

Samarajiva, Rohan (2002). Redefining universal service: Policy and Regulatory Actions. Retrieved November 8, 2006, from http://www.itu.int/TELECOM/ast2002/cfp/pap5195.doc

Spiwak, Lawrence J. (1999). Why Local Loop Unbundling is no telecom Panacea. Communications Week International, 26 April 1999.

The Boston Consulting Group, Inc. (2001). Boosting Asset Productivity: A Call to Action for Telcos. Retrieved November 3, 2006, from http://www.bcg.com

Ureta, Sebastian (2005). Variations on expenditure on communications in developing countries. Retrieved September 20, 2006, from http://www.regulateonline.org

Document(s) 16 of 21