Monitoring and Evaluation Working Paper 2 August 2000 21749 The GEF Solar PV Portfolio: Emerging Experience and Lessons ~~~~~- . Eric Martinot Ramesh Ramankutty Frank Rittner :fl|: ~~~~~~~~~GEF The GEF Solar PV Portfolio: Emerging Experience and Lessons Eric Martinot Ramesh Ramankutty Frank Rittner Monitoring and Evaluation Working Paper 2 August 2000 Authors Eric Martinot, GEF Secretariat, Climate Change Team Ramesh Ramankutty, GEF Secretariat, Monitoring and Evaluation Team Frank Rittner, GEF Secretariat, Climate Change Team Steering Committee Dennis Anderson, GEF Scientific and Technical Advisory Panel Tom Hamlin, United Nations Environment Programme Richard Hosier, United Nations Development Programme Chandra Shekhar Sinha. World Bank Published 2000 Global Environment Facility This working paper may be reproduced in whole or in part and in any form for educational or nonprofit uses, without special permission, provided acknowledgment of the source is made. The Global Environ- ment Secretariat would appreciate receiving a copy of any publication that uses this publication as a source. Copies may be sent to GEF Secretariat. G 6-148, 1818 H Street NW, Washington, DC 20433. No use of this publication may be made for resale or other commercial purpose without prior written consent of the Global Environment Facility Secretariat. The designations of geographical entities in this document, and the presentation of materials, do not imply the expression of any opinion whatsoever on the part of the GEF concerning the legal status of any country, territory, or area, or its authorities, or conceming the delimitation of its frontiers or boundaries. The views expressed in this paper are not necessarily those of the GEF or its associated agencies. ISBN 1-884122-88-4 ISSN 1020-0894 Preface The GEF Monitoring and Evaluation (M&E) team is tasked with analyzing and documenting GEF results. Until now, conclusions of these efforts have been in the form of evaluation and study reports, annual Project Performance Reports, and GEF Lessons Notes. With the introduction of the M&E series of Working Papers, we are publishing reports that are not full-fledged evaluations, but nevertheless deserve attention. Many of the issues and early results that these reports identify will be pursued later in broader evaluations to arrive at more definite conclusions. We expect the M&E working papers to be a valuable catalyst for promoting dialogue on issues and results of importance within GEF's operational areas and efforts. We therefore look forward to your feedback and suggestions. Please contact us through the coordinates listed below and visit the GEF Web site to find out more about the Monitoring and Evaluation program. The GEF Solar PV Portfolio: Enmerging Experience and Lessonis is the result of a 1999 thematic review. Thematic reviews are not comprehensive evaluations-when many projects in a portfolio are relatively new, such evaluations would be premature. Rather, such reviews are more modest attempts to take stock of progress to date and identify lead indicators of achievements, if any. Additionally, reviewers may identify issues related to project design and implementation, thereby enabling discussion and reexamination of strategic issues within the GEF operational programs. This review was based on data and information collected from a variety of sources: (a) desk reviews of project documents, including proposals for funding project preparation, (b) Project Implementation Reports and evaluations of completed projects; (c) interviews with project managers in the implementing agencies: and (d) visits to Bangladesh, China, Ghana, India. Sri Lanka, Uganda, and Vietnam. Jarle Harstad Senior Monitoring and Evaluation Coordinator GEF Corporate Monitoring and Evaluation Team 1818 H Street. NW Washington, DC 20433, USA Telephone: (202) 458-2548 Fax: (202) 522-3240 E-mail: geflessons@gefweb.org Web: http://www.gefweb.org iii Acknowledgements This thematic review would not have been possible without the support and valuable contribution of several individuals and organizations. The review benefited greatly from the advice and guidance provided by the Steering Committee. During the preparation of this review, several colleagues in the GEF implementing agencies shared with us their valuable experiences in project design and implementation, and facilitated country visits to review projects. They include: Wolfgang Bertelsmeier, Susan Bogach, Anil Cabraal, Aeneas Chapinga Chuma, Mac Cosgrove- Davies, Stephen Duah-Yentumi, Suresh Hurry, Magdalena Manzo, Subodh Mathur, Ademola Salau, and Doug Salloum. We would like to thank the Governments of Bangladesh, China, Ghana, India, Sri Lanka, Uganda, and Vietnam for facilitating country visits. We are particularly grateful to the following project managers who organized project reviews during country visits: * Dewan Alamgir, General Manager, Grameen Shakti, Bangladesh - Clement Abavana, Project Manager, Renewable Energy Services Project, Ghana V. Bakthavatsalam, Managing Director, Indian Renewable Energy Development Agency, India * Jayantha Nagendran, Project Manager, Energy Services Delivery Project, Sri Lanka - Steve Hirsh, Chief Technical Advisor, Ministry of Energy and Mineral Development/United Nations Development Programme/GEF, Uganda Shawn Luong, Managing Director, Solar Electric Light Company, Vietnam Additional support from the World Bank's Climate Change Thematic Group contributed to prior research that went into this report. Finally, we would like to acknowledge the contributions of our colleagues at the GEF Secretariat: Alan Miller and Scott Smith for providing contrasting viewpoints and inputs, and Elizabeth Mook for editing the report. Ramesh Ramankutty Task Manager GEF Thematic Review of Solar PV Portfolio il' Contents Executive Summary I Introduction 3 Solar Home Systems Delivery Models in the GEF Portfolio 7 Ten Emerging Lessons from the Review of Solar PV Projects 13 Conclusions and Recommendations 17 Annex 1: Project List and Status 21 Annex 2: Approaches to Solar Home Systems in Africa 23 Annex 3: Consumer Credit through Microfinance Organizations in Sri Lanka 25 Annex 4: Solar PV Businesses and the IFC/GEF Small and Medium Scale Enterprise Program 27 Annex 5: Rural Energy Service Concessions in Argentina 29 List of Tables, Boxes, and Figures Table 1. Portfolio by Implementing/Executing Agency and Region (number of projects) 4 Table 2. Portfolio by Implementing/Executing Agency and Region (financing) 5 Table 3: Delivery Models. Relevant Issues, and Country Applications 9 Box 1: A Fee-for-Service Approach in Kiribati 10 Box 2: Consumer Costs of Solar Home Systems: Credit Sales vs. Fee-for-Service 12 Figure 1: Typology of GEF-supported Solar PV Projects 8 1V vi Executive Summary Since 1991, the GEF has provided grant financing for credit. With a service model (10 projects), an energy- 23 off-grid solar photovoltaic (PV) projects in 20 service company provides electricity for a monthly countries. In addition, four more projects are under fee to rural households. In this case, the system is preparation in the pipeline. Though specific objec- owned, financed, and maintained by the energy-ser- tives vary, the projects, in general, are aimed at stimu- vice company. Since a large share of the portfolio is lating and achieving commercialization of solar PV still under implementation, it is too early to draw systems for rural households (called "solar home sys- definite conclusions regarding impacts. Nevertheless, tems"). In all, these 23 projects together account for the review of the GEF solar PV portfolio suggests ten about US$210 million of GEF allocation, and about emerging lessons: $1.4 billion in total project costs. 1. Viable business models must be demonstrated to The review highlighted a number of important issues sustain market development for solar PV. associated with solar home systems projects that fu- ture projects should explicitly address to varying de- 2. Delivery/business model development, evolution, grees, depending on the delivery model employed: and testing require time and flexibility. * Technological credibility 3. Institutional arrangements for project implemen- * The role of solar PV within rural electrification tation can greatly influence the value of the project programs in terms of demonstrating viable business models * Government policies vis-a-vis electricity as a ba- and thus achieving sustainability. sic need * Household affordability and willingness to pay 4. Projects must explicitly recognize and account for * Marketing strategies, costs, and purveyors the high transactions costs associated with market- * Concession selection and regulation ing, service, and credit collections in rural areas. * Commercial creditworthiness and access to busi- ness finance 5. Consumer credit can be effectively provided by * Sustainability of development finance. microfinance organizations with close ties to the local communities if such organizations already Projects in the portfolio employ either or both of two have a strong history and cultural niche in a spe- primary approaches: with a sales nmodel (eight cific country. projects), private dealers sell solar home systems to rural households. The system is owned and main- 6. Projects have not produced adequate experience tained by the household, which is also responsible for on the viability of dealer-supplied credit under a servicing any debt if the system is purchased with I sales model, and no project in the portfolio ap- credit approaches pears set to provide such experience. * Use of GEF resources for non-recurring costs re- lated to business and market development 7. Rural electrification policies and planning have a * Access to finance and incremental risk sharing major influence on project outcome and * Explicit linkages to rural electrification policies sustainability, and must be explicitly addressed in and planning project design and implementation. * Commercially feasible business models that are sustainable and can be replicated. 8. Establishing reasonable equipment standards and certification procedures for solar home system Projects must be careful to avoid an "equipment dem- components that ensure quality service while onstration" mentality where the main objective is maintaining affordability is not difficult, and few installation and maintenance of a certain number of technical problems have been encountered with systems. By project completion, the number of sys- systems. tems installed is much less significant than whether the business, delivery, and credit models are viable, 9. Substantial implementation experience is still sustainable and being replicated. This emphasis re- needed before the success of the service approach quires implementing agencies to rethink traditional can be judged. development assistance patterns and evaluation techniques. 10. Post-project sustainability of market gains achieved during projects has not yet been demon- We question whether purely private delivery models, strated in any GEF project: it is too early in the by themselves, are able to achieve the widespread evolution of the portfolio. market penetration in poorer countries that will sat- isfy both global environmental and development ob- The global environmental benefits from rural solar jectives. We recognize that experience from some PV projects are primarily indirect, and depend on the countries, such as Kenya, shows that the private- degree to which the GEF can help catalyze markets sector can achieve substantial market penetration for rural PV applications that serve large shares of the without much support from subsidies or government two billion people in developing countries currently or multilateral agencies. Still, we hypothesize that without electricity. Development benefits to house- projects that involve government measures will result holds and rural communities are clearly connected in greater penetration and larger shares of rural with these projects, although further surveys of households able to benefit from PV than purely pri- households are desirable to quantify income-genera- vate-sector models. Regardless of government in- tion effects, customer satisfaction, and delivery costs. volvement, service models seem more likely to result in larger markets because they provide greater We recommend that future projects in the GEF port- affordability for poorer households. More definitive folio focus on the following key issues: conclusions must await further experience from the portfolio in the coming years. * Affordability through fee-for-service or consumer 2 Introduction Role of Solar Photovoltaics fluorescent lights, a radio or cassette player, television, and perhaps small fans or other small appliances. Elec- Photovoltaics (PV) already provide electricity to an tricity is drawn from rechargeable batteries recharged estimated 500.000 to I million rural households in through an electronic controller by PV modules developing countries who lack access to electricity mounted on a pole beside the house or on the rooftop. grids. Worldwide, two billion people lack access to The total capacity of the unit is usually in the range of 30 electricity, so the potential for continued application Wp to 100 Wp but can be smaller or larger.3 of PV is large. with resultant economic, local envi- ronmental, and global environmental benefits. Tradi- Solar home systems can eliminate or reduce the need tionally, bilateral donor assistance has resulted in for candles, kerosene, liquid propane gas (LPG), and/ hardware installations but has placed less emphasis or battery charging. Direct economics benefits in- on the key ingredients for sustainability, such as vi- clude avoided costs of battery charging and LPG or able service networks and trained personnel. In the kerosene purchases; other significant benefits include past several years, several examples of sustainable increased convenience and safety, improved indoor commercialization of PV systems have emerged air quality, a higher quality of light than kerosene without much direct donor assistance, notably in lamps for reading, and reduced carbon dioxide emis- China, Indonesia, Kenya, and Zimbabwe. But the sions. Improved lighting quality can assist reading reality is still much smaller than the potential.2 and provide additional educational benefits, espe- cially to children, or allow income-generating activi- "Solar home systems" (SHS) are one of the most ties to occur at night. PV systems can also power common forms of PV application in rural areas. An lights and vaccine refrigerators in medical clinics, run SHS usually provides electricity for two or three water pumps, and assist other applications.4 I Gerald Foley. Photov oltaic Applications in Rur-al Areas of'the Delveloping World, World Bank Technical Paper No. 304 (Washington. D)C. 1995); GTZ. Basic Electrification fbrRural Houselholds: Experience with the Dissemination of Sniall-Scale Photovoltaic Systems ( Eschborn, Germany: Deutsche Gesellschaft fur Technische Zusammenarbeit. 1995) ; World Bank. Rural Energy and Development: Improving Eneigy Supplies for 2 Billion People (Washington. DC. 1996); Dan Kammen. "Promoting appropriate energy technologies in the developing world." Environment vol. 41. no. 5 (June 1999), pp. 11- 1 5. 34-41; unpublished World Bank Group documents. 2 Other common applications of PV include water pumping in agriculture. commercial applications like telecommunications, and village mini-grids serving small numbers of households (often integrated with diesel and wind systems). 3 See A. Cabraal. M. Cosgrove Davies. and L. Schaeffer, Best Prac tices for Plhotovoltaic Household Electrification Programs: Lessonsfirom Experiences in Selected Countries, World Bank Technical Paper No. 324 (Washington, DC, 1996); Foley 1995, op. cit. note 1; Christoper Flavin and Molly O'Meara. "Shining examples," World Watch, May/June (1997), pp. 28-36. 3 The GEF Solar PV Portfolio of the portfolio among the implementing agencies and the geographical portfolio allocation are shown Since 1991, the GEF has provided grant financing for in Tables 1-2.5 23 off-grid solar photovoltaic (PV) projects in 20 countries (Annex 1). In addition, four more projects The projects in the GEF portfolio target total direct are under preparation. Though specific objectives installations of 500,000 or more solar home systems, vary, the projects, in general, are aimed at stimulating comparable in magnitude to the current installed base and achieving commercialization of solar PV systems of such systems in developing countries.6 However, for rural households. Some of the projects supported most projects are just starting implementation and by the GEF fully fund solar PVs and/or related com- only a few are nearing completion (the Zimbabwe ponents, while others contain significant solar PV project was the first to be completed, in 1998). Direct components. In all, these 23 projects together account installations from the projects in the portfolio by the for about US$210 million of GEF allocation, and end of 1999 amounted to about 18,000 systems (see about $1.4 billion in total project costs. The division Annex ]).7 Table 1. Portfolio by Implementing/Executing Agency and Region (number of projectsl Region UNDP World Bank World Total Bank/IFC Africa 4 4 8 East Asia & Pacific 1 3 4 Latin America & Caribbean 2 2 4 South Asia 1 2 3 Global 4 4 Total 8 11 4 23 4 Refer to Annexes 1-4 for more detaits of the projects covered under the review. For further descriptions and evaluations of these projects, see the GEF web site (www.gefweb.org). See also Eric Martinot and Omar McDoom, Promoting Energy Efficiency and Renewable Energy: GEF Climate Clhange Projects and Impacts (Global Environment Facility. Washington, DC, 2000); Eric Martinot. Anil Cabraal, and Subodh Mathur, "World Bank/GEF solar home systems projects: Experiences and lessons learned 1993-2000" (Washington, DC: World Bank, 2000). 5 The total targeted installations does not include 200.000 systems originally expected from the Indonesia project, which was never implemented because of Indonesia's macroeconomic crisis and will now be cancelled. 6 Most of these installations are stand-alone solar home systems, except in India, where installations directly financed by the project have included 1500 solar lanterns, five village power schemes of 25 kWp each supplying electricity to about 500 families, and 200 SHS sold by a private dealer. 4 Table 2. Portfolio by Implementing/Executing Agency and Region (million US$) UNDP World Bank World Bank/lFC Total GEF Total GEF Total GEF Total GEF Total Region allocation project allocation project allocation project allocation project Af rica 14.7 17.4 9.1 86.4 23.8 103.8 East Asia & Pacific 8.8 27.7 60.4 528.1 69.2 555.8 Latin America and Caribbean 8.5 17.7 22.2 251.6 30.7 269.3 South Asia 1.5 1.5 31.9 241.3 33.4 242.8 Global 56.5 222.5 56.5 222.5 Total 33.5 64.3 123.6 1107.4 56.5 222.5 213.6 1394.2 The Thematic Review * Identify whether (and how) projects that have com- pleted implementation have sustained project During the 1998 GEF Project Implementation Re- achievements or promoted replication and/or ex- view, several questions were raised regarding the les- pansion sons being derived from this portfolio. It was * Identify the roles played by different stakehold- recommended that a thematic review of the portfolio ers in different parts of the project cycle and their be conducted with the overall objective of document- impact on project performance ing the experience with the design, implementation, * Estimate the potential impacts of these projects on and impacts (for completed projects). The review is global efforts to reduce greenhouse gas emissions not a comprehensive evaluation of the portfolio- * Make specific recommendations for GEF project such an exercise would be premature given the cur- design, programming, and performance indicators rent status of most projects in the portfolio. Rather, in solar PV projects. the review is a more modest attempt to take stock of the current progress, identify lead indicators of The review was based on data and information col- achievements, if any. and identify issues arising from lected from a variety of sources: (i) desk reviews of project design and implementation to foster a discus- project documents, including proposals for funding sion and reexamination of strategic issues within the project preparation; (ii) Project Implementation Re- GEF climate change program. The specific objectives ports and evaluations of completed projects; (iii) in- of the review are to: terviews with project managers in the implementing agencies; and (iv) visits to Bangladesh, China, * Identify elements and characteristics of project Ghana, India, Sri Lanka, Uganda, and Vietnam. The design, implementation, policy, and institutional data and views obtained from country visits are de- environments that have been responsible for scribed in Annexes 1-4.9 project performance 7 In addition to the sources of information and country visits documented in Annexes 1-4 for Bangladesh, Ghana, Sri Lanka, Uganda. Vietnam, and Zimbabwe. the team visited China (met with UNDP. World Bank Project Implementation Unit, State Economic and Trade Commission) and India (met with India Renewable Energy Development Agency). Substantial information was already available on the Argentina project from parallel work done by the World Bank (source noted in Annex 5). 5 6 Solar Home Systems Delivery Models in the GEF Portfolio GEF Operational Programs * High transaction costs Lack of consumer financing The solar PV projects in the GEF portfolio have * High first-cost and affordability been funded under GEF Climate Change Opera- . Unwillingness of utilities to provide off-grid elec- tional Program 6, Promoting the Adoption7 of Re- tricity services newable Enierg-y Byl Removing Barr-ier-s and . Lack of experience regulating rural energy-service Reducing Implementation Costs, which has two concessions. major objectives:s Models for Delivery of Solar Home Systems 1. Remove the barriers to the use of commercial or near-commercial renewable energy technologies. Projects in the portfolio employ either or both of two 2. Reduce any additional implementation costs for primary approaches.9 renewable energy technologies that result from a lack of practical experience, initial low volume Sales models (8 projects). With a sales model, private markets, or from the dispersed nature of applica- dealers sell solar home systems to rural households. tions, such that economically profitable "win-win" The system is owned and maintained by the house- transactions and activities increase the deployment hold, which either pays cash in full or obtains con- of renewable energy technologies. sumer credit and is responsible for servicing the debt. The sales model is employed in projects in Bangladesh, The output of a GEF-supported project in this opera- China, India, Indonesia, Sri Lanka, Uganda, Vietnam, tional program is expected to be the removal of a and Zimbabwe. Under the sales model, consumer barrier to a particular renewable energy application. credit may be provided by the dealer (Bangladesh, In- Some of the key barriers usually addressed through dia. Indonesia, Sri Lanka), by a micro-finance organi- off-grid solar PV projects in the portfolio are: zation (Sri Lanka and Uganda), or by a development finance institution (Uganda, Vietnam, Zimbabwe). * Lack of established market * Lack of proven business models Service models (10 projects). With a service model, * Lack of business financing and business skills an energy-service company provides electricity for a 8 See GEF. Operational Programs (Washington, DC, 1996); Martinot and McDoom, op. cit. note. 4. 9 Not all projects explicitly specify in their design the delivery model to be employed. We treat sub-projects of the SME project as separate "projects" in our accounting here. Since there are three SME solar PV subprojects, we account for 25 "projects" rather than 23. Sri Lanka overlaps both sales and service categories, and thus 8 projects are left as unspecified as to delivery model. 7 monthly fee to rural households. The system is owned Solar Development Group (a global project), do not and maintained by the energy-service company. The explicitly define the delivery models; eligible busi- service model is employed in projects in Argentina, nesses financed through these projects may employ Benin, Cape Verde, Dominican Republic, Guinea, models of any type. Ghana, Lao PDR, Peru, Sri Lanka, and Togo. In al- most all of these projects, the energy-service company Of the 12 solar PV projects most recently approved is regulated by government and awarded monopoly by the GEF (since 1996), nine employ service models status for specific geographic regions. The main alter- and three employ sales models. In addition, at least native to regulated energy-service concession is an two of the four existing Project Development Funds open-market approach without regulation, which oc- (Block B) related to rural energy will lead to projects curs in Dominican Republic and is supposed to occur with service models. Thus the recent trend among in Peru after the project is completed. (See also Box I implementing agencies is to pursue service models."' for a non-GEF project using a service model.) Figure I and Table 3 depict the different types of Two projects, the Photovoltaic Market Transforma- sales and service models employed in the solar PV tion Initiative (in India, Kenya, and Morocco) and the portfolio. Figure 1. Typology of GEF-supported Solar PV Projects Solar PV Project Sales Models Service Models Cash Sales Service by Regulated Concession (Model A) Credit Sales _ ~~~Concession to Existing IJtility _ ~~~~~~~~~~~~~~~~~~~~~~~(Model E) Credit by Supplier Concession to Private Firm Competitively Selected (Model B) (Model F) Credit by Development Finance Institution Service by Unregulated Open Market Provider (Model C) (Model G) Credit by Microtinance Organization Service by Community-based Provider (Model D) (Model H) 10 Sales models are employed in China. Malawi. and Sri Lanka. The Sri Lanka project also employed a service model during initial phases. The Malawi project envisions considering service approaches later in the project but excludes them in the beginning. The Bolivia project does not specify in advance the models to be developed. 8 Table 3: Delivery Models, Relevant Issues, and Country Applications l ~~~~~~~~~~~~~~~Country Delivery Models Most Relevant Issues as Indicated by Emerging Lessons Applications SALES MODELS A. Cash * Demographics and household income limit demand. China * Sales to niche markets do not render comprehensive rural service. India * Marketing can be costly and labor-intensive. Uganda Vietnam Zimbabwe B. Consumer * Credit delivery and collection can be costly and risky. Bangladesh credit by supplier * Additional business finance required for extending consumer credit; India supplier must be creditworthy. Indonesia * Marketing can be costly and labor-intensive. Malawi * NGO suppliers may lack appropriate business skills for credit delivery. Sri Lanka C. Consumer * Development finance institution must be willing and able to continue Malawi credit by lending post-project. Uganda development * Continued subsidies for concessional finance may be needed. Vietnam finance institution * Development finance institution can assist with marketing. Zimbabwe D. Consumer * Microfinance organization must be creditworthy. Sri Lanka credit by * Credible microfinance organizations must exist. Uganda microfinance * Existing microfinance organizations may not be commercially oriented organization for rapid delivery of credit. * Microfinance organization can assist with marketing. SERVICE MODELS E. Regulated * Utility must have interest and experience in rural areas. Argentina concession by * How should government regulate concession and set tariffs? Ghana (asdesiged) existing utility * How should government specify/ensure quality of service in Lao PDR concession contracts? F. Regulated * Concession must be creditworthy and commercially viable. Argentina concession by * How should government regulate concession and set tariffs? Benin private firm * How should government specify/ensure quality of service in Cape Verde competitively concession contracts? Guinea selected * What should be the terms and conditions for renegotiation of Peru concession contracts? Togo * How should government conduct competitive selection and attract qualified bidders? G. Unregulated, * Marketing can be costly and labor-intensive. Benin open market * Long-term business finance required for initial capital investments. Dominican Rep. * Long-term recurring service costs may threaten continued profitability India or limit financial ability to expand. Peru (post-project) Togo H. Community * Supplier should have strong ties to the community, which may India overcome other barriers. * Supplier may need development of technical skills. * May be community-driven if government lacks ability or interest to develop concessions for community. 9 Box 1: A Fee-for-Service Approach in Kiribati" In the outer islands of Kiribati, 500 solar home systems have been installed by a government-owned energy- service company. The initial capital for the 100 Wp PV solar home systems was donated by the European Union (EU) under its Pacific Regional Energy Programme (PREP). This program also provided training and technical assistance to the company in installation, maintenance, competitive procurement, and other busi- ness practices. The company charges households an equivalent of roughly $6-7 per month, which covers direct maintenance costs and replacement capital costs for the batteries, controllers, and lights over a period of 20 years.12 The company maintains an interest-bearing account in which part of the fees are kept to provide financing for future equipment replacements. The monthly fees do not yet cover the full central overhead costs of the company, but they do include a component to finance PV panel replacement after 20 years. The company was supplementing its operating costs with income from other PV projects, such as installation of PV pumping systems and sales of controllers it was manufacturing itself. An evaluation by the EU estimated that an installed base of 1,000 systems minimum would allow the company to cover these overhead costs as well. The program provides a sustainable mechanism for continued maintenance (and perpetual replacement) of donor-supplied systems and provides customers with affordable energy services. However, the scheme is not commercially replicable because it relies on donor-supplied capital equipment to get started. If the company were to finance the initial capital costs on its own, the monthly fees needed to amortize these costs would be higher (perhaps double by one estimate), and some additional financing mechanism would be needed. The European Union is planning to extend this program and supply additional PV systems to extend the installed base to the 1,000 systems necessary for sustainable operation of the energy-service company. Issues Associated with the Delivery Models The role of solar PV within rural electrification programs (all models). Without the government de- The review suggested a number of important issues marcating an identified niche for solar PVs in its associated with solar home systems in general and policies, rural consumers are enticed by the promise, with specific delivery models in particular (refer to often false, of electricity provision through the grid, Table 3). These include: adversely affecting demand for solar PV through pri- vate suppliers. This has been observed in several Technological credibility (all models). It is impor- countries, including Sri Lanka and Vietnam. Grid tant to assure the quality both of hardware and service expansion is often not a cost-effective option for pro- to establish long-term credibility of the technology. vision of electricity to the most remote of rural areas. The quality of panels and PV-related hardware has See also lesson #7. been more or less well-established, though the reli- ability of locally produced hardware, such as charge Government policies vis-a-vis electricity as a basic controllers, may have to be established. Even where need (models E, F, and G). If government policy reliability has been established, it is strongly influ- gives rural electrification priority even in areas where enced by maintenance, hence the need to focus on grid extension is impractical, then regulated conces- reliable service as a goal to establish long-term cred- sions (models E and F) are more likely to be fa- ibility. For sales models (models A-D), customer vored-as has happened in Argentina. Subsidies may education on proper maintenance is especially impor- be provided to concessions with the clear understand- tant. See also lesson #8. 11 Box I based on Bill Gillet and Gill Wilkins, "Solar so good-An EC funded solar utility succeeds in Kiribati." TheACP- EC Courier No. 177 (Oct-Nov. 1999), pp.5-7; personal communication with Suresh Hurry, UNDP (New York), January 2000; personal communication with Gill Wilkins, AEA Technology-ETSU (Oxfordshire, UK), April 2000. 12 The program expects PV panels to last 20 years. Lifetimes are estimated at 5 years for batteries and lighting ballasts and 8 years for controllers and DC/DC converters. I0 ing that electricity is part of rural development policy. Concession selection and regulation (models E If national policy is lacking, local communities may and F). For concession models, numerous issues decide to promote community-based energy service must be resolved: setting tariffs, finding and attract- delivery as a way of obtaining service for an entire ing capable bidders. and conducting competitive bid- community (model G). See also lesson #7. ding procedures, plus ensuring service quality and regulating concessions on an ongoing basis. If an Household affordability and willingness to pay (all existing utility is being considered for rural energy models). The choice of delivery model is influenced service delivery, then the interest and experience of by the affordability and willingness of households to that utility in operating in rural areas becomes a fac- pay. Affordability is most at issue with cash sales tor. These issues are clearly visible in Argentina. (model A). as only a small fraction of the rural popu- lation may be able to afford to purchase a system Commercial creditworthiness and access to busi- outright (depending on rural demographics). Other ness finance (all models). Business finance is an sales models with credit (models B, C, and D) may important issue for all delivery models except per- improve but still limit affordability. depending on the haps model E, where an existing utility would be term of credit available. In absence of government presumed to have access to credit. Supplier-provided interest in and/or regulation of concessions, an "open- credit (model B), as well as private concessions market" energy-service company (model G) may be (model F) and open-market energy service companies an attractive alternative. Under this model, a firm (model G) in particular require large amounts of busi- provides fee-for-service to specific territories but ness finance because they must finance installed sys- without monopoly status granted by a regulator (as tems over extended periods. Creditworthiness also has been seen in the Dominican Republic). Refer to affects the terms of consumer credit that can be pro- Box 2 for typical cost comparisions facing a con- vided. and thus the market demand (this is clearly sumer between the sales and service models. seen in Bangladesh; see also lesson #6). If a microfinance organization is providing consumer Marketing strategies, costs, and purveyors (mod- credit (model D), then it must itself be creditworthy els A, B, C, D, and G). Private firms may need to enough to obtain credit from the banking or financial expend substantial resources on marketing their prod- system, a major issue in Sri Lanka. ucts, especially if rural households are highly dis- persed or difficult to reach (Bangladesh is one such Sustainability of development finance (model C). case). Mass media methods may be ineffective; door- If development finance organizations provide con- to-door campaigns may prove most effective. For sumer credit, then their willingness or ability to con- sales models with credit, the purveyor of credit may tinue after the project completes (including the need provide marketing services in addition to or instead of for any ongoing subsidies) is an important issue-as the dealer (this has happened in Sri Lanka and Viet- seen in Zimbabwe and potentially in Vietnam. See nam). Under regulated concession models, marketing also lesson #5. may still be important but not as essential to the survival of the business. See also lesson #4. The next section amplifies these issues in the context of emerging lessons from the thematic review. II Box 2: Consumer Costs of Solar Home Systems: Credit Sales versus Fee-for-Service How much will a consumer pay for a solar home system over its lifetime if purchased on credit compared to the same system provided with a fee-for-service arrangement? There is little available cost data with which to make such an analysis, but the following numbers are illustrative. Based upon data collected during this review and published sources, a typical 50 Wp solar home system can be expected to cost between $450 to $800 fully installed, depending upon several factors: * level of import duties on imported components like PV panels * level of profit margins incorporated by system integrator and/or system installer firms * degree to which components are produced locally and are cheaper than imported components * quality standards applied to components and installation * complexity/features of battery controller. Assuming an installed cost of $600, a lifetime of 10 years, a 3-year loan at 20 percent real interest with 25 percent down-payment, an average annual cost of battery replacement of $20 ($50 battery with 2.5 year life), and estimated O&M costs of $20/year, the lifecycle costs of a purchased system are about $1200. With fee-for- service, a $17/month charge for the system over a 10-year period totals $2040, or almost double the lifecycle costs of a purchased system.13 But fee-for-service has several advantages: * the energy-service company assumes the performance risk * an initial down payment may not be required (or is smaller than for a credit sale) * service is guaranteed as long as the firm is providing fee-for-service * consumers can return the system at any time, minimizing the risk of electric-grid extensions. Given a choice, some consumers may prefer these benefits even though total lifecycle costs may be higher. 13 Monthly charges vary in the projects examined. In the Argentina Renewable Energy in Rural Markets Project, an estimated monthly recovery cost to the concession is given as $17 for a 50-Wp system (15-year SHS life. 14% return on investment, various other assumptions); this figure is only calculated ("Argentina Renewable Energy in the Rural Market: Project Appraisal Document." World Bank, 1999). In the Dominican Republic, Soluz Dominicana has been charging customers $20/month for a 50-Wp system (see Annex 4). 12 Ten Emerging Lessons from the Review of Solar PV Projects The review of the GEF solar PV portfolio suggests dies are required to sustain it, then the extent of these ten emerging lessons. Since a large share of the port- subsidies must be clearly documented within an in- folio is still under implementation, it is too early to come-expense framework, and arrangements for their draw definite conclusions regarding impacts. How- continuation must be secured. ever, from trends identified during project implemen- tation, and early indication of some impacts on the Virtually all GEF projects have explicit or implicit ground, some initial lessons can be drawn. goals directed towards developing viable business models (see Annexes 1-4). But these pilot models Lesson 1. Viable buisiniess models must be demon- have yet to be tested. Delivery is clearly linked with strated to sustain niar-ket developnmentfor solar PV. profitability and business model development in projects in Bangladesh, China, Dominican Republic, Demonstration of a viable business model, whether Sri Lanka, and Zimbabwe. For example, the firm that business is public, private, utility, or even perma- Soluz Dominicana in the Dominican Republic is nently subsidized, is key to achieving project demonstrating a "proof of concept" for its fee-for- sustainability and achieving GEF's programmatic ob- service business model for up to 5,000 systems and is jective of transforming (or developing) markets for seeking to adapt the model to a larger scale-perhaps solar PV. Viability means that expenses and receipts, 25,000 systems (see Annex 4). cash flow, profits (or subsidies), management and service arrangements demonstrate an entity can con- Lesson 2. Delivervlbusiness model development, tinue to exist and function on commercial terms. evolution, aid testing Irequire time and flexibility. "There is a high value-added [by the GEF] in terms of developing and improving business models.. .you Building markets and identifying viable sustainable want to stimulate markets based on these business delivery models in specific contexts are slow and models" said one dealer being supported by the GEF. time-consuming processes requiring a much greater degree of flexibility and adaptation than currently For profit-oriented commercial businesses, profit is allowed by most GEF project designs. GEF projects the ultimate measure of whether a business model is do not allow sufficient time or resources to first iden- viable, which in turn depends on maximizing income tify the most promising approaches and, then, de- (which depends on demand, pricing, and competi- velop those approaches to a point where their tion) and minimizing expenses (for marketing, ser- viability and sustainability are clear and tested. vice, training, procurement, and operations). If a Projects should explicitly allow for testing multiple business receives public support because of economic models and for adapting and modifying models over development or other public objectives associated time until viable approaches become clear and are with solar home systems, such that continuing subsi- tested sufficiently. Insufficient project durations, de- 13 lays in project start-up due to unanticipated circum- ticed by GEF or implementing agency staff; but the stances, and fixed project completion dates can overriding importance of sustainability and replica- hinder this process. tion in a GEF project demand a much greater aware- ness of how implementation arrangements affect the This lesson is clearly illustrated in Sri Lanka, where demonstration value of a project. dealer credit and fee-for-service approaches were tried early in the project without much success and Lesson 4. Projects must explicitly recognize anid the project began to emphasize consumer credit account for the high transactions costs associated through microfinance organizations (see Annex 3). with ma-keting, service, and credit collectionis in ru- This later approach appeared to the dealers and the ral areas. project to show the most promise, and the rate of system installations under this approach was acceler- Long distances, poor transport infrastructure, impass- ating through one main microfinance organization. able roads during monsoons, low literacy rates, cash- But more time would be needed to replicate and and-barter based transactions, and lack of technical amplify the microfinance model to provide greater skills all mean that transaction costs of operating a volumes of installations and include greater numbers rural PV business, whether sales or service-for-fee, of microfinance organizations. Unfortunately, the can be quite high. The costs and staff time needed for project is scheduled to close before the microfinance marketing, credit or fee collections, service, estab- model can be adequately tested. In particular, there is lishing business infrastructure, and training staff can a question as to whether commercial financiers will easily eat away already-slim profit margins. This les- consider microfinance organizations beyond the main son is illustrated in Sri Lanka, where dealers decided one participating in the project good enough credit not to offer consumer credit, citing the high costs of risks to extend commercial credit to them. credit collections in remote rural areas (see Annex 3), and in Bangladesh, where a dealer was investing Lesson 3. Inistituitionial ari-angemenits for project heavily in marketing out of its own operating budget implenmentation can greatly influence the value of the (without government or grant assistance), delaying its project in term-ls of demonistratinig viable blusiness ability to begin to make a profit (see Annex 4). models andcl thus achievinig sustainahilitv. No one has figured out how to conduct a standard- This lesson can be illustrated by the Ghana project, ized-product/high-volume approach to low transac- which was originally designed to demonstrate a busi- tion costs. Some form of subsidies (e.g., multilateral ness model in which the national utility would pro- or host-country government) may be a permanent and vide fee-for-service to rural households using solar essential feature of SHS delivery for this reason home systems (see Annex 2). At the conclusion of the alone. Dealers and energy-service companies need project, the costs, service, cash flow, and manage- experience, training, and developed business infra- ment of these installations could be assessed in terms structure in rural areas to be able to operate effec- of the viability of this model from the utility's per- tively with low transaction costs. spective. The demonstration of this business model could also be used to convince other private compa- Lesson 5. Consumer credit can he effectively pro- nies to enter the market, which is an explicit project vided by microfinance organizations with close ties to objective. But project implementation responsibility the local communities if such organizations already was transferred to a "project" established under the have a strong histo-y an?d cultural niche in a specific Ministry of Mines and Energy early in the project country. implementation. Although this office may succeed in installing and servicing a given number of systems, Under the sales model, a few projects are successfully given that it is subject to the rules and regulations of a providing consumer credit through microfinance or- government ministry, demonstrating and judging ganizations (Sri Lanka) and development-finance or- business viability in a transparent commercial man- ganizations (Zimbabwe and Vietnam). However, the ner is bound to be difficult, meaning that sustainabilityoftheseconsumercreditmechanismsis sustainability is seriously called into question. questionable in two of the three. The Agricultural Changes in institutional arrangements can go unno- Finance Corporation in Zimbabwe has not been able '4 to replenish their credit revolving fund, which will Annex 4). Of course, all else being equal, households wind down otherwise (see Annex 2). In Vietnam, would prefer to be connected to a grid than obtain consumer credit by the Vietnam Bank for Agriculture energy services from a solar PV system. Still, in most and Rural Development is partly dependent on countries. 100percentgridextensionistoocostly and dealer-provided credit guarantees (see Annex 4). In unrealistic. Policy development, in conjunction with Sri Lanka, consumer credit by microfinance organi- solar home system delivery models, is thus crucial so zations appears sustainable, but perhaps because Sri that areas of planned rural electrification are clear and Lanka has a strong and long-standing microfinance realistic, and rural electrification planning explicitly industry (see Annex 3). accounts for the potential of solar home systems in providing a least-cost path to rural electrification in Lesson 6. Projects have oot produced adequate some areas (and ultimately, the utility itself may ex- experience otn the viability of dealer-supplied plicitly plan to install solar home systems in certain credit under a sales model, and no project in the areas as an alternative to grid extension). portfolio appears set to provide such experience. Lesson 8. Establishing reasonable equipnment The Indonesia Solar Home Systems project was one of standards anid certification procedures for solar the first projects to utilize dealer-supplied credit as a honme svstenm components that ensure quality ser- delivery mode. The experience from this project would vice while maintaining affordabilitv is 7not difficult, have been extremely valuable. Unfortunately, this andfew technical problems have been encountered project never really got started because of the macroeco- with systems. nomic difficulties in Indonesia in the late 1990s, and now will be cancelled."4 The Solar PV project in Suppliers in all projects have generally been able to Bangladesh is the only one that shows dealer-supplied comply with standards and certification procedures credit to be working (see Annex 4). The dealer receives established under projects. The technologies have three-year credit from the IFC. Once this credit is com- worked with few problems. Where difficulties with pleted, the dealer may depend on continued develop- standards compliance have occurred in projects, ment institution assistance unless commercial business projects were able to slightly relax standards without financing for the dealer becomes available for longer sacrificing quality of installations or performance of terms thus increasing profitablilty. Longer term com- systems. Few customer complaints or technical prob- mercial financing depends on the dealer's ability to lems have been encountered in those projects where overcome high overhead and marketing costs. substantial systems have been installed (Zimbabwe, Sri Lanka, Dominican Republic, and Bangladesh). Lesson 7. Rr-al electrification policies and planninig Certification has been often slower than anticipated; have a major influence on7 project outcome and for example, in Sri Lanka it took almost the whole sustainiabilitv, and must be explicitlh addressed in first year of the project before suppliers had certified project design and inmplementationi. products on hand to deliver to customers. Participants in some projects cited unrealistic politi- Lesson 9. Substanitial implementation experience cal promises or planning about rural grid extension as is still needed before the success of the service ap- a serious barrier to solar-home-system market expan- proach can be judged. sion, one that was not anticipated adequately in project design. "Our main competition is the false The best experience with the service model has taken promise of the grid and kerosene and battery charg- place in the Dominican Republic, where 3,500 sys- ing, not other companies" said one supplier in Sri tems were installed from 1996-2000, about 1,700 of Lanka when asked about competition (see Annex 3). these on a fee-for-service basis (see Annex 4). In A private dealer in Vietnam was also encountering 2000, the great majority of new installations were problems as rural electrification encroached upon po- being installed on a fee-for-service basis and the firm, tential customers and interfered with marketing (see Soluz Dominicana, had passed the break-even point 14 A project completion report for the Indonesia project is expected in 2000. 15 of profitability."5 In Sri Lanka, one dealer attempted a Most projects are in early implementation and those service approach in 1998-99 but quickly (perhaps few completed or nearly completed are not yet dem- prematurely) gave up on this approach and switched onstrating market sustainability. Even in Zimbabwe, to a sales approach, citing the high administration and where the private dealer market was greatly expanded transaction costs and other difficulties of monthly fee and 10,000 systems were sold under the project, the collections in rural areas (see Annex 3). Early experi- question of continued consumer credit mentioned ence in Argentina also suggests that selecting and above-and the sustainability of many of the busi- effectively regulating energy-service concessions in nesses created during the project-is still in question rural areas can be a formidable regulatory challenge (see Annex 2). Sri Lanka appears to be closest to that requires significant assistance and capacity demonstrating sustainability, based upon consumer building for regulatory and institutional development credit through microfinance organizations and the (see Annex 5). The Argentina project should provide entry of Shell International Renewables into the Sri a wealth of new experience with the service ap- Lanka market (see Annex 3). Shell attributed their proach, but was just getting under way in 2000. entry to the World Bank/GEF project there, and other dealers see Shell's entry as helpful to market maturity Lesson 10. Post-project stistainiabilitv of mar-ket and sustainability after the project.'6 (gains achieved dur-inig projects has not yet been den7- onistrated in any GEF project; it is too earlY in the evolution of the portfolio. 15 IFC/GEF assistance to Soluz has taken the form of a $75.000 convertible loan with a 12% interest rate and a six-year tern. through the IFC Small and Medium Scale Enterprise program. Counting all sources, Soluz has raised over $1.5 million to invest in Soluz Dominicana's operations. 16 For a fuller treatment of methodologies for monitoring and evaluating post-project market gains, see Eric Martinot, Monitoring and Evaluation of Market Development in World Bank-GEF Climate Change Pro jects: Framework and Guidelines, World Bank Environment Department Paper No. 66 (Washington. DC, 1998). 16 Conclusions and Recommendations The global environmental benefits from rural solar Ghana, Sri Lanka, Uganda, and Vietnam showed PV projects are primarily indirect. That is, the direct early purchasers of systems to be among the wealthi- global environmental benefits from solar home sys- est households in rural areas. Some historical surveys tems projects, in terms of avoided carbon dioxide have shown substantial income-generation benefits from displaced kerosene, candles and batteries, are are possible, but further survey work will be required small relative to other sources of carbon dioxide to assess the economic benefits achieved in GEF emissions in these countries.'7 The emergence of sig- projects.'" Such surveys should also include other nificant global environmental benefits is primarily unknowns that influence market sustainability, such dependent on the degree to which a market for PV as customer satisfaction, system performance, dealer emerges, serving large shares of the two billion rural marketing costs, and system prices. Surveys con- population currently without electricity. This is ex- ducted in Kenya, for example, by the UNDP/World pected to have two results: (i) avoidance of fossil-fuel Bank Energy Sector Management Assistance Pro- use for providing electricity in rural services; and (ii) gram, have shown high levels of customer satisfac- lowering of global PV cost and spurring of applica- tion and technical performance.2' tions in both developed and developing countries that would otherwise be delayed. How well suited are GEF-supported projects for achiev- ing both global environment and development goals? Clearly there are immediate welfare-enhancing ben- The evidence from emerging project experience sug- efits from rural solar PV systems or services. In addi- gests that some GEF projects will demonstrate delivery tion to the direct benefits to households discussed in models with significant replication potential, but it is too the introduction, local economic benefits from em- early to be more definitive. Based upon this review, we ployment by PV dealers and service firms are signifi- recommend that future projects in the GEF portfolio cant. The extent of income-generation benefits for focus on five key issues: rural households is more uncertain.'8 Field visits in 17 Steven Kaufman, "Rural electrification with solar energy as a climate protection strategy." Renewable Energy Policy Project Research Report No. 9 (Washington, DC, 2000). Due to the inefficiency of kerosene lighting, avoided CO, emissions per installed Wp of PV in rural households are greater than for grid-connected applications, in some cases by a factor of ten, says Kaufman. This presumes that PV is used to displace kerosene rather than provide additional energy services; the issue of fuel displacement vs. added services has been inadequately studied. 18 We have not looked at rural development literature to examine income-generation and income-distribution effects. 19 Kaufman, op. cit. note 17. 20 Robert J. van der Plas and Mark Hankins, "Solar electricity in Africa: A reality." Energy Policy vol. 26, no. 4(1998), pp. 295-300. 17 I. Affordabilitv. Affordability through fee-for-service tainable and whether replication mechanisms are or consumer credit approaches will continue to effective. This emphasis requires implementing be a central issue. Projects should experiment with agencies to rethink traditional development assis- different approaches to affordability and show a tance patterns and evaluation techniques. reasonable chance of making solar PV available to more than just the wealthiest households. The We conclude the review by questioning whether potential for widespread affordability exists; some purely private delivery models, by themselves, are private sector studies claim 25-50 percent of rural able to achieve the widespread market penetration in households currently without electricity would be poorer countries that will satisfy both global environ- able to afford systems under fee-for-service or con- mental and development objectives. In many coun- sumer credit arrangements.2" tries, rural populations are simply too poor to afford solar home systems on their own. This can be seen in 2. Use of GEF resources for non-recuriring c-osts. particular in the African GEF projects. Short-term GEF resources should pay for incremental, non- profits may exist for a few dealers serving the recurring business and market development costs. wealthiest households, but market penetration may be rather than partial equipment subsidies. Such non- limited. However, if sales to the wealthiest house- recurring costs include, for example, business holds provides a future "pathway" to more wide- planning, feasibility studies, consumer awareness, spread sales because of significant cost reductions credit delivery pilot schemes, and initial market- (there are analogies in diffusion of technologies like ing and market development efforts. mobile phones), then private delivery models may lead to further market gains. 3. Access to cr edit and incremental risk sharinig. Ini- tial market development efforts can be assisted by One cannot dispute the private-sector-led experience providing financial services such as partial risk in Kenya, where an estimated 80,000 households had guarantees. performance incentives, and other solar PV systems in 1999. This experience shows that forms of contingent finance to local PV businesses. the private sector can achieve substantial market pen- etration without much support from subsidies, gov- 4. Explicit linkages to rural electrification policies ernments, or multilateral agencies (although training ald plannling. Policy development and rural util- and performance standards are still important compo- ity planning should be explicitly linked to solar nents of market facilitation there). In Kenya, most PV delivery models, so that areas of planned ru- households have purchased systems for cash, and a ral electrification are clear and realistic and pro- thriving market has emerged, now growing at 10-18 vide greater certainty for off-grid markets. Rural percent annually. A modular system of buying has electrification planning should account for the po- emerged where households can invest small sums in tential of solar home systems and other rural en- modest systems and upgrade as income allows. After ergy options, and consider incentive mechanisms, 10 years, the commercial market has reached about like energy-service concessions, to provide a least- one percent of rural households.22 cost path to rural electrification. Still, we hypothesize that projects involving govern- 5. Commercially feasible buzsiness models. Projects ment measures will result in greater penetration and must be careful to avoid an "equipment demon- larger shares of rural households able to benefit from stration" mentality where the main objective is in- PV than purely private sector models. Such measures stallation and maintenance of a certain number of may include government support for local industry, systems. By project completion, the number of policy approaches like regulated concessions, favor- systems installed is less significant than whether able rural development or power sector reform poli- the business, delivery, and credit models are sus- cies, and even continuing government subsidies for 21 Kaufman, op. cit. note 17. 22 Van der Plas and Hankins 1998, op. cit. note 20; Dan Kammen. op. cit. note 1. For other experiences with solar PV in rural areas, also see GTZ, op. cit. note 1; Geerling Loois and Bernard van fiemert, eds., Stanid-Alone Photovoltaic Applications: Lessons Learned (London: James and James. 1999). 18 the rural poor as part of poverty reduction objectives poorer households, particularly if available credit (perhaps equivalent to those for grid-connected cus- terms under sales models are short. Evidence for tomers). Regardless of government involvement, ser- these conclusions from GEF experience is still lack- vice models seem more likely to result in larger ing and may not exist for several more years as ser- markets because they provide greater affordability for vice models in the portfolio are tested. 19 20 Annex 1: List of Projects and Status (as of 12/99) Total Approximate GEF project number of Responsible contribution cost installations by Project (year approved by GEF) agency ($mn) ($mn) Status end of 1999 India: Alternate energy/renewable WB 26 186 under 2,200 resources development (1991) implementation Zimbabwe: PV for household and UNDP 7 7 completed 10,000 community use (1991) Small and Medium Scale Enterprise IFC 1.6 4.8 under 5,100 Program (1994) implementation Indonesia: Solar home systems (1995) WB 24 118 to be cancelled - Uganda: PV pilot project for rural UNDP 1.8 3.6 under electrification (1995) implementation Ghana: Renewable energy-based UNDP 2.5 3.1 under electricity for rural, social and economic implementation development (1996) PV Market Transformation Initiative (1996) IFC 30 120 under implementation Renewable Energy and Energy Efficiency IFC 30 130 soon operational Fund (1996) Sri Lanka: Renewable energy capacity UNDP 1.5 1.5 under - building (1996) implementation Sri Lanka: Energy services delivery (1996) WB 5.9 55 under 1,000 implementation Argentina: Renewable energy in rural WB 10 120 under - markets (1997) implementation Bolivia: Rural electrification with UNDP 4.5 8.5 under renewable energy (1997) \implementation China: Capacity building for renewable UNDP 8.8 28 under energy commercialization (1997) implementation Lao PDR: S. provinces renewable WB 0.7 2.1 under - energy pilot (1997) implementation Benin: Decentralized rural energy (1998) WB 1.1 5.7 pending - approval by WB Cape Verde: Energy & water sector WB 4.9 65 under - reform and development (1998) implementation China: Renewable energy development WB 35 445 under - (1998) implementation Peru: PV-based rural electrification (1998) UNDP 4 9.2 under - implementation Solar Development Group (1998) IFC 10 50 under - implementation Togo: Decentralized rural energy (1998) WB 1.1 5.7 pending - approval by WB Guinea: Rural energy (1999) WB 2 10 pending - approval by WB Malawi: Barrier removal to Malawi UNDP 3.4 10.7 pending renewable energy program (1999) approval by UNDP Mexico: Renewable energy for agriculture WB 8.7 26 pending (1999) approval by WB 21 (Nj C(NJ Annex 2: Approaches to Solar Home Systems in Africa23 Three UNDP/GEF African projects illustrate the two In contrast, the Uganda project is based on the sales primary approaches to solar home systems. The model. Consumer credit is provided through two lo- Ghana project employs the fee-for-service model, cal credit institutions: a private rural development while the Uganda and Zimbabwe projects employ a bank and a credit-union type of women's trust. In dealer-sales model. While Ghana and Uganda have addition to the GEF grant, UNDP has provided been under implementation for less than a year, the cofinancing to guarantee credit lines of these institu- Zimbabwe project was the first solar home system tions. The project is at very early stages of implemen- project in the GEF portfolio to be completed (in tation, and hence it is not possible to assess the 1998). likelihood of overall project success. But there is a clear contrast in this approach to the Ghana project; in The goal of the Ghana project is to establish a sustain- Uganda, the project is clearly targeted towards those able capacity in Ghana to provide decentralized re- who are credit worthy and can afford the cost of credit newable energy-based electricity services to rural (perhaps only the top 10% wealthiest households). communities through the fee-for-service model. The Households that cannot afford commercial credit still project is under implementation through a special constitute the vast majority of the rural population. office-the Renewable Energy Services Project (RESPRO)-established in the Ministry of Mines The Zimbabwe project was also based on the sales and Energy (MOME). RESPRO is intended to act as a model. From 1995 to 1998, over 10,000 solar home for-profit enterprise to be "spun-off' as a private systems were sold, primarily through private dealers. sector company towards the end of GEF project de- A utility-sales model was also piloted, through the sign. This is a departure from the original project national electric utility, which sold about 200 systems design, where the project was to have been imple- under the project but appeared to lack sufficient inter- mented by the Volta River Authority/Northern Elec- est to continue after the project completed. Expected tric Department (VRA/NED), the electricity utility in experience with sales by NGOs was limited. Con- Ghana, which is expected to be privatized in the sumer credit was provided by the Agricultural Fi- future. The current implementation structure does nance Corporation (AFC), a development institution, raise questions about the potential for privatization of through a revolving fund mechanism. The AFC pro- RESPRO, as it is currently housed within a ministry. vided credit to 4,200 households but has been unable to replenish the fund, which will be depleted without The project, which has just started implementation, replenishment. targets some of the poorest households in northern Ghana, and expects to sell electricity through installa- The Zimbabwe project was designed to enhance and tion of 50 Wp (for the equivalent of US$7 per upgrade indigenous solar manufacturing and delivery month) or 100 Wp (US$12 per month) in house- infrastructure, to develop an expanded commercial holds. Willingness-to-pay surveys and demand market in rural areas for affordable domestic solar from households show that these rates are afford- electric lighting by providing low-interest financing able. However, it is unclear whether these rates can through existing institutions, and to establish new generate enough revenue to offset expenses. includ- credit mechanisms at the grassroots level to benefit ing capital, operation, and maintenance costs. lower income groups in rural areas (both households 23 Annex 2 is based on: field visit to Ghana. October. 1999 (met with UNDP; World Bank; Ministry of Finance and Economic Planning: Ministry of Environment. Science and Technology; Ministry of Mines and Energy; a member of parliament; University of Science and Technology: MICAP Coral Technology: Solarvent: Solarco; Wilkins Engineering. Gold River Solar Electric. Deng Limited, and Solar Light Co.): field visit to Uganda, October. 1999 (met with UNDP: World Bank: African Development Bank: Ministry of Finance. Planning and Economic Development: Ministry of Energy and Mineral Development; Uganda Renewable Energy Association; Uganda Women's Finance and Credit Trust Ltd.: Solar Energy Uganda Ltd; Solar Energy for Africa: Impact Solar Systems; Shell Uganda Ltd.); Martinot and McDoom, op. cit. note 4; Resource Futures International. "Lessons learned during the GEF pilot phase" (Ottawa. Ontario, 1998). 23 and community-based institutions). The project has poorer of the rural population than the sales model. had a number of impacts on the market for PV sys- Also, the fee-for-service model looks affordable to tems in Zimbabwe, including a greatly expanded net- larger sections of the rural population, and hence work of dealers, reduced market prices (partly might have better potential for developing large mar- through elimination of import duties on imported kets for rural solar PV applications. Regardless of the components), improved technical knowledge among model used, continued finance after the project, either firms, establishment of PV module standards for cer- from private or public sources, will be essential for tifying and guaranteeing installed systems, develop- the sustainability of the energy-service businesses or ment of equipment certification institutions and the delivery of consumer credit through develop- procedures, and much greater awareness of PV by ment-finance or credit-union institutions. consumers, NGOs and government. In comparing these projects. the fee-for-service deliv- ery model seems to be oriented more towards the 24 Annex 3: Consumer Credit Through Microfinance Organizations in Sri Lanka24 The Sri Lanka project demonstrates the initial viabil- other MFIs in the market, credit delivery through ity of a "microfinance model" in which solar home Sarvodaya may simply be too slow for the market system (SHS) dealers market, sell, service, and guar- expansion desired by suppliers. antee their products to rural consumers through their own local sales/service offices. Consumers obtain The project also demonstrates the initial failure of a loans from Sarvodaya, a national microfinance insti- fee-for-service model in that country. Initially, one tution (MFI) with many local branches and strong ties dealer provided 140 systems on a fee-for-service ba- to the communities in which it operates. A customer sis and thought this approach held promise. But it signs a credit agreement with Sarvodaya, Sarvodaya soon stopped offering systems this way because it did pays the supplier, and Sarvodaya is responsible for not want the expense of monthly collections in a fee- repayment and collections. The supplier provides for-service scheme. "Collection costs were eating up maintenance service for the first three years, a one- our entire profit margin." the dealer said. "You need a year warranty for the system and a 10-year warranty strong fee collection system with good timing, other- for the PV module. The credit provided by the wise customers will spend the money on something microfinance organization for purchase of solar else (if your timing is off) and default. Or they say home systems is similar in kind to that provided for they will pay next month and ask us to wait, or cite enterprise development: $500 with 20 percent down poor performance. It's a continuing problem. Also, payment, terms of up to five years, and 24 percent we found that if customers don't own the system, they interest rate. "This is the only way to go," said the won't take proper care of it and this increases our two major SHS suppliers in the market, who today costs." sell more than 90 percent of their systems this way (now 50-100 systems per month). For 2000, Dealer credit through the project suffered the same Sarvodaya has signed agreements with the two major fate. In the early stages of the project, suppliers found SHS suppliers to provide credit for an additional collections too difficult and time-consuming. "Build- 5,000 systems and is looking at extending credit for ing a rural service infrastructure with technicians is a 10.000 systems in 2001. very different business from building a rural credit delivery and collection infrastructure," said the sup- However, Sarvodaya is currently "the only game in pliers. "Credit is not [the suppliers] business." echoed town" in terms of consumer credit, and there is a need one industry observer, "it is the business of for other MFIs to participate in the SHS market. It microfinance institutions, and the success of credit appears that the market is being constrained by the depends on local connections, knowledge, and insti- lack of other MFIs with whom suppliers can sell tutions already in place." One factor affecting the systems on credit. "Help us to strengthen the rural viability of dealer credit is a very low rural population credit structure," advised the two suppliers. density in Sri Lanka. Transport and labor costs in- "Sarvodaya is a social mobilization organization, not volved in collections are substantial because of the really a business," said one SHS industry observer. long distances and time required to travel those dis- Sarvodaya sees these projects as primarily social tances by supplier personnel. Population density and projects and thus does not approach the market with transport costs also greatly influence supplier costs the aggressiveness of a private company. Without for marketing and service. 24 Annex 3 is based on: field visit to Sri Lanka, November 14-19. 1999 (met with World Bank Project Management Unit. Shell Renewables Lanka Ltd., Resco Asia Ltd., Alpha Thermal Systems Ltd., Sarvodaya Economic Enterprises Development Services (central office and Mayangaya district office). UNDP. World Bank, Lalith Gunaratne & Associates, DFCC Bank, Sanasa Development Bank Ltd., Hatton National Bank, Browns Ltd., and three households that purchased solar PV systems); World Bank project supervision reports; Lalith Gunaratne, "Funding and repayment management of PV system dissemination in Sri Lanka." paper presented at Financial Services for Decentralized Solar Energy Applications 11, 20-23 October 1998. Harare, Zimbabawe; Jayantha Nagendran. "Building local capacity in rural and renewable energy: Emerging lessons from Sri Lanka." paper presented at World Bank Energy Week, 6-9 April 1999. Washington. DC. 25 Both SHS suppliers and Sarvodaya have had no prob- one directly attributable to the World Bank/GEF lem obtaining business financing from commercial project.25 "Anybody who is really going to advance banks and do not anticipate having problems in the the market here has to have deep pockets," said one future. But other MFIs besides Sarvodaya may face long-time participant and observer of Sri Lanka's more difficulty, as commercial financiers see lending SHS market. Shell also signed a memorandum of to MFIs as too risky and a marginal business. Because understanding with the national electric utility that MFIs do not have assets, some observers felt that included a statement by the prime minister that the commercial banks would not lend to them, but this government would promote the private SHS industry. has not been tested yet in the project. "The credibility of SHS has increased several notches" within the government due to the World In 1999, Shell International Renewables purchased Bank/GEF project and Shell's entry into the market, one of the existing SHS dealers and observers saw said an industry observer. Shell's entry into the market as a very promising sign, 25 Shell International Renewables has stated that they would not have entered the Sri Lanka market if the World Bank/GEF project had not taken place, according to the World Bank task manager for the project. 26 Annex 4: Solar PV Businesses and the IFC/GEF Small and Medium Scale Enterprise Program26 Business financing is being provided under the IFC/ to be enormously draining on their time, resources, GEF Small and Medium-Scale Enterprise Program to and profitability. Grameen Shakti is finding that after three solar home systems businesses in Bangladesh, a "critical mass" is reached in a particular community Vietnam, and the Dominican Republic. (perhaps 100 systems). word spreads among friends and relatives, people see systems in operation, and The Bangladesh project demonstrates a dealer-credit mark-eting is easier. model in which one organization (Grameen Shakti, legally a non-profit), performs all functions: market- In Vietnam, sales by a private dealer (SELCO) are ing, sales, service, credit provision, collections, and assisted by a complex credit delivery scheme involv- guarantees. From 1997 to 1999. Grameen Shakti in- ing the Vietnam Women's Union (VWU), an NGO, stalled 1.500 systems using this model (about 1,100 and the Vietnam Bank for Agriculture and Rural De- systems since IFC financing began in July 1998). and velopment (VBARD), a development finance institu- plans to install 2.000-2,500 systems in 2000 (consis- tion. VWU markets SELCO's systems and tent with their original business plan). Grameen administers consumer loans provided by VBARD Shakti is so far the only player in the Bangladesh SHS (VWU collects fees for these services). SELCO pro- market. Before the IFC loan, Grameen Shakti was vides systems (receiving full cash payments) and is installing about 20 systems per month using Grameen responsible for service. VBARD provides consumer Bank financing exclusively, which was for one-year loans, assuming risk for 75 percent of the purchase terms only. Therefore, Grameen Shakti could only price. Of the remaining 25 percent of the purchase extend credit for one-year terms, limiting demand price, SELCO provides a collateralized guarantee to greatly. The IFC SME loan enables Grameen Shakti VBARD for 5-10 percent and the customer pays 15- to extend three-year credit to customers, which has 20 percent as a down-payment. SELCO covers its made a large difference in its business. Grameen collaterized guarantee to VBARD with IFC/GEF fi- Shakti believes they will ultimately be able to receive nancing. If a purchaser defaults on the VBARD loan, loans from commercial banks in perhaps another 3-4 SELCO repossesses and refurbishes the system, and years. after they demonstrate profitability. VWU finds a new buyer for it. If there is any loss in this repossession/refurbishing/resale process. In Bangladesh, Grameen Shakti is selling to house- VBARD has access to the SELCO guarantee. Despite holds that have incomes two or three times higher instructions from the head office, some conservative than Grameen Bank "members" (those eligible to branch managers of VBARD have been reluctant to borrow from the Grameen Bank). Grameen Shakti's participate in the SELCO business. Where this has customers represent the top 10 percent to top 15 happened. SELCO has extended consumer credit it- percent of income status among rural households. self. So far SELCO has sold 500 systems in Vietnam. Grameen Shakti's biggest problem is the cost of mar- keting and consumer education. They are spending all In the Dominican Republic, the U.S. firm Soluz has their own money (financed through business loans) been developing a subsidiary, Soluz Dominicana, on this. They don't receive any grants from the gov- into a successful fee-for-service business that targets ernment or Grameen Bank. Grameen Shakti finds the up to 50 percent of the population in the rural commu- process of building customer demand and confidence nities it serves and charges $10 to $20 per month for 26 Annex 4 is based on: field visit to Bangladesh November 20-22, 1999 (met with Grameen Shakti, Bangladesh Center for Advanced Studies, and World Bank); field visit to Vietnam October. 1999 (met with IFC; World Bank; UNDP; Japan Bank for International Cooperation; Electricity of Vietnam; Ministry of Science, Technology and Environment; Vietnam Women's Union; Vietnam Bank for Agriculture and Rural Development; BP Solar Vietnam; Selco Vietnam; Shell Vietnam; Trans Energ); Richard Hansen, "Solar Electric Energy Delivery-A Business Model," paper presented at Village Power '98, 6-8 October 1998. Washington, DC (Golden. CO: National Renewable Energy Laboratory); personal communications with Richard Hansen of Soluz in October 1998. February 2000 and April 2000. 27 electricity service from SHS.27 The Soluz business model R&D" very difficult to fund from operating model revolves around a "service center" for up to revenue alone at the proof-of-concept scale. Thus 2,000 customers and "zones" of about 500 customers Soluz sees the need for concessional funding to help served by technicians collecting payments at "collec- it cover first-time commercialization costs, including tion points" covering 20 to 100 customers. Collection financial transaction engineering and optimization of rates have been typically over 95 percent, although to its business and technical systems. Said Soluz of its maintain high rates Soluz Dominicana has needed to commercialization efforts, "We are doing the work to make household visits to a portion of customers. prepare for a $5-l million company (25,000-50.000 Through continuous improvement to the business customers), but we are concerned about burdening a model, including business and technical systems opti- $1 million company (5,000 customers) with the over- mization, Soluz expects to complete a robust "proof head and first-time costs of building an energy-ser- of concept" with Soluz Dominicana at a scale of vice company on a larger scale. The business is risky, 5,000 fee-for-service customers. As of April 2000, and the GEF still has a legitimate role. This is a lean Soluz Dominicana had installed over 3,500 systems margin business, so a small difference can affect and had passed the break-even point where revenues profitability greatly. Over the next three years we cover the direct costs of operations. About 1,70() of need to create a strong franchise. There are necessary these systems have been installed on a fee-for-service overheads as well as first-time innovation and trans- basis. Soluz also established Soluz Honduras to enter action costs to keep the thing going on a solid path- the Central American market and diversify. the question is how to cover these costs. To rely only on equity would place high pressure for rapid and Soluz is now working on developing its business more difficult-to-achieve growth to meet return re- model to the point where eventually it will be able to quirements. which adds risk." support 25.000 customers. Soluz finds such "business 27 Monthly fees are $10 fora 20 Wp system, $15 for40 Wp. and $20 for 5OWp. Customers own and are responsible forthe battery, although Soluz Dominicana can include financing for batteries in the monthly fees. 28 Annex 5: Rural Energy Service Concessions in Argentina28 The World Bank/GEF Renewable Energy in the Ru- Concessions are eligible to re-bid for their business ral Market project aims to supply electricity to 66,000 every 15 years (for up to a total of 45 years) competi- households with individual solar home systems (of tively against other eligible firms. The 15-year period 5OWp to 400Wp), 1,100 public facilities with solar was seen as a compromise between the need for a photovoltaic systems, and 3,500 households with vil- short period for the quasi-monopoly and a long period lage power systems (using mini-hydro or hybrids for the annuity calculations of the concession. After such as solar/wind. wind/diesel, or solar/diesel) 15 years, the government may modify the concession through province-level energy service concessions. rules to account for new technological developments, Concessions are free to select which technology to or may even decide to abandon the concession system apply in any given situation, including diesel-only and open the market to competition. During the 15 village power systems. Concessions will be obligated year period, the concession, provincial government, to: and provincial utility regulatory agency renegotiate the tariffs every 2 years. * provide electricity services to rural off-grid cus- tomers anywhere in the province for a period of Eight provincial governments (out of 22 total) are at least 15 years, upon request; eligible to participate in the project. Each of these * carry out all necessary maintenance, repairs, or provinces has privatized or is in the process of priva- replacement of components as needed to ensure tizing its power sector, or at least has made a legal the continuity of the electricity service to each and commitment to privatize.Four of these provinces every customer; have existing private concessions serving the concen- * provide "state-of-the-art commercial service stan- trated (urban) market that are regulated by the provin- dards" forconnection requests, billing, collection, cial governments. Under the project, these and claims handling; and governments will first try to negotiate a rural conces- * provide the provincial utility regulatory agency sion contract with their existing concessions (as an (ENRESP) with periodic reports on the status amendment to the existing contract). If such negotia- of the concession including but not limited to tion fails. or if there is no existing concession for that performance indicators such as number of con- province, then a new concession contract will be nections by type of consumer and method and awarded according to international competitive bid- technology supply, outages statistics, and finan- ding procedures. cial results. 28 Annex 5 is based on Eric Martinot and Kilian Reiche, "Regulatory Approaches to Off-Grid Electrification and Renewable Energy: Case Studies from Six Developing Countries," (Wk'ashington. DC: World Bank. 2000). 29 Global Environment Facility 1818 H Street, NW Washington, DC 20433 USA tel: I (202)473-0508 fax: 1(202)522-3240/3245 www.gefweb.org Sg_~~~~~~~_