THE WORLD BANK 45400 The Welfare Impact of Rural Electrification: A Reassessment of the Costs and Benefits An IEG Impact Evaluation RECENT IEG PUBLICATIONS 2006 Annual Report on Operations Evaluation Annual Review of Development Effectiveness 2006: Getting Results Addressing the Challenges of Globalization: An Independent Evaluation of the World Bank's Approach to Global Programs Assessing World Bank Support for Trade, 1987­2004: An IEG Evaluation Books, Buildings, and Learning Outcomes: An Impact Evaluation of World Bank Support to Basic Education in Ghana Brazil: Forging a Strategic Partnership for Results--An OED Evaluation of World Bank Assistance Bridging Troubled Waters: Assessing the World Bank Water Resources Strategy Capacity Building in Africa: An OED Evaluation of World Bank Support China: An Evaluation of World Bank Assistance The CGIAR at 31: An Independent Meta-Evaluation of the Consultative Group on International Agricultural Research Committing to Results: Improving the Effectiveness of HIV/AIDS Assistance--An OED Evaluation of the World Bank's Assistance for HIV/AIDS Control Country Assistance Evaluation Retrospective: OED Self-Evaluation Debt Relief for the Poorest: An Evaluation Update of the HIPC Initiative A Decade of Action in Transport: An Evaluation of World Bank Assistance to the Transport Sector, 1995­2005 The Development Potential of Regional Programs: An Evaluation of World Bank Support of Multicountry Operations Development Results in Middle-Income Countries: An Evaluation of the World Bank's Support Economies in Transition: An OED Evaluation of World Bank Assistance Engaging with Fragile States: An IEG Review of World Bank Support to Low-Income Countries Under Stress The Effectiveness of World Bank Support for Community-Based and -Driven Development: An OED Evaluation Evaluating a Decade of World Bank Gender Policy: 1990­99 Evaluation of World Bank Assistance to Pacific Member Countries, 1992­2002 Extractive Industries and Sustainable Development: An Evaluation of World Bank Group Experience Financial Sector Assessment Program: IEG Review of the Joint World Bank and IMF Initiative From Schooling Access to Learning Outcomes: An Unfinished Agenda--An Evaluation of World Bank Support to Primary Education Hazards of Nature, Risks to Development: An IEG Evaluation of World Bank Assistance for Natural Disasters How to Build M&E Systems to Support Better Government IEG Review of World Bank Assistance for Financial Sector Reform Improving Investment Climates: An Evaluation of World Bank Group Assistance Improving the Lives of the Poor Through Investment in Cities Improving the World Bank's Development Assistance: What Does Evaluation Show? Maintaining Momentum to 2015: An Impact Evaluation of Interventions to Improve Maternal and Child Health and Nutrition Outcomes in Bangladesh New Renewable Energy: A Review of the World Bank's Assistance Pakistan: An Evaluation of the World Bank's Assistance Pension Reform and the Development of Pension Systems: An Evaluation of World Bank Assistance Poland Country Assistance Review: Partnership in a Transition Economy The Poverty Reduction Strategy Initiative: An Independent Evaluation of the World Bank's Support Through 2003 The Poverty Reduction Strategy Initiative: Findings from 10 Country Case Studies of World Bank and IMF Support Power for Development: A Review of the World Bank Group's Experience with Private Participation in the Electricity Sector Putting Social Development to Work for the Poor: An OED Review of World Bank Activities Small States: Making the Most of Development Assistance--A Synthesis of World Bank Findings Social Funds: Assessing Effectiveness Sourcebook for Evaluating Global and Regional Partnership Programs Water Management in Agriculture: Ten Years of World Bank Assistance, 1994­2004 World Bank Assistance to the Financial Sector: A Synthesis of IEG Evaluations The World Bank in Turkey: 1993­2004--An IEG Country Assistance Evaluation World Bank Lending for Lines of Credit: An IEG Evaluation All IEG evaluations are available, in whole or in part, in languages other than English. For our multilingual section, please visit http://www.worldbank.org/ieg W O R L D B A N K I N D E P E N D E N T E V A L U A T I O N G R O U P The Welfare Impact of Rural Electrification: A Reassessment of the Costs and Benefits An IEG Impact Evaluation 2008 The World Bank http://www.worldbank.org/ieg Washington, D.C. ©2008 The International Bank for Reconstruction and Development / The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org E-mail: feedback@worldbank.org All rights reserved 1 2 3 4 5 11 10 09 08 07 This volume is a product of the staff of the International Bank for Reconstruction and Development / The World Bank. The findings, interpretations, and conclusions expressed in this volume do not necessarily reflect the views of the Executive Di- rectors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denomina- tions, and other information shown on any map in this work do not imply any judgment on the part of The World Bank con- cerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The International Bank for Reconstruction and Development / The World Bank encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center Inc., 222 Rosewood Drive, Danvers, MA 01923, USA; telephone: 978-750-8400; fax: 978-750-4470; Internet: www.copyright.com. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2422; e-mail: pubrights@worldbank.org. Cover photo: A woman in Bangladesh stands beside a household electric meter. Photo courtesy of NRECA International Programs. ISBN-13: 978-0-8213-7367-5 e-ISBN-13: 978-0-8213-7368-2 DOI: 10.1596/978-0-8213-7367-5 Library of Congress Cataloging-in-Publication Data have been applied for. World Bank InfoShop Independent Evaluation Group E-mail: pic@worldbank.org Knowledge Programs and Evaluation Capacity Telephone: 202-458-5454 Development (IEGKE) Facsimile: 202-522-1500 E-mail: eline@worldbank.org Telephone: 202-458-4497 Facsimile: 202-522-3125 Printed on Recycled Paper Contents vii Abbreviations ix Acknowledgments xi Foreword xiii Executive Summary xix Chairperson's Summary: Committee on Development Effectiveness (CODE) 1 1 Introduction 3 The Shifting Rationale and Returns to Lending for Rural Electrification 4 Evaluation Questions 5 The Study Approach 7 2 World Bank Lending for Rural Electrification 9 The Bank's Evolving Energy Strategy 10 The Portfolio 11 Objectives 13 Project Design: Analysis of Components 14 Outputs and Outcomes 17 3 Who Benefits from Rural Electrification? 19 The Distribution of Electrification 19 Which Communities Get Electricity? 23 Which Households Get Electricity? 27 The Distribution of Benefits from Electrification 27 Concluding Comment 29 4 What Is Electricity Used for in Rural Areas? 31 What Types of Connection Are There? 33 Domestic Uses of Electricity 34 Community Uses 34 Productive Uses 37 5 The Benefits of Rural Electrification 39 Domestic Uses: Lighting and TV 42 Health Benefits 45 Time Use 46 Education Benefits 46 Productive Uses 47 Global Benefits 48 Adding Up the Benefits 50 How Do the Benefits Compare with the Costs? 50 Off-Grid Connections iii THE WELFARE IMPACT OF RURAL ELECTRIFICATION 53 6 Conclusion and Lessons Learned 55 Answering the Evaluation Questions 56 Lessons Learned 59 Appendixes 61 A: Rural Electrification Portfolio 67 B: Statistical Overview of Portfolio 83 C: Distributional Analysis of Rural Electrification 89 D: Uses of Electricity 97 E: Literature Survey 109 F: Impact of Rural Electrification on Microenterprise 117 G: Health and Education 131 H: Calculating Consumer Surplus 141 I: Evaluation Approach Paper 147 Endnotes 151 Bibliography Boxes 21 3.1 Successful RE through a Multisectoral CDD Project 22 3.2 Selection of Projects under the Peru Rural Electrification Project 23 3.3 Chile Rural Electrification Fund 25 3.4 India's Experience with the Single Point Light Connection Scheme: Kutir Jyoti 26 3.5 Overcoming the Connection Cost Barrier 27 3.6 Poor Communication of Tariff Structures Can Disadvantage the Poor 34 4.1 The Cold Chain 35 4.2 Electrification and Worker Absenteeism in the Social Sector 40 5.1 Shedding Light on Lumens 43 5.2 The Health Risk of Candles 47 5.3 Micro Home Enterprises 52 5.4 Technical Problems Reduce the Benefits from Off-Grid Investments Figures 10 2.1 A Growing Number of Rural Electrification Projects Are in Latin America and Sub-Saharan Africa 11 2.2 Increased Energy Supply and Institutional Development Account for the Largest Share of Objectives 12 2.3 Gender Issues Are Increasingly Taken into Account but Still Affect the Design of Only a Minority of Energy Projects 20 3.1 Pattern of Electrification Favors the Non-Poor, but This Bias Generally Reduces over Time as Electrification Coverage Expands 20 3.2 Share of Poor of On-Grid Electricity Consumption Is Low 24 3.3 A Large Proportion of Households Connect to the Grid Immediately after It Becomes Available . . . But Some Remain Unconnected after Many Years 32 4.1 The Energy Ladder 32 4.2 Nearly All Projects Provide Residential Connections, but also Other Connections for Productive Purposes 34 4.3 Pattern of Consumption in Rural Households (Distribution Total kWh) 40 5.1 Consumer Surplus 40 5.2 Producer Surplus 51 5.3 Willingness to Pay Exceeds Supply Cost iv CONTENTS Tables 24 3.1 Relative Price of Grid, Off-Grid, and Kerosene ($/kWh) for Selected Countries 41 5.1 Willingness to Pay Calculation for Lighting 42 5.2 Willingness to Pay Calculation for TV 45 5.3 Fertility Impact of Electrification 46 5.4 Hours Watching TV by Electrification Status 49 5.5 Rural Electrification Benefits (US$ per Household per Month) v Electricity distributors at a mini hydroelectric plant in Sri Lanka. (Photo from the World Bank Photo Library.) Abbreviations CDD Community-driven development DHS Demographic and Health Survey DSM Demand-side management ERR Economic rate of return ESMAP Energy Sector Management Assistance Program GEF Global Environment Facility HAZ Height for age IBRD International Bank for Reconstruction and Development (World Bank) IEG Independent Evaluation Group klh Kilo lumen hours km Kilometer kWh Kilowatt hours LSMS Living Standard Measurement Survey (Peru) Norad Norwegian Agency for International Development O&M Operations and maintenance PAD Project Appraisal Document PPAR Project Performance Audit Report PV Photovoltaic RE Rural electrification REF Rural Electrification Fund RET Renewable energy technologies SHS Solar home system TFR Total fertility rate USAID United States Agency for International Development WAZ Weight for age Wp Watt peak WTP Willingness to pay All dollar amounts are U.S. dollars unless otherwise indicated. v i i Villagers extending cables to provide village with electricity from hydro power. (Photo from the World Bank Photo Library.) Acknowledgments The report was prepared by Howard White (Task Thanks are due to the following for providing en- Manager), Nina Blöndal, Morgan Rota, and Anju ergy survey data and related documentation: Doug Vajja, with inputs from Robin Banerjee, Mollie Barnes (the Philippines); Morten Larsen, Jie Tang, Fair, Tara Lonnberg, and Andrew Waxman under and Voravate Tuntivate (Lao People's Democratic the supervision of Alain Barbu. Administrative Republic); and René Masse (Sri Lanka). Thanks are support was provided by Soon Won-Pak, and also due to Gustavo Angeles (Carolina Population the report was edited by William Hurlbut and Center, University of North Carolina) for running Heather Dittbrenner. The external reviewers for tabulations using data from the health facility sur- the report were Andrew Barnett and Stein vey for Nicaragua, and to Chris Edwards for advice Hansen. at various stages of the analysis. Director-General, Evaluation: Vinod Thomas Director, Independent Evaluation Group, World Bank: Cheryl Gray Manager, Sector, Thematic, and Global Evaluation: Alain Barbu Task Manager: Howard White ix Village home in Sri Lanka lit by solar energy. (Photo from the World Bank Photo Library.) Foreword This report is part of the Independent Evaluation evidence from a number of different countries; it Group's (IEG) impact evaluation series. These uses data from a range of sources, both existing studies fit under the category of "rigorous but studies and reanalyses of existing survey data. Al- relevant" evaluations, seeking to use a variety of though the report touches on aspects of sector data sources both to demonstrate impact and performance, it does not claim to be a compre- to deliver policy-relevant conclusions. This study hensive sector review. is the first of the impact evaluations to combine xi A woman in Bangladesh beside a household electric meter. (Photo courtesy of NRECA International Programs.) Executive Summary I t has long been claimed that rural electrification greatly improves the quality of life. Lighting alone brings benefits such as increased study time and improved study environment for school children, extended hours for small businesses, and greater security. But electrification brings more than light. Its sec- ing from project objectives; when present, they are ond most common use is for television, which assumed to follow automatically from the outputs. brings both entertainment and information. The But the connection cannot be taken for granted. people who live in rural areas greatly appreciate these benefits and are willing to pay for them at Project design components to ensure that out- levels more than sufficient to cover the costs. puts do result in the intended outcomes are rare, However, the evaluation of these and other ben- though they are increasing. To give this results ori- efits (for example, in terms of public goods), as entation further impetus, this assessment by the In- well as of their distribution, has been sparse. dependent Evaluation Group (IEG) examines anew the costs and benefits of RE for Bank-supported This report reviews recent methodological ad- projects in several Regions of the world. vances made in measuring the benefits of rural elec- trification (RE) and commends them. It also notes Background to the Study that the understanding of the techniques shown The World Bank has made loans for power gen- in project documents is sometimes weak, and eration, transmission, and distribution since its ear- quality control for the economic analysis in proj- liest years. By the 1980s it was lending substantial ect documents lacking. This study shows that amounts for expanding coverage into rural areas. willingness to pay (WTP) for electricity is high, ex- However, a 1994 IEG report, Rural Electrification ceeding the long-run marginal cost of supply. in Asia, cast doubt on these investments, arguing that the rates of return were low because many of Hence, in principle, RE investments can have the claimed benefits were not realized and that the good rates of return and be financially sustainable. costs of these programs imposed a financial bur- But caveats are in order. The first caveat is that at- den on the provider. Since that time, financial re- tention needs to be paid to ensuring least cost sup- forms have been implemented in a number of ply, including limiting system losses. Second, countries, and the RE portfolio has seen significant continued attention needs to be paid to achiev- shifts in terms of project objectives and design. ing the right balance between financial sustain- ability and reaching the poor. In addition, in response to that IEG report, the Energy Sector Management Assistance Program The World Bank has been financing RE for decades (ESMAP) carried out a study in the Philippines to in Asia, and it has been expanding such activities quantify a broader range of benefits from RE. Most in Latin America and Africa. Its support for RE has notably, that study developed a new methodology focused on outputs--building infrastructure and for measuring the benefits of electric lighting that institutions. Yet outcomes have often been miss- has been widely adopted in project appraisals, xiii THE WELFARE IMPACT OF RURAL ELECTRIFICATION giving very acceptable rates of return. The main Another recent trend is the growth of support for focus of IEG's current study is to review these off-grid electrification, usually as a subcompo- claims and examine the extent to which changes nent of a larger project, as in the Southern in the portfolio have addressed earlier concerns re- Provinces Rural Electrification Project and follow- garding the limited poverty impact of lending to RE. on Rural Electrification Project in Lao People's Democratic Republic. Most off-grid projects rely The study analyzed data from a range of sources, on renewable energy technologies, which have including IEG's own analysis of existing data sets also become more prominent in the Bank's lend- for a dozen countries (three energy surveys, nine ing in the last 15 years. Demographic and Health Surveys, and two in- come and expenditure surveys) and a review of Three-quarters of RE projects have objectives re- Bank and external studies. The analysis unpacks lated to improving energy supply, and the same the causal chain from the provision of electricity proportion has objectives related to institutional to the various benefits it is claimed to bring, and development. Only 60 percent have the objective quantifies these benefits where possible to address of increasing welfare (including environmental the balance of costs and benefits. The data were benefits), and this objective is mostly stated in gen- used to test the impact of RE on several variables, eral terms, such as improving incomes. More- such as the quantity of lighting used, opening over, this objective is most common in the hours of clinics, female health knowledge, and in- multisectoral projects. Only 7 percent of dedi- come from home businesses. cated RE projects and energy sector projects have an explicit poverty-reduction objective. Hence, The Bank's Portfolio poverty has not become a central concern of RE The Bank's strategy for the energy sector has projects, and there is rarely any explicit consid- evolved considerably in the last 15 years. In 1993 eration either of how the poor will be included two policy papers were published that gave greater or of any poor-specific activities. Similarly, al- emphasis to the role of the private sector and high- though mention of gender in project documents lighted environmental concerns (World Bank has increased greatly in the last decade, these 1993a, 1993b). A 1996 paper discussed the 2 bil- concerns rarely affect project design. lion poor people around the world lacking access to modern energy services and how the Bank Where the Bank finances a series of dedicated proj- may best meet their needs (World Bank 1996), and ects it can make a substantial contribution to in- a 2001 sector board paper increased the empha- creasing RE coverage: in Indonesia coverage rose sis on both poverty and the environment (World from 33 percent in 1991 to 85 percent by 2003, Bank 2001b). How have these strategy changes with about 45 percent of these additional con- been reflected in the RE portfolio? nections being paid for with Bank financing. In Bangladesh, the number of rural connections IEG identified 120 Bank-supported projects with grew from practically zero in 1980 to more than RE activities since 1980, falling roughly equally into 4 million by 2002; 600,000 of these connections three categories: dedicated projects (such as were made with Bank financing. Bangladesh Rural Electrification I, II, and III), en- ergy sector projects with RE components (such By and large, Bank-supported projects have suc- as the Jordan Energy Development Project), and cessfully created the physical infrastructure for RE, multisector projects with RE components (such although technical problems have often meant as Brazil's Northeast Rural Poverty Alleviation high system losses--which have reached as high Projects). A growing number of these projects as 50 percent in Albania and India (Rajasthan). are in Latin America, where RE is common in These losses drive a wedge between the cost of multisectoral community-driven development generation and the cost of supply, thus under- projects, and Sub-Saharan Africa. mining financial performance. Many Bank projects xiv EXECUTIVE SUMMARY have components to address this problem of sys- est areas. In other cases, such as the Ghana Na- tem loss, but not all have been successful. tional Electrification Project, the Bank has ac- ceded to the government's request to ensure There has been less success with institutional geographically equitable coverage. In a small num- development, with the majority of unsatisfactory ber of cases, RE funds have been used to offset projects being rated such for this reason. The the financial loss incurred by private companies poor overall performance of the subsector-- that extend coverage to less advantaged rural with just 68 percent of projects rated satisfactory areas. from 1996 to 2006 (compared with 75 percent for the Bank as a whole)--mainly reflects institu- Although off-grid connections can serve remote tional problems. These problems commonly re- communities that may not be connected to the grid late to the lack of financial sustainability of the for some years, they do not necessarily reach the utility responsible for distribution, as tariffs are poor better than grid extension does. Bank sup- set below cost recovery. But the situation is chang- port to off-grid electrification is typically through ing; some countries have introduced higher tar- a private business model, so social concerns have iffs and others, such as Lao PDR, are on track to to be weighed against financial viability. do this. But there also remain a number of coun- tries in which financial performance requires fur- In most countries, increases in coverage come ther attention. from extensive growth (extending the grid to new communities) rather than intensive growth (con- Who Benefits from Rural Electrification? necting the unconnected in already electrified vil- It is widely recognized that the larger share of ben- lages). Once electricity arrives in a village, the efits from RE is captured by the non-poor. IEG connection charge is a hurdle that prevents the poor analysis shows that this continues to be the case, from connecting to the grid, even though the ben- although the gap closes as coverage expands. efits they would derive--and so their WTP--would Two factors underpin this anti-poor pattern in exceed the cost of supply. electrification: which communities get connected and which households can afford the connection Even in villages that have been connected for once the grid is available. 15­20 years, it is not uncommon for from 20 to 25 percent of households to remain unconnected In many countries communities to be connected (for example, in Lao PDR). The absence of credit to the grid are identified on a "least cost" basis, markets means households cannot borrow to pay which favors which larger communities nearer the connection charge. Only a very small number to the existing grid, roads, and towns. The Bank of Bank-supported projects have either extended has promoted this approach, which is often nec- credit to customers (for example, the Second Ac- essary to secure the financial viability of the RE pro- celerated Rural Electrification Project in Thailand) gram, in a number of countries. For example, the or allowed the connection charge to be paid over recent Peru Rural Electrification Project changed a number of years. Because the poor do not con- community prioritization from the government's nect, progressive tariff structures have proved to "social criteria" to a least cost approach. be regressive subsidy schemes--so better-targeted connection charges would be consistent with the Although this is necessary for the financial health Bank's priority of ensuring that the poor benefit of the service provider, there is a clear trade-off directly. with reaching the more disadvantaged. Hence, some countries include social variables in their el- The same point applies to off-grid schemes, which igibility criteria; in Bank-supported projects this are more expensive to the consumer than grid has most often been the case for community- electricity. In some countries, the subsidy pro- driven development projects that target the poor- vided to these schemes is tilted toward the smaller xv THE WELFARE IMPACT OF RURAL ELECTRIFICATION systems likely to be chosen by poorer house- positive impact of RE on service provision comes holds. For example, this is the case with the Philip- from the greater willingness of health and edu- pines Rural Power Project. Also, credit or extended cation workers to stay in communities that have repayment periods for installation costs are more electricity. common for off-grid projects than for grid extension. The lack of large-scale productive uses for rural electricity remains a constraint on the financial vi- The poor who do connect benefit from a "lifeline ability of RE because of low load factors resulting tariff," a low tariff rate--commonly a fixed charge-- from consumption being heavily concentrated in for consumers who use below a certain level, usu- the evening peak hours. ally 25 kilowatt hours (kWh) per month. But poor customer information means that many consumers RE does not drive industrial development, but it unnecessarily restrict consumption to save money, can provide an impetus to home businesses, even when in fact it saves them nothing. though few households use electricity for pro- ductive purposes. IEG's analysis shows that the The full benefits of providing electricity to the poor number of enterprises grows as a result of elec- are not being realized: first, poorer households are trification and that these enterprises operate for not enabled to connect to the grid, and second, more hours. There is, therefore, a positive impact consumers do not get information that allows on household income. However, the broader lit- them to obtain their maximum benefit. Bank- erature has found these effects to be less than ex- supported projects that claim to have the objec- pected, except when there has been a specific tive of bringing RE to the poor have typically program to promote productive uses of electric- neglected to include components that would ity. This is, then, another example of how an ad- help to achieve this objective. ditional project component can help achieve the welfare objective. What Is Electricity Used for in Rural Areas? Benefits of Rural Electrification The dominant use of electricity in rural house- IEG's review endorses the approaches advocated holds is lighting. All households use it for this pur- in the ESMAP study (2003) for measuring the pose, and many use little electricity for anything benefits of lighting and TV; this involves measur- else. The next most common use is TV. Lighting ing them as WTP for lumens (a measure of the and TV account for at least 80 percent of rural elec- quantity or intensity of lighting) in the case of light- tricity consumption and thus the bulk of the ben- ing and hours of TV. There is a caveat that the efits delivered by electrification. Electricity is rarely shape of the demand curve matters (although used for cooking in rural areas, though East Asia the evidence as to its shape is still thin) and that is something of an exception with the use of rice assuming a linear demand curve, as in some stud- cookers. Fans and irons are also used for a minority ies, most likely results in an overestimation of of consumption. project benefits. In one notable case, the claimed economic rate of return of 60 percent fell to 12 The pattern of use has implications for the ben- percent in IEG's recalculation. efits from RE. The potential benefits to be gained from displacing firewood or kerosene stoves are It is also evident that some authors of project eco- not realized in the vast majority of cases. Again, nomic analyses have a weak grasp of the method- consumer education may enable these consumers ology, so the Bank's economic analysis does not to achieve a greater range of benefits. match the quality of the available analytic work. Quality control mechanisms are not in place to stop Electricity is also used in community facilities-- weak analysis appearing in Board documents. But notably for the cold chain for vaccines, though this this view must be balanced with the observation does not appear to affect immunization rates. A that some project documents, such as that for xvi EXECUTIVE SUMMARY the Peru Rural Electrification Project, are best · RE investments can generate sufficient bene- practice examples of cost-benefit analysis. fits for the investment to be warranted from an economic standpoint--and they often The ESMAP approach yields a WTP of around have. $0.10­0.40 per kWh for lighting and TV alone. · The value of these benefits to households is This figure is already well in excess of the average above the average long-run supply cost, so cost- long-run supply cost, which is usually in the range recovery tariff levels are achievable, even if po- of $0.05­0.12 kWh. litically unpopular in countries with a history of low tariffs. This study also considers education benefits (as did · Analysis of feasible tariff levels can be informed the ESMAP study) and health and fertility benefits. by good quality economic analysis of the sort More studies are required to better understand pioneered by the Philippines ESMAP study. But these channels. Other benefits are harder to quan- the quality analysis of that study is not uni- tify. But many of them are most likely internalized formly replicated, as the quality of project-level by the household and so reflected in the WTP. analysis is uneven, with apparent weak quality The exceptions are public good benefits, such as control. street lighting, which increases security, and the · The evidence base remains weak for many of so-called "global benefits" of reduced carbon diox- the claimed benefits of RE. Tailor-made sur- ide emissions, where applicable. Including these veys, designed to test these benefits, need to benefits means the benefit for an average house- be built into a greater number of Bank projects hold consuming 30­40 kWh a month is about $60 and designed to allow rigorous testing of the per month per household. This level is sufficient impact of electrification. to ensure an adequate rate of return for most · Countries with low coverage rates--now mostly grid-extension schemes. in Africa--still have to make investments in generation, transmission, and distribution, Off-grid schemes fare less well because they have which implies relatively high average supply higher costs but lower benefits. Benefits are fur- costs and low coverage, increasing slowly by ther reduced by technical issues, including sup- extensive growth for some years to come. The ply problems. The economic rationale for funding principal challenge is to balance financial sus- off-grid components alongside grid extension tainability with growing coverage, requiring when the latter has the higher economic rate of efficiency by limiting system losses. Grid con- return is far from clear. Such a decision might be nections will grow slowly, so many areas may justified on social grounds, but the case is far be eligible for off-grid connections, but the lo- from proven, especially when much lower subsi- gistics of maintaining technical quality will be dies would be required to reach the poor who are challenging. unconnected in electrified villages. An alterna- · Some countries in Asia and Latin America are tive argument to support these investments is reaching the limits of grid extension. Further in- that these are mostly small-scale programs to creases in coverage require intensive growth, enable learning by doing, which, together with which requires instruments designed for that general cost reductions and technological devel- purpose, or off-grid schemes, which need de- opments, will eventually make off-grid electricity sign improvements if they are to be financially more competitive. sustainable. · There are project design options that have Lessons Learned been uncommon but that would enhance proj- It is difficult to generalize about RE, because both ect benefits. These include financing schemes costs and benefits are context specific. However, for connection charges, consumer education, some broad statements can be made. and support for productive uses. xvii Solar power generates electricity for village shop in Sri Lanka. (Photo from the World Bank Photo Library.) Chairperson's Summary: Committee on Development Effectiveness (CODE) O n December 17, 2007, the Informal Subcommittee of the Commit- tee on Development Effectiveness considered a study entitled The Welfare Impact of Rural Electrification: A Reassessment of the Costs and Benefits, prepared by the Independent Evaluation Group (IEG)-World Bank. IEG Findings and Recommendations pact on the poor. Questions were raised about the The IEG report reviewed recent methodological Bank's support for development of the new advances made in measuring the benefits of rural sources of energy, particularly renewable energy. electrification (RE) and commends them, while Members highlighted the importance of using noting that the understanding of these techniques targeted and well-focused cross-subsidies, which shown in project documents is sometimes weak could increase the positive impact of RE on the and quality control for the economic analysis in poor. Speakers also underlined the need to project documents is lacking. One of the main strengthen cross-sectoral collaboration and pro- findings was that RE investments could generate mote local development initiatives. Members also sufficient benefits to households and the value of stressed the importance of combining qualita- these benefits was above the average long-run sup- tive and quantitative methods while analyzing ply costs. IEG also noted that analysis of feasible the impact of the Bank's programs. tariff levels could be informed by good-quality economic analysis and endorsed the measure- Main Issues ment approaches pioneered by the Philippines The following main issues were raised during the Energy Sector Management Assistance Program meeting. (ESMAP). IEG found that the Bank RE projects have become more explicitly focused on poverty Impact Analysis. A number of speakers appre- reduction. However, complementary measures ciated the methodology used in ESMAP. At the to ensure the highest poverty reduction and so- same time, a member underlined the importance cial impacts--such as educational campaigns, of continued improvement of the analytical tools promotion of productive use, and smart subsi- and asked management to ensure that staff would dies--have been lacking in Bank projects. use such tools in day-to-day work. He also called for the rapid development and application of Overall Conclusions improved evaluation techniques to contribute to Speakers welcomed the IEG impact evaluation and setting up informed and clear objectives and the methodology used for estimating measurable strategies in this sector. A member considered the costs and benefits, particularly with regard to im- scope of the IEG report narrow and would have xix THE WELFARE IMPACT OF RURAL ELECTRIFICATION preferred to see more in-depth analysis of the op- tice. These subsidies are more feasible in the portunity costs. He also remarked that develop- countries with high electrification rates, where ment effectiveness of infrastructure projects the cross-subsidies from industries and com- should be evaluated by the nationwide economic mercial and high-income customers are possible. impact, not by the Regional social impact. A Alternatives also adopted are using low-cost speaker encouraged management to conduct an methods for household connections and fi- analysis of microscale positive effects that elec- nancing to spread the connection cost over sev- trification has on small businesses. eral months. Use of New Technologies. Some speakers noted Community-Driven Mechanisms. Some speak- that the IEG report finding about higher costs but ers sought more information about the decision- lower benefits of off-grid connection compared to making process in RE. They wondered whether grid extension was disputable. They also would such processes should be community driven with have preferred to see recommendations that help the involvement of local institutions and pro- the Bank develop its operations in renewable en- ducer organizations. In this regard, a member ergy. In addition, a member remarked that the underlined the importance of public awareness Global Environmental Facility grants can be used campaigns and noted that the poor should be to promote use of new technologies for electrifi- made aware of the services available to them for cation in remote areas. Management explained the low basic tariffs. IEG responded that in some that grid and off-grid electrification is comple- Regions, particularly in Latin America, com- mentary, and in sparsely populated, remote, or munity-driven development programs provide mountainous areas the off-grid connection is financing for RE. Implementation of the com- the only solution. IEG clarified that there is often munity-driven RE programs also increases con- a trade-off between connecting, at a higher cost, sumers' awareness of their potential benefits. better-off people in remote locations and poorer ones in nearer locations (including already Other Issues connected villages) and that this trade-off needed A member expressed disappointment that gender to be made clear and a rationale provided. issues had not been more central to Bank proj- ects. He also sought more information about the Use of Subsidies. Several speakers commented impact that access to television has on rural house- on the need to further analyze subsidies, because holds. Management agreed that the gender the poverty dimension of RE can be addressed element is important in the electrification sector through cross-subsidization. A member noted and stressed the importance of developing the that recently the International Finance Corpora- appropriate mechanisms to incorporate gen- tion (IFC) Board of Directors approved the Per- der dimensions in analytical and operational formance-Based Grants Initiative, authorizing the work. Regarding the impact of television, IEG IFC to provide subsidies to the private sector to clarified that the study had found that TV in- extend its infrastructure subsidies to the poor. He creased women's health knowledge and so had suggested the IFC share its experience in this area a fertility-reducing impact, but a similar effect with the Bank. Management clarified that the was not found for radio. targeted subsidies for connection charges for low-income households, which the report advo- cates, face implementation difficulties in prac- Jiayi Zou, Chairperson xx Chapter 1 Evaluation Highlights · A 1994 Independent Evaluation Group assessment found that rural electrification projects had lower economic rates of return than ex- pected and benefited the non-poor. · This evaluation calculates returns on rural electrification to determine whether the earlier finding still holds true. Rural power station at a mini hydroelectric power station in Sri Lanka. (Photo from the World Bank Photo Library.) Introduction T he World Bank started as a lender for infrastructure investments, so some of its first loans were to the power sector, such as Loan 0005 for the Power and Irrigation Project in Chile, signed in 1948.1 During the 1950s and 1960s the Bank was heavily involved in electrification projects around the world. Examples include the first loan to Ghana in 1962 paper did acknowledge that stimulation of pro- for the Volta Power Project, the 1957 Philippines duction by RE had in general been disappointing, Binga Hydroelectric Project, and two projects in albeit with exceptions. If these calculations did Nicaragua in the 1950s (the Diesel Power and not provide an acceptable rate of return, then jus- Thermal Power Projects). By the 1970s attention tification might be provided on the grounds of was turning to electrification of rural areas: cu- social benefits, although the paper recognized mulative investments had reached $10 billion by that electricity was not a basic need to be com- 1971, with another $10­15 billion expected to be pared with clean water or health. disbursed during the 1970s. At the time of the 1975 paper many Early analysis of RE The Shifting Rationale and Returns to Latin American countries had estab- suggested that its very Lending for Rural Electrification lished countrywide networks linking high up-front costs A 1975 paper entitled "Rural Electrification" (World major demand centers and were mov- would drop rapidly as Bank 1975) reviewed the rationale for Bank sup- ing to connect smaller rural centers and connections expanded. port to the sector.2 The paper argued, "There is outlying areas; Asian countries were plenty of scope for successful investments in rural then in the process of establishing systems to reach electrification (RE), provided that they are prop- the major demand centers, and African countries erly selected and prepared" (World Bank 1975, were still at the stage of creating their own power- p. 3). The paper also recognized that these in- generation facilities. Hence, by the 1980s the focus vestments would often be loss making, at least ini- of Bank lending for RE was in Asia, such as the tially. The up-front investment costs were very Malaysia Rural Electrification Project (1982­1988) high, and rural demand was considerably lower and the first two Bangladesh RE projects (Bangla- than that in urban areas, resulting in low load desh RE I, 1981­1990, and RE II, 1985­1993). factors and high unit costs. However, a report published by the Indepen- However, marginal costs could fall rapidly as cov- dent Evaluation Group (IEG) in 1994 presented erage expanded and once the main grid was es- generally pessimistic findings regarding these tablished, so connecting neighboring areas could projects. The study noted that-- be relatively inexpensive. Project selection should rest initially on estimation of the economic rate · Ex post ERRs were much lower than those at of return (ERR), valuing benefits as the revenues appraisal, as many of the indirect and external from domestic consumption and the incremen- benefits had not materialized. Notably, there tal value added from productive uses. The 1975 was little impact on industrial development. 3 THE WELFARE IMPACT OF RURAL ELECTRIFICATION · RE projects ignored financial aspects. As had rates of return, reaching levels as high as 95.3 been recognized in the 1975 paper, unit in- percent for the Bangladesh Third Rural Electrifi- vestment costs for RE were much higher than cation Project and 60.5 percent for the Lao Peo- those in urban areas because of lower popu- ple's Democratic Republic Southern Provinces lation density and the low ratio of average Rural Electrification Project. demand to peak demand (rural use is con- centrated in early evening, whereas urban de- A primary intention of this report is to subject mand is spread across the day). Cost recovery these new approaches to critical scrutiny. Does the was low (between 10 and 50 percent), thus 1994 IEG finding that ERRs from RE investments imposing a financial burden on electricity util- are too low still stand? Or do recent changes in ities or governments. the Bank's energy strategy, the nature of its sup- · The direct benefits of RE went to the non- port for RE, and methodological developments poor. Even with low tariffs, the poor cannot af- and evidence overturn that earlier position? ford connection costs. The poverty-reduction benefits of RE were thus indirect and came This question remains extremely relevant. The through rising rural incomes; these effects Bank has an active portfolio in the area. Mean- were found to be limited. while, coverage rates across most of Sub-Saharan Africa are extremely low, with RE rates of well Early projects had lower In the decade following IEG's report, below 5 percent in many countries (see attach- ERRs than expected, the Bank's strategy shifted toward a ment 1 of appendix I). Should the Bank support and benefits went to the stronger poverty focus. This shift has RE in these countries? Or are the returns to such non-poor. been one factor behind changes in the investments insufficient to justify them? portfolio, such as the development of off-grid programs. Evaluation Questions For the purposes of this study, the broad question In a more direct response to IEG's report, further of the justification for RE lending is broken down technical work was carried out to identify and into a number of evaluation questions, leading to quantify the benefits of RE, most notably the sem- the ultimate objective of calculating private and inal report by the Bank's Energy Sector Man- social rates of return from investments in RE. Fol- agement Assistance Program (ESMAP), "Rural lowing are specific questions: Electrification and Development in the Philip- pines: Measuring the Social and Economic Ben- · What is the rationale for World Bank support efits," published in 2003. RE had long been claimed of RE? to have many diverse benefits for health, educa- · What has been the growth in the coverage of tion, nutrition, security, and so on--one study RE in countries receiving Bank support? To provides a list of more than 50 discrete benefits what extent has the Bank contributed to these (Saunier 1992). But there was little rigorous evi- connections? What is the distributional profile dence regarding these benefits and no attempt at of those taking connections? What are the unit all to quantify them. costs of connection by type of supply to the user and the supplier? The ESMAP study of the Philippines changed that. · What are the direct economic benefits from RE? Although the full range of estimates made in that Who gains these benefits? What are the indirect report, which are discussed below, are quite data economic benefits (employment generation), intensive, it developed techniques for measur- and who gains them? How does the distribu- ing the main benefits from improved lighting and tion of benefits change as coverage of electri- access to television, which have since been used fication programs expands? in a number of appraisal documents. Application · What is the impact of RE on time use, and of these methods has resulted in very respectable what are the welfare implications of these 4 INTRODUCTION changes for health, education, and increased The report combines an over- The ultimate objective of leisure? view of the Bank's RE portfolio this evaluation is to · How does RE affect the quality of health and with an analysis of the impact of calculate private and education services? this lending. It does so using a public rates of return · How do the aggregate private benefits and the theory-based approach, iden- from RE investments. public good benefits compare to the willingness tifying inputs and outputs and to pay? What is the distributional profile of then the outcomes (benefits) from those out- these benefits? puts and who receives them. The impact analy- · What are the private and social rates of return sis is carried out on several levels, relying on the from investments in RE? various survey data mentioned above. Most of this analysis is on single survey cross-sectional The Study Approach data, although panel data are available for two This study adopted a multilayered approach to ad- countries (Ghana and Peru). dress these questions. First, a portfolio review was conducted to identify Bank lending for RE The challenge for most impact evaluations is to since 1980, allowing quantification of the scale of overcome possible selection bias. In the case of this support in both monetary terms and the electrification, selection is very clearly on the number of beneficiaries. Second, 10 country case basis of observables, most notably income and lo- studies were compiled based on a desk review of cation.6 When the selection criteria are observable, Bank documentation and other documents on RE as in this case, then the regression-based ap- to capture the variety of experiences in different proach adopted in this study overcomes selection settings.3 Third, a review of existing evidence on bias. Hence, the regression-based approach is the impacts of RE was carried out. Fourth, analy- largely used to capture the impact of electrifica- sis was made of Demographic and Health Survey tion compared with the counterfactual of no elec- (DHS) data for nine countries to examine the im- trification. Some of the possible benefits examined pacts of RE on health and family planning out- in this study--for example, through media ac- comes.4 In addition, household income and cess to increased health knowledge and improved expenditure surveys for two countries (Ghana health and fertility outcomes--have not been and Peru) were analyzed to examine impacts on previously explored. The report acknowledges rural income generation. Fifth, RE-specific data sets weaknesses in the available data, calling where ap- were examined for Lao PDR, the Philippines, and propriate for more data collection specifically de- Sri Lanka.5 signed to examine these impacts. 5 Chapter 2 Evaluation Highlights · Over the last 15 years the Bank's strategy has increasingly empha- sized poverty and environmental issues. · The objectives of Bank-supported projects have increasingly focused on welfare outcomes. · The design of project components, however, has continued to focus on outputs. · Where the Bank has had a continued presence, it has made a significant contribution to RE. · Project performance has been low relative to the Bank-wide average, mainly because of poor institutional performance. Villagers in Sri Lanka using solar panels to light houses. (Photo from the World Bank Photo Library.) World Bank Lending for Rural Electrification U ntil the 1980s Bank lending in the energy sector was largely to pub- lic sector monopolies for power generation and supply. During the eighties, attention turned to institutional issues, aiming to improve economic efficiency and financial sustainability in the sector by encouraging least-cost planning, marginal-cost pricing, and other practices.1 The Bank also tightened its policies on environ- Bank; this was reflected in the energy sector by mental and resettlement standards and imple- the publication of a second policy paper in 1993, mentation arrangements. These changes were "Energy Efficiency and Conservation in the De- reflected in a power sector support strategy paper veloping World." This focus on the environment in 1983 and the power sector Operations Direc- has deepened over time, marked first by the pub- tive of 1987. lication of Fuel for Thought: An Environmental Strategy for the Energy Sector (World Bank 2000) The Bank's Evolving Energy Strategy and second with "protecting the environment" Two significant shifts occurred in the 1990s. First, being listed as one of the four business lines of there was a shift toward promotion of private Bank energy lending in the document "The World sector participation in power generation and sup- Bank's Energy Program: Poverty Reduction, Sus- ply, as laid out in the policy paper "The World tainability, and Selectivity" (World Bank 2001b). Bank's Role in the Electric Power Sector" (World This focus included programs to promote efficient Bank 1993b). The 2003 IEG report Power for De- energy use, fuel switching, and emissions trading. velopment: A Review of the World Bank Group's Experience with Private Participation in the In accordance with the re-establishment of poverty Electricity Sector reviewed the experience with on the development agenda, the Bank's energy this policy; it concluded that, with the appropri- sector has paid increased attention to the fact ate commitment from government, the expected that the poor have often been left out of the di- benefits had been achieved. However, reform in rect benefits of RE programs (IEG 1994). The 1996 many countries was in the early stages, and re- publication Rural Energy and Development: Im- forms in the distribution sector had lagged behind proving Energy Supplies for Two Billion People those in generation, possibly jeopardizing the lat- (World Bank 1996) set out the links between en- ter. The report also pointed to the need to main- ergy and poverty (operating through the high stream poverty and environmental concerns. economic and health cost of biomass energy sources for poor households) and proposed steps The second shift was increased attention to en- in addition to market liberalization to enhance vironmental issues. The publication of the World rural energy supplies and ensure that the poor Development Report (World Bank 1992) on sus- would benefit. Finally, the 2001 sector board paper tainable development marked a shift across the (World Bank 2001b) moved poverty closer to the 9 THE WELFARE IMPACT OF RURAL ELECTRIFICATION center stage with "helping the poor directly" as the Ghana National Electrification Project, which in- first of four priorities for Bank support to energy stalled distribution systems in both urban and supply. The elaboration of measures related to rural areas. Finally, a project covering several sec- this priority include "gender issues related to ac- tors may include RE, most usually community cess to energy" (p. 23). demand-driven projects in which electricity sup- ply is one option, such as the Brazil Northeastern The Portfolio Rural Poverty Alleviation Program. Bank support to RE is provided through dedi- cated projects, general energy projects with RE A review of the portfolio since 1980 identified 120 components, and multisector projects that in- RE projects,2 with roughly a third falling into clude RE. Examples of dedicated projects include each of the three categories: 42 (35 percent) the three Bangladesh RE projects, which brought were dedicated RE projects, 44 (37 percent) were electricity to more than half a million households larger energy projects with RE components, and between 1982 and 2000, and the two RE projects 34 (28 percent) were multisector projects that in- in Indonesia, which reached more than 10 million cluded RE components. It is not always possible households. to identify the amount of the loan or project budget allocated to RE. A lower estimate is given Since 1980 the Bank has RE components of larger en- by taking the dedicated projects only, amounting financed 120 projects that ergy projects may be very small, to $798.3 million from 1980 to 2006. An upper es- include rural such as conducting a feasibil- timate was reached using the total budget of all electrification. ity study or supporting devel- 120 projects, which comes to $5.97 billion; these opment of a strategy, as in amounts are equivalent to 0.14 and 1.6 percent, the Kenya Geothermal Development and Pre- respectively, of the Bank's total lending over this Investment Project; or they may be large, as in the period. A large and growing proportion of RE projects are in the Latin America and Caribbean and Sub- Figure 2.1: A Growing Number of Rural Saharan Africa Regions, with a falling share in Electrification Projects Are in Latin America South Asia and East Asia; many projects in Latin and Sub-Saharan Africa American and the Caribbean are multisectoral, 25 community-driven development (CDD) projects (see figure 2.1). There are very few RE projects in the Middle East and North Africa and Europe and 20 Central Asia Regions. 15 projects This pattern broadly reflects coverage rates, which of are high in the Regions in which there are fewer 10 projects. However, there are many countries, no- Number tably in Africa, with low coverage rates where the 5 Bank is not supporting RE. And substantial in- vestments are still needed to reach the many 0 millions of unconnected households in South LAC MENA AFR SA EAP ECA Asia. Although there are a number of new energy 1980­1995 1996­2006 projects in African countries--such as Ethiopia, Tanzania, and Uganda--the scale of the Bank's in- Source: IEG portfolio review. vestments in the sector does not match the chal- Note: Figures refer to year of approval. AFR = Africa; EAP = East Asia and Pacific; ECA = Europe and Cen- lenge, suggesting the need for a review of the tral Asia; LAC = Latin America and the Caribbean; MENA = Middle East and North Africa; SA = South Asia. priorities in the Bank's lending program against the availability of funds.3 10 WORLD BANK LENDING FOR RURAL ELECTRIFICATION Objectives Figure 2.2: Increased Energy Supply and Objectives can be broadly classified into four Institutional Development Account for the Largest categories: Share of Objectives (percentage of projects given objectives in each category) · Improved welfare: This includes objectives such as "enhancing Madagascar's prospects 80 for economic recovery and growth by ensuring 70 an adequate supply of electricity in the medium 60 term, for both businesses and households" (Madagascar Energy Sector Development Proj- 50 projects ect) and to "improve welfare, enhance income- of 40 earning capacity, and help alleviate poverty" (Vietnam Rural Energy Project). Included in 30 this category are environmental objectives, Percentage 20 such as "reduce deforestation and increase ac- 10 cess and diversify choice to renewable and cleaner fuels to the household and [small and 0 Improved Increased Institutional Other medium enterprise] sectors" (Benin Energy welfare energy supply development Services Delivery Project). · Improved energy supply: This category in- 1980­95 1996­2006 cludes objectives such as "helping to bring Source: IEG portfolio review. about an improvement in supply and distribu- Note: Categories do not total 100 because each project can have objectives in more than one category. tion of electricity over the medium term" (Guinea-Bissau Energy Project) and "expand rural electricity service in seven central and southern provinces of Lao PDR, where ergy supply (73 percent) and institutional devel- economically justified, through grid extension opment (75 percent), compared with 60 percent and off-grid electrification" (Lao PDR Southern that include improving welfare among their ob- Provinces Rural Electrification Project). jectives. The percentage for the first two cate- · Institutional development: This includes ob- gories has been constant across the study period, jectives such as "strengthen government ca- but there has been a marked increase in projects pacity to implement its national RE strategy" with welfare objectives (figure 2.2; appendix table (Nicaragua Off-Grid Rural Electrification) and B.5). Before 1995 only half of all projects men- "support transformation of [electric coopera- tioned welfare objectives, but since 1995 two- tives] through institutional and operational im- thirds (68 percent) do so; this increase is driven provements" (Philippines Rural Power Project). by the increase in multisectoral projects, which are · Other objectives not related to electrification: more likely to have welfare objectives than are en- These objectives are from the multisectoral ergy projects. projects with RE components. Examples in- clude "increasing the capacity of communes for The objectives were classified into subcategories decentralized and participatory planning and under these four main headings (see appendix management of development activities" (Viet- table B.3 for full details). The most common wel- nam Community-Based Rural Infrastructure) fare objective is to increase growth or incomes, and "foster the sound management of water re- which is found in 30 percent of all projects, fol- sources" (Cape Verde Energy and Water Sector lowed by environmental effects (in 23 percent), Reform and Development Project). reducing poverty (22 percent), and a general statement of improving welfare (21 percent). Analysis of the 120 projects shows that three- However, the poverty-reduction objective is quarters have objectives related to increasing en- mostly associated with multisector projects; just 11 THE WELFARE IMPACT OF RURAL ELECTRIFICATION 7 percent of dedicated RE and energy sector cent of energy projects in the period 1980­95 projects include poverty reduction among their (figure 2.3). objectives. The situation has changed somewhat: from 1997 There has been a marked Increased access is the most to 2006 just under half (48 percent) of appraisal increase in projects that common objective under in- documents for energy projects mention gender, have the objective of creased energy supply, either in although there has been a clear impact on design improving welfare. general (30 percent) or for rural for just 20 percent. More recent projects oriented areas (24 percent). Improving toward providing modern energy services, rather efficiency is also a common objective, mentioned than focusing solely on electricity connections, are for nearly one-third (29 percent) of projects. The the most usual exceptions to the neglect of gen- most common institutional development objec- der. For example, the Energy Access Project in tive is "institutional development," either in gen- Ethiopia includes training and technical assis- eral (34 percent of projects) or for the utility tance for the women to promote sustainable man- company (23 percent), followed by promoting pri- agement and exploitation of woodfuel plantations. vate sector involvement (28 percent). In Uganda, the Energy for Rural Transformation Project has promoted gender-specific TV and Another notable shift is the in- radio communications to raise health awareness. Though the focus on crease in the attention paid to welfare outcomes has gender. Traditionally, energy proj- Multisectoral projects are far more likely to con- increased, a focus on ects have paid no explicit atten- tain a gender aspect, with gender influencing the outputs continues to tion to gender issues--even design of more than half all such projects. In ad- dominate the portfolio. though men and women may dition, the monitoring and evaluation systems of use electricity differently. This such projects are more likely to focus on gender- was true of the appraisal reports for 83 per- specific effects on impact than energy sector proj- Figure 2.3: Gender Issues Are Increasingly Taken into Account but Still Affect the Design of Only a Minority of Energy Projects 20 18 16 14 12 projects of 10 8 Number 6 4 2 0 RE projects Other energy Multisector RE projects Other energy Multisector 1980­95 1996 onward No mention of gender Gender mentioned, but did not influence design Gender influenced design Source: IEG portfolio review. Note: RE = rural electrification. 12 WORLD BANK LENDING FOR RURAL ELECTRIFICATION ects do (with some notable exceptions, such as · Institutional development: This includes, for the Benin Energy Services Delivery Project and example, "operational support, training, and Nepal Power Development Project). technical assistance to the Rural Electrification Agency to enable the agency to carry out the Indeed, evaluations of multisectoral projects have RE program" (Senegal Electricity Services for found benefits from women's involvement in the Rural Areas project) and "development of the selection, prioritization, execution, and operation institutional framework and regulations for and maintenance of subprojects (for example, rural provision of electricity service on and off Peru Second Social Development and Compen- grid ... and capacity building for demand-driven sation Fund Project), although not all projects and decentralized identification, planning and have managed to address gender-related weak- development of projects" (Peru Rural Electri- nesses in project design during implementation fication Project). (for example, the Panama Social Investment Fund · Electrification financing: This includes, for Project). example, "provide medium and long-term fi- nancing to private sector firms, [nongovern- The main conclusion from this discussion is that mental organizations], and cooperatives for projects have historically focused on objectives re- solar home system and village hydro pre-grid lated to outputs (energy supply and institutional electrification, grid-connected mini-hydro development) rather than to outcomes (welfare). schemes and other renewable energy invest- This focus is changing, though not much more ments" (Sri Lanka Energy Services Delivery than one-fifth (22 percent) of projects have ex- Project). plicitly included poverty reduction among their · Other: This category includes components re- objectives. Similarly, a growing number of project lated to consumers, including demand-side appraisals mention gender, but it only has an in- management, and other activities, including fluence on project design in a minority of cases. resettlement. This relatively low percentage may reflect recog- nition that the poor frequently do not benefit di- Not surprisingly, most projects (82 per- Project components and rectly from electrification, but as coverage rates cent) contain infrastructure compo- subcomponents focus on increase, this topic deserves some more atten- nents, and 85 percent have institutional building infrastructure, tion--see chapter 3--especially as directly ben- development components (appendix institutional develop- efiting the poor is now a priority in the Bank's table B.6).4 The most common infra- ment, and electrification energy lending. structure component is grid expansion, financing. present in 63 percent of projects, fol- Project Design: Analysis of Components lowed by renewable energy (in 32 per- The nature of the objectives has implications for cent). Institutional development is most commonly project design. Project design is analyzed by clas- supported by technical assistance for general man- sifying project components and subcomponents, agement (71 percent), engineering (26 percent), which fall under four broad headings: or financial/ commercialization (19 percent). · Building infrastructure: This includes, for ex- The two main developments in the portfolio in the ample, "distribution networks to electrify about last decade have been an increase in projects uti- 120 rural villages and small towns that at pres- lizing renewable energy technologies (RETs), such ent lack electricity service, through extension as solar, wind power, and hydropower, and support of existing transmission and subtransmission fa- to off-grid schemes. Promotion of RETs is in line cilities" (Colombia Village Electrification Project) with the growing emphasis on environmental pro- and "... to support investments in small power tection in the Bank's energy strategy: of the 120 generation, decentralized grids and stand-alone projects, 13 percent of those from 1980 to 1995 uti- RET systems, most notably [photovoltaic] sys- lized RETs, compared with nearly half (46 per- tems" (Philippines Rural Power Project). cent) from 1996 onward (see appendix table B.4).5 13 THE WELFARE IMPACT OF RURAL ELECTRIFICATION Support to off-grid electrification has grown are pilot projects (appendix tables B.10 and B.11). quickly in recent years: there were only two such In such cases the relative budget share is often projects prior to 1995, but 31 since then. Although small. For example, under the Lao PDR Southern off-grid systems in the developing world often Provinces Rural Electrification Project, the off- rely on diesel generators, Bank support for off-grid grid component was 6 percent, compared with 76 connections has been linked to RETs: three- percent for grid expansion, rising to 17 percent quarters of all Bank off-grid projects have pro- under the follow-on Rural Electrification Project.6 moted photovoltaic energy (usually solar home These off-grid components have often attracted systems, SHS), nearly half (47 percent) micro cofinancing from the Global Environment Facil- hydro, and one-third (31 percent) wind power ity (GEF), which has provided $270 million to (see appendix table B.26). Bank-supported RE projects. Where there has been a choice of technologies, Outputs and Outcomes SHSs have been the dominant one. For example, The ratings for RE projects on completion are in Lao PDR, the Bank-financed off-grid program slightly lower than those for other Bank projects.7 provided electricity to 46 villages, all but one of Before 1995, 73 percent of projects were rated as which opted for SHS. In contrast, just two proj- satisfactory (highly satisfactory, satisfactory, or ects (6 percent) promoted diesel power, one of moderately satisfactory), which is comparable to which was the Lao PDR Southern Provinces Rural the 74 percent for the Bank as a whole (appendix Electrification Project. table B.2). But since 1995 the percentage has slipped slightly to 68 percent, whereas the Bank's Off-grid electrification has The increased focus on RETs has been overall performance has improved modestly. Rat- been growing rapidly. driven by two factors. First, the cost of ings are lower for energy sector projects (that is, these technologies has decreased sub- excluding multisector), with the figures for the two stantially since the 1970s, so they have become least periods being 70 and 64 percent, respectively. cost energy solutions, at least in areas difficult to reach with the grid. Projections in Rural Energy The principle reasons given for unsatisfactory rat- and Development: Improving Energy Supplies ings are poor institutional performance; this was for Two Billion People (World Bank 1996) sug- cited for 11 of the 16 projects rated unsatisfactory gested that, if cost reductions for RETs continued, or highly unsatisfactory.8 In some cases institu- then by 2020 they will in general be as cheap as con- tional problems undermine physical implemen- ventional methods of power generation, a con- tation (failure to meet physical targets was a clusion made more likely by the higher price of oil problem in 5 of 16 unsatisfactory projects), but in recent years. Second, during the 1990s the Bank many projects manage to implement their infra- began to take environmental issues more seri- structure components even while the service ously; they now have a prominent position in the provider is ailing. This was so for a number of Bank's energy strategy. cases, such as the first and second Brazil Elec- torbras Power Distribution projects, the Philip- The greater economic feasibility of RETs has also pines Transmission Grid Reinforcement Project, given a boost to off-grid programs. A second rea- and, in Lao PDR, the Provincial Grid Integration son for the growth of off-grid investments is the Project and the follow-on Southern Provinces explicit attention now being given to providing Rural Electrification Project. It is possible to sus- modern energy services to the poor, who live tain service expansion despite weak underlying fi- disproportionately in remote areas beyond the nancial performance when concessional finance reach of the grid. is being used to fund the expansion, but it will not prove sustainable in the long run. Usually support for off-grid activities is a compo- nent of a larger project--this is true for 28 of the So there is a largely positive story to be told re- 33 off-grid projects; in many cases these activities garding the physical achievements of Bank- 14 WORLD BANK LENDING FOR RURAL ELECTRIFICATION supported RE projects, although with an impor- the United States). Better-performing low- and tant caveat regarding technical problems in some middle-income countries achieve similar figures cases. In particular, where the Bank has a contin- (for example, Thailand)--but many lie in the ued presence through a series of dedicated proj- range of between 20 and 50 percent (for example, ects, it can make a substantial contribution to Albania and Rajasthan at the upper end). increasing the country's electrification rate. From 1993 to 2004, coverage in Lao PDR rose from 14 Because system losses are electricity Ratings for RE projects to 45 percent as an additional 325,000 households bought or generated for which no rev- are low relative to other got electricity; Bank-supported projects supplied enue is collected, they represent a loss projects--and are getting 90,000 of these households (28 percent). to the utility, a loss in potential bene- worse. fits to consumers, and a source of ex- In Bangladesh the number of rural connections cess environmental costs. RE programs are likely grew from close to zero in the early 1980s to more prone to these losses--long transmission and than 4 million by 2002 (and is now growing at the distribution lines are most likely to suffer both rate of 700,000 households a year) (Barnes 2007). technical and nontechnical losses, with the prob- The three Bank-supported RE projects financed lem declining as the network gets "filled in." But more than 600,000 of these connections, or 15 per- the problem is also one of poor system con- cent of the total. struction, which should be addressed through better design and supervision. Most notably, the RE rate in Indonesia rose from 33 to 85 percent between 1991 and 2003, as ap- Explicit attention to the issue in Bank projects can proximately 11 million households got new elec- help reduce losses: one-quarter of the Bank's RE tricity connections. Between 1991 and 2000 the projects had either a rehabilitation or system loss first and second Indonesian RE projects brought component (see appendix table B.4). For instance, electricity to more than 10 million households; the the Jordan Energy Development Program Bank's support of a "time slice" of the Indonesian (1984­91) succeeded in reducing transmission program means that Bank finance paid for about line losses from 16 to 12 percent. But not all proj- half of all the new connections in this period. ects have been successful. For instance, in the third Bangladesh RE project (1990­2000), despite re- The scale of Bank support can also be measured duction of system losses being an objective, the by outputs, although these figures are underesti- percentage of system loss stayed nearly steady, 16.8 mates because the data are not available for all percent in 1990 and 16.2 percent in projects. Bank projects have brought electricity to 2000; in the Pakistan RE project, where Where the Bank has had a more than 130,000 villages, reaching nearly 20 mil- reduced system losses was also an ob- continued presence, it has lion households, supplied more than 600,000 kilo- jective, the percentage actually in- made a substantial meters (km) of new lines, and installed more than creased from 23 to 26 percent between contribution to increasing 500 substations (appendix tables B.15 and B.16). 1990 and 2000. countries' RE rates. However, success in increasing access has been In the 1980s and early 1990s many electricity marred by technical problems in some countries. supply agencies were experiencing financial prob- System losses are both technical (electricity "evap- lems, which undermined sustainability, as oper- orates" during transmission and distribution) and ations and maintenance (O&M) was weak and nontechnical, which is a euphemism for theft sometimes hindered implementation of planned through illegal connections. The percentage of grid extension. In some cases these problems nontechnical loss can vary widely between coun- were tackled both through the Bank's adjust- tries--for example, from 3 percent of total system ment lending and the conditionality attached to losses in Vietnam to 30 percent in Mozambique. power sector investment loans. A recent study In developed countries, total system losses are states that 70 countries have undertaken power less than 10 percent (for example, 7 percent in sector reform since the early 1990s (Besant-Jones 15 THE WELFARE IMPACT OF RURAL ELECTRIFICATION 2006), though in several cases it is too early to proved, the necessity of Bank conditionality in this judge the success of these changes (as also sug- area has declined. A Bank study (Komives and oth- gested in the 2003 IEG review). And in some ers 2005) found that current tariff levels were in- cases, provision has been made for private sector sufficient to cover O&M in less than a third (29 participation in generation without addressing percent) of low- and low middle-income countries, underlying structural issues.9 whereas in a quarter (24 percent) they were suf- ficient to cover O&M and make a contribution to But several countries have made progress, which capital cost (see appendix table B.22). is demonstrated by the changing focus of Bank- supported RE projects. There has been a marked But it should also be noted that a more compre- decline in tariff conditionality, that is, require- hensive approach to the issue of financial sustain- ments to raise or review electricity tariff rates, in ability is required, focusing not just on tariff reform, Bank loans since 1995. Bank loans required a re- but also on explicit recognition of the possible view of tariffs in 35 percent of projects need for subsidies (including cross-subsidies) and Financial performance before 1995 and in only 14 percent improving system design and revenue collection. is certainly better than after. The decline in required increased The story in relation to financial performance is thus it was a decade ago, tariff rates is even more striking--from an uneven one: the picture is certainly better than but in many countries 17 percent of projects to 9 percent. As it was at the time of the last IEG review, but in many progress is still needed. utility financial management has im- countries progress is still needed. 16 Chapter 3 Evaluation Highlights · Electrification favors the non-poor, although more of the poor are in- cluded as the grid is extended. · Emphasis has been given to ex- tending the grid to areas where it will cost least and communities can most afford it. · A majority of households that are going to connect do so in the first three years that the grid is available. · Project benefits would be greater if explicit attention were paid to ex- tending the grid to those least able to connect and to ensuring that poor customers use electricity efficiently. Solar panels in Mali provide rural power. (Photo from the World Bank Photo Library.) Who Benefits from Rural Electrification? I t is widely recognized that the immediate benefits of RE seldom go to the poor. IEG's analysis supports the finding that the poor are less likely to have access to electricity. But the analysis also shows that distribution im- proves as coverage expands. The Distribution of Electrification Which Communities Get Electricity? In 1992 in Bangladesh, the poorest 40 percent of Of the 120 projects, information is available for 29 rural households accounted for just 7 percent of on how the communities to be electrified are to all electrified rural households, but this share in- be chosen. These eligibility criteria can be classi- creased to 17 percent by 2004 (see figure 3.1). For fied as follows: Ghana these figures are 5 percent for 1988 and 23 percent for 2003.1 · Cost-effectiveness: Criteria are developed to identify which communities it will be most cost The share of the poor in electricity consumption effective to connect. These criteria typically in- is lower still if the level of consumption is taken clude distance to the existing grid, population into account: although there are substantial vari- size, affordability (average community income), ations by country, the expenditure by the poor on and productive potential. This approach was electricity is typically one-half to two-thirds that widely promoted in the 1960s by the US Agency of the non-poor (Komives and others 2005, annex for International Development (USAID) under B.2). Figure 3.2 plots the share of electricity con- the name the Demand Assessment Model, for sumption against the population share for the example, in El Salvador and Guatemala. The Philippines and Lao PDR. In the former, the bot- Bank used a similar approach later, sometimes tom 40 percent accounted for just 14 percent of adopting earlier work by USAID, such as in the electricity consumption; in the latter, that fig- Bangladesh. For example, the Cambodia Rural ure was 15 percent. Electrification and Transmission Project com- bined most of these factors in its decision that The rural poor are less likely to have grid connec- villages should be within 40 kilometers (km) of tions for two reasons. First, in nearly all countries, the existing grid, be reasonably accessible by communities are ranked by a number of criteria that road, have development potential from agri- usually favor the better-off communities. Second, culture or handicrafts, and be able to pay their within a community connected to the grid, there electricity bills. will be some households that cannot afford to con- nect. Despite the fact that energy expenditures The Pakistan Rural Electrification Project se- are typically less for electrified households, the lected communities with I/K ratios greater than connection fee acts as a barrier, preventing the 24, where I is the population size and K the poorest from switching to the lower-cost source. distance to the nearest medium voltage wire. 19 THE WELFARE IMPACT OF RURAL ELECTRIFICATION Figure 3.1: Pattern of Electrification Favors the Non-Poor, but This Bias Generally Reduces over Time as Electrification Coverage Expands 30 Ghana 25 in 40% 20 Peru Philippines bottom households Nepal 15 the of 10 electrified Bangladesh Share 5 0 0 10 20 30 40 50 60 70 Rural electrification rate The graph plots the share of the poorest 40 percent against the RE rate for five countries at dif- ferent points in time. In all but one case (Nepal), as coverage expands, so does the share of the bottom 40 percent--when there is universal coverage, their share will reach 40 percent. Source: Appendix C. Figure 3.2: Share of Poor of On-Grid Electricity Consumption Is Low (Lorenz curves for rural electricity consumption) Philippines Lao PDR 1.0 1.0 0.9 0.9 0.8 0.8 Grid connection 0.7 0.7 grid grid Connection 0.6 The poorest 40% of the 0.6 share population accounts for 25% of share 0.5 0.5 grid connections and 15% of grid electricity consumption. 0.4 0.4 Grid consumption 0.3 0.3 Cumulative Cumulative connection/consumption connection/consumption 0.2 0.2 Consumption 0.1 0.1 0.0 0.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Cumulative share of population Cumulative share of population Source: Appendix C. 20 WHO BENEFITS FROM RURAL ELECTRIFICATION? Cost-effectiveness allocation rules make com- Box 3.1: Successful RE through a Multisectoral munities with many poor people less likely to CDD Project be connected, sometimes explicitly, as when the Indonesian government targeted those In Ceara, a Brazilian state, more than 1,500 rural communities were elec- within 10 km of the district capital and a poverty trified through a multisectoral CDD project. As part of the project, com- rate of less than 20 percent, compared with munities were directly involved in selecting, preparing, and overseeing the national average of 45 percent (Meier 2001). the implementation of electrification subprojects, which in turn were ex- · Social allocation: The decision rule includes ecuted by private firms contracted by the community associations. Under poverty or other social indicators, giving a pref- this arrangement 91,000 families were connected and power provided for erence to the more deprived areas. One ex- street lighting as well as schools, village shops, and small-scale pro- ample is the eligibility criteria for the North East cessing units at an average cost of about $425 per family. Rural Poverty project in Brazil, which included low socioeconomic indicators, a poor natural Source: World Bank 2001a. resource base, and communities of fewer than 7,500 people. Alternatively, the decision rule may strive for geographical balance, which will also favor areas that would not satisfy strictly grid connection, but the infrastruc- Common eligibility economic criteria. ture is managed by the utility. criteria for connection · Combined allocation rule: This takes into ac- include cost-effectiveness, count both financial viability and social consid- On the other hand, a growing number social allocation, and use erations. Examples include the eligibility criteria of projects have adopted a combined of a combined allocation under the Infrastructure for Territorial Devel- approach to allocation. But this is not rule. opment Project in Chile and the Honduras Rural an unambiguous trend; the case of the Infrastructure Project; both included high Peru Rural Electrification Project is a recent ex- poverty incidence but also productive potential. ample in which there was a strictly cost-efficient The Vietnam Rural Energy Project included af- decision rule, the government favoring this ap- fordability and productive potential but also proach to prevent the political interference that communes in the government's list of the poor- plagued the previous social allocation (box 3.2). est communes and those that had made great contributions to, or suffered from, the war. More socially oriented allocations have been as- sisted for both grid and off-grid connections A cost-effectiveness approach is justified on the through Rural Electrification Funds (REFs), which grounds of financial sustainability. By going first sometimes (though not always) have the inten- to communities that cost the least to reach and tion of subsidizing connections to less-well-off where load factors will be highest, RE does not put communities. The best known fund has been undue strain on the utility's finances. that in Chile (see box 3.3), which was not a Bank- supported initiative. The Bank has supported The minority of projects (17 percent) that do such funds through the Uganda Energy for use a social allocation rule are mostly multisec- Rural Transformation Project and the Nicaragua toral projects, that is, CDD projects targeting Off-grid Rural Electrification Project. But to poor communities that include electrification date, such funds have only been employed in among their possible subprojects. In the Brazil a minority of cases, and sometimes with a dif- Northeast Rural Development Project, 28 per- ferent focus, such as to support private sector cent of communities selected electrification-- development. and an ex post evaluation suggests that this has been successfully implemented with the expected In addition, growing support for off-grid electri- benefits (see box 3.1). Under such projects the fication may favor less-well-off communities, be- community is usually responsible for the cost of cause these projects benefit those that do not 21 THE WELFARE IMPACT OF RURAL ELECTRIFICATION be connected to the grid for at least 10 years are Box 3.2: Selection of Projects under the Peru Rural eligible. Electrification Project But three caveats are needed regarding the The Peru Rural Electrification Project stresses efficient provision of rural poverty focus of off-grid connections. First, the electricity. One means of achieving this end is to change the current ap- scale of off-grid investments remains small com- proach of selecting areas for electrification, which the government was pared with those in grid extensions, so the num- basing largely on social grounds. Under the project, the emphasis is ber of connections from the latter is far greater. being shifted to prioritizing cost-effectiveness by selecting first those Hence, the number of disadvantaged households communities that are near existing distribution systems. The appraisal re- reached through off-grid systems will remain rel- port showed that if communities are selected this way rather than by using atively small. the ordering chosen by government, the $92.4 million subsidy to be pro- vided by the project could finance the electrification of 150,000 households Second, affordability considerations must also be compared with 100,000 under the government's existing program. present for off-grid supplies, particularly as the fa- Source: World Bank 2006. vored model is one of private sector supply. For example, the project appraisal document (PAD) for the Nicaragua Off-Grid Rural Electrification satisfy the criteria for grid connection. This Project explicitly states that the need for com- approach is formalized in the "least cost fron- mercial viability means that the project cannot ex- tier." The smaller the community and the further clusively target the poor. In Lao PDR, communities it is from the existing grid, the more expensive the identified as eligible are subject to an affordabil- grid connections are. Using data on connection ity assessment, with 80 percent of households costs, a cost-effectiveness frontier can be con- having to sign up before an off-grid scheme can structed between grid connection and be introduced. Off-grid solutions may photovoltaic (PV) sources of energy. Var- favor less-well-off ious Bank documents present this analy- Finally, although off grid is least cost for those com- communities. sis, for example, for Brazil, the Philippines, munities receiving it, it costs more than grid ex- Vietnam, and Senegal. tension to other areas, where there may also be concentrations of poor. For the one country for One of these reports states that PV is competitive which comparable data are available (Sri Lanka), for communities larger than 45 households when connections to SHSs are less equitably distrib- the distance to the grid is more than 11.5 km. That uted than are those to the grid (see appendix distance decreases to 6.5 km if systematic PV rural H). Hence, off-grid investments are not necessarily electrification by a regional operator works with the most pro-poor allocation of funds. local nongovernmental organizations (NGOs) that have lower overheads. The Vietnam example The emphasis in RE projects has been on ex- plots the cost of grid extension (which is a func- tending the grid to areas where it will cost least tion of population size and distance) against av- to do so and to communities that can best afford erage consumption: at a typical consumption it. This emphasis can be seen as necessary because level of 30 kilowatt hours (kWh) per month, SHS many electricity utilities were in poor financial is the least cost option if grid extension costs health. Indeed, the report Rural Electrification more that $600 per household. However, these in Asia (IEG 1994) criticized the Bank for failing programming approaches are not adopted in all to consider the financial consequences of RE for cases (and are only explicitly presented in one electricity supply companies and the govern- project appraisal document). Other projects use ments that subsidize them. a more rule-of-thumb approach to identifying communities for off-grid connections; for in- But in some countries circumstances are chang- stance, in Lao PDR those communities that will not ing. The financial situation of utilities has 22 WHO BENEFITS FROM RURAL ELECTRIFICATION? improved, and electricity is now being provided to those communities that meet the cost- Box 3.3: Chile Rural Electrification Fund effectiveness criterion. Hence, social considera- tions are creeping into sector projects through Chile's RE program, launched in 1994, included the creation of a spe- combined allocation rules, and multisector proj- cial REF that links subsidies to output targets. This fund is used to com- ects have shown that RE can be viable even among petitively allocate one-time direct subsidies to private distribution communities selected as being the poorest (see companies to cover part of their investment costs in RE projects. Local box 3.1). operators apply for a subsidy by presenting their proposed project; these in turn are scored against a checklist of objective criteria, including cost- Which Households Get Electricity? benefit analysis, operator investment commitment, and social impact. The second factor behind low connection rates The central government allocates subsidy funds to the regions based on for the poor is that, once electricity becomes the number of unelectrified households and the progress each region has available in a community, the poor may not be able made in RE during the preceding year. to afford the service; high connection charges are a frequent barrier. For example, in Lao PDR an Sources: Jadresic 2000; Tomkins 2001. estimated 30 percent of the population cannot af- ford the $100 connection charge. This pattern is at best only partially overcome by peak (Wp)3 systems, P5000 for 31- to Emphasis has been given the development of off-grid electricity sources. 50-Wp systems, and nothing for sys- to extending the grid to Remote communities are among the poorest tems higher than 50 Wp. areas where it will cost and most expensive to connect to the grid, so they least and communities will be the last to be reached under schemes These cost differentials mean that can most afford it. that set the order in which communities are con- those who can afford to do so connect nected on the basis of cost-effectiveness. Off- to the grid once it becomes available. Analysis of grid sources provide the opportunity to bring data from Lao PDR shows that around 60 percent electricity to these communities. It may be the of households connect within the first year; the case that unit costs in these schemes are lower vast majority of households that will connect do than those of bringing the grid to these com- so in the first three years of the grid reaching the munities (see table 3.1), but they are invariably community (figure 3.3). In the Philippines a higher than the price of electricity for those who smaller percentage connect in the first year but can access the grid.2 still account for half of all those who connect in the first 20 years; the connection rate is 50 per- So the second barrier of cost still prevents many cent after three years, but it has still not reached from accessing off-grid services: in Namibia house- 80 percent after 20 years. In Thailand 25 percent holds must have an annual income of at least of households in electrified villages remained $2,500 to be eligible for an SHS. Off-grid activities unconnected after more than 20 years (Green in Lao PDR, supported by the Bank's Southern 2005). In India 90 percent of villages Provinces Rural Electrification Project and the have electricity, but only 40 percent of High connection charges Rural Electrification Project, undertake an af- rural households have access (ESMAP are a frequent barrier to fordability survey of a village before deciding 2002). connecting the poor. whether to provide services to the community. In some projects this barrier is reduced somewhat So evidence from several countries shows that ex- by tilting the program subsidies to smaller systems tending coverage to the remaining households that are more likely to be chosen by poorer con- takes some years--in communities with electric- sumers. For example, under the Philippines Rural ity for more than 10 years, between 15 and 20 per- Power Project, a P8000 subsidy was provided to cent remain without electricity connections. help meet the connection cost for 20- to 30-watt Countries that are expanding their RE rates are 23 THE WELFARE IMPACT OF RURAL ELECTRIFICATION Table 3.1: Relative Price of Grid, Off-Grid, and Kerosene ($/kWh) for Selected Countries Price ratio Off-grid Off-grid: Kerosene: Kerosene: Grid (SHS) Kerosene Grid Grid Off-grid Indonesia 0.0580 0.0137b 0.0370b n.a. n.a. 2.7 Philippines 0.0075 n.a. 0.3600 n.a. 48.0 n.a. Nicaraguaa 0.0040 0.0350 0.3300 8.75 82.5 9.4 Honduras n.a. 0.0400 0.5000 n.a n.a. 12.5 Bolivia n.a. 0.0400 0.4800 n.a. n.a. 12.0 Mozambique 0.0400 0.0400b 0.1000b n.a. n.a. 2.5 Peru 0.0100 n.a. 0.5700 n.a. 57.0 n.a. Lao PDR 0.0003 n.a. 0.1950 n.a. 650.0 n.a. Senegal n.a. 0.0220b 0.2358b n.a. n.a. 10.7 Malaysia 0.0800 n.a. 0.5800 n.a. 7.25 n.a. Source: Project documents. Note: n.a. = not available. a. Grid is mini grid. b. Cost per klh, not kWh. largely doing so by extending electricity to IEG analyzed data from four countries5 and found previously unconnected communities; it is only that only in the Philippines, where more than when a high proportion of communities are cov- half of the population lives in electrified com- ered that intensive growth takes over.4 munities, does the majority of the increase in Figure 3.3: A Large Proportion of Households Connect to the Grid Immediately after It Becomes Available . . . But Some Remain Unconnected after Many Years Lao PDR Philippines 100 90 . . . then it takes 7 years for the next 10% to connect: 80 80 Another 10% connect in the next two years . . . All households 70 54% connect (%) All households in first year . . . (%) 60 rate 60 rate Poor households 50 40 40 Poor households 30 Electrification Electrification 20 20 10 0 0 0 2 4 6 8 10 12 14 1 3 5 7 9 11 13 15 17 19 21 23 Years since grid connection Years since electrified Source: REP I baseline data. Source: ESMAP 2003. 24 WHO BENEFITS FROM RURAL ELECTRIFICATION? Box 3.4: India's Experience with the Single Point Light Connection Scheme: Kutir Jyoti Under Kutir Jyoti, a social welfare program by the Indian gov- of the state electricity boards. Consequently, utilities became re- ernment for families below the poverty line, India's Rural Electri- luctant to promote RE, and the number of villages being electri- fication Corporation supplied state electricity boards with a full fied dropped from 100,000 during 1985­90 to 11,000 between 1997 subsidy to cover the cost of low-voltage connections for house- and 2002. In response, the government reformulated its RE scheme, holds below the poverty line. More than 5.8 million households in keeping a single point light component whereby free connec- rural areas have benefited, although it has proven difficult for tions would still be available to households below the poverty line, the utilities to sustain this level. The increased kilometers of line but increasing the government's share of the cost burden for new exposed the utilities to the risk of theft and the cost of upkeep, mak- infrastructure to 90 percent; the other 10 percent still fell to the ing the scheme expensive and threatening the financial position state power utility. Sources: Bhattacharyya 2006a, 2006b; http://recindia.nic.in/rggvy.htm. coverage come from intensification (see appen- ginal cost of extensification, the cost A majority of households dix C). But in Bangladesh, Nepal, and Peru the bulk of extending the 20 kV network to an that connect do so in the of increased coverage comes from grid extension un-electrified village" (World Bank first 3 years that the grid to new communities. 1995). Under this project the average becomes available. Even cost per new connection in already after 20 years, some 20 The focus on extensive growth has been deliber- connected villages is a third of that in percent of households are ate policy in some countries. A sector review in In- newly connected villages ($53 per still not connected. donesia that laid the basis for the First and Second household versus $157 per household). Rural Electrification Projects stated, "Initial con- nection rates are assumed to be 33 percent and The appraisal report for the Accelerated Elec- 50 percent of village households for average and tricity Access Rural Expansion Project in Ethiopia above average income villages, respectively. These contains a graph showing how the marginal cost connection rates increase to 60 percent and 75 per- of connection falls rapidly as more households cent, respectively, by the 20th year of electrifica- connect. If tariff levels are sufficient to cover tion" (World Bank 1986). That is, the connection O&M, then the provider will lose little by providing rate will grow slowly over time as incomes rise with these connections. Even if O&M is not covered, growth and electricity becomes affordable to a the government may feel the social benefits war- greater proportion of the village. But even after rant subsidizing these final connections--such 20 years, between 25 and 40 percent of households as the single lightbulb schemes in several Indian in the village will remain unconnected. states (box 3.4). This pattern exists despite the fact that once a vil- An alternative argument is an economic one that lage is electrified, the marginal cost of electrifica- a monopoly supplier should practice price dis- tion of each additional household is low. As crimination to maximize profits, charging a lower explained in the appraisal report for Indonesia's price to those who have a higher elasticity of de- Second Rural Electrification Project: "Given the rel- mand. The problem for the supplier is usually to atively low levels of household electrification today, identify a consumer's "type"--but that is readily the marginal costs of intensification--the incre- done in this case. "Late connectors" are those mental cost of connecting one additional house- who cannot afford the higher connection fee hold within a village that already has access to (and so have a higher elasticity), so a connection electricity--are substantially lower than the mar- tariff differentiated across time from the village 25 THE WELFARE IMPACT OF RURAL ELECTRIFICATION Box 3.5: Overcoming the Connection Cost Barrier Meeting the high cost of connection can be eased by allowing the utility. It is estimated that the five-year plan boosted the take households to spread payments, either by adjusting the tariff to an up in electrified villages by 20 percent and the two-year plan by installment basis or by providing credit for this purpose. Two coun- 10 percent. The Bank has not financed this subsidy, but under the tries have taken the former route. forthcoming Electricity Expansion Project II is seeking funding During the Second Rural Electrification Project in Morocco from the Global Partnership on Output-Based Aid for this purpose. rural consumers were allowed to pay the connection charge in The Electricity Access Rural Expansion in Thailand piloted monthly installments of 40 dirhams over a seven-year period (com- a credit program. Loans were made available one year after the ing to a total of 3,360 dirhams). The Ethiopian Electric Power Cor- village had been connected to the grid, with interest charged at poration introduced a similar program, with the connection charge one percent the local rate paid by the utility. However, the scheme paid over a five-year period (with no interest); it later reduced this was discontinued because it was found not to be sufficiently to two years because of the financial burden of the program for encouraging. being connected would increase the benefits from two other countries--Morocco and Senegal--the the project and the profits to the supplier. connection charge is repaid over a longer period (15­20 years in Morocco). However, few Bank projects have taken this issue on board. Connection costs benefit from a blan- In contrast, in off-grid programs, subsidy schemes ket subsidy, as the charges are rarely sufficient to are common. Although households may have to meet the investment cost. Neither the connection pay some up-front costs, it is not the whole in- charge nor the tariff structure, however, is con- stallation cost, which is typically in the range of structed in such a way as to target these benefits $200­$500 (appendix table B.30). The cost is par- toward the poor (see box 3.5). It is, however, tially subsidized, credit is available, payments are true that the tariff structures of many countries spread over several years, or a combination thereof have contained cross-subsidies from commercial exists. Of 33 projects with off-grid components, or urban domestic consumers to rural customers; 22 specified having subsidies in the project doc- the Cambodia Rural Electrification and Trans- uments. Of those, 86 percent specified an up- mission Project has created an REF that is an ex- front capital cost subsidy, usually declining over plicit cross-subsidy mechanism to help finance RE. time and using an output-based aid approach But at the same time, in neighboring Lao PDR, the (see appendix table B.31). Bank urged the government to increase real tar- iffs most rapidly for the lowest "lifeline" rate so it Seven projects employed credit support facilities would reach cost recovery levels. for off-grid energy. For example, in Indonesia the SHS program provided credit to enable private However, Bank staff in Lao PDR are now explicitly providers to offer their customers the option of considering the issue of late connectors. Similarly, spreading out the cost of the SHS over several the Ethiopia Accelerated Access (Rural) Expan- years. In Sri Lanka, the Renewable Energy for sion Project includes a study of possible connec- Rural Economic Development Project also pro- tion subsidies for rural areas. An alternative to an vided credit to solar dealers and microfinance or- outright subsidy is a loan to meet connection ganizations but did so in conjunction with a charges: two Bank projects--Thailand Second subsidy phased out over five years. Rural Electrification and Ethiopia Accelerated Ac- cess (Rural) Expansion--provided credit to rural Likewise, Bank projects in both Nicaragua and consumers to meet the $98 connection fee. In Honduras employ a combination of microfinance 26 WHO BENEFITS FROM RURAL ELECTRIFICATION? and subsidies to reach the most remote users. Fi- Box 3.6: Poor Communication of Tariff Structures nally, in Lao PDR, there is an up-front payment of Can Disadvantage the Poor around $50, but most installation charges are spread across monthly payments of $1 over 10 The Zanzibar State Fuel and Power Corporation applies a flat rate tariff years. up to 50 kWh per month. However, many consumers are unaware of this, The Distribution of Benefits partly as their monthly bills vary because of irregular meter readings. On from Electrification average, villagers consume only 25 kWh per month, even though they could double their consumption and not pay any more. One villager decreased Because consumption patterns favor the better off, his electricity usage to just 3 kWh a month--equivalent to burning one subsidies to electricity providers also go dispro- lightbulb for 1.5 hours per day--in a futile attempt to save money. portionately to the better off. Evidence from a In Tambo, South Africa, consumers had a choice between a connec- number of national-level studies shows that elec- tion fee of 200 rand (R) and a metered charge per kilowatt hour, or a tricity subsidies are invariably less well distrib- lower connection fee of R10 and a fixed monthly charge of R15. Given ac- uted than a random allocation of funds would tual consumption levels, most households would have been better off tak- be, though performance improves as coverage in- ing the first option, but most opted for the second because they could not creases and can be improved through geographic afford the R200 connection charge and were not sure how much they would targeting or means testing in the subsidy scheme. use. To make matters worse, many low-income consumers cannot always However, connection subsidies perform much afford the R15 a month and so are disconnected and have to pay the R10 better, having a positive distributional impact. again to be reconnected. Apparently progressive tariff structures may ac- Sources: Winther 2005; James 1997. tually mean the poor pay more per kilowatt hour if there is a minimum monthly payment. The poor also end up paying more because they are more likely to be disconnected and subsequently face reconnection charges, especially as the constant in the utility's load shape. For example, Some progressive tariff monthly payment does not match the seasonal the Mali Household Energy and Uni- structures actually mean fluctuations in rural income. Payment problems are versal Access Project sought to pro- the poor pay more per exacerbated if tariff structures are not transparent mote use of low-energy consumption kilowatt hour. or are improperly understood, so consumers may lamps and energy-efficient air coolers make poor choices or unnecessarily reduce their at the household level to reduce peak consumption (see the examples from South Africa hour power use and lower electricity and Zanzibar in box 3.6)--a problem exacerbated bills. In addition, the project included grassroots by bills that are complicated even for those who information campaigns to raise awareness about are literate and numerate. efficient energy use. Another example is the Viet- nam System Efficiency, Equitization, and Renew- These examples illustrate the importance of con- ables Project, which sought to achieve system sumer education that will both stimulate demand peak reduction of 120 megawatts by implement- and ensure that consumers derive maximum ben- ing several DSM measures, including promotion efit at least cost, which also, of course, increases of energy-efficient lamps and time-use meters for the return to the project. Such issues have typi- large and medium-size customers. cally been ignored in Bank projects, though they have begun to emerge in recent years in demand- Concluding Comment side management (DSM) components. The direct benefits of RE programs have tradi- tionally gone to the non-poor. This continues to DSM comprises activities designed to influence the be the case, but the poor gain a greater share of customer's timing and amount of electricity use benefits as coverage increases. The distribution of in a way that will simultaneously increase cus- benefits is affected both by the manner of se- tomer satisfaction and produce beneficial changes lecting communities to be electrified and by the 27 THE WELFARE IMPACT OF RURAL ELECTRIFICATION connection cost barrier preventing poor house- as programs become established with a secure holds in electrified villages from connecting. financial footing, then smart subsidies can be used--including funds to subsidize connec- Because RE programs have historically been a fi- tions to more remote communities and connec- nancial burden on utility companies, strategies tion charge subsidies for late connectors--to such as identifying the most cost-effective ex- increase the volume of benefits and improve their pansion pattern help relieve this burden. However, distribution. 28 Chapter 4 Evaluation Highlights · Most connections in rural areas are residential. · The most common uses of electric- ity are lighting and television; there is some resistance to using elec- tricity for cooking. · Electrification has beneficial impacts for clinics and for the attraction and retention of skilled staff in schools and health centers. · Although some countries have seen greater productive use of RE, it has not led to industrial rural development. Rural home powered by Sri Lankan village hydroelectric scheme. (Photo from the World Bank Photo Library.) What Is Electricity Used for in Rural Areas? T he energy ladder refers to the phenomenon of households and firms-- and so, in aggregate, countries--shifting from low-efficiency fuels to high-efficiency ones as income per capita increases. Biomass fuels such as dung and fuelwood are at the bottom of the energy ladder and elec- tricity at the top. Electricity should be differentiated by how it is gen- and restaurants), or they can be for The poor are the least erated, because, for example, solar and hydro- public and social facilities and for street able to take advantage of power are less polluting than thermally generated lighting. Bank projects nearly always moving up the energy electricity. Figure 4.1 (panel A) shows the energy support residential connections, with ladder from biofuels to ladder from cross-country data; the greater re- a small number of exceptions being electricity because of the liance of poorer countries on biomass fuels and, focused on agricultural connections. Of connection charge conversely, the greater amount of electricity use the 120 projects reviewed, end uses are barrier. per capita in higher-income countries are evident. identified in the project documents The same point applies intracountry, as shown in for 75, and all but three of these include resi- figure 4.1 (panel B), whereby the non-poor are less dential connections (the three exceptions are reliant on inefficient fuels than are the poor. Mexico Renewable Energy for Agriculture, Pak- istan Private Tubewell Development, and Brazil Moving up the energy ladder has implications for Irrigation Subsector Project). However, only a the benefits of RE. More efficient fuels pollute minority (11 percent) of the 72 projects under- less. Moreover, for basic fuels, the transforma- taking residential connections are restricted to tion of matter into energy takes place in or near those end uses. Most projects also seek to sup- the home, so the pollution is not only more con- ply business, agriculture, and so on; the pro- centrated but also nearer to the user. Moving up portion of agricultural connections has fallen the energy ladder therefore has positive envi- over time, as these were a strong feature of Asian ronmental and health effects. projects whose share in the portfolio is falling (figure 4.2). In addition, more efficient fuels are cheaper per unit of energy consumed. Ironically, however, the However, in terms of the number of connections, poor are least able to afford these cheaper fuels it is residential connections that usually domi- because of the connection charge barrier. But nate. Data from selected projects (appendix table even for the better off, moving up the energy lad- B.22) show that more than 95 percent of con- der takes time and is differentiated by end use. nections are residential. For example, under the Tunisia Fourth Power Project, the RE component What Types of Connection Are There? was to make 35,000 new domestic connections Connections can be residential, industrial, agri- and to connect 1,500 water pumping stations and cultural (usually irrigation), or commercial (retail 50 small industrial or commercial customers. 31 THE WELFARE IMPACT OF RURAL ELECTRIFICATION Figure 4.1: The Energy Ladder (A) Poor countries use less-efficient biomass fuels, whereas richer ones rely (B) The poor are lower down the energy ladder (ratio of consumption of the more on electricity. non-poor to poor). 30 2.0 6.0 1.8 capita) 25 5.0 1.6 per capita) 1.4 20 per 4.0 (MWh 1.2 non-poor:poor 15 1.0 meters 3.0 0.8 10 (cubic 2.0 0.6 consumption consumption, 0.4 5 oodfuel Ratio1.0 0.2 W Electricity 0 0.0 0 4 5 6 7 8 9 10 11 Biomass Diesel/kerosene Electricity Income per capita (logged) Energy source Electricity per capita Electricity (fitted) Lao PDR Philippines Woodfuel per capita Woodfuel (fitted) Sources: World Development Indicators, FAO statistics. Source: IEG analysis of survey data. Note: Mwh = Megawatt hours. Figure 4.2: Nearly All Projects Provide Residential Connections, but also Other Connections for Productive Purposes 100 90 80 70 60 project of 50 40 Percent 30 20 10 0 Residential Agricultural Commercial SME Public Type of connection Before 1995 1995 and after Source: IEG portfolio review. Note: SME = small and medium enterprise. 32 WHAT IS ELECTRICITY USED FOR IN RURAL AREAS? In contrast, nonresidential users typically demand spillover is context specific and cannot The use of electricity also more energy: data presented for the Second In- be taken for granted. IEG analysis of progresses up a ladder, donesian Rural Electrification Project show non- DHS data for nine countries showed that from lighting through residential customers on Java consume on average though 60 percent of women in a house- television to other small nine times as much as residential customers (417 hold with a television watched almost appliances and then to kWh per month versus 45 kWh per month)--sug- every day, this was so for only 10 percent refrigeration, cooking, gesting that residential consumption is two-thirds of women in houses without their own and air conditioning. of total consumption. Data reported for the Thai- TV (appendix figure D.2). land Second Rural Electrification Project show res- idential consumption to be 78 percent of the total. Electricity is used for cooking only in a small mi- nority of homes, less than 1 percent in most coun- Domestic Uses of Electricity tries (appendix table D.2). Asian countries are a There is also an energy ladder in the pattern of partial exception, because rice cookers are a com- electricity use. The bottom rung is lighting, the mon acquisition in electrified homes, thus partially basic purpose to which RE is put in all homes. Elec- displacing the demand for wood or other cook- tricity provides more and better lighting at lower ing fuels. Refrigerator ownership is low, with fewer cost than the next available alternative, kerosene than one in five electrified homes having one; that lamps, for most households. proportion is higher in middle-income countries. The next most common use of electricity is tele- This pattern of end use is evident in the pattern vision: on average, close to half of all electrified of energy use found in energy surveys. In Lao homes in rural areas have a television (see ap- PDR lighting accounts for 66 percent of electric- pendix D, figure D.1.B). A similar proportion of ity use among households connected to the grid; all homes have radios, ownership of radios not but that figure is only 24 percent in the Philippines, being related to grid connection because they which has been connected longer.2 This share is can be powered by batteries. However, grid elec- higher for the poor, reaching 72 percent in Lao tricity is cheaper than battery power, so house- PDR. The most important household item in the holds connected to the grid are likely to listen for Philippines is TV, accounting for on average 39 per- longer. For example, in the Philippines those con- cent of electricity consumption; it is second most nected to the grid used the radio for 105 hours a important in Lao PDR, where it accounts for 14 month, compared with just 13 hours for non- percent (see figure 4.3). electrified households. The finding that electricity is not used for cooking Television may be viewed at home, in the home comes from a wide range of data in a variety of of a friend or relative, or in a public place such as countries (see appendix D). One reason a bar. In some cases, the government may use mo- is cost: households are conscious of Resistance to the adoption bile TVs for public education. Viewing television the rapidly spinning wheel of the elec- of electricity for cooking outside the home is most common for teenage tricity meter if a heating ring is turned is partly economic and children, and it is more common for men than for on. But tradition also plays a part: peo- partly social. Overcoming women. Women's ability to watch elsewhere is lim- ple say they prefer the taste of food that resistance requires ited by restrictions on their mobility, especially in cooked over wood or charcoal. In other consumer education. the evening, and especially in conservative soci- places, cooking with electricity is said to eties (such as Zanzibar and Bangladesh)1 and by be dangerous, either because poor connections husbands feeling shamed because they cannot mean that it actually is or because of fear of spir- provide for their family if women have to go to an- its (angering the kitchen gods in Lao PDR) or other's house to watch (Mensah-Kutin n.d.), writ- witchcraft (neighbors giving the evil eye to those ing about Ghana. acquiring electrical appliances in Zanzibar).3 So although in principle any benefits from television That there is some resistance to adopting new may spill over to nonelectrified households, such a electricity-based technologies is not unique to 33 THE WELFARE IMPACT OF RURAL ELECTRIFICATION Figure 4.3: Pattern of Consumption in Rural Box 4.1: The Cold Chain Households (distribution total kWh) 65 Vaccines are sensitive to both heat and cold and so need to be kept between 2° and 8°C from the point of 55 manufacture to the point of use. The system for doing 45 this is called the cold chain. To maintain a store of vaccine, health clinics need 35 month refrigerators, which are most easily and cheaply op- per 25 erated by electricity but may also run on gas or, less kWh reliably, on kerosene. A cold box can keep vaccines at 15 the required temperature for between two and seven 5 days and is needed in case of interrupted power sup- ply or equipment maintenance. They can also be used ­5 to supply health posts without refrigeration, provided Lao PDR Philippines the vaccine is to be used immediately, as would be the Lighting TV Fan Rice cooker Other case for a National Immunization Day. Source: IEG analysis of survey data. Note: kWh = kilowatt hour. than in those without electricity. But the propor- developing countries. The mass RE programs of tion of clinics offering immunization services did the 1930s in the United States and the United not differ between the two groups. Kingdom both ran into substantial resistance, based on opposition to "urban values"--beliefs So RE can help bring down the cost of providing about possible harmful effects and conflicts over immunization services and be a part of the rou- land use (Kline 2002; Luckin 1990). Consumer ed- tine services offered by a clinic. But it does not in- ucation regarding electrification, discussed in crease the extent to which such services are chapter 3, can help overcome this resistance and offered or (for the one case where the causal so increase program benefits. chain could be followed that far) the immuniza- tion rate. Community Uses Health facilities can benefit directly from RE in two Of course these benefits are not dependent on the ways: by having longer opening hours and by grid, or even a dedicated RE project. For example, having equipment that requires electricity. These the Bank-supported social fund in Zambia provides links have not been subject to previous empirical solar power to all clinics it constructs. The same investigation. IEG analysis of health facility survey project also constructs a house with solar power data for two countries--Bangladesh and Kenya-- for the health worker, which points to an under- found that electrified clinics are indeed open for, appreciated benefit from RE: the positive effect it on average, one hour longer each day. has on attracting and retaining skilled staff. Two studies--one of health workers in Bangladesh The most commonly claimed benefit for and one of teachers in Ghana--find a clear and sig- Electrified clinics are health clinics for electrification is that nificant link from electrification to reduced ab- open longer and, with it helps preserve the cold chain (see box senteeism (see box 4.2) (Chaudhury and Hammer refrigeration, can offer 4.1) for vaccines; again, this claim has 2003; IEG 2004). less-costly immunization; not been tested. IEG examined the data this does not necessarily for six countries (appendix table D.5) Productive Uses lead to a higher and found that the cold chain was sig- The data above show that the vast majority of rural immunization rate. nificantly stronger in electrified clinics connections are residential, though some pro- 34 WHAT IS ELECTRICITY USED FOR IN RURAL AREAS? grams have had a stronger productive focus than Box 4.2: Electrification and Worker others. The most notable example is India, whose Absenteeism in the Social Sector RE program was strongly linked to the promotion of high-yield varieties of crops and the spread of Two studies provide evidence of how RE can reduce irrigated agriculture, facilitated by electric water absenteeism in the social sector. pumps with subsidized or free electricity. Surprise visits to health facilities in Bangladesh were used to collect data on absenteeism. It was Support for industrial development has been lim- found that health workers were significantly more ited, even in countries that do have rural-based likely to live in the same community as the facility if the industries. During the 1980s Sri Lanka began a rate of electrification was greater--and that living lo- phase of rapid growth of garment and textile ex- cally greatly reduced the probability of the worker ports. In the early 1990s, the 300 Factories Pro- being absent from the facility for the whole day. gram sought to spread the employment and IEG's impact study of support to basic education in income benefits from the export sector to rural Ghana estimated a multivariate model of the determi- areas by providing incentives for producers to nants of teacher absenteeism. A teacher's living con- locate in rural areas. Despite the relatively high ditions, including whether his or her home had level of RE in Sri Lanka, these factories continue electricity, affected the incidence of absenteeism and to use their own diesel generators because the teacher morale. During fieldwork for the study, the electric power supply is not sufficiently reliable. chair of the school committee in a community without A similar story can be told about Bangladesh, electricity--where only one of the four teachers allo- which has seen rapid growth in textile production, cated to the school had taken up his post--com- with many factories in semirural locations. But plained: "What teacher will come here and live in a these manufacturers again rely on their own place with no electricity?" power supply rather than trusting the grid. So it cannot be argued that RE led, or even facilitated, Sources: Chaudhury and Hammer 2003; IEG 2004. industrial rural development. Small-scale enterprises, including home busi- nesses, are more plausibly influenced by the avail- (the load factor is the ratio of average consump- ability of electricity. As shown in the next chapter, tion to the total possible consumption). From a fi- there is evidence of RE increasing both the num- nancial point of view it is preferable that demand ber of businesses and the hours they are open. be evenly spread throughout the day (because in- stalled capacity has to meet maximum demand, but Relative lack of productive uses means that elec- is idle for much of the time with a low load factor). tricity consumption is heavily concentrated in the The financial viability of RE is therefore linked to peak evening hours, resulting in a low load factor promoting productive uses. 35 Chapter 5 Evaluation Highlights · Connections to RE lower the price of operating lighting and TV. · The impacts of RE on indoor air qual- ity, health, and knowledge, and fer- tility reduction are quantifiable and significant. · RE has some long-term impact on home businesses. · Off-grid solutions have demonstra- ble environmental benefits. · Willingness to pay is high and exceeds the average supply cost where grid connection is feasible. · Reducing consumption by high-end users through higher tariffs can have a net welfare benefit. · Off-grid investments usually have a lower rate of return than grid exten- sion because the costs are more and the benefits less. A Sri Lankan village shop at dusk, lit by solar lamps. (Photo from the World Bank Photo Library.) The Benefits of Rural Electrification T he main domestic uses of electricity are lighting and TV. In the Bank's economic analysis, the valuation of the benefits of lighting have typi- cally been based on the willingness to pay (WTP), which is calculated on the basis of the cost of lighting using the existing source, usually a kerosene lamp. Domestic Uses: Lighting and TV consumer surplus associated with the new con- Older analyses instead compared the cost of kilo- sumption, Qe ­ Qk. watt hours from a diesel generator with that from the grid. This approach changed with the ESMAP The benefit to the consumer is B + C. It is com- study of the Philippines, which instead measured mon in project analysis to also include areas D and lighting as lumens consumed (see box 5.1). E, sometimes referring to the whole area B + C + D + E as WTP; that is not quite cor- The approach is illustrated in figure 5.1, which rect, as WTP includes area A also. It is Electricity supply reduces shows the demand for lumens. Electricity supply perfectly acceptable to include areas D the cost of energy to the lowers the cost of energy to the user, resulting in and E. This is the amount paid by the user. an increase in consumer surplus, which is the consumer, which is simply a transfer difference between what consumers are willing to payment to the utility and so a neutral flow for eco- pay and what they actually do pay. Data from an nomic analysis.1 The cost side of the analysis will energy survey give two points on the demand capture the cost of consumption. Assuming that curve: price of lumens and the quantity con- the average cost of supply (Ce) is less than tariff sumed using either kerosene as the source (Pk,Qk) rate, there will be a positive producer surplus, or electricity (Pe,Qe). Using these two points, the which is being captured in this calculation (figure demand curve can be interpolated. 5.2). The alternative is to deduct the payments (D + E) from consumers and add them to pro- The amount the consumer is willing to pay for a ducers, so when summing across all flows these quantity Q is the area under the demand curve payments/receipts cancel out. from 0 to Q. Hence, the consumer is willing to pay A + B + D for consumption of Qk, but actually The calculation of WTP clearly depends on the pays just B + D (= Pk Qk), leaving a consumer sur- shape of the demand curve, which determines the plus of A. Once electricity becomes available, the extent of area C (see appendix H for more dis- consumer surplus is A + B + C, so the increase cussion). The evidence base for the shape of this in consumer surplus as a result of electrification curve is still thin. Currently the most satisfactory is B + C. This consumer surplus has two parts: approach is to take two points on the curve, as in that arising from the reduction in the price of figure 5.1, and assume a constant elasticity (or the Qk units already being consumed and the log linear) demand curve. This approach has been 39 THE WELFARE IMPACT OF RURAL ELECTRIFICATION A selection of results is reported in table 5.1, show- Box 5.1: Shedding Light on Lumens ing an average household WTP of $9­16 a month. These figures are comparable for both grid and off- grid sources. The latter might be expected to be A lumen is a measure of light emitted: a candle emits around 12 lumens, lower, given that the level of service is lower; the a kerosene lamp from 30 to 80 lumens, and a 60-watt lightbulb 730 lumens. similar size of the WTP estimates partly reflects the So by using a single 60-watt lightbulb for four hours a day for one month means by which the demand curve is estimated.2 (30 days), a household is consuming 88 kilolumen hours (klh) (=4 x 30 x 730/1,000). Electricity consumption is 7.2 kWh per month (=4 x 60 x 30/1,000). Electrification projects are among those that still Suppose electricity costs the consumer $0.05 per kWh; then she has routinely report an ERR. More than 80 percent of a monthly lighting bill of $0.36, equivalent to a cost of $0.004 per klh. In con- the 120 projects did so, and most of the excep- trast, burning a kerosene lamp for four hours a night yields just 6 lumens tions were multisector projects. but costs about the same as the monthly electricity bill, giving a cost of $0.06 per klh. Moving from kerosene to electricity cuts the cost by more IEG examined the ERR calculations for 13 proj- than a tenth and increases consumption more than tenfold. ects.3 The following approaches were found to be used: IEG found four used in a number of Bank studies, such · Estimate WTP assuming a nonlinear demand approaches to as the PAD for Peru Rural Electrification. curve or a linear demand curve but taking only consumption used in ERR Some more recent studies rely on more a percentage of the estimate for area C to allow calculations. points on the curve (where a range of for overestimation. This approach conforms fuel options is available, say kerosene, to best practice. It was used in 5 of the 13 cases car batteries, and electricity) to estimate a kinked examined, including Peru Rural Electrification demand curve. However, other studies have as- and Senegal Electricity Services for Rural Areas sumed a linear demand curve, which results in an Projects. overestimation of WTP and so of total project · Estimate WTP assuming a linear demand curve. benefits, sometimes by a substantial magnitude This approach results in an overestimate of (see appendix H). project benefits. In one case the ERR fell from Figure 5.1: Consumer Surplus Figure 5.2: Producer Surplus 12 12 10 10 8 8 Pk A 6 6 Price Price B C 4 4 Pe Pe Producer surplus 2 2 D E Ce Cost of production 0 0 0 Qk 20 40 60 80 Qe 100 120 140 0 20 40 60 80 Qe 100 120 140 Quantity Quantity Note: Pe = price of electricity from the grid; Pk = price of kerosene; Qe = quantity of electricity Note: Ce = average cost of supply; Pe = price of electricity from the grid; Qe = quantity of elec- used from the grid; Qk = quantity of kerosene consumed. tricity used from the grid. 40 THE BENEFITS OF RURAL ELECTRIFICATION Table 5.1: Willingness to Pay Calculation for Lighting Grid Off-grid Lao PDR Peru Philippines Indonesia Bolivia Honduras Mozambique Nicaragua Quantity (lumen hours/month) Nonelectricity 20 4.6 4.1 8.8 7.0 5.5 48.7 2.4 Electricity 435 363 204 38 90 115 122 125 Price ($/lumen hour) Nonelectricity 0.20 0.57 0.36 0.55 0.48 0.50 0.10 0.33 Electricity 0.00 0.01 0.01 0.10 0.04 0.04 0.04 0.04 Expenditure ($/month) Nonelectricity 3.90 2.62 1.48 4.80 3.36 2.75 4.87 0.78 Electricity 1.31 3.63 1.53 3.81 3.60 4.60 4.87 4.36 WTP Total 11.20 16.16 7.36 11.08 12.24 13.68 9.73 9.01 Per klh 0.03 0.04 0.04 0.29 0.14 0.12 0.08 0.07 Per kWh 0.81 1.11 0.47 0.71 3.02 3.37 2.06 1.90 Average kWh 13.81 14.50 15.50 15.50 4.06 4.06 4.74 4.74 Sources: Calculations based on figures in project documents. Note: klh = kilolumen hours; kWh = kilowatt hour; WTP = willingness to pay. 60 percent in the project Implementation Com- There has been a change in appraisal methods pletion Report to just 12 percent when the over time, with more recent studies adopting the correct approach was used (and for the off-grid ESMAP approach. However, this approach has component of that project the ERR fell from 26 been unevenly applied, with the understanding percent to 1.5 percent); this approach was also of the approach among some task managers being used in 5 of the 13 cases. weak.4 There is more than one case of inappro- · Estimate WTP based on the alternative energy priate application of the approach, resulting in an source, and then value the whole of expected overestimation of project benefits. This suggests energy consumption with electricity at that a failing of the quality control mechanism of re- level. This approach neglects the downward ports that go to the Board--indeed, many proj- sloping demand curve, resulting in a substan- ect documents contain insufficient information for tial overestimate of project benefits. The only the analysis to be replicated, though these data example found was Peru Rural Electrification, are sometimes available in separate documents. but the analysis also underestimated benefits, as costs were double counted. But lighting is only one use of electricity, albeit the · Estimate benefits as the cost savings on cur- main one from a domestic perspective. If other rent consumption levels (that is, area B in fig- benefits are not also captured, then the return to ure H.1). This approach underestimates project the project might be underestimated. This has in- benefits because it ignores additional consumer deed been the case for many projects. surplus from new consumption (area C). Sev- eral older projects used this approach, such as Recent studies use the lumen-based approach to Malawi Power V. value the WTP per kilowatt hour and value all sales 41 THE WELFARE IMPACT OF RURAL ELECTRIFICATION Table 5.2: Willingness to Pay Calculation for TV Lao PDR Mozambique Peru Philippines Modern supply source Grid SHS Grid Grid Q units hours hours kWh hours Quantity: car battery 55 36.5 4.2 41.3 Quantity: electricity 106 91.25 11 129 Price: car battery 0.032 0.050 1.04 0.22 Price: electricity 0.001 0.025 0.18 0.0125 Exp(cb) 1.78 1.83 4.37 9.09 Exp(e) 0.11 2.28 1.98 1.61 Elasticity ­5.30 ­0.76 ­1.82 ­2.52 Intercept 17.79 ­0.27 2.65 7.86 Ba 1.72 0.91 3.61 8.57 Ca 0.39 1.87 2.91 4.92 Total 2.11 2.79 6.52 13.49 WTP Total Per month 2.22 5.07 8.5 15.11 Per hour 0.021 0.056 0.06 0.12 Per kWh 0.26 0.69 0.77 1.46 Sources: Calculations based on figures in project documents. Note: TV uses 80 watts per hour (that is, 0.08 kWh per hour). a. See figure 5.1. The price of operating a at that amount. That is, all sales are val- is $0.77 per kWh for TV usage, compared to $1.11 TV is lower and the ued at the WTP for electricity for light- per kWh for lighting--the WTP for a kilowatt amount of usage is higher ing. Such an approach may yield either hour for lighting is almost twice that for TV). But with a grid connection. an underestimate or an overestimate, for the Philippines the WTP for TV is higher than depending on the WTP for other end for lighting. uses. Where possible, it is preferable to measure the benefits from these end uses separately. Health Benefits The health benefits from RE operate through a An approach similar to that based on the demand number of channels: curve can be used to calculate the benefits from television, where the unit of consumption is hours · Improvements to health facilities of television watched per month. In the absence · Better health from cleaner air as households re- of a grid connection, TVs are operated using car duce use of polluting fuels for cooking, light- batteries, which will provide the price and quan- ing, and heating (Hutton and others 2006) tity against which to compare grid electricity. As · Improved health knowledge through increased with lighting, the price is lower and the quantity access to television higher with a grid connection (see table 5.2), al- · Better nutrition from improved knowledge though the change in quantity is not as and storage facilities from refrigeration. Household electrification marked as with lighting. was found to have a Each of these benefits is explored here. The find- significant impact on For two of the cases, the WTP for a ings support the view that there are health ben- health outcomes in kilowatt hour is greater for lighting efits from RE, including fertility reduction (next Bangladesh. than for TV (for example, in Peru WTP section)--but the survey instruments were not de- 42 THE BENEFITS OF RURAL ELECTRIFICATION signed with the intention of examining these is- sues and so should be seen as suggestive; further Box 5.2: The Health Risk of Candles research and evaluations are needed to strengthen the evidence base. The health risks from candles have only recently been appreciated, since a 1999 Australian study showed that the lead used in candle wicks results Indoor air quality in air lead concentrations at levels far in excess of established safety stan- The use of traditional solid fuels such as fuel- dards. Burning a candle for a few hours in an enclosed room results in wood crop residue and dung exposes people-- lead concentrations sufficient to cause fetal damage or to harm the men- especially women and young children--to indoor tal development of children. air pollution, with consequent health risks: prin- Since the Australian study was done, many developed countries have cipally acute lower respiratory infections, but also banned the use of lead in wicks, but these bans do not affect candles made low birth weight, infant mortality, and pulmonary for developing county markets. tuberculosis. A review of existing studies showed that exposure to indoor cooking using traditional methods increased the risk of premature death by a factor between two and five. These diseases adult work days average 3 per year, and the ad- caused by indoor air pollution cause between 1.6 ditional under-five mortality is 2.2 per 1,000. So and 2 million excess deaths each year,5 more than substituting electric lighting for kerosene lamps half of them among children younger than five. has a quantifiable health benefit of $2.50 per This figure accounts for 2.7 percent of the global household. burden of disease. There is also a fire risk. In ad- dition, fuel collection imposes a costly time bur- A second, relatively unrecognized Where RE can help is in den of up to 8 hours a week (appendix D), once health benefit from RE comes from the replacement of again usually mainly on women. displacing candles (see box 5.2). How- kerosene lamps with ever, candles are not used for lighting electric lamps, a change In principle, RE can tackle both of these issues, pro- that much. They do have nonlighting that reduces indoor air moting better health through reduced indoor air uses, such as for ceremonial pur- pollution. pollution and reducing the time burden on women poses--but these are not affected by of fuel collection. However, in practice, these ben- electrification, so these effects are not captured efits have been little realized because (as shown in this study. in the previous chapter) electricity is largely not used for cooking in rural areas.6 Knowledge and fertility reduction IEG's impact evaluation of health outcomes in But improvements in indoor air quality can also Bangladesh (IEG 2005) found a significant impact come about through changes in lighting source. of household electrification on mortality. One pos- Kerosene lamps emit particles that cause air pol- sible channel for this effect is that access to media lution; these are measured by the concentration improves health knowledge. Chapter 4 demon- of the smallest particles per cubic meter (PM10). strated the link between electrification and access Burning a liter of kerosene emits PM51 micro- to television. IEG analyzed DHS data for eight grams per hour, which is just above the World countries to examine how access to media (radio, Health Organization 24-hour mean standard of TV, and newspapers) affects women's health knowl- PM10 of 50 micrograms per cubic meter. But edge (see appendix G for full details). these particles do not disperse, so burning a lamp for four hours can result in concentrations several The causal chain for the first possible health im- times the World Health Organization standard. pact is as follows: The extra risk of respiratory sickness from expo- · Access to electricity increases time spent watch- sure to these levels of PM10 is captured in the haz- ing TV and listening to the radio. ard ratio (the relative probability of the exposed · Increased access to media increases awareness versus unexposed being sick), which is 3.5. Lost of health issues. 43 THE WELFARE IMPACT OF RURAL ELECTRIFICATION · This increased awareness results in changed cause there is more light and because TV and health behavior. radio provide an "alternative to sex" for recre- · Changed behavior improves health outcomes ation. However, the data do not support this and reduces fertility. point of view. TV watching only significantly affects sexual activity in one of the eight cases, and house- Access to television The link between electrification, TV hold electricity is never significant. To the contrary, significantly increases ownership, and TV viewing was demon- electrification indirectly increases sexual activity, women's knowledge of strated in chapter 3. It was also shown as coital frequency is higher for women using health and family that the access to TV in villages where modern contraception, the knowledge of which planning. not all people have a TV cannot be comes in part from TV. taken for granted. Multivariate analysis of the determinants of women's knowl- These results can be used to estimate the impact edge of health and family planning provides very electrification has on fertility (table 5.3). The strong evidence that access to television signifi- total effect is the combination of the direct im- cantly increases this knowledge; this variable is sig- pact from the fertility equation and the indirect nificant in all but one of the 11 cases examined impact via higher knowledge (which is the knowl- (see appendix G). edge coefficient from the fertility equation mul- tiplied by how electricity affects knowledge, taken The household electrification variable is not sig- as the coefficient on the household electricity nificant but becomes so when television is dropped variable in the absence of the media variables). from the equation, which shows that television is These calculations show a median impact of a re- the channel through which electrification affects duction in fertility of 0.6 children as a result of health knowledge. But the percentage of house- electrification. holds that have television is significant in only one case (Ghana), so, in general, the channel of However, the link between electrification and other households having television does not op- mortality does not appear strong; the results are erate. As noted above, women are the least likely not robust. Immunization and knowledge are to view television in someone else's home. both significant in a few cases, but not over- whelmingly so, as earlier links in the chain are. The next step is to examine the extent to which knowledge affects practice. Two health practice Nutrition is an outcome that may also be affected variables are examined: use of modern contra- by knowledge, both because health knowledge ceptives and child immunization. The contra- proxies for nutrition knowledge and because ill ceptive knowledge variable was significantly, health (notably diarrhea) is a major factor in poor positive in all 11 equations estimated. For im- nutrition. In addition, electricity may positively af- munization status, the knowledge variable was sig- fect nutrition by allowing refrigerated food storage. nificant, with the expected sign in 53 of the 55 estimated equations. The link between knowl- Two nutrition measures are used to evaluate the edge and practice is thus firmly established. effect of nutrition: the height for age z score (HAZ) and the weight for age z score (WAZ). The z score The final step is the link between knowledge and is a standardized measure; being more than two z outcomes. In seven of the nine cases, the health scores below the reference value constitutes being knowledge variable has a significant negative im- undernourished; being more than three scores pact on fertility. The household electricity variable below constitutes severe undernourishment. HAZ is also significant and negative in seven cases. What is taken as a measure of long-run nutritional sta- are the possible reasons for this latter finding? tus; WAZ indicates short-run status. The condi- tioning variables are similar to those used in the A possible reason is that electricity reduces coital mortality equations. These are in turn similar to frequency by increasing waking hours, both be- those used throughout this analysis but with more 44 THE BENEFITS OF RURAL ELECTRIFICATION Table 5.3: Fertility Impact of Electrification Bangladesh Ghana Indonesia Morocco Nepal Nicaragua Peru Philippines Senegal Knowledge equation Electrification status 0.28 0.38 0.02 ­0.10 0.17 0.10 0.23 0.16 0.08 Child ever born: TFR equation Electrification status ­0.02 ­0.08 ­0.08 ­0.11 ­0.09 0.04 ­0.01 ­0.16 ­0.16 Knowledge variable ­0.02 ­0.03 ­0.01 ­0.07 0.02 ­0.04 ­0.03 ­0.03 0.00 TFR at age 50 5.83 6.35 4.63 6.32 5.98 7.25 6.57 5.54 7.42 Impact of electrification 5.68 5.77 4.26 5.67 5.46 7.51 6.45 4.63 6.23 Reduction ­0.15 ­0.58 ­0.37 ­0.65 -0.52 0.26 ­0.11 ­0.91 ­1.19 Reduction (only significant variables) 0.00 ­0.07 ­0.37 0.00 ­0.52 ­0.03 ­0.05 ­0.91 ­1.19 Source: Appendix H. Note: TFR = total fertility rate. demographic variables, because there is possible (the United States is a clear example-- Calculations show that competition for resources between siblings. child-raising costs are very high and the electrification results in pecuniary advantages negligible). fertility reduction. The knowledge variable affects HAZ in four of the six countries for which data are available; it af- An alternative approach is to take the value of fer- fects WAZ in three of the six. There is also an in- tility reduction to be the cost of achieving the direct effect, with immunization status affecting same decline in fertility through a reproductive nutrition in two of the six countries.7 Data on health intervention. To take a successful example, households owning a refrigerator were available approximately $5 billion was spent on the Bangla- for four countries: the coefficient was significantly desh family planning program, bringing the total positive on HAZ in two of the four countries and fertility rate down from around seven to three chil- in just one of the four for WAZ. There is thus ev- dren per family. As a conservative estimate, at idence that electrification improves child nutri- least half of this fertility decline is attributable to tional status, but the channels are not operating the family planning program (see IEG 2005). The in all countries. number of averted births from this fertility decline was 60 million. Hence, the cost of a one-child re- Is it possible to put a value on these effects? Valu- duction in fertility is $167 per averted birth. So ing fertility decline is a difficult matter. A straight- the fertility-reduction benefit of RE is approxi- forward economic (Beckerian) approach is to mately $100 per household. However, only 10 assume that families wish to avert births with a neg- percent of the RE impact was through clearly ative net present value, so the benefit of averted identifiable channels, so a lower limit of the RE births is the avoidance of incurring this negative impact is $10 per household. net present value. This approach was used to cal- culate the benefits of family planning programs in Time Use the 1970s, including by Bank analysts. However, it Electrification can affect time use in a variety of has fallen out of favor--at the Bank there is no re- ways: watching TV, greater participation in com- quirement to perform cost-benefit analyses for so- munity activities and socializing, reducing time cial sector interventions. Anyway, conceptually it spent on household work or shifting it to the is clear that parents want to have children even evening, increasing time spent reading or--for when they have a huge negative net present value children--doing homework, and extending hours 45 THE WELFARE IMPACT OF RURAL ELECTRIFICATION of home businesses. These additional activities are most a two-year difference (8.5 versus 6.7 years). made possible by the longer waking hours elec- However, this is a single difference estimate that tricity makes possible, with households reporting does not allow for other factors such as parental they stay up, on average, an additional one to education, household income, and school facil- two hours. ities. But IEG's analysis of DHS data for nine countries also found that electricity has a direct The main use of this additional time is watching impact on rural education once these factors TV; indeed, time spent watching TV is greater are controlled for. What are the reasons for this than the increased time available, suggesting that impact? it cuts into other activities (table 5.4). One such activity is reduced time on housework--the Philip- In low-income countries rural schools often lack pines study found that women spent one hour less basic equipment, such as furniture and adequate on such tasks as a result of electrification. But textbooks--the presence of electricity does not other studies have suggested that women's work affect these important constraints.9 The failure of burden can actually increase, as household ac- teachers to take up posts in remote locations and tivities can be carried out in the evening, allow- frequent absenteeism from such postings are ing more working hours on other activities.8 problems in many countries, and the evidence pre- Indeed, the latter may be one explanation for in- sented in the last chapter, albeit for just one coun- creased business hours, which are found in IEG's try, supports the argument that the availability of analysis of data from Ghana and the Philippines electricity makes rural positions more attractive (but not in Peru). to teachers. This is thus one possible reason for the higher education levels, with improved school Education Benefits quality encouraging students to stay on longer or The main channels through which electrification enabling them to do so as their grades improve may affect education are (1) by improving the from better teaching. quality of schools, either through the provision of electricity-dependent equipment, or increasing The other possible explanation is that increased teacher quantity and quality; and (2) time alloca- study time at home results in better grades, so chil- tion at home, with increased study time, though dren stay in school longer. There is indeed evi- the availability of TV may decrease that time (but dence that electricity increases study time (by at the same time it may also possibly provide ed- approximately an hour an evening in the case of ucational benefits). the Philippines--see appendix G), but no study follows the causal chain through to improved re- Children in electrified households have higher ed- sults and higher educational attainment. ucation levels than those without electricity. The ESMAP Philippines study (ESMAP 2003) finds al- Productive Uses A general conclusion from analysis of RE pro- grams is that the impact on productive activities Table 5.4: Hours Watching TV by Electrification is limited. But three caveats are needed to this con- Status clusion. The first is that some irrigation programs (and Bank projects) have focused on RE for irri- Lao PDR Car battery: 1.78 hours/day gation programs, and--in India, at least--were SHS: 2.20 hours/day linked to the spread of Green Revolution tech- Grid: 3.55 hours/day nologies (see Barnes 1988; Binswanger and Regression estimate: 1.26 hours/day = 37.8 hours/month Khandker 1993). Philippines Car battery: 1.85 hours/month Grid: 129 hours/month Second, in cases where there has been a com- Regression estimate: 2.25 hours/day = 67.5 hours/month plementary program to assist productive uses of Sources: ESMAP 2002; IEG analysis of survey data for Lao PDR. electricity, there has been more success--an early 46 THE BENEFITS OF RURAL ELECTRIFICATION example being USAID support in Colombia in the 1960s. However, only a minority of Bank- Box 5.3: Micro Home Enterprises supported projects have had such components. Electrification may bring the chance for small busi- Finally, considering home enterprises, the effects ness activities that help defray the costs of electrifi- are greater than those from medium and large cation. In Ghana, the woman of the household firms, although these enterprises may be small in- prepares snacks to be sold to people who come to her deed, such as renting out refrigerator space (see house to watch television in the evenings. In South box 5.3). Africa, households sell cold drinks and rent out re- frigerator space. IEG's analysis of household survey data does find evidence of a positive impact of RE on home busi- nesses. The finding is strongest for the 15-year panel data from 1988 to 2003: the number of home businesses grew significantly more in communities · GEF funding is taken as the international com- that became electrified than in either those com- munity's WTP for reduction in global CO2 emis- munities that did not or those that were already sions; that is, the total value of electrified in 1988 (appendix F). Similar evidence reduced emissions equals the GEF Off-grid renewable was not found in the other panel data set (Peru), contribution to the project. The energy activities but the year between surveys occurred at a time value per ton of CO2 avoided is have positive when rural areas were experiencing considerable based on an estimate of the amount environmental unrest. In addition, the presence of electricity ex- of CO2 avoided divided by the benefits. tends the work hours of home businesses, and amount of GEF funding. Examples this increases the net income from these activities. include Argentina Renewable Energy in Rural Markets Project, China Renewable Energy, In- Global Benefits donesia Solar Home Systems, and Honduras RE largely involves transmission and distribution, Rural Electrification. so, unlike power-generation projects, it has lim- · As an alternative approach, global environ- ited direct environmental impact. To the extent mental benefits are estimated using GEF that electrification promotes increased energy incremental costs. One example is the Nic- consumption, it increases CO2 emissions, though aragua Off-Grid Project, where GEF's incre- these are at least partially offset by the fuel dis- mental cost for minigrids is based on a placed. Grid-extension projects have not entered Prototype Carbon Fund WTP of $7 per ton of into these calculations, so the net balance of cost CO2. The same approach was used in the Bo- and benefits cannot be reported here but would livia Decentralized Infrastructure for Rural be a useful area for further analysis. Transformation Project. · Avoided emissions are calculated and valued at Off-grid activities are an exception, because Bank carbon prices currently observed in the carbon support for off-grid energy supply mainly relies market. Using this approach, the Senegal Elec- on RET, the most common being SHSs (see ap- tricity Services for Rural Areas Project values 1 pendix table B.26). Installation of RET generation ton of avoided CO2 emissions at $4.50 per ton. capacity displaces existing nonrenewable energy · The emission factor from the project (which for sources, mostly kerosene, thus creating an envi- the Mexico Carbon Fund Project is 0.584 tons ronmental benefit. The main benefit is averted of CO2emissions/MWh) is calculated, and then CO2 emissions, the value of which should be in- the value of CO2 reductions at the price of en- cluded in the benefit stream. ergy sales to the grid ($0.057 per kWh). Bank project documents value this amount in one The most common methods are based on GEF's of four ways, the first two being the most common: decision of how much to allocate to the project. 47 THE WELFARE IMPACT OF RURAL ELECTRIFICATION GEF's country allocations are based on a two- was only calculated for the Philippines, where part formula: the potential for reducing CO2 emis- it was found to be zero. sions (the product of the baseline emissions and · Improved health comes from the value of re- the rate of reduction over the previous decade) duced mortality as a result of improved indoor and an institutional quality measure of the ca- air quality from reduced reliance on kerosene pacity to implement environmental programs. lamps, which has a monthly annuity value of How much money a project gets from GEF de- $0.02. pends on the country allocation and the number · Reduced fertility coming from knowledge from of projects. Hence, the GEF-based estimation of channels accessed using electricity, valued at the environmental benefits is not project specific and cost of achieving fertility reduction through thus bears no relation to the actual level or value reproductive health programs. The lower limit of carbon emissions averted. of these benefits was placed at $10, which is equivalent to a monthly "annuity" of $0.08. Calculation of CO2 Application of the method results in dif- · Public goods benefits, such as increased secu- emissions averted is done ferent valuations of the worth of avert- rity, have not been estimated in any of the in several different ways, ing CO2 emissions from project to cases, but are listed for completeness. Global resulting in ERRs that are project. This figure varies as well, be- benefits from reduced CO2 emissions apply not comparable. cause different documents use differ- only to off-grid components. As argued in the ent approaches. As the benefit is a text, calculations should also reflect the net global one, it should be expected that a ton of CO2 environmental impact of increased energy con- emissions saved carries the same value regardless sumption as a result of grid extension, but this of the source. Using a variety of methods to arrive has not been calculated in any Bank studies. at different values per ton of CO2 undermines the comparability of the resulting ERRs. Obtaining the total benefits from RE is difficult for two reasons. First, some of the benefits are diffi- Adding Up the Benefits cult to put a value on. The rationale for the ESMAP Table 5.5 summarizes the data for selected coun- study of the Philippines was that it was the re- tries for which a range of benefits have been sulting systematic undervaluation of the benefits quantified. These benefits are as follows: from electrification that made these projects appear unattractive investments. Accordingly, · Benefits from lighting and TV/radio, calculated the study valued a broader range of benefits, as WTP. showing these to be substantially in excess of · Education benefits from higher educational at- those from lighting alone (see table 5.4). However, tainment by the children of electrified house- the second problem is that there can be double holds, which results in higher future earnings. counting. The present value of these incremental future earnings is calculated and imputed to a monthly Double counting can occur because households' figure.10 WTP for lighting or TV includes the value they at- · Time saved from household chores (additional tach to longer waking hours, better indoor air leisure time), valued at the opportunity cost of quality, greater study time, and the informational labor, that is, the average wage.11 benefits flowing from watching television. Project · Productivity of home business includes total net documents often calculate the benefits based on revenue from new businesses and incremen- WTP for lighting and sometimes TV, noting that tal revenue to existing businesses. Because the these are underestimates because there are many results are for the average household, they other benefits not included. In fact, though, many have to be adjusted to reflect the proportion are included in the household's valuation of its of households with a home business.12 WTP. It is only the public good elements of house- · Similarly increased agricultural productivity hold consumption that are not included, which calculated as incremental revenue. This figure might include, for example, knowledge benefits 48 THE BENEFITS OF RURAL ELECTRIFICATION Table 5.5: Rural Electrification Benefits (US$ per household per month) Benefit Philippines Peru Lao PDR Bolivia Lighting 7.36a 16.16 5.60 12.24 TV 15.11 8.5 2.22 4 Radio (included in TV) Not estimated Not estimated Not estimated Education 12.46 Not estimated Not estimated Not estimated Time saved for household chores/increased leisure 5.30 5.5 5.5 5.5 Productivity home business: existing business 6.30 0.0 3.40 Not estimated Productivity home business: new business 5.25 0.0 2.35 Not estimated Improved health 0 0.02 0.02 Not estimated Reduced fertility Not estimated 0.08 0.08 Not estimated Increased agricultural productivity 0 0 Not estimated Not estimated Public good benefits (including security) Not estimated Not estimated Not estimated Not estimated Reduced pollution (global benefits)b Not estimated 0.24 0.15 0.20 Source: IEG data. a. IEG estimates for the Philippines differ from those by ESMAP (2002) because that study used a linear demand curve. b. Applies to off-grid beneficiary households only. Assumes 0.6 ton of CO2/MWh priced at $8/ton of CO2. to outsiders watching the household television, and there are not missing benefits but Calculating the total household members spreading this knowledge by a mispricing of some output, with an benefits of electrification word of mouth (though IEG's analysis suggests unknown bias on the ERR. But if the is difficult because some these are very much second-order benefits), or benefits are truly omitted, then the benefits are difficult to public benefit from an external light. omission is not serious if the ERR is value and there is a above the threshold anyway, which is potential for double Even if the health and fertility benefits are addi- usually the case. counting. tional, their monthly value is quite small. However, this study is the first to quantify these, so these Total household WTP depends on the extent to estimates should be treated with caution until which it is believed other benefits are internalized further data and analysis are available to strengthen in the WTP for lighting and TV. Assuming they are the evidence base. In contrast, the education ben- so internalized, with an allowance for unaccounted efit is sizeable. Again, further research is needed benefits, gives a WTP of $10­30 per household to fully understand this channel. For those house- per month (excluding home business benefits), holds that do have a home business, there appears corresponding to around $0.20­0.60 per kWh, to be a reasonable income impact from RE. depending on whether other benefits are inter- nalized in the WTP. Assuming these benefits are Table 5.5 shows that there are other, unquantified, not internalized adds up to another $30 to this benefits. Project documents typically speak of amount (though the Philippines study derived ERR estimates being conservative because they rather higher figures), giving a range of $40­70 per omit some benefits. Of course, total consumption household per month (or around $1.00 and higher is being valued at the household WTP for lighting, per kWh). 49 THE WELFARE IMPACT OF RURAL ELECTRIFICATION How Do the Benefits Compare than $0.03 a day ($0.78 a month) if amortized over with the Costs? 20 years, and thus is well within the affordability Actual connection costs vary between $150 and of even most poor households, the absence of $2,000 per household, depending on the loca- credit markets means these households are not tion and size of the community; costs are even less in a position to spread the payment in this way. in already connected communities. The estimate The two solutions are to fill the gap in the credit of household benefits at a midpoint of $50 a market and to subsidize the connection fee for month equals $600 a year, meaning that the break- poorer households. As argued earlier, the mar- even point is between one and three years and that ket can be segmented by the connection lag, al- discounting net benefits over, say, a 20-year period lowing an increase in both the utility's financial will give a good rate of return. Put another way, the performance and the economic return to the household WTP is well above the average supply project. cost. Rural grid connections can indeed New approaches be good investments, though each in- But, as this report stresses, decision making is convincingly demonstrate vestment is context specific regarding context specific. Many African countries have yet that WTP is high and both costs (generation and supply costs to embark on RE. In these cases, connection costs generally exceeds the vary widely and depend on tackling will be high, and many areas may not be con- average supply cost where technical issues, most notably system nected to the grid for some years. The emphasis grid connection is losses) and benefits (which vary ac- in the coming years will be on putting in place the feasible. cording to alternative energy sources, basic infrastructure for grid connections with an potential for productive use, and so on). eye on financial sustainability, which will imply relatively low community connection rates, in- What the new approaches have convincingly creasing coverage by extensive rather than in- demonstrated is that the WTP for RE is high and tensive growth. almost invariably exceeds the average supply cost for areas in which grid connection is considered At the same time, there will be areas beyond the feasible (see figure 5.3). The immediate implica- reach of the grid that will be suitable for off-grid tion of this finding is that the ERR will exceed the connections. In these schemes the subsidy ele- financial rate of return, a state of affairs that points ment can be tilted toward the bottom end to in- to the policy conclusion that the financial rate of crease connection rates. A second group of return can be raised by increasing tariffs if a countries, mostly in Asia, is still struggling to es- stronger financial rate of return is needed. tablish the financial sustainability of its grid pro- grams. The analysis shows that the market can Reducing consumption of But assuming that return is sufficient bear tariff increases, though these may be politi- the high-end consumers for sustainability, then the policy con- cally unpopular. through a higher tariff clusion is that the project ERR can be will raise consumption of raised through a cross-subsidy-based re- But there is another group of countries that has the low-end consumers, distribution. The argument is a simple put the utility on a sound financial footing and is providing a net welfare utilitarian one: the marginal benefit to in a position to reap the full potential benefits of benefit. the low-level consumer exceeds that to RE. These benefits will be realized by shifting to the high-end consumer; reducing the intensive growth, which is made possible by re- consumption of the latter by charging a higher duced or monthly connection charges for late price to raise the consumption of the former re- connectors, increasing and diversifying patterns sults in a net welfare gain. of electricity use through consumer education, and providing support to productive uses. The strength of this argument is reinforced by the fact that low-income households may be willing Off-Grid Connections to pay but cannot afford to. The connection fee The rate of return to RET investments can be ex- is typically $100. Although this comes to a less pected to be lower than that in grid electricity be- 50 THE BENEFITS OF RURAL ELECTRIFICATION Figure 5.3: Willingness to Pay Exceeds Supply Cost 0.45 0.40 0.35 0.30 0.25 $/kWh 0.20 0.15 0.10 0.05 0.00 Bangladesh Lao PDR Philippines India India Indonesia 1990 2005 1994 (Nathpa Jhakri) (Rajasthan) 2000 1989 2000 WTP Price Cost of supply Source: Project documents and IEG calculations. Note: kWh = kilowatt hours; WTP = willingness to pay. cause, at present, the costs are usually higher and The same survey found that satisfac- Africa generally is the benefits lower. To clarify the statement on tion with quality and quantity of elec- far from having the costs: the cost of providing electricity to the com- tricity supply is much lower for infrastructure needed to munity should be lower using a RET than con- households connected by SHS: nearly build a rural grid, and necting that community to the grid (otherwise all (92 percent) of such households say many Asian countries it should be part of a grid-extension program), they need more kilowatt hours to still struggle with but the cost per kilowatt hour will likely be greater cover their needs, which is the case for financial sustainability than the cost for those who are connected to less than a third of grid-connected of their grids. the grid. The benefits are lower because the households. capacity is less, so off-grid connections support fewer appliances and possibly provide fewer In addition, off-grid projects have suffered from hours of lighting. But of course the choice for technical problems resulting from lack of technical these communities is not between grid and capacity in rural areas and the logistical difficul- off-grid, but among grid, kerosene, and car ties of servicing equipment (see box 5.4). These batteries. problems mean systems fall into disuse or run below planned capacity, further reducing the re- For example, the IEG analysis of survey data for turn. For example, IEG reanalyzed the rate of re- Sri Lanka showed that the median household turn for off-grid investments under the Lao PDR total wattage of all lightbulbs used is 360 watts for Southern Provinces Rural Electrification Project; grid-connected households and just 60 watts the review by Bank operational staff estimated the for SHS users.13 In addition, households with ex post ERR to be 26 percent. But allowing for the SHS own practically no electrical appliances other technical problems reduced the return to 16 per- than a TV, whereas a large proportion of grid- cent (which fell to ­9.4 percent once the revalu- connected households own a range of appli- ation of consumer surplus, discussed at the ances including water heaters, irons, and grinders. beginning of this chapter, was applied). 51 THE WELFARE IMPACT OF RURAL ELECTRIFICATION Box 5.4: Technical Problems Reduce the Benefits from Off-Grid Investments More than 6,000 households were given off-grid connections under more than four years old. Most households have not replaced these the Southern Provinces Rural Electrification Project in Lao PDR, batteries (which cost $20­50), but continue to charge them, get- virtually all of them SHSs. This went beyond the appraisal target ting just 30 minutes to an hour's electricity a day, which is a great by nearly one-third. reduction in project benefits. However, a survey undertaken in Vientiane Province in 2006 In Thailand, 59 micro hydroelectricity schemes have been im- revealed a number of operational problems. More than 80 percent plemented. Of these, only 25 remain in operation. Most of those that of SHSs were not working properly or were working at a low are now defunct ceased operation when the grid reached the vil- level of service. The large majority of controllers were no longer lage. There are a number reasons for the preference for grid elec- working: 40 percent were simply missing, and most of the re- tricity: (1) the lower cost to villagers, which is heavily subsidized mainder had been bypassed so the panel was connected directly but also does not require community management, which micro to the battery. The resultant power fluctuations shorten battery life hydro does; (2) technical problems with micro hydro, in part aris- (which was only two to three years anyway), as does excess use ing from the last reason; and (3) "a tragedy of the commons" by connecting too many lights or appliances. As a result, nearly whereby households consume "excess electricity" by using rice all batteries were past their useful life, with more than half being cookers and heaters, causing shortages. Source: Greacan 2003. In all cases where projects This statement is supported by looking It might be argued that off-grid investments serve have both grid and off- at Bank projects that contain both grid other environmental or social objectives, but grid components, the ERR and off-grid components (appendix these should be explicitly factored into the analy- is greater for the grid table B.20). In virtually all cases, the sis by the valuation of environmental benefits component. ERR for grid connections is greater and by using social weights (reflecting policy than that for the off-grid component. makers' preferences) for different groups. An This observation raises a question about the use alternative argument might be that these are of the results of economic analysis. If a project has small programs that enable learning by doing, one component with a high return and another which, together with technological develop- with a low return, the policy implication is that ments, will improve competitiveness. This ar- funds should be diverted to the high-return ac- gument is not, however, found in the project tivity until the rates of return are equalized. documents. 52 Chapter 6 Rural Chinese village with electric poles. (Photo from the World Bank Photo Library.) Conclusion and Lessons Learned Answering the Evaluation Questions multisectoral investments--respectively; appen- dix table B.3). A final perspective is given by the What is the rationale for Bank support of RE? benefits included in the project analysis. Why does the Bank support RE? The policy paper "Rural Electrification" (World Bank 1975) argued Quantification of benefits is most commonly re- that investments had to be justified by the bene- stricted to lighting benefits, with a small number fits to consumers and increased production. If the of analyses including TV viewing. Other benefits resulting return on investment (ERR) was insuffi- are sometimes mentioned but not quantified. cient, a case might be made on social grounds. However, it should be recognized that electricity The largely private quantified benefits appear is not a necessity like water or health, although it rather distant from the broad claim that clean does benefit consumers and results in increased energy is central to poverty reduction and eco- production. In contrast, the two 1993 policy papers, nomic growth. This is especially so because the which were not specifically concerned with RE, pro- poor are still excluded from direct benefits, and ceed straight to their core subject matter without few Bank projects have taken explicit steps to in- offering a rationale for energy investments. clude them. More recent policy papers have stressed the links In summary, the economic case for investments between energy and poverty. Most notably, Rural in RE is proven, provided technical problems in Energy and Development (World Bank 1996) service provision are adequately addressed. But the documented the time burden and adverse health evidence base for the links between RE and poverty implications of relying on biomass fuels. The 2001 remains thin. Improved evaluation tools--of the sector board paper begins with the bolder state- sort already adopted in some recent projects--are ment: "Efficient and clean energy supply is cen- needed to build the case that RE should be a pri- tral to the reduction of poverty through many and ority for a poverty-oriented lending institution. varied linkages, as well as being important for economic growth." What has been the growth in coverage of RE in countries receiving Bank support? To what Alternatively, project documents can be exam- extent has the Bank contributed to these ined to understand the rationale for Bank lend- connections? What is the distributional profile ing. The majority of projects take the benefits as of those taking connections? What are the unit self-evident, as the objectives are restricted to costs of connection by type of supply to the the outputs of improved access or institutional de- user and to the supplier? velopment. A minority of RE projects have welfare RE has increased substantially in many coun- objectives, the most common of which are in- tries receiving Bank support. Where the Bank creasing growth and a general improvement in has had a series of dedicated projects, it has welfare (cited in the objectives of 21 percent and made a significant contribution to increases in 19 percent of energy projects--that is, excluding coverage. 55 THE WELFARE IMPACT OF RURAL ELECTRIFICATION Support for electrification has mostly been pro- longer. Again, more evidence is needed to sub- vided to communities where connection was stantiate these findings. deemed most cost-effective, leaving remote com- munities--often among the poorest--the last How does RE affect the quality of health and ones connected. This pattern is at best only par- education services? tially overcome by the development of off-grid RE benefits the quality of health services and electricity sources, but per unit costs of off-grid lowers costs by extending opening hours and sig- sources are significantly higher than the price of nificantly strengthening the cold chain for vac- electricity on the grid. Once a community is con- cines--though it does not increase the extent to nected, however, electricity from both grid and off- which such services are offered. Electrification grid sources represents substantial cost savings was also found to reduce worker absenteeism in compared with kerosene. However, the connec- both health clinics and schools by improving liv- tion charge barrier prevents many of the poorest ing conditions and morale. However, the cases from connecting once electricity is available, and studied are few, so further analysis is required. few Bank projects have introduced mechanisms to help overcome this barrier. How do the aggregate private benefits and the public good benefits compare with the WTP? What are the direct economic benefits from What is the distributional profile of these RE? Who gains these benefits? What are the benefits? indirect economic benefits (employment The WTP internalizes many of the benefits and ex- generation), and who gains them? How does ceeds the supply cost. The benefits, like electric- the distribution of benefits change as coverage ity consumption, favor the better-off. of electrification programs expands? Direct economic benefits from RE occur as elec- What are the private and social rates of return tricity supply lowers the cost of energy to the from investments in RE? user, resulting in an increase in consumer sur- The private returns to electrification are high-- plus. Such benefits tend to favor the well-off, be- as indicated by the fact that most, if not all, house- cause connection charges and tariffs are often holds that can afford electricity take it once it prohibitive for the poorest. The pattern of elec- becomes available. Great value is placed on the trification favors the non-poor, but distribution improved lighting it makes available and the pos- becomes more equitable as electrification cover- sibility of watching TV. Because these benefits are age expands. mostly private, the economic and social returns would be close if electricity prices reflected eco- RE does not in general drive industrial develop- nomic cost. The divergence between the two re- ment, but it can spur growth of home businesses. turns emerges because of the subsidy element and Such businesses mostly employ family labor and makes a case for reducing the amount of that increase their hours once electricity becomes subsidy, which could be better targeted to getting available. Electrification thus provides a small, poorer households connected. but not negligible, boost to the incomes of some households. However, the evidence base on this Lessons Learned point remains thin. It is difficult to generalize about RE because both costs and benefits are context specific. However, What is the impact of RE on time use, and what some broad statements can be made. are the welfare implications of these changes for health, education, and increased leisure? · RE investments have often generated sufficient Electrification extends waking hours, with a prin- benefits for the investment to be warranted ciple impact being more time spent watching from an economic standpoint. television. Time is also saved from chores, but this · The value of these benefits to households gain is limited, and the time spent in home busi- is above the average supply cost, so cost- nesses increases. Health clinics remain open recovery tariff levels are achievable, even if 56 CONCLUSION AND LESSONS LEARNED they are politically unpopular in countries with by extensive growth for some years to come. a history of low tariffs. The principal challenge is to balance financial · Analysis of feasible tariff levels can be informed sustainability with growing coverage, requiring by good quality economic analysis of the sort efficiency by limiting system losses. Because pioneered by the Philippines ESMAP study grid connections will grow slowly, many areas (ESMAP 2003). But the quality of that study may be eligible for off-grid connections, but the is not uniformly replicated, as the quality of logistics of maintaining technical quality will be project-level analysis is uneven, with apparent challenging. weak quality control. · Some countries in Asia and Latin America are · The evidence base remains weak for many of reaching the limits of grid extension. Further in- the claimed benefits of RE. Tailor-made sur- creases in coverage require intensive growth, veys, designed to test these benefits, need to which requires instruments designed for that be built into a greater number of Bank projects purpose, or off-grid schemes, which need de- and designed so they allow rigorous testing of sign improvements if they are to be financially the impact of electrification. sustainable. · Countries with low coverage rates, which are · There are project design options that have now mostly in Africa, still have to make in- been uncommon but that would enhance proj- vestments in generation, transmission, and dis- ect benefits. These include financing schemes tribution, which implies relatively high average for connection charges, consumer education, supply costs and low coverage, increasing slowly and support for productive uses. 57 Turbine on stream near village generating energy as part of hydro scheme. (Photo from the World Bank Photo Library.) Appendixes APPENDIX A: RURAL ELECTRIFICATION PORTFOLIO 61 62 THE Table A.1: Project Portfolio Data WELF IDA IBRD GEF Prototype ARE Planned Actual funding funding funding carbon fund Approval closure closure amount amount amount amount IMP Country Project date date date ($ millions) ($ millions) ($ millions) ($ millions) ACT Africa Benin Energy Services Delivery Project 2005 2009 45.0 OF Burundi Energy Sector Rehabilitation Project 1991 1996 1999 20.4 RURAL Cape Verde Energy and Water Sector Reform and Development Project 1999 2004 17.5 4.7 Eritrea Power Distribution and Rural Electrification 2005 2009 50.0 Ethiopia Energy 2 1998 2004 2006 200.0 ELECTRIFICA Ethiopia Energy Access Project 2003 2009 132.7 4.9 Ethiopia Electricity Access (Rural) Expansion 2006 2010 133.4 Ghana Northern Grid Extension Project 1987 1992 1992 6.9 Ghana Sixth Power Project 1990 1996 1998 9.2 Ghana National Electrification Project 1993 1998 2000 80.0 TION Guinea Decentralized Rural Electrification Project 2003 2008 5.0 2.0 Guinea-Bissau Energy Project 1991 1995 1998 13.3 Kenya Geothermal Development and Energy Pre-Investment Project 1988 1994 1996 41.3 Madagascar Energy Sector Development Project 1996 2002 2006 44.2 Malawi Power V Project 1992 1998 2000 91.5 Malawi Infrastructure Services Project 2006 2012 40.0 Mali Household Energy and Universal Access 2004 2009 35.7 3.5 Mauritania Energy, Water, and Sanitation Sector Reform Technical Assistance 2000 2003 2005 9.9 Mozambique Energy Reform and Access 2004 2008 40.3 3.1 Niger Energy Project 1988 1995 1997 17.5 Nigeria Privatization Support Project 2001 2008 114.3 Nigeria National Energy Development Project 2006 2008 172.0 1.0 Senegal Electricity Services for Rural Areas Project 2005 2009 29.9 5.0 Tanzania Songo Songo Gas Development and Power Generation 2002 2006 2008 183.0 Uganda Energy for Rural Transformation 2002 2009 49.2 12.1 East Asia and Pacific Cambodia Rural Electrification and Transmission 2004 2009 40.0 5.8 China Daguangba Multipurpose Project 1992 1998 1999 69.3 China Tarim Basin Project 1992 2008 2008 128.0 China Second Red Soils Area Development Project 1993 2001 2002 150.0 China Southwest Poverty Reduction Project 1995 2002 2006 200.0 47.5 China Western Poverty Reduction--Gansu and Inner Mongolia 1999 2006 2006 100.0 60.0 China Renewable Energy Development 1999 2008 100.0 35.0 China Hubei Hydropower Development in Poor Areas 2002 2009 105.0 China Renewable Energy Scale-Up Program 2005 2011 87.0 40.2 China Follow Up to Renewable Energy Scale-Up 2006 2011 86.3 Indonesia Power Sector Efficiency Project 1989 1995 1996 309.3 Indonesia Rural Electrification Project I 1990 1994 1995 261.3 Indonesia Second Rural Electrification Project 1995 1999 2000 244.7 Indonesia Solar Homes System Project 1997 2002 2004 0.1 4.5 Lao PDR Southern Provinces Electrification Project 1987 1993 1994 0.5 25.9 Lao PDR Provincial Grid Integration Project 1992 1998 1999 34.9 Lao PDR Southern Provinces Rural Electrification Project 1998 2004 2005 34.7 0.7 Lao PDR Rural Electrification Phase 1 2006 2010 10.0 3.8 Malaysia Rural Electrification Project 1982 1986 1988 36.3 Malaysia Eleventh Power Project 1984 1989 1990 70.0 Papua New Guinea Yonki Hydroelectric Project 1986 1993 1993 28.5 APPENDIX Philippines Rural Electrification Revitalization Project 1992 1997 1998 54.6 Philippines Leyte Luzon Geothermal Project 1994 1999 2000 177.3 31.2 Philippines Transmission Grid Reinforcement Project 1996 2001 2004 203.9 Philippines Rural Power Project 2004 2010 10.0 9.0 A: Philippines Electric Cooperative System Loss Reduction Project 2004 2011 12.0 RURAL Thailand Second Accelerated Rural Electrification Project 1980 1986 1987 75.0 Thailand Provincial Power Distribution Project 1983 1987 1988 30.6 Thailand Power Transmission Project 1988 1993 1993 110.0 ELECTRIFICA Thailand Distribution System Reinforcement Project 1995 1999 1999 47.6 Vietnam Power Development Project 1996 2000 2000 179.6 Vietnam Rural Energy 2000 2007 150.0 Vietnam Community-Based Rural Infrastructure 2001 2008 102.8 Vietnam System Efficiency, Equitization, and Renewables 2002 2008 225.0 4.5 TION Vietnam Second Rural Energy Project 2005 2012 220.0 5.3 POR Europe and Central Asia Albania Power Transmission and Distribution 1996 2001 2003 17.7 TFOLIO Portugal Tras os Montes Regional Development 1989 1997 1997 30.7 63 (Table continues on next page) 64 THE Table A.1: Project Portfolio Data (continued) WELF IDA IBRD GEF Prototype ARE Planned Actual funding funding funding carbon fund Approval closure closure amount amount amount amount IMP Country Project date date date ($ millions) ($ millions) ($ millions) ($ millions) ACT Latin America and the Caribbean Argentina Renewable Energy in the Rural Market Project 1999 2009 30.0 7.2 OF Bolivia Decentralized Infrastructure for Rural Transformation 2003 2008 20.0 RURAL Brazil Rural Poverty Alleviation--Sergipe 2002 2006 20.8 Brazil Sergipe Rural Poverty Alleviation 1995 2001 2001 36.0 Brazil Pernambuco Rural Poverty Reduction 2001 2010 30.1 ELECTRIFICA Brazil Northeast Rural Poverty Alleviation Project--Pernambuco 1997 2001 2002 36.0 Brazil Ceara Rural Poverty-Alleviation Project 1995 2001 2001 70.0 Brazil Ceara Rural Poverty-Alleviation Project 2001 2009 37.5 Brazil Itaparica Resettlement and Irrigation Project 1988 1995 1998 132.0 Brazil Bahia State Integrated Project: Rural Poverty 2006 2011 54.4 TION Brazil Bahia Rural Poverty Reduction 2001 2005 2005 54.4 Brazil Irrigation Subsector Project 1988 1994 1998 195.0 Brazil Rural Poverty Alleviation--Paraiba 1998 2003 2006 60.0 Brazil Northeast Rural Poverty-Alleviation Program--Muranhao 1998 2003 2004 80.0 Brazil Rural Poverty Reduction--Minas Gerais 2006 2010 35.0 Brazil Rural Electrification 1984 1990 1990 222.0 Brazil Energy Sector Reform Loan 2002 2003 2003 454.6 Brazil Electrobras 1st Distribution 1982 1988 1990 172.1 Brazil Electrobras 2nd Distribution 1984 1989 1992 237.0 Chile Infrastructure for Territorial Development 2005 2010 50.3 Colombia Village Electrification Project 1981 1986 1988 30.1 Colombia Private Sector Energy Development Project 1996 2003 2003 249.3 Colombia Jepirachi Carbon Offset 2003 2020 3.2 Dominican Republic Power Sector Technical Assistance 2004 2009 7.3 Ecuador Power and Communications Sectors Modernization and Rural Services 2001 2007 23.0 2.8 Honduras Rural Infrastructure Project 2006 2010 47.0 2.3 Mexico Decentralization and Regional Development 1991 1996 1996 350.0 Mexico Renewable Energy for Agriculture 2000 2004 2006 13.7 8.9 Mexico Waste Management and Carbon Offset 2005 2015 0.9 Nicaragua Off-Grid Rural Electrification 2003 2009 12.0 4.0 Panama Social Investment Fund 1997 2002 2004 21.9 Panama Utilities Restructuring Technical Assistance 1998 2001 2004 16.0 12.7 Peru Social Development and Compensation Fund Project 1993 1997 1997 100.0 Peru Second Social Development and Compensation Fund 1997 1999 2000 124.4 Peru Rural Electrification Project 2006 2013 50.0 10.0 Middle East and North Africa Jordan Energy Development Project 1984 1988 1991 29.3 Jordan 6th Power Project 1986 1991 1992 27.5 Morocco Village Electrification Project 1979 1984 1986 36.3 Morocco Second Rural Electrification Project 1991 1997 1997 36.5 Morocco Irrigation-Based Community Development 2001 2008 32.6 Morocco Integrated Solar Combined Cycle Power Project 2007 2009 43.2 Tunisia 4th Power Project 1984 1989 1991 21.5 South Asia APPENDIX Bangladesh Rural Electrification Project 1981 1989 1990 49.4 Bangladesh Second Rural Electrification Project 1985 1991 1993 80.1 Bangladesh Third Rural Electrification 1990 1998 2000 110.2 Bangladesh Rural Electrification and Renewable Energy Development 2002 2008 191.0 8.2 A: India Third Rural Electrification Project 1982 1986 1988 295.5 RURAL India Maharashtra 1989 1997 1999 336.6 India Nathpa Jhakri Power Project 1989 1998 2002 427.0 India Renewable Resources Development Project 1993 2000 2002 110.4 75.0 26.2 ELECTRIFICA India Second Renewable Energy 2000 2008 50.0 80.0 5.0 India The Rajasthan Power Sector Restructuring Project 2001 2005 2006 166.2 Nepal Power Sector Efficiency Project I 1992 1998 1999 56.5 Nepal Power Development 2003 2009 75.6 Pakistan Private Tubewell Development Project 1989 1995 1995 35.3 TION Pakistan Rural Electrification I 1990 1996 1997 37.0 123.0 Pakistan Transmission Extension and Reinforcement 1990 1996 1997 160.4 POR Sri Lanka Energy Services Delivery Project 1997 2003 2003 22.1 5.7 Sri Lanka Renewable Energy for Rural Economic Development Project 2002 2008 22.1 5.7 TFOLIO Note: IBRD = International Bank for Reconstruction and Development; IDA = International Development Association; GEF = Global Environment Facility. 65 APPENDIX B: STATISTICAL OVERVIEW OF PORTFOLIO Table B.1. Number and Percentage of Projects by Region and Period Table B.2. Number and Percentage of Projects by Rating and Period Table B.3. Proportion of Projects by Objective and Period Table B.4. Proportion of Projects by Component and Period Table B.5. Number and Percentage of Projects by Period and Objective Category Table B.6. Number and Percentage of Projects by Period and Component Category Table B.7. Proportion of Projects by Region and Component Category Table B.8. Number and Percentage of Projects by Planned Use and Period Table B.9. Number of Projects by Planned Use and Region Table B.10. Number of Projects On and Off Grid Table B.11. Projects with an Off-Grid Component by Period Table B.12. Number and Percentage of Projects by Promotional Activities and Approval Period Table B.13. Number and Percentage of Projects by Region and Scope Table B.14. Number of Percentage of Projects by Period and Scope Table B.15. Number of Beneficiaries Table B.16. Lines and Substations Built Table B.17. Projects by Objective Category and Scope Table B.18. Projects with Poverty-Reduction Objective by Scope Table B.19. Number of Projects by Scope and Rating Table B.20. Economic Rate of Return for Grid and Off-Grid Components Table B.21. Connections by Type Table B.22. Utility Cost Recovery by Income Level and Region Table B.23. Number and Percentage of Projects with RET by Period and Region Table B.24. Number and Percentage of Projects with RET by Period and Scope Table B.25. Number and Percentage of Projects with RET by Period and Scope Table B.26. Projects with Off-Grid Components and Type of Energy Source Table B.27. Gender as a Design Feature in Projects Table B.28. Number and Percentage of Projects by Rating and Objective Category Table B.29. Comparisons for PV System Retail Prices Table B.30. Connection Costs and Charges in Project Countries Table B.31. Subsidy Types for Off-Grid Project Components 67 THE WELFARE IMPACT OF RURAL ELECTRIFICATION Table B.1: Number and Percentage of Projects by Region and Period Number of projects Percentage 1980­95 1996­2006 Total 1980­95 1996­2006 Total Entire RE portfolio Latin America and Caribbean 10 25 35 19.2 36.8 29.2 Middle East and North Africa 5 2 7 9.6 2.9 5.8 Sub-Saharan Africa 8 17 25 15.4 25.0 20.8 South Asia 11 6 17 21.2 8.8 14.2 East Asia and Pacific 17 17 34 32.7 25.0 28.3 Europe and Central Asia 1 1 2 1.9 1.5 1.7 Total 52 68 120 100 100 100 Energy sector only Latin America and Caribbean 4 10 14 10.0 21.7 16.3 Middle East and North Africa 5 1 6 12.5 2.2 7.0 Sub-Saharan Africa 8 13 21 20.0 28.3 24.4 South Asia 10 6 16 25.0 13.0 18.6 East Asia and Pacific 13 15 28 32.5 32.6 32.6 Europe and Central Asia 0 1 1 0.0 2.2 1.2 Total 40 46 86 100 100 100 Table B.2: Number and Percentage of Projects by Rating and Period Number of projects Percentage 1980­95 1996­2006 Total 1980­95 1996­2006 Total Entire RE portfolio Highly satisfactory 4 1 5 7.7 1.0 7.0 Satisfactory 27 10 37 51.9 52.6 52.1 Moderately satisfactory 7 2 9 13.5 10.5 12.7 Moderately unsatisfactory 3 1 4 5.8 5.3 5.6 Unsatisfactory 11 5 16 21.2 26.3 22.5 Total 52 19 71 100 100 100 Energy sector only Highly satisfactory 3 1 4 7.5 9.1 7.8 Satisfactory 20 5 25 50.0 45.5 49.0 Moderately satisfactory 5 1 6 12.5 9.1 11.8 Moderately unsatisfactory 3 0 3 7.5 0 5.9 Unsatisfactory 9 4 13 22.5 36.4 25.5 Total 40 11 51 100 100 100 Source: IEG ratings database. Note: No projects were rated unsatisfactory. Some totals do not equal exactly 100 because of rounding. 68 APPENDIX B: STATISTICAL OVERVIEW OF PORTFOLIO Table B.3: Proportion of Projects by Objective and Period By sector Excluding All Objective multisector Multisector 1980­95 1996­2006 projects Increase general welfare 0.19 0.26 0.15 0.25 0.21 Increase incomes, economic growth 0.21 0.53 0.21 0.37 0.30 Reduce poverty 0.07 0.59 0.13 0.28 0.22 Improve health 0.00 0.03 0.02 0.00 0.01 Increase agricultural production 0.03 0.18 0.13 0.03 0.08 Increase industrial/SME production 0.03 0.03 0.02 0.04 0.03 Beneficial environmental effects 0.30 0.06 0.17 0.28 0.23 Increase access: general 0.15 0.03 0.12 0.12 0.12 Increase access: rural 0.37 0.06 0.27 0.29 0.28 Support electricity supply--general 0.31 0.03 0.33 0.16 0.23 Support electricity supply--poor/disadvantaged 0.01 0.00 0.02 0.00 0.01 Support electricity supply--rural 0.19 0.00 0.15 0.12 0.13 Improve sector efficiency 0.36 0.12 0.31 0.28 0.29 Develop energy resources 0.08 0.03 0.10 0.04 0.07 Promote RET 0.28 0.06 0.06 0.34 0.22 Institutional development 0.42 0.15 0.42 0.28 0.34 Capacity building of utility 0.29 0.06 0.38 0.10 0.23 Capacity building: pricing 0.08 0.03 0.13 0.01 0.07 Capacity building: private sector 0.08 0.03 0.04 0.09 0.07 Capacity building: RET 0.06 0.00 0.04 0.04 0.04 Capacity building: other 0.09 0.41 0.17 0.19 0.18 Promote private sector involvement 0.35 0.12 0.10 0.43 0.28 Third sector (NGOs/CSOs/ECs) involvement 0.10 0.32 0.12 0.21 0.17 Nonelectrification objective 0.00 0.65 0.15 0.21 0.18 Note: CSO = civil society organization; EC = electric cooperative; NGO = nongovernmental organization; RET = renewable energy technology; SME = small and medium enterprise. 69 THE WELFARE IMPACT OF RURAL ELECTRIFICATION Table B.4: Proportion of Projects by Component and Period Proportion of projects Component 1980­95 1996­2006 Total Hydro plant 0.13 0.03 0.08 Grid expansion 0.88 0.44 0.63 Off-grid generation 0.00 0.16 0.09 RET--wind, mini hydro, SHS 0.13 0.46 0.32 Other (dispatch, substations, etc.) 0.40 0.10 0.23 Rehabilitation 0.29 0.18 0.23 System loss reduction 0.02 0.01 0.02 Facilities n.e.s. 0.19 0.03 0.10 Independent grid 0.00 0.04 0.03 Study--tariff policy 0.13 0.07 0.10 Study--project feasibility 0.19 0.06 0.12 Study--other 0.38 0.26 0.32 Technical assistance: staff/management/IT 0.56 0.82 0.71 Technical assistance: engineering design/increase efficiency 0.33 0.21 0.26 Technical assistance: commercialization 0.06 0.29 0.19 Reform/Create government agency 0.04 0.09 0.07 Policy development--RET 0.02 0.21 0.13 Policy development--general 0.08 0.16 0.13 Procure equipment, instruments, software 0.33 0.15 0.23 Consumer subsidies 0.00 0.01 0.01 Credit to private providers 0.04 0.16 0.11 Open market to private tenders 0.02 0.03 0.03 Finance pilot projects 0.08 0.38 0.25 End user training/education 0.00 0.19 0.11 Resettlement/compensation 0.08 0.10 0.09 Demand-side management 0.00 0.12 0.07 Nonelectrification activities 0.25 0.26 0.26 Note: IT = information technology; n.e.s. = not otherwise specified; RET = renewable energy technology; SHS = solar home system. 70 APPENDIX B: STATISTICAL OVERVIEW OF PORTFOLIO Table B.5: Number and Percentage of Projects by Period and Objective Category Number of projects Percentage 1980­95 1996­2006 Total 1980­95 1996­2006 Total Entire RE portfolio Improve welfare 26 46 72 50.0 67.6 60.0 Energy supply 39 49 88 75.0 72.1 73.3 Institutional development 39 51 90 75.0 75.0 75.0 Nonelectrification objective 8 14 22 15.4 20.6 18.3 Total 52 68 120 100 100 100 Energy sector only Improve welfare 15 30 45 37.5 65.2 52.3 Energy supply 38 42 80 95.0 91.3 93.0 Institutional development 33 34 67 82.5 73.9 77.9 Nonelectrification objective 0 0 0 0.0 0.0 0.0 Total 40 46 86 100 100 100 Table B.6: Number and Percentage of Projects by Period and Component Category Number of projects Percentage Component 1980­95 1996­2006 Total 1980­95 1996­2006 Total Infrastructure 49 49 98 94.2 72.1 81.7 Institutional development 41 61 102 78.8 89.7 85.0 Project financing mechanisms 7 34 41 13.5 50.0 34.2 Consumer relations 4 26 30 7.7 38.2 25.0 Nonelectrification activities 13 18 31 25.0 26.5 25.8 Total number of projects 52 68 120 100 100 100 Table B.7: Proportion of Projects by Region and Component Category Other Other Non- Institutional electrification electrification electrification Region Infrastructure development components components activities Latin America and Caribbean 0.57 0.74 0.54 0.23 0.43 Middle East and North Africa 1.00 0.86 0.00 0.00 0.29 Sub-Saharan Africa 0.84 1.00 0.24 0.32 0.32 South Asia 1.00 0.76 0.41 0.29 0.06 East Asia and Pacific 0.91 0.88 0.24 0.26 0.12 Europe and Central Asia 1.00 1.00 0.50 0.00 0.50 Total 0.82 0.85 0.34 0.25 0.26 Note: Figures are proportions expressed as between 0 and 1 rather than 0 and 100. 71 THE WELFARE IMPACT OF RURAL ELECTRIFICATION Table B.8: Number and Percentage of Projects by Planned Use and Period Number of projects Percentage Planned use 1980­95 1996­2006 Total 1980­95 1996­2006 Total Residential 31 40 71 93.9 97.6 95.9 Agricultural 24 15 39 72.7 36.6 52.7 Commercial 12 7 19 36.4 17.1 25.7 SME 17 27 44 51.5 65.9 59.5 Public 5 22 27 15.2 53.7 36.5 Total number of projects 33 41 74 100 100 100 Note: SME = small and medium-size enterprises. Table B.9: Number of Projects by Planned Use and Region Planned use Region Residential Agricultural Commercial SME Public Total Latin America and Caribbean 20 14 1 14 12 35 Middle East and North Africa 4 2 0 2 2 7 Sub-Saharan Africa 13 3 2 9 8 25 South Asia 15 11 10 10 4 17 East Asia and Pacific 18 8 6 9 1 34 Europe and Central Asia 1 1 0 0 0 2 Total 71 39 19 44 27 120 Note: SME = small and medium-size enterprises. Totals may not sum across because each project can have more than one planned use. Table B.10: Number of Projects On and Off Grid Number of projects Percentage No off grid Off grid Total No off grid Off grid Total No grid 5 6 11 7.6 18.2 11.1 Grid 61 27 88 92.4 81.8 88.9 Total 66 33 99 100.0 100.0 100.0 Note: No grid = grid extension was not in project design. Table B.11: Projects with an Off-Grid Component by Period Number of projects Percentage 1980­95 1996­2006 Total 1980­95 1996­2006 Total No off-grid component 45 21 66 95.7 40.4 66.7 Off-grid component 2 31 33 4.3 59.6 33.3 Total 47 52 99 100 100 100 Note: No off grid = off-grid generation was not in project design. 72 APPENDIX B: STATISTICAL OVERVIEW OF PORTFOLIO Table B.12: Number and Percentage of Projects by Promotional Activities and Approval Period Number of projects Percentage Type of promotional activity 1980­95 1996­2006 Total 1980­95 1996­2006 Total Primary Workshops to educate consumers 0 8 8 0.0 11.8 6.7 Newspaper/radio/TV broadcast to educate consumers 0 6 6 0.0 8.8 5.0 Demonstrations/village visits 0 1 1 0.0 1.5 0.8 Total 0 15 15 0.0 22.1 12.5 Secondary Workshops to educate consumers 0 1 1 0.0 1.5 0.8 Newspaper/radio/TV broadcast to educate consumers 4 7 11 7.7 10.3 9.2 Demonstrations/village visits 0 1 1 0.0 1.5 0.8 Total 4 9 13 7.7 13.2 10.8 Tertiary Demonstrations/village visits 0 4 4 0.0 5.9 3.3 Total 0 4 4 0.0 5.9 3.3 Table B.13: Number and Percentage of Projects by Region and Scope Number of projects Percentage Larger Larger energy energy project project Dedicated w/ RE Multisectoral Dedicated w/ RE Multisectoral Region to RE component project Total to RE component project Total Latin America and Caribbean 8 6 21 35 22.9 17.1 60.0 100.0 Middle East and North Africa 2 4 1 7 28.6 57.1 14.3 100.0 Sub-Saharan Africa 6 15 4 25 24.0 60.0 16.0 100.0 South Asia 8 8 1 17 47.1 47.1 5.9 100.0 East Asia and Pacific 18 10 6 34 52.9 29.4 17.6 100.0 Europe and Central Asia 0 1 1 2 0.0 50.0 50.0 100.0 Total 42 44 34 120 35.0 36.7 28.3 100.0 Note: RE = rural electrification. 73 THE WELFARE IMPACT OF RURAL ELECTRIFICATION Table B.14: Number and Percentage of Projects by Period and Scope Number of projects Percentage Project scope 1980­95 1996­2006 Total 1980­95 1996­2006 Total Dedicated to RE 17 25 42 32.7 36.8 35.0 Larger energy project w/RE component 23 21 44 44.2 30.9 36.7 Multisectoral project 12 22 34 23.1 32.4 28.3 Total 52 68 120 100.0 100.0 100.0 Note: RE = rural electrification. Table B.15: Number of Beneficiaries Number Number of projects for of beneficiaries which data available Villages 137,173 15 New customers 35,000 1 Rural customers 383,179 6 Off-grid customers 22,685 1 Off-grid villages 46 1 RET-supplied customers 11,230 2 Newly electrified town 150 1 People 18,600,000 13 Households 17,200,000 38 Note: RET = renewable energy technology. Table B.16: Lines and Substations Built Number of projects for Built which data available Kilometers of lines 607,069 55 Number of substations 526 20 74 APPENDIX B: STATISTICAL OVERVIEW OF PORTFOLIO Table B.17: Projects by Objective Category and Scope Number of projects Column percentage Larger Larger energy energy project project Dedicated w/RE Multisectoral Dedicated w/RE Multisectoral Objective to RE component project Total to RE component project Total Improve welfare 27 18 27 72 64.3 40.9 79.4 60.0 Energy supply 38 42 8 88 90.5 95.5 23.5 73.3 Institutional development 30 37 23 90 71.4 84.1 67.6 75.0 Nonelectrification objective 0 0 22 22 0.0 0.0 64.7 18.3 Total number projects 42 44 34 120 100.0 100.0 100.0 100.0 Note: RE = rural electrification. Table B.18: Projects with Poverty-Reduction Objective by Scope Number of projects Percentage of projects w/poverty-reduction w/poverty-reduction objective Total objective Dedicated to RE 3 42 7.1 Larger energy project w/RE Component 3 44 6.8 Multisectoral project 20 34 58.8 Total 26 120 21.7 Note: RE = rural electrification. Table B.19: Number of Projects by Scope and Rating Large energy Dedicated project Multisectoral Rating to RE w/RE component project Total Highly satisfactory 1 3 1 5 Satisfactory 13 12 12 37 Moderately satisfactory 0 6 3 9 Moderately unsatisfactory 1 2 1 4 Unsatisfactory 5 8 3 16 Total 20 31 20 71 Note: RE = rural electrification. 75 THE WELFARE IMPACT OF RURAL ELECTRIFICATION Table B.20: Economic Rate of Return for Grid and Off-Grid Components Country Project Grid Minigrid Off-grid Cambodia Rural Electrification and Transmission 19.8 22.3 11 Honduras Rural Infrastructure Project 33 20 30 India Renewable Resources Development 14 n.a. 41 Lao PDR Rural Electrification Phase 1 687a n.a. 36 Lao PDR Southern Provinces Rural Electrification 60.5 n.a. 26 Mozambique Energy Reform and Access 22.7 n.a. 14.5 Nicaragua Off-Grid Rural Electrification 40 23 27 Peru Rural Electrification Project 22.6 n.a. 23.8 Philippines Rural Power Project 26.4 21.5 48 Sri Lanka Renewable Energy for Rural Energy 15.2 n.a. 10.9 Sri Lanka Energy Services Delivery Project 23.5 n.a. 8.5 Source: Project documents. a. See project PAD, p. 12. Table B.21: Connections by Type Connection Public Industrial/ Percent Country Project Domestic institutions Agricultural commercial residential Tunisia Fourth Power Project 35,000 0 1500 50 95.8 Thailand Second Accelerated Rural Electrification 534,496 0 463 26,809 95.1 Pakistan Rural Electrification 153,000 0 127 299 99.7 Mali Household Energy and Universal Access 40,000 1,199 0 0 97.1 76 APPENDIX B: STATISTICAL OVERVIEW OF PORTFOLIO Table B.22: Utility Cost Recovery by Income Level and Region Percentage of utilities whose average tariffs appear to be... Too low to cover Enough to cover Enough to cover Electricity utilities basic O&M costs most O&M O&M and partial capital Global 15 44 41 By income High 0 17 83 Upper-middle 0 71 29 Lower-middle 27 50 23 Low 31 44 25 By Region OECD 0 17 83 Latin America and the Caribbean 0 47 53 Eastern and Central Asia 31 38 31 East Asia and Pacific 29 65 6 Sub-Saharan Africa 29 71 0 South Asia 33 67 0 Source: Komives and others 2005, p. 26. Note: O&M = operation and management. Table B.23: Number and Percentage of Projects with RET by Period and Region Number of projects Percentage Region 1980­95 1996­2006 Total 1980­95 1996­2006 Total Latin America and Caribbean 0 11 11 0.0 26.2 18.3 Middle East and North Africa 1 1 2 5.6 2.4 3.3 Sub-Saharan Africa 5 13 18 27.8 31.0 30.0 South Asia 4 6 10 22.2 14.3 16.7 East Asia and Pacific 8 11 19 44.4 26.2 31.7 Total 18 42 60 100.0 100.0 100.0 Note: RET = renewable energy technology. Table B.24: Number and Percentage of Projects with RET by Period and Scope Number of projects Percentage Region 1980­95 1996­2006 Total 1980­95 1996­2006 Total Dedicated to RE 4 21 25 22.2 50.0 41.7 Larger energy project w/RE component 12 14 26 66.7 33.3 43.3 Multisectoral project 2 7 9 11.1 16.7 15.0 Total 18 42 60 100.0 100.0 100.0 Note: RE = rural electrification; RET = renewable energy technology. 77 THE WELFARE IMPACT OF RURAL ELECTRIFICATION Table B.25: Number and Percentage of Projects with RET by Period and Scope Number of projects Percentage Larger energy Multi- Larger energy Multi- Dedicated project w/RE sectoral Dedicated project w/RE sectoral Region to RE component project Total to RE component project Total Latin America and the Caribbean 6 1 4 11 54.5 9.1 36.4 100.0 Middle East and North Africa 0 2 0 2 0.0 100.0 0.0 100.0 Sub-Saharan Africa 5 11 2 18 27.8 61.1 11.1 100.0 South Asia 3 7 0 10 30.0 70.0 0.0 100.0 East Asia and Pacific 11 5 3 19 57.9 26.3 15.8 100.0 Total 25 26 9 60 41.7 43.3 15.0 100.0 Note: RE = rural electrification; RET = renewable energy technology. Table B.26: Projects with Off-Grid Components and Type of Energy Source Infrastructural Country MH PV W Bio D D/RET outputs Argentina Renewable Energy in the Rural 65,500 SHS, Market Project 2 wind home systems Bangladesh Rural Electrification and Renewable 64,000 SHS Energy Development Bolivia Decentralized Infrastructure for 15,000 SHS Rural Transformation Brazil Bahia Rural Poverty Reduction 16,000 SHS Cambodia Rural Electrification and Transmission 12,000 SHS 0.85 MW micro hydro Cape Verde Energy and Water Sector Reform and 7.8 MW wind energy Development Project Chile Infrastructure for Territorial Development Ecuador Power and Communications Sectors Modernization and Rural Services Honduras Rural Infrastructure Project 5,000 SHS 2 micro hydro systems India Renewable Resources Development Project 2.145 MW 87.2 MW wind energy Indonesia Solar Homes System Project 8,054 SHS Lao PDR Southern Provinces Rural Electrification Project Lao PDR Rural Electrification Phase 1 9,000 SHS (Table continues on next page) 78 APPENDIX B: STATISTICAL OVERVIEW OF PORTFOLIO Table B.26: Projects with Off-Grid Components and Type of Energy Source (continued) Infrastructural Country MH PV W Bio D D/RET outputs Madagascar Energy Sector Development Project 82 diesel units Malaysia Rural Electrification Project Mali Household Energy and Universal Access 10,150 SHS Mexico Renewable Energy for Agriculture 1,545 SHS 92 wind systems 150 MW from solar/diesel hybrid Mexico Waste Management and Carbon Offset Mozambique Energy Reform and Access 2,800 SHS Nepal Power Development 0.2 MW solar/diesel hybrid 125 micro hydro systems Nicaragua Off-grid Rural Electrification 6 solar charging stations 7 micro hydro systems Nigeria National Energy Development Project 1 SHS Peru Rural Electrification Project 15 MW micro hydro Philippines Rural Power Project Senegal Electricity Services for Rural Areas Project Sri Lanka Energy Services Delivery Project 20,953 SHS 0.35 MW micro hydro Sri Lanka Renewable Energy for Rural Economic 85,000 SHS Development Project 21 MW wind energy Tanzania Songo Songo Gas Development and 1,600 SHS Power Generation Uganda Energy for Rural Transformation 7,496 SHS Vietnam Community Based Rural Infrastructure 1 SHS Vietnam System Efficiency, Equitization, and Renewables Vietnam Rural Energy Number of projects 15 24 10 4 2 5 As a percentage of RET projects 46.9 75.0 31.3 12.5 6.3 15.6 Note: Bio = biomass; D = diesel; D/RET = diesel/renewable hybrid; MH = micro hydro; MW = megawatts; PV = photovoltaic; RET = renewable energy technology; SHS = solar home system; W = wind. 79 THE WELFARE IMPACT OF RURAL ELECTRIFICATION Table B.27: Gender as a Design Feature in Projects 1980­96 1996 onward RE Other RE Other projects energy Multisector projects energy Multisector No mention of gender 15 18 4 12 12 7 Gender mentioned, but did not 2 2 1 7 6 2 influence design Gender influenced design 0 3 7 6 3 13 Total 17 23 12 25 21 22 Note: RE = rural electrification. Table B.28: Number and Percentage of Projects by Rating and Objective Category Increase Energy Institutional Other Rating welfare supply development (non RE) Total Number of projects Highly satisfactory 3 3 5 1 12 Satisfactory 19 25 29 8 81 Moderately satisfactory 5 8 7 1 21 Moderately unsatisfactory 1 3 4 1 9 Unsatisfactory 8 13 10 2 33 Total 36 52 55 13 156 Percentage (by objective) Highly satisfactory 8.3 5.8 9.1 7.7 7.7 Satisfactory 52.8 48.1 52.7 61.5 51.9 Moderately satisfactory 13.9 15.4 12.7 7.7 13.5 Moderately unsatisfactory 2.8 5.8 7.3 7.7 5.8 Unsatisfactory 22.2 25 18.2 15.4 21.2 Total 100 100 100 100 100 Note: RE = rural electrification. Totals may not sum 100 exactly because of rounding. 80 APPENDIX B: STATISTICAL OVERVIEW OF PORTFOLIO Table B.29: Comparisons for PV System Retail Prices Watt peak China Philippines Indonesia Sri Lanka India Kenya Zambia 10 85 15 120 20 150 300 302 300 30 203 40 520 303 419 307 50 660 300­408 480 360 822 75­80 640 750­1000 686 Source: World Bank 2002. Table B.30: Connection Costs and Charges in Project Countries Country Year Connection cost Connection charge Morocco 1979 350 300 Thailand 1980 98 98 Thailand 1982 348 Tunisia 1984 82 urban/164 rural Pakistan 1989 45 Thailand 1989 22 Burundi 1991 500­750 Malaysia 1992 1,284 Malawi 1992 19 Brazil 1993 393 Jordan 1994 135 Morocco 1998 --monthly charge of $4.20 over 7 years-- Bangladesh 2001 417 Pakistan 2001 1,729 Lao PDR 2002 746 urban/1,047 rural ~100 Cambodia 2003 214 Bolivia 2003 150 Eritrea 2004 68 Lao PDR 2005 75­122 Honduras 2005 300­400 urban/768 rural Ethiopia 2006 --500 (50­100 in electrified village)-- Peru 2006 940 81 THE WELFARE IMPACT OF RURAL ELECTRIFICATION Table B.31: Subsidy Types for Off-Grid Project Components Country Project Subsidy type Argentina Renewable Energy in the Rural Market Project Declining capital cost subsidy--OBA approach Bangladesh Rural Electrification and Renewable Energy Development Declining capital cost subsidy Bolivia Decentralized Infrastructure for Rural Transformation Capital cost subsidy Brazil Bahia Rural Poverty Reduction No mention Cambodia Rural Electrification and Transmission Capital cost subsidy Cape Verde Energy and Water Sector Reform and Development Project Declining capital cost subsidy Chile Infrastructure for Territorial Development Capital cost subsidy China Renewable Energy Development Subsidized loans for technology development Ecuador Power and Communications Sectors Modernization and Rural Services Capital cost subsidy for demonstration projects Honduras Rural Infrastructure Project Capital cost subsidy--OBA approach, micro credit to providers India Renewable Resources Development Project Capital cost subsidy 50% and subsidized credit scheme Indonesia Solar Homes System Project Credit at market rates Lao PDR Southern Provinces Rural Electrification Project No mention Lao PDR Rural Electrification Phase 1 Capital cost subsidy 100% Madagascar Energy Sector Development Project No mention Malaysia Rural Electrification Project No mention Mali Household Energy and Universal Access Capital cost subsidy--OBA approach Mexico Renewable Energy for Agriculture Matching grants for capital investments Mexico Waste Management and Carbon Offset No subsidy Mozambique Energy Reform and Access Capital cost subsidy 50%--OBA approach Nepal Power Development No mention Nicaragua Off-Grid Rural Electrification Capital cost subsidy for part, down payment by consumer 5­10%, micro credit to spread rest out over 3 years Nigeria National Energy Development Project Some subsidy Peru Rural Electrification Project Capital cost subsidy--average $457 Philippines Rural Power Project Some targeted subsidies Senegal Electricity Services for Rural Areas Project Capital cost subsidy--OBA approach Sri Lanka Energy Services Delivery Project No subsidy Sri Lanka Renewable Energy for Rural Economic Development Project Declining scale capital cost subsidy--OBA approach; micro credit to providers Tanzania Songo Songo Gas Development and Power Generation No mention Uganda Energy for Rural Transformation Capital cost subsidy and credit support facility Vietnam Community Based Rural Infrastructure No mention Vietnam Rural Energy No mention Vietnam System Efficiency, Equitization, and Renewables Capital cost subsidy Note: OBA = output-based aid. 82 APPENDIX C: DISTRIBUTIONAL ANALYSIS OF RURAL ELECTRIFICATION Intensive versus Extensive Growth in ual. These two terms may be interpreted as in- Electrification Coverage tensification and extension, respectively. That is, The rural electrification (RE) rate may be broken coverage may expand either by households in down as follows: connected villages taking connections or by ex- tending connections to new villages. P P Pec e = e , (C.1) P Pec P Table C.1 summarizes the analysis of this de- where Pe are households with electricity, Pec the composition. As might be expected, when the population of communities with electricity, and P coverage rate is low, then the bulk of the increase the total population. The first term on the right- comes from extension, meaning that there re- hand side of the equation is the electrification rate mains a sizeable number of unconnected house- in electrified communities and the second term holds in communities with electricity. This was is the share of the population of electrified evident in figure C.1, looking at the ratio Pe/Pce, communities in the total population (that is, the where in Bangladesh and Nepal only about half population-weighted proportion of electrified of households in electrified communities have communities). electricity. Only in the Philippines, where more than half the population lives in connected com- Figure C.1 shows these three terms for four coun- munities, does the majority of the increase in tries at two times using Demographic and Health coverage come from intensification. Survey (DHS) data. The final bar in each set shows coverage of rural communities: the Philippines had achieved a high level of coverage (more than Lorenz Curves 80 percent of the population lived in communi- Which households are connecting to the grid can ties that had electricity) in the first year shown be further analyzed using Lorenz curves. In the (1993), and Bangladesh achieved a substantial first set of curves presented here, households increase in the proportion of the population liv- are ranked according to a wealth index, starting ing in electrified villages, from 52 to 90 percent, with the poorest. The curve plots the cumulative between 1992 and 2004. share of electrified households against the cu- mulative share of households (ranked by wealth). A closer examination of the change in the electri- If this line lies below the 45o line, then poorer fication rate can be made through a decomposi- households are less likely to have electricity. DHS tion analysis. From equation C.1, the percentage data are used to show these curves for five coun- change in the electrification rate can be decom- tries (figure C.2), drawn for rural areas only. All posed as follows: graphs show the same two points: (1) access to the grid is bias toward the non-poor, and (2) the P^ P ^ P ^ec e = e + , (C.2) bias reduces over time (as the electrification rate P Pec P increases) (Nepal is an exception). This reduction where the ^ denotes the percentage change in the in bias is most marked in Ghana, where the gov- ratio. For large changes there will also be a resid- ernment embarked on a program to bring 83 THE WELFARE IMPACT OF RURAL ELECTRIFICATION Figure C.1: Rural Electrification Rate Ratios 1.0 0.9 0.8 0.7 0.6 0.5 Ratio 0.4 0.3 0.2 0.1 0.0 Pe/P Pe/Pce Pce/P Pe/P Pe/Pce Pce/P Pe/P Pe/Pce Pce/P Pe/P Pe/Pce Pce/P Bangladesh Nepal Peru Philippines Earliest year Earliest year Note: P = total population, Pce = population of communities with electricity; Pe = households with electricity. Table C.1: Decomposition of Changes in Rural Electrification Rate Electrification rate Latest Percent Earliest year increase Change in rate attributable to: Country Period year (%) (%) in rate Intensification Extension Residual Bangladesh 1992­2004 10 30 207 28 46 26 Nepal 1996­2001 12 17 45 ­10 116 ­5 Peru 1992­2004 19 35 84 20 68 12 Philippines 1993­2003 46 60 29 56 37 6 Source: DHS data. electricity to all districts. This pattern is also shown or not (see figure 3.2 in the main report for the in table C.2, which reports the Gini coefficient (an cases of Lao PDR and the Philippines). inequality measure between 0 and 1, which would be 0 if grid access were equitably distributed). In one case (Sri Lanka) there are sufficient ob- servations to draw separate curves for grid and off- For the countries for which an energy survey is grid connections (see figure C.3). The sample for available, the Lorenz curve may also be calculated this survey is not representative--connected for electricity consumption. Because the better-off households are oversampled--and survey weights consume greater amounts of electricity, the dis- are not available, so the general shape of the line tribution of consumption is more unequal than is is misleading (the true share of unconnected is shown by the dichotomous variable of connected understated, making the distribution look more 84 APPENDIX C: DISTRIBUTIONAL ANALYSIS OF RURAL ELECTRIFICATION Figure C.2: Electrification Lorenz Curves (A) Bangladesh (B) Ghanaa 1.0 1.0 0.8 0.8 of of 0.6 0.6 share share households 0.4 households 0.4 0.2 0.2 Cumulative Cumulative electrified 0.0 electrified 0.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 Cumulative population share Cumulative population share 1992 2000 2004 45° 1988­89 2003 45° (C) Nepal (D) Peru 1.0 1.0 0.8 0.8 of of 0.6 0.6 share share households 0.4 households 0.4 0.2 0.2 Cumulative Cumulative electrified 0.0 electrified 0.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 Cumulative population share Cumulative population share 1996 2001 45° 2000 2004 45° (E) Philippines 1.0 0.8 of 0.6 Access to electricity is biased toward the better off. But that bias reduces as electrification rates increase. share households 0.4 0.2 Cumulative electrified 0.0 0.0 0.2 0.4 0.6 0.8 1.0 Cumulative population share 1993 1998 2003 45° Source: DHS data. a. Sources for this section are Ghana Livings Standards Survey and IEG 2004. 85 THE WELFARE IMPACT OF RURAL ELECTRIFICATION Table C.2: Gini Coefficients Electricity Pricing and Subsidy for Access to Electricity Distribution Economists generally argue that efficiency is ob- Country Year National Rural tained through marginal cost pricing. However, for Bangladesh 2000 0.40 0.44 industries with large fixed costs, there may be in- 2004 0.32 creasing returns to scale, so that average costs fall Ghana 1999 0.37 0.40 as supply expands. Under these circumstances 2003 0.22 0.20 marginal cost pricing implies running at a loss, and Ghanaa 1989 n.a. 0.58 in public economics this has long been taken as 2003 n.a. 0.20 a case for state subsidy. Governments have also Morocco 2004 0.17 0.21 been keen to subsidize to ensure public utilities Nepal 1996 0.48 0.38 achieve good coverage; once established, subsi- dies become entrenched and politically difficult 2001 0.52 0.49 to remove. Fiscal pressures in developing coun- Nicaragua 2001 0.32 0.44 tries have pushed some governments to seek Peru 2000 0.31 0.40 greater levels of cost recovery, but electricity sup- 2004 0.23 0.29 ply is still subsidized in virtually every country in Philippines 1998 0.24 0.33 the world. 2003 0.20 0.29 Senegal 2005 0.45 0.57 Foster and Yepes (2006) analyze cost structures Nepal 1996 0.48 0.38 to derive the following basic categorization: (1) tar- Source: DHS data. iffs below $0.04 per kilowatt hour (kWh) are a. Source for this category is Ghana Living Standards Survey. insufficient to cover basic operations and man- agement (O&M) costs for either residential or in- dustrial customers; (2) tariffs above $0.05/kWh Figure C.3: Lorenz Curve for Grid are sufficient to cover O&M and make a significant and Off-Grid Connections, Sri Lanka contribution toward capital costs for industrial users; and (3) tariffs above $0.08/kWh are sufficient 1.0 to cover O&M and make a significant contribution toward capital costs for residential users. This cat- 0.8 egorization applies to most, though not all, grid- of based distribution systems. 0.6 share population 0.4 Using these figures a regional classification can be produced, as shown in table C.3. There is a clear Cumulative connected0.2 pattern showing that countries in poorer Regions are less likely to achieve cost recovery, with nearly 0.0 a third of countries in East Asia and Pacific, Sub- 0.0 0.2 0.4 0.6 0.8 1.0 Saharan Africa, and South Asia charging insufficient Cumulative population share fees to cover O&M and virtually all the remainder Grid SHS 45° unable to make a contribution toward capital Source: IEG calculation from Sri Lanka household survey data. costs. Note: SHS = solar home system. Who benefits from this subsidy clearly depends both on who uses electricity and how much they equal than it is), but the curves for grid and off- use and on the nature of the subsidy scheme. grid can be compared. These curves show that off- Subsidies may be either to the cost of connection grid connections are less equitably distributed or on the tariff. A linear tariff (that is, a fixed amount than are grid connections. per kilowatt hour) will give the subsidy propor- 86 APPENDIX C: DISTRIBUTIONAL ANALYSIS OF RURAL ELECTRIFICATION Table C.3: Percentage of Countries Covering O&M and Capital Costs through Electricity Charges Too low to cover Enough to cover Enough to cover basic O&M most O&M O&M and partial capital OECD (n = 23) 0 17 83 Latin America and Caribbean (n = 19) 0 47 53 Europe and Central Asia (n = 18) 31 38 31 East Asia and Pacific (n = 13) 29 65 6 Sub-Saharan Africa (n = 8) 29 71 0 South Asia (n = 3) 33 67 0 Global (n = 84) 15 44 41 Source: Komives and others 2005, table 2.5. tional to consumption, or more than proportional the subsidy goes to the poor exactly in proportion if the tariff is a monthly fixed charge plus a linear to their population share (which is what a random tariff (because then the average charge per kilo- allocation would achieve). In contrast, ø < 1 means watt hour falls as consumption rises). But the ma- the poor get a lower share of the subsidy than their jority of electricity companies use differentiated population shares; that is, it disproportionately tariffs. benefits the non-poor. Quantity-based tariff sub- sidies have an average ø of just 0.62, though this These tariffs take two basic forms. Most common performance can be improved (to ø > 1) if com- is the increasing block tariff, whereby consumers bined with geographic targeting or means testing. pay more for units of consumption above a certain threshold. The alternative is a volume- The reason for this finding is clear. The poor are differentiated tariff, where those who consume less likely to be connected (as shown in the analy- above the threshold pay a higher rate on all con- sis above) and so are excluded from the subsidy sumption, not just that above the threshold. altogether. And the better-off consume more, so even if the tariff rises, as long as it is still subsidized, Komives and others (2005) present a compre- the non-poor may obtain a greater absolute hensive analysis of who benefits from electricity tar- benefit. iffs. The general finding is that these subsidies are poorly targeted, being less pro-poor than a simple Connection subsidies appear to be a different story, random allocation of money among the popula- although few studies have been done. Because tion. The targeting measure used is ø, which is the connection rates among the poor are low, even un- ratio of the share of the subsidy going to the poor targeted subsidies have ø > 1, with still higher to their share in the population. Hence if ø = 1, values if combined with geographic targeting. 87 APPENDIX D: USES OF ELECTRICITY Fuelwood Collection and Cooking East Asian countries may be something of an ex- Table D.1 shows that the time spent collecting ception, because rice cookers are a common pur- wood each day varies quite considerably. But the chase in electrified households: in Lao People's average time is more than 2.5 hours, and it can Democratic Republic just over a quarter (27 per- be eight times that. The largest share of the bur- cent) of electrified households have a rice cooker, den is carried by women. so electrified households do use significantly less fuelwood in these countries. If electrification means that less firewood is col- lected, then there are time savings to the house- holds. However, the main use for firewood is for Ownership of Electric Goods cooking, and, as table D.2 shows, in most coun- Figure D.1 uses DHS data for 53 countries, sev- tries less than 1 percent of the rural population eral with data from more than one survey, giving (and usually not a much larger share of those a sample of up to 113 observations. The graphs with electricity) uses it for cooking. Therefore, plot the percentage of people in rural areas own- these benefits will not be realized in the short to ing the specified good against the RE rate. The 45o medium term. line and fitted line are also shown. Table D.1: Time Spent Collecting Wood (hours/day) Country Boys Girls Men Women Total Benin n.a. n.a. 0.47 1.60 2.07 Benin n.a. n.a. 0.50 2.90 3.40 Ghana (rural)a n.a. n.a. 3.23 4.60 7.83 Guinea (rural) 2.50 1.50 1.60 2.40 8.00 Madagascar n.a. n.a. 1.35 0.82 2.17 South Africa n.a. n.a. 0.35 0.70 1.05 Indonesia n.a. n.a. 0.21 0.09 0.30 Burkina Faso n.a. n.a. 0.03 0.10 0.13 India n.a. n.a. 0.65 0.65 1.30 Nepal n.a. n.a. 0.83 2.37 3.20 Himachal Pradeshb n.a. n.a. n.a. n.a. 1.30 Tamil Naduc n.a. n.a. n.a. n.a. 0.65 Rajastanc n.a. n.a. n.a. n.a. 1.64 Sources: Bardasi and Wodon 2006; Charmes 2006; Dutta 2005; GSS 2000; Nathan and Kelkar 1997; Laxmi and others 2003; Parikh 2005; Parikh and Laxmi 2000; Blackden and Wodon 2006. Note: n.a. = not available. a. Average for forest and savannah. b. Average across Regions. Gender division of labor varies by Region. Daily average over a month. c. Average across Regions. Daily average over a month. 89 THE WELFARE IMPACT OF RURAL ELECTRIFICATION Table D.2: Use of Electricity for Cooking Percent of those with electricity using Percent of rural population Percent of total population electricity for cooking Use electricity Use electricity Country Year Electricity for cooking Electricity for cooking Rural Total Armenia 2000 98.6 20.4 98.9 37.4 20.7 37.8 Bangladesh 2004 30.4 0.0 40.6 0.1 0.0 0.2 Bolivia 1998 29.0 0.0 71.2 0.9 0.0 1.3 Bolivia 2003 35.7 0.0 72.3 0.5 0.0 0.7 Burkina Faso 2003 0.8 0.0 11.4 0.0 0.0 0.0 Cambodia 2000 9.0 0.1 16.6 0.1 1.1 0.6 Colombia 2000 83.8 5.0 95.2 13.8 6.0 14.5 Colombia 2005 89.2 4.1 96.8 7.2 4.6 7.4 Dominican Republic 2002 80.6 0.1 92.3 0.1 0.1 0.1 Egypt, Arab Rep. of 2000 95.9 0.3 97.7 0.4 0.3 0.4 Eritrea 2002 3.0 0.0 32.2 0.6 0.0 1.9 Ethiopia 2000 0.4 0.0 12.7 0.1 0.0 0.8 Ghana 2003 24.1 0.1 48.3 0.3 0.4 0.6 Haiti 2000 5.2 0.0 33.7 0 0.0 0.0 India 1998 48.1 0.2 60.1 0.4 0.4 0.7 Indonesia 2002 84.5 0.2 90.7 0.4 0.2 0.4 Jordan 2002 98.7 0.0 99.5 0.1 0.0 0.1 Kenya 2003 4.6 0.1 16 0.3 2.2 1.9 Madagascar 2003 10.8 0.1 20.3 0.3 0.9 1.5 Malawi 2000 1.0 0.2 4.8 2.1 20.0 43.8 Mali 2001 2.2 0.0 10.8 0 0.0 0.0 Morocco 2003 51.3 . . 78.2 7.8 . . 10.0 Mozambique 2003 1.1 0.2 8.1 0.8 18.2 9.9 Namibia 2000 13.2 5.5 36.5 26.4 41.7 72.3 Nepal 2001 17.4 0.0 24.6 0.1 0.0 0.4 Nicaragua 2001 40.1 0.1 72.6 0.9 0.2 1.2 Nigeria 2003 33.8 0.1 52.2 0.3 0.3 0.6 Peru 2000 28.9 0.0 69.3 1.1 0.0 1.6 Rwanda 2000 0.9 0.0 6.2 0.2 0.0 3.2 Tanzania 2004 1.6 0.0 11.4 0.3 0.0 2.6 Turkmenistan 2000 99.6 0.4 99.6 0.5 0.4 0.5 Uganda 2000 2.4 0.1 8.6 0.8 4.2 9.3 Zambia 2001 2.9 1.9 17.4 14.1 65.5 81.0 Zimbabwe 1999 8.3 1.9 38.4 27.4 22.9 71.4 Source: DHS data: ORC Macro, 2006. MEASURE DHS STATcompiler. http://www.measuredhs.com, October 28, 2006. Note: . . = excluded because of discrepancy between published data and IEG calculations. 90 APPENDIX D: USES OF ELECTRICITY Figure D.1: Electrification and Consumer Goods Ownership in Rural Areas (A) Radio (B) TV 100 100 90 90 80 80 70 70 (%) (%) 60 60 50 50 40 40 Ownership Ownership 30 30 20 20 10 10 0 0 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 Electrification rate (%) Electrification rate (%) (C) Refrigerator 100 90 80 70 (%) The graphs plot the percentage of people in rural areas owning the 60 specified good against the RE rate. The 45o line and fitted line are 50 also shown. As expected, electrification is close to a necessary con- 40 dition for TV and refrigerator ownership (no points above the 45o Ownership 30 line) but not for radio, for which there is a very weak association between ownership and electrification. 20 10 0 0 10 20 30 40 50 60 70 80 90 100 Electrification rate (%) Source: DHS data (MEASURE DHS STATcompiler, http://www.measuredhs.com, March 8 2007). As expected, electrification is close to a necessary quire a TV in most countries (again Indonesia condition for TV and refrigerator ownership (no and Vietnam are outliers, but not as extreme as points above the 45o line). But it is not a sufficient in the case of fridge ownership). By contrast, condition, because many data points lie below the there is a very weak association between radio 45o line. This is especially so for refrigerator own- ownership and electrification. Many observations ership, which only rises above low levels for coun- lie above the 45° line, because radio ownership is tries with very high electrification rates--reflecting, not dependent on a grid connection. of course, the income effect. The two relatively poor countries (Indonesia and Vietnam) with high Because the surveys are relatively close together, electrification rates have low fridge ownership. many variables can be controlled for by modeling the changes in the variables between survey rounds In contrast, TV ownership rises quite quickly with (changes in other determinants will generally be electrification; the data support the argument small compared with the intracountry variation). The that the majority of electrified households ac- results from regressing the change in ownership of 91 THE WELFARE IMPACT OF RURAL ELECTRIFICATION TV watching is unsurprisingly associated with TV Table D.3: Change in Electric Goods ownership. Table D.4 reports DHS data (women's Ownership as a Function of Change in survey) for nine countries, showing this associa- Electrification Rate tion, which is summarized in figure D.2. On av- Radio TV Refrigerator erage, 60 percent of women in households with Intercept ­0.03 3.42* 3.01* TV watch almost every day, whereas less than 10 (­0.56) (2.66) (4.24) percent of women in households without TV Coefficient 6.15* 0.47** 0.93* watch that frequently. These data thus support the (1.24) (2.48) (4.99) argument that "communal viewing," while not R2 0.00 0.18 0.47 unknown, is not the norm. N 59 59 59 Note: Ordinary least squares with robust standard errors. Figures in paren- theses are t-statistics. Health Facilities *Significant at 1 percent. RE is argued to facilitate the cold chain and so im- **Significant at 5 percent. prove health status. Analysis of health facility data each of the three changes in electrification are from six countries shows that the cold chain is shown in table D.3. Consistent with the above analy- significantly more present in electrified rural fa- sis, there is a significant impact on TV and refrig- cilities than in those without electricity (see table erator ownership but not for radio. D.5)--this difference is significant in all six Table D.4: TV Watching by TV Ownership (Women) Frequency of watching Less than At least Television once a once a Almost Country ownership Not at all week week every day Number Bangladesh Do not own TV 64.2 10.6 19.9 5.2 6,193 Own TV 9.7 5.7 19.0 65.4 1,343 Ghana Do not own TV 70.4 15.4 10.1 4.0 2,985 Own TV 13.7 14.9 24.3 47.1 329 Indonesia Do not own TV 30.3 32.1 18.4 18.8 9,279 Own TV 2.0 7.4 6.4 83.8 7,886 Morocco Do not own TV 49.3 6.5 11.5 32.7 4,715 Own TV 5.4 3.5 9.9 81.2 3,085 Nepal Do not own TV 90.4 0.0 9.6 0.0 6,943 Own TV 12.4 0.0 87.6 0.0 629 Nicaragua Do not own TV 84.6 3.2 4.2 7.9 4,111 Own TV 5.1 2.0 9.1 83.7 1,662 Peru Do not own TV 50.2 43.9 1.3 4.5 9,751 Own TV 3.3 33.1 1.8 61.7 5,735 Philippines Do not own TV 52.3 31.3 9.5 6.4 7,843 Own TV 3.9 47.9 7.8 40.0 5,601 Senegal Do not own TV 58.0 18.2 13.9 9.8 6,071 Own TV 5.6 9.3 17.4 67.7 2,218 Unweighted average Do not own TV 61.1 17.9 11.0 9.9 Own TV 6.8 13.8 20.4 59.0 Source: IEG analysis of DHS data. 92 APPENDIX D: USES OF ELECTRICITY Figure D.2: TV Watching Is Far Greater in TV-Owning Households 70 60 50 40 watching 30 Percent 20 10 0 Not at all Less than once At least once Almost every day a week a week Watching frequency Do not own TV Own TV Source: IEG analysis of DHS data. Table D.5: Electrification of Rural Health Clinics and the Cold Chain by Country Ghana (2003) Egypt (2002) Kenya (2004) No No No Electricity electricity Electricity electricity Electricity electricity Electricity 72.8 27.2 98.6 1.4 77.5 22.5 Refrigerator 64.2 40.7*** 51.3 0.0*** 71.9 67.3 Ice 2.6 6.2* 0.6 0.0 0.6 0.0 No storage 21.9 37.2** 11.6 0.0*** 3.3 7.1*** Immunization 88.7 84.1 63.4 0.0*** 75.7 74.5 Electrified (%) Not electrified (%) Bangladesh (2000) Cold chain equipment available and operational 55.9 10.0*** Nicaragua (2001) Electric refrigerator 65.5 10.2*** Solar refrigerator 15.2 20.7* Any refrigerator 80.7 31.0*** Cold box 29.2 14.1* Termo 92.6 65.9*** Rwanda (2001) Refrigerator 80.8 96.6*** Immunization 80.8 98.8*** Source: DHS data except Nicaragua, from Measure Evaluation Health Facility Survey. *Significant difference at 10 percent. **Significant difference at 5 percent. ***Significant difference at 1 percent. 93 THE WELFARE IMPACT OF RURAL ELECTRIFICATION Table D.6: Average Number of Hours from three countries show a significant difference Rural Clinics Are Open in source of water. The electrified facilities were more likely to have access to piped water (index Electricity No electricity = 1 for piped water; two-thirds for protected Bangladesh 7.1 6.1a well/borehole; one-third for unprotected well) Kenya 15.1 11.0a compared with the nonelectrified facilities. Avail- Source: DHS facility surveys. ability of water within 500 meters of the facility was a. Significant difference at 1 percent. significantly higher for electrified facilities in two of the four countries. However, availability of water countries. However, the difference in the pro- year 'round was better for electrified facilities only portion of clinics providing immunization ser- in one of four countries. Causality is not established vices does not vary according to electrification here, as both electricity and better water supply status--meaning that clinics are used for immu- may reflect some third factor. nization days and so forth. The impact of RE on immunization rates is thus not likely to be strong, but it can help build the immunization into rou- Access to Water tine service delivery at the clinic level and thus re- Water supply frequently needs pumping, which duce the cost of delivering immunization services. can be done using either diesel or electric pumps. Data are not available to explore this issue further. Figure D.3 shows a positive association between rural access to water and RE in cross-country It is also argued that electrification allows longer data. It might be thought that this is a spurious opening hours for community facilities. Data for correlation, explained by income per capita, which health facilities were only available for two coun- is an explanatory factor for both these variables. tries, but in both cases electrified clinics had sig- However, electrification remains a significant de- nificantly longer opening hours (table D.6). terminant of water supply once income is allowed for (table D.8). The relationship between water Finally, access to electricity may improve the water supply and electrification deserves further study supply (see table D.7). Data from health facilities but is not pursued in this report. Table D.7: Electrification of Rural Health Centers and Water Availability Water available on site Water available all year round Source of water index No No No Country and year Electricity electricity Electricity electricity Electricity electricity Ghana 2003 77.4 47.5a 37.7 24.6a Bangladesh 2000 0.54 0.38a Egypt 2002 96.3 33.3a 4.2 0.0 0.97 0.44a Kenya 2004 64.3 69.4 71.1 68.4 0.86 0.58a Rwanda 2001 16.4 13.8 58.6 52.1 Source: DHS facility surveys. a. Significant difference at 1 percent. 94 APPENDIX D: USES OF ELECTRICITY Table D.8: Regression of Rural Water Figure D.3: Rural Electrification and Access to Piped Supply Using Cross-Country Data Water Coefficient t-stat 1.0 Electrification rate 0.192 2.04a 0.9 Floor material ­0.026 ­0.34 0.8 Nonelectric assets ­0.025 ­0.10 0.7 GDP per capita 0.000 2.60a 0.6 Note: GDP = gross domestic product. water 0.5 a. Significant difference at 5 percent. Piped 0.4 0.3 0.2 0.1 0.0 0.0 0.2 0.4 0.6 0.8 1.0 Rural electrification (cumulative share) Source: DHS data. 95 APPENDIX E: LITERATURE SURVEY 97 98 THE Latin America and Caribbean WELF Topic Mexico Peru Colombia El Salvador Guatemala Costa Rica ARE Poverty targeting, Government, via the Fed- Large subsidy for poor USAID CARES project in- USAID CARES project in- Rural cooperatives give pro-poor govern- eral Power Commission, through "social tariffs" troduced the DAM, which troduced the DAM, which concessionary connection IMP ment policy, and subsidized capital costs on electricity use of up to allowed communities to allowed communities to fees to those below the ACT end-user and tariffs in rural areas. 30 kWh/month. Cost with be ranked on the benefit be ranked on the benefit poverty line (Barnes 2007, knowledge (Barnes 2007, p. 156). social tariff is more than cost ratio, thus prioritiz- cost ratio, thus prioritiz- p. 38). OF 20 times less than real ing those communities in ing those communities in cost, but lack of informa- which benefits will be which benefits will be RURAL tion on the amount of greatest (Shonder and greatest (Shonder and electricity that can be Wilbanks 1996). Wilbanks 1996). used within the minimum ELECTRIFICA tariff, so does not have intended effect. Sixty- eight percent pay the minimum tariff (Valencia and others 1990). TION Usage patterns In rural areas: household Electricity was mainly Increased use of appli- Rural families: lighting, use, especially lighting, used for lighting, leading ances increases produc- television, ironing, refriger- and pumping for irriga- to more leisure for men tive hours (Barnes 1988), ation, household appli- tion (Barnes 2007, and children but also through the use of hot ances, and cooking only p. 146). longer workdays, espe- plates, irons, and sewing when fuelwood is unavail- cially for women. Local machines. Results in able (wood preferred). economy is based on staying up later at night Larger farms: irrigation, agriculture, but there is (20 minutes more than milling refrigeration, saw- a missing link between non-electrified house- ing, welding, and so on. electrification and holds) (Barnes 1988). Dairy farms: milking ma- agriculture. chines, refrigerators, boil- ing water, fences. Ninety percent of electric- Communities benefited ity consumed by house- from public lighting, educa- holds, 5 percent commer- tional institutions, and cial, and 3 percent indus- night classes (Barnes 2007, trial. For commercial use, pp. 25, 34). there has been no change in use of electricity, but mainly in productivity (Va- lencia and others 1990). Time savings from Although there was use not collecting of hot plates by some biomass richer households, most households still used tra- ditional means for cook- ing (Barnes 1988). Promotional Solar program includes Each rural cooperative activities promotional information has an education commit- --how photovoltaic sys- tee to inform members of tems work, their benefits, technical information/ and how they differ from new ways of using elec- grid systems (Barnes tricity for residential, 2007). farm, and commercial use (Barnes 2007, p. 27). Household income Higher agricultural pro- Very little impact on agri- effect ductivity brought about cultural development by electrification raised (Barnes 1988) rural income (Barnes 2007). Productive uses: Absolute business activ- Small farms did not have agricultural ity was positively corre- a marked increase in pro- targeting lated with year of electri- ductivity once electrified fication; percent of (Barnes 2007, pp. 32, 34). households with business higher in electrified than in unelectrified villages; APPENDIX most business started after electrification. About two-thirds of sur- veyed businessmen indi- cated that they would E: face difficulty without LITERA electricity supply (Barnes 1988). TURE Productive uses: FUNDAP carried out pro- business targeting motion activities on ben- SUR (industry and/or efits of electrification and SME) uses of electrical machin- VEY 99 (Table continues on next page) 100 THE Latin America and Caribbean (continued) WELF Topic Mexico Peru Colombia El Salvador Guatemala Costa Rica ARE ery and gave loans to small businesses sup- IMP ported by technical assis- tance, targeting those ACT that would not otherwise OF have access to credit (creating a likely down- RURAL ward bias compared to control). The results sug- gest higher electricity ELECTRIFICA usage among loan bene- ficiaries and a conse- quent positive impact on income (Shonder and Wilbanks 1996). TION Health benefits More refrigeration for Lighting and vaccine cold Increase in the number of food and medicines and storage for small hospi- health centers (Barnes reduction in kerosene use tals; fuelwood is still the 2007, pp. 32, 34). lead to better health main source for cooking, (Barnes 2007). so there is no effect on indoor air quality; street lighting led to fewer at- tacks at night (Valencia and others 1990). Alternative energy Since 1990, the Federal Consortium of coopera- Power Commission has tives building a hydro- been incorporating re- plant to connect to grid newables in electrifica- for lower-cost energy. tion planning for rural Some users are also con- areas (Barnes 2007, necting SHS and wind to p. 150). grid (Barnes 2007, pp. 34, 40). South Asia Region Topic India Bangladesh Sri Lanka Poverty targeting, pro- poor government policy and end-user knowledge Usage patterns Increased working hours, because cooking is delayed to evening; increased productivity of women making handicrafts using night light; increased reading time for children and adults; increased fan use (Barnes 1988). Time savings from not Electricity did not replace firewood and kerosene as collecting biomass the energy source for cooking (Barnes 1988). Promotional activities RE resulted in positive agricultural productivity be- Potential new consumers are informed about how to cause it was complemented as part of development obtain service. Meetings are held with community programs (for example, hybrid seeds/fertilizer in the leaders and rural industries, farming groups, and Green Revolution) and government policies (credit for commercial leaders to disseminate key information electric pumps and subsidized tariffs, resulting in an and ensure that their interests are considered. Rural increase in the number of electrified fields/pump Electrification Board personnel organize membership sets in India from 500,000 in 1966 to 10 million in committees to encourage community participation. 1992) (Barnes 1988). India Grameen Shakti, an out- Awareness of key issues is also promoted through growth of the Grameen microfinance group, offers distribution of information bulletins and through con- APPENDIX improved credit terms for SHS, with concessional sumer meetings and village electrification commit- loan from GEF/IFC. The program includes intensive tees (Barnes 2007). outreach and a training program through 52 branch offices. A variety of artisan, rural industry, and agri- E: cultural uses are reported. Industrial charges are triple that of agricultural, but much lower compared LITERA to the alternative energy source, diesel (Barnes 1988). TURE Household income effect Impact on household income is indirect, through changes in irrigation patterns (that is, active promo- SUR tion of electric pumps since 1966) and agricultural in- 101 VEY (Table continues on next page) 102 THE South Asia Region (continued) WELF Topic India Bangladesh Sri Lanka ARE novations (increased access through improved distri- bution of fertilizer and hybrid seeds--Green Revolu- IMP tion). RE without pumping for irrigation means that ACT new agricultural technology is less likely to improve agricultural productivity. Growth of agricultural pump OF sets was slow, a gradual, cumulative process (Barnes RURAL 1988). Traditional methods of irrigation were re- placed by electrical pump sets in electrified villages, resulting in increased agricultural productivity. ELECTRIFICA Productive uses: Government policy was that electricity should be Promotion of productive uses has become priority for agricultural targeting used for agricultural and rural productivity, specifi- the Rural Electrification Board. This is reflected in the cally. RE found to have direct impact on investment in requirement that all rural communities meet certain pumps and then, indirectly, on well irrigation. Study revenue requirements. Emphasis has been on the suggests that electricity had an impact on multiple promotion of irrigation and rice milling (Barnes 2007). TION cropping although findings are not robust. Same for impact on level of agricultural innovations, which seem to reach high levels a number of years after electrification (Barnes and Binswanger 1986). RE was found to be positively associated with business de- velopment, resulting in an increase in both new busi- nesses and businesses using electricity. Electrified businesses have twice the capital, more employees, lower fuel costs, and higher productivity. The cottage industry also benefited from extended hours and im- proved income. In the short run, the cottage industry does not seem to be affected by the increase in rural industry due to electrification (Barnes 1988). Productive uses: busi- RE and education programs had mutually reinforcing In addition to the above, electricity cooperatives pro- Mini-grid was used for grinding mills, radio re- ness targeting (industry consequences, through lighting for reading and the vide technical assistance to potential small industrial pair shops, and carpenters, but its potential and/or SME) village environment (Barnes 1988). RE appeared to and commercial clients, for example, assistance in was not realized. Reasons for this include that lead to the setting up of a larger number of shops in purchasing of electric motors and machinery (Barnes the suppliers were not oriented to promoting the village and engendered a structural shift in em- 2007). productive uses, and locations using minihydro ployment to the tertiary sector. In individual service were remote so there was limited market establishments, RE seems to lead to larger firm size, access. For off-grid, there were few productive more items sold, longer working hours, higher in- uses because of low voltage of supply. come, and higher incomes per employee (Barnes and Binswanger 1986). Social service staffing Some substitution of electricity for kerosene in light- The proportion of households in the community with (education and health ing (Barnes 1988). electricity has a strong positive impact on the likeli- staff) hood of living in the community, which in turn re- duces absenteeism (Chaudhury and Hammer 2003). Asia Topic Indonesia Thailand China Poverty targeting, pro- Government policy was to target the early stages of Government plan was to electrify whole of Thailand by poor government policy, RE toward villages that were already "relatively well the end of the 20th century. However, within villages, and end-user knowledge developed"--villages in study were all within 10 km no major pro-poor policy was in effect, and as a result of the district capital and 20 percent below the a quarter of the households remain unconnected even poverty line as opposed to 45 percent, the rate in when the village has had electricity for 20+ years. This rural Java overall (Meier 2001). RE progress initially was largely because of high up-front costs and (the 1980s) was in areas that had potential--first, monthly bills and inappropriate dwellings (Piampiti ones near the grid, second, ones with non-oil energy and others 1982). The government provided solar bat- potential, and third, others using diesel power tery charging stations to indigenous peoples in Thai- (Barnes 1988). land (these are village-level systems, from which households take electricity rather than household level APPENDIX SHS) (Green 2005). Usage patterns High use of appliances such as TV, fridge, sewing Increased use of appliances (for example, rice cook- Stage 1: (1949­57): Lighting and food process- machine, and water pumps (Barnes 1988) ers) and productive hours (Cecelski 1992). Electrified ing. Stage 2: (1958­78) also: Flood prevention, E: villages are generally better equipped than nonelec- irrigation, productive uses (see below) (Barnes LITERA trified: 37:26 piped water; 65:29 concrete roads; 2007). 40:14 health centers; 8:0 secondary schools. Econo- metric modeling indicates that the availability of TURE electricity had a positive impact on migration, though not on the crude birth rate at village level, couples' income, female labor force participation, fertility, SUR mortality, and migration at the household level (Pi- 103 VEY (Table continues on next page) 104 THE Asia (continued) WELF Topic Indonesia Thailand China ARE ampiti and others 1982). There was no observable ef- fect on child literacy rates, and there are a lot of IMP problems with low educational achievement in this ACT Region--having electricity is a low priority for help- ing education (Green 2005). OF RURAL Time savings from not Households rarely use electricity for cooking; they collecting biomass typically use charcoal or firewood instead (Piampiti and others 1982). ELECTRIFICA Promotional activities Indonesia subsidizes diesel and has high tariff and Stage 2: As part of strategy to promote pro- installation charges for electricity, which makes elec- ductive uses of RE, the government organized tric power an unattractive option (Barnes 1988). workshops to disseminate successful experi- ences in the development of small hy- dropower. Also developed a quota scheme for TION raw materials, giving priority to the rural in- dustrial sector. Financial and technical support for local manufacturers producing goods needed for development of RE. Stage 3: Start- ing in 1950 the government used professional personnel to disseminate technologies through national and provincial meetings, training courses, and workshops (Barnes 2007). Household income effect Very little impact on agricultural development Irrigation is insufficient in villages in northeast Thai- Stage 2: RE to support irrigation and agricul- (Barnes 1988) land (electrified and nonelectrified) (Piampiti and oth- tural mechanization and water conservation. ers 1982). There was no income-generation effect, as General: Subsidies for sectors that consider people did not have access to appliances, and the energy critical including agricultural product power supply from SCBS was too low for many appli- processing (Barnes 2007). ances anyway. It was also risky because supply was not reliable (Green 2005). Productive uses: agricul- Electrification resulted in extension of work hours, Noted increase in household industry (Cecelski 1992). Stage 2: RE as support for development of tural targeting but not employment. Electric tools not used, except About a third of the industries electrified in the elec- rural small industry and steel making (Barnes for lighting (Barnes 1988). Business development hin- trified villages. 2007) dered by subsidization of diesel, capital scarcity, poor product marketing channels, and inadequate informa- tion (Barnes 1988). Productive uses: Kiosk owners reported sales increases, as they could Staffing of doctors and nurses was better at the elec- business targeting extend their product ranges to include wiring mate- trified health stations than unelectrified (12:0.1) (Pi- (industry and/or SME) rial, lightbulbs, and kerosene cookers (some electric- ampiti and others1982). ity consumers have substituted firewood for kerosene previously used for lighting). The most prevalent productive use of electricity was ice pro- duction in freezers used for ice lollies and ice cubes (Meier 2001). Productive uses of electricity were found within various sectors, with the largest per- centage in the services sector. At the household level 26 percent of households with home enterprise in- creased income following electrification, as opposed to 4 percent without electrification (Brodman 1982). Obstacles in optimizing productive uses included dif- ficulty obtaining credit, need for marketing assis- tance, lack of information on, for example, electricity use, amount of power available, comparative costs, maintenance requirements for electric equipment. Alternative energy Mini-hydro (Meier 2001) Solar (Green 2005) Small hydro power (Barnes 2007) APPENDIX Sub-Saharan Africa Topic Tunisia Kenya Botswana Ethiopia Zambia Nambia Lifeline tariff for rural Tariffs are based on suc- Coverage existed to rural Government wants RE Financing RE is a major Grid users enjoy free con- E: Poverty targeting, low-voltage grid con- cessive consumption areas where economic but has left subsidization problem for ZESCO, so nections. Solar users have a pro-poor govern- sumers using less than bands where decreases and industrial activities to the utilities, resulting the normal criterion used down payment of 20 per- LITERA ment policy, and 50 kWh/month. Tariffs occur as consumption in- were easy to stimulate in long waiting lines and in the selection of proj- cent of the system cost. end-user have increased less and creases--favoring large (Ramasedi 1992). reduced hours of opera- ects for implementation Households must have a TURE knowledge more slowly for these consumers. Consumers tion. Energy is supplied is economic viability. The minimum income of $2,500. consumers than for other are organized in five dif- (through diesel in Bonga) only users that are better groups. Also, more than ferent tariff categories SUR 105 VEY (Table continues on next page) 106 THE Sub-Saharan Africa (continued) WELF Topic Tunisia Kenya Botswana Ethiopia Zambia Nambia ARE 90 percent of PV systems (< 7 M kWh/month; for 6 hours per day to me- off are those connected for remote rural areas are 7­100, >100, off-peak in- tered consumers (Mariam to the grid. Some eco- IMP heavily subsidized terruptible service, street 1992). nomically impractical ACT (Barnes 2007). lighting), with tariff uni- projects were included by form within each cate- government on social or OF gory. As it is cheaper to political grounds but supply services to urban were then subsidized RURAL areas, rural consumers under special funding are effectively subsidized. (Foley 1993). However, tariffs are high ELECTRIFICA and increasing (Walu- bengo and Onyango 1992). Stem migration, water Electricity has meant de- The limited supply in Electricity mainly used for Usage patterns heating (accounted for 20 creased water supply in- rural areas has supported lighting. Just under 10 per- TION percent of all consumption terruptions. The average the municipal water sup- cent of grid-electrified in 1986), provision of income of a household of ply, central clinics, street households use electricity clean water through elec- five has increased from lighting, adult education, for cooking, and none of tric pumping, lighting, $15 per month to $30 per and other civic systems. solar powered households refrigeration, communica- month. (Ramasedi 1992). Drop-out rate has im- do, as voltage is insufficient. tion, cinema, employment, proved over the years Solar also is not used for and convenience (Walu- (Mariam 1992). radio, as it provides 12 volts bengo and Onyango 1992) and radio requires 9 volts. Majority of population Time savings from still uses wood for cook- not collecting ing (Walubengo and biomass Onyango 1992). Price of paraffin (kero- Instead of load-promoting Some projects were initi- Household income sene) has been kept low, connection policy, the ated on a cost-sharing effect and any attempt to re- Business Partnership basis with private sector move the subsidy on it Center has a load- industries or agricultural would be politically diffi- deterring connection pol- enterprises (Foley 1993). cult. Liquefied petroleum icy of making the gas is not subsidized, subscriber pay for the however. The electrifica- long-term costs, distribu- tion program has been to tion transformer, and ser- electrify all 42 district vice connection. A con- headquarters, then rural sumer has to make an ini- industries formerly de- tial payment of $800 for pendent on diesel, then single-phase connection small market centers. Do- and incur a cost of mestic consumers are ex- US$5591 for a 30 KVA pected to be targeted only pump station. The corre- in the 21st century. The sponding supply costs are government currently pays half these amounts--im- a higher subsidy/grant to plying 100 percent profit those whose demand is margin (Ramasedi 1992). higher (Walubengo and Onyango 1992). Since the early days of There has been signifi- Some evidence of the use Productive uses: RE, tariff policies have cant growth in small- of electricity for produc- agricultural encouraged agricultural scale activities such as tive purposes on commer- targeting activities such as irriga- dairy farming, horticul- cial farms close to urban tion, oil pressing, and ture, and pig raising over areas, but overall elec- milling/grinding as part 10 years (Ramasedi tricity has had no dis- of a broader program to 1992). cernible development stimulate rural develop- impact in most areas ment (Barnes 2007). (Foley 1993). APPENDIX In most areas with elec- New activities attracted The range of electrically Productive uses: trification, diesel motors because of electrifica- operated coffee process- business targeting have been replaced by tion, such as a defense ing equipment has in- (industry and/or electric ones, which are air base, a police college, creased Bebeka Coffee SME) cheaper and more and a Rural Industries Plantation's production by E: efficient (Walubengo and Office (using PC) 35­40 percent over 10 LITERA Onyango 1992). (Ramasedi 1992). years. However, in most places the main use is for lighting, cutting out all TURE productive uses (Mariam 1992). SUR 107 VEY (Table continues on next page) 108 THE Sub-Saharan Africa (continued) WELF Topic Tunisia Kenya Botswana Ethiopia Zambia Nambia ARE There are new and im- There are new education Lighting at educational The main beneficiaries of There was no evidence of Social service proved health and educa- centers and schools and officer's residence re- RE were public service impact: no new businesses IMP staffing (education tion centers operated on suitable accommodation sulted in having an in- workers or employees of started after electrification ACT and health staff) cheaper and more effi- for staff who teach agri- house educational officer private companies (Foley (and home business were cient electricity, rather cultural skills, hygiene by (Mariam 1992). 1993). even more among the non- OF than diesel (Walubengo medical staff, and other electrified). and Onyango 1992). education staff RURAL (Ramasedi 1992). There has been a decline ELECTRIFICA Health benefits in the crude death rate over past 10 years (Ra- masedi 1994). Photovoltaic systems Analysis reveals that, As electricity little used for TION Alternative energy (Barnes 2007) where possible, hydro is cooking, these benefits are the best, least-cost op- limited. tion for Ethiopia, fol- lowed by grid and then diesel (Mariam 1992). Note: DAM = Demand Assessment Model; FUNDAP = Fundacao de Desenvolvimento Administrativo do Estado de Sao Paulo; GEF = Global Environment Facility; IFC = International Finance Corporation; km = kilometer; PV = photovoltaic; RE = rural electrification; SBCS = solar battery charging station; SHS = solar home system; SME = small and medium-size enterprises. APPENDIX F: IMPACT OF RURAL ELECTRIFICATION ON MICROENTERPRISE While stimulation of productive enterprise is RE on microenterprises. A difference-in-difference claimed to be among the benefits of electrification, approach (the change over time in the difference few studies have tried to quantify these benefits between electrified and nonelectrified commu- using an impact evaluation methodology. For ex- nities) was used to analyze the impact of electri- ample, the USAID evaluation of RE in Bangladesh fication, because only two conglomerates were (Barkat and others 2002) identifies those enter- newly electrified during the three years 1991­94. prise activities that use electricity and attribute the The 1994 nationally representative LSMS was an- total income from these to electrification, thus ig- alyzed by examining the links in the casual chain. noring the possibility of substitution of either one activity for another or energy sources--and The Ghana education impact evaluation (IEG so overestimating the benefit. An exception is 2004) covered the same communities (though not the ESMAP study of the Philippines, the data from the same households) that were covered by the which are reanalyzed in this appendix. LSMS for Ghana in 1988. This community-level panel was used to analyze the long-term impacts The Independent Evaluation Group (IEG) ana- of RE on growth and development of microen- lyzed three different effects considered important terprises. The 1998 nationally representative LSMS to achieving higher economic benefits: (i) com- for Ghana was analyzed using the casual chain. In plementary infrastructure--such as roads, trans- each case, only those communities classified as port, markets, bank, and adult literacy; (ii) stock rural were used. Table F.1 presents the break- of equipment and tools of microenterprises; and down of communities by electrification status for (iii) hours of operation. The empirical evidence each year for the two countries. Also analyzed relating to each of these points is discussed in the were the Lao PDR and the Philippines RE- following sections. specific data sets. Data Sources Are complementary infrastructure--such The Living Standard Measurement Survey (LSMS) as roads, transport, markets, bank--and for Peru in 1994 covered 112 conglomerates, with adult literacy more likely in electrified the sample drawn from all three rural regions-- communities? mountain, coast, and forest--of the country. The 1991 LSMS for Peru covered 43 conglomerates re- Complementary infrastructure such as roads, stricted to the rural mountain regions of the coun- transport, markets, buildings, equipment, and try. The 1994 LSMS sample, however, covered all training and information--often not provided in 43 conglomerates and all dwellings within these tandem with electricity--are important to achieve conglomerates that were interviewed in the 1991 economic benefits from electrification (Cecelski LSMS, thus giving panel data for the rural moun- 2004). There are two issues here. First, is general tain regions of the country. infrastructure, such as roads for access to markets, available in electrified communities? Second, are It is these data that are used here to analyze the business-specific services more available? The short-term growth and development impacts of first question is clearly a matter of correlation 109 THE WELFARE IMPACT OF RURAL ELECTRIFICATION Table F.1: Breakdown of Communities by Electrification Status (Number of Communities) Peru Ghana 1994 1988 Community status 1991 (1991 sample) 1994 1998 (2003 sample) 2003 Electrified 22 24 42 64 4 21 Nonelectrified 17 15 69 126 32 16 Total 39 39 111 190 36 37 Table F.2: Complementary Infrastructure Peru Ghana Facility Electrified Nonelectrified Electrified Nonelectrified Post office 0.071 0.140 0.368 0.056** Restaurants 0.262 0.246 0.887 0.403** Bank 0.000 0.000 Market 0.047 0.058 0.351 0.103** Road 0.809 0.695* 0.958 0.818** Transport 0.895 0.605** Water 0.509 0.024** Primary school 0.905 0.927 0.949 0.878* Secondary school 0.714 0.464** 0.775 0.488** Health staff 0.657 0.410** Health facility 0.635 0.440** Adult literacy 0.619 0.580 0.737 0.831 Wage 3.002 2.553** Microenterprises 0.435 0.375 0.580 0.359* Number of communities 42 69 57 124 *Significant differences at 1 percent. **Significant differences at 5 percent. rather than causation--the point of the analysis households operating a microenterprise as their being to see whether conditions for successful uti- primary or secondary occupation. lization of electricity for business purposes are in place. In Peru in 1994, electrified communities had sim- ilar complementary infrastructure, compared with The data used are from LSMS community ques- the nonelectrified communities, except for roads tionnaires for Peru 1994 and Ghana 1998. In and secondary schools. Electrified communities Ghana in 1998, electrified communities had a sig- have significantly better roads and higher proba- nificantly higher number of facilities--post of- bility of having a secondary school (table F.2). Al- fice, restaurant, market, roads, transport, water, though electrified communities had a significantly school, and health--than the nonelectrified com- higher reservation wage, there was no significant munities and a significantly higher percentage of difference between electrified and nonelectrified 110 APPENDIX F: IMPACT OF RURAL ELECTRIFICATION ON MICROENTERPRISE Table F.3: Ghana Community Panel Analysis by Electrification Status (Considering only Nonelectrified Communities in 1988) Nonelectrified Electrified in Electrified Difference-in-difference in 2003 2003, not in 1988 in 1988 t-statistics (1) (2) (3) (1) & (2) (1) & (3) Enterprises in 1988 9.7 5.8 9.2 Enterprises in 2003 9.1 12.6 10.2 ­2.76b ­0.33 Hours worked/day in 1988a 4.1 4.9 5.0 Hours worked/day in 2003a 4.6 4.8 5.3 0.80 0.21 Number of communities 15 17 4 32 19 a. Number of hours reported by self-employed members who are working in manufacturing, service, or retail industry. b. Significant difference at 1 percent. communities in percentage of households oper- youth promoters on starting up and managing a ating a microenterprise as their primary or sec- microenterprise and establishing network of col- ondary occupation. laborators. Significant differences in access to road, trans- Does electricity increase productivity/prof- port, and even market between electrified and itability through increased hours of opera- nonelectrified communities are not surprising. tion and use of equipment and tools? The economics of extending the grid to rural areas is least prohibitive for communities closer The Ghana panel (1988­2003) has 48 communi- to a road. Thus, communities closer to a road ties, of which 32 were not electrified in 1988 (see are likely to be electrified first, and other facilities table F.3). Of these 32 communities, 17 were elec- and infrastructure usually expand over time. trified as of 2003. The number of people report- ing self-employment in areas of manufacturing, The second question relates to business support service, or retail businesses has increased signif- services such as microfinance--a common find- icantly in the electrified communities, compared ing in the literature is that these services are nec- with the nonelectrified communities, but the essary to ensure that RE has the desired impact hours worked per day are not significantly different on microenterprise development. Few Bank RE between the two. projects have contained explicit components on either nurturing enterprise development or pro- The Peru panel has 224 rural households that viding training in using electricity for productive were interviewed in 1991 and again in 1994. Of purposes, though of course another scheme may these, 113 (50 percent) did not operate a mi- provide microfinance, for example, the Grameen croenterprise in 1991 or in 1994. In 1994, 35 Bank in Bangladesh. households reported a microenterprise but did not report having one in 1991; and 26 house- Credit and concessional loans allowed local en- holds reported a microenterprise in 1991 but not trepreneurs to explore possibilities for electrifi- in 1994. Although the turnover was higher in cation in India Grameen Shaki and the Sri Lanka electrified communities, the absolute increase in SEEDS/ESD Project; knowledge and training on number of microenterprises was higher in the how to use newfound electrical and motive power nonelectrified communities. increased profitability for households under the Nepal Home Employment and Lighting Package. The balanced panel results in the 50 house- In Morocco, a United Nations Development Pro- holds with a microenterprise in both periods. gramme program sponsored workshops for the However, the industry code used to classify 111 THE WELFARE IMPACT OF RURAL ELECTRIFICATION industry activities changed between 1991 and vation wage in the case of Peru [proxy for op- 1994, so analysis based on microenterprise-level portunity cost of doing business]; price of alter- data (change in profits, equipment, and so forth) nate fuels and some infrastructure variables, for is difficult. The number of hours worked by these example, distance to road; community electrifi- households before and after (irrespective of in- cation status). These variables are selected as dustry codes) was found to be independent of elec- being exogenous with respect to electrification. trification status. In a further analysis carried out In a theoretical household model, income would for all common households between 1991 and enter the model but be endogenous; here the in- 1994, for change in self-employed hours reported come term is instrumented by the education of by household (irrespective of industry code), dif- the household head and a housing index (rather ferences between the electrified and nonelectri- than a more extended wealth index, which is ar- fied households were also found to be insignificant. guably endogenous). Because of limited observations in Ghana and In the revenue/profit equation, electricity appears mismatched data in Peru, the panel data do as a variable both in its own right and through the not shed light on the impact of electrification on channels affecting hours worked (Peru, Ghana, productivity/profitability of microenterprises. The and Philippines), equipment (Peru and Ghana), causal chain for the impact using cross-sectional and distance to road as a proxy for community de- data was thus analyzed--for Peru 1994, Ghana velopment. The channels are tested by running 1998, Philippines 2002, ESMAP data and Lao PDR the regression with and without hours of opera- 2002--as follows: (1) access to electricity increases tion and value of equipment owned. If the elec- hours household members put into the business; tricity channel is measured only through hours of (2) access to electricity increases use of equipment operation and value of equipment owned, then and tools, thereby increasing productivity; (3) the household electricity variable will be signifi- access to electricity improves community infra- cant when these variables are dropped but in- structure required to reap economic benefits; significant when they are included. and (4) improved community environment, in- creased productivity, and hours of operation re- The two-step Heckman model is used to correct sult in increased profits. for sample selection bias, estimating the equation for those rural households with a microenter- The variable used to measure hours put in by prise. Tables F.4­F.7 present the results for Peru, household members in the microenterprise Ghana, the Philippines, and Lao PDR, respec- comes from data on the economic activity mod- tively, for two or three variables: total hours ule, which records average hours worked per worked by household members on the business, week by each member of the household, by in- value of equipment and tools owned by the busi- dustry code. These were matched to industry ness, and business returns. codes specified in the business module and summed over households. The variable used to The housing index does not capture much except measure value of equipment and tools owned by in Lao PDR. Households with more dependents the business came directly from the business are more likely to operate a microenterprise in module. The data in LSMS are deflated to correct Ghana and the Philippines and less likely to op- for inflation during the survey period. erate a business in Lao PDR. Education of the household head has a positive influence on the The explanatory variables are (1) household char- propensity to operate a microenterprise in all acteristics (housing index, education of household countries but Peru. The probability of a household head, dependent-adult ratio, and household elec- running a business was found to be positively re- tricity status),1 (2) entrepreneur characteristics lated to electrification status in the Philippines and (age, marital status, and education), and (3) com- to community electrification status in Ghana. The munity characteristics (regional dummies; reser- price of alternative fuel was a deterrent for oper- 112 APPENDIX F: IMPACT OF RURAL ELECTRIFICATION ON MICROENTERPRISE Table F.4: Peru: Impact on Microenterprises Propensity to Hours Value operate ME worked of worth Revenue Revenue + House quality index 0.7 Household has electricity 0.59 0.41 ­0.23 ­0.08 Education of household head = 1 ­0.11 Education of household head = 2 0.05 Entrepreneur education level = 1 0.22 0.68 0.63a 0.33 Entrepreneur education level = 2 ­0.16 0.82c 0.45 0.19 Entrepreneur head of household 1.65a ­0.16 ­0.58a ­0.41 Age of head of household ­0.28 Age of entrepreneur 1.69b 0.74a ­0.03 Dependency ratio [0 ­ 14/15+] 0.09 Male entrepreneur 0.3 0.87c 0.64b 0.12 Business is the main occupation of entrepreneur 2.61b 0.72c 0.35a 0.11 Years of operation (business) 0.19 0 0 ­0.04 Business outside home (fixed) 2.05a 0.17 0.57a 0.24 Mobile business ­0.54 ­0.83 0.24 0.48a Retail business 1.95a 0.76c 0.79b 0.53a Service business 1.09 0.25 ­0.2 ­0.34 Household hours in business 0.06a Value of business 0.48b Road index ­0.73a ­0.27 Price of candle ­1.68 Price of kerosene 0.27 Reservation wage (community average) 0.08 Community electrified 0.24 3 regional dummies suppressed Constant 0.73 2.49 ­0.29 4.14b 2.01 Observations 1338 524 524 524 524 F-stat 1.63 5.22 6.23 8.12 22.44 Note: ME = microenterprise. a. Significant at 5 percent. b. Significant at 1 percent. c. Significant at 10 percent. ating a microenterprise in all countries, but sig- older or educated entrepreneurs were more likely nificant only in Ghana. to put in fewer hours. Households operating re- tail enterprises are more likely to put extra hours Entrepreneurs who are heads of households were in Peru and the Philippines.2 Lao PDR data do more likely to put more hours into the enter- not have information on total hours worked by the prise than others in both Peru and Ghana, where household in running a microenterprise. information was available on the person respon- sible for the microenterprises; however, the result In Ghana, the stock of equipment a microenter- was significant only in Peru. In the Philippines, prise owned was likely to be greater for older 113 THE WELFARE IMPACT OF RURAL ELECTRIFICATION Table F.5: Ghana: Impact on Microenterprises Propensity to Equipment Hours operate ME purchased worked Revenue Revenue + House quality index 0.17 Household has electricity 0.12c 0.32 0.08b 0.07 Education of household head ­0.13 Education of household head ­0.25 Education of household head 0.01 Education of household head 0.89a Entrepreneur uneducated ­0.09b 0.26 ­0.01 ­0.01 Female head of household ­0.1 Entrepreneur head of household 0.03 0.36 0.03 0.02 Male entrepreneur 0.24a ­0.55 0.11c 0.10c Dependency ratio (0­14s/15+) 0.08b Age of head of household ­0.62a Age of entrepreneur 0.11b ­0.09b ­0.08 Years of operation (business) 0.02b 0.18b 0.04a 0.04a Manufacturing business 0.19c ­0.06 ­0.16a ­0.18a Service business 0.28a 0.35 ­0.26a ­0.30a Value of equipment 0.10b Household hours in business 0.01b Distance to road ­0.03 0.01 Price of kerosene ­0.20b Community electrified 0.35a 10 regional dummies suppressed Constant 2.17a ­0.31 4.83a 0.44a 0.38c Observations 3,938 1,742 1,742 1,742 1,675 F-stat 5.39 6.81 3.62 28.57 8.1 Note: ME = microenterprise. a. Significant at 1 percent. b. Significant at 10 percent. c. Significant at 5 percent. enterprises and enterprises that were operating The revenue of the enterprise was higher for fixed a manufacturing or service unit. Male entrepre- and retail businesses in Peru, lower for manufac- neurs were likely to invest more in equipment, as turing and service units in Ghana, and higher for were entrepreneurs who were educated. In Peru, retail businesses in Lao PDR. Older enterprises because of lack of data on equipment, the net were likely to have higher earnings, as were en- worth of the microenterprise was regressed in- terprises operated by men in Ghana. Household stead. The worth of the enterprise was likely to hours put into the business were positively related be higher for older and educated entrepreneurs. to revenue/profit earnings of the microenterprise Mobile businesses, not surprisingly, were likely to for Peru, Ghana, and the Philippines. Stock of be of low worth. Philippines ESMAP data and the equipment and worth of the microenterprise were Lao PDR data set do not have information on positively related to revenue/profit earnings of equipment owned by the microenterprise. the microenterprise in both Peru and Ghana. 114 APPENDIX F: IMPACT OF RURAL ELECTRIFICATION ON MICROENTERPRISE Table F.6: The Philippines: Impact on Microenterprises Propensity to Hours operate ME worked Income Income + House quality index 0.18 Household has electricity 0.59a Male household head ­0.08 ­0.99 0.64 0.76 Age of head of household 0.41b ­2.83a 2.19 2.2 Household education (in years) 0.04b ­0.15c 0.34a 0.34a Dependency ratio [0­14/15+] 0.07c Cultivators ­0.13 Commercial farming 0.19 Livestock rearing 0.18c Finland fishing ­0.54 Sari-sari store 3.11a 1.86b 1.60c Business is the main occupation of entrepreneur 0.77 0.4 0.05 Business uses electricity 1.47b 1.74b 1.64b Business uses kerosene ­0.21 1.02 Business uses biomass ­0.43 0.53 Business uses liquefied petroleum gas ­2.09b 1.02 Hours worked in business 0.14c Price of dry-cell batteries ­0.03 Time taken to collect water 0.01b 3 regional dummies suppressed Constant ­3.29a 15.11a ­10.44 ­10.53 Observations 1,979 250 250 250 F-stat 8.29 6.75 11.2 10.23 Note: ME = microenterprise. a. Significant at 1 percent. b. Significant at 5 percent. c. Significant at 10 percent. Electrification status of the household was sig- Philippines, the variable capturing use of elec- nificantly and positively related to equipment and tricity in home business is significant even when earnings in Ghana, but not hours worked; it was hours worked is added as a regressor. One pos- positively related to hours worked and revenue sible explanation, stemming from the Ghana analy- earnings in the Philippines and to revenue earn- sis, may be that electricity is capturing electrical ings in Lao PDR. All three equations were found tools used in businesses. to be independent of the electrification status in Peru. The channel of electrification is clear in Electrification has a small but significant impact Ghana, where the variable capturing electrification on the revenue earnings of the microenterprise. status of the household becomes insignificant The possible channels are increase in number of once the hours and value of equipment are added hours worked per day by the household members as regressors in the revenue equation. In the and use of electrical equipment. 115 THE WELFARE IMPACT OF RURAL ELECTRIFICATION Table F.7: Lao PDR: Impact on Microenterprises Propensity to operate Business microenterprise income House quality index 0.41a Household has grid 0.53b Household has electricity 0.17 Household education (in years) 0.07b Dependency ratio (0­14/15+) ­0.10c Number of businesses ­0.11 Business is the main occupation of entrepreneur 1.45b Home-based business ­0.19 Retail business 1.39b Pieces of farmland owned 0.00 Distance to road ­0.01 Price of electricity 0.13 Price of diesel ­0.03 Community electrified ­0.13 15 regional dummies suppressed Constant ­2.26b 14.02b Observations 5,446 529 a. Significant at 5 percent. b. Significant at 1 percent. c. Significant at 10 percent. 116 APPENDIX G: HEALTH AND EDUCATION DHS data were analyzed for 11 countries (table The second knowledge variable is a simple aver- G.1). DHS survey instruments are reasonably age of four separate questions from the survey: standardized across countries, allowing the same (1) the contraceptive knowledge variable already model specification to be used for different coun- mentioned, but scaled to be between 0 and 1; (2) tries, albeit with some variations in variable defi- knowledge of the timing of ovulation (0 = in- nition. In each case only those households correct answer "do not know, any time, or mid- classified as rural were used. The sample size var- dle of the periods," 0.5 = "after period ends or ied from 2,801 (child nutrition in Peru) to 17,165 before period starts," and 1 = correct answer (women in Indonesia). "middle of cycle") (this question was not asked in Bangladesh, where instead a question was asked Three different effects were examined using these on knowledge of signs of a risky pregnancy); data: first, how access to information from TV (3) knowledge of the symptoms of AIDS, scaled and radio influences health knowledge and so between 0 and 1; and (4) knowledge of oral re- health behavior and health outcomes; second, hydration salts. how the availability of refrigerated storage im- proves nutrition; and third, how electricity af- The explanatory variables are the same in each fects schooling. case covering (1) household characteristics (hous- ing index, education of household head, and The causal chain for the first possible impact is as whether the household has electricity); (2) indi- follows: (1) access to electricity increases time vidual characteristics (age, marital status, liter- spent watching TV and listening to the radio; acy, agency, and whether the women listen to the (2) increased access to media increases awareness radio at least once a week and watch TV at least of health issues; (3) this increased awareness re- once a week); and (3) community characteris- sults in changed health behavior; and (4) changed tics (regional dummies and the share of women behavior improves health outcomes and reduces who have heard of family planning by family plan- fertility. The empirical evidence relating to each ning worker by cluster). The share of women vis- of these points is discussed in turn. ited by a family planning worker is a proxy for active health or family programs in the area. Most of these variables are self-explanatory, but some Knowledge-Media Link require a little discussion. A more general wealth Two variables are used to measure health knowl- index is not used, to avoid problems of endo- edge. The first is the response to the DHS question geneity, and the housing index, together with ed- regarding knowledge of modern contraceptives. ucation of the household head, may be regarded Women are asked to name modern contraceptive as an instrument for income. methods they know; the questionnaire records those they name against a list in the survey. The Electricity appears as a variable in its own right, number of possible methods ranges from 0 to 12 through the channels of affecting women's access (Ghana). The variable is simply the number of to TV directly and indirectly through someone else methods named. The coefficients are thus not owning a TV (proxied by household electrification comparable between the regressions. rate in the community). This channel is tested by 117 THE WELFARE IMPACT OF RURAL ELECTRIFICATION Table G.1: DHS Sample Sizes Country Survey year Eligible women Children under 5 Bangladesh 2004 7,536 4,835 Ghana 2003 3,317 2,801 Indonesia 2002­03 17,165 9,636 Morocco 2003­04 7,801 3,496 Nepal 2001 7,572 6,294 Nicaragua 2001 5,775 3,973 Peru 2000 10,749 7,467 Peru 2004 4,737 2,717 Philippines 1998 7,253 5,004 Philippines 2003 6,197 3,854 Senegal 2005 8,290 7,364 Source: DHS data. running the regression with and without the vari- pooled across surveys, so estimates become more ables capturing women's access to TV (direct and precise as they are based on a larger sample. This indirect). If a TV channel is the only one through results in a nine-country analysis for each variable. which electricity affects knowledge, then the household electricity variable will be significant Most of the conditioning variables have the ex- when the access to TV variables are dropped but pected sign (tables G.2 and G.3). The better off significant when they are included. and literate women have more health knowledge, as do those with higher mobility, control over The estimation method used is as follows: The decisions affecting their lives (agency),1 and some contraceptive knowledge equations are estimated urban living. For both contraceptive knowledge using an ordered probit, an extension of the bi- and health knowledge variables, TV is significant nary probit model that is used in cases with mul- for all nine countries. The electricity status, in its tiple and ranked discrete dependent variables. own right, significantly increases both contra- The ordered probit model is of the form: ceptive and health knowledge in the Philippines and Indonesia and health knowledge in Bangla- p = (1 + x) desh as well. When the regressions are run drop- 1 p ping the variable capturing access to TV as the i>1 = ( i + x ) - ( i -1+ x ) source of information, then the household elec- ......... tricity coefficient is positive and significant for p = 1 - (k + x), k -1 most of the knowledge equations. where denotes the cumulative standard normal Taken as a whole, the regressions provide sufficient distribution function and pi is the probability of evidence that access to TV increases health and the event i occurring; in this case it would denote family planning knowledge and that it is this that that probability that women know i contraceptive is the channel through which electrification affects methods. The health knowledge equations are es- health knowledge. timated using ordinary least square. The next step is to examine the extent to which The estimates for Peru and the Philippines are cal- knowledge affects practice. Two health prac- culated from two rounds of the DHS, and data are tice variables are examined: (1) use of modern 118 Table G.2: Health Knowledge [Range 0­1] Bangladesh Ghana Indonesia Morocco Nepal Nicaragua Peru Philippines Senegal House quality index 0.01a 0.00 0.03b 0.07b ­0.01 0.02a 0.08b 0.02b 0.03b Education of the head (0­1.7) 0.03b 0.05b 0.09b 0.03b 0.03b 0.05b 0.07b 0.05b 0.06b Household has electricity 0.01c 0.02 0.02b 0.00 0.00 0.01 0.01 0.02b 0.00 Woman can read and write 0.04b 0.07b 0.09b 0.06b 0.07b 0.05b 0.12b 0.07b 0.07b Woman's current age 0.09b 0.24b 0.11b 0.16b 0.10b 0.10b 0.31b 0.21b 0.34b Woman's age square ­1.39b ­3.32b ­1.76b ­2.20b ­1.54b ­1.25b ­4.39b ­2.97b ­4.73b Woman is head of household 0.00 0.03c 0.01 0.00 0.01 0.01a 0.00 -0.02 0.02a Woman is divorced/widowed/not living together ­0.03b ­0.01 ­0.01 ­0.02c ­0.03b ­0.02b ­0.01 0.00 ­0.04b Woman never married ­0.05b 0.07b ­0.06b ­0.02b ­0.07b ­0.07b Some urban living 0.04b 0.01 0.03b 0.02c 0.01 0.04b 0.01c 0.04b Getting medical help is small problem 0.02b 0.01c 0.01b 0.01b 0.00c 0.01b 0.01c Woman alone has the final say 0.00c 0.00 0.00 0.00 0.01b 0.00 0.01b 0.01a Reads paper at least once a week 0.04b 0.01 0.08b 0.04b 0.04b 0.05b 0.04b 0.04b 0.03a Listens to radio at least once a week 0.02b 0.08b 0.03b 0.02b 0.04b 0.03b 0.03b 0.01a 0.05b Watches TV at least once a week 0.04b 0.04b 0.04b 0.02c 0.03b 0.02b 0.05b 0.02b 0.03b Share of women visited by family planning worker by cluster 0.01 -0.01 0.02 ­0.13b 0.06 0.12b 0.00 0.09b 0.06 APPENDIX Cluster probably has electricity 0.01c 0.01 0.02 0.00 0.00 0.01 0.01 0.00 ­0.01 Survey 0.09b ­0.10b Constant 0.19b ­0.02 ­0.02 0.18b 0.33b 0.19b ­0.40b 0.02 ­0.32b Observations 7,535 3,312 17,073 7,778 7,550 5,762 15,470 13,413 8,262 G: R-squared 0.23 0.27 0.28 0.16 0.20 0.25 0.31 0.19 0.27 HEAL F-stat 80.30 34.15 83.52 25.99 38.06 49.34 162.64 85.06 110.40 Sources: DHS data; IEG calculations. TH Note: Regional dummies suppressed. a. Significant at 10 percent. AND b. Significant at 1 percent. c. Significant at 5 percent. EDUCA TION 119 120 THE WELF ARE IMP Table G.3: Knowledge of Modern Contraceptives [Range 0­12] ACT Bangladesh Ghana Indonesia Morocco Nepal Nicaragua Peru Philippines Senegal OF House quality index ­0.04 0.12 0.11 0.35a ­0.02 0.28a 0.42a 0.11a 0.16a RURAL Education of the head 0.23a 0.21a 0.60a 0.03 0.32a 0.37a 0.30a 0.32a 0.21a Household has electricity ­0.11b 0.00 0.11a ­0.09 0.09 0.04 ­0.03 0.09a 0.04 ELECTRIFICA Woman can read and write 0.10c 0.17a 0.43a 0.42a 0.65a 0.35a 0.58a 0.42a 0.32a Woman's current age 1.80a 1.05a 0.91a 1.06a 0.77a 0.93a 1.47a 1.18a 1.62a Woman's age square ­25.87a ­14.44a ­13.04a ­13.96a ­12.35a ­12.31a ­22.68a ­15.70a ­23.01a Woman is head of household 0.00 0.09 0.14 0.01 0.08 0.06 0.05 ­0.06 0.08 Woman divorced/widowed/not living together ­0.36a 0.01 ­0.14b ­0.19a ­0.23a ­0.05 ­0.05 ­0.09 ­0.18a TION Woman never married ­0.34a ­0.42a ­0.53a ­0.25a ­0.25a ­0.41a Some urban living 0.07 0.09c 0.12b 0.37a 0.09c 0.22a 0.11a 0.31a Getting medical help is small problem 0.25a 0.04b 0.09a 0.07a 0.03b 0.07a 0.05a Woman alone has the final say 0.04a 0.03b 0.01 0.02 0.07a 0.03 0.05a 0.00 Reads paper at least once a week 0.13c 0.19c 0.52a 0.34a 0.53a 0.34a 0.20a 0.29a 0.18c Listens to radio at least once a week 0.17a 0.41a 0.18a 0.12a 0.55a 0.27a 0.18a 0.16a 0.29a Watches TV at least once a week 0.14a 0.33a 0.38a 0.20a 0.21a 0.18a 0.28a 0.19a 0.20a Percent women visited by family planning worker by cluster 0.21 0.32c 0.01 ­0.53a 0.16 0.72b ­0.01 0.65a 1.18a Cluster probably has electricity 0.16a ­0.03 0.09 0.01 0.05 0.12c 0.09 0.07 ­0.17b Survey 2.01a 1.32a Observations 7,535 3,312 17,073 7,778 7,550 5,762 15,470 13,413 8,262 F-stat 40.81 33.85 71.62 41.2 50.82 45.92 199.63 199.68 77.01 Source: DHS data; IEG calculations. Note: Regional dummies suppressed. a. Significant at 1 percent. b. Significant at 5 percent. c. Significant at 10 percent. APPENDIX G: HEALTH AND EDUCATION contraceptives by women who have had at least because there is more light and because TV and one intercourse and (2) child immunization for radio provide an "alternative to sex" for recre- children older than 18 months. Contraceptive ation. However, the data do not support this practice is a dichotomous variable of whether a point of view. TV watching only significantly affects woman has ever used modern contraception. Im- sexual activity in one of the eight cases, and house- munization is similarly defined, corresponding hold electricity is not significant. On the contrary, to the child's status for BCG (for tuberculosis), electrification indirectly increases sexual activity, DPT (for diphtheria, pertussis, and tetanus), as coital frequency is higher for those women measles, and polio vaccinations, and two vari- with higher contraceptive knowledge. ables corresponding to having all vaccines or no vaccines at all, respectively. Hence, for immu- These results can be used to estimate the impact nization, six separate regressions were estimated electrification has on fertility (table G.8). The for each of the nine countries for which data were total effect is the combination of the direct impact available, making 54 equations in all. from the fertility equation and the indirect impact via higher knowledge (which is the knowledge co- The right-hand side variables for contraceptive efficient from the fertility equation multiplied by practice are similar to those used for the knowl- how electricity affects knowledge, taken as the co- edge regressions, plus the knowledge variable, in- efficient on the household electricity variable in formation capturing children, and partner's the absence of the TV variables). These calcula- education (see table G.4). The general determi- tions show an impact on fertility reduction from nants are as expected and are similar to those for a low of 0.04 in Nicaragua to about 2.00 in Sene- knowledge. Of interest here is the knowledge gal as result of electrification. variable, which is significantly positive in all equa- tions (table G.4). For immunization status, the The health outcomes used are nutrition and knowledge variable was positive and significant in under-five mortality. Electricity may positively af- 44 of the 45 "have vaccination" regressions. For fect nutrition directly by allowing refrigerated "no vaccination," knowledge was, as expected, food storage and indirectly through knowledge. significantly negative in all but one case (table Two nutrition measures are used: the height for G.6). The link between knowledge and practice age z score (HAZ) and the weight for age z score is thus firmly established (see table G.5). (WAZ). The z-score is a standardized measure; being more than two z scores below the reference Some surveys interviewed men, and for four of the value constitutes being undernourished and more nine countries, men's knowledge variable was than three, severely undernourished. HAZ is taken also included as a regressor in the contraceptive as a measure of long-run nutritional status; WAZ practice equation.2 In three of the four cases, the indicates short-run status. knowledge variable is significantly positive. The explanatory variables are similar for the two Fertility outcomes are measured as total children equations, which in turn are similar to those used ever born as a ratio to the total fertility rate for that throughout this analysis but with more demo- age group of women, using 5-year age ranges graphic variables because there is possible com- starting at age 20. In five of the nine cases, the petition for resources between siblings. In each health knowledge variable has a significant neg- case the variables cover (1) household character- ative impact on fertility. The household electric- istics (housing index, education in years of house- ity variable is also significant and negative in eight hold head, number of young children, and whether of the nine cases (table G.7). What are the possi- the household has electricity and a refrigerator); ble reasons for this latter finding? (2) mother's characteristics (age, height, marital status, literacy and knowledge, mother is head of A possible explanation is that electricity reduces household, mobility, and agency); (3) child's char- coital frequency by increasing waking hours, both acteristics (gender, birth order, gap between own 121 122 THE WELF ARE IMP Table G.4: Practice of Modern Contraceptives ACT Bangladesh Ghana Indonesia Morocco Nepal Nicaragua Peru Philippines Senegal OF House quality index 0.17 0.07 0.14 0.26 0.16 0.07 0.42a 0.05 0.17a RURAL Education of the head 0.08 0.16b 0.18a 0.21c 0.25a 0.19 0.05 0.21a 0.33a Household has electricity 0.15a ­0.04 0.11b ­0.20b 0.08 0.09 ­0.04 0.11a 0.20a ELECTRIFICA Women's health knowledge 2.30a 1.35a 0.69a 1.36a 1.69a 2.04a 0.85a 1.10a 1.34a Woman can read and write ­0.20b 0.09 0.20a ­0.06 0.22a 0.13b 0.06 0.07 0.19a Couple is uneducated ­0.15a ­0.32a ­0.02 ­0.17b ­0.11b ­0.15b ­0.17 ­0.62a ­0.36a Woman's current age 2.31a 0.43c 2.05a 2.29a 2.62a 1.52a 1.32a 1.91a 0.97a Woman's age square ­35.92a ­7.37c ­29.75a ­31.15a ­35.16a ­25.17a ­23.05a ­27.48a ­13.21a TION Woman divorced/widowed/not living together ­0.69a ­0.02 ­0.33 ­1.69a ­0.86a ­0.54a ­0.55a ­0.54a ­0.13 Woman never married ­0.03 ­1.42a ­1.17a ­1.69a 0.21 Woman is head of household ­0.30a ­0.07 ­0.16 ­0.09 ­0.16c 0.01 0.05 ­0.05 ­0.05 Woman alone has the final say 0.03c 0.00 0.04b 0.05c 0.05a 0.02 0.07a 0.05b Getting medical help is small problem 0.20a 0.03c 0.01 0.02 0.02c 0.02 0.01 0.01 Some urban living 0.03 0.20a ­0.15 0.19b ­0.03 0.11a 0.07c 0.19a Woman has only living girls ­0.12 ­0.05 0.11 ­0.02 ­0.38a ­0.17 ­0.24a ­0.37a ­0.14 Number of children born--only girls ­0.03 ­0.01 0.03 0.09 ­0.10a 0.11b 0.06c 0.13a 0.01 Muslim 0.02 ­0.14 0.22a ­0.81a ­1.08a ­0.04 Source of FP method by cluster 0.99a 0.91a 1.33a ­0.04 0.72a 0.41b 0.65a 0.47a 0.15 Share of women accessing health care by cluster 0.34 0.25c 0.18 0.23 0.50b 0.37b 0.89a 0.12 0.74a Cluster probably has electricity ­0.02 ­0.13 ­0.07 0.12 0.12 0.05 ­0.09 0.11c ­0.19b Survey ­0.10b ­0.07 Constant ­4.56a ­2.43a ­4.89a ­3.58a ­6.40a ­3.07a ­2.81a ­4.14a ­3.89a Observations 7,098 2,963 16,287 4,533 7,549 4,641 12,789 9,843 6,717 F-stat 36.94 14.71 25.86 18.16 39.84 22.17 56.38 42.22 28.03 Source: DHS data; IEG calculations. Note: Regional dummies suppressed. a. Significant at 1 percent. b. Significant at 5 percent. c. Significant at 10 percent. Table G.5: Immunization: Child Received All Vaccinations Bangladesh Ghana Indonesia Morocco Nepal Nicaragua Peru Philippines Senegal House quality index ­0.08 ­0.02 ­0.06 0.04 ­0.04 0.01 0.02 0.10 0.05 Education of the head 0.19a 0.11 0.18b 0.07 0.14c ­0.10 0.10c 0.20a 0.04 Household has electricity 0.05 0.12 0.00 0.07 0.05 0.01 0.02 0.11b 0.11 Women's health knowledge 1.02a 1.00a 1.24a 0.78a 1.21a 0.77a 0.32a 0.82a 0.61a Woman can read and write ­0.04 0.12 0.26b ­0.02 0.32a 0.00 0.03 0.19a 0.15b Woman's current age 0.80a ­0.24 0.70b 0.66b 0.89a 0.38 0.26 0.94a 0.22 Woman's age square ­11.78b 6.09 ­11.11b ­8.80c ­13.44a ­3.93 ­1.66 ­13.34a ­2.57 Woman has married more than once ­0.26b ­0.18b ­0.28b 0.07 ­0.13c 0.04 0.09 ­0.20b ­0.05 Woman is head of household ­0.06 0.10 ­0.23 ­0.16 0.05 ­0.11 0.04 0.19 ­0.07 Woman alone has the final say 0.03 ­0.02 ­0.03 0.08b 0.04c 0.00 0.01 ­0.04c Getting medical help is small problem 0.06 0.02 0.07a 0.04b 0.05b 0.04b 0.02 ­0.02 Some urban living ­0.02 ­0.14c ­0.09 0.01 ­0.05 ­0.07c ­0.04 ­0.02 Muslim 0.42c 0.25c 0.03 ­0.36 ­0.59a ­0.23b Hindu 0.66a ­0.05 Christian 0.28b 0.09 ­0.06 Share of women accessing health care by cluster 0.58c 0.63a 0.62a 0.37a 0.39 0.36b 0.23c 0.31a 0.32c Percent women visited by family planning worker by cluster 0.52b 0.17 0.80b 0.31 1.10a ­0.29 ­0.11 0.27c 0.30 APPENDIX Cluster probably has electricity 0.01 0.02 ­0.03 0.08 ­0.05 ­0.10 ­0.03 0.09 0.00 Survey ­0.06 ­0.12a Constant ­2.16a ­0.95 ­2.91a ­1.44a ­2.30a ­1.07a ­1.03a ­2.21a ­0.77b G: Observations 3,475 2,032 6,924 2,408 4,454 2,817 7,001 6,333 5,428 HEAL F-stat 4.53 4.93 9.95 2.99 8.08 5.79 4.5 20.4 4.71 Source: DHS data; IEG calculations. TH Note: Regional dummies suppressed. a. Significant at 1 percent. AND b. Significant at 5 percent. c. Significant at 10 percent. EDUCA TION 123 124 THE Table G.6: Immunization: Child Received No Vaccinations WELF Bangladesh Ghana Indonesia Morocco Nepal Nicaragua Peru Philippines Senegal ARE House quality index 0.04 0.16 0.07 ­0.19 0.28 0.08 ­0.05 ­0.09 ­0.22c IMP Education of the head ­0.26a ­0.30a ­0.28a ­0.61b ­0.17 ­0.08 0.05 ­0.19a 0.06 Household has electricity 0.05 -0.09 0.01 ­0.16 0.48c 0.34a ­0.41a ­0.26c ­0.18 ACT Women's health knowledge ­1.01b ­0.79c ­1.06c ­1.05a ­1.41c ­0.47 ­1.27c ­1.13c ­1.04c OF Woman can read and write 0.31a ­0.08 ­0.17 ­0.47b ­0.21a ­0.07 ­0.11 ­0.22c ­0.41c Woman's current age ­0.73 0.58 ­0.34 ­0.69 ­0.39 ­0.77 0.14 ­0.65a 0.27 RURAL Woman's age square 12.57 ­8.80 5.51 10.39 4.08 12.58 ­3.45 11.01a ­5.07 Woman has married more than once 0.20 0.05 0.10 0.19 0.01 ­0.10 0.17 0.17b 0.02 ELECTRIFICA Woman is head of household 0.11 0.04 0.45 0.11 ­0.28 ­0.25 ­0.19 0.04 0.05 Woman alone has the final say 0.00 0.00 ­0.02 ­0.15 ­0.04 ­0.07 0.01 0.02 Getting medical help is small problem ­0.15 ­0.08a ­0.05b ­0.05 ­0.04 ­0.03 ­0.03 ­0.05a Some urban living 0.17 0.01 0.11 ­0.34 0.02 0.12 ­0.10 ­0.01 Muslim 4.17c ­0.35b ­0.27 0.18 0.58c TION Hindu 4.01c ­0.10 Christian ­0.30a ­0.17 -0.09 Share of women accessing health care by cluster ­1.24c ­1.01c -0.75c ­0.44 ­0.72 ­1.36c ­1.07c ­0.97c ­0.26 Percent women visited by family planning worker by cluster ­1.12b 0.03 ­1.20c ­0.07 ­0.23 ­0.12 ­0.82b ­0.69c 0.11 Community has electricity ­0.08 0.19 0.09 ­0.25 ­0.30b ­0.01 ­0.03 ­0.12 ­0.17 Survey ­0.22a 0.03 Constant ­4.20 ­1.30 0.95 0.21 0.28 0.14 ­0.96 1.29c ­0.99b Observations 3,475 2,032 6,924 2,408 4,454 2,817 7,001 6,333 5,428 F-stat 184.15 2.49 3.68 2.12 4.75 6.06 7.37 18.34 6.96 Source: DHS data; IEG calculations. Note: Regional dummies suppressed. a. Significant at 5 percent. b. Significant at 10 percent. c. Significant at 1 percent. Table G.7: Fertility Rate Bangladesh Ghana Indonesia Morocco Nepal Nicaragua Peru Philippines Senegal House quality index ­0.24a ­0.12 ­0.10c ­0.63a ­0.11b ­0.38a ­0.35a ­0.22a ­0.11a Education of the head ­0.11a ­0.13c ­0.07a 0.17 ­0.02 ­0.22a ­0.03 ­0.33a ­0.03 Household has electricity ­0.07b ­0.16c ­0.11a ­0.19b ­0.15a ­0.07 ­0.08b ­0.31a ­0.27a Women's health knowledge ­0.34b ­0.19 ­0.23a ­0.36 0.17b ­0.26c ­0.46a ­0.42a 0.04 Couple is uneducated 0.02 ­0.1 0.02 ­0.42a 0.05b 0.04 ­0.09c ­0.25 ­0.03 Woman divorced/widowed/not living together ­0.24a 0.07 ­0.1 ­1.03a ­0.04 ­0.09b ­0.38a ­0.31a Woman has married more than once ­0.16a ­0.01 ­0.07c ­0.48a ­0.12a ­0.01 ­0.02 ­0.15a ­0.18a Woman is head of household ­0.05 ­0.18b ­0.17a ­0.03 ­0.03 ­0.15a ­0.04 ­0.09b Getting medical help is small problem ­0.03c ­0.03a ­0.05a ­0.01 0.01 Woman alone has the final say 0.03a 0.00 0.01 ­0.07a ­0.01 ­0.03 ­0.02b 0.00 Woman is working ­0.05b ­0.19 ­0.13a 0.09 ­0.05 ­0.15a ­0.32a ­0.06c Some urban living ­0.02 ­0.13 ­0.03 0.01 ­0.09a ­0.08 ­0.04 Christian 0.00 0.08 -0.01 Hindu 0.04 ­0.01 Muslim 0.20a ­0.06 0.18a 0.08 0.01 0.23a Source of family planning method by cluster ­0.23 0.07 ­0.12c 0.49c ­0.13 ­0.23b ­0.01 0.13 0.03 Child mortality rate by cluster 0.69b 0.5 0.26 1.3 0.39c 0.43 0.07 0.79c 1.32a APPENDIX Survey 0.01 ­0.07 Constant 1.23a 1.71a 1.51a 2.65a 1.15a 1.77a 1.52a 2.32a 1.07a Observations 7,098 2,963 16,287 4,533 7,549 4,641 12,789 9843 6717 F-stat 19.23 1.99 16.38 15.66 8.1 9.61 26.67 21.46 10.27 G: Source: DHS Data; IEG calculations. Note: Regional dummies suppressed. HEAL a. Significant at 1 percent. b. Significant at 5 percent. TH c. Significant at 10 percent. AND EDUCA TION 125 THE WELFARE IMPACT OF RURAL ELECTRIFICATION Table G.8: Impact of Electrification on Fertility Rate Bangladesh Ghana Indonesia Morocco Nepal Nicaragua Peru Philippines Senegal TFR at age 50 5.83 6.35 4.63 6.32 5.98 7.25 6.57 5.54 7.42 Knowledge equation Electrification status 0.02 0.04 0.03 0.00 0.01 0.02 0.03 0.02 0.01 Child ever born: TFR equation Electrification status ­0.07 ­0.16 ­0.11 ­0.19 ­0.15 ­0.07 ­0.08 ­0.31 ­0.27 Health knowledge ­0.34 ­0.19 ­0.23 ­0.36 0.17 ­0.26 ­0.46 ­0.42 0.04 Reduction ­0.45 ­1.06 ­0.54 ­1.20 ­0.89 ­0.55 ­0.62 ­1.76 ­2.00 Reduction (only significant variables) ­0.45 ­1.06 ­0.54 ­1.20 ­0.89 ­0.04 ­0.62 ­1.76 ­2.00 Note: TFR = total fertility rate. birth and birth of previous sibling); and (4) com- tween first and fifth birthdays), so there are 27 munity characteristics (regional dummies, share mortality regressions. Although DHS includes of households with electricity). questions on self-reported sickness, there are well-known biases in such variables, so they are Most of the child conditioning variables have the not used in this analysis. The explanatory variables expected sign. The relationship between birth for the three mortality equations are similar to order and nutrition is as expected, decreasing those used in the nutrition equations. with increase in birth order, reflecting food avail- ability for older children. Similarly, children born Most of the child-conditioning variables have the at short intervals show a smaller HAZ. Taller expected sign. A female child is more likely to sur- women have taller children and so do women who vive in infancy than at an older age. For biologi- are literate and with higher agency. The contra- cal reasons, males have a higher natural risk of ceptive knowledge variable affects HAZ in four of death at all ages. However, preference for boys the six countries for which data are available, and over girls, especially in Asia, overrides biological WAZ in two of the six. Data on the household own- factors at later stages in infancy. The relationship ing a fridge were available for four countries; the between birth order and mortality risk is convex coefficient was significantly positive on HAZ and (U-shaped) as expected, reflecting mother's age, WAZ in three of the four countries. The household social preferences, and food availability for older electricity variable is also significant and positive children. in four of the six cases. Similarly, children born at short intervals and In two countries, a possible reason can be refrig- twins/triplets show a higher risk of mortality. erator ownership. However, what are other pos- Women's factors send a mixed message with the sible channels for this finding in two countries with mortality regressions. Although cluster averages significant household electricity impact and the show expected signs, the household variables do presence of a refrigerator? There is evidence that not behave as expected. The results in general are electrification improves child nutritional status, but not that good. Immunization and knowledge are the channels are not fully explained in all coun- both significant in some cases but are not over- tries (table G.9). whelming, as earlier links in the chain were. Separate equations are estimated for neonatal Electricity and Education mortality (death in first month), infant mortality A Cox hazards model (where the hazard for a (death in first year), and child mortality (death be- child between the ages of 6 and 15 is dropping out 126 APPENDIX G: HEALTH AND EDUCATION of school) was estimated to see the impact of it is not possible to confirm the impact through electrification, if any, on the propensity of a child this channel. to stay in school. Although the model takes care of the censoring problem, the DHS data do not Time use provide much information on education-specific Electrification was found to increase the reading information for children. time of both adults and children in the household once the adult and/or child decides to read. Multi- Nevertheless, a simple model was estimated with variate regressions of the effect of electrification largely time-invariant variables as the independ- on adult reading and children's studying--con- ent variables. The estimated model indicates that trolling for factors such as housing index, educa- RE indirectly improves the propensity of child to tion, and age of the head of household--showed stay in school via increase in the mother's knowl- that the availability of electricity in the house- edge and education. Moreover, electricity itself is hold had no significant effect on adults' and chil- significant in seven of the nine regressions (table dren's propensity to read and study, respectively. G.10). This might be capturing an increase in However, once individuals choose to read or reading/studying hours due to illumination after study, electricity was also found to increase the dawn. However, because of lack of time-use data, time the children spend studying by 77 minutes Table G.9: HAZ Bangladesh Ghana Morocco Nepal Nicaragua Peru House quality index 0.52a ­0.09 0.73a 0.07 0.33a 0.32a Education of the head 0.13a 0.03 0.05 0.07 0.10 0.14a Household has electricity 0.20a 0.27c 0.08 0.20a 0.04 0.10c Household owns refrigerator 0.24 0.35a 0.38a 0.38a Number of household members age 0 ­ < 5 ­0.10a ­0.02 ­0.05 ­0.03 ­0.02 ­0.10a Contraceptive knowledge 0.02 0.03a 0.04b 0.03c 0.03c 0.00 Woman can read and write 0.18a 0.06 0.00 0.19a 0.04 0.08c Woman divorced/widowed/not living together 0.04 ­0.18 0.01 ­0.16 ­0.05 ­0.04 Woman's current age 0.02 ­0.10 0.00 ­0.05 0.06 0.11a Woman is head of household 0.16b ­0.17b ­0.29b 0.01 ­0.05 0.04 Woman alone has the final say ­0.01 0.04c 0.08c 0.00 ­0.03 ­0.01 Getting medical help is small problem 0.01 0.01 ­0.02 0.00 0.00 0.00 Log of woman's height (cm) 6.25a 4.10a 5.96a 5.51a 7.15a 7.14a Female child 0.06b 0.22a 0.22a 0.02 0.08b 0.09a Order of birth ­0.07b 0.00 ­0.06 ­0.06c ­0.09a ­0.16a Birth order 0.01 0.00 0.00 0.01c 0.00c 0.01a = 1 if < 24 months gap with preceding sibling ­0.14a ­0.11 ­0.10 ­0.14a ­0.08b ­0.12a Cluster probably has electricity ­0.13a ­0.21c ­0.12 0.02 0.01 0.01 Constant ­33.23a ­22.30a ­31.61a ­29.63a ­37.58a ­37.32a Observations 4,002 2,302 3,000 5,196 3,453 8,794 F-stat 22.08 5.25 8.38 18.08 16.07 33.1 Source: DHS data; IEG calculations. Note: Regional dummies suppressed. a. Significant at 1 percent. b. Significant at 10 percent. c. Significant at 5 percent. 127 THE WELFARE IMPACT OF RURAL ELECTRIFICATION (1.28 × 60) and the time adults spent reading the choice to study, electricity was found to in- by 27 minutes per electrified household per crease the time he or she spent reading or study- day, compared with nonelectrified households ing by 48 minutes per day, compared with (table G.11). nonelectrified households. For adults, the study found an increase in time spent reading of close The ESMAP report (2000) conducted a Heckman to 15 minutes per day. procedure using child-level information to study the effect of electrification on children's reading The study and reading time was found to be sig- and studying--controlling for factors such as in- nificantly higher for the children and adults of elec- come, parents' education, type of dwelling, and trified households in electrified villages than for price of fuel. The analysis concluded that the both children and adults in the nonelectrified availability of electricity in the household had a households in nonelectrified villages and non- negative effect on children's propensity to read electrified households in electrified villages using or study--which in turn is presumed to be caused the nearest neighbor matching technique (see by more time spent watching TV and other forms table G.12). The nearest match was based on ed- of entertainment. Nevertheless, once a child made ucation (in years) of the head of the household. 128 Table G.10: Education Bangladesh Ghana Indonesia Morocco Nepal Nicaragua Peru Philippines Senegal House quality index ­0.95a ­0.23c ­0.55a ­0.70a ­1.14a ­0.60a 0.26 ­0.60a ­0.07a Education of the head ­0.89a ­0.34a ­0.58a ­0.20 ­0.72a ­0.64a ­1.17a ­1.25a ­0.37a Household has electricity ­0.55a ­0.18 ­0.15c ­0.30a ­0.33a ­0.37a ­0.30 ­0.49a ­0.10c Household size 0.02b 0.01b 0.03c 0.01 0.01 0.03a 0.03 0.09a 0.00 Women's health knowledge ­1.26a ­0.32a ­1.11a ­0.58a ­0.97a ­0.87a ­1.16a ­0.71c ­0.18a Woman's age (in log) 0.36c ­0.29c ­1.09a 0.01 0.33a ­0.53a ­0.91c ­0.56c 0.04 Some urban living ­0.24 ­0.10b ­0.42a ­0.23 ­0.14b 0.11 ­0.12 ­0.08b Couple is uneducated 0.41a ­0.06 0.19b 0.22a 0.25a 0.28a 0.12 ­0.04 0.27a Female child ­0.46a ­0.02 ­0.06 0.26a 0.25a ­0.14a ­0.11 ­0.28a 0.02 Muslim 0.41c ­0.10 ­0.12 0.21c 0.31b 0.57 Hindu 0.36b 0.00 Christian ­1.10 ­0.17a ­0.21b ­0.16 0.01 Distance to school ­0.03 Cluster probably has electricity 0.08 0.13 0.03 0.17c 0.04 0.12b ­0.69a 0.00 0.08c Share of women accessing HC by cluster 0.00 ­0.05 ­0.26c ­0.46a ­0.63a ­0.16 0.47 ­0.05 ­0.07 Share of children in school by cluster ­2.50a ­2.91a ­2.76a ­2.67a ­2.62a ­2.29a ­3.49a ­3.15a ­2.06a Observations 7,334 3,053 15,805 6,096 8,722 6,925 5,126 6,481 7,119 APPENDIX Wald Chi 1,193.06 1,188.31 1,578.32 1,178.15 2,623.88 1,667.93 473.73 1,328.13 1,463.64 Source: DHS data; IEG calculations. Note: Regional dummies suppressed. a. Significant at 1 percent. b. Significant at 10 percent. G: c. Significant at 5 percent. HEAL TH AND EDUCA TION 129 THE WELFARE IMPACT OF RURAL ELECTRIFICATION Table G.11: Reading and Studying Time Children Adult Propensity Propensity Study time to study Read time to read Housing index 0.37 ­0.03 0.41 0.15 Education (no. of years) of head of household 0.12a 0.01 0.07a 0.07a Age of the head (log) of household 0.03 0.10 Occupation of head of household Farmer 0.30 0.01 ­0.36b 0.00 Home business 0.33 0.11 0.01 0.18c Children in the household Number 0.66a ­0.16a Child of more than 10 years age (0/1) 2.56a Children in school (0/1) 0.97a Household size 0.04a Source of light: Grid 1.28a 0.11 0.45b 0.06 Kerosene ­0.37 ­0.30 Candle ­0.34 ­0.17 Others 0.02 0.26 Regional dummies suppressed Share of households with electricity (cluster average) 0.14 0.26 Constant ­4.18a ­0.80b 2.14a ­1.27b Observations 1,152 1,992 1,204 1,992 Wald Chi 10.50 4.76 Sources: ESMAP 2000, 2003. Note: Ideally this equation should be estimated at the child level. Given data limitation (missing household roster and time-use section), the model uses household- level information and total study time per household. The analysis is based on data collected from 2,000 households in four regions of the island of Luzon. a. Significant at 1 percent. b. Significant at 5 percent. c. Significant at 10 percent. Table G.12: Reading and Studying Time (Propensity Score Matching) Electrified Nonelectrified Nonelectrified household in household in Regression household in electrified villages electrified villages coefficient nonelectrified villages ATT Study time 1,366 260 1.405a 346 1.374a Reading time 1,366 224 0.950a 298 0.960a a. Significant at 1 percent. 130 APPENDIX H: CALCULATING CONSUMER SURPLUS The Theory which is simply a transfer payment to the utility The theory of calculating the benefits of RE as the and so a neutral flow for economic analysis.1 The increase in consumer surplus is relatively straight- cost side of the analysis will capture the cost of forward. The difficulties come in applying that consumption. Assuming that the average cost of theory. supply (Ce) is less than the tariff rate, there will be a positive producer surplus, which is being cap- Electricity supply lowers the cost of energy to tured in this calculation (figure H.2). The alter- the user, resulting in an increase in consumer native is to deduct the payments (D + E) from surplus, which is the difference between what consumers but then add them to producers, so the consumer is willing to pay and what they ac- when summing across all flows, these pay- tually do pay. Assume that before electricity, en- ments/receipts cancel out. ergy is supplied from a single source, kerosene, at price Pk with consumption Qk (figure H.1). A complication in the above is that, in drawing the Once electricity is available at lower price Pe, con- demand curve between the two observed points, sumption rises to Qe. Using these two points, the it is assumed that other characteristics that affect demand curve may be interpolated; exactly how demand are the same for those households for this is done is one of the important issues con- which (Pk,Qk) is observed and those for which sidered below. (Pe,Qe) is observed. This is almost certainly not the case. Electrified households have higher incomes The amount the consumer is willing to pay for a than nonelectrified, and the average income in quantity Q is the area under the demand curve electrified communities is higher than that in non- from 0 to Q. Hence the consumer is willing to pay electrified ones. Energy is a normal good, demand A + B + D for consumption of Qk but actually pays for which rises with income, so (Pe,Qe) in fact lies B + D (= Pk Qk), leaving a consumer surplus of on a higher demand curve than does (Pk,Qk). A. Once electricity becomes available, the con- sumer surplus is A + B + C, so the increase in con- The consumer surplus for customers who are al- sumer surplus as a result of electrification is B + ready connected is thus underestimated by this C. This consumer surplus has two parts: that aris- approach. But the consumer surplus for those ing from the reduction in the price of the Qk who will become connected (if they do despite units already being consumed and that associ- their lower income) is overestimated. In addi- ated with the new consumption, Qe ­ Qk. tion, when benefits are projected into the future, real income growth will shift the demand curve The benefit to the consumer is B + C. It is com- to the right over time so that consumer surplus mon in project analysis to also include areas D + is increasing. The value of this additional con- E, sometimes referring to the whole area B + C sumer surplus can be calculated if the income elas- + D + E as the willingness to pay (WTP), which ticity of demand for energy is known. If this is not quite correct because WTP includes area A elasticity is known, along with the average in- also. It is perfectly acceptable to include areas D come of currently connected and unconnected + E. This is the amount paid by the consumer, households, then different demand curves can 131 THE WELFARE IMPACT OF RURAL ELECTRIFICATION Figure H.1: Consumer Surplus Figure H.2: Producer Surplus 12 12 10 10 8 8 Pk A 6 6 Price Price B C 4 4 Pe Pe Producer surplus 2 2 D E Ce Cost of production 0 0 0 Qk 20 40 60 80 Qe 100 120 140 0 20 40 60 80 Qe 100 120 140 Quantity Quantity Note: Pe = price of electricity from the grid; Pk = price of kerosene; Qe = quality of electricity Note: Ce = average cost of supply; Pe = price of electricity from the grid; Qe = quantity of elec- used from the grid; Qk = quantity of kerosene consumed. tricity used from the grid. also be drawn for these different groups to more Units for the horizontal axis accurately measure consumer surplus. The above example puts energy on the horizon- tal axis. To apply the approach, a common met- Issues in Applying the Theory ric is needed. Nonelectrified households rely on There are several complications in applying this a variety of energy sources, often woodfuel for approach in practice: data availability, the common cooking, kerosene for lighting, and, among the metric to be used on the horizontal axis, the better off, car batteries for TV. The mix among shape of the demand curve, and the way in which these varies by household. In principle, total en- consumer surplus is expressed. ergy consumption can be converted to a common metric such as kilowatt hour or kilograms of Data availability oil equivalent, and the average price can be cal- Although the analysis is based on just four pieces culated by dividing total consumption by energy of information (Pk, Qk, Pe, and Qe), these are not expenditure. readily come by. A recent energy household sur- vey (or expenditure survey with a good energy There are two problems in this approach. First, module) is necessary to estimate average prices the energy consumption mix varies across house- and quantities for electrified and unelectrified holds, so taking a representative household might households. But it is also necessary to control for be misleading as coverage expands and different income differences between these two groups. households become connected to the grid. Sec- The simplest way to do this is to restrict the sam- ond, electrified households typically still use other ple to communities whose average income is energy sources, so the comparison is not a within a certain range or to take a small number straightforward one between electricity and an al- of subsamples with different average incomes, ternative energy source. These problems might which is a better approach if future RE will ben- be partially sidestepped by valuing consumers efit communities with a different profile than separately by end use. In this case, end use- those that have already been electrified. specific units can be used for the horizontal axis, 132 APPENDIX H: CALCULATING CONSUMER SURPLUS such as lumens for lighting (see below) or hours To calculate total project benefits, this value can for TV watching. be multiplied by the cumulative number of house- holds connected to the grid each year. The analy- Shape of the demand curve sis can be a bit more sophisticated if there are The value of area C in figure H.1 clearly depends different household types with different levels of on the shape of the demand curve. The simplest consumer surplus, but this is quite a demand in assumption is to assume a linear demand curve terms of data. There is also the problem that not (that is, a straight line) between the two observed all consumption is residential, so these calculations points, making the calculation of area C very need to be repeated for different end users (com- straightforward = 0.5(Pk­ Pe)(Qk­ Qe). But if the mercial, agricultural, and so forth). demand curve is convex to the origin, as theory suggests, then the linear demand curve overesti- The alternative approach is to express consumer mates the amount of consumer surplus. An al- surplus per kilowatt hour and then multiply the ternative functional form is a constant elasticity resulting value by total electricity sales each year. (that is, log linear) demand curve. The equation This approach does not, of course, bypass the for this demand function can be written as follows: problem of their being different types of con- sumer: the average consumer surplus per kilowatt P = K Q, hour should be a weighted average for the dif- ferent end users. There are two problems here: where is the elasticity and K a constant. The elas- (1) if the share of end users varies over time, the ticity can then be calculated as weighted average is no longer the appropriate one, although the resulting bias is unlikely to be large; = ln( P )- ln( Pe ) k and (2) the calculation of the average consumer . ln(Q )- ln(Qe ) surplus/kilowatt hour is itself biased, though again k probably not by much. Once the parameters and K are obtained, then area C may be calculated as Consider the case of three households shown in table H.1; consumer surplus rises with con- Qe sumption but at a diminishing rate, a fact that C = KQ dQ -(Qe - Qk )Pe arises from the shape of the demand curve. Av- Qk erage consumption is 25 kWh/month, the same = K +1 as household 2, and corresponds to an average + 1 (Q+1 - Qk )-(Qe - Qk )Pe. e consumer surplus of Rs 2/kWh. Because total con- sumption is 75 kWh, applying this average suggests Note that the deducted amount (Qe ­ Qk)Pe (the a total consumer surplus of 150, whereas the true amount paid for incremental consumption = figure is 145. The inaccurate result comes be- area E in figure H.1) is usually included in proj- cause the method of calculation implies a constant ect benefits (because the economic costs of pro- duction are deducted separately) so the second Table H.1: Consumer Surplus for term need not be deducted, the first term giving Different Households C + E. Consumption Consumer Expressing consumer surplus Household (kWh/month) surplus (Rs) The example given here uses either the market 1 20 40 demand curve or that for a representative house- 2 25 50 hold. Assume the latter. The calculation therefore 3 30 55 gives the value (in local currency) of the increase Total 75 145 in consumer surplus from connecting to the grid. Note: Rs = rupees. 133 THE WELFARE IMPACT OF RURAL ELECTRIFICATION increase of consumer surplus with respect to con- Table H.2: Wattage of Common sumption. The same problem arises if multiply- Household Appliances/Tools ing the number of connections by the consumer surplus per connection; such an approach also Appliance Resistive load (watts) yields an estimate of a total consumer surplus of Clock radio 5 150 in this example. Computer-PC 300 Deep fryer 1,800 In summary, the preferred method would be to Iron 1,200 calculate consumer surplus by end user (possibly TV, color 300 even different categories of each type of end Fan 30­70 user) and thus calculate consumer surplus based on cumulative connections for each type of end user. In practice, however, the data requirements (by a factor of 2.5 or more) and the price far for this approach are substantial, so a more com- lower (by a factor of X or more) for electricity than mon approach is to estimate consumer surplus for the next lighting source (see table H.3 for per kilowatt hour (though this should be a examples). weighted average of different end users) and mul- tiply this amount by sales. This approach suffers Once these data are available it is a straightforward from an unknown bias from the sales composition matter to calculate the change in the consumer effect and an upward bias from ignoring dimin- surplus and the WTP. Table H.3 shows the results ishing consumer surplus as consumption rises. of this calculation, assuming both a linear and a This bias is offset by the downward bias from ig- log-linear demand curve, showing that assuming noring growing consumer surplus as income rises. a linear demand curve can overestimate increase Although the balance is not known, the two can- in the consumer by a factor of up to eight times. cel each other out to some extent, so the net bias will not be too great. Overview of Bank Approaches to Measuring Electrification Benefits Applying the Theory: Examples Bank project documents adopt a range of ap- proaches to valuing electricity benefits. The focus Lighting here is on approaches based on consumer surplus. Lighting is the main use of RE (see appendix D) A detailed review is presented in the table of ERR and has been the focus of Bank efforts to calcu- calculations beginning on page 136. late the consumer surplus from electricity. In- deed, the lighting benefits are often the only ones As noted earlier, studies commonly estimate area valued for residential consumption. In the 1990s, B + C + D + E as the benefit, calling this WTP. The project documents typically compared the cost of same terminology is used here, noting that WTP lighting using a kerosene lamp with the cost of the should in fact also include area A, but that area A same amount of lighting from electricity. But the is not part of the project benefits (that is, it is the ESMAP Philippines study (ESMAP 2003) proposed terminology that is slightly wrong, rather than a a new approach based on lumens, which is a conceptual error in the method of calculation). The measure of emitted light. Table H.2 shows the lu- following main approaches can be identified: mens emitted by some typical light sources. · Estimate WTP assuming a nonlinear demand Using these conversion factors and data from a curve or assuming a linear demand curve but household energy survey, it is possible to obtain taking only a percentage of the estimate for estimates of Qk and Qe (quantity of lumens con- area C to allow for the overestimation. This ap- sumed for kerosene and electricity, respectively) proach conforms with what is considered by this and the corresponding prices (Pk and Pe). This ap- review as best practice, although allowance proach finds that the consumption is far higher should also be made for the different income 134 APPENDIX H: CALCULATING CONSUMER SURPLUS Table H.3: Demand for Lumens from Different Energy Sources Quantity (kilolumens) Price (US$ per lumen) Consumer surplus Ratio Grid Grid Linear linear/log Country Kerosene connection Kerosene connection curve Log-linear linear Bolivia 7 90 0.48 0.04 21.3 8.6 2.5 Lao PDR 20 435 0.195 0.003 43.7 9.9 4.4 Peru 5 363 0.57 0.01 102.9 12.5 8.2 Philippines 4 204 0.36 0.0075 36.8 5.8 6.3 Source: Project documents. of current and future consumers and income level. This approach neglects the downward growth among consumers. sloping demand curve, resulting in a substan- · Estimate WTP assuming a linear demand curve. tial overestimate of project benefits. This approach results in an overestimate of · Estimate benefits as the cost savings on current project benefits. consumption levels (that is, area B in figure · Estimate WTP based on the alternative energy H.1). This approach underestimates project source, and then value the whole of expected benefits, as it ignores additional consumer sur- energy consumption with electricity at that plus from new consumption (area A). 135 136 THE Economic Rate of Return Calculations WELF WTP ERR ARE IEG revised IEG Country Project Document Year Method Comment Estimate estimate Estimate revised IMP ACT Lao PDR Southern Provinces ICR 2004 Area under linear demand A single curve should be 72.00 5.60 60.5 12.6 Rural Electrification curve, using three separate used passing through the OF curves for four data points four points, assuming constant elasticity rather RURAL than linear. Malawi Power V SAR 1992 Cost of avoided diesel USc33/kWh ELECTRIFICA generation Indonesia Rural Electrification SAR 1990 USc15/kWh 15 Project TION Indonesia Rural Electrification ICR 1995 Linear demand curve be- Should use constant elas- Residential: Project tween two points ticity demand curve, and USc19/kWh given data they had Commercial: should estimate actual USc16/kWh Qk and Qe. Indonesia Solar Home ICR 2004 Begins with an avoided cost $168/million 184.56 40.9 40.9 Systems Project estimation of CS based both lumen hrs. on energy expenditures data and on response of TV/radio use to presence of SHS. Then computes WTP using log- linear demand curve based on average monthly energy expenditures. Nepal Power Sector SAR Takes price currently paid as 12.1 6.6 Efficiency Project WTP for all consumption (but also does lower bound based on actual tariff). Peru Rural Electrification PAD 2006 Constant elasticity demand Best practice PAD analy- TV: 2.26 S./kWh, Project curve, estimated separately sis, including Monte lighting: USc111/kWh for lighting and TV. Presents Carlo simulations for sen- Total USc179/kWh separate estimates for WTP sitivity analysis. by level of consumption (not clear which are used). Pakistan Rural Electrification SAR 1989 CS estimated separately for Overestimate as WTP de- USc10/kWh USc13/kWh 10.9 22.9 Project different end users, based on clines as consumption expenditure on current en- rises, but underestimate ergy source. Benefits are CS, as double count costs (as not WTP. use CS not WTP); latter effect is larger. Bangladesh Rural Electrification SAR 34 Project III Bangladesh Rural Electrification SAR 1990 They use a discounted sales- Insufficient data to con- Tk6.2/kWh 36.5­58 Project III weighted average WTP firm that estimates are based on avoided cost esti- correct mation for kerosene and the change in CS due to a switch to electricity averaged as an APPENDIX upper bound, and the fiscal 1990 electricity tariff rate as the lower bound. Bangladesh Rural Electrification ICR Similar method used as in 95.3­102.5 H: Project III SAR, but with benefits de- CALCULA fined for only the first 10 kWh/month out of an aver- age of 37 kWh/month. TING Bangladesh Rural Electrification SAR 12.7 Project II CONSUMER Bangladesh Rural Electrification PCR 1995 WTP for rural and residential Should use constant elas- Residential: Residential: 13.6 ­ Project II consumers separately using ticity demand curve. Lim- USc105/kWh USc54 /kWh linear demand curve ited data given to estimate this, but WTP SURPLUS approximately 50% that used in PCR. This lower 137 (Table continues on next page) 138 THE Economic Rate of Return Calculations (continued) WELF WTP ERR ARE IEG revised IEG Country Project Document Year Method Comment Estimate estimate Estimate revised IMP value means net benefits ACT are always negative (hence ­ ERR). OF RURAL Bangladesh Rural Electrification PCR 1993 WTP for rural and residential Should use constant elas- n.a n.a 12.6 n.a. Project consumers separately using ticity demand curve. In- linear demand curve sufficient data in PCR for ELECTRIFICA re-estimation. Bangladesh Rural Electrification SAR 11.4 Project TION Morocco Rural Electrification SAR 1990 Benefits are revenues plus Insufficient information n.a n.a 17.8 n.a Project II consumer surplus. No further to determine method or details given. perform re-estimation Morocco Rural Electrification ICR 1998 Benefits are revenues plus Insufficient information n.a n.a 15.3 n.a Project II consumer surplus. No further to determine method or details given. perform re-estimation Senegal Electricity Services SAR/ WTP calculated from linear Insufficient information USc72/kWh n.a. 28.4 24 for Rural Areas PAD and linear in log-log scale. to recalculate WTP Project Sri Lanka Energy Services ICR 2003 Mini hydro appraisal benefits Not enough data is pro- USc57/kWh (avoided 0.001 n.a. Delivery Project calculated using avoided cost vided to accurately recre- cost) for mini hydro; Rs/lumen hour approach, and ICR uses finan- ate net benefit stream, so 0.001 Rs/lumen hour for SHS; 0.30 cial model to create financial levelized stream used, re- for SHS, 0.70 Rs/TV- Rs/TV hour for internal rate of return to con- sulting in higher ERR. hour for SHS SHS firm ERR from PAD. Village- based hydro uses avoided cost for assessing consumer benefit, but source not given--appears to be from a linear demand curve. SHSs component uses log-linear demand curve to derive WTP. Bolivia Decentralized PAD 2003 Taken as one-quarter of n.a. n.a. 29.1 29.1 Infrastructure for (Pk ­ Pe)*(Qe ­ Qk) to avoid Rural overestimate using a linear Transformation demand curve. Cape Verde Energy and Water PAD 1999 Not calculated n.a. n.a. 12 n.a. Sector Reform and Development Project Mozambique Energy Reform and PAD 2003 For EdM grid intensification EdM Grid: 23 23 Access Project component WTP is area USc13/kWh; Mini- under semi-log demand curve grid: USc18/kWh times factor of 0.8 to mini- (weighted avg. for mize overestimation; mini- residential and com- grid components estimated mercial) in N. Inham- APPENDIX similarly for North Inham- bane and also bane and Mocimboa da Praia USc18/kWh in for residential and commer- Mocimboa da Praia cial users, and photovoltaic (estimated sepa- component estimates also rately); PV: H: using semi-log, but for both USc13/lumen-hour for CALCULA lighting and TV/radio con- TV/lighting sumption. WTP estimates re- flect weighted averages of subgroups/components. TING Guinea-Bissau Energy Project ICR 1998 WTP from linear demand Data for determining n.a. n.a. n.a. CONSUMER curve WTP not provided Malaysia Rural Electrification PCR 1990 WTP not provided, but Data for determining USc58/kWh n.a. 26 n.a. Project avoided cost calculated WTP and ERR not provided SURPLUS Note: All dollar amounts are US dollars. CS = consumer surplus; ERR = economic rate of return; ICR = Implementation Completion Report; kWh = kilowatt hour; PAD = project appraisal document; PCR = project completion report; PV = photovoltaic; Rs = rupees; SAR = staff appraisal report; S. = soles; SHS = solar home system; Tk = taka; USc = US cents; WTP = willingness to pay. 139 APPENDIX I: EVALUATION APPROACH PAPER Rationale · Increasing income This evaluation is the fourth in IEG's current pro- · Contributing to better health gram of impact evaluations.1 Infrastructure was se- · Supporting education lected, as the sector has not been covered in the · Improving women's quality of life program to date. The RE subsector was selected · Reducing environmental harm. for the following reasons: Accordingly, it is argued that investment in RE · It has not been the subject of an IEG evaluation can make a major contribution to achieving sev- since 1994. eral of the Millennium Development Goals, notably · The 1994 IEG study found that the costs of in- in Africa, where coverage rates in rural areas in vestments in RE did not appear to be justified many countries are between 1 and 2 percent (see by the benefits, although there was need for fur- attachment 1). In accordance with the theory- ther investigation. In response, more recent based evaluation approach adopted in the IEG work by the Bank in the Philippines quantifies impact evaluations, the study will seek to unpack a broader range of benefits, stating that the re- the channels though which these poverty impacts sults demonstrate the possibility that "benefits can be felt (see table I.1). will outweigh the costs of extending electricity service" (ESMAP 2003). This evaluation will But when IEG last reviewed RE 12 years ago, it was make an independent assessment of this state- critical of the limited benefits realized by such in- ment and the methodology used to reach it. vestments, which appeared insufficient to justify · The question of the viability of these invest- the costs (see box I.1). Despite that finding, lend- ments is of operational significance because the ing for RE has grown since the mid-1990s, spurred RE portfolio is growing in size, especially given in part by the growth of a portfolio of projects sup- the development community's new emphasis porting renewable energy. There were just 10 on renewable energy sources. Off-grid electri- projects with an RE focus in the years 1990­94, fication was ignored in the 1994 IEG study but compared with 23 for 2000­04; the number of will be covered in this evaluation. IEG's recent projects with an RE component grew from 14 to review of renewable energy (which is largely off 42 over the same period. grid) concluded that the "poverty reduction im- pact is largely nonevaluable" on account of In response to the IEG report, operational staff lack of evidence (IEG 2006). This evaluation will have introduced new evaluation tools to cap- help fill that gap. ture a broader range of benefits,2 with results so far available for the Philippines and work ongo- Background ing in Bangladesh and Vietnam. This impact eval- Energy policy and services are linked to poverty uation will take a critical look at these new reduction by the following possible benefits findings, undertaking new analysis of existing (World Bank 2005): and new data. 141 THE WELFARE IMPACT OF RURAL ELECTRIFICATION Table I.1: Benefits from Rural Electrification RE-affected input Channel Direct Indirect Lighting Time use Time-saving devices Richer social life Increased study time Electric lighting replaces Improved indoor air quality other fuels (in principle for cooking also, but rarely in practice) Media access (radio, TV, and Improved health knowledge Health and nutrition Health and nutrition Internet) Fertility Fertility Entertainment Fan/air conditioning Improved living conditions Greater comfort Facilities Better social facilities with Clinics: longer hours and better equipment more equipment, including cold chain for vaccines, Internet access Schools: available for adult literacy in evenings; computer facilities Water pumps: cleaner water supply Productive enterprise Electrical equipment for Enhanced productivity (in- Increased income and workshops and agriculture cluding irrigation) employment (including lighting and Longer business hours pumps) Food preparation Refrigeration and boiling Better nutrition and reduced water ill health Community lighting Improved security Richer social life An important development in the portfolio in This evidence will combine new analysis of new the last decade has been the growth of lending for data for one country (Lao PDR) and reanalysis of off-grid electrification. These investments were not existing data for 10 others. These findings will be considered in the 1994 study (which was not an combined with existing evidence to form a com- impact evaluation), but this new study will con- prehensive summary of what is known about the sider both on-grid and off-grid electrification. A impact of RE, and the part played by external final rationale is that there are few impact stud- agencies, in particular the World Bank. ies on RE. This approach departs from the single-country The evaluation will be a meta-impact analysis, focus of previous IEG impact studies. This alter- drawing on evidence from a number of sources. native approach is being used in this particular 142 APPENDIX I: EVALUATION APPROACH PAPER Box I.1: The 1994 Study Rural Electrification in Asia The main findings of the 1994 IEG study were as follows: to peak demand (rural use is concentrated in early evening, · Ex post economic rates of return were much lower than those whereas urban demand is spread across the day). Cost recov- at appraisal, as many of the indirect and external benefits had ery has been low (10­50 percent), thus imposing a financial bur- not materialized. Notably, there was little impact on industrial den on the electricity utility or government. development. · The direct benefits of RE go to the non-poor. Even with low tar- · RE projects ignored financial aspects. Unit investment costs for iffs the poor cannot afford connection costs. The poverty- RE are much higher than for urban electrification because of reduction benefits are thus indirect through rising rural incomes, lower population density and the low ratio of average demand and these effects have been found to be limited. case to broaden the operational relevance of the connections? What is the distributional profile study and because it is feasible in this case, given of those taking connections? What are the unit the limited range of other studies to be covered. costs of connection by type of supply to the user and to the supplier? Channels for the Welfare Impact of RE · What are the direct economic benefits from RE? The direct benefits from RE flow to households Who gains these benefits? What are the indirect or businesses that get connections. Indirect ben- economic benefits (employment generation), efits arise either from the income opportunities and who gains them? How does the distribu- overspilling to others or from benefits to uncon- tion of benefits change as coverage of electri- nected households from a connection in the com- fication programs expands? munity. For example, villagers may watch television · What is the impact of RE on time use, and what in a community rather than household setting. are the welfare implications of these changes for health, education, and increased leisure? Given the difficulty of quantifying all these ben- · How does RE affect the quality of health and efits, most studies use estimates of the WTP to cap- education services? ture electrification benefits. WTP is an indirect · How do the aggregate private benefits and the measure, assuming that how much people are will- public good benefits compare to the WTP? ing to pay gives a good measure of the value of What is the distributional profile of these the benefits. However, estimating WTP requires benefits? some strong assumptions and, even if it is done · What are the private and social rates of return correctly, it ignores the public good benefits from from investments in RE? electrification.3 Hence, direct measurement of the benefits, as proposed here, is to be preferred. Evaluation Approach and Data Requirements Evaluation Questions The evaluation approach mostly relies on new The evaluation questions address the realization and existing survey data to quantify the benefits of the claimed benefits of RE and the extent to from RE. Qualitative information shall come from which they are gained by the poor. The ultimate existing material through the desk review and objective is calculation of private and social rates Project Performance Audit Reports (PPARs) and of return from investments in RE. Specific ques- a qualitative component in the in-depth country tions are as follows: case study of Lao PDR. · What has been the growth in the coverage of The two main challenges in conducting an impact RE in countries receiving Bank support? To evaluation are contagion (the control becoming what extent has the Bank contributed to these treated) and endogeneity (the selectivity bias in 143 THE WELFARE IMPACT OF RURAL ELECTRIFICATION who is treated). Contagion takes two forms: among the countries included for analysis of spillover effects, which are dealt with directly in DHS data, and new analysis of the LSMS (a this study, and treatment in control communi- household income and expenditure survey) ties, which is not an issue for an ex post analysis data will be undertaken regarding test scores such as this because the control can be restricted and possibly rural enterprises. to uncontaminated communities. Those who re- · Sri Lanka: Sri Lanka has a well-documented RE ceive electrification (at both household and com- program, especially its experience with off-grid munity levels) are better off than average, so electrification. The Bank has supported four RE there is a problem of selectivity bias. However, the projects in the country. A survey was under- determinants of selection (income and geo- taken in 2002 as part of the EnPoGen project; graphical location) are observed, so the bias can IEG may reanalyze this rich but underexploited be removed. data set.5 In-depth country case study The case study evidence shall be synthesized on An in-depth country study comprising a house- a thematic basis. The synthesis will also include hold survey and a qualitative study will be con- impact studies conducted for other countries. ducted in Lao PDR, which has had four RE projects, one of which is ongoing. Further analysis of DHS data The DHS contain data on a range of health and In Lao PDR, IEG will commission a structured fertility outcomes, as well as output measures on survey with a sample of 1,200­1,500 households, knowledge and practice. The household data following up a survey ESMAP conducted in 1997. contain a variable on electrification (though not A qualitative study of the impact of RE shall also the source) and variables that can be used to be undertaken. construct an asset index to proxy for income (which needs to be controlled for).6 Several coun- Desk reviews tries have DHS data for more than one year, which Desk reviews will be undertaken for the follow- will allow analysis of the changing distributional ing countries:4 pattern of access to electricity. · Bangladesh: The Bank has supported three Because the DHS questionnaire is standardized RE projects in the 1990s and a fourth project across countries, the data are suitable for pool- with an RE component. There is a continuing ing--or at the very least in realizing economies of program. Along with the Bank, USAID was the scale in estimating the same models for multiple main financer of the program and undertook data sets. Whether pooling the data will strengthen one of the most substantial RE impact evalua- the analysis will be determined once the country- tions of any program anywhere in the world. level analysis is completed. In addition, there are PPARs of the last two Bank projects. RE was covered in the IEG im- The countries to be included in this analysis are pact study, and Bangladesh shall be included Bangladesh, the Philippines, Indonesia, Ghana, among the countries in which analysis of DHS Vietnam, Pakistan, Morocco, Uganda, Guinea, data is carried out (see below). and Senegal. All these countries have received · Philippines: The Bank supported RE through Bank support for RE. For each country a mini two projects in the 1990s, for which there are desk review will be undertaken to provide the PPARs. The only ESMAP study on economic country context. and social benefits published so far is for the Philippines (2003). The Philippines is also Review of PPARs among the countries included for analysis of A review shall be made of PPARs for RE projects DHS data. completed since the 1994 IEG review. These cover · Ghana: RE was supported through two Bank (see attachment 2 for a complete list): Bangladesh, projects, one of which has a PPAR. Ghana is also Ghana, India, Indonesia, Kenya, Lao PDR, Mozam- 144 APPENDIX I: EVALUATION APPROACH PAPER bique, Níger, Pakistan, the Philippines, and Sri at least at the level of sharing approaches and Lanka (and one for Uganda, which is currently findings. The study has been discussed with staff being prepared). of ESMAP and the Global Environment Facility, and we shall share findings as they emerge. Review of existing impact evaluations This review will be conducted early in the study Advice from both internal and external peer re- to inform the evaluation design, but the results viewers will be sought on intermediate and final will also be utilized. The identified studies come findings. from three sources:7 · ESMAP: Quantification of benefits for the Schedule and Task Management Philippines and women's time-use in India, The schedule for the study is as follows: and other pieces of ongoing work (for exam- ple, study time in Ecuador and indoor air qual- · Inception phase, June­September 2006: De- ity in India) velopment of approach and design papers, · EnPoGen (Energy, Poverty, and Gender):8 and visit to selected case study country (pos- Country case studies for China, Indonesia, and sibly October); review of existing impact eval- Sri Lanka produced in 2003 uations · USAID: Impact studies for USAID's support to · Preparatory phase: September­October 2006: RE in Bangladesh (Barkat and others 2002) Questionnaire design and collection of and Colombia (Davis and Saunders 1978). materials · Desk reviews: September­December 2006 The meta-analysis · Analysis of existing data sets: October­ This meta-analysis draws together the evidence November 2006 from the approximately 20 countries included in · Survey: fielded by December 2006, with clean the study. A matrix will be constructed in which data by February 2007 the research questions (or more detailed ques- · Data analysis and synthesis: February­March tions derived from those questions) are the rows 2007 and the countries are the columns. As many cells · Report writing: First draft by mid-April, final re- as possible will be filled, but using only those port to CODE by June 2007 findings based on technically rigorous methods. · Dissemination: September­December 2007. Thus, a summary of all available evidence on each Dissemination includes the usual report dis- question can be made in a systematic manner. tribution internally and externally, plus pre- sentations to targeted agencies with an interest Collaboration with Other Agencies and in RE (for example, USAID, Norad, and the Peer Review Swiss aid agency Seco). Additional publications Collaboration shall be sought with relevant gov- from the study shall be prepared as part of ernment officials or research institutions in the dissemination. country selected for in-depth analysis. Funding from the Norwegian Agency for Development Co- The evaluation will be carried out by a team of IEG operation (Norad) partnership has been sought. staff and consultants with the assistance of in- Norad is embarking on its own impact study of RE, country consultants for the survey under the task and possibilities for collaboration shall be sought, management of Howard White. 145 THE WELFARE IMPACT OF RURAL ELECTRIFICATION Attachment 1: Rural Electrtification Coverage (DHS Data) 100 90 80 70 60 50 Percent 40 30 20 10 0 4 7 2004 200098/98 2003 200190/91 2003 200 2Bolivia 2003 1996 2005 200298/99 2000 200 u 2000 1 1999 2000 2000 2002 199 1999 1977 2000 1996 an an tan an lic 19 Yemen hst Brazil Haiti ragua Per India 19onesiaNepalan19 Benin 2001o 2003 200494/95 2004 1991998/99 2002 2000 2000 2003 1999 1998 2003r 2004 2Mali 20a 2000/01 2003aNigerNigeria 2003SenegalAfrica 1998ia 2004o 199800/01 6 200 1997 0 000 01 2000 1998 19 20 2001/02 Fas eroonCAR Chad que ibi Comorosoire Eritrea ea 2003­200 Eth iopiaGabonGhanaGuin KenyaKenya ascaMalawi Tog Rwanda eniaEgyptJord Vietnam Cam d'Iv ritani zambi Nam Tanzan Arm occo Republicnis Uzbekist Colombia Bangladesh Cambodia Ind Philippines Repubtemala Nica Burkina Madag South UgandaZambiaZimbabwe Mo Kazak Turkme Pakist can Cote Mau Mor Kyrgyz Gua Domini Sub-Saharan Africa Middle East Central East and Latin America and Asia South Asia and the North Africa Caribbean Attachment 2: Available PPARs for Review ICR Fiscal outcome PPAR Loan size Country Project name years rating rating (millions) Bangladesh Third Rural Electrification 1990­2000 HS HS 105.0 Ghana Fifth Power Project 1990­97 S MS 40.0 India Renewable Resources Development 1993­2002 S S 190 Indonesia Kecamatan Development Project 1998­2002 HS MS 225 Kenya Geothermal Development and Energy Pre-Investment Projecta 1989­96 S S 40.7 Lao PDR Provincial Grid Integration Project 1993­99 S U 36 Mozambique Urban Household Energy Project 1989­98 U MS 22 Niger Energy Project 1988­97 S MS 31.5 Pakistan Rural Electrification Project 1990­97 U U 160.0 Philippines Rural Electrification Revitalization Project 1992­98 U U 91.3 Philippines Energy Sector Project 1990­96 S MU 390 Sri Lanka Energy Services Delivery Project 1997­2003 S HS 24.2 Sri Lanka Second Power Distribution and Transmission Project 1992­98 S MS 50 Note: HS = highly satisfactory, MS = moderately satisfactory, MU = moderately unsatisfactory, S = satisfactory, U = unsatisfactory. a. RE component was a study. 146 ENDNOTES Chapter 1 covery policies for public sector projects: general as- 1. World Bank Group Historical Chronology pects" (March 1977) and 3.72 "Energy, Water Supply, and http://siteresources.worldbank.org/EXTARCHIVES/ Telecommunications" (September 1978), though these Resources/WB_Historical_Chronology_1944_2005.pdf. made allowance for flexible application with below- 2. Rural electrification (RE) is not precisely defined cost recovery charges permissible on distributional but may often include some areas that are at the very grounds. least peri-urban, either because, as in Bangladesh, vil- 2. Projects excluded were those belonging to the lages are very large or, as in Ghana, RE programs include following categories: projects for power plant con- district towns that serve surrounding areas. struction with little or no grid extension, even when this 3. These countries were Bangladesh, Ghana, Lao plant would eventually supply rural areas; projects for PDR, Morocco, Nepal, Nicaragua, Peru, the Philippines, rehabilitation of existing electricity infrastructure only; Senegal, and Sri Lanka. projects that did not seem to contain a specific RE as- 4. These countries were Bangladesh, Ghana, In- pect, even where these cover large areas; and projects donesia, Morocco, Nepal, Nicaragua, Peru, the Philip- with no project appraisal document. pines, and Senegal. 3. One issue is the substantial scale of investments 5. These data sets were of variable quality. The needed against available International Development As- ESMAP survey of the Philippines (ESMAP 2003) has sociation resources, which limits the Bank's ability to been the best designed to date, though unfortunately tackle RE in Africa. But it should also be recognized that the complete data set is no longer available, which re- unit costs in many African countries are high and rural stricted some of the analysis. The Lao PDR survey was population densities and incomes low, so structural based on the Philippines survey, but problems in sur- inefficiencies need to be addressed before RE can be vey design and execution also limited the analysis (for tackled on a financially sustainable basis. example, the agricultural production and time use 4. The fact that a smaller percentage of projects have modules). The Sri Lanka survey had a rather different institutional development (ID) components than have focus and thus had limited usefulness for impact analy- ID objectives suggests some mismatching of objec- sis. A recommendation of this report is for more tailor- tives and components. But examination of projects made surveys to assess the impact of RE programs. with ID objectives but no ID category shows specific 6. Income may be regarded as endogenous with re- reasons, such as multisectoral projects whose ID com- spect to electrification, creating a problem if data prior ponents were classified as "nonelectrification." to electrification are not available. However, income may 5. The 120 projects do not include all RET projects, be instrumented either with assets or, if these are also because some of these do not qualify as RE. IEG's re- thought to be endogenous, then fixed household char- view of RETs listed 65 projects for the period 1990­2005 acteristics such as education and sex of household (IEG 2006). Although many of these activities are rel- head. atively new to the Bank, the exception has been hy- dropower. This has been a major investment line for Chapter 2 the Bank that had declined in recent years because of 1. Cost recovery policies were embodied in the mounting pressure against large dams. During the policy statements in Operational Manual 2.25: "Cost re- three decades up to 1995, the Bank financed 110 147 THE WELFARE IMPACT OF RURAL ELECTRIFICATION hydroelectric projects in 50 countries (Besant-Jones Chapter 4 1995). From 1970 to 1985 the Bank was involved in an 1. "It is not easy for [women] to watch TV in other estimated 3 percent of the new dam projects around people's houses after dusk, defying the purdah barrier" the world, but this had fallen to 0.6 percent by March (Siddiqui 2000, p. 253). 2001 (World Commission on Dams 2001). 2. It is surprising that average consumption for 6. From the International Bank for Reconstruction lighting is less in the Philippines than in Lao PDR. This and Development (IBRD), $2.2 million, and from GEF may be a seasonal effect. The Philippines survey took $0.7 million for off-grid and $31.2 for grid extension for place in June/July, when daylight hours are at their Southern Provinces Rural Electrification Project; and peak of around 12.5 hours. The timing of the Lao PDR $6.13 million from IBRD and $3.75 from GEF for off-grid survey is not known, but if it were early in the year, then for the Rural Electrification Project. daylight at the time of the survey could be 1.5 hours 7. In the Implementation Completion Report pre- less. For a household using just two 60-watt bulbs, this pared by operational staff, projects are rated on a four- difference in daylight hours would account for a dif- point scale--"highly unsatisfactory," "unsatisfactory," ference of 5.4 kWh per month of electricity use for "satisfactory," and "highly satisfactory." IEG independ- lighting. ently verifies these ratings based on a desk review, and 3. The most detailed treatment of this topic is in 25 percent of cases carries out its own fieldwork, the Winther's (2005) anthropological study of RE in Zan- findings from which are reported in a Project Perfor- zibar. In a discussion of fear of witchcraft among those mance Assessment Report. The ratings given here are acquiring electrical goods, a footnote reveals that the those from IEG's database. anthropologist herself and her husband switched to 8. The increased focus on welfare objectives is not cooking with kerosene because of the frequent elec- a driving factor behind deteriorating performance-- tric shocks they and their cook received from the heat- projects with welfare objectives perform no worse on ing plate. average than other projects. 9. A review of best practice approaches to reduc- Chapter 5 ing the costs of grid extension is given in ESMAP (2000). 1. This would not be so if distributional weights were being used. Chapter 3 2. Estimates of a constant elasticity curve passing 1. A more comprehensive picture is given by plot- through PQ combinations for grid electrified and non- ting the whole distribution, as is done in a Lorenz electrified pass below the PQ combinations obtained curve. These curves are given in appendix C. for PQ combinations for off-grid electrified and non- 2. Pan-territorial pricing means that customers in re- electrified. Where both sets of data are available for a mote areas will enjoy these low charges once they are single country, then a kinked demand curve could be connected to the grid, benefiting from cross subsidies estimated, which would show the WTP for grid con- in the service provider's tariff structure. For this reason, nections to be higher than that for off grid. schemes to market off-grid sources are usually unsuc- 3. More were examined, but it is not always possi- cessful if a community believes it will be connected to ble to determine the method being employed. the grid in the near future. 4. To increase understanding of the methodology, 3. Watt peak is the wattage of the solar panel at max- ESMAP produced "A Primer on Consumer Surplus and imum (peak) performance. Demand: Common Questions and Answers" (Peskin 4. In appendix C extensive growth is defined as 2006). growth achieved by extending electricity to previously 5. The figure of 1.6 million is commonly cited (for unconnected communities and when intensive ex- example, Ahmed and others 2005), but the higher fig- pansion reaches households in already connected ure of 2 million is given in the survey of Larson and households. The analysis shows that most of the in- Rosen (2000). crease in coverage comes from extensive growth-- 6. Rather than RE, the strides that have been made only once a majority of communities are connected does in this area are through the adoption of improved intensive growth play a more important role. wood stoves (which also have higher efficiency) or the 5. Bangladesh, Nepal, Peru, and the Philippines. spread of liquefied petroleum gas. 148 ENDNOTES 7. Immunization is instrumented as it is endogenous Appendix F with respect to child nutrition. 1. Income is the variable of interest, but because it 8. However, other studies suggest there may be is endogenous, it is instrumented using education of time savings in cooking (ESMAP 2004). head of the household and construction material of 9. For an elaboration of this argument in the con- walls, roof, and floor of the house. text of a project in South Africa to electrify rural schools, 2. The ESMAP study (2003) also found that house- see Bedford (1997b). holds spend more time running sari-sari shops (retail 10. This is not the approach used in the ESMAP shops) than other home-based businesses. Further- study, which simply attributes the current value of in- more, female household heads were found to spend cremental higher earnings from higher educational at- more hours engaged in home business activity than tainment to the household. That procedure is incorrect, males, and older adults spend less time than younger because it is not adults currently in the labor market who adults. Compared with household heads who are un- earn more as a result of electrification but the children employed or working part time, fully employed house- who will go on to earn more in the future. The IEG cal- hold heads were found to spend about two hours culation is based on these future earnings. Specifically, more per day running their home businesses. The the incremental earnings are assumed to occur from ESMAP study also found a direct relationship between ages 16 to 60, which are discounted back to their pre- the hours spent working in a home business and the sent value. This PV is then converted to a monthly "an- amount of household income from other sources. nuity." The IEG analysis for the Philippines also adjusts for the overestimate of the educational impact of elec- Appendix G trification resulting from the single difference approach, 1. The agency index is the simple sum of indicators assuming it to be one year rather than two. of whether the woman has final say on her own and her 11. Time should be valued at its opportunity cost, children's health care; household purchases for large which clearly depends on whose time it is. This study items and daily needs; what should be cooked each day; takes average income per capita as the opportunity cost and visits to family, friends, or relatives. (rather than the wage) to allow for the distribution of 2. The k knowledge variable for men was a simple household tasks across all household members. Clearly average of three to four separate questions from the a more age- and sex-specific analysis would be prefer- survey: (1) the contraceptive knowledge variable, able, but the data do not allow that. (2) knowledge of timing of ovulation in three of the 12. The figures reported here for the Philippines are four cases and the pregnancy problem in one, (3) lower than those in the ESMAP report. The latter report recognition of diarrhea problems, and/or (4) cough found no significant impact on business revenue from problems. multivariate analysis (although there was for hours worked) and so used single difference estimates. These Appendix H are certainly an overestimate, partly because of the su- 1. This would not be so if distributional weights were perior location of households in electrified villages. An being used. imperfect control group is provided by businesses in electrified villages that do not use electricity in their busi- ness (which may still overestimate the electrification ef- Appendix I fect because of other differences in characteristics). It 1. The previous evaluations have been IEG 2004, is not clear how the ESMAP study allowed for the pro- 2005, and forthcoming. portion of households having a home business. 2. The study was undertaken under the auspices of 13. The difference in lumen capacity (the lumen ESMAP, a technical assistance program of the World power of all lightbulbs owned by the household) is Bank and UNDP, with the secretariat based in the Bank's rather less, as SHS households are far more likely to use Washington, DC, headquarters. energy-efficiency compact fluorescent lightbulbs. How- 3. The problems in this approach, and the alterna- ever, the data collected do not allow an accurate cal- tive "cost-savings" approach, are detailed in the ESMAP culation of this figure (or the more relevant figure of study on the Philippines (ESMAP 2003). lumen hours). 149 THE WELFARE IMPACT OF RURAL ELECTRIFICATION 4. The choice of case study countries is still under are a function of electrification status, they will be ex- review. Indonesia may be substituted for the Philippines. cluded from the indices. The latter is included, as it is the subject of the ESMAP 7. GEF has made some estimates of impact at both study, but the Bank's operations in the country have local and global levels that shall be included in the made little direct contribution to RE coverage. review, but GEF has not used intensive field-level data 5. The report relies on bivariate tabulations and so collection. does not control for other determinants of the outcome 8. This was a World Bank research initiative to bet- variables (Massé 2003). The data will allow multivariate ter understand the links between energy and poverty, analysis, which will do so. for which case studies were undertaken in the three 6. These asset indices usually contain items such as countries mentioned. ownership of a radio and TV. Because these variables 150 BIBLIOGRAPHY Ahmed, Kulsham, Yewande Awe, Douglas F. Barnes, Besant-Jones, John. 2006. "Reforming Power Markets Maureen L. Cropper, and Masami Kojima. 2005. En- in Developing Countries: What Have We Learned?" vironmental Health and Traditional Fuel Use in Energy and Mining Sector Board Discussion Paper Guatemala. Washington, DC: World Bank. No. 19, World Bank, Washington, DC. Bardasi, Elena, and Quentin Wodon. 2006. "Measuring ------. 1995. "Attracting Finance for Hydroelectric Time Poverty and Analyzing Its Determinants: Con- Power." Energy Note 3, World Bank, Washington, DC. cepts and Applications in Guinea." Economics Bul- Bhattacharyya, S.C. 2006a. "Energy Access Problems of letin 10 (12): 1­7. the Poor in India: Is Rural Electrification a Rem- Barkat, Abul, S.H. Khan, M. Rahman, S. Zaman, A. Pod- edy?" Energy Policy 34 (18): 3387­3397. der, S. Halim, N.N. Ratna, M. Majid, A.K.M. Maksud, ------. 2006b. "Universal Electrification: Will the New A. Karim, and S. Islam. 2002. Economic and Social Electrification Programme Succeed in India?" OPEC Impact Evaluation Study of the Rural Electrifica- Review 30 (2): 105­123. tion Program in Bangladesh. Arlington, VA: Human ------. 1997. "Electrification of Clinics and Schools in Development Research Centre and NRECA Inter- the Former Transkei: Pre-Electrification Study in national, Ltd. Nmceba, Ludeke, AmaNdengane and Ama- Barnes, Douglas, ed. 2007. The Challenge of Rural Ntshangase." Energy & Development Research Cen- Electrification: Strategies for Developing Coun- tre Report, University of Cape Town, South Africa. tries. Washington, DC: Resources for the Future. Binswanger, H., and Shahidur Khandker. 1993. "How ------. 1988. "Electric Power for Rural Growth: How Infrastructure and Financial Institutions Affect Agri- Electricity Affects Rural Life in Developing Countries." cultural Output and Investment in India." Journal Rural Sociology 54 (1): 132­135. of Development Economics 41: 337­366. Barnes, D. F., and H. Binswanger. 1986. "Impact of Blackden, C. Mark, and Quentin Wodon, eds. 2006. Rural Electrification and Infrastructure on Agricul- Gender, Time Use, and Poverty in Sub-Saharan tural Changes, 1966­1980." Economic and Political Africa. Washington, DC: World Bank. Weekly (India) 21: 26­34. Brodman, J. 1982. "Rural Electrification and the Com- Bedford, Laura. 1997a. "An Assessment of the Process mercial Sector in Indonesia." Discussion Paper of the Electrification of Clinics in Region E of the East- D-73L, Resources for the Future, Washington, DC. ern Cape." Energy & Development Research Cen- tre Report, University of Cape Town, South Africa. Cecelski, Elizabeth. 2004. "Re-Thinking Gender and ------. 1997b. "An Assessment of the Process of the Energy: Old and New Directions." ENERGIA/EASE Electrification of Schools in Region E of the Eastern Discussion Paper, Leusden, the Netherlands. Cape." Energy & Development Research Centre Re- ------.1992. "Energy and Rural Women's Work: Crisis, port, University of Cape Town, South Africa. Response, and Policy Alternatives." International Berrie, Tom. 1993a. "Energy Efficiency and Conservation Labor Review 126 (1): 41­64. in the Developing World: The World Bank's Role." Charmes, Jacques. 2006. "A Review of Empirical Evidence World Bank Policy Paper, World Bank, Washington, DC. on Time Use in Africa from UN-Sponsored Surveys." ------. 1993b. "The World Bank's Role in the Electric In Time Use and Poverty in Sub-Saharan Africa, ed. Power Sector." World Bank Policy Paper, World Bank, C.M. Blackden and Quentin Wodon, 39­72. Wash- Washington, DC ington, DC: World Bank. 151 THE WELFARE IMPACT OF RURAL ELECTRIFICATION Chaudhury, Nazmul, and Jeff Hammer. 2003. "Ghost Series No 16, Stockholm Environment Institute, Doctors: Absenteeism in Bangladeshi Health Facil- Stockholm, Sweden. ities." World Bank Research Working Paper 3065, Foster, Vivien, and Tito Yepes. 2006. "Is Cost Recovery World Bank, Washington, DC. a Feasible Objective for Water and Electricity? The Davis, J.M., and John Saunders. 1978. Rural Electrifi- Latin American Experience." World Bank Policy Re- cation: An Evaluation of Effects on Economic and search Working Paper 3943, Washington, DC. Social Changes in Costa Rica and Colombia. Boul- Gordon, Adele. 1997. "Facilitating Education in Rural der, CO: Westview Press. Areas of South Africa: The Role of Electricity and Davis, Mark. 1997a. "A Financial and Economic Analy- Other Sources of Energy." Energy & Development sis of Two Electrification Projects." Energy & De- Research Centre Report, University of Cape Town, velopment Research Centre Report, University of South Africa. Cape Town, South Africa. Greacan, Chris. 2003. "Small Is Pitiful: Micro- ------. 1997b. "The Financial Impacts of Rural Elec- Hydroelectricity and the Politics of Rural Electricity trification: Draft Report." Energy & Development Re- Provision in Thailand." Presented at "Politics of the search Centre Report, University of Cape Town, Commons: Articulating Development and Strength- South Africa. ening Local Practices," Chiang Mai, Thailand, July ------. 1995. "Electricity Consumption Growth in 11­14. Newly Electrified Settlements." Energy & Develop- Green, Donna. 2005. "The Politics of Environmental ment Research Centre Report, University of Cape Technology Choice for Rural Electrification in Thai- Town, South Africa. land." Conservation and Society 3 (2): 265­279. Dutta, S. 2005. "Energy as a Key Variable in Eradicating GSS (Ghana Statistical Service). 2000. Ghana Living Extreme Poverty and Hunger: A Gender Perspective Standards Survey: Report of the Fourth Round. on Empirical Evidence on MDG#1." DFID Discus- Accra: Ghana Statistical Service. sion Paper, U.K. Department for International De- Hutton, G., E. Fehfuess, F. Tediosi, and S. Weiss. 2006. velopment, London. Evaluation of the Costs and Benefits of Household ESMAP (Energy Sector Management Assistance Pro- Energy and Health Interventions at Global and Re- gram). 2004. "Power Sector Reform and the Rural gional Levels. Geneva: World Health Organization. Poor in Central America." ESMAP Report 297/04, IEG (Independent Evaluation Group). Forthcoming. An World Bank, Washington, DC. Impact Evaluation of India's Second and Third ------. 2003. "Rural Electrification and Development Andhra Pradesh Irrigation Projects: A Case of in the Philippines: Measuring the Social and Eco- Poverty Reduction with Low Economic Returns. nomic Benefits." ESMAP Report 255/03, World Bank, IEG Study Series. Washington, DC: World Bank. Washington, DC. ------. 2006. New Renewable Energy: A Review of ------. 2002. "Energy Strategies for Rural India: Evi- dence from Six States." ESMAP Report 258/02, World the World Bank's Assistance. IEG Study Series. Bank, Washington, DC. Washington, DC: World Bank. ------. 2000. "Reducing the Cost of Grid Extension for ------. 2005. Maintaining Momentum to 2015: An Im- Rural Electrification." ESMAP Report 227/00, World pact Evaluation of External Support to Maternal Bank, Washington, DC. and Child Health and Nutrition in Bangladesh. IEG Fluitman, F. 1983. "The Socio-Economic Impact of Study Series. Washington, DC: World Bank. Rural Electrification in Developing Countries: A Re- ------. 2004. Books, Buildings, and Learning Out- view of Evidence." World Employment Programme comes: An Impact Evaluation of World Bank Sup- Research Working Paper 126, International Labour port to Basic Education in Ghana. IEG Study Series. Office, Geneva. Washington, DC: World Bank. Foley, Gerald. 1993. "Rural Electrification in Mozam- ------. 2003. Power for Development: A Review of bique, Tanzania, Zambia, and Zimbabwe: Synthesis the World Bank Group's Experience with Private Report from the SEI/BUN Workshop on Rural Elec- Participation in the Electricity Sector. IEG Study Se- trification." Energy, Environment and Development ries. Washington, DC: World Bank. 152 BIBLIOGRAPHY ------. 1994. Rural Electrification in Asia: A Review Luckin, Bill. 1990. Questions of Power: Electricity and of Bank Experience. IEG Study Series. Washing- Environment in Inter-War Britain. Manchester, ton, DC: World Bank. United Kingdom: Manchester University Press. Jadresic, A. 2000. "A Case Study on Subsidizing Rural Mariam, H. 1992. "Rural Electrification in Ethiopia." In Electrification in Chile." In Energy Services for the Rural Electrification in Africa, ed. V. Ranganathan. World's Poor. Washington, DC: World Bank. London: Zed Books, Ltd. James, Bronwyn. 1997. "Current Limited Supplies of Massé, Réné. 2003. "Impacts of Rural Electrification on Electricity in the Context of South African Rural Poverty and Gender in Sri Lanka." Country Study, Areas." Energy & Development Research Centre World Bank, Washington, DC. Report, University of Cape Town, South Africa. Mbewe, A. 1992. "Rural Electrification in Zambia." In ------. 1996a. "An Assessment of the Pre-Electrification Rural Electrification in Africa, ed. V. Ranganathan. Process in the Tambo Village Pilot Project." Energy London: Zed Books, Ltd. & Development Research Centre Report, University Meier, T. 2001. Mini Hydropower for Rural Develop- of Cape Town, South Africa. ment: A New Market-Oriented Approach to Maxi- ------. 1996b. "Criteria Used for the Selection of Elec- mize Electrification Benefits--With Special Focus trification Projects in Rural Areas." Energy & De- on Indonesia. Berlin: Lit-Verlag. velopment Research Centre Report, University of Mensah-Kutin, Rose. n.d. "Gender and Rural Electrifi- Cape Town, South Africa. cation in Ghana." Unpublished doctoral dissertation, ------. 1995. "The Impacts of Rural Electrification: Centre for West African Studies, University of Birm- Exploring the Silences." Energy & Development ingham, United Kingdom. Research Centre Report, University of Cape Town, Nathan, Dev, and Govind Kelkar. 1997. "Wood Energy: South Africa. The Role of Women's Unvalued Labor." Gender, James, Bronwyn, and Pamela Ntutela. 1997. "Rural Technology, and Development 1 (2): 205­224. Households' Response to the 2.5A Electricity Sup- National Rural Electric Cooperation Association. 1982. ply Option in the Tambo Village Pilot Project." En- "Productive Uses of Electricity in Rural Areas." Paper ergy & Development Research Centre Report, prepared for the workshop, Dhaka, Bangladesh, University of Cape Town, South Africa. Nov. 15­19. Kjellström, Bjorn, Maneno Katyega, Henry Kadete, Dolf Parikh, J. 2005. The Energy, Poverty, and Gender Nexus Noppen, and Abu Mvungi. 1992. "Rural Electrifi- in Himachal Pradesh, India: The Impact of Clean cation in Tanzania: Past Experience--New Ap- Fuel Access Policy on Women's Empowerment. proaches." Energy, Environment and Development New Delhi: Integrated Research and Action for Series No.15, Stockholm Environment Institute, Development. Stockholm, Sweden. Parikh, J., and Laxmi, V. 2000. "Biofuels, Pollution and Kline, Ronald. 2002. "Resisting Development, Rein- Health Linkages--A Survey of Rural Tamil Nadu." venting Modernity: Rural Electrification in the United States before World War II." Environmental Values Economic and Political Weekly, November 18. 11: 327­344. Peskin, Henry M. 2006. "A Primer on Consumer Surplus Komives, K., Vivien Foster, Jonathan Halpern, and and Demand: Common Questions and Answers." Quentin Wodon. 2005. Water, Electricity and the ESMAP Knowledge Exchange Series No. 5, World Poor: Who Benefits from Utility Subsidies? Wash- Bank, Washington, DC. ington, DC: World Bank. Piampiti, Suwanlee, Twatchai Yongkittikul, Laddwal Larson, Bruce A., and Sydney Rosen. 2000. Household Rodmanee, and Vanawipha Pasandhanatovn. 1982. Benefits of Indoor Air Pollution Control in Devel- "Demographic and Economic Impact of Rural Elec- oping Countries. Washington, DC: USAID. trification in Northeastern Thailand." National In- Laxmi, Vijau, Jyoti Parikh, Shyam Karmakar, and Pramod stitute of Development Administration Report, Dabrase. 2003. "Household Energy, Women's Hard- Bangkok, Thailand. ship, and Health Impacts in Rural Rahastan, India: Ramasedi, B. R. 1992. "Rural Electrification in Botswana." Need for Sustainable Energy Solutions." Energy for In Rural Electrification in Africa, ed. V. Ran- Sustainable Development 7 (1): 50­68. ganathan, 112­140. London: Zed Books, Ltd. 153 THE WELFARE IMPACT OF RURAL ELECTRIFICATION Rogerson, Christian Myles. 1997. "Rural Electrification Winther, Tanja. 2005. "Current Styles: Introducing Elec- and the SMME Economy in South Africa." Energy & tricity in a Zanzibari Village." Unpublished doctoral Development Research Centre Report, University of dissertation, Department of Social Anthropology, Cape Town, South Africa. University of Oslo. Ross, Fiona, Richard Matzopoulos, and Rozett Phillips World Bank. 2006. "Peru--Rural Electrification Proj- 1997. "The Role of Rural Electrification in Promot- ect." Project Appraisal Document Report No. 32686- ing Health in South Africa." Energy & Development PE, World Bank, Washington, DC. Research Centre Report, University of Cape Town, ------. 2005. "Energy and Poverty: Myths, Links, and Policy Issues." Energy and Mining Sector Board South Africa. Working Notes, World Bank, Washington, DC. Saunier, G., ed. 1992. Rural Electrification Guidebook ------. 2002. Renewable Energy Toolkit. http:// web. for Asia and the Pacific. Bangkok: Asian Institute of worldbank.org/WBSITE/EXTERNAL/TOPICS/ Technology. EXTENERGY/EXTRETOOLKIT/0,,contentMDK: Shonder, J.A., and T.J. Wilbanks. 1996. Final Evaluation 20768499~menuPK:2069822~pagePK:64168427 of the Central American Rural Electrification Sup- ~piPK:64168435~theSitePK:1040428,00.html. port Project (CARES). Washington, DC: USAID. ------. 2001a. "Brazil--Rural Poverty Alleviation Proj- Siddiqui, Kamal. 2000. Jagatpur 1977­1997: Poverty and ect: State of Ceara." Implementation Completion Social Change in Bangladesh. Dhaka, Bangladesh: and Results Report No. 23397, World Bank, Wash- University Press Ltd. ington, DC. Singh, D. 1997. "Renewable Energy for Village Electri- ------. 2001b. "The World Bank's Energy Program: fication." International Centre for Application of Poverty Reduction, Sustainability, and Selectivity." Solar Energy Report, New Delhi, India. Energy and Mining Sector Board Paper, World Bank, Thom, Cecile. 1996. "Pre-Electrification Research in Washington, DC. Mafefe: Energy Use, Livelihoods, Power Relations and ------. 2000. Fuel for Thought: An Environmental the Electrification Process." Energy & Development Strategy for the Energy Sector. Washington, DC: Research Centre Report, University of Cape Town, World Bank. South Africa. ------. 1996. Rural Energy and Development: Im- Thom, Cecile, M. Davis, and M. Borchers. 1995. "Review proving Energy Supplies for Two Billion People. of South African Experience in Rural Electrification." Washington, DC: World Bank. Energy & Development Research Centre Report, ------. 1995. "Indonesia--Second Rural Electrifica- University of Cape Town, South Africa. tion Project." Staff Appraisal Report No. 12920-IND, Tomkins, R. 2001. "Extending Rural Electrification: A World Bank, Washington, DC. Survey of Innovative Schemes." In Contracting ------. 1993a. "Energy Efficiency and Conservation in for Public Services: Output-Based Aid and Its Ap- the Developing World." Policy Paper, World Bank, Washington, DC. plications, ed. P. J. Brook and S. M. Smith, 47­56. ------. 1993b. "The World Bank's Role in the Electric Washington, DC: World Bank. Power Sector: Policies for Effective Institutional, Valencia, Alfonso Carrasco, Kimmo J. Juvas, and Maija Regulatory, and Financial Reform." Working Paper, Seppo. 1990. Electrification and Rural Develop- World Bank, Washington, DC. ment--Electrification Project in the Rural Areas of ------. 1992. World Development Report: Develop- Cusco, Peru. Helsinki: International Development ment and the Environment. Washington, DC: World Agency, FINNIDA. Bank. Walubengo, D., and Onyango, A. 1992. Energy Systems ------. 1986. "Indonesia Rural Electrification Review." in Kenya: Focus on Rural Electrification--Kengo East Asia and Pacific Regional Office Working Paper, Regional Wood Energy Programme for Africa. World Bank, Washington, DC. Nairobi: KENGO. ------. 1975. "Rural Electrification." Working Paper Wentzel, Marlett, and C. Wood. 1997. "A Post-Electrifi- No. PUB-517, World Bank, Washington, DC. cation Study of the Mafefe Electrification Project." World Commission on Dams. 2001. http://siteresources Energy & Development Research Centre Report, .worldbank.org/INTWRD/903857-1112344791813/ University of Cape Town, South Africa. 20424164/WB&WCDQ&A.pdf. 154 THE WORLD BANK GROUP WORKING FOR A WORLD FREE OF POVERTY The World Bank Group consists of five institutions--the International Bank for Reconstruction and Development (IBRD), the International Finance Corporation (IFC), the International Development Association (IDA), the Multilateral Investment Guarantee Agency (MIGA), and the International Centre for the Settlement of Investment Disputes (ICSID). Its mission is to fight poverty for lasting results and to help people help themselves and their environment by providing resources, sharing knowledge, building capacity, and forging partnerships in the public and private sectors. THE INDEPENDENT EVALUATION GROUP ENHANCING DEVELOPMENT EFFECTIVENESS THROUGH EXCELLENCE AND INDEPENDENCE IN EVALUATION The Independent Evaluation Group (IEG) is an independent, three-part unit within the World Bank Group. IEG-World Bank is charged with evaluating the activities of the IBRD (the World Bank) and IDA, IEG-IFC focuses on assessment of IFC's work toward private sector development, and IEG-MIGA evaluates the contributions of MIGA guarantee projects and services. IEG reports directly to the Bank's Board of Directors through the Director-General, Evaluation. The goals of evaluation are to learn from experience, to provide an objective basis for assessing the results of the Bank Group's work, and to provide accountability in the achievement of its objectives. It also improves Bank Group work by identifying and disseminating the lessons learned from experience and by framing recommendations drawn from evaluation findings. ISBN 978-0-8213-7367-5 SKU 17367 THE WORLD BANK