Energy Policies and MultitopicHousehold SurveysGuidelines for Questionnaire Design in Living

Standards Measurement Studies

Kyran O’Sullivan and Douglas F. Barnes

E N E R G Y A N D M I N I N G S E C T O R B O A R D D I S C U S S I O N P A P E R

P A P E R N O . 1 7

A P R I L 2 0 0 6

Energy and MiningSector Board

THE WORLD BANKGROUP

AUTHORS

Doug Barnes (dbarnes@worldbank.org) and Kyran O’Sullivan(kosullivan@worldbank.org) are Senior Energy Specialists atthe World Bank.

DISCLAIMERS

The findings, interpretations, and conclusions expressed inthis paper are entirely those of the authors and should not beattributed in any manner to the World Bank, to its affiliatedorganizations, or to members of its Board of ExecutiveDirectors or the countries they represent.

CONTACT INFORMATION

To order additional copies of this discussion paper pleasecontact the Energy Help Desk: +1.202.473.0652energyhelpdesk@worldbank.org

This paper is available online www.worldbank.org/energy/

The material in this work is copyrighted. No part of this work may be reproducedor transmitted in any form or by any means, electronic or mechanical, includingphotocopying, recording, or inclusion in any information storage and retrieval system,without the prior written permission of the World Bank. The World Bank encouragesdissemination of its work and will normally grant permission promptly. For permis-sion to photocopy or reprint, please send a request with complete information tothe Copyright Clearance Center, Inc, 222 Rosewood Drive, Danvers, MA 01923,USA, fax 978-750-4470. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, World Bank,1818 H Street N.W., Washington DC, 20433, fax 202-522-2422, e-mail: pubrights@worldbank.org

The World Bank, Washington, DC

E N E R G Y A N D M I N I N G S E C T O R B O A R D D I S C U S S I O N P A P E R

P A P E R N O . 1 7

A P R I L 2 0 0 6

Energy Policies and MultitopicHousehold SurveysGuidelines for Questionnaire Design in Living

Standards Measurement Studies

Kyran O’Sullivan and Douglas F. Barnes

Energy and MiningSector Board

THE WORLD BANKGROUP

Copyright © 2006 The International Bank for Reconstruction and Development/The World Bank. All rights reserved

CONTENTS

ACKNOWLEDGMENTS ..............................................................................................iiFOREWORD ................................................................................................................11 . INTRODUCTION ..................................................................................................3

Potential Value of LSMS Surveys for Analysis of Household Energy Use ............................3Report Audience and Organization ..............................................................................4

2 . CHARACTERISTICS OF LSMS SURVEYS AND SPECIALIZED HOUSEHOLD ENERGY SURVEYS..................................5Characteristics of LSMS Surveys ..................................................................................5Characteristics of Specialized Household Energy Surveys................................................6Other Distinguishing Features of Both Types of Surveys ..................................................7Existing and Missing Energy Questions in LSMS Surveys ................................................7Energy and Development Outcomes: The Issue of Causality ..........................................9

3 . IMPORTANCE OF LSMS SURVEY DATA FOR ENERGYPOLICY ANALYSIS ............................................................................................11Household Energy Transitions ....................................................................................11Using LSMS Surveys for Energy Policy Analysis ............................................................13Rural Electrification and Development ........................................................................16

4 . THE WAY FORWARD ........................................................................................19Fuel Sources Module ................................................................................................20Electricity Sources Modules (Grid and Off-grid) ..........................................................24Durable Goods (Light Bulbs and Appliances) Module ..................................................29Electricity and Fuels in the Community Module ............................................................31

5 . CONCLUSION ....................................................................................................346 . REFERENCES ......................................................................................................357 . APPENDIX ..........................................................................................................38

i i

ACKNOWLEDGMENTS

These guidelines were prepared by Kyran O’Sullivanand Douglas F. Barnes. Dale Whittington reviewedenergy questions in LSMS surveys and prepared abackground report. The financial support of ESMAP isgratefully acknowledged. The authors wish to thankKathleen Beegle, Aline Coudouel, Olivier Dupriez,Dominique Lallement, Lucio Monari, StefanoPaternostro, Peter Roberts, Diane Steele, Caroline vanden Berg and Robert van der Plas for their guidance andencouragement. We are grateful to Vivien Foster andKinnon Scott, who reviewed earlier drafts of the paperand made many useful suggestions. Norma Adamsedited the paper.

1

FOREWORD

Accurate data on household energy use, combined with other data on household well-being (includingconsumption, income, health, and education), is essentialto monitor progress in the household energy transitionfrom traditional biomass fuels to modern fuels andelectricity and to evaluate the effect of government energypolicies on living conditions. Multitopic socioeconomichousehold surveys, such as the World Bank’s LivingStandards Measurement Study (LSMS), can provide datawith which to make these measurements. Designers ofLSMS and other multitopic household surveys can usethese guidelines to help ensure that their surveys providemore extensive and reliable data on household energyuse than they do at present. The guidelines highlightweaknesses in current LSMS surveys with respect toenergy questions and discuss how such questions canbe better formulated to yield more useful data for energypolicy analysis. Household energy surveys implementedover the years offer lessons on which formulations ofquestions work best and provide the most consistentresults. This experience has been drawn on to developthe prototype fuel and electricity modules contained inthese guidelines. Indicators that may be constructedfrom the data are also discussed; in this regard, thepresent report contributes to international efforts to defineenergy indicators for sustainable development. Theseefforts include the publication Energy Indicators forSustainable Development: Guidelines and Methodologies(IAEA 2005), which drew attention to using householdsurveys to provide data for constructing indicators.

It is anticipated that these guidelines will help LSMSdesigners incorporate energy modules of the typeproposed herein into LSMS survey questionnaires. Over time, as more surveys containing these modulesare implemented, more experience will be gained onwhich questions work best in particular country settingsand which are most useful for policy analysis.

Jamal Saghir Director, Energy and WaterChairman, Energy and Mining Sector Board

2

3

1. INTRODUCTION

Adequate and affordable supplies of electricity andmodern fuels must be available to households if they are to have a good living standard. Safe, reliable, andgood-quality energy services1—lighting, heating, cooking,and motive and mechanical power—are only availableto households when they use electricity and modernfuels. In developing countries, however, some 2 billionpeople cook and heat their homes with wood or othertraditional biomass; 1.6 billion people do not haveelectricity in their homes (World Bank 1996). For thesepeople, obtaining energy is costly and time-consuming;and they may be denied the opportunity to earn income,read, and access information and entertainment.

Recognizing the role that energy services play in povertyreduction, governments everywhere wish to implementpolicies and make investments that will accelerate the transition from use of traditional fuels to modernfuels and electricity. As a consequence, policymakersincreasingly seek empirical evidence of the relationshipbetween investments made in energy infrastructure, the energy policies they implement, and welfareimprovements at the household level.

Analysis using data on household energy use from LivingStandards Measurement Study (LSMS) surveys can helppolicymakers to identify which households are affectedby energy poverty and then design policies that willaccelerate the household energy transition. For householdsthat have already made the transition to using modernfuels and electricity, government policies in the sectorwill affect how much households pay and consume. For example, regulatory policy for service providers maycause energy supply to improve or deteriorate, and energyprice and tax adjustments will affect households’ energyexpenditure and consumption. Data from LSMS surveyscan be used before a policy is implemented to constructhypotheses of the welfare changes that may occur. If data is available from two successive LSMS surveys(the first carried out before the policy intervention andthe second after), the change in household welfare as a result of the intervention may be observed.

Energy services also affect household welfare throughtheir role in the education and health status of householdmembers. Literacy is much higher in households withelectricity than in those without, and children in householdsthat use modern cooking fuels are less likely to sufferacute respiratory disease than children in households thatuse biomass. Other effects on household welfare includetime saving and increased productivity and income.

Potential Value of LSMS Surveys for Analysis ofHousehold Energy Use

Unfortunately, data sets on which to base evaluation ofenergy services’ role in contributing to these householdwelfare outcomes are in remarkably short supply. LSMSsurveys offer one possible source of better informationgathering on providing energy services to households indeveloping countries. If LSMS surveys, which alreadyprovide high-quality comprehensive data on many aspectsof household welfare, could also provide better data on household energy use, the role of energy services in household welfare could be better understood. LSMSsurveys with fuels and electricity modules containingquestions on the fuels and electricity sources availableto households, quantities consumed, and householdcost and expenditures could be combined with otherhousehold socioeconomic characteristics and theirrelationships analyzed.

At present, household energy data collected throughLSMS surveys is insufficient for extensive energy policyanalysis. At the same time, specialized household energysurveys often are not designed for such analysis; theyare seldom national in scope and do not cover thebreadth of topics involving other sectors, which is thehallmark of LSMS surveys. Incorporating the proposedenergy modules found in this report into future LSMSsurveys will provide greater insight into the role energyservices play in household welfare and thus will contributeto energy infrastructure planning and energy policyanalysis so that they achieve the highest possible socialand development impact.

1 Primary energy sources (crude oil, natural gas, coal, modern biomass and renewable energy) are transformed into fuels and electricity to providehouseholds energy services. Modern biomass includes sustainable and safe use of biomass (e.g., biogas and wood used in improved cookingstoves).

4

Report Audience and Organization

These guidelines advise on how an LSMS or othermultitopic household survey can be modified to improveits usefulness to the energy sector. Thus, the guidelinestarget two main audiences. The first are designers ofLSMS and other multitopic household surveys who maynot be familiar with the data needs of policy analystsworking in the household energy sector. The second areenergy policy analysts who wish to ensure that a plannedmultitopic household survey includes an adequate set ofquestions essential to their work.

This report begins by examining the characteristics and relative merits of LSMS and specialized householdenergy surveys. It then reviews the types of questionscurrently used in LSMS surveys, identifying importantdata gaps. Next, the issue of causality as it relates toenergy services and development is examined. Thereport then turns to examples of the types of energyanalysis that benefit from enhanced LSMS survey data.Prototype energy modules (accompanied by notes oneach question contained in them) are presented. It isproposed that the few ad-hoc questions on energy usecurrently found in LSMS surveys—which yield littleinformation useful to policymakers—be replaced by thequestions presented in the prototype modules. Mindfulthat LSMS surveys are already lengthy, expensive, andtime-consuming and that increased demands on theLSMS should be carefully considered, the additionalquestions in the proposed modules have been kept to a minimum while still being sufficient for the types ofpolicy analysis that are most frequently conducted.

5

2. CHARACTERISTICS OF LSMS SURVEYS ANDSPECIALIZED HOUSEHOLD ENERGY SURVEYS

Since inception of the LSMS program in 1980 and pilotingof the first surveys in Côte d’Ivoire and Peru in 1985, theWorld Bank LSMS surveys have been used for povertyanalysis that underpins poverty-reduction strategies inmore than 50 countries. Integrated household surveys,similar in design to the LSMS, have also been developedand implemented in many countries.

LSMS surveys aim to provide a comprehensive picture of living standards. They are one of the most extensivelyused tools for measuring and monitoring householdmonetary and non-monetary welfare (including income;expenditure; employment; migration; housing conditions;school enrollment; educational attainment; health status;nutrition; and use of transport, water, sanitation andenergy services) (Deaton 1997). Thus, governmentpolicymakers can use the LSMS survey data to determineliving conditions of the poor and other disadvantagedgroups. The data can be used to model economicbehavior in order to design better policies or choosebetween alternative public investments. Such analysisenables one to assess the efficiency and efficacy ofalternative policy designs and investments. For example,analysis of anthropometric data can be used to assessprograms to reduce malnutrition or analyze health statusused to evaluate immunization or HIV prevention.

Specialized household energy surveys, by contrast, are implemented to inform a particular energy policy or investment. For example, they are used to assess the efficiency and efficacy of fuel-price subsidies or toestablish baseline information for and monitor ruralelectrification programs.

The distinguishing characteristics of LSMS and specializedhousehold energy surveys are important in explainingthe rationale for advocating a greater depth of energycoverage in LSMS and other multitopic household surveys.To this end, it is also helpful to examine the relativestrengths and weaknesses of such surveys from theperspective of energy policy analysts.

Characteristics of LSMS Surveys

LSMS surveys are both multitopic and multi-level (Groshand Glewwe 2000; Grosh and Munoz 1996). Theycommonly use up to four survey instruments with whichto examine a broad range of household welfare andbehavioral factors, as follows:

Household questionnaire. This questionnaire collectsdetailed information on household members. It iscomposed of modules that focus on specific topics. The questionnaire always collects detailed informationto measure household consumption, which is the bestmonetary indicator of household welfare. The householdquestionnaire typically includes modules on education,health, employment, migration, anthropometry, fertility,housing, agriculture, household enterprises, income,and savings and credit. Certain information (includingemployment, health, and education) is collected at theindividual level. Frequently, the housing module coversinformation on fuels used for cooking and heating andelectricity services available to the household.

Community characteristics questionnaire. Thisquestionnaire gathers information on local conditionscommon to all households living in the same community.Like the household questionnaire, the communityquestionnaire contains modules on specific topics,including schools, health facilities, agricultural practices,and infrastructure (e.g., roads, fuel sources, electricity,and water). The questionnaire asks key community leadersand groups about available services, economic activities,access to markets, and (more recently) social capital.

Price questionnaire. This questionnaire aims to gatherinformation from market vendors on prevailing prices ofcommonly purchased items in shops and markets.

Facilities questionnaire. This questionnaire isadministered to local service providers, such as schoolsand health clinics.

6

Characteristics of Specialized HouseholdEnergy Surveys

Specialized household energy surveys typically containthe following sections:

Core information. Whatever the purpose of a specializedhousehold energy survey, information usually consideredessential includes the fuel and electricity sources availableto households, supply (e.g., pricing, quality, and reliability)and demand (e.g., the quantities consumed) characteristicsof these sources, the energy services households derivefrom the fuels and electricity sources they consume, andhousehold expenditures on these fuels and electricity.

Initial cost of access. Since the cost of obtaining service(e.g., initial purchase of a liquefied petroleum gas [LPG]cylinder or electricity connection fee) is recognized as asignificant barrier to access, specialized household energysurveys frequently collect this information, along with dataon sources of financing that households use to defraythese costs.

Alternatives to grid electricity. In developing countries,many urban areas often face unreliable grid-electricitysupplies (i.e., they are available infrequently or subjectto interruptions); in rural areas, electricity infrastructuremay be lacking entirely. For these reasons, specializedhousehold energy surveys often focus on alternatives togrid electricity. The surveys commonly investigate the useof diesel gensets, solar home photovoltaic (PV) systems,and car (storage) batteries. The questionnaire usuallycovers access to these energy sources and expendituresfor acquiring and operating them.

Time use. In areas where biomass is used extensivelyfor cooking or heating, specialized household energysurveys often contain large sections that investigate thetime household members spend collecting biomass andtending cooking fires.

Energy end use. Specialized household energy surveyscan investigate in detail the end uses of the energysources consumed in order to quantify the benefits thathouseholds obtain from them. For example, to measureelectric lighting’s effects on rural households, one mustknow the number of lighting appliances customers used

before and after electrification. For households withoutelectricity, these appliances include non-electricequipment (e.g., candles, simple kerosene wick lamps,regulated wick lamps [hurricane lanterns], andpressurized kerosene lamps). For households withelectricity, they include incandescent, fluorescent, andcompact fluorescent lamps. The questionnaire mustcollect data on the length of time each lamp is usedduring a typical day. With this information, researcherscan compare electrified and non-electrified ruralhouseholds with regard to price and quantity of lumenhours used. Households with electric lamps typicallyconsume more than 20 times more lighting thanhouseholds with kerosene lamps; however, the price per unit of light is substantially lower. Analysts can usesuch information to quantify the benefits of electriclighting for rural households, using methods involvingwillingness to pay.

Attitudes. Specialized household energy surveys oftencontain sections that explore household attitudes towardenergy services. For example, households may have theoption to connect to grid electricity service, but forvarious reasons choose not to. Respondents can bequeried about the reasons for their choices. Theiranswers may reveal high up-front connection charges,side payments to utility providers, or inability to wiretheir houses internally for lack of qualified electricians or because of safety issues involving the materials usedto construct their houses.

Socioeconomic information of households.Specialized household energy surveys usually contain alimited set of questions on sources of income andexpenditures for non-energy goods and services.2 Inaddition, specialized household energy surveys usuallycollect cursory information on other householdsocioeconomic characteristics (e.g., housing or level ofeducational attainment); however, they rarely containextensive modules on such topics.

2 LSMS surveys, by contrast, aim at a comprehensive measure of household consumption; thus, they contain more reliable income andexpenditure data, which is essential for distributional analysis.

7

Other Distinguishing Features of Both Types ofSurveys

Questionnaire design. Questionnaires of LSMS surveysand specialized household energy surveys are not staticor uniform. Both types vary considerably by country andare customized to the identified analytical needs thatmake a survey necessary. Typically, a national energyministry or energy utility commissions specialized householdenergy surveys and retains control over their questionnairedesign. This control can ensure that sufficient questionsare included to permit a detailed investigation of energyend use, time use, attitudes, initial cost of obtainingservice, and other information important for energy policy.The design process should allow for adequate consultationwith other sector ministries and research institutions indetermining questionnaire content. In practice, however,such consultation is often ad-hoc.

By contrast, design of LSMS survey questionnaires usuallyinvolves forming a committee chaired by the nationalplanning or statistical agency; the committee typicallyincludes members from sector ministries, civil society,research institutions, universities, and developmentagencies. It oversees questionnaire design through aniterative process. Since LSMS questionnaires can yielddata useful for analysis of a great range of public-policyissues, many more modules and questions are usuallyproposed than can be accommodated since the lengthof the questionnaire must be limited to avoid respondentfatigue. Thus, it is important that energy specialists fullyparticipate in the LSMS survey design process to ensure thatessential, well-formulated energy questions are included.

Sample design. Both types of surveys are based onmulti-stage stratified probability designs, using smallsample sizes of 2,000-5,000 households to minimizenon-sampling errors.3 Their small size usually precludesdisaggregating the results to small geographical areas.Domains of study (e.g., urban/rural) are identified; withineach, a stratified two-stage cluster design is used (clusteringinterviews helps to reduce cost). LSMS surveys are typicallynational in coverage, while specialized household energysurveys are often implemented in a particular region(e.g., where a pilot rural electrification or improvedstove program will be implemented).

Survey cost. LSMS surveys are significantly more expensive(US$0.5-1.5 million) than specialized household energysurveys (US$50,000-150,000). Cost factors includesample and questionnaire size, local per diem, andsalaries. Collecting data on household energy usethrough a well-designed LSMS survey offers anextremely cost-effective opportunity.

Quality control. Extensive quality control is a hallmarkof LSMS surveys. In terms of questionnaire design, qualitycontrol includes use of verbatim questions, explicit skippatterns, and translation of questions into respondents’languages. Survey modules provide the enumeratorexplicit suggestions for further probing. A custom-designed,data-entry program performs automated range andconsistency checks within and across modules. Otherquality-control features include enumerator training(typically four weeks), close field supervision, questionnairepre-testing, piloting, and rigorous documenting of allsurvey stages. While specialized household energysurveys have adopted many of these features, theirsmaller budgets are a limitation that may sometimescompromise quality control.

Because the LSMS surveys are much longer, two fieldvisits at two-week intervals to each household arerequired to complete the household questionnaire.Between visits, data for the first half of the householdquestionnaire are entered and checked for errors.During the second visit, when the second half of thequestionnaire is completed, these errors are corrected.For specialized household energy surveys, householdsare visited only once; thus there is no re-interviewingopportunity to correct for errors.

Existing and Missing Energy Questions in LSMSSurveys

In developing these guidelines, completed LSMS surveyswere reviewed to determine what energy questions theycontained. The review found that most surveys containedsome energy questions. In the household questionnaire,energy questions centered mainly on fuel types used forcooking or lighting, along with types of householdappliances. In the community questionnaire, energyquestions typically focused on local pricing of variouscommodities, including energy.

3 However, for LSMS surveys, larger samples are not uncommon. For example, the 2001 Bosnia and Herzegovina LSMS Survey sampled 5,402households, and the 1994 South Africa Integrated Household Survey covered about 9,000 households.

8

Household questionnaires: What energy questions do they currently contain?

The typical LSMS household survey instrument containsfour types of energy questions that analysts can use tobetter understand household energy-use patterns. Theserelate to:

• electricity service providers,

• type of energy the household uses,

• expenditures on commercial fuels, and

• energy-using durable goods.

Electricity service providers. Nearly all LSMS householdquestionnaires ask households whether they have electricityservice. This question is sometimes followed by anotherto determine the service provider (utility, self-generated,or other), as well as other questions on monthlyhousehold expenditures.

Type of energy the household uses. Questionsregarding the types of energy households use askrespondents about their households’ energy sources for various end uses (e.g., “What is your principal sourceof lighting?” or “What is your primary fuel for cooking?”).In some cases, these questions may be followed byothers regarding the secondary fuels used for suchactivities (Komives, Whittington, and Wu 2002). Sinceany household that has electricity uses it for lighting, thequestion on source of energy used for lighting can beused to determine whether the household has an electricityconnection. In countries where space heating is common,respondents usually are asked such questions as “What isyour primary fuel for space heating?” or “What is yourmain energy source for heating?”

Expenditures on commercial fuels. Questions onexpenditures ask respondents how much money theirhouseholds spent during the last week or month on listednon-food items, which often include a variety of fuels(e.g., candles, batteries, bottled gas, kerosene, charcoal,and fuelwood). The responses reveal whether thehousehold purchased a particular fuel and the monthlyexpenditures on it. Certain fuels are sometimes lumpedtogether in ways that make it difficult to disaggregateexpenditures on individual fuels.

Energy-using durable goods. Questions on durablegoods present respondents a long list from which theyare asked to indicate which goods their householdsown. The list typically includes electrical appliances(e.g., fans and televisions) and cooking equipment (e.g.,stoves and hot plates). To enable a calculation of the flowof services from such durable goods, further questions onthe age and present value of the item are asked.

Although most LSMS household surveys contain all foursets of questions, the level of sophistication varies widely.Some include only two questions with which to determinewhether a household has an electricity connection and,if so, the level of monthly expenditure. Others include10 or more questions to probe whether the householdhas a working electricity meter, whether it shares its billwith other households, household billing frequency, andnumber of hours of electricity service. Some instrumentsinclude small enterprise modules to collect informationon the electricity households use for business or otherproductive purposes.

Some LSMS household questionnaires ask energyquestions with regard to time use (Harvey and Taylor2000). These questions involve how much time householdmembers spend collecting fuelwood or other biomassfuels used for cooking and heating. The questionssometimes detail which household members do this typeof work, where the fuel is collected, and the type ofbiomass stove used for cooking.

Community questionnaires: What energy questions do they currently contain?

In the LSMS community questionnaire, the most commonenergy-related questions involve availability and prices.They ask respondents about the availability and pricingof such fuels as charcoal, fuelwood, kerosene, LPG,candles, coal, and batteries. Some survey instrumentsinclude questions on the percentage of households inthe community with electricity service or ask about thenumber of hours electricity service is usually available. A few ask how many years the community has hadelectricity service and whether electricity utility ownershipis public or private.

9

LSMS surveys: What energy questions are currently missing?

Most LSMS surveys ask only about the main source ofenergy for a particular purpose. Because the surveypermits only one answer, it is not possible to deducewhich minor energy sources households use for lighting,cooking, and other end uses. Community questionnairesseldom include a comprehensive list of all energy sourceswhen they ask which fuels are available in the community;thus, it is not possible to deduce which energy sourcesavailable in the community households choose not touse. To model the determinants of households’ fuel-choice thoroughly, it is important to accurately characterizeall of the available energy options that a householddoes not choose.

In addition, LSMS surveys typically lack sufficientlydetailed questions on:

• pricing in order to accurately determine unit priceshouseholds pay for fuels and electricity (communityquestionnaires do not always include questions on pricesof all the energy sources available in the community);

• connection fees;

• service quality and supply reliability from serviceproviders and retail distributors;

• seasonal variation, as it relates to pricing and servicereliability;

• household coping costs (e.g., how households behaveduring power outages or fuel shortages);

• quantities of fuels and electricity consumed;4 and

• attitudes toward various energy sources (e.g., whetherhouseholds perceive particular types of energy asrelatively clean or polluting, convenient, expensive, or reliable).

Energy and Development Outcomes: The Issue of Causality

The issue of causality is a fundamental concern fordevelopment research. Causal models are needed tolink improvements in energy infrastructure services withsuch desired outcomes as improved household health,education, or income. An analyst must know how apolicy intervention will affect household decisions and,in turn, energy markets. The LSMS survey can be usedto develop such causal models if better energyinformation is available.

However, the statistical and econometric work associatedwith using LSMS survey data on pricing and quantitiescan be challenging, and the socioeconomic informationwith which to estimate household demand functionsmay not be available. To identify the demand function,policy analysts must have access to variations in theprice of fuel or electricity resulting from changing costconditions. These may not be available because ofunchanging, nationally uniform tariffs. Also, increasingblock tariff structures pose difficulties because the marginaland average prices a household pays depend on itsown decision about how much to use. At a minimum,estimating demand functions necessitates gathering dataon the tariff structures applied to households in theLSMS survey sample.

In addition to estimating development outcomes (e.g.,improved health or educational attainment), causality isimportant in analyzing the costs and benefits of energyservices. Most cost-benefit analysis is based on “withand without” or “before and after” comparisons, implyingthat the intervention is causally related to the change instatus of an individual, household, business, or othersocial entity. Seriously evaluating the intervention’s effectrequires some type of control group. After the causal (or at least attributed) benefits are identified, the issuebecomes how to measure them in order to comparethem to the cost of the energy service.

LSMS survey data can be used to estimate the benefitsof infrastructure investments and policy interventions—and thereby help infrastructure analysts and task managersto appraise new projects—in three ways. The simplestmethod is to use data on the household costs of coping

4 Quantities of fuels consumed by the household must be back-calculated from data on household expenditures, using estimates of prices paidby households that sometimes are taken from a source other than the LSMS. In practice, this approach often proves problematic becauseinformation on the applicable tariff or price is not available or responses to the fuel expenditure questions are unreliable.

1 0

with unreliable services as a measure of the benefits ofservice quality improvement. The argument is that, ifservice quality is improved, households will experienceeconomic benefits in the form of cost savings becausethey will no longer spend financial resources on copingwith unreliable services.

The second method involves estimating a demand curve for a particular service based on price-quantityrelationships derived from LSMS energy questions. Inaddition to its use in predicting behavioral changes inresponse to changes in pricing policy, this demand curvecan be used to estimate welfare gains and losses usingstandard cost-benefit methods (Boardman et al. 2001).With an estimated demand curve for a service such aselectric lighting, it is possible to estimate the benefit fora rural household that switches from one lighting sourceto another. An example of this type of work was developedin the Philippines from a specialized household energysurvey conducted in four regions geographicallyrepresentative of that country.

In special circumstances, a third method may be used,whereby an energy or infrastructure analyst includes a“stated preference” module in the LSMS survey anddirectly queries respondents about their demand(willingness to pay) for improved energy services. Forexample, respondents could be asked if they wouldagree to a specified monthly increase in their monthlyelectricity bill in exchange for improved service reliability.Alternatively, choice modeling techniques might be usedto explore the conditions under which a householdwould switch fuels for various uses. These techniquesuse survey techniques to value willingness to pay forlevels and types of services (Mitchell and Carson 1988;Kopp, Schwarz, and Pommerehne 1997). The techniqueswere first pioneered to assess the value of such publicgoods as a clean environment. In this case, levels ofpollution were described (sometimes aided by picturesfor a given situation) to individuals to elicit their responsesregarding how much they would pay to attain a cleanenvironment. Stated preference techniques, such as thecontingent valuation method (CVM) and choice modeling(CM), would probably not require the use of an entireLSMS survey sample. CM techniques in particular wouldrequire samples of only a few hundred households, notthousands as in many LSMS surveys (Haab and McConnell2002; Bennett and Blamey 2001), so the questions

could be administered to a subsample of a largesample survey.

To date, the household energy sector has rarely usedstated preference methods. Experience using thesemethods in other sectors in developing countries suggeststhey can be used successfully in the household energysector. However, this does not imply that such work iseasy or straightforward. Indeed, many applications ofstated preference methods in both developing anddeveloped countries have not provided policymakersreliable information. Generally, the problems haveinvolved unsophisticated sampling and questions thatask respondents how much they are willing to pay forvarious levels of energy service. People generally knowthe national prices for electricity and other fuels andthus are often reluctant to express their willingness topay much higher prices in fear that the hypotheticalsituation posed by the question will become a reality.Sophisticated sampling techniques and well-craftedquestions are necessary to ensure reliable information(Whittington 1998, 2002).

A possible alternative to these techniques requires theavailability of high-quality, cross-sectoral informationwith which to develop models of household demand for specific fuel types.5 Using this approach, householddemand models can potentially explain household fuelchoices, as well as the quantities of each fuel type usedas a function of fuel prices, household income, andother socioeconomic factors. Panel data from repeatedLSMS surveys are particularly useful in evaluating theeffects of policy interventions over time. However, oneshould not underestimate the complexity involved informulating and estimating such household demandmodels in the energy sector.

5 This is possible with LSMS surveys.

1 1

3. IMPORTANCE OF LSMS SURVEY DATA FORENERGY POLICY ANALYSIS

Worldwide, as many as 2.4 billion poor people live inhouseholds that have yet to make the energy transition.The traditional fuels poor people use come at a highcost in time and labor. Hundreds of millions of people—mainly women and children—spend several hours dailygathering wood for cooking. Because of these demandson their time and energy, women and children aredenied opportunities to improve their quality of life.

Household transitions to cleaner, more efficient, andmore productive energy uses can dramatically affect the quality of life of people in developing countries.Cooking with LPG instead of biomass and using efficientwood stoves with chimneys instead of open fires reduceindoor air pollution, a major cause of respiratory illness.Irrigating using electric motors and diesel enginesinstead of traditional methods that use animal andhuman power can increase agricultural productivity.Using electric lighting rather than kerosene lampsencourages evening reading and school attendance. In addition, electricity supply enables households toengage in income-generating activities.

Household Energy Transitions

A significant body of literature on household energytransitions has focused on how households ascend whathas been called an energy ladder (Barnes 1990; Leach1986). At the bottom of the ladder, many people uselow-grade biomass fuels (e.g., straw, dung, and wood)in inefficient and unhealthful ways. At intermediatepoints along the ladder, they turn to commercial fuels(e.g., charcoal, coal, and kerosene), finally switching tocleaner and more convenient fuels (e.g. LPG andelectricity) that can significantly improve their quality of life.

This model theorizes that households switch from low-grade fuels to ones higher up the ladder in incrementalsteps. The main factors determining the type and levelof fuel-switching include income, price, and urbanization(Barnes, Krutilla, and Hyde 2005). For example, highlevels of household income make fuels higher up theenergy ladder more affordable. Similarly, as smallsettlements grow into large cities, fuelwood becomesscarcer while commercial fuels become more readily

available as a result of improved economies of scale ininfrastructure development and fuel distribution. In short,this simple model views the household energy transitionas a smoothly sequenced evolution from fuelwood tocharcoal and kerosene and ultimately to LPG andelectricity.

Transitional Pathways

More recent empirical studies have found that householdsretain a portfolio of fuel options; that is, at any giventime, they are likely to consume a range of fuels alongthe energy ladder (Barnes and Qian 1992; Dewees1989; Hosier and Kipondya 1993). These findings haveled researchers to propose more nuanced models of thehousehold energy transition to explain the variety oftransitional pathways encountered. Reliability of supplyis just one of the factors thought to play a role. Anotheris the cost of energy forms for various uses. Energysources have different levels of cost effectiveness andsubstitutability for the energy services that householdsrequire (e.g., heating, cooking, lighting, and motivepower). Degree of convenience, level of safety, andamount of pollution emitted into the home associatedwith the various energy sources may also be key factorsaffecting household choices and levels of consumption.For example, it is not unusual for households to usecharcoal to prepare tea or coffee early in the morningand use fuelwood to prepare a stew cooked over severalhours, which is eaten later in the day. It has been foundthat even well-off households use fuelwood to bake certaintypes of bread and meats because of the taste it imparts.

Lower-income households, which rely more ontraditional fuels than do higher-income households, aredisproportionately burdened by the costs (both monetaryand non-monetary) of household energy use. Poorhouseholds are more vulnerable to price increases andvolatility and unreliable energy supply. High start-up costsalso present a significant obstacle for poor households.

Greater focus on the energy service or utility that thevarious energy sources provide in various end-useapplications is also important for government policiesthat encourage the household energy transition tocleaner and more efficient energy use. For example,improved fuelwood stoves (e.g., stoves fitted withchimneys that are more efficient than traditional three-

1 2

stone cooking fires) reduce the amount of firewoodneeded to prepare a meal and levels of indoor smoke(Wallmo and Jacobson 1998; Barnes et al. 1993).Similarly, pressurized lanterns are safer than simplekerosene wick lamps and produce more light output perunit of fuel used.

Indicators To Measure the Energy Transition

Indicators constructed from LSMS survey data sets—access, consumption, pricing, and reliability, and levelsof household pollution—can be used to diagnose andtrack the household energy transition. The indicatorsand their variables are described below, and the questionsin the prototype modules that may be used to constructthe indicators are noted.

Access Indicators

The reasons why a particular household lacks access toa modern fuel or electricity involve either a supply-sidefailure—the community where the household lives lacksa local distribution network—or a demand-side failure—the household chooses not to use the available servicebecause it is too expensive, culturally unfamiliar, orotherwise inappropriate. Because the policy implicationsof these supply- and demand-side failures vary sogreatly, access indicators should differentiate betweenthe two situations.

Access ratio. This indicator refers to the percentage ofall households that currently use a specific fuel orelectricity. It does not discriminate between demand-and supply-side explanations for any shortfall in access[Question F01, Fuel Sources Module; Question E01,Electricity Sources Module].

Availability ratio. This indicator refers to the percentageof all households for whom a particular fuel or electricityservice is available, regardless of whether they use it.The indicator captures the extent to which electricityservice or a particular fuel supply has reached all areasof a country, and hence the extent to which householdshave the option of using the particular fuel or electricitysource [Question C11, Electricity and Fuels in theCommunity Module].

Usage ratio. This indicator refers to the percentage ofhouseholds in communities where the fuel or electricityis available that choose to use it [Question F01 andF11, Fuel Sources Module; Question E01, ElectricitySources Module; Question C11, Electricity and Fuels inthe Community Module].

Appliance use. Several appliance-use indicators may becalculated. The stock of electrical appliances may beused as a proxy measure of household economic well-being (similar to the ways in which housing buildingmaterials and housing type are used). In addition, aproxy indicator of indoor air pollution is the percentageof households using biomass fuels for cooking that ownan efficient wood stove fitted with a chimney that extractssmoke to the exterior of the dwelling. Another proxyindicator is the percentage of all households that useelectric stoves or LPG burners and stoves [QuestionA04, Durable Goods (Light Bulbs and Appliances)Module; Question F01 and F11, Fuel Sources Module].

Consumption Indicators

Consumption indicators provide policymakers a pictureof the relative importance of various fuels and theirhousehold end uses.

Gross energy consumption. This indicator is the totalenergy content of fuels and electricity that a householdpurchases. To calculate gross energy consumption, onemust convert the quantity of each of the fuels purchasedby the household into a common unit (such as kilojoules)for the purposes of aggregation (Appendix). Low valuesof gross energy consumption may indicate that householdscannot afford to consume sufficient energy to meetbasic needs of cooking, lighting, and heating. Conversely,high values may indicate that some energy forms areunderpriced, causing wasteful use [Questions F02, F03,F04, F12, F13 and F14, Fuel Sources Module;Questions E03, Electricity Sources Module].

Net energy consumption. This indicator is an adjustmentto the gross energy consumption variable, reflecting theefficiency factors of various fuels in different end uses. Itis therefore a measure of the useful energy obtained bythe household in end-use application. To calculate netenergy consumption from gross energy consumption,

1 3

one must disaggregate gross energy consumption byend-use activity, identify the appliance associated witheach end-use activity, and scale down the gross energyconsumption for each activity by the efficiency factor for the corresponding fuel and associated appliance(Appendix) [Questions F02, F03, F04, F12, F13 andF14, Fuel Sources Module; Question E03, ElectricitySources Module; Question A04, Durable Goods (LightBulbs and Appliances) Module].

Price Indicators

Because LSMS and other multitopic household surveyscan be used to collect information on prices paid,quantities used, and expenditures for all fuels used by a specific household, various price indicators can becalculated. High values or rapid changes in theseindicators may signal an affordability problem amongcertain groups. Analysis of variations in fuel prices mayreveal the presence of spatial monopolies. Price indicatorsinclude unit price, mean-weighted unit price, energyexpenditure, and start-up costs.

Unit price. This indicator is the market price of each fueland of electricity per unit purchased. The unit pricesmay vary geographically and may be an indicator oflocal monopolies or fuel scarcity [Questions F05, F06,F15 and F16, Fuel Sources Module; Questions E03and E05, Electricity Sources Module.

Mean-weighted unit price. The household’s averagegross unit price of energy is the average of the unitprices for each fuel purchased by the household,weighted by the share of that fuel in the household’soverall gross energy consumption [Questions F05, F06,F15 and F16, Fuel Sources Module; Questions E03and E05, Electricity Sources Module.

Energy expenditure. This indicator is the price of eachfuel, multiplied by the quantity consumed. When summedfor all energy sources, one can estimate the aggregatehousehold energy expenditures as a percentage ofhousehold expenditure or income [Questions F04, F06,F14 and F16, Fuel Sources Module; Question E05,Electricity Sources Module.

Start-up costs. This indicator represents the investmentcosts involved in securing access to an energy source(e.g., electricity connection charges or cost of an initialcylinder of LPG). Connection costs can be expressed asa percentage of a poor household’s monthly income[Question C06 (for electricity connection charge only),Electricity and Fuels in the Community Module].

Reliability Indicators

Reliability ratio. This indicator is the average percentageof time for a given period (usually a month) that anygiven fuel or electricity is available for use Questions C03and C04, Electricity and Fuels in the Community Module].

Reliability perception. This indicator refers to gaugingperceptions on the relative availability and scarcity ofvarious fuel and electricity sources [Question C11,Electricity and Fuels in the Community Module].

Pollution Indicators (Measurement of Indoor AirPollution)

In India, a recent pilot study that tested an inexpensiveair pollution monitor found it accurate and easy to use(Balakrishna et al. 2005). If LSMS surveys can use thisdirect way of estimating indoor air pollution, it may bepossible to construct indicators that combine concentrationsof air pollutants with the various types of cooking fuelsthat households use.

Using LSMS Surveys for Energy Policy Analysis

LSMS surveys can be used to understand the relationshipsbetween use of energy services and household welfare;these analysis, in turn, can be used to adjust energypolicies. For example, LSMS surveys can be used toinvestigate associations between use of biomass fuelsand incidence of acute respiratory infection (ARI), withconsequences for public health policy. LSMS data canalso be used to estimate prices paid and expendituresincurred, with implications for energy pricing policy. Itcan also help to establish the household benefits ofrural electrification, which can inform rural electrificationpolicy. This section outlines these types of analysis.

1 4

Biomass Fuel Use and Acute Respiratory Infection

Exposure to smoke from the burning of traditional solidfuels (e.g., fuelwood) for cooking and heating increasesthe risk of disease, most notably acute lower respiratoryinfection (ARI), in both children and adults (Smith andMehta 2003; Kammen 2001; WHO 2002).6 Womenand children are disproportionately affected, in part,because of women’s primary role in food preparation andchild care, which exposes them to indoor air pollution.

Using data from a 2000 LSMS survey in Guatemala, Table 1 illustrates the association of fuel use and respiratoryillness (Ahmed et al. 2005). Incidence is less in householdsthat use mainly LPG and that have a separate kitchen. Bycontrast, incidence is higher among households that usefuelwood only and that lack a separate kitchen.

A 1999 study, based on information from the fourthZimbabwe Demographic and Health Survey, illustratesanalysis that can be performed to assess the effects ofcooking fuel type on ARI (Mishra 2003). In this study,control variables, identified in the research literature ascovariates of ARI, included age of child, sex of child, birthorder of child, nutritional status of child, mother’s age atchildbirth, mother’s education in completed years ofschooling, mother’s religion, household standard of living,

and region of residence.7 Information on fuel types wasused to group households into three categories (Table 2).

The results show that the unadjusted odds of havingsuffered from ARI are almost twice as high amongchildren who live in households using high-pollutionbiomass fuels than among those living in householdsusing low-pollution LPG/natural gas or electricity forcooking. After adjusting for the nine above-mentionedsocioeconomic variables, children in households usingwood, dung, or straw for cooking were more than twiceas likely to have suffered from ARI as those in householdsusing LPG/natural gas or electricity.

The mechanism by which cooking smoke can increaserisk of ARI is not fully understood. Disentangling thecausal associations with income, ventilation, kitchen-usepatterns, exposure times, location of kitchen, and otherfactors remains the subject of much experimental research,including carefully designed epidemiological studies.Nevertheless, it is clear that LSMS studies can be used to investigate associations between household fuel-usepatterns and ARI, and all cooking fuels consumed by thehousehold must be accurately measured.

6 For a comprehensive review of studies on indoor air pollution and ARI, see Smith et al. (2000).7 Typical socioeconomic variables for which LSMS survey data is collected.

FUELWOOD ONLYMIX OF FUELWOOD,

LPG, AND OTHERCLEAN ONLY(MAINLY LPG) TOTAL

BIOMASS, NO KITCHEN

MIXED, NO KITCHEN

BIOMASS, KITCHEN

MIXED,KITCHEN

LPG,KITCHEN TOTAL

Percentage

Total No. Cases

Percentage

Total No. Cases

Source: 2000 LSMS (Ahmed et al. 2005).

48

3,631

50

1,167

53

661

47

2,314

43

1,069

42

841

47

6,052

46

1,769

40

649

47

6,049

Table 1. Respiratory Symptoms in Children, by Fuel and Cooking Space, Guatemala (2002)

CHILDREN WITHRESPIRATORY ILLNESS

FUEL AND ROOM FOR COOKING

FUEL

1 5Energy Prices and Household Expenditure

The questions set out in the fuels and electricity prototypemodules in this report are designed to obtain accuratedata on household use, quantities consumed, andprices paid. LSMS data sets that provide good-qualitydata on these variables can be useful in analyzing theeffect of price changes on household welfare. Aspecialized household energy survey conducted inYemen in 2003 used such questions, illustrating thetypes of analysis possible using such data.

Tables 3 summarizes, for each income decile, the percentof households reporting kerosene use and their averageconsumption and expenditure on kerosene purchases.What varies by income decile is not the quantity ofkerosene used, but the percentage of households usingkerosene. Thus, for example, 92 percent of householdsin the lowest decile and 57 percent in highest decile usekerosene. Rural households are more likely than urbanhouseholds to use kerosene (83 percent compared to46 percent on average across all income deciles)reflecting, in part, easier access to LPG in urban areas(survey questions on LPG access confirm this finding).

Since kerosene is essential to the poor for cooking andlighting, many countries subsidize kerosene prices withthe policy objective of protecting the poor. The rationaleis that the poor should not have to pay the full cost ofthe fuel. However, subsidized kerosene prices haveimplications for public finances since the differencebetween the economic and subsidized price must beabsorbed by the budget. Thus, policymakers would liketo know who benefits from the subsidy and, if the subsidywere reduced, what the reduction in benefits would be.Analysis using the Yemen data illustrates one method ofcalculating the distribution of benefits.

Table 3. Kerosene Use by Income Decile, Yemen (2003)

INCOMEDECILE

1

2

3

4

5

6

7

8

9

10

AVERAGE

Source: World Bank (2005).

URBAN

79

66

50

56

52

46

52

40

27

35

46

RURAL

94

93

85

90

86

80

80

74

71

68

83

ALL

92

89

80

83

79

71

72

64

57

57

75

URBAN

8

10

10

6

6

7

9

8

7

7

8

RURAL

10

11

11

11

9

10

11

11

11

11

10

ALL

10

11

11

11

8

9

11

10

10

10

10

URBAN

282

368

313

233

268

259

360

304

230

303

297

RURAL

394

446

433

475

367

397

415

418

377

406

414

ALL

386

436

422

442

354

374

404

398

354

384

397

High 66.1 17.9

Medium 10.6 15.0

Low 23.3 10.1

* High = wood, dung, or straw; medium =kerosene or charcoal; low = electricity or LPG.Source: Mishra (2003).

Table 2. ARI Prevalence among Children under Five Years of Age, Zimbabwe (1999)

HOUSEHOLDSREPORTING USE

(PERCENT)CONSUMPTION

(LITERS PER MONTH)EXPENDITURE

(YR PER MONTH)

PERCENT OFSAMPLE

ARI PREVALENCE

COOKING FUEL TYPE (POLLUTION LEVEL)*

1 6

In Yemen, the economic price of kerosene is itsinternational price (the cif price at the port of Aden),plus distribution costs and taxes. Since the 2003 ex-distributor price (17 rials per liter) was substantiallybelow that year’s economic price (40 rials per liter), the difference represents the subsidy consumers received.Since the total quantity of kerosene consumed by eachincome decile is known, it is straightforward to calculatethe distribution of the total amount of the subsidy. AsFigure 1 shows, the poorest decile receives 13 percentof the total subsidy, while the richest receives 7 percent.The poverty rate in Yemen in 2003 was about 40 percent,meaning that 52 percent of the total subsidy goes to thenon-poor. Using the data, it is also possible to estimatehow removing the subsidy affects household expenditure.Figure 1 shows the increase in expenditure as a percentof total household budget. The subsidy is equivalent to2.5 percent of the total expenditure of the pooresthouseholds, but only 0.2 percent of the richest.8

Rural Electrification and Development

Rural electrification can transform rural people’s lives in several major ways. Electrification of households candirectly improve household welfare. Electricity for suchproductive uses as irrigation or rural enterprises canboost incomes. Electrification of rural health clinics,schools, and local government offices facilitates thedelivery of public services.

For newly electrified households, the predominant use islighting (Barkat et al. 2002; Barnes and Floor 1996;Denton 1979). Numerous studies have found a correlationbetween electricity used in rural households and theamount of time adults read and children study (Barnes1988; World Bank 2002, 2004). The probable reasonis that kerosene lamps and candles, the usual alternativeto electricity in rural households, provide inadequatelight for reading (Nieuwenhout, Van de Rijt, andWiggelinkhuizen 1998). The significantly higher levels of lighting that electric lamps provide enable comfortablereading, which can potentially improve education andschool attendance (Khandker, Lavy, and Filmer 1994;Khandker 1996; Gordon 1997).

After lighting, household priority uses of electricity includetelevisions, fans, and other household appliances(Sathaye and Tyler 1991; World Bank 2002; Saunderset al. 1975). Television, increasingly found in electrifiedrural households, is an important source of entertainmentand information.

Households with electricity may experience time savingsif they no longer must travel to battery-charging stationsin a nearby town. Rural electrification can potentiallysupport improvements in rural productivity through theuse of electric pumps for irrigation and small electricalappliances in rural businesses (Ranganathan andRamanayya 1998; World Bank 2002; Cabraal, Barnes,and Agarwal 2005; World Bank 1994).

The improvements that electrification brings are amplifiedwhen accompanied by complementary investments,such as infrastructure (USAID 1983). A 1999 study inPeru found that providing a combination of infrastructureservices—such as electricity, water, sanitation, andtelephones—has a greater effect on poverty reduction

0

3

6

9

12

15

% S

hare

13%

2.5%

% Total subsidy

% Household budget

13%

1.4%

13%

1.2%

9%

1.0%

9%

0.8%

9%

0.7%

11%

0.7%

8%

0.6%

7%

0.4%

7%

0.2%

Income Decile1 2 3 4 5 6 7 8 9 10

FIGURE 1. Share of Total Subsidy Received and Shareof Subsidy in Household Budget – by Income Decile

Source: World Bank (2005).

8 The calculation assumed no fuel switching and a price elasticity of – 0.2.

1 7

than providing a single service (World Bank 1999).Providing a second complementary service is important,but more significant is providing a third or fourth service.Indeed, the study found that adding the fourth servicecan bring welfare improvements seven times greaterthan those delivered by the second service. A morelimited amount of evidence concerns electricity andchanges in fertility rates (Herrin 1979).

Thus, the benefits of electrification are many, some ofwhich may be more easily quantified using data from LSMSsurveys. The sections below describe several recent studiesthat measured the benefit of electric lighting and therelationship between rural electrification and education.

Willingness To Pay for Lighting

A 2002 study in the Philippines used survey data todevelop a method for estimating the household benefitsof electrification in monetary terms (World Bank 2002).Increasingly, rural electrification projects use this methodto establish their economic returns. This section describesthe theoretical framework for estimating the benefits ofelectric lighting using the willingness-to-pay method,along with a practical way to calculate the benefits. Theresults of applying the method in three recent WorldBank rural electrification projects are also provided.

Consumer surplus—the difference between what thehousehold pays for lighting service and the maximumamount it would be willing to pay—is the measure usedto gauge the household benefits of electrification.Willingness to pay is based on the actual demand for aservice, such as lighting. Based on the willingness-to-paymethod, the monetary benefit to a household is the areaunder the demand curve, as shown in Figure 2. The onlyway to estimate such a demand curve is through ahousehold survey that quantifies the amount of fuelsand electricity consumed for lighting. This information is combined with information on the lighting serviceobtained from the various lighting devices the householduses. For the sake of simplicity, this example assumesthat, before electrification, the household only useskerosene in kerosene lamps for lighting (in addition tokerosene, various other lighting sources, such as candlesand dry-cell batteries, are usually used).

Before a household obtains electricity from the nationalor local grid, it uses kerosene or another form of energyfor lighting. Thus, before electrification, the quantity oflighting services (kilolumen hours [klm-hr] per monthconsumed) is QKERO, at the price PKERO. Thus, totalhousehold expenditure on lighting is QKERO x PKERO,which equals area B + D. When a household obtainselectricity service, its demand for lighting changes toQELEC at price PELEC.

Total willingness to pay for the service at level QKERO isthe total area under the demand curve to that level ofconsumption; that is, area A + B + D. This is the totalbenefit to the consumer. However, the cost is area B +D, and therefore the net benefit or consumer surplus isthe difference between the two; namely area A.

After electrification, the level of service (in the case oflighting, the number of lumen-hours) typically increasessubstantially; consumption increases from QKERO toQELEC, but the price paid for the electrified service also falls from PKERO to PELEC. Now the household’sexpenditure for electricity is PELEC x QELEC, which equalsarea D + E.

At this level of consumption, the total area under thedemand curve to QELEC—that is, the total benefit—isnow area A + B + C + D + E. Therefore, the net benefitor consumer surplus, after subtracting the cost or D +E, equals A + B + C. Thus, it follows that the neteconomic benefit of electrification is the increase in

Price

QKERO QELEC

Service Level

PELEC

PKEROA

y

B

D E

C

x

FIGURE 2. Demand Curve for Lighting

Source: World Bank (2002).

1 8

Electricity and Education

As pointed out above, the benefits of electrificationinclude time saving, increased productivity, and improvedhealth and education outcomes. Thus, the willingness-to-pay measure for lighting represents only a portion of the benefits. The Philippines study described above is one of the few to have estimated the education andtime-saving benefits of electrification in monetary terms.When it estimated the combined effect of electricity andeducation on income, it found that wage earners inhouseholds with electricity can expect to earn $37-47more per month than their counterparts without electricity.This amount represents about 25 percent of a typicalhousehold’s monthly income in the survey areas, which is

consumer surplus, or area B + C. Areas B and D canbe calculated using survey data on prices and quantitiesof kerosene used for lighting. The types of kerosenelanterns that households use may also be known fromthe LSMS survey (Durable Goods [Light Bulbs andAppliances] Module) or can be otherwise determined.The service obtained—expressed as klm-hr per month—can be obtained by multiplying lamp light output by thenumber of hours per month that lamps owned by thehousehold are used.9

Similarly, area E is calculated by multiplying the total cost of electricity used for lighting by the service obtained.To calculate the service obtained by the households, one must acquire information on the number of electriclamps the household uses, their rating, and hours ofusage per month. Manufacturer information on lightoutput from various types of light bulbs can then be used to calculate QELEC (klm-hr per month).10

The shape of the demand curve cannot be determinedwhen only two demand points (for lighting obtainedusing kerosene and electricity) are known. In the absenceof empirical data for intermediate points along the demandcurve (i.e., for lighting obtained using dry-cell batteries,diesel gensets, or car batteries), assumptions on theshape of the curve between points x and y are critical tocalculate area C; in such a case, a standard functionalspecification with a constant elasticity may be used.

Since its development using survey data from thePhilippines, this method has been applied in economicanalysis of several World Bank projects. Table 4summarizes survey results from three projects. Thesecalculations used data from specialized householdenergy surveys; however, similar calculations could bemade using data from LSMS surveys that incorporatethe questions presented in the prototype modules.

These household benefits are for lighting alone, andthey are well above the cost of electricity service. As iswell known, households can derive an array of otherbenefits from electricity, including entertainment, foodpreservation, and comfort from fans.

9 A simple kerosene lamp may provide up to 40 lumens and a hurricane lamp 10-100 lumens, depending on the type of wick and lamp. A kerosene mantle lamp may provide approximately 400 lumens. Actual values vary and should be obtained from experimental data.

10 Approximate light output of a 40-W incandescent lamp is 400 lumens, 630 lumens for a 60-W incandescent lamp, 600 lumens for a 15-W fluorescent lamp, and 900 lumens for a 16-W compact fluorescent lamp (CFL).

Table 4. Willingness-to-pay Resultsfrom Three Projects

FACTOR

QKERO

(klm-hr per month)

QELEC

(klm-hr per month)

PKERO ($ per klm-hr)

PELEC ($ per klm-hr)

Elasticity

B (US$)

C (US$)

D (US$)

E (US$)

Total willingness to

pay ($ per household

per month)

Sources: World Bank project appraisal documents:

ERTIC Project PAD (Bolivia data), 2003; Second

Southern Provinces Rural Electrification Project

(Lao PDR data), 2004; Peru Rural Electrification

Project (Peru data), 2005.

BOLIVIA

7

90

0.48

0.04

-1.03

3.08

5.56

0.28

3.32

12.24

LAO PDR

20

435

0.195

0.003

-1.74

3.84

6.05

0.06

1.25

11.20

PERU

4.6

363

0.57

0.01

-1.08

2.58

9.95

0.05

3.58

16.16

PROJECT COUNTRY

1 9

well above the price households pay for electricity service.It should be cautioned that this is an estimate of the valueof all the lighting service under the demand curve, andcannot be used as a proxy for increase in income.

Data from the LSMS and specialized household energysurveys can provide some evidence of the associationbetween electrification and education, even if falling shortof estimating monetary values of the benefits (Table 5).

The 1996 India survey found that, in households withoutelectricity, virtually no women read, regardless of incomeclass. By contrast, in households with electricity, literacywas much higher, regardless of income class. Findingsfrom the 1998 Nicaragua LSMS survey were similar.Electricity in the home was highly correlated with schoolattendance and family literacy. Obviously, thesecorrelations constructed from survey data do not establishcausality; nevertheless, the findings are compelling andare consistent with those in other countries. The implicationis that this could be an important research area forfuture LSMS surveys with better energy questions.

4. THE WAY FORWARD

The purpose of these guidelines is to design a set ofenergy questions for use in LSMS surveys that, while notas extensive as those in the specialized household energysurveys, could still provide sufficient information withwhich to improve analysis of household energy use.

It is proposed that the few ad-hoc questions on energyuse currently found in LSMS surveys—which yield littleinformation useful to policymakers—be replaced by thequestions presented in the prototype energy modules that follow. The rationale for including these particular questions is based on the review of the key policy issues that household data can inform and an assessment ofthe feasibility of including such questions in LSMSsurveys. The format of the proposed modules is similarto that often used in current LSMS surveys in order tofacilitate their adoption.

Formulation of the questions presented in the prototypemodules draws on experience from recent LSMS andspecialized household energy surveys. The notesprovided point out commonly encountered problemsand how they may be overcome. The notes caution that

INDIA SPECIALIZED HOUSEHOLD ENERGY SURVEY, 1996

Table 5. Electricity and Education Links in India and Nicaragua

INCOMEDECILE

Lowest

Middle

Highest

WITH ELECTRICITY

48.2

57.3

64.4

WITHOUTELECTRICITY

7.7

14.1

26.5

WITH ELECTRICITY

29.9

46.3

39.0

WITHOUTELECTRICITY

4.1

2.1

2.2

FAMILY LITERACY (%)WOMEN READING IN THE

LAST 24 HOURS (%)

REGION

Atlantic

Central

Pacific

Total

Sources: Kulkarni and Barnes (2005); World Bank (2002).

WITH ELECTRICITY

74

74

77

73

WITHOUTELECTRICITY

46

50

62

53

WITH ELECTRICITY

77

77

73

72

WITHOUTELECTRICITY

40

46

62

50

HOUSEHOLD LITERACY (%)CHILDREN ENROLLED

IN SCHOOL (%)

NICARAGUA LSMS SURVEY, 1998

2 0

the questions must be customized to the particularcountry for which a survey is designed, taking intoaccount its unique circumstances of fuel and electricityavailability and use.

The prototype modules presented below have beendesigned for insertion in an LSMS survey as stand-alonemodules. Since the questions in each prototype modulecomplement each other, their potential to yield the greatestamount of data on household energy use will be realizedif they are kept whole. However, because of the necessityto limit the number of modules, LSMS survey designersmay not always be able to insert stand-alone modules.Instead, the electricity and fuels questions on access,consumption, prices, costs, and end use in the prototypemodules may be integrated into another module, suchas the Housing Module, and questions on time spentcollecting fuels into the Time Use and Labor Module.

Questions in the Durable Goods (Light Bulbs andAppliances) Module may be integrated into the LSMSDurable Goods Module. It should be kept in mind that,in addition to the question “Does your household own[item]?,” the essential complementary questions (in orderto estimate a monetary value of the benefits of energyuse) are “How many [items] do you own?” and “Howmany hours during the last week did you use the [items]?”

For best results, it is recommended that all of thequestions in the prototype modules be included.

Fuel Sources Module

The purpose of the Fuel Sources Module is to collectinformation on household access, consumption, andpayment for all the fuels used (Figure 3). Those householdmembers that pay for or collect fuels are usually thebest-informed respondents.

The list of household fuels in this module is indicative(Table 6). The LSMS survey designer may need to deletefuels that are not available in the country for which theLSMS is being designed and add others. The surveydesigner should recall that fuel shortages or priceadvantages of one fuel over another can induce fuelsubstitution. For example, if LPG is subsidized andgasoline is not, households can have their car fuelsystems adapted to use LPG instead of gasoline.

Notes on Fuel Sources Prototype Module

F01 and F11: During the last 30 days, has yourhousehold used [for each fuel]?

These questions establish which energy sources thehousehold uses. It is important to collect informationabout all fuels used by the household, not simply thosemost frequently used, as households tend to use aportfolio of fuels. Design of the module ensures thatenumerators will first determine which fuels and electricitysources households use. Only after establishing whichfuels households have used in the last 30 days andwhich electricity sources are available to them will theenumerator ask the follow-up questions for each fueland electricity source.

F02 and F12: What is the typical unit of measure[for each fuel]?

Sometimes atypical units are collected (e.g., large treelimbs in the case of firewood). The enumerator will haveto adjust these for a commonly used unit. Instructionsshould be given to the enumerators to make notes of allsuch cases.

FUEL

Firewood, animal dung, and other biomass

Charcoal

Candles

Kerosene

Biogas

Liquefied petroleum gas (LPG)

Diesel

Gasoline

Distinct heating

Natural gas

HOUSEHOLD USE

Cooking and hot-water heating

Cooking and hot-water heating

Lighting

Lighting (wick andhurricane lamps) and cooking

Cooking

Cooking and lighting (less often)

Lighting (wick andhurricane lamps) and electricity (diesel generators)

Transport (motor bikes and cars) and electricity(gasoline generators)

Space heating

Cooking and space heating

Table 6. Fuels and Their CommonHousehold Uses

21

F03: What is the approximate weight of a typicalunit of [for each fuel]?

F13: What is the approximate weight or volumeof a typical unit of [for each fuel]?

Accurate measurement of the units in which energy isconsumed is critical to data quality. The complexity ofaccurately capturing this information varies substantiallyacross fuels. Some are measured in formal scientificunits (e.g., liters in the case of kerosene or kilograms forLPG). However, respondents may think more in terms ofthe cylinders or bottles in which the fuel is sold ratherthan the scientific units they represent. For example,kerosene is often sold in several sizes of plastic bottles,and respondents may not know their volumetric content.It is important to develop appropriate promptingstrategies so that the enumerator can establish whichsizes the household uses and ensure that the correctweight is entered.

Specific fuels present their unique challenges. Forexample, the weight contained in LPG cylinders candiffer markedly from that marked on the cylinders. UsingLSMS survey data, the analyst (not the field enumerator)will need to establish the average net weight of LPGcylinders sold. Biomass fuels present the most difficultcase in that they tend to be sold or collected in measures—that is, bags of charcoal, bundles of firewood, or cakesof dung—for which only approximate conversion toscientific units may be possible. It is therefore essentialto begin the questionnaire design process with a detailedunderstanding of the units in which fuels are measuredand sold in local markets, as well as a method forconverting these into scientific units. For example, incertain countries, it is not unusual for charcoal to be soldin discarded fertilizer bags. It is not difficult to establishan average weight for the amount of charcoalcontained in these bags.

F04 and F14: How many units of [for each fuel]has your household used in the last 30 days?

These questions record the household’s physicalconsumption of each fuel during the last 30 days. Differentfuels tend to have distinct time cycles for purchase andconsumption, which make accurate responses problematic.Households tend to make frequent purchases of certainfuels, such as kerosene. They purchase these in smallquantities and consume them entirely within days or weeks.

Households (especially poor ones) tend to make lessfrequent purchases of such fuels as LPG, which may beavailable only in 11-kg cylinders. The household mayconsume a single cylinder over a period longer than 30days. The difficulty is to accurately estimate what fractionof the 11-kg cylinder the household consumed over the last 30-day period. As pointed out above, seasonalitycan be an important factor in fuel consumption.Consumption of fuels for space heating peaks duringcold seasons. Religious festivals and harvests are often a factor that causes increased consumption of transportand cooking fuels. Thus, local habits for obtaining fuelsshould be considered during the questionnaire designstage so that the choice of time period can be adaptedas needed and appropriate prompting strategies can bedeveloped for the interview.

In deciding whether to specify the last 30 days or anothertime period for this question, as well as for QuestionsF05, F07, F15 and F17, the survey designer should takeinto account the particular country’s energy-use patternsand whether it has marked seasonal variations of use. Ifhouseholds have marked seasonal variations in fuelsconsumption, it may be necessary to ask an additionalquestion on the household’s average consumption permonth.

F05 and F15: How many of these units of [foreach fuel] did your household purchase in thelast 30 days?

F06 and F16: What is the typical price yourhousehold pays per unit of [for each fuel]?

F07 and F17: What was the total cost of all theunits of [for each fuel] that your householdpurchased in the last 30 days?

Questions F05, F06, F07, F15, F16 and F17 are used torecord the household’s expenditure on each fuel sourceduring the last 30 days. Expenditure is calculated bymultiplying the number of units purchased (F05 or F15)by the price paid (F06 or F16). The result of thiscalculation should be consistent with the response toQuestions F07 and F17. If the results are not consistent,the analyst must decide which is more reliable. Theanswers are straightforward for fuels purchased frequentlyin cash. However, as noted above, some fuels may bepurchased less frequently than once per month, making itharder for households to estimate monthly expenditure.

22

ID CODE

FUEL

SOU

RCES

Agr

icul

tura

l Re

sidu

e

F01

Dur

ing

the

last

30 d

ays,

has

your

hou

seho

ldus

ed [

...]?

F02

Wha

t is

the

typ

ical

unit

of m

easu

re

for

[...]?

Uni

t

F03

Wha

t is

the

appr

oxim

ate

wei

ght

of a

typi

cal u

nit

of[..

.]? U

sede

cim

als.

Kilo

gram

s

F04

How

man

yun

its o

f [..

.]ha

s yo

urho

useh

old

used

in t

hela

st 3

0 da

ys?

(incl

ude

both

purc

hase

dan

d co

llect

edun

its).

Use

deci

mal

s.

Num

ber

F05

How

man

y of

the

seun

its o

f [..

.]di

d yo

urho

useh

old

purc

hase

duri

ng t

hela

st 3

0da

ys?

Use

deci

mal

s.

If no

tpu

chas

ed>

>F0

8

Num

ber

F06

Wha

t is

the

typi

cal

pric

e yo

urho

useh

old

pays

per

unit

of[.

..]? Am

ount

F07

Wha

t w

as t

heto

tal c

ost

ofal

l the

uni

ts

of [

...]

that

your

hous

ehol

dpu

rcha

sed

inth

e la

st 3

0da

ys? Am

ount

F08

How

muc

htim

e di

d al

lm

embe

rs o

fyo

urho

useh

old

spen

dco

llect

ing

[...]

in t

he la

st 3

0da

ys?

Incl

ude

roun

dtri

ptr

avel

tim

e.

Hou

rsM

en

W

omen

F09

Wha

t is

the

one-

way

dist

ance

mem

bers

of

you

rho

useh

old

typi

cally

trav

el t

oco

llect

[...

]?

Kilo

met

ers

Ligh

ting

Coo

king

Spac

eH

eatin

gH

ot W

ater

Hea

ting

Oth

er

F10

Wha

t pe

rcen

tage

of

[...]

was

use

d fo

r th

e fo

llow

ing

purp

oses

? En

ter

9999

for

no

use

; 1,

2,

3, o

r 4

[1=

25%

2=50

%3=

75%

4=10

0%]

Ask

que

stio

n F0

1 f

or e

ach

sour

cefir

st.

Then

>>

to

ques

tions

F02 -

F10

for

each

que

stio

n fo

r w

hich

the

resp

onse

to

F01

was

“ye

s” [

1].

Yes.

..1

No.

..2>

>N

ext

Item

Cod

e

BIO

MA

SSPi

ece

......1

Bund

le..2

Stac

k......3

Bag

........4

Sack

......5

CA

ND

LESm

all..

....6

Med

ium

..7

Larg

e....8

OTH

ER..14

Dun

g

Fire

woo

d

Oth

ertr

aditi

onal

biom

ass

Cha

rcoa

l &

Coa

l

Can

dles

FIG

URE

3A

. Fue

l Sou

rces

(bio

mas

s an

d ca

ndle

s)–

Prot

otyp

e M

odul

e

23

ID CODE

FUEL

SOU

RCES

LPG

Kero

sene

Gas

olin

e

Die

sel

Nat

ural

Gas

FIG

URE

3B.

Fuel

Sou

rces

(g

aseo

us a

nd li

quid

fuel

s) –

Pro

toty

pe M

odul

e (c

ontin

ued)

F11

Dur

ing

the

last

30 d

ays,

has

your

hou

seho

ldus

ed [

...]?

F12

Wha

t is

the

typ

ical

unit

of m

easu

re

for

[...]?

Uni

t

F13

Wha

t is

the

appr

oxim

ate

wei

ght

orvo

lum

e of

aty

pica

l uni

tof

[...]?

Use

deci

mal

s. Lite

rs

F14

How

man

yun

its o

f [..

.]ha

s yo

urho

useh

old

used

in t

hela

st 3

0 da

ys?

Use

dec

imal

s.

Num

ber

F15

How

man

y of

the

seun

its o

f [..

.]di

d yo

urho

useh

old

purc

hase

duri

ng t

hela

st 3

0da

ys?

Use

deci

mal

s.

If no

tpu

chas

ed>

>F1

8 Num

ber

F16

Wha

t is

the

typi

cal

pric

e yo

urho

useh

old

pays

per

unit

of[.

..]? Am

ount

F17

Wha

t w

as t

heto

tal c

ost

ofal

l the

uni

ts

of [

...]

that

your

hous

ehol

dpu

rcha

sed

inth

e la

st 3

0da

ys? Am

ount

F18

Wha

t is

the

one-

way

dist

ance

mem

bers

of

you

rho

useh

old

typi

cally

tr

avel

to

purc

hase

[...]? Kilo

met

ers

F19

Wha

t pe

rcen

tage

of

[...]

was

use

d fo

r th

e fo

llow

ing

purp

oses

? En

ter

9999

for

no

use

; 1,

2,

3, o

r 4

[1=

25%

2=50

%3=

75%

4=10

0%]

Ask

que

stio

n F1

1fo

r ea

ch s

ourc

e fir

st.

Then

>>

to

ques

tions

F12 -

F19

for

each

que

stio

n fo

r w

hich

the

resp

onse

to

F01

was

“ye

s” [

1].

Yes.

..1

No.

..2>

>N

ext

Item

Cod

e

LPG

12kg

......9

16kg

....10

LIQ

UID

FU

ELS

Litr

es....11

Bott

les.

.12

Jugs

....13

OTH

ER..14

Ligh

ting

Coo

king

Spac

eH

eatin

gH

ot W

ater

Hea

ting

Irrig

a†io

nTr

ansp

ort

Oth

er

Kilo

-gr

ams

24

F08: How much time did all members of yourhousehold spend collecting [for each biomassfuel] in the last 30 days? Include time spentpurchasing and collecting, as well as round-triptravel.

This question identifies the extent to which householdspay for different fuels in terms of the time they expendcollecting them. This question is expected to be relevantprimarily for traditional biomass fuels, including dungand fuelwood. However, it may sometimes be relevantfor other fuels (e.g., car batteries, which may need to betaken to the nearest town for recharging). This question,combined with other LSMS survey information, can beused to explore which types of activities are curtailed byhouseholds who engage in wood collection. Identity ofthe household members involved in fuel collection isimportant because it affects the way in which the timededicated to this activity is valued. Depending on thecontext, it may be important to distinguish not onlybetween men and women and adults and children, butalso adults that earn income outside the household andchildren in and out of school. Households membersmay combine time spent collecting or purchasing fuelswith other tasks. Local habits for obtaining fuels shouldbe considered during the questionnaire design stage sothat the question can be adapted as needed.

F09: What is the one-way distance members ofyour household typically travel to collect [for eachbiomass fuel]?

F18: What is the one-way distance members ofyour household typically travel to purchase foreach liquid fuel]?

These questions record the distance a household travelsto reach the place where each type of energy is collectedor purchased. One should note that it may be difficult toobtain accurate distance estimates from respondents.Research addressing this issue and the relative merits ofasking for “time to source” is being undertaken andmay be helpful to review. Some objective testing of thismeasure prior to conducting a survey is recommended.

F10 and F19: What percentage of [for each fuel]was used for the following purposes?

These questions document the uses to which each fuel isput. The end-use categories are country-specific; thus, the

module must be carefully reviewed so that it is adapted tothe particular country. For example, if the country’s use offuels for space heating is negligible, that end-use categorycan be removed from the questionnaire. In-country testingmay be needed to determine if the percentages can be leftopen or if set categories (1 = 25%, 2 = 50%, 3 = 75%,and 4 = 100%) are needed.

Electricity Sources Modules (Grid and Off-grid)

The Electricity Sources (Grid) Module pertains toelectricity the household obtains from a utility (e.g., grid electricity), community system (e.g., mini- andmicro-hydro systems), or neighbor or relative (e.g.,informal connections to a neighbor or relative’s gridsupply or genset) (Figure 4a). The purpose of this moduleis to collect all relevant information on electricity access,consumption, payment, and uses, bearing in mind thathouseholds may rely on a variety of sources. Thehousehold member that pays for electricity purchasesis usually the best-informed respondent.

E01: Does your household have electricity from[for each electricity source]?

This question is asked first for all possible electricitysources (Box 1). Only after answering for all possiblesources does the interviewer ask Question E02 (for grid,genset, and micro-hydro sources). If the householdobtains no electricity from the grid, the interviewer willmove on to Question E07 (for off-grid sources).

Box 1: Electricity Sources for Household Use

• Car (storage) batteries

• Household-owned electric generator (genset)

• Electric generator owned by neighbor

• Electricity from mini-grid or village/community grid

• Electricity from national, regional, or town grid

• Pico-and micro-hydro electric generator

• Solar (photovoltaic) home system (SHS)

25

Households may have access to grid electricity from a utility without being utility customers if they obtainelectricity illegally (unofficial grid connection) orinformally (by wiring up to a neighbor’s system). Themodule somewhat artificially distinguishes between utilitycustomers and non-customers by treating grid electricityin the latter case as a distinct electricity source. Theappropriate expression for these types of arrangementsmust be determined by taking local conditions intoaccount. If the grid supply is particularly unreliable, thehousehold may have both grid and non-grid sources,with the latter used as a back-up system. Question E01also clarifies the name of the electric utility that suppliesthe household so that it is possible to later verify thetariffs and charges faced by the household. The namesof all available formal electricity-service providers shouldbe listed on the printed questionnaire. The questionshould therefore be adapted to local circumstances toensure that this information is obtained.

Notes on Electricity Sources (Grid) PrototypeModule

E02: What is electricity used for in your home?

List up to five in order of importance. The end-usecategories are country-specific and must be reviewedcarefully when the module is adapted for use in anyparticular country. For example, if there is negligibleelectricity use for space heating, hot-water heating, orcooking in the country, these end-use categories can beremoved from the questionnaire. Because substitution ofelectricity with fuels for certain applications may involveimportant policy issues, design of this question and thecorresponding one in the fuels sources module (F10)may need to be considered together (for example,policymakers want to discourage use of electricity forcooking [because it causes demand to spike duringevening hours, making it impossible for the utility tomatch the demand load during that time] and promotesubstitution by LPG or natural gas). Informationobtained through the Durable Goods (Light Bulbs andAppliances) Module will complement and serve tovalidate the information obtained with this question.

E03: How much electricity did the householdconsume from [for each electricity source] during30 days of the last billing period?

For formal metered connections to grid electricity,consumption is recorded on the monthly bill received by the customer. If possible, the household should beasked to show the enumerator the bill. If the norms ofconfidentiality permit, it may be desirable to record thehousehold’s customer number so that its consumptionrecord can be matched with that of the customerdatabase. When the bill cannot be shown by therespondent (or when the household has a pre-paymentmeter and does not receive a bill), it is unlikely thatrespondents can answer this question accurately.Therefore, when the bill is unavailable, this box shouldbe left blank. During analysis of the data set, thequantity of electricity consumed may be inferred byapplying the charges detailed in the published tariffschedule to the reported monthly expenditure (responseto Question E06).

E04: What is the basis for the electricity chargesthat you pay for [for each electricity source]?

Households that live in rented accommodations may not be utility customers. Instead, they may pay forelectricity as part of their monthly rental; some householdsmay not pay at all. Households that share a meter maypay a neighbor. It is important to record this informationas it can help interpret the information provided on theamount the household paid for electricity (response toQuestion E06).

FIG

URE

4A

.Ele

ctric

ity S

ourc

es (G

rid) -

Pro

toty

pe M

odul

e

26

ID CODE

ELEC

TRIC

ITY

SOU

RCES

Now

I w

ould

like

to

ask

you

abou

tyo

ur s

ourc

e(s)

of

elec

tric

ity.

Firs

t I

will

ask

you

que

stio

ns a

bout

you

rel

ectr

icity

usa

ge f

rom

gri

d so

urce

s.

E01

Doe

s yo

ur h

ouse

hold

have

ele

ctri

city

fro

m[.

..]?

Ask

que

stio

n E0

1 f

or e

ach

sour

ce f

irst

. Th

en >

> t

oqu

estio

ns E

02 -

E05 f

orea

ch q

uest

ion

for

whi

chth

e re

spon

se t

o E0

1 w

as“y

es”

[1].

Yes.

..1

No.

..2

>>

E06

E02

Wha

t is

ele

ctri

city

use

d fo

r in

you

rho

me?

Lis

t up

to

five

in o

rder

of

impo

rtan

ce.

Ligh

ting

....

....

....

....

....

....

....

....

....

1TV

/Rad

io..

....

....

....

....

....

....

....

....

2Fa

ns..

....

....

....

....

....

....

....

....

....

....

3Re

frig

erat

ion

............................4

Coo

king

....

....

....

....

....

....

....

....

....

5H

ot W

ater

Hea

ting

(for

was

hing

/bat

hing

)................6

Wat

er P

umpi

ng(n

ot in

cl.

Irri

gatio

n)..................7

Irri

gatio

n..

....

....

....

....

....

....

....

....

8Sp

ace

Hea

ting

..........................9

Oth

er S

mal

l App

lianc

es..........10

E03

How

muc

h el

ectr

icity

did

the

hous

ehol

d co

nsum

efr

om [

...]

duri

ng 3

0 da

ys

of t

he la

st b

illin

g pe

riod

?

Cal

cula

te f

rom

bill

. If

bill

unav

aila

ble,

leav

e bl

ank.

Kilo

wat

t ho

urs

(kW

h)

E04

Wha

t is

the

bas

is f

or t

heel

ectr

icity

cha

rges

tha

t yo

u pa

y fo

r [..

.]?

Util

ity M

eter

......

......

......

....1

Pay

Nei

ghbo

r....

......

......

....2

Incl

uded

in R

ent.

......

......

...3

Don

’t Pa

y...

......

......

......

.....4

Oth

er...

......

......

......

......

.....5

E05

How

muc

h di

d th

e ho

useh

old

pay

for

30 d

ays

of e

lect

rici

ty

use?

Cal

cula

te f

rom

bill

. If

bill

unav

aila

ble,

as

k re

spon

dent

to

estim

ate.

Am

ount

Estim

ated

1st U

seC

ode

2nd

Use

Cod

e3r

d U

seC

ode

4th

Use

Cod

e5t

h U

seC

ode

GRI

D E

LEC

TRIC

ITY

[Nam

e]:

from

a P

ublic

Com

pany

,M

unic

ipal

Cor

pora

tion,

Etc.

[Hou

seho

ld is

a U

tility

Cus

tom

er]

GRI

D E

LEC

TRIC

ITY

from

Nei

ghbo

r or

Rel

ativ

e[H

ouse

hold

is n

ota

Util

ity C

usto

mer

]

GEN

SET

ELEC

TRIC

ITY

from

Nei

ghbo

r or

Rel

ativ

e

MIC

RO-H

YDRO

from

Com

mun

ity

Cal

cula

ted

from

bill

FIG

URE

4B.

Elec

tric

ity S

ourc

es (O

ff-gr

id) -

Pro

toty

pe M

odul

e

ID CODEEL

ECTR

ICIT

Y SO

URC

ES(C

ontin

ued)

Now

I w

ould

like

to

ask

you

abou

t an

y of

f-gr

idso

urce

s of

ele

ctri

city

tha

tyo

u us

e.

E06

Doe

s yo

ur h

ouse

hold

have

ele

ctri

city

fro

m[.

..]?

Ask

que

stio

n E0

6 f

or e

ach

sour

ce f

irst

. Th

en >

> t

oqu

estio

ns E

07 -

E011 f

orea

ch q

uest

ion

for

whi

chth

e re

spon

se t

o E0

6 w

as“y

es”

[1].

Yes.

..1

No.

..2

E07

Wha

t is

ele

ctri

city

use

d fo

r in

you

rho

me?

Lis

t up

to

five.

Ligh

ting

....

....

....

....

....

....

....

....

..1

TV/R

adio

....

....

....

....

....

....

....

....

2Fa

ns..

....

....

....

....

....

....

....

....

....

3Re

frig

erat

or............................4

Oth

er

(Spe

cify

)................................1

0

E08

How

muc

h di

d yo

upa

y fo

r yo

ur [

...]?

purc

hase

cos

tof

syst

em if

obt

aine

dfo

r fr

ee:

code

999

9

Am

ount

E09

How

did

you

pay

for

th

is [

...]?

Cas

h Pu

rcha

se...

......

.....1

Leas

e...

......

......

......

......

.2C

redi

t...

......

......

......

......

.3O

ther

......

......

......

......

....4

E10

How

muc

h di

dyo

u pa

y to

oper

ate

your

[...]

in t

he la

st30

day

s?

Am

ount

E11

Wha

t is

the

ra

ting

of y

our

[...]?

SOLA

R (P

V)

HO

ME

SYST

EM

STO

RAG

E (C

AR)

BA

TTER

Y

GEN

SET

(PER

SON

AL)

E12

Do

you

use

elec

tric

ity in

you

r ho

useh

old

for

hom

ebu

sine

ss p

urpo

ses?

Yes.

..1

No.

..2

Wat

t pe

ak(W

p)

Batt

ery

char

ging

cost

/30 D

ays

Volts

Am

p H

ours

Fuel

cos

t/30

Day

skW

27

1st U

seC

ode

2nd

Use

Cod

e3r

d U

seC

ode

4th

Use

Cod

e5t

h U

seC

ode

28

E05: How much did the household pay for 30days of electricity use from [for each electricitysource]?

Ideally, the answer to this question can be obtainedfrom the utility bill, which the enumerator should ask to see. In some countries, it may be possible and usefulto record the household’s customer number so thattime-series information on the household’s consumptioncan be obtained from the utility’s database. The amountcharged to the household is determined by applying thetariff schedule, which may have fixed charges; variablegeneration, transmission, and distribution charges; and taxes and fees. The total amount of the bill (i.e.,inclusive of all charges) should be recorded. If it is not possible to view the bill, the respondent should be asked to estimate this amount. As noted above,information on the quantity of electricity consumed mayhave to be inferred from payment information; thus, it iskey to obtain accurate information in response to thisquestion. Households that share a meter may also sharethe bill (i.e., for billing purposes, one household isrecorded as the customer in the utility’s billing system;this customer receives the bill, but other householdsconnected to the meter share payment). The questionmay need to be adapted to take account of this.

Notes on Electricity Sources (Off-grid)Prototype Module

The Electricity Sources (Off-grid) Module refers toelectricity sources that the household does not sharewith neighbors, such as its own genset, car (storage)battery, or solar (PV) home system (Figure 4b).

E06: See notes for Question E01 in the ElectricitySources (Grid) Module.

E07: What is electricity used for in your home?

This question is similar to Question E02 in the ElectricitySources (Grid) Module. However, since gensets, car(storage) batteries, and solar home systems (SHSs)usually have sufficient capacity only to power lights,televisions, radios, and other appliances, the list ofpossible end uses is not as extensive. As before, theend-use categories are country-specific and must bereviewed carefully when the module is adapted for usein any particular country.

E08: How much did you pay for your [for eachelectricity source]?

Answers to this question can help establish a household’swillingness to pay for grid electricity, which is useful inplanning electrification programs.

E09: How did you pay for this […]?

Answers to this question can help gauge the effectivenessof efforts to establish credit facilities and SHS dealernetworks.

E10: How much did you pay to operate your [foreach electricity source] in the last 30 days?

This question is relevant for car batteries and gensetsbut not for SHSs (the cost of replacing the battery everyfew years is the greatest running cost).

E11: What is the rating of your [for eachelectricity source]?

For solar panels, the quantity of electricity consumptioncan be directly inferred from system capacity. For dieselgeneration, it can be inferred from system capacity andthe monthly expenditure on diesel (Appendix).

E12: Do you use electricity in your household forhome business purposes?

As part of collecting basic information on householdcharacteristics, LSMS surveys usually contain a questionthat asks if the household has a home business andother questions that probe the nature of a household’shome business. Thus, it may be unnecessary to ask thisquestion in the electricity module. If the household hasa home-based business, electricity consumption willoften be much greater than otherwise expected.

29

Durable Goods (Light Bulbs and Appliances)Module

The purpose of the Durable Goods (Light Bulbs andAppliances) Module is to determine the uses of electricity(i.e., to complement and validate the information obtainedthrough Question E03 above) (Figure 5). The electricitycan be from any source (grid electricity, own or neighbor’sgenset, solar [PV] home system, and car [storage] batteryare the most common sources). Household membersthat are most often in the household are usually thebest-informed respondents. Information on applianceuse provides insights into the household’s priority usesof electricity service and can also be used to calculatemonetary estimates of the benefits of electrification. Thequantity of electricity consumed by the various appliancesis calculated as follows: watt rating of appliance x hoursof use per day x 30 days/1,000 = kWh consumed permonth. The light output from the various classes of lightbulbs is calculated as follows: efficiency rate of lightbulb (lumen/watt) x hours of use per month = lightoutput (klm per month).

Notes on Durable Goods (Light Bulbs andAppliances) Prototype Module

A01: Does your household own [for each class oflight bulb]?

The wattage can usually be read from manufacturermarkings on the light bulb. The listed classes of lightbulbs should be those that are available locally. The listshould be comprehensive, and not limited to the mostcommon classes of light bulbs owned by households.Otherwise, it will not be possible to construct a completepicture of electricity use for lighting.

A02: How many [for each class of light bulb] doyou own?

The enumerator should enter the number of all light bulbsin each class owned by the household. Experience withthis question in specialized energy surveys has obtainedreasonably good results. Poor households usually canreadily answer this question and those below. Respondentsin non-poor households with many light bulbs andappliances experience greater difficulty in accuratelyanswering this question.

A03: How many hours during the last week didyou use the [items]?

The enumerator should enter the sum of hours of usefor all light bulbs in the same class. Thus, if the householdowns two 25-W light bulbs, one of which is used 14hours per week and the other 28, the enumeratorshould enter 42.

A04: How many of the following items does yourhousehold own? [for each appliance]?

The items listed in the module should be adapted toinclude appliances that are locally available. The listshould be comprehensive and not just list the mostcommon appliances owned by households. Local namesof appliances (e.g., type of kerosene lamp or improvedwood stove) should be used. One should note thatthese questions have been integrated into the standardmodule on durable goods for an LSMS survey. The listof items should contain all durable goods for which theLSMS requires data, plus any specific items relevant tothe energy needs (such as energy-efficient mud stoves).For all items for which electricity is used, the two additionalquestions on wattage (A06 and A07, respectively) willbe needed.

A05: How many years ago did you acquire the[item]?

Household preferences for electricity use may be inferredfrom the answers to this question.

A06: According to current prices, what do youthink you could get if you sold it?

It is standard in household surveys to establish the marketvalue of household assets.

30

ID CODE

A01

Doe

s yo

urho

useh

old

own

[item

]?

A02

How

man

y[it

ems]

do

you

own?

Num

ber

A03

How

man

yho

urs

during

the

last

wee

kdi

d yo

u us

eth

e [it

ems]

?

Hou

rs

No.

..2

(>>

A0

4)

A05

How

man

yye

ars

ago

did

you

acqu

ire

the

[item

]?

Num

ber

A06

Acc

ordi

ng t

ocu

rren

t pr

ices

,w

hat

do y

outh

ink

you

coul

d ge

t if

you

sold

it?

Am

ount

A07

Wha

t is

the

aver

age

wat

tage

ra

ting

of t

heap

plia

nce?

Wat

tage

A08

How

man

y ho

urs

duri

ng t

he la

stw

eek

did

you

use

the

[item

s]?

Hou

rs

ID CODE

A04

How

man

y of

the

follo

win

g ite

ms

does

you

r ho

useh

old

own?

Num

ber

10W

Stra

ight

20W

Stra

ight

22W

Circ

ular

32W

Circ

ular

Flu

ore

scen

t tu

bes

12

Wat

t

18

Wat

t

20

Wat

t

Energ

y sa

vin

g b

ulb

s (C

FLs)

25

Wat

t

50

Wat

t

75

Wat

t

Inca

nd

esc

en

t lig

ht

bu

lbs

Elect

rici

ty A

pp

lia

nce

s

Lig

hti

ng

& C

ookin

g A

pp

lian

ces

LIG

HTB

ULB

SA

PP

LIA

NC

ES

Radi

o

TV (B

&W

)

TV (C

olor

)

Vide

o/D

VD

Fan

Elec

tric

iron

Refr

iger

ator

Elec

tric

hot

pl

ate

or s

tove

Dom

estic

wat

er p

ump

Hot

wat

erge

yser

Elec

tric

pum

pfo

r irr

igat

ion

App

lianc

e (S

peci

fy)

Kero

sene

lam

p (w

ick)

Pres

suriz

ed(P

etro

max

) lam

p

Impr

oved

cook

ing

stov

e

LPG

bur

ner

orst

ove

FIG

URE

5.D

urab

le G

oods

(Lig

htbu

lbs

and

App

lianc

es) -

Pro

toty

pe M

odul

e

31

A07: What is the average wattage rating of the[for each appliance]?

Manufacturer markings that include the rating are usuallyaffixed to the appliance. However, it may be difficult toread some markings if they are worn or otherwise illegibleor affixed to an inaccessible part of the appliance. Trainingshould enable enumerators to readily identify the ratings ofthe most common brands or types of appliances, such aslocally available televisions and radios. Information onthese ratings must be collected and provided to theenumerators during training.

A08: How many hours during the last week didyou use the [for each appliance]?

This question establishes the number of hours of applianceuse. In terms of televisions and radios, this question differsfrom those that ask households how many hours theywatch television or listen to the radio.

Electricity and Fuels in the Community Module

The Electricity and Fuels in the Community Module isdesigned to collect data on items common to a clusterof households interviewed; it does not refer to a communityin any administrative sense. These questions provide asnapshot of the choice set of energy sources availableat the community level (Figure 6).

Notes on Electricity and Fuels in the CommunityPrototype Module

C01: Is electricity available in this community?

Answers to this question help interpret individual householdresponses to Question E01 (“Does your household haveelectricity from?”) in the electricity sources module.

C02: What is the main source of electricity in this community? (i.e., Which source of electricityis available to the most households in thiscommunity?)

Depending on local circumstances, the questionnaireshould be adapted to collect information on the nameof the provider, wherever a formal company is involved,to provide a means of verifying information about theprice of services.

C03: On average, how many hours a day iselectricity available from [service provider]?

Questions C03 and C04 establish the reliability ofelectricity supply in the community.

C04: How many times in the last 30 days didelectricity from […] fail for more than 15minutes?

Load shedding (electricity blackouts) may affect onlypart of the community. During design, blackout patternsshould be considered so that the question can beadapted accordingly.

C05: In what year did [electricity source] becomeavailable to this community?

This question makes it possible to reconstruct thehistorical diffusion of electricity services across communitiesin the country, even in the absence of a long series ofhousehold surveys over time. If the electricity sourcesmodule also asks when the household obtained electricityservice, it will be possible to examine the historicaldiffusion of coverage within communities.

C06: How much is the one-time, connection-costfee for [electricity source]?

The cost of gaining access to electricity service varies bycountry location. For any particular community, connectioncost is a function of the community’s distance from theexisting grid, population density, and other demographicfactors. The cost per household may vary between US$400and $1,000. Subsidies can defray the cost for individualhouseholds. Repayment of the hook-up fee may bespread over several months (or in some cases, overseveral harvests), and it may be important to capturethis information.

In addition, it may be useful to establish the start-upcost for other types of energy services, such as LPG (asit is necessary to purchase a cylinder to initiate fuel use)or car batteries (where the battery must be purchasedinitially). One should note that start-up costs do notrefer to the cost of appliances that may be needed touse the fuel (e.g., light bulbs, hurricane lamps, orstoves) since these are covered separately under theappliance inventory in the LSMS survey.

32

ID CODE

C01

Is g

rid

elec

tric

ityav

aila

ble

in t

his

com

mun

ity?

Yes.

..1

No.

..2

(>>

C0

7)

ID CODE

ELEC

TRIC

ITY

SOU

RCE:

Com

plet

eC

09 f

or

each

fa

cilit

y/bu

sine

ss.

C09

Doe

s th

isco

mm

unity

have

the

[...

]lis

ted?

Yes.

..1

No.

..2

(>>

C0

11

)

C10

Wha

t ty

pe

of e

lect

rici

tyso

urce

doe

s th

e […

] us

e?

NO

NE

..........1

GRI

D............2

GEN

SET.

....

...3

OTH

ER........4

Hos

pita

l/he

alth

clin

ic

Post

off

ice

Scho

ol

Com

mun

ityce

nter

Loca

lgo

vern

men

tof

fices

Flou

r m

ill/

grai

n m

illin

g

Bake

ry

Furn

iture

mak

ing

Gro

cery

sho

p

Barb

er s

hop

Oth

er (s

peci

fy)

ID CODE

HO

USE

HO

LDEN

ERG

YSO

URC

ES:

Com

plet

e C

11 -

C14

for

each

ene

rgy

sour

ce.

C11

In y

our

opin

ion,

is

[…

] re

adily

avai

labl

e in

thi

sco

mm

unity

?

C12

In y

our

opin

ion,

is

[…

]ex

pens

ive

in t

his

com

mun

ity?

C13

In y

our

opin

ion,

do

es the

us

e of

[…

]ca

use

heal

th

prob

lem

s?

C14

Is u

sing

[…

]sa

fe in

thi

sco

mm

unity

?

Gri

del

ectr

icity

Fire

woo

d

Cha

rcoa

l &co

al

LPG

Nat

ural

gas

Ker

osen

e

Die

sel

Oth

er(s

peci

fy)

ID CODE

C03

On

aver

age,

how

man

yho

urs

a da

y is

ele

ctric

ityav

aila

ble

from

[…

]

C04

How

man

y tim

es in

the

last

30 d

ays

did

elec

tric

ity f

rom

[…]

fail

for

mor

e th

an 1

5m

inut

es?

C05

In w

hat

year

did

[…]

beco

me

avai

labl

e to

thi

sco

mm

unity

?

C06

How

muc

h is

th

e on

e-tim

eco

nnec

tion

cost

fee

for

[…]?

C07

Is L

PGav

aila

ble

inth

isco

mm

unity

?

Yes.

..1

No.

..2

(>>

C0

9)

C08

How

muc

his

the

on

e-tim

eco

st f

or a

n in

itial

LPG

cylin

der?

Hou

rs/D

ayN

umbe

rYe

arA

mou

ntA

mou

nt

ELEC

TRIC

ITY

SOU

RCES

:C

ompl

ete

C03

- C

06 fo

rth

e so

urce

of

elec

tric

ity t

hat

is u

sed

mos

tin

this

com

mun

ity.

(Res

pons

e fr

om C

02)

Yes.

..1

No.

..2

Yes.

..1

No.

..2

Yes.

..1

No.

..2

Yes.

..1

No.

..2

FIG

URE

6.E

lect

ricity

and

Fue

ls in

the

Com

mun

ity P

roto

type

Mod

ule

C02

Wha

t is

the

mai

n so

urce

of

elec

tric

ity in

thi

s co

mm

unity

?(i.

e. W

hich

sou

rce

of e

lect

rici

ty is

avai

labl

e to

the

mos

t ho

useh

olds

in t

his

com

mun

ity?)

STA

TE U

TILI

TY [

Nam

e]............1

MU

NIC

IPA

L U

TILI

TY [

Nam

e]....2

CO

OPE

RATI

VE

........................3

PRIV

ATE

CO

MPA

NY

[Nam

e]....3

OTH

ER (

Spec

ify).

.....................5

33

C07: Is LPG available in this community?

The module focuses on LPG access because of LPG’soften critical role as a clean cooking fuel.

C08: How much is the one-time cost for an initialLPG cylinder?

This information is important because the cost of theinitial cylinder is often key to the household’s decision to make the transition from kerosene for cooking. Thisquestion will reveal if cost of LPG cylinders variessignificantly between regions in a country, indicatingsome market failure in regions with high costs.

C09: Does this community have the[facility/business] listed?

C10: What type of energy source does the[facility/business] use?

Questions C09 and C10 are important because thedelivery of services may depend on the facilities havingelectricity access.

C11: In your opinion, is […] readily available inthis community?

Questions C11 can reveal local shortages of particularfuels and electricity.

C12: In your opinion, is […] expensive in thiscommunity?

Opinions of which fuels are expensive can differmarkedly by region within a country and can beinformative to policy makers.

C13: In your opinion, does the use of […] causehealth problems?

“Health problems” may imply burns that people sufferfrom cooking fires or use of kerosene lamps andcooking appliances. They may also imply a respondent’sawareness of the health effects of indoor air pollutionfrom use of biomass in cooking fires.

C14: Is using […] safe in this community?

“Safe” may imply household accidents (e.g., fires)associated with using kerosene or wood fires. In certaincountries or regions within a country, LPG use isconsidered unsafe because damaged LPG bottles maybe in circulation.

34

5. CONCLUSION

One goal of these guidelines is to promote the use ofthe modules in forthcoming LSMS surveys and tomonitor which questions work best in different countrycontexts. It is anticipated that countries preparing LSMSsurveys will make use of these guidelines and may seekWorld Bank assistance in evaluating alternative energymodule designs.

It is anticipated that data sets from these enhancedLSMS survey modules can be used to formulate indicatorsthat analysts in the energy sector can use as key decision-making tools. Data monitoring can ensure that welfarechanges are taken into account when new policyinterventions are designed and that equity issues areaddressed. For a program designed to reduce fuelwoodconsumption and indoor air pollution, LSMS data canmonitor changes in consumption after households adoptimproved charcoaling techniques and cook stoves. Inaddition, LSMS data can monitor whether the expectedcost savings from the competitive procurement ofpetroleum products and the introduction of competitionin the marketing of petroleum prices are passed on tohouseholds in the form of lower prices and increasedreliability of supply. In terms of shifts in consumption,LSMS data can monitor changes in household behaviorwhen fuel-price subsidies are removed or electricity-service charges raised, thereby gaining insight into thedynamics of the energy portfolios households maintainto minimize risk. Finally, LSMS data can monitor theextent to which various groups gain access to electricityand the effects that policy tools—such as lifeline rates,credit financing schemes, or improved design standardsand technologies—have on household welfare.

Summing up, LSMS surveys containing better informationon household energy use will provide important insightsinto the role that energy services play in householdwelfare and the policies that would be most effective in accelerating the household transition to use ofmodern fuels.

35

REFERENCES

Ahmed, Kulsum, Yewande Awe, Douglas Barnes,Maureen Cropper, and Masami Kojima. 2005.Environmental Health and Traditional Fuel Use inGuatemala. Directions in Development Series.Washington, DC: World Bank.

Balakrishna, Kalpana, Douglas Barnes, Kyra Naumoff,Kirk Smith, R. Uma, and Reeve Vanneman. 2005.“Measuring Indoor Air Pollution and Lung Function inIndian Field Settings.” Paper prepared for the 2006Population Association of America Annual Meetings, LosAngeles.

Barkat, Abdul, M. Rahman, S. Zaman, A. Podder, S.Halim, N. Ratna, M. Majid, A. Maksud, A. Karim, andS. Islam. 2002. “Economic and Social ImpactEvaluation Study of the Rural Electrification Program inBangladesh.” Report to National Rural ElectricCooperative Association (NRECA) International, Dhaka.

Barnes, Douglas. 1988. Electric Power for Rural Growth:How Electricity Affects Rural Life in DevelopingCountries. Boulder, Colorado: Westview Press.

_______. 1990. “Population Growth, Wood Fuels, andResource Problems in Developing Countries.” InPopulation Growth and Reproduction in Sub-SaharanAfrica: Technical analysis of Fertility and ItsConsequences, eds. G. T. E. Acsadi, G. Johnson-Acsadi, and R. A. Bulatao. Washington, DC: WorldBank.

Barnes, Douglas, and Willem Floor. 1996. “RuralEnergy in Developing Countries: A Challenge forEconomic Development.” Annual Review of Energy andthe Environment 21: 497–530.

Barnes, Douglas, Kerry Krutilla, and William Hyde.2005. The Urban Energy Household Transition: Socialand Environmental Impacts in the Developing World.Washington, DC: Resources for the Future Press.

Barnes, Douglas, and Liu Qian. 1992. “Urban InterfuelSubstitution, Energy Use, and Equity in DevelopingCountries.” In International Issues in Energy Policy,Development, and Economics, eds. James P. Dorian andFereidun Fesharaki, 163–81. Boulder, Colorado:Westview Press.

Barnes, Douglas, Keith Openshaw, Kirk R. Smith, andRobert van der Plas. 1993. “The Design and Diffusionof Improved Cooking Stoves.” The World Bank ResearchObserver 8(2): 114–41.

Bennett, J., and R. Blamey. 2001. The Choice ModelingApproach to Environmental Valuation. Cheltenham, UK:Edward Elgar.

Boardman, A., D. Greenburg, A. Vining, and D.Weimer. 2001. Cost-benefit Analysis: Concepts andPractice. Toronto: Prentice-Hall.

Cabraal, Anil, Douglas Barnes, and Sachin Agarwal.2005. “Productive Uses of Energy for Rural Development.”Annual Review of Environment and Resources.

Deaton, A. 1997. The Analysis of Household Surveys: A Microeconometric Approach to Development Policy.Baltimore, Maryland: The Johns Hopkins University Press.

Dewees, P. A. 1989. The Woodfuel Crisis Reconsidered:Observations on the Dynamics and Abundance andScarcity. World Development 17: 1159–72.

Denton, Frank H. 1979. Lighting Up the Countryside:The Story of Electric Cooperatives in the Philippines.Development Academy of the Philippines. Manila:Academy Press.

Freeman, R. 2003. The Measurement of Environmentaland Resource Values: Theory and Methods. 2nd ed.Washington, DC: Resources for the Future Press.

Gordon, Adele. 1997. “Facilitating Education in RuralAreas of South Africa: The Role of Electricity and OtherSources of Energy.” University of Cape Town EDRCReport Series. Cape Town: Energy and DevelopmentResearch Centre.

Grosh, M., and P. Glewwe, eds. 2000. DesigningHousehold Survey Questionnaires for DevelopingCountries. 3 vols. Washington, DC: World Bank.

Grosh, M., and J. Munoz. 1996. “A Manual forPlanning and Implementing the Livings StandardsMeasurement Study Survey.” LSMS Working Paper No.126. Washington, DC: World Bank.

36

Haab, T. C., and K. E. McConnell. 2002. ValuingEnvironmental and Natural Resources: The Econometricsof Non-market Valuation. Cheltenham, UK: Edward Elgar.

Harvey, A., and M. Taylor. 2000. “Time Use.” InDesigning Household Survey Questionnaires forDeveloping Countries, vol. 2, eds. M. Grosh and P.Glewwe, 249–72. Washington, DC: World Bank.

Herrin, Alexandro. 1979. “Rural Electrification andFertility Change in the Southern Philippines.” Populationand Development Review 5(1): 61–87.

Hosier, Richard H., and W. Kipondya. 1993. “UrbanHousehold Energy Use in Tanzania: Prices, Substitutesand Poverty.” Energy Policy 21(5): 454–73.

IAEA. 2005. Energy Indicators for SustainableDevelopment: Guidelines and Methodologies. Vienna:International Atomic Energy Agency.

Kammen, Daniel. 2001. “Quantifying the Effects ofExposure to Indoor Air Pollution from BiomassCombustion on Acute Respiratory Infections inDeveloping Countries.” Environmental HealthPerspectives 109(5): 481–8.

Khandker, Shahidur. 1996. “Education Achievements andSchool Efficiency in Rural Bangladesh.” World BankDiscussion Paper No. 319. Washington, DC: World Bank.

Khandker, Shahidur, Victor Lavy, and Deon Filmer.1994. “Schooling and Cognitive Achievements ofChildren in Morocco: Can the Government ImproveOutcomes?” World Bank Discussion Paper No. 264.Washington, DC: World Bank.

Komives, K., D. Whittington, and X. Wu. 2002.“Infrastructure Coverage and the Poor: A GlobalPerspective.” In Infrastructure for Poor People: Public Policyfor Private Provision, eds. Penelope J. Brook and TimothyC. Irwin, 77–124. Washington, DC: World Bank.

Kopp, Raymond, Norbert Schwarz, and Werner W.Pommerehne. 1997. Determining the Value of Non-marketed Goods: Economics, Psychological and PolicyRelevant Aspects of Contingent Valuation Methods.London: Springer.

Kulkarni, V., and D. Barnes. 2005. The Impact of RuralElectrification on Education Participation in Nicaragua.ESMAP Draft Paper (processed), World Bank,Washington, DC.

Leach, Gerald. 1986. Household Energy in South Asia.London: International Institute for Environment andDevelopment.

Millennium Project. 2005. Energy Services for theMillennium Development Goals. New York. UNMillennium Project.

Mishra, V. 2003. “Indoor Air Pollution from BiomassCombustion and Acute Respiratory Illness in PreschoolAge Children in Zimbabwe.” International Journal ofEpidemiology 32: 847–53.

Mitchell, Robert, and Richard T. Carson. 1988. UsingSurveys to Value Public Goods: The ContingentValuation Method. Washington, DC: Resources for theFuture Press.

Nieuwenhout, F., P. Van de Rijt, and E. Wiggelinkhuizen.1998. “Rural Lighting Services.” Paper prepared for theWorld Bank, Netherlands Energy Research Foundation,Petten.

Ranganathan, V., and T. V. Ramanayya. 1998. “Long-term Impact of Rural Electrification: A Study in UP andMP.” Economic and Political Weekly 33: 3181–84.

Sathaye, Jayant, and Stephen Tyler. 1991. “Transition inHousehold Energy Use in Urban China, India, thePhilippines, Thailand, and Hong Kong.” Annual Reviewof Energy and Environment 16: 295–335.

Saunders, John, J. Michael Davis, Galen C. Moses, andJames E. Ross. 1975. Rural Electrification andDevelopment: Social and Economic Impact in CostaRica and Colombia. Boulder, Colorado: Westview Press.

Smith, Kirk, and Sumi Mehta. 2003. “The Burden ofDisease from Indoor Air Pollution in DevelopingCountries: Comparison of Estimates.” InternationalJournal of Hygiene and Environmental Health 206:279–89.

37

Smith, Kirk, Jonathan M. Samet, Isabelle Romeiu, andNigel Bruce. 2000. “Indoor Air Pollution in DevelopingCountries and Acute Lower Respiratory Infections inChildren.” Thorax 55: 518–32.

USAID. 1983. “Power to the People: Rural ElectrificationSector Summary Report.” Program Evaluation ReportNo. 11. Washington, DC: United States Agency forInternational Development.

Wallmo, K., and S. Jacobson. 1998. “A Social andEnvironmental Evaluation of Fuel-efficient Cook-stovesand Conservation in Uganda.” EnvironmentalConservation 25(2): 99–108.

Whittington, D. 1998. “Administering ContingentValuation Surveys in Developing Countries.” WorldDevelopment 26(1): 21–30.

_______. 2002. “Improving the Performance ofContingent Valuation Studies in Developing Countries.Environmental and Resource Economics 22(1-2):323–67.

WHO. 2002. The World Health Report 2002. ReducingRisks, Promoting Healthy Life. Geneva: World HealthOrganization.

World Bank. 1994. Rural Electrification in Asia: AReview of Bank Experience. Summary of Sector StudyNo. 13291. Operations Evaluation Department.Washington, DC: World Bank.

_______. 1996. Rural Energy and Development: ImprovingEnergy Supplies for Two Billion People. Development inPractice Series. Washington, DC: World Bank.

_______. 1999. “Poverty and Social Development inPeru, 1994–1997.” World Bank Country Study,Washington, DC.

_______. 2002. “Rural Electrification and Developmentin the Philippines: Measuring the Social and EconomicBenefits.” Energy Sector Management AssistanceProgramme (ESMAP) Report No. 255/02. Washington,DC: World Bank.

_______. 2004. “The Impact of Energy on Women’sLives in Rural India.” Energy Sector ManagementAssistance Programme (ESMAP) Report No. 276/04.Washington, DC: World Bank.

_______. 2005. “Household Energy Supply and Use inYemen” (ESMAP) Report No. 315/05. Washington, DC:World Bank.

APPENDIX

NOTES ON ENERGY APPLIANCES ANDCONVERSION FACTORS

Definitions

Car Battery. Commonly used in areas of developingcountries without an electricity grid (motorcycle or lorrybatteries are also used) to provide household electricityfor lighting and to power televisions and radios.Recharging often occurs at battery-charging stationsusing grid electricity in the nearest grid-connected town.The amount of electricity per recharge depends on thesize of the battery, depth of discharge, and efficiency ofthe charger. Wind generators, SHSs, or diesel generatorscan also be used to recharge batteries. The most commonbattery rating is ampere-hours; this unit of measurementfor battery electrical storage capacity is obtained bymultiplying a current flow in amperes by the time in hoursof discharge. A battery that delivers 5 amperes for 20hours delivers 100 ampere-hours. Since watts equalvolts times amps, a 12-volt battery with a 400 amp-hourcharge can deliver 12 x 400 or 4,800 watts (W) for anhour or 4.8 kWh.

Horsepower. Unit for power or for measuring the rate ofwork; 1 horsepower equals 746 watts. Diesel generatorsused by households to generate electricity and dieselpumps used to pump water are often rated in horsepower.

Joule (J). Standard international (SI) unit of energy andheat; 1 J equals 0.2389 calories.

Kilowatt hour (kWh). Work equals energy per time period;one kWh is equivalent to 1,000 watts (joules per second)over a 1-hour period; thus, 1 kWh equals 3.6 x 106 J.

Lumen (lm). Quantity of light (luminous flux) emitted bya lamp. The quantity of light radiated or received for aperiod of 1 hour is a lumen-hour (lm hr). A 60-Wattincandescent lamp and a 12-Watt compact fluorescentlamp emit approximately the same amount of lumens(720 lm).

Peak Watt (Wp). Power output that a photovoltaic (PV)module produces at Standard Test Conditions (STC) of amodule operating temperature of 25°C with an irradianceof 1,000 W per square meter (usually attained at noon).

Power. Rate at which work is done (or rate at which heatis released or energy converted). A 100-W light bulb, forexample, draws 100 J of electrical energy per second.

Solar Home System (SHS). System whose maincomponents typically include a solar (PV) module,inverter, battery, and charge controller (sometimesknown as a regulator). The PV module may range insize from 20 to 100 peak watts (Wp). SHSs with 400Wp or more are necessary to operate a refrigerator andare not common in developing countries. A 50-Wpsystem can supply lighting and can power a smalltelevision or radio for several hours per day, and a 12-Wp system can operate a lamp or a radio forseveral hours per day.

Watt (W). Standard unit of measurement of electricalpower; 1 W equals 1 ampere of current flowing at 1-volt. Electric appliances (e.g., radios, televisions,electric heating coils, and light bulbs) are rated in watts; a 60-W light bulb or a 60-W black-and-whitetelevision set operated 5 hours per day uses 9 kWh of electricity per month (60 W x 5 hours x 30 days).

Conversion Equivalents

Units of mass 1 kilogram (kg) = 2.2046 pounds (lb)1 pound (lb) = 0.454 kilograms (kg)

Units of volume 1 liter (l) = 0.2642 gallons (gal) (U.S.) = 0.22 gal (UKor imperial)1 gallon (gal) (U.S.) = 0.8327 gal (UK) = 3.78528liters (l)

Units of energy1 megajoule (MJ) = 238.84 kilocalories (kcal) =0.2777 kilowatt hours (kWh)1 kilowatt hour (kWh) = 3.6 megajoules (MJ) = 860kilocalories (kcal)

Cooking Efficiency

The term useful energy refers to work harnessed for thepurpose of which the fuel is consumed. In the case ofcooking, useful energy is the heat actually used forheating the food (transmitted to the food-cooking process).The annual amount of energy required for cooking varieswith the type of food, fuel, and stove used and the

38

39

specific cooking practices of a household (MillenniumProject 2005). Diet is also a factor in energy needs.Cooking hard staples such as maize and potatoes requiresmore energy than cooking fish. The annual energyrequirements for a family of five is about 5 gigajoules ofuseful energy (i.e. energy "into the pot"). Typically,households use a combination of fuels. Table A-1 provides a comparison of typical efficiencies ofdifferent fuels in cooking.

Household Lighting Appliances

In developing countries, households in areas not served by grid electricity use a great variety of lightingappliances. Carbide (acetylene) lamps are simple lampsthat produce and burn acetylene gas through the reactionof calcium carbide with water. Kerosene wick lamps aresimple lamps consisting of a small receptacle forcontaining the kerosene and a wick (usually made ofcotton). The lower half of the wick is dipped, absorbingthe kerosene, while the top part extends out of the topof the receptacle. Instead of kerosene, diesel or animal

fat are also used. A kerosene or gasoline mantle lamphas a fuel tank and a small pump to pressurize thekerosene. A burner sets atop the lamp, directly underneathof which is the mantle, which incandesces when heatedby the gas flame. The most widely used gas lamps usecylinders of propane, butane, or a mixture thereof (LPG).

Table A-2 compares characteristic values of variouslamps. Performance varies considerably, depending onthe luminous conditions in rooms and such behavioralpatterns as adjusting the lamp setting, cleaning glasscovers, replacing mantles, and polishing reflectors. Inaddition, characteristics of the same type of lamp (e.g., kerosene lamps) vary by lamp size.

Table A-3 compares characteristic values of variousbatteries.

Table A-1. Typical Efficiencies at the Final Consumption Stage of Cooking

LPG

Natural gas

Kerosene (pressure)

Kerosene (wick)

Biogas (60% methane)

Charcoal (efficient)

Charcoal (traditional)

Bituminous coal

Fuelwood (efficient), 15% moisture

Fuelwood (traditional), 15% moisture

Crop residue (straw, leaves, and grass),

5% moisture

Dung, 15% moisture

Sources: “Energy Statistics: A Manual for Developing Countries,” Series F, No. 56, United Nations,

New York, U.S. and authors estimates.

45.5

38 MJ/M3

43.0

43.0

22.8 MJ/M3

30.0

30.0

22.5

16.0

16.0

13.5

14.5

60

60

55

35

60

30

20

25

25

15

12

12

27.3

23.6

15.1

9.0

6.0

5.6

4.0

2.4

1.6

1.7

180

219 M3

210

330

365 M3

550

830

880

1250

2000

3000

2900

ENERGYCONTENT

(MJ PER KG)FUEL SOURCE

CONVERSION EFFICIENCY

(%)

APPROXIMATEQUANTITY OF FUEL

NECESSARY TOPROVIDE 5

GIGAJOULES OF USEFUL ENERGY

FOR COOKING(KILOGRAMS)

USEFUL ENERGYAT FINAL

CONSUMPTIONSTAGE OFCOOKING (MJ PER KG)

40

Table A-2. Comparison of Non-electric and Electric Lamps for Domestic Lighting in Developing Countries

NON-ELECTRIC

Paraffin candle

Kerosene wick

Kerosene hurricane

Kerosene pressure

INCANDESCENT (WATTS)

25

40

50

60

100

FLORESCENT (WATTS)

10

20

40

COMPACT FLORESCENT

Philips lamp (15-W)

Philips lamp (9-W)

Osram sol lamp (6.14-W)

Source: Nieuwenhout, Van de Rijt, and Wiggelinkhuizen (1998).

11.8

11.4

32

2,040

230

430

580

730

1,280

600

1,200

1,613

894

369

240

2.33

1.15

1.92

17.53

109.12

127.50

137.58

144.30

151.80

711.63

711.63

478.27

706.88

486.28

463.60

.20

.10

.16

1.48

9.20

10.75

11.60

12.17

12.80

60.00

60.00

40.33

59.60

41.00

39.09

LIGHTING OUTPUT(LUMENS)LIGHTING TYPE

klmh PERKgOE

klmh PERkWh

OUTPUT BASED ON LUMENS PER WATT

Table A-3. Battery Electricity

High Quality D Cell (1.5 v)

Poor Quality D Cell (1.5 v)

Car battery (12 v)

Note: A battery can only discharge to 80 percent of capacity.

7.5

3.5

60

11.25

5.25

720

AMP HOURS (Ah)BATTERY TYPE WATT HOURS

RATINGSTORAGECAPACITY

The World Bank1818 H Street N.W.Washington, D.C. 20433USA

Energy and MiningSector Board

THE WORLD BANKGROUP