handbook for integrating poverty impact assessment · handbook for integrating poverty impact...

Handbook

for

Integrating PovertyImpact Assessment

in the

Economic Analysisof Projects

Economics and Development Resource Center

(EDRC)

July 2001

Handbook Poverty Impact Prelims.p65 5/17/02, 10:27 AM1

© Asian Development Bank 2001

All rights reserved

ISBN 971-561-285-7Publication Stock No. 020100

Published by the Asian Development BankP.O. Box 789, 0980 Manila, Philippines


Foreword

This Handbook is the second output of EDRC’s study on integrating

poverty impact assessment in the economic analysis of projects. ManabuFujimura led the study under the overall supervision and guidance of

David Edwards, then Assistant Chief Economist. John Weiss (Staff Consultant)provided technical inputs and advice throughout the study. Nigel Raynercontributed Appendix 2. Olivier Dupriez and Christopher Edmonds contributedAppendix 3. Tumurdavaa Bayarsaihan contributed Case 2 in Appendix 6. Case4 in Appendix 6 benefited from the poverty analysis included in NanakKakwani’s (Staff Consultant) work. Marcelia Garcia and Ma. Virginita Capulongprovided assistance for literature search. Anneli Lagman did a consistencycheck on the technical presentation of the final draft. Regina Sibal contributedthe glossary of terms and was the sole undertaker of word processing andgraphics. Lily Bernal did the proofreading.

Preparation of the Handbook underwent three stages. In the first stage,a draft issues paper was prepared and received interdepartmental comments,which resulted in the EDRC paper “Integration of Poverty Impact in ProjectEconomic Analysis: Issues in Theory and Practice.” In the second stage, apreliminary draft of the Handbook was presented in an EDRC seminar, whichenhanced interactions with operational departments. Sean O’Sullivan and BoLin shared their work on ADB’s power projects. Stephen Curry, FrancescaAgnello, Jeffrey Miller, and Hiromi Sakurai shared their recent work on ADB’stransport projects. In the third stage, the final draft of the Handbook wascirculated and received detailed comments. The Handbook benefitedparticularly from comments and suggestions provided by Stephen Curry, RitaNangia, Etienne Van de Walle, Adrian Ruthenberg, Mandar Jayawant, TyrellDuncan, Lourdes Adriano, Sultan Hafeez Rahman, Alfredo Perdiguero, andCindy Houser. Arvind Panagariya, Chief Economist, and Xianbin Yao, AssistantChief Economist, provided guidance in the final stage.

The Handbook will serve as a reference material in preparing projectsunder ADB’s renewed mandate of poverty reduction and augments the currentpractice of project economic analysis. It illustrates applications of Appendixes25 (Distribution Analysis) and 26 (Poverty Impact Analysis) of ADB’s Guidelines

for the Economic Analysis of Projects (1997). It is primarily intended for theguidance of ADB staff and officials of its developing member countries, but itmay also be of interest to others involved in development assistance.


iv HANDBOOK FOR INTEGRATING POVERTY IMPACT ASSESSMENT IN THE ECONOMIC ANALYSIS OF PROJECTS

Abbreviations and Acronyms

ADB Asian Development BankCBO community-based organizationCPI consumer price indexCWIQ Core Welfare Indicators QuestionnaireDALY disability adjusted life yearsDHS demographic and health surveyDMC developing member countryEDRC Economics and Development Resource CenterEIRR economic internal rate of returnFIRR financial internal rate of returnGDP gross domestic productHIES household income and expenditure surveyILO International Labuor OrganisationISA initial social assessmentLao PDR Lao People’s Democratic RepublicLECS Lao Expenditure and Consumption SurveyLSMS living standards measurement surveyMICS multiple indicator cluster surveyMOLISA Ministry of Labor, Invalids and Social Affairs, Viet NamNGO nongovernment organizationNPV net present valueOPEC Organization of Petroleum Exporting CountriesPIA poverty impact assessmentPIR poverty impact ratioPPTA project/program preparatory technical assistanceRMS rapid monitoring surveyRRP report and recommendation of the PresidentRSA rapid social assessmentSCF standard conversion factorSERF shadow exchange rate factorSWRF shadow wage rate factorTA technical assistanceTLSS Tajikistan Living Standards SurveyVOC vehicle operating cost

Note: In this Handbook, “$” refers to US dollars.


1 Introduction v

Contents

PPPPPageageageageage

FFFFForororororeworeworeworeworeworddddd iiiiiiiiiiiiiii

AbbreviationsAbbreviationsAbbreviationsAbbreviationsAbbreviations i vi vi vi vi v

11111 INTRODUCTIONINTRODUCTIONINTRODUCTIONINTRODUCTIONINTRODUCTION 11111

22222 PRO-POOR CONTEXT CHECK FOR OTHERPRO-POOR CONTEXT CHECK FOR OTHERPRO-POOR CONTEXT CHECK FOR OTHERPRO-POOR CONTEXT CHECK FOR OTHERPRO-POOR CONTEXT CHECK FOR OTHERINTERINTERINTERINTERINTERVENTION PROJECTSVENTION PROJECTSVENTION PROJECTSVENTION PROJECTSVENTION PROJECTS 55555

33333 APPROAPPROAPPROAPPROAPPROAAAAACH TO POCH TO POCH TO POCH TO POCH TO POVERVERVERVERVERTY IMPTY IMPTY IMPTY IMPTY IMPAAAAACT ANALCT ANALCT ANALCT ANALCT ANALYSIS FORYSIS FORYSIS FORYSIS FORYSIS FORPOPOPOPOPOVERVERVERVERVERTY INTERTY INTERTY INTERTY INTERTY INTERVENTION PROJECTSVENTION PROJECTSVENTION PROJECTSVENTION PROJECTSVENTION PROJECTS 99999

Pre-PPTA Stage 12PPTA Stage 13Steps for Poverty Impact Analysis 14

44444 DDDDDAAAAATTTTTA REQUIREMENTSA REQUIREMENTSA REQUIREMENTSA REQUIREMENTSA REQUIREMENTS 1717171717

Benefit Incidence 18Government Net Benefit and Effects on the Rest of the Economy 20

55555 ADADADADADVVVVVANTANTANTANTANTAAAAAGES AND LIMITGES AND LIMITGES AND LIMITGES AND LIMITGES AND LIMITAAAAATIONS OF THE ANALTIONS OF THE ANALTIONS OF THE ANALTIONS OF THE ANALTIONS OF THE ANALYSISYSISYSISYSISYSIS 2323232323

Improving Project Quality at Entry 23Complementing the Inadequacy of Headcount Approach 24Variety of Projects 26Caution on Interpretation of PIR index 27Risk for the Poor 27

66666 LINKLINKLINKLINKLINKAAAAAGE TO POLICGE TO POLICGE TO POLICGE TO POLICGE TO POLICYYYYY-BASED LENDING-BASED LENDING-BASED LENDING-BASED LENDING-BASED LENDING 2929292929


vi HANDBOOK FOR INTEGRATING POVERTY IMPACT ASSESSMENT IN THE ECONOMIC ANALYSIS OF PROJECTS

APPENDIXESAPPENDIXESAPPENDIXESAPPENDIXESAPPENDIXES 3131313131

1. Distribution and Poverty Impact Analysis (reproduced from ADB Guidelines for the Economic Analysis of Projects) 33

2. Benchmark Criteria for Good Practice Project Preparation 433. Application of Existing Primary Survey Data in Poverty Impact

Analysis at the Project Level 454. Case Illustrations of Distribution Analysis 67

Case 1: Mongolia – Energy Conservation Project 67Case 2: Bangladesh – Jamuna Bridge Project 74

5. Incorporating Project Financing in Distribution and Poverty Impact Analysis 83

6. Case Illustrations of Poverty Impact Analysis 89Case 1: Philippines – Transmission Interconnection and Reinforcement Project 90Case 2: Viet Nam – Second Red River Basin Water Resource Project 98Case 3: Sri Lanka – Coastal Resource Management Project 109Case 4: Lao PDR – Primary Health Care Expansion Project 121Case 5: Tajikistan – Road Rehabilitation Project 128

7. Approximation of Income Share of the Poor 1378. Sample Terms of Reference and Resource Requirements 141

GLOSSARGLOSSARGLOSSARGLOSSARGLOSSARYYYYY 145145145145145

REFERENCESREFERENCESREFERENCESREFERENCESREFERENCES 151151151151151


DB's adoption of poverty reduction as its overarching objective meansthat all its interventions must be prepared with poverty impact as theprimary focus in one form or another. ADB's Poverty Reduction Strategy

(October 1999) appropriately sets out the three pillars of poverty reduction:pro-poor sustainable economic growth, social development, and goodgovernance. However, it should be recognized that existing economic theoryand cross-country empirical evidence are not likely to be a satisfactory guidein terms of informing policy options for individual developing member countries(DMCs) of ADB toward maximum poverty reduction. Pro-poor growth that isfavored by most international agencies including ADB can be interpreted as

1Introduction

Untitled-1 5/17/02, 10:25 AM1

2 HANDBOOK FOR INTEGRATING POVERTY IMPACT ASSESSMENT IN THE ECONOMIC ANALYSIS OF PROJECTS

an outcome from a mix of interventions, but identification of such mix is acomplex undertaking in practice. A challenging operational task for ADB inthe short to medium term is to find a good balance between the twointervention categories adopted: Poverty Interventions (of which Core PovertyInterventions are a subset) and the other interventions (Loan ClassificationSystem, November 2000). (This Handbook will refer to the second group simplyas Other Interventions throughout.) It can be broadly considered that PovertyInterventions will be targeted projects while Other Interventions will be non-targeted projects. ADB's Poverty Reduction Strategy (paragraphs 55-56)recognizes that "it is often difficult in the case of individual countries to decidehow much emphasis to place on poverty interventions and how much ongrowth-oriented investments. Where past performance in poverty reductionhas been weak or inequality is rising, the emphasis will be on governanceand social development. In countries where essential reforms have beenundertaken or are under way, growth-oriented investments will reducepoverty…In each country, the mix and nature of projects will be shaped bythe poverty analysis (to be carried out for each DMC)."

As an intermediate step before country-specific poverty analysis canbetter inform the appropriate intervention mix for poverty reduction, one couldthink of some country grouping with which to broadly guide ADB's countryprogramming. For example, DMCs with demonstrated growth performancebut potential or alarming pace of increasing inequality and poor record ofpoverty reduction might need more poverty interventions than growth-oriented projects. DMCs with relatively equitable institutional setup but weakgrowth performance might need more growth-oriented projects than povertyinterventions. DMCs with less than mediocre record on both growth andequitable institution and governance in general, including the economies intransition, might need governance reform interventions before anything elseas a prerequisite to the possibility of poverty-reducing growth process. However,truly useful information must come from country-specific studies. Some earliercountry-specific poverty studies undertaken by ADB include Quibria (1993,1994) and Pernia (1994) covering Bangladesh, India, Sri Lanka, Indonesia,Republic of Korea, Philippines, and Thailand. More knowledge at countryprogramming level must be accumulated over time to inform the exercise ofcountry strategy and program through economic and sector work (ESW).

This Handbook is intended as a reference material to assist ADB projectpreparation in light of the economic analysis under the mandate of povertyreduction. It provides workable recommendations for augmenting the currentpractice. In accordance with the twofold loan classification, two separate

Untitled-1 5/17/02, 10:25 AM2

1 Introduction 3

treatments are recommended. The main part of the Handbook (Chapters 3-5) isdevoted to the treatment of Poverty Intervention projects. It can be consideredas a detailed application of the ADB Guidelines for the Economic Analysis of

Projects (reproduced in Appendix 1). Treatment of Other Intervention projectsis discussed in Chapter 2. It should be noted that due to the specific focus onthe poverty impact of projects, this Handbook does not cover directly otherequally important aspects of project appraisal such as financial sustainability,which needs continued attention. For general guidance on overall projectanalysis, readers are referred to the Guidelines for the Economic Analysis of

Projects (ADB 1997).

Untitled-1 5/17/02, 10:25 AM3

arious policy and institutional contexts of project investments must beinformed by country- and sector-level knowledge based onmacroeconomic performance record, public expenditure review,

governance review, etc. In light of ADB’s poverty reduction objective, theanalyst should first check whether the project-induced growth effects willlead to poverty reduction, e.g., with at least one-to-one relationship. If suchenvironment can be confirmed, a reasonable strategy would be to look forinvestments with maximum growth impact, provided that care is taken toensure obvious negative externalities will not arise. A presumed pro-poorrationale for Other Intervention (primarily nontargeted) projects must be thattheir policy and institutional contexts are such that growth is the most effective

2Pro-PoorContext Check forOther Intervention

Projects

Untitled-1 5/17/02, 10:25 AM5

channel for poverty reduction. If, for example, social service delivery (e.g.,primary education, primary health expenditure) and transfer mechanisms(social safety net) in a project's target area(s) is demonstrated to be pro-poor,there is less need to design the project in a pro-poor way especially wheresuch design tends to narrow the room for growth impact, which in turn narrowthe scope for government social expenditure targeted for the poor. Undersuch environment, selection of projects with maximum net present value (NPV)with a 12 percent cutoff economic internal rate of return (EIRR) can continueto be the primary economic criteria.

In determining whether a project in question fits in such environment,the analyst may proceed with a two-tier context check. The first-tier checkwill be at economy-wide level, and the second-tier check will be atgeographically disaggregated local level. Much of the information requiredin the first-tier check should be covered in country-level studies at CSP level.Such information can be broadly categorized into the following:

Macroeconomic context: It is generally accepted that inflation has anadverse impact on the poor involved in market economy as it works as aregressive tax due to loss in what little purchasing power they have. Monetaryauthorities cannot necessarily control the size and nature of the impact ofmacroeconomic policy changes on the poor. Reconciliation of short-run andlong-run effects on the poor will not be straightforward. It may be best toassign some macro policies to a limited policy goal such as controlled inflationand fiscal stabilization. The analyst should at least check whethermacroeconomic management is broadly sound in this context.

Public expenditure and tax incidence on the poor: The public spendingincidence is often used as a shortcut to welfare measurement of public services.While this is certainly an approximation of true benefit incidence, where dataor resources are not available for full-fledged behavioral analysis, conventionalbenefit incidence results would be informative in judging the first-cutredistributive environment of public spending. Methodological problems withthis approach pointed out, for example, in van de Walle (1996), may not bea serious concern if its use is limited to checking a broad public spendingtendency instead of using it to guide actual public spending decisions. Taxincidence analysis is the other side of the public expenditure and can beapplied in combination with public spending analysis. [See, for example, thepublic spending chapter in the World Bank's Poverty Reduction Strategy Source-

book (World Bank 2000d.)]Institutional/governance context: The way rules of the game are written

and effectively enforced determines the overall parameters within which


Untitled-1 5/17/02, 10:25 AM6

economic activities bring about growth and equity outcomes. While manyagencies have developed governance indicators and checklists, they couldbe broadly categorized into two groups in terms of poverty impact: poverty-neutral indicators and proactive indicators. The former group could includeaccountable and contestable government, credibility of budget process, efficacyof legal institutions, factor market efficiency, anticorruption legal frameworkand enforcement, etc. The latter group could include social stratification/classsystem, voice of the poor, asset distribution, credibility of social andenvironmental protection; social safety net system, etc.

In line with the stipulation in the Advisory Notes on Poverty Analysis

(SPD April 2000), it is desirable that as much as possible of the followinginformation be covered in country poverty analysis:

(i) macroeconomic stability – inflation rate and its trend; exchangerate depreciation trend and its impact on rural and urban poor;

(ii) asset distribution (especially landownership profile), preferablywith geographical breakdown, and its implication on the poor'scapability to participate in market activities;

(iii) labor market condition: market competitiveness; location anddensity of labor-intensive industries and small and mediuminterprises, etc., and their implication for employment of the poor;

(iv) public spending and tax incidence, preferably with geographicalbreakdown (ideally covered by public expenditure review or social

expenditure review);(v) government antipoverty programs: magnitude, location, sectors,

and types of antipoverty programs;(vi) social safety net availability for the poor, preferably with

geographical breakdown (e.g., World Bank's social sector reviews,International Labour Organisation’s social protection expenditure

reviews);(vii) effectiveness of the regulatory regimes and implication on the

poor: e.g., existence and enforcement status of anticorruption laws(ideally covered in country governance review);

(viii) indicators of risk-coping capacity of the poor, preferably withgeographical breakdown: social indicators such as educationlevels and health status;

2 Pro-Poor Context Check for Other Intervention Projects 7

Untitled-1 5/17/02, 10:25 AM7


(ix) support of the civil society and private sector: existence ofnongovernment and community-based organizations etc., thatrepresent and promote the interests of the poor, with geographicalbreakdown; and

(x) ongoing and planned external assistance: existence of targetedpoverty reduction initiatives with geographical breakdown.

In the second-tier context check, where information is inadequate atdisaggregate levels, the project analyst is responsible for collecting andcomplementing the information specific to the local situation and examiningwhether the project environment is conducive to the poor's access to theservices produced by the project.

It is expected that these information will be increasingly availablethrough ESW. In the meantime, the project analyst should rely on as muchexisting evidence as meaningful to make a sensible judgement on the policyand institutional context of the project at hand. Existing ADB documents forprogram loans or sector development program loans with poverty impactassessment (PIA) matrixes may include some clues to inform the projectenvironment. Where a series of ADB projects have been implemented withina sector or subsector, evaluation reports may yield useful information ongovernance performance over time and help examine its implication on thepoor.

A rigorous economic calculation of distribution and poverty impactanalysis, which follows in the next section, need not be essential for OtherIntervention projects. However, project economists and consultants areencouraged to undertake such analyses wherever possible and meaningfulfor many reasons (see discussion in Chapter 5). For benchmarks for goodproject preparation with a poverty perspective, see Appendix 2.

Untitled-1 5/17/02, 10:25 AM8

1 Introduction 9

or Poverty Intervention (including Core Poverty Intervention) projects,they should be subjected to an analysis specific to the poor beneficiariesin addition to the conventional efficiency analysis represented by the EIRR

or aggregate NPV indicator. It would be ideal if a consistent yardstick couldbe applied to rank all interventions. Economic logic requires that such yardstickbe based on the efficiency of delivering poverty reduction combined with aweighting system that incorporates contributions to all other relevantobjectives, of which growth will be the most important. However, the practiceof necessity must fall short of this theoretical ideal. Apart from the intrinsicproblem of weighting different objectives there remains the methodologicaldifficulty of comparing interventions among (i) those with monetized net

3Approach toPoverty Impact

Analysis for PovertyIntervention Projects

Untitled-1 5/17/02, 10:25 AM9

benefits, (ii) those with quantifiable but nonmonetized net benefits, and (iii)those with only beneficiary headcounts. Due to the diverse nature of theinterventions at ADB's disposal it is impractical to attempt to develop acomprehensive methodology or a single criterion. Operational poverty impactanalysis cannot be expected to be any more standardized than the existingpractice of efficiency-based analysis. In addition there is a range of interventionsincluding policy-based lending and institutional strengthening that are noteasily amenable to rigorous quantitative methods, which is addressed in aseparate study undertaken by the Economics and Development ResourceCenter (EDRC).

As early as in the late 1970s, ADB recognized the importance of bringingin beneficiary identification and distributional impacts in project analysis (ADB1978). There has been a recent attempt to promote a rigorous approach toestimating poverty reduction impact of agricultural projects (Ali 1990). Thecurrent ADB Guidelines (Appendix 1) introduces the methodology. But thepractice has not taken off on a regular basis until very recently. This Handbookexpands upon the Guidelines. It sets out briefly the main steps involved inincorporating a poverty dimension into the economic analysis of projects,identifies some of the main technical difficulties, and suggests some rules ofthumb as a practical means by which they can be overcome. The appendixesdiscuss more specialist issues and illustrate the approach by an analysis ofcase studies drawn from recent appraisals of ADB projects. Readers interestedin theoretical aspects and experiences of other international agencies in thisarea are referred to Fujimura and Weiss (2000).

Economic analysis of projects uses a money-metric measure—i.e., asfar as possible all project effects should be expressed in terms of economicbenefits and costs expressed in monetary units. Hence it is logical that for thepurposes of project analysis, poverty should be defined also in income/expenditure terms as opposed to headcount terms. This requires identifyinga poverty line level of income/expenditure and defining all those who fallbelow this line as the poor and those whose income/expenditure is above itas the nonpoor. For the purposes of defining poverty for ADB appraisals, it isrecommended that wherever possible the poverty line used should be a nationalone agreed between ADB and the DMC government concerned.

Nonetheless, there may be circumstances in which a national povertyline does not exist or where household income/expenditure data for projectbeneficiaries are not available. Here it is sometimes possible to deriveapproximate average income/expenditure data for aggregate groups ofbeneficiaries, for example, by inferring income from data on household assets,


Untitled-1 5/17/02, 10:25 AM10

such as hectares of land available for cultivation or adult family membersavailable for casual wage employment. Lack of access to key basic need items—such as clean drinking water or primary education or to assets like typesof dwelling or animals—could be used as a means of defining which groupsfall into the category of the poor. The precise proxy to be applied would haveto be determined by the circumstances of the case (see Appendix 3 for availablesurvey data). The implication is not that this is a superior measure to income/expenditure but that if households had access to these assets they would useit to purchase the basic need concerned. As far as possible for consistency,the lack of access to the same basic need indicator—such as clean drinkingwater—should be used for projects of similar category. It will be theresponsibility of project teams to establish how the poor affected by a projectcan be identified, and where an income/expenditure measure of poverty isnot to be used, this omission should be justified.

The remaining part of the Handbook should be directly relevant for awide range of projects at present financed by ADB. These are projects thatare well-defined in the sense of having identifiable and quantifiable outputsproduced by tangible inputs. Even projects for which outputs are quantifiable(for example, numbers of pupils, patients, houses, etc.) but cannot be readilyvalued in monetary terms can be incorporated into the framework set outbelow. However, projects outside of this category, where outputs are definedmore broadly, such as institutional development in a policy reform context,would need a different approach under a separate study. Projects to promotesmall enterprise development, while potentially highly significant in terms ofpoverty reduction, also have benefits that are too diffused to make a meaningfulex-ante assessment. Financial analysis of such projects normally proceedson the basis of calculations on representative projects, whose returns may beaggregated to form an economic return. However, this practice is tooapproximate for meaningful distribution analysis. Here it is probably far moreeffective to concentrate on setting clear priorities for financial intermediariesand ensuring that they operate within a well regulated financial environment.For ex-post evaluation, however, there can be a range of evaluation techniquesfor this category of projects (e.g., Khandker 1999).

Resource implications are important in operationalizing poverty impactanalysis. The World Bank handbook (Baker 2000) for ex-post poverty impactevaluation studies indicates that the share of such studies in the total projectcost can vary widely from 0.2 to 1.3 percent. It is likely that the analysis relevantfor ex-ante project appraisal will not be able to match the time and resourcesrequired for such ex-post studies. Appendix 8 provides a preliminary indication

3 Approach to Poverty Impact Analysis for Poverty Intervention Projects 11

Untitled-1 5/17/02, 10:25 AM11

of resource requirements under project preparatory technical assistance. Thenext few years could be considered as a pilot test period for this Handbook.For distribution and poverty impact analyses to become meaningful, it isimportant that ESW is strengthened to cover poverty issues associated witheach country and major sectors where project identifications are proposed.It is recommended that future ESW carry out a diagnostic analysis on issuesthat are discussed in Chapter 2 to provide poverty-focused contexts of projectpreparation.

Pre-PPTA Stage

If poverty focus is to be built successfully into project work it is obviousthat it needs to be considered as early as possible in the project planningcycle. At the fact-finding stage of the project preparatory technical assistance(PPTA) the question must be raised as to whether or not a project should beconsidered as a Poverty Intervention. Technical nature of the project typemay well constrain the extent of targetability. At this stage it is recommended

Box 1

��

�� • Explain definition of poverty used (e.g., poverty line, or some basic needs indicator).• Identify and estimate the coverage of the poor groups affected to the extent meaningful.• Analyze causes of poverty in project area of influence to the extent meaningful.• Explain the mechanism through which the poor are affected (e.g., as consumers

through lower prices, as nonpaying users, as workers through new jobs, as producersusing services of the project as inputs).

��• Explain the critical assumptions required to achieve poverty impact (e.g., policies for

targeting, uptake by the poor, willingness to pay by the poor, financial sustainabilityof project).

��• Explain the risks of failure in achieving poverty objectives.• Examine scope for leakage of benefits to nonpoor.• Consider possibility of project encountering financial difficulties.• Discuss preliminary measures available to reduce the risks.


Untitled-1 5/17/02, 10:25 AM12

that the questions described in Box 1 be addressed in the initial socialassessment (ISA). It is recommended that the project's envisaged impact, themechanism through which it will improve the position of the poor and theassumptions required for this to be achieved, be presented. Considering thetypical time constraint for PPTA fact-finding missions, it is recommended thatas much existing poverty data relevant to the envisaged project as possiblebe collected prior to the mission (Appendix 3).

PPTA Stage

At this stage, projects that are to be Poverty Interventions will requiredetailed socioeconomic assessments and detailed questions on poverty impact.This will require a more precise indication of the poverty impact of the projectin terms of numbers of the poor affected and, wherever possible, estimatesof their net benefits (expressed in money-metric terms) due to the project.The assessment should provide the basic data that can be used to extendproject economic analysis to incorporate poverty impact analysis. There mustbe a discussion of ways of reaching the poor and ensuring that leakage ofbenefits to nonpoor is minimized (minimization of "type I error" relative toperfect targeting). Refined leakage minimization measures should be addressedat this stage. In the final project design, it should be ensured that credibleinstruments exist for targeting and monitoring poverty impacts.

A poverty focus requires that poverty concerns are fully incorporatedat the project design stage. Where possible, alternative ways of implementinga project, for example in choice of technology, in location or in type of service,should be considered with a view to maximizing net benefits to the poor,subject to the overall constraint that where net benefits can be monetizedthe design selected must achieve a minimum acceptable EIRR of 12 percent.(Maintaining the current economic decision criteria is unquestionable in lightof ADB's opportunity cost of development fund.) At the stage when projectdesign and location are being decided, poverty issues should be consideredwherever possible. The economic analysis appendix of the report andrecommendation of the President (RRP) should discuss project (design)alternatives that are considered. While discussion of the alternatives has beenthe standard requirement for all projects, poverty reduction should now bethe added dimension for consideration. Box 2 illustrates the type of questions


Untitled-1 5/17/02, 10:25 AM13

to be addressed. The discussion can be brief provided a clear case can bemade to justify the final design and packaging of the project.

Poverty impact analysis should be applied to the extent the analystconsiders practicable in the economic analysis appendix of the RRP for allPoverty Interventions. It could also be applied to Other Interventions whereverdata are available and the analyst finds it a useful tool in considering alternativeprojects or project designs. It should aim to calculate a poverty impact ratio(PIR), that is, the share of the poor in net benefits of the project (note, however,that the PIR in itself is not the target for maximization as cautioned in Chapter5). Where, as for example in primary health or sanitation projects, benefitscannot be valued realistically in monetary terms, the second-best will be aheadcount approach, that is, to estimate the poor as a proportion of totalbeneficiaries.

Steps for Poverty Impact Analysis

The extension of standard economic analysis of a project to povertyimpact analysis requires first a distribution of project net benefits—settingout the effects of a project for the groups that gain or lose—and second, apoverty impact analysis in which the proportion of each group's gains or lossesthat goes to the poor is estimated. Hence a poverty impact analysis cannotbe carried out without first undertaking a distribution analysis on a project(as can be seen from Step 5 in Box 3). Gains and losses arise first from thefinancial arrangement of a project. In addition they can arise from the distortionsor externalities that are captured in an economic analysis. The economic NPVof a project, which measures its full contribution to national welfare, will bethe sum of its financial effects and its external economic (economic minus

Box 2

��

• Explain the alternatives considered (e.g., location, type of service, technology, scale).• Explain the poverty impact of the various alternatives; this can be quantitative (e.g.,

number of jobs) or qualitative where firm data are lacking (e.g., access to improvedfacilities).

• Justify the alternative selected on both economic efficiency (e.g., higher economicNPV) and poverty reduction grounds (e.g., cost per unit of net benefits received bypoor.


Untitled-1 5/17/02, 10:25 AM14

financial) effects. The ADB Guidelines (Appendix 1) explain how the analysisworks. The conceptual issues associated with these procedures arediscussed in Fujimura and Weiss (2000). Rather than repeat these discussions


Box 3

��

��

�� . Set out the annual financial data on the project showing inflows (revenueand loan receipts) and outflows (investment, operating costs, loan interest andprincipal repayments, and tax both on profits and purchased inputs) from theperspective of the project owners. This is sometimes termed a return to equitycalculation at constant prices. As this aspect of financial analysis is in fact a departurefrom current normal ADB practice, Appendix 4 illustrates what is involved indistribution analysis with reference to actual ADB projects. Appendix 5 illustrateswhat is required in the project financial analysis to obtain the data needed indistribution analysis. See also Appendix 23 of the ADB Guidelines (1997) forillustration of financial returns to equity. This part of distribution analysis can onlybe done fully after the project financing plan is finalized (since only then will theloan-equity split be known), but in practice its omission may not make a greatdeal of difference to poverty impact estimates.�� Discount each annual inflow and outflow to derive present values for eachcategory and a net present value (NPV). Normally a 12 percent discount rate shouldbe used for these calculations. The resulting NPV will be a financial NPV showing theincome change for project owners. In addition there will be a gain to government fromtax payments and where subsidized loans are provided, a loss to lenders.Note: The choice of the 12 percent discount rate is for the purpose of analytical consistency.As the methodology applied in the Handbook is ultimately concerned with economicanalysis, the 12 percent is the natural choice. However, obviously the financial opportunitycost of capital faced by the project entity (or various stakeholders) need not coincide with12 percent. This issue is discussed in Fujimura and Weiss (2000). Financial viability testof the project should follow the usual practice (with the weighted average cost ofcapital, WACC, being the proxy for financial opportunity cost of capital). See ADB(2001a) for strictly financial aspects of project analysis.�� Identify the economic value to be used for each project input/output category.The ratio between this economic value and the financial price for actual transactions isthe conversion factor (CF) for the item concerned. Normally for distribution analysis itis simpler to conduct economic appraisals in the domestic price numeraire (which meansthat income from the financial and economic calculations will then be in the same priceunits). If a world price numeraire is required for the economic calculations, to carry outa consistent distribution analysis, all financial data from steps 1 and 2 must be converted

Untitled-1 5/17/02, 10:25 AM15


to world prices by multiplication by the standard conversion factor (SCF) (see Case 1 inAppendix 6). Nonetheless, it is recommended that the domestic price numerairebe used consistently.�� Express all project items in economic terms. This can be done by applying CFsto revalue the financial data from step 1. If CFs are taken as constant over the project'slife, only the present value figures at step 2 need to be adjusted. For items for which thereis no financial value at step 1 (for example, an environmental cost for which a projectitself is not charged), their economic value, wherever estimated, should be entereddirectly in the economic benefit flows.In practice, where project analyst did not foresee the need for distribution analysis andhas done the conventional economic analysis first, as would be the case for most projectpreparations prior to the wider application of distribution analysis, the analyst could workbackward to arrive at financial benefit and cost streams using conversion factors andtransfer payments (see Case 2 in Appendix 6).�� Allocate any difference between financial and economic values to particulargroups. These plus the changes for project owners and others at step 2 give the netbenefits created by the project. The net benefits to different groups must sum to theeconomic NPV of the project, since this measures the total net benefits of the project.This can be seen as an identity: economic NPV = financial NPV + (economic NPV -financial NPV).Note that where there is no financial revenue for the project agency, as in the case ofroad rehabilitation projects, net benefits still need to be distributed between differentstakeholders (see Case 5 in Appendix 6).

�� !��

A poverty focus requires that for each identifiable group affected by a project, theproportion of those who can be classified as poor be estimated.�� "� Estimate for each group affected by a project the proportion of netbenefits that will go to those below the poverty line. Groups involved will varybetween projects but will typically include consumers, workers, producers,government and the rest of the economy. For the government what is requiredis an estimate of the counterfactual; i.e., what proportion of government expenditurediverted from other uses by the project under consideration would have otherwisebenefited the poor. Similarly if a project generates government income a proportionof this will create benefits for the poor, which will be indirectly caused by theproject concerned (see Chapter 4).�� #� Finally, sum all net benefits going to the poor and divide by the totalnet benefits (economic NPV). This result is termed the poverty impact ratio (PIR).

here the basic steps required to carry out these analyses are set out in summaryform (Box 3). Appendix 4 illustrates actual distribution analyses of ADB projects.Appendix 5 shows how a financial analysis desirable from a distributionperspective can be structured. Appendix 6 illustrates poverty impactcalculations, based on materials from recent ADB project appraisals.

Untitled-1 5/17/02, 10:25 AM16

1 Introduction 17

though the methodology for distribution and, by extension, povertyimpact analysis has been around for more than two decades (seeFujimura and Weiss 2000), there remains a fair amount of data problems

in application. Of the parameters to be estimated in Step 6 of the analysis (Box3), two are the most fundamental. One is benefit incidence—the extent towhich it is possible to distinguish between the poor and nonpoor among directproject beneficiaries. The other is the problem of establishing the indirectimpact on the poor of the gains/losses that accrue to the government and therest of the economy.

4Data Requirements

Untitled-1 5/17/02, 10:25 AM17

Benefit Incidence

The estimation of how the poor among direct beneficiaries are affectedby a project can be complex. For certain types of projects with a direct povertyfocus, the poor should be the sole direct beneficiaries. This is most obviouswhere there is geographic targeting, so a school or health clinic is locatedin areas in which only the poor reside. Similarly, there may be effective self-targeting, where only the poor take up the services of the project; for examplea local clinic, where the better-off would use a hospital, or a public worksscheme, where only the poor would seek employment at the wage offered.Means of effectively targeting the poor would have to be part of the initialproject design.

On the other hand, there may be less clearcut cases. Provision of wateror electricity to an area may involve the connection of households, some ofwhich are above and some below the poverty line. The project socioeconomicassessment or similar field data collection during PPTA would have to identifythe approximate distribution of users and how the project will affect thesedifferent groups. Box 4 gives an indicative terms of reference for suchsocioeconomic assessment from the poverty impact perspective.

It may be that for some projects regional or district level estimates ofthe headcount index of poverty may be available (e.g., poverty mappingdiscussed in Appendix 3; Bigman and Fofack 2001), which can be used toinfer the degree of poverty in the population served by the project. In the

Box 4

��

The assessment should• identify the potential beneficiary (and/or loser) groups, separating those below the

poverty line; in doing this, collect and use as much existing survey data as possible(Appendix 3);

• describe the current status of the target population (either in income/expenditure ormore general socioeconomic characteristics);

• examine the need or demand for the project by the target population and whereappropriate assess how much they are willing to pay or forego to receive its services;and

• generate information to allow community participation in the design of project activitiesand in its implementation.


Untitled-1 5/17/02, 10:25 AM18

absence of such data, however, alternatives will have to be found. A specialhousehold survey, as part of ADB's PPTA, is an option. However, given thedifficulties in quantifying household income (e.g., where there is significantsubsistence activity and multiple family jobs) or expenditure (e.g., recall bias)in one-time surveys, it may be appropriate and more cost-effective to usesuch a survey to establish specific household characteristics (such as lack ofclean water, adequate shelter or access to primary education), which can beused to identify the existence of poverty. For example, a proportion of thefamilies without access to clean water in an area can become the approximateheadcount poverty measure. Such proxy indicators were found to be useful,for example, in the World Bank study on Poverty in Cambodia (World Bank1999). Appendix 3 discusses how existing survey data can be utilized to assistpoverty impact analysis prior to and during the PPTA.

An important part of any project plan will be to ensure that not onlyare target beneficiaries identified but also that there is a delivery mechanismthat ensures that benefits accrue primarily to the target group. However,whether or not the delivery mechanism includes user charges can have asignificant impact on the uptake of project services. Here consideration needsto be given to the balance between the project's financial sustainability andthe poverty impact on the direct beneficiaries, since even if price elasticityof demand may be low it will not be zero (see ADB August 2000, for a summaryof different approaches to demand elasticity estimation), while of course usercharges can be pro-poor from a broad fiscal perspective. Another aspect ofa socioeconomic assessment must establish how much the poor currentlypays for use of services, where payment may be either direct in fees or chargesto existing providers or indirect, for example in terms of travel costs or theopportunity cost of waiting time. In addition, it will be important to estimatethe demand for new services at likely levels of charge. In many instances,estimates of actual payments and potential willingness to pay may need tobe only approximate. However, in some cases (e.g., water and sanitation),willingness to pay for new connections may be an essential variable in bothtariff setting and benefit valuation. Detailed estimates of willingness to payare best left to specialist studies such as application of contingent valuationmethod (CVM) that would be additional to and complementary with socio-economic assessments. [ADB (March 1996) provides appendixes on CVMapplication for environmental valuation. ADB (March 1999) provides anappendix on CVM in water supply projects. Choe et al. (1996a and 1996b)provides detailed applications of CVM. See also Anand and Perman (1999),Carson (2000), Singh et al. (1993) and Whittington et al. (1993).]

4 Data Requirements 19

Untitled-1 5/17/02, 10:25 AM19

Where projects have effects that are spread widely, targeting mayinvolve considerably more than simply location. For example, an irrigationproject may in principle benefit farmers from a range of income strata. Ensuringthat a significant number of beneficiaries are poor may require provision ofa package of support services to poor farmers to allow them to gain themaximum crop increase from the availability of irrigation water. Seeds andfertilizer and the credit to purchase these will be required, as well as theinformation on their application. A road project would have inherent difficultyin associating its beneficiaries with geographical areas the road passes through.In such case, a special study such as origin/destination survey may be requiredin addition to the usual traffic projection. A socioeconomic assessment neednot set out the exact mechanisms involved but it should draw attention to theissues.

Government Net Benefit and Effects on the Rest of the Economy

Since governments will be major participants in ADB projects evenwhere they have no direct equity commitment, a critical parameter in povertyimpact analysis is the extent to which government expenditure and taxationaccrue to the poor. In other words, to apply the approach rigorously we requirean estimate of the benefit/tax incidence for the poor from marginal governmentexpenditure and finance. This is particularly important where poverty-focusedprojects receive subsidies and therefore draw income from the governmentbudget. However, this parameter is not available for any country and itsderivation on a country-specific basis would need to be the subject of separateand extensive research. To avoid complications, marginal governmentexpenditure/tax can be treated as distribution-neutral: additional unit ofgovernment net benefit induced by the project will be distributed back to theeconomy in a neutral way. Such distribution-neutral incidence of governmentnet benefit on the poor will conceptually be represented by the weightedby the weightedby the weightedby the weightedby the weightedaverage of the expendituraverage of the expendituraverage of the expendituraverage of the expendituraverage of the expenditure and tax incidencee and tax incidencee and tax incidencee and tax incidencee and tax incidence on the poor, which will needto be estimated in country-specific studies. The analyst might make an upwardor downward adjustment from the weighted average if she/he is convincedof a specific way in which project-induced income/financing will be distributed(somehow pinning down the fungible public fund), provided that it can besubstantiated by convincing empirical evidence.


Untitled-1 5/17/02, 10:25 AM20

However, the weighted average incidence parameter is unlikely to beavailable in the current state of knowledge in, for example, public expenditureand finance reviews. Until such knowledge becomes available, furthersimplifying assumption (admittedly with no rigorous theoretical underpinning)can be applied: if the government spends additional funds created by a projector forgoes spending as funds are displaced by a project, these effects will notalter the distribution of income and the poor will share in the government netbenefit in direct proportion to their share in the gross domestic product (GDP.)This simplification amounts to regarding the incidence of government netincome as being equivalent to the effects of the rest of the economy on thepoor (that can be referred to as economy effects). This is in line with theconvention of project appraisal in which the shadow exchange rate factor(SERF) premium is treated as equivalent of any other form of governmentincome. The assumption used is that the current state of income distributionis considered as the natural outcome of how an exogenously available extraunit of resources will be distributed in the economy. While this parameter israrely known or documented, Appendix 7 shows how an approximate estimatemay be derived from country-specific survey data and gives a list of preliminarycalculations for DMCs where necessary data can be obtained. In most casesthe income share of those below the poverty line will be low, and well belowtheir headcount share in population. (For example, the estimate for Viet Namis 12.6 percent while the headcount poverty incidence is 37.4 percent.) Thesecan be used as the proxy for the unknown parameter for the economy effects,and if no alternatives are available, also for the incidence of government netincome on the poor. Note, however, that this figure will only be available infinancial terms while project net benefits will have to be in economic terms.Where country-specific poverty data/poverty lines are not available, it isproposed that 10 per10 per10 per10 per10 percent could be used as a rcent could be used as a rcent could be used as a rcent could be used as a rcent could be used as a rule of thumbule of thumbule of thumbule of thumbule of thumb. This is roughlythe upper bound for the share of the lowest quintile in income distribution forDMCs (e.g., World Bank 2000f, p. 282, Table 5).

4 Data Requirements 21

Untitled-1 5/17/02, 10:25 AM21


Box 5

�� !�� "��

The assumption made about how the government net benefit would otherwise havebeen distributed and hence the degree to which it would have accrued to the poor canbe significant in determining a project's poverty impact ratio (PIR). Where projects arenet users of government income rather than generators of additional income for thegovernment, the lower this parameter the higher will be a project's PIR. This can beillustrated with a simple numerical illustration. We assume a project ENPV is distributedas follows.

��

$�� $�� $�� $��

Farmers 200 0.8 0.8 160 160Consumers 5 0 0.1 0.1 5 5Governmen t (100) 0.5 0.1 (50) (10)

ENPV 150Gain to Poor 115 155PIR 0 .77 1 .03

In case 2 with the lower proportion of the poor in government net benefits (0.1 ratherthan 0.5) the PIR rises to above 1.0. On the other hand, projects which are net incomegenerators for the government, for example, through a commercial pricing policy bypublic utility projects, will have a lower PIR, the lower the value assumed for the governmentnet benefit parameter.

Where government income is only marginally affected by a project (e.g., due to costrecovery pricing and low distortions in the economy) the overall PIR will not be sensitiveto this parameter. Nonetheless, when the assumption about the government incomeparameter is uncertain, a sensitivity analysis for PIR with respect to this parameteris recommended (see Case 5 in Appendix 6).

Untitled-1 5/17/02, 10:25 AM22

1 Introduction 23

Improving Project Quality at Entry

There are several advantages in carrying out distribution and povertyimpact analysis of projects that are relevant regardless of project classification.First, the analysis forces a more thorough cost-benefit analysis. For example,in the analysis of agricultural projects, focusing on distribution and povertyimpact forces separate consideration of different farm sizes when preparingfarm-income statements, which in turn may lead to a more detailed analysisof the rate of adoption of technology, contributing to better accounting for

5Advantagesand Limitationsof the Analysis

Untitled-1 5/17/02, 10:25 AM23

differences in expected productivity increases among farmers (Londero 1995).Second, the estimation of net benefits accruing to different stakeholders willhelp clarify who are the major beneficiaries and losers of the project andsharpen the judgement of project sustainability from the financial and politicaleconomy perspective. The analysis can assess the consequences of differentfinancial and institutional arrangements on the incidence of net economicbenefits on the poor. By establishing the link between financial and economicanalysis and rigorously eliciting the transfer payments, especially direct andindirect subsidies entailed, such analysis will highlight financial sustainabilityissue in a much stronger manner (Case 2 in Appendix 6). Third, the analysisbrings up policy questions in an illuminating way. For example, anoncompetitive market structure may prevent adequate benefits from reachingthe poor (Case 5 in Appendix 6).

Complementing the Inadequacy of Headcount Approach

The World Bank and Inter-American Development Bank (IDB) use projectclassification systems similar to ADB’s, in which a category of targeted projectsis given unique attention under a poverty reduction objective. Criteria forthis category of interventions rely crucially on the beneficiary headcountapproach. The World Bank’s criteria rely either on beneficiary headcount ora specific mechanism of targeting the poor (World Bank 2001), while IDB’scriteria are based on the combination of three criteria: sectoral, geographicaland headcount (IDB 2001). In the case of ADB, the classification (November2000, page 3) states, “The criteria used for poverty classification for loans are(i) the proportion of poor among the beneficiaries, and (ii) impact or benefitanalysis.” This foresees that both headcount and money-metric measure ofnet benefit incidence are applicable. This is based on the recognition that aheadcount beneficiary measure is a blunt and often misleading measure ofpoverty impact. It fails to capture the depth of poverty impact and can eitheroverstate or understate the money-metric measure. For example, overstatementoccurs when project net benefits are spread thinly over a large number ofthreshold-poor, while understatement occurs when project net benefits areconcentrated on a small number of hard-core poor. Therefore, one wouldprefer to apply the money-metric measure of poverty impact wherevertechnically feasible and practicable under available data and resources.


Untitled-1 5/17/02, 10:25 AM24

There are some obvious inadequacies with the headcount approach:beneficiaries must include those affected negatively as well as those whoare targeted for service delivery. In many projects that are net users of publicfunds, the population outside the projects’ influence areas are the losers. Shouldthe number of such losers be deducted from the number of gainers accordingto poverty level? This will not likely be done in practice. This vagueness remainsin defining what is meant by beneficiaries itself when looking at the cost sideof the project (Londero 1999). Furthermore, the headcount approach tellsnothing about efficiency of delivering poverefficiency of delivering poverefficiency of delivering poverefficiency of delivering poverefficiency of delivering poverty impactty impactty impactty impactty impact. While some targetedprojects may provide tangible benefits to the identified beneficiaries, thequestion remains whether their delivery mechanism is cost-effective. A fullpoverty impact analysis set out in this Handbook can check the efficiency ofpoverty reduction in this regard. By using the information obtained in thepoverty impact analysis, the indicator for efficiency of poverty impact can becalculated as the net benefits going to the poor per unit project cost. This canbe a ranking index among, for example, alternative subprojects (this point isilluminated in the discussion in Case 2 of Appendix 6).

With care and qualifications (page 27), poverty impact analysis can beused to provide economic underpinning of Poverty Interventions. Providedthat PIR estimation has been judged practical and carried out, the requirementof a project disproportionately benefiting the poor translates into an EIRR ofan EIRR ofan EIRR ofan EIRR ofan EIRR of12 percent or above plus a PIR significantly greater than the poor’s share12 percent or above plus a PIR significantly greater than the poor’s share12 percent or above plus a PIR significantly greater than the poor’s share12 percent or above plus a PIR significantly greater than the poor’s share12 percent or above plus a PIR significantly greater than the poor’s sharein GDPin GDPin GDPin GDPin GDP..... Here the PIR works as a money-metric counterpart of headcountpoverty incidence among the project beneficiaries, while the poor’s share inGDP works as a money-metric counterpart of national headcount povertyincidence. Appendix 7 provides a list of approximations for the poor’s incomeshare in GDP for selected DMCs. Depending on the agreed poverty line to beused in the analysis, the figure for the poor’s share in GDP may be excessivelylow or high. Where the figure for this parameter is unavailable or unreliablefor one reason or another, 10 per10 per10 per10 per10 percent could be used as a rcent could be used as a rcent could be used as a rcent could be used as a rcent could be used as a rule of thumbule of thumbule of thumbule of thumbule of thumb. Asin the headcount approach, how to interpret “significantly higher” is opento discussion and review.

For many Poverty Intervention projects it may not be possible to monetizebenefits and hence an EIRR is not calculated. For example, in many socialsector projects, benefits will be identified in quantity form (e.g., patients, yearsof life, pupils, years of schooling, number of dwellings). Where benefits arenot in money-metric terms, the twin requirements of efficiency and povertyimpact must still be applied, but in a different manner from that described

5 Advantages and Limitations of the Analysis 25

Untitled-1 5/17/02, 10:25 AM25

above. Here efficiency requires that the project be cost-effective in providingthe service concerned, e.g., in cost per patient, cost per year of life saved,or cost per slum dwelling cleared. Cost-effectiveness should in principle becalculated by taking the ratio of the discounted stream of project costs overits operating life to the discounted stream of physical output. Hence the unitcost of service provision must be derived in discounted terms as an averageincremental economic cost (AIEC). The comparator for testing whether a costis excessive will have to be built up from experience of similar projectinterventions in the country and elsewhere, as is recommended for healthsector projects (ADB August 2000). For such projects where project benefitsare not monetized, full PIR analysis is impossible and poverty impact estimationwould be limited to a headcount basis (Case 4 in Appendix 6).

Variety of Projects

One obvious limitation of the distribution and poverty impact analysisis that it cannot be expected to cover all types of projects and comparabilityacross projects cannot be any better than that of conventional project economicanalysis as it is practiced. This limitation is nothing new and is a carryoverfrom the limitation in the current practice. At one end of the range are projectsin sectors such as power, water and irrigation, where full benefit-cost analysesare regularly applied and where the use of distribution and poverty analysismay be a natural extension of current work. At the other end of the range areprojects in social sectors such as health and sanitation and primary education,for which EIRRs are rarely calculated for practical purposes. Such projectscan be subject to cost-effectiveness analysis and increasingly a form of suchanalysis is being applied. As explained above, alternative criteria can be appliedfor poverty-focused projects where monetary estimation of benefits is notpossible and beneficiaries may be, for example, in terms of number of poorpatients or poor pupils. Between these extremes will be a range of intermediatesituations, such as in roads, where there may be technical difficulties relatingto distribution and poverty analysis. However, it is encouraging that ADB’srecent transport projects are testing the methodology (see Appendix 8).Projects for which the methodology is very difficult to apply will be, for example,in relation to institution building and private sector development, since it willbe difficult to relate investment expenditures to tangible outputs and incomeflows.


Untitled-1 5/17/02, 10:25 AM26

Caution on Interpretation of PIR Index

Early experience at ADB has provoked two concerns. First, some mayhastily interpret that the PIR index is a summary indicator for poverty impactas the EIRR is a summary indicator for project economic viability. This is notthe case. The PIR by itself is merely the proportion of the NPV accruing to thepoor against the total project NPV and does not inform poverty impact rankingor efficiency of poverty reduction among alternative projects (designs). It isconceivable that in some cases, a deliberate attempt to raise the PIR wouldreduce poverty impact in absolute terms and defeat the purpose. What projectsshould be maximizing is not the PIR index but the NPV going to the poor(absolute poverty impact), or the NPV going to the poor per project cost(efficiency of poverty impact).

Second, while in principle the PIR is a superior index to the headcountpoverty index in interpreting the meaning of “disproportionately benefitingthe poor,” it is often an uncertain point estimate. As explained in Chapter 4(Box 5), the PIR is usually sensitive to crucial parameters, assumptions forwhich are uncertain. Therefore, the analyst needs to use a sensible judgementon whether the full poverty impact analysis makes sense and avoid mechanicalapplication of the PIR. When it is calculated, sensitivity tests are stronglyrecommended with respect to uncertain parameters such as the poorproportion of net benefits accruing to the government. Sensitivity test on thePIR can be useful in illuminating policy discussion if it is done on policyparameters such as utility pricing (Appendix 1) and market competitiveness(Case 5 in Appendix 6).

Risk for the Poor

Poverty is normally associated with vulnerability, so that unexpectedunfavorable outcomes may have very serious consequences for those at orbelow the poverty line. ADB project appraisals in the past have rarely addressedissues of risk of project failure with a few exceptions in the energy sector.ADB Guidelines (1997, Appendix 21) have only a brief discussion on riskanalysis. The possibility of pooling risk across projects provides a rationalefor omitting risk calculations from the decision criteria, but it does not apply

5 Advantages and Limitations of the Analysis 27

Untitled-1 5/17/02, 10:25 AM27

when one is considering particular groups. Fujimura and Weiss (2000)illustrate how in principle a risk of failure, defined as the probability that aproject makes the poor worse off, can be estimated. The idea is to replacepoint estimates of project net benefits accruing to particular groups ofbeneficiaries by their probability distribution. This subject is taken up in asubsequent EDRC study (ADB 2001).


Untitled-1 5/17/02, 10:25 AM28

CHAPTER 1 Introduction 29

t is worth noting that project ideas emerge from sector-level work andthe proposed project should address some type of constraint identifiedin a sector-level analysis, which cannot be addressed if left to private

sector activities. Logical framework requires that projects be identified toimprove sector economic performance for the benefit of the entire economy.When in some cases project identification is based on inadequate sector-level analysis or project formulation is based on unclear economic justification,the process of project economic analysis can proactively raise policy issues.For example, the extent of difference between financial and economic prices(and therefore the consequences for estimates of FIRR and EIRR) usuallyimplies something fundamental about the nature of the policy environment

6Linkage to Policy-Based Lending

Untitled-1 5/17/02, 10:25 AM29


within which the project is being prepared (ADB June 1999). If such differencecan be traced to either sector-level and economy-wide policy or institutionalanomalies (that hinder efficient workings of the economy), there may beroom to address the former within the project design (e.g., removal of statemonopoly in a particular project input or output market) before the fullinvestments take place. Or alternatively, in consultation with the counterpartgovernment, consideration can be given to resorting to sector developmentprogram modality with a policy loan component. The project analyst shouldexamine whether the project’s envisaged benefits could be jeopardized bythe current policy and institutional environment and whether this shouldbe addressed through policy- based interventions including high-level policydialogue. The early findings from a separate EDRC study (Bolt and Fujimura2001) point to the importance of political economy process in assessing linksbetween policy-based lending and poverty reduction.

Untitled-1 5/17/02, 10:25 AM30

1 Introduction 31

Appendixes

Untitled-1 5/17/02, 10:25 AM31


Untitled-1 5/17/02, 10:25 AM32

1 Introduction 33

APPENDIX�1��

(reproduced from Appendixes 25 and 26 of the ADB

Guidelines for the Economic Analysis of Projects)

APPENDIX 25: DISTRIBUTION OF PROJECT EFFECTS

The costs and benefits of a project are shared among different groups.There are several ways in which the distribution of project effects can beanalyzed. First, the project effects can be allocated among different projectparticipants, usually suppliers, consumers, owners, lenders, workers orproducers, and the government representing the rest of the economy. It isusual to expect owners, lenders, workers or producers, and the governmentall to share in the net project effects. Frequently, consumers and suppliersalso do. Second, for projects that involve foreign investors, lenders,management, and labor, the distribution of net project effects between nationalsand foreigners can be demonstrated. Third, project effects can be allocatedbetween the public and the private sectors. This may be particularly importantfor infrastructure developments where public sector expenditures are madein support of private sector operations. Fourth, the net project effects can beallocated not only among different project participants but also amongparticipants with different income levels. Fifth, net project effects can beallocated according to whether the project net benefits are likely to beconsumed or saved. Finally, costs and benefits can be allocated among differentcountries participating in subregional projects.

Considerable effort was expended in the past to include the distributionof net benefits between savings and investment into project analysis. Thepurpose was to identify and give priority to projects that would enhancesavings, and therefore investment in the economy, by applying a premium toproject effects resulting in extra savings. Considerable effort was also expendedto include the distribution of net benefits by income group into project analysis.The purpose here was to identify and give priority to projects that wouldenhance incomes for lower income groups, by applying a different weight to

Untitled-1 5/17/02, 10:25 AM33

the incremental incomes of different groups. However, both forms of analysisdepend on specifying premia that are essentially subjective and open todisagreement. In addition, enhancing savings can lead to priorities thatcontradict the enhancement of incomes for lower income groups, and bothsavings and the distribution of income can be affected more directly by policychanges at the national level rather than through the net effects of newinvestment projects.

It is recommended that simply a statement of the distribution of projecteffects be given, without applying any premium to either incomes that aresaved or to incomes accruing to particular income groups. There are threereasons for providing such a statement. The first is to assess whether thelikely distribution of project effects corresponds with the objectives of theproject. The second is to bring the financial and economic analyses of projectstogether to ensure that the consequences for the economic benefits of projectsof changes in financial arrangements are assessed. The third is to assess thelikely impact of policy changes on the distribution of project effects.

The following example illustrates the construction of a statement onthe distribution of project effects. For simplification, the project excludes theeffects of project financing, that is, it does not consider the possible net costsor benefits to lenders. Neither does it include direct tax payments. Theillustration concerns a telecommunications project involving 50,000 new linesand associated exchanges that will extend the national network into a ruralarea through the provision of publicly accessible telephones in villages andrural towns. The analysis of the distribution of project effects is based on theincremental number of calls for the telecommunications corporation and theincremental costs of providing the new telephone facilities. Values for thecosts and benefits of the project, at both financial and economic prices, areall given as present values calculated at a discount rate of 12 percentrepresenting the economic price of investment funds in the economy.

The forecast project financial statement at constant domestic marketprices is summarized in Table A1.1. At the projected future charge level, whichwill apply across the whole telecommunications network and not just in theproject area, the telecommunications corporation will not recover the fullincremental costs of the project at financial prices inclusive of the opportunitycost of capital. The corporation will have a loss on resources in present valueof 100.

The economic analysis of the project introduces three majorconsiderations. First, with the project telephone calls will be made at a costthat includes the telephone charge going to the corporation plus the costs of


Untitled-1 5/17/02, 10:25 AM34

Table A1.1

� �� !�� #��

�� %��&�� '� �� (

BenefitsRevenue 700

CostsEquipment 400Installation 100Operating Labor 100Other Operating Costs 200

)�� $�� *++

,�� %�� &�++(

traveling to the telephone. Without the project a high proportion of telephoneusers would continue to communicate through other means, includingtraveling to the call destinations. The difference between the cost ofcommunication without the project and the full costs with the project, includingthe costs of reaching the telephone, represents an economic benefit to telephoneusers that is not incorporated in the financial charge for the telephone calls.In addition, a further economic benefit will stem from the fact that severalsmall businesses and farmers will benefit from the better access tocommunication relating to input and output markets and prices, and transportschedules. Taken together these additional economic benefits can be addedon to the financial revenues as a consumer surplus. Second, there is a differencebetween the economic price of foreign exchange and the official exchangerate. A shadow exchange rate factor (SERF) of 1.3 has been estimated for thecountry, implying that foreign exchange costs have a higher economic thanfinancial cost to the economy. Third, there is a surplus of labor that couldeasily be trained for telecommunication operations in the area. The opportunitycost at domestic prices for operating labor has been estimated as 90 percentof the wage level, in other words a shadow wage rate factor of 0.9.

APPENDIX 1 Distribution and Poverty Impact Analysis 35

Untitled-1 5/17/02, 10:25 AM35

The financial project statement has been adjusted by the consumersurplus and the appropriate conversion factors to derive the project economicstatement in Table A1.2. The economic values have been expressed at thedomestic price level in national currency. Table A1.2 also shows the differencesbetween the financial and economic value of resources. These differencesgive rise to losses and gains among the project participants. As indicated bythe consumer surplus, consumers of the new telephone services benefit tothe extent to which the economic value of communication cost savings andbusiness efficiency improvements exceed the full cost of making calls. Theeconomic valuation of the equipment for the project exceeds its financial valueto the extent of the SERF; the consequent loss because of the overvaluationof the exchange rate is borne by the government representing others in theeconomy, especially importers. The financial cost of labor exceeds itsopportunity cost; the difference accrues as a gain to operating labor. Thesegains and losses are complemented by the loss to the corporation becausenot all the full financial costs, including capital costs, are recovered. The right-hand-side section of Table A1.2 summarizes these gains and losses to differentproject participants.

The overall results for the project are a financial net present value (NPV)of minus 100 and an economic NPV of 40. The economic NPV exceeds thefinancial NPV by 140. More specifically, as at present structured, twoparticipants lose from the project. The corporation will suffer a loss of 100,and the rest of the economy will suffer a loss of 120 because foreign exchangeis available at a price lower than its economic price. On the other hand, twoparticipants will gain. Operating labor will gain by 10 at the project wagelevel, while consumers will gain from their consumer surplus of 250. Thesegains and losses in part compensate for each other; the net gain is positiveand is equal to the economic NPV of 40.

The differences between the financial and economic values, and theconsequent gains and losses for different project participants, provide thebasis for considering the impact of policy changes. First, there is a small gainto labor. If there were a completely competitive market for labor this wouldnot occur. However, this is not a major source of difference between financialand economic outcomes. Second, a substantial gain has been identifiedaccruing through the corporation to the consumers as a result of revaluingthe foreign exchange element of project inputs. This gain could only be correctedby a general realignment of domestic and world prices, outside the contextof project level changes.


Untitled-1 5/17/02, 10:25 AM36

Table A1.2

�� "�� !��

$� �� %&' �� (

Difference Distribution of Project Effects

��

��

Benefits

Revenue 700 1.00 700 0

Consumer Surplus 1.00 250 250 250

�� ! �

Costs

Equipment (400) 0.30 (520) (120) (120)

Installation (100) 1.00 (100) 0

Operating Labor (100) 0.90 (90) 10 10

Other Operating

Costs (200) 1.00 (200) 0

� �� 800 910 110

"�� (100) 40 140 #$��%

�� &� �� (100) (120) 10 250

Finally, the main beneficiary of the project will be the consumers. Inpart, their benefits could be incorporated into the telephone charges theymust pay. If the telephone charges were raised to cover the financial loss of100, that is, by 100/700 or 14.3 percent on a project basis, most consumerswould still be making substantial gains. However, the most marginal consumersmay not use the new telephone in these circumstances and so some of theeconomic benefits would be lost. Moreover, any change in telephone chargesin real terms would also impact on existing network users.

The distribution of project effects is of interest for its own sake. It alsoassists those designing projects by drawing attention to the effects of currentpolicies on the financial and economic results, in this case, the exchange rateand pricing policy. Changes in project design within the current policies canbe assessed within the same framework. The effects of policy alternativescan be presented in this way to inform policy dialogue with governmentsand the stipulation of loan covenants.


Untitled-1 5/17/02, 10:25 AM37

APPENDIX 26: IMPACT ON POVERTY REDUCTION

Poverty reduction is the most formidable development challenge. Toreduce poverty some projects target the poor directly, but most aim at economicgrowth, benefiting the poor indirectly as well as directly. This appendix showshow to trace the economic impact of growth projects on the poor.

The poverty-reducing impact of a project is traced by evaluating theexpected distribution of net economic benefits to different groups. With financialprices determining who controls net economic benefits, the first step is toestimate the present value of net financial benefits by participating group.Next, the difference between net benefits by group at economic and financialprices is added to net financial benefits by group to give the distribution ofnet economic benefits by group. Finally, the net economic benefits accrue tothe poor according to the proportion of each group that is poor. A povertyimpact ratio expressing the proportion of net economic benefits accruing tothe poor can be calculated by comparing net economic benefits to the poorwith net economic benefits to the project as a whole.

This can be illustrated through a publicly funded water utility projectselling piped water. The water supply project serves a small rural town. Allcapital equipment is imported, subject to an import tariff. Labor and electricityaccount for total operating and maintenance (O&M) costs. Wages are controlledby a minimum wage law, with the economic price of labor being a proportionof the minimum wage. Electricity is subject to a sales tax and a productiontax. The water utility is not subject to income tax. All financial and economicvalues are given in constant year-of-appraisal prices and in present valueterms. Tradables are valued at border prices at the domestic price level andnontradables at domestic market prices. Net financial benefits (NFB) and neteconomic benefits (NEB) are expressed in domestic currency (rupees).

For the purpose of poverty impact analysis, project beneficiaries aredivided into three national groups: the poor, the nonpoor, and the government.Net economic benefits by group are distributed between the poor and thenonpoor, according to the extent that they benefit the poor. In the case of neteconomic benefits to the government, it is assumed that 50 percent potentiallybenefit the poor.

The present value of project capital costs is $25 million at border prices.Import duties are 30 percent, the official exchange rate (OER) is Rs20/$ andthe SERF is 1.20. The market value of electricity is Rs300 million, including


Untitled-1 5/17/02, 10:25 AM38

a production tax of 20 percent and there is a sales tax of 10 percent. Wagesamount to Rs80 million and the supply price of labor is 70 percent of theaverage wage rate. Water sales are Rs1,000 million. The quantity of waterillegally consumed is 20 percent of revenue water. The economic cost ofwater consumed and paid for is Rs1,500 million.

The NFB is equal to sales revenue of Rs1,000 million minus capitalcosts of Rs650 million ($25 million multiplied by the OER of Rs20/$ plus theimport tariff of 30 percent), electricity costs of Rs330 million (the marketvalue of electricity plus sales tax), and labor costs of Rs80 million. The NFBof the project shows a loss of Rs60 million in present value (see Table A1.3).

The NEB of the project expressed at the domestic price level is Rs894million. It is equal to gross benefits of Rs1,800 million (the cost of water increasedby the proportion of water consumed but not paid for) minus capital costs ofRs600 million (capital imports converted to local currency at the OER multipliedby the SERF), electricity costs of Rs250 million (market value of electricityless production tax), and labor costs of Rs56 million (wages valued at thesupply price of labor).


Table A1.3

�� )��

$� �� %&' �� (

'( )�� )��

��*�� +�� +��

Output 1,000 1,800 800 800

Capital costs (650) (600) 50 50

Electricity (330) (250) 80 80

Labor (80) (56) 24 24

�� #,�% -�. � . -�� $/� 24

�( �� 0�� +��

NEB-NFB 800 130 24 954Financial return (60) (60)Benefits 800 70 24 894Proportion of poor 0.25 0.50 0.333Benefits to poor 200 35 8 243

�� 0�� +��1� !./� �� -�.� 2� �(!�$� �� !��

Untitled-1 5/17/02, 10:25 AM39

The difference between the NEB and the NFB is distributed by group.The difference of Rs954 million is made up of (i) consumer surplus of Rs800million (the difference between the without project cost of water and the withproject expenditure on piped water, plus the value of water consumed butnot paid for); (ii) government tax revenues from capital imports of Rs150million; (iii) government tax revenue from electricity production of Rs80 million(production tax of Rs50 million plus sales tax of Rs30 million); (iv) benefits tolabor of Rs24 million (wages of Rs80 million less opportunity cost of Rs56million); and (v) loss in the economy to government of Rs100 million throughovervaluation of the exchange rate.

The NEB-NFB difference is added to the NFB by group to arrive at thedistribution of the NEB by group. The government's financial losses frominvesting in the water supply project amount to Rs60 million. Adding Rs130million in taxes results in a net economic benefit to the government of Rs70million. Consumers gain Rs800 million in consumer surplus and laborers earnRs24 million more than they would have without the project. The NEB bygroup is Rs894 million.

The final step is to distribute the NEB by group between the poor andthe nonpoor. One quarter of consumer surplus and one third of surplus forlabor go to those living below the poverty line. Fifty percent of the returnto the government is assumed to benefit the poor. The NEB accruing to thepoor is therefore Rs243 million. The PIR of the project is Rs243 million/Rs894million or 27 percent.

Charges, Benefits, and the Poverty Impact Ratio

The government has decided it can no longer sustain the financiallosses of the water supply corporation. The level of water charges is to beraised by 50 percent. It is predicted that, with a price elasticity of demandof -0.4, this will result in a decline in the volume of revenue water of 20 percent.Table A1.4 depicts the financial and economic returns, and the PIR, in thesenew circumstances. It is assumed that capital and labor costs are fixed, whileelectricity costs are fully variable. It is also assumed that nonrevenue waterwill remain in the same proportion as revenue water.

The new level of charges captures some of the consumer surplus.Financial returns become positive and substantial while economic returns,though still positive, are reduced. The distribution of the net benefits betweengroups changes significantly. The government receives less tax revenue butnow receives a surplus from the water supply corporation instead of a loss.


Untitled-1 5/17/02, 10:25 AM40

Table A1.4

�� *�+�� +� ,��

$� �� %&' �� (

'( )�� )��

��*�� +�� +��

Output 1,200 1,440 240 240

Capital costs (650) (600) 50 50

Electricity (264) (200) 64 64

Labor (80) (56) 24 24

�� !�, -. /�- !.� $$. !.

�( �� 0�� +��

Beneficiaries NEB-NFB 240 114 24 378 Financial return 206 206Benefits 240 320 24 584Proportion of poor 0.25 0.50 0.333Benefits to poor 60 160 8 228

�� 0�� +��1� !!-� �� -.� 2� �(/�� /��

Its share of the benefits increase considerably. The benefits to labor remainthe same, while the benefits to consumers decrease substantially, both becauseof the reduction in consumer surplus per unit of water consumed and becauseof the decrease in consumption.

The PIR in these new circumstances is 39 percent instead of 27 percent.It has increased significantly but it is not the main parameter to be affectedby the increase in charges, which has transferred more benefits to the ownerof the water supply corporation, the government. In fact, the absolute amountof benefits going to the poor has decreased with the increase in water charge.This suggests two things. First, the charges may have been raised by toomuch; given the new financial returns, a lower increase in charges couldhave ensured the financial sustainability of the corporation. Second, the tariffstructure is as important for the PIR as the tariff level. In this case the tarifflevels were increased for all types of consumers. An increase in tariff togetherwith a different tariff structure could have captured some of the consumersurplus from the higher income groups while leaving the poor groups


Untitled-1 5/17/02, 10:25 AM41


unaffected. In other words, the increase in charges could be designed to leavea higher proportion of benefits going to the poor.

By focusing attention on cost recovery mechanisms and tariffstructures, PIR analysis can help improve project design by identifying whobenefits and who pays, and by how much. Pricing policy can affect thepoverty impact of a project; it can also affect the distribution of benefitsbetween the private and public sectors, for example, where the water supplycorporation is privately not publicly owned. However, projects designed tohave a significant impact on the poor may at the same time have to beprovided at a different scale or in a different location, to raise the proportionof benefits going to the poor.

Untitled-1 5/17/02, 10:25 AM42

1 Introduction 43

APPENDIX�2Benchmark Criteria for

Good Practice Project Preparation

The following is adapted from the study undertaken by a staff consultant(Hansen 2000) who reviewed RRPs for ADB's transport projects. Generic pointsin the study's final report are extracted here. Based on the review of RRPs expost, the report identified seven benchmark criteria that can be used to judgewhether the design of a project is such that it is likely to make a significantcontribution to poverty reduction.

The criteria are as follows:(i) Has the project drawn on evidence about, and addressed, the

causes of poverty?(ii) Has the project, in its choice of site and design, explicitly addressed

poverty reduction?(iii) Has the project design been developed so as to reduce possible

adverse impacts on poor people?(iv) Is the project linked to poverty-focused policy reforms and

institution building?(v) Is the project designed as part of a package of integrated projects

and programs? Has the possibility that the project will crowd outother poverty reduction projects been addressed and assessed?

(vi) Does the background data for the project establish the extent towhich the situation of the poor, in general, and that of targetgroups, in particular, can be assessed?

(vii) Have poverty impact distribution assessment and benefitsincidence assessment been carried out?

Based on these criteria, the report recommends a checklist of tencontrols, for quality at entry, which would help to identify any weaknessesand shortcomings in the design of projects, so that the design could becorrected before it is too late in the project cycle.

Untitled-1 5/17/02, 10:25 AM43


The ten controls are as follows:(i) The project selection, design, and implementation arrangements

should incorporate key social issues and the views of majorstakeholders, as determined through a participatory process.

(ii) The social impact of the project should be disaggregated by socialgroup (including gender) and adequate provision made formitigation of any adverse impacts.

(iii) The project should be consistent with the poverty reductionstrategy as discussed in the country operational strategy (COS).

(iv) If the project is poverty-targeted, the design should be adequateto reach the target beneficiaries.

(v) The direct impact on the poor should be clearly articulated andquantified to the extent meaningful.

(vi) The indirect impact on the poor should be clearly articulated andquantified to the extent meaningful.

(vii) Adequate arrangements should exist for monitoring and evaluationof social impacts, including poverty impacts. These should includea baseline survey, clearly specified targets, provision for datacollection on outcome indicators, and provision for financing ex-post evaluation of project impacts.

(viii) Project design should comply with The Bank’s Policy on Indigenous

Peoples.

(ix) Project design should comply with ADB's policy on Involuntary

Resettlement.

(x) Project design should comply with ADB's policy on culturalproper ty.

Untitled-1 5/17/02, 10:25 AM44

1 Introduction 45

APPENDIX�3Application of Existing Primary Survey Data

in Poverty Impact Analysis at the Project Level

Fully Exploiting Existing Household Survey Data

Development assistance agencies invest considerable effort and expensein collecting social and economic data in ADB's DMCs. Although a substantialamount of data is collected, the lack of relevant, timely, and reliable informationis still often invoked as a constraint to conducting detailed empirical analysisin project formulation. Lack of detailed quantitative information in areas whereprojects are being implemented hinders effective targeting, and the assessmentand monitoring of the effects of poverty reduction programs and projects.However, the potential of available data for application in poverty assessmentsin support of ADB operations is significantly underexploited. The reasons whythis is the case include the high fixed costs to initially load and understandavailable data, the tight time schedules typically faced in loan preparation,and the lack of easy-to-use analytical protocols (i.e., programs for using dataneeded to parameterize worksheet-based poverty analyses) to assist in theanalysis of previously collected data to support project activities. Table A3.1describes the characteristics of surveys and censuses suitable for povertyanalysis.

The use of existing data and information for policy and project designcould be increased by facilitating the ease of access to primary data throughcentralized cataloguing and storage of the data in a data archive or data bank(Box 3.1), and development of analysis tools. More effective management andsharing of the available data could improve the efficiency and quality ofresearch carried out in support of loan preparation. It assists in avoidingduplication in data collection efforts, and minimizes the need to collect newdata. The cost and time savings from using available data rather than initiatingnew data collection are other important reasons for making use of availabledata in project preparation as much as possible.

Untitled-1 5/17/02, 10:25 AM45


Typical Sample Size

2,000–5,000 householdsnationally

2,000–5,000 householdsnationally

Varies greatly dependingupon population of countrywhere applied

Entire population of thecountry

Frequency of DataCollection

Possible Geographic Disaggregation

Large agroclimatic zones/regions—usuallycannot be used at state/provincial level orlower

Large agroclimatic zones/regions—usuallycannot be used at state/provincial level orlower

State or provincial level in some countries,regional/national in all

Village level

Num

ber

of

Inte

rvie

ws

with

ea

ch H

ouse

ho

ld S

urv

eye

d

Reco

mm

en

ded

Fre

quen

cy o

fIm

ple

men

tatio

n1

Multiple

Multiple

Single

Single

,��+ �� -�� $,�-�(

Table A3.1

��

*�� ".�� $*�"�(

�� -�� + �� $�-�(

�� *��+ ��

Every 5 years

Annually

Annually

Every 10years

Untitled-1 5/17/02, 10:25 AM46

1 Introduction 47

Time, Expertise, and ResourcesRequired2

1–2 years required to carry out. Highlevel of expertise necessary. Veryexpensive to implement.

1–2 years required to carry out. Highlevel of expertise necessary. Expensive toimplement.

6–12 months required to carry out.Moderate level of expertise necessary.Relatively inexpensive to implement.

About 2 years required to carry out.Moderate-to-high level of expertisenecessary. Very expensive to conduct.

Other Principal Characteristics

Comprehensive survey intended forpoverty analysis. Panel data availablefor some countries. Many secondaryapplications (e.g., poverty maps).Complexity of survey can raise validityproblems.

Mainly used for CPI and NationalAccounts. Alone has limited applicationin poverty analysis. Often supplementedwith data collection on other topics.

Quick and easy-to-implement survey thatprovides poverty indicators but no orlimited information on income/consumption. Covers many of the sametopics as LSMS but in much less detail.

Collects little information, but coversentire population. Data access can besensitive politically. Can be combinedwith LSMS to generate poverty maps.

Topic Usually Covered3

Application of Existing Primary Data inPoverty Impact Analysis at the Project LevelAPPENDIX 3

Ed

uca

tio

n &

Em

plo

ymen

t

Co

mm

un

ity

Leve

l D

ata

(e.g

., p

rice

s)

Pu

blic

Serv

ices/

Infr

ast

ruct

ure

Fa

mil

y C

om

po

siti

on

4

Inco

me

Sa

vin

gs

& A

ssets

Co

nsu

mp

tio

n &

Exp

en

dit

ure

Hea

lth

/Nu

trit

ion

Sta

tus

Ho

usi

ng

Co

nd

itio

ns

Untitled-1 5/17/02, 10:25 AM47


Typical Sample Size

5,000 – 8,0005

Roughly 15,000

4,000 – 5,000

Frequency of DataCollection

Possible Geographic Disaggregation

Large agroclimatic zones/regions—usuallycannot be used at state/provincial level

Generally state/provincial level

State/provincial level in some countries,regional/national in all

Num

ber

of

Inte

rvie

ws

with

ea

ch H

ouse

ho

ldSu

rveye

d

Reco

mm

en

ded

Feq

uen

cy o

fIm

ple

men

tatio

n1

Usually asinglevisit

Single

Single

��+ �� *�� $�*�(

Table A3.1 (cont’d.)

��

�� )�� $�)��(

-�� $-��(

No regularschedule

Annually

Every 5 years

��1 Actual frequency of survey/census implementation varies markedly from recommendations, and often

depends upon donor support. When administered annually, RMS applies a core questionnaire each yearwith rotating specialized modules. Some countries now carry out a partial-census every 5 years.

2 Particular expertise required to complete different surveys differ greatly. For example, the sampleselection and survey enumeration for the LSMS and HIES are difficult due to the complexity of thequestionnaire used, while the CWIQ has a simple questionnaire but requires expertise to identify povertypredictors. Carrying out a census requires mapping skills and massive logistical support. DHS askssensitive health history questions requiring well-trained interviewers.

6

Untitled-1 5/17/02, 10:25 AM48

1 Introduction 49

Time, Expertise, and Resources Required2

About 6 months required to carry out.High level of expertise necessary.Relatively inexpensive to conduct.

About 4 months required to carry out.Moderate level of expertise necessary.Relatively inexpensive to implement.

About 6 months required to carry out.Relatively low level of expertisenecessary. Relatively inexpensive.

Other Principal Characteristics

Demographic and health profile, butlacks economic characterization. Datapublicly available. Now coordinatedwith MICS.

Quick and inexpensive to apply survey.Requires valid poverty predictors fromLSMS survey. New type of surveyimplemented in only a few countries todate.

UNICEF-sponsored focusing on childand maternal welfare information. Datanot widely disseminated.

Topic usually covered3

3 Darkly shaded boxes indicate topic is well covered in the survey/census, lightly shaded boxes indicatelimited data on the topic is collected.

4 Family composition refers to demographic characteristics such as are typically captured in an inventoryof household residents.

5 Sample sizes reported are rough figures. Sample used for DHS can be much larger—in the case of India,the DHS covers about 90,000 households.

6 The CWIQ does not collect information on total consumption, but solicits data on easy-to-collect variablesfound to be highly correlated with consumption.


Fa

mil

y C

om

po

siti

on

4

Ed

uca

tio

n &

Em

plo

ymen

t

Inco

me

Sa

vin

gs

& A

ssets

Co

nsu

mp

tio

n &

Exp

en

dit

ure

Ho

usi

ng

Co

nd

itio

ns

Hea

lth

/Nu

trit

ion

Sta

tus

Pu

blic

Serv

ices/

Infr

ast

ruct

ure

Co

mm

un

ity

Leve

l D

ata

(e.g

., p

rice

s)

Untitled-1 5/17/02, 10:25 AM49


Use of Household Survey Data at Different Steps of Loan Processing

At the pre-design stage

Available household survey data can provide useful input into the initialassessment of the poverty impact of a proposed project. Data can be usedto (i) help identify beneficiaries, (ii) define the constraints and barriers facedin increasing income and improving the welfare of the poor, (iii) assist in thedesign of effective targeting mechanisms, and (iv) contribute to preliminaryproject classification. Issues of data confidentiality and access are problemsthat prevent use of existing data, and must be overcome in some of the DMCswhere ADB operates. In addition, a modest commitment of personnel timeand resources will be required to make available data easily accessible toADB staff developing projects, but an important constraint to their applicationwill be the readiness of staff and consultants to use the statistical and analyticaltechniques needed. The availability of well-documented and standardized

Box 3.1

�� $� �� !��/(

Data archives acquire, store, and disseminate data for secondary research. Secondaryresearch refers to research by individuals or groups besides those originally collectingthe data for research purposes that may not have been envisioned when the data werebeing collected. Secondary research may seek to replicate analyses already carried outto verify results, to extend original research, or to analyze the data for an entirely newpurpose or from an entirely new perspective. For example, DMCs periodically conductnational censuses to measure the size, composition, and basic socioeconomic indicatorsof their population. Because census data represent the sole source of survey informationat the household level which covers all localities in a country, several secondary researchtechniques have been developed that combine the data from the census with other datato consider issues beyond population and simple characterization of living conditions.Archiving of data involves collection of information from previously conducted surveysat the original level of observation at which the data were collected (e.g., households,individuals, municipalities, etc.) and making it available in a data format desired by theuser. Archiving also requires documentation regarding the meaning and properinterpretation of information collected in the surveys, along with details concerning sampleselection techniques and the extent to which data may be properly disaggregated.

Untitled-1 5/17/02, 10:25 AM50

1 Introduction 51

data (to the extent possible) will foster use of the data in preliminary analysisand description of the project environment before pre-PPTA fact-finding.The ready availability of data at ADB can reduce time spent to collect andprocess data during the mission; at the same time, it improves the quantityand quality of the data available at this stage of loan processing. The resultsof preliminary analysis could provide material for discussion with thegovernment agencies and stakeholders concerned during the pre-PPTA fact-finding mission. By providing an empirical characterization of the targetedbeneficiaries and the environment in which they operate based on data thatwere collected using proper statistical techniques, the usefulness ofdiscussions at the fact-finding stage could be focused and made moreoperationally useful.

One particular task where using existing data could be valuable is incarrying out the initial social assessment (ISA). These data are often suitablefor use in identifying explicit poverty reducing and social developmentobjectives, outcomes, and indicators. Careful consideration of what can begleaned from previously collected data should precede design of data collectionon poverty indicators, and the design of monitoring and evaluation strategies.

At the design stage

At the PPTA stage, there is sufficient time to carry out detailed analysisof available data and to collect limited new information if necessary. Analysisat this stage of project preparation can be carried out to clarify the expectedimpact and effects of proposed projects on different stakeholders, includingthe poor. Following in-depth analysis of existing household survey data, it ispossible to assess whether collection of additional data to fill gaps isappropriate. New data collection may be necessary for confirming the loanrationale, to confirm the dimensions of poverty and vulnerability in the projectservice area, and to refine the identification of people affected by the project.New data can provide information needed for subsequent monitoring andevaluation of the project. Survey data can be used at this stage of the analysisto estimate the poverty impact ratio (wherever project net benefits can bevalued) or project cost-effectiveness for specific groups targeted by the project,particularly the poor.


Untitled-1 5/17/02, 10:25 AM51


Available Surveys Suitable for Project Level Analysis

Living Standards Measurement Surveys

The LSMS was established by the World Bank in 1980 to explore waysof improving the quality of data on indicators of household welfare collectedby government statistical offices in developing countries. The LSMS developednew methods for monitoring progress in raising the standards of living, andgathered quantitative data useful in identifying the consequences ofgovernment policies on households differentiated by their level of assets andincome, and fostered improved communication between survey statisticians,analysts, and policymakers. The LSMS also served to increase the comparabilityof welfare indicators across developing countries. The surveys gather detailedconsumption and expenditure information and a great number of quality oflife indicators needed to measure household income and assess householdwelfare and changes in welfare at national and sometimes regional level.Implementation of LSMS requires a high level of expertise in data collectionand processing, and a large commitment of resources. Typically, 2 years andmultiple visits (ranges from as few as 2, to as many as 16) to each respondenthousehold are required to carry out an LSMS. Sample sizes of the LSMSconducted are included in Table A3.2, which lists household survey data setsfor ADB's DMCs that are currently available. For additional details about thesampling procedure, instrument used, and other relevant characteristics ofthe LSMS that account for the quality and higher validity of the data,readers are referred to http://wwwhttp://wwwhttp://wwwhttp://wwwhttp://www.worldbank.or.worldbank.or.worldbank.or.worldbank.or.worldbank.org/html/prg/html/prg/html/prg/html/prg/html/prdph/lsms.dph/lsms.dph/lsms.dph/lsms.dph/lsms. Groshand Glewwe (2000) is a good source book for LSMS application to policy analysisand potentially to project analysis.

LSMS data allow measurement of poverty in terms of household money-metric income with sample validity at the national and in most cases regionallevels. Use of LSMS at levels of geographic disaggregation below the regionallevel is generally impossible due to the relatively small sample sizes. A generalrule of thumb applied in LSMS is that a minimum of 350-400 households shouldbe surveyed from each state or province of a country in generating the overallsample. The information collected in the LSMS is ideally suited for preparingdetailed country poverty profiles. The data can be potentially applied toestimate the proportion of the poor within a project's intended clientele, andto identify project stakeholders and the likely effects of the project on different

Untitled-1 5/17/02, 10:25 AM52

1 Introduction 53

stakeholder groups. Poverty incidence analysis and examination of the localand household characteristics that are highly correlated with poverty statusare other common applications of the data that are useful in project preparation.

Household Income and Expenditure Surveys

The HIES (or household budget surveys) are single topic surveys aimedat collecting detailed information on household consumption and income. Likethe LSMS, which generally includes the HIES as part of its longerquestionnaires, collection and data preparation for these surveys is expensiveand requires both time and expertise. Data are collected during multipleinterviews with the surveyed households. HIES data allow measurement ofhousehold money-metric income and poverty status, but the principal reasonsthese surveys are usually collected are for use in formulating the nationalconsumer price indices or to provide information needed to produce nationalaccounts information. Data from the surveys can sometimes be used inanalyses at the subnational level depending upon the sample size. The datacan be used to calculate headcount indices of poverty, but are of limited useby themselves in generating poverty profiles as they neither provide informationon household characteristics expected to influence household income andpoverty status nor quality of life indicators besides income and consumptionpoverty. However, it is possible to supplement HIES with data collectionmodules to gather information on household characteristics and livingconditions. Such data can be used in analyses useful for project preparationincluding assessment of the risk that income shocks will move householdsinto poverty, poverty incidence across the particular groups in the population,and examination of the local and household characteristics that are highlycorrelated with poverty status.

Rapid Monitoring Surveys (or Priority Surveys)

These surveys are similar to LSMS, but are shorter and require muchless time and fewer resources to conduct—although they are still quite costly.Usually one visit only to each household is required to collect the informationincluded in the surveys. This requires survey questions to rely on respondentrecall more than other surveys that use multiple interviews with eachrespondent, which increases measurement error. Sample sizes for them arelarger, so smaller subsamples (e.g., households with selected characteristics,or surveys collected in a particular jurisdiction) can be studied and remain


Untitled-1 5/17/02, 10:25 AM53


statistically valid. Rapid monitoring surveys have a more limited set ofquestions to measure household incomes and expenditures than the surveysdiscussed above, but yield data that allow measurement of income povertyand provide information useful in generating a country poverty profile. Likethe LSMS and HIES, data from these surveys are suited for a number ofother analyses for poverty impact assessment (e.g., poverty risk analysis,poverty correlation, and poverty incidence analysis).

Population and Housing Censuses

Data collected in periodic population and housing censuses carriedout in most DMCs include information on the demographic characteristics ofhouseholds as well as information about the literacy and education ofindividuals and basic characterization of the economic activities and housingconditions. These variables are useful in computing basic needs povertyindicators. Data from the census can be used to create detailed (low scale)maps showing the levels of basic service provision and population across allregions of a country.

Censuses usually collect no data or very limited data on householdincome, expenditure, or consumption. They cannot be used to measurehousehold income or its relationship to a poverty line. The crucial advantageof census data is that they cover all households and regions of a country, sodata from the census can be analyzed at low levels of aggregation (i.e., censusdata can provide valid information on local conditions about all householdsin an area where a project is being prepared). Used in conjunction with datafrom other surveys reviewed in this appendix, it is often possible to producegeographically disaggregated income based poverty indicators. Themethodology for combining census and income survey data to createcomprehensive and detailed maps of income poverty incidence in a countryhas been termed poverty mapping and is discussed in Box 3.2. Details areprovided in Bigman and Fofak (2001). Minot (2000) provides a case studyapplied to Viet Nam.

Untitled-1 5/17/02, 10:25 AM54

1 Introduction 55

Box 3.2

� ��+ �� -�� + ��

Poverty mapping combines survey and census data to generate poverty and inequalityprofiles at low levels of aggregation.

In the past, data shortcomings have made it impossible to generate detailed povertyand inequality profiles. On the one hand, surveys like the LSMS conducted by the WorldBank in many developing countries have solid expenditure or income data, but at lowlevels of geographic aggregation they are not representative and lack sufficient samplesize to construct poverty and inequality profiles. On the other hand, the national censusescarried out in many countries have sufficient population coverage, but do not includehigh quality information on expenditure or income.

The poverty mapping methodology requires the following steps:(i) Select all variables which exist in both the household survey and the census data

set (pay attention to variable definitions).(ii) Use the household surveys (LSMS) to run a series of linear regressions explaining

household consumption in each region that is designed to be representative. Forthese regressions, the left hand side (dependent) variable is the natural log of percapita expenditure in each household. The right hand side variables are householddemographic variables selected by the user.

(iii) Apply the estimated coefficients from the above step to the census data to imputea value of log per capita expenditure for each census household.

(iv) Use these imputed values to produce poverty or inequality profiles for the desiredaggregation units of the census data. The technique allows calculation of standarderrors for whichever welfare measure is estimated. This offers a means to assessthe statistical reliability of estimates as well as comparisons across estimates fordifferent geographic areas.Examples of poverty maps created from this procedure include South Africa,

Guatemala, Panama, Nicaragua, Ecuador and Peru, and Brazil that led to some actualprogram applications by the governments. The process of developing poverty mapresults in transfer of skills and capacity necessary to refine, revise and update the method.In the Asia-Pacific region, the methodology is being applied to Cambodia and Pakistanand is being considered for World Bank assistance as of February 2001.

Requirements for poverty mapping:- census and household survey that were conducted roughly in the same time period;- at least one person who has statistical and/or econometric skills;- 2-6 months for the whole exercise depending on quality of the data and analysts'

experience; and- a statistical analysis package capable of estimating multiple regression models.

Sources: Demombynes (2000), Henninger (1998), Lanjouw (2001), and Ozler (2000).


Untitled-1 5/17/02, 10:25 AM55


Table A3.2

,�� 0�� *��

�� -�� $%1234&333(

��

Afghanistan

Azerbaijan

Bangladesh

Bhutan

Cambodia

China, People’s Rep.of

Cook Islands

��

198919951999

198119841986198919911991199419951997199920002001

1999

1991199319971998199819992000

19821985199019901992199319941995199519981999

��

Population and Housing CensusLiving Standard Measurement SurveyPopulation and Housing Census

Population and Housing CensusHousehold Expenditure SurveyHousehold Expenditure SurveyHousehold Expenditure SurveyHousehold Expenditure SurveyPopulation and Housing CensusDemographic and Health SurveyHousehold Expenditure SurveyDemographic and Health SurveyDemographic/Health Service Provision AssessmentDemographic and Health SurveyDemographic and Health Survey (special)

Household Income and Expenditure Survey

Demographic and Health SurveySocio-Economic SurveySocio-Economic SurveyDemographic and Health Survey (special)Population and Housing CensusSocio-Economic SurveyDemographic and Health Survey

Population and Housing CensusRural/Urban Household SurveyRural/Urban Household SurveyPopulation and Housing CensusRural/Urban Household SurveyRural/Urban Household SurveyRural/Urban Household SurveyRural/Urban Household SurveySample Population and Housing CensusRural/Urban Household SurveyRural/Urban Household Survey

�� (households)

n.a.2,016n.a.

n.a.---

5,760n.a.

9,1747,4408,682

350 facilities9,854

100,000

3,854

-5,5786,000

7,630 womenn.a.

6,00013,500 women

n.a.90,980

102,138n.a.

130,780102,960102,360102,860

-100,000100,000

Untitled-1 5/17/02, 10:25 AM56

1 Introduction 57Application of Existing Primary Data inPoverty Impact Analysis at the Project LevelAPPENDIX 3

Fiji Islands

Hong Kong, China

India

Indonesia

19811986199119941996

1981series1991199319971999

19801980198119821983198419851986198719871988198919901990199119921993199319941994199519961997199719971997199819981998199819991999

Population and Housing CensusSample Population and Housing CensusPopulation and Housing CensusHousehold Expenditure SurveySample Population and Housing Census

Population and Housing CensusNational Sample SurveyPopulation and Housing CensusDemographic and Health SurveyUP-BIHAR Survey of Living ConditionsDemographic and Health Survey

SUSENAS - Indonesia's Socio-Economic SurveyPopulation and Housing CensusSUSENAS - Indonesia's Socio-Economic SurveySUSENAS - Indonesia's Socio-Economic SurveySUSENAS - Indonesia's Socio-Economic SurveySUSENAS - Indonesia's Socio-Economic SurveySUSENAS - Indonesia's Socio-Economic SurveySUSENAS - Indonesia's Socio-Economic SurveySUSENAS - Indonesia's Socio-Economic SurveyDemographic and Health SurveySUSENAS - Indonesia's Socio-Economic SurveySUSENAS - Indonesia's Socio-Economic SurveySUSENAS - Indonesia's Socio-Economic SurveyPopulation and Housing CensusDemographic and Health SurveySUSENAS - Indonesia's Socio-Economic SurveySUSENAS - Indonesia's Socio-Economic SurveyIndonesian Family Life Survey (IFLS I)SUSENAS - Indonesia's Socio-Economic SurveyDemographic and Health SurveySUSENAS - Indonesia's Socio-Economic SurveySUSENAS - Indonesia's Socio-Economic SurveySUSENAS - Indonesia's Socio-Economic SurveyDemographic and Health Survey100 Village Survey IIndonesian Family Life Survey (IFLS II)100 Village Survey II100 Village Survey IIISUSENAS - Indonesia's Socio-Economic SurveyIndonesian Family Life Survey (IFLS II+)100 Village Survey IVSUSENAS - Indonesia's Socio-Economic Survey

n.a.-

n.a.5,591

-

n.a.25,000

n.a.88,5622,250

97,282

-n.a.

---

50,000--

60,00014,142

--

46,000n.a.

26,858-

65,600 7,224

-33,738

-65,66465,66434,25512,0007,224

12,00012,00065,6642,000

12,000205,000

Untitled-1 5/17/02, 10:25 AM57


Kazakhstan

Kiribati

Korea, Rep. of

Kyrgyz Republic

Lao PDR

Malaysia

Maldives

Marshall Islands

Micronesia, Fed.States of

Series19891995199619991999

1963-2000198519951996

19891993199619961997199719991999

1985199219931995

1997-1998

19731976-1977

198019841987

1988-1989198919911991199519971999

1985199019951997

1999

2000

Family Budget SurveyPopulation and Housing CensusDemographic and Health SurveyLiving Standards Measurement SurveyPopulation and Housing CensusDemographic and Health Survey

Quarterly Income and Expenditure SurveyPopulation and Housing CensusPopulation and Housing CensusNational Survey of Family Income and Expenditure

Population and Housing CensusPoverty Monitoring SurveyPoverty Monitoring Survey (Spring)Poverty Monitoring Survey (Fall)Poverty Monitoring SurveyDemographic and Health SurveyPoverty Monitoring SurveyPopulation and Housing Census

Population and Housing CensusLao Expenditure and Consumption Survey ISocial Indicators SurveyPopulation and Housing CensusLao Expenditure and Consumption Survey II

Household Expenditure SurveyMalaysian Family Life Survey IPopulation and Housing CensusHousehold Income/Basic Amenities SurveyHousehold Income SurveyMalaysian Family Life Survey IIHousehold Income/Basic Amenities SurveyHousehold Income/Basic Amenities SurveyPopulation and Housing CensusHousehold Income/Basic Amenities SurveyHousehold Income/Basic Amenities SurveyHousehold Income/Basic Amenities Survey

Population and Housing CensusPopulation and Housing CensusPopulation and Housing CensusVulnerability and Poverty Survey

Population and Housing Census

Income and Expenditure SurveyPopulation and Housing Census

-n.a.

4,1782,000n.a.

5,844

5,200n.a.n.a.

30,000

n.a.2,1002,0002,0002,0003,672

-n.a.

n.a.2,9372,869n.a.

8,882

-1,262n.a.

12,000-

92612,000

-n.a.

39,139--

n.a.n.a.n.a.

2,600

n.a.

-n.a.


,�� 0�� *��

�� -�� $%1234&333(

Untitled-1 5/17/02, 10:25 AM58

1 Introduction 59

Mongolia

Myanmar

Nauru

Nepal

Pakistan

Papua New Guinea

Philippines

19891992-1999

19951999

1999-2000

19891997

19811984198719911991

1995-19961995-1996

19962001

19811984

1986-199119871990199119911995199619961998

19901996

19801985198819901991199319941995199519971998199819992000

Population and Housing CensusMonthly Income and Expenditure SurveyHousehold SurveyHousehold SurveyPopulation and Housing Census

Household Income and Expenditure SurveyHousehold Income and Expenditure Survey

Population and Housing CensusMultipurpose Household Budget SurveyDemographic and Health SurveyPopulation and Housing CensusRural Credit SurveyNepal Living Standards SurveyArun Living Standard SurveyDemographic and Health SurveyDemographic and Health Survey

Population and Housing CensusHousehold Integrated Economic SurveyPanel SurveyHousehold Integrated Economic SurveyHousehold Integrated Economic SurveyHousehold Integrated Economic SurveyDemographic and Health SurveyHousehold Integrated Economic SurveyHousehold Integrated Economic SurveyChild Labor SurveyPopulation and Housing Census

Urban Household Expenditure SurveyLiving Standards Household Survey

Population and Housing CensusFamily Income and Expenditure SurveyFamily Income and Expenditure SurveyPopulation and Housing CensusFamily Income and Expenditure SurveyDemographic and Health SurveyFamily Income and Expenditure SurveyPopulation and Housing CensusSurvey on Children 5-17 Years Old (SCL)Family Income and Expenditure SurveyDemographic and Health SurveyAnnual Poverty Indicators SurveyAnnual Poverty Indicators SurveyPopulation and Housing Census

n.a.1,500

--

n.a.

-25,470

n.a.5,3234,709

-7,3363,3881,2008,082

-

n.a.12,000

80012,00012,0004,8007,193

12,50012,500

-n.a.

1,2001,146

n.a.--

n.a.24,78912,99524,797

n.a.-

41,00012,407

-41,000

n.a.


Untitled-1 5/17/02, 10:25 AM59


Samoa

Singapore

Solomon Islands

Sri Lanka

Taipei,China

Tajikistan

Thailand

Tonga

Turkmenistan

1980199019921995

1981198519871990199519961999

19801990

198919992000

19081981198619871988199019901992199419961998

199719982000

Population and Housing CensusPopulation and Housing CensusHousehold Expenditure SurveyGeneral Household Survey

Population and Housing CensusLabor Force and Socio-Economic SurveyDemographic and Health SurveyHousehold Income and Expenditure SurveySocio-Economic SurveyConsumer Finance and Socio-Economic SurveyChild Labor Survey

Population and Housing CensusPopulation and Housing Census

Population and Housing CensusLiving Standards SurveyPopulation and Housing Census

Population and Housing CensusThailand Socio-Economic SurveyThailand Socio-Economic SurveyDemographic and Health SurveyThailand Socio-Economic SurveyPopulation and Housing CensusThailand Socio-Economic SurveyThailand Socio-Economic SurveyThailand Socio-Economic SurveyThailand Socio-Economic SurveyThailand Socio-Economic Survey

Living Conditions SurveyLiving Standards Measurement SurveyDemographic and Health Survey

n.a.n.a.

7,000-

n.a.19,4987,669

36,35214,00010,000

-

n.a.n.a.

n.a.2,041n.a.

---

9,045---

16,27231,143

--

2,0002,3506,303


,�� 0�� *��

�� -�� $%1234&333(

Untitled-1 5/17/02, 10:25 AM60

1 Introduction 61

Tuvalu

Uzbekistan

Vanuatu

Viet Nam

1989series1996

1989199219971998series1999

Population and Housing CensusFamily Budget Survey (annual since 1950)Demographic and Health Survey

Population and Housing CensusViet Nam Living Standards SurveyDemographic and Health SurveyViet Nam Living Standards SurveyMulti-Purpose Household Survey (annual)Population and Housing Census

n.a.-

3,703

n.a.4,8007,0016,000

45,000n.a.

Notes:n.a. – not applicable.

Sources:


Socio-Economic Surveys: World Bank, Poverty Monitoring Database (On-line). Available: �� Demographic and Health Surveys: Macro International Inc. (On-line). Available: �� !��Child Labour Surveys: International Labour Organisation (On-line). Available: ��!��!��!Population and Housing Census: US Census Bureau (On-line). Available: ��!��!��!��!��

Untitled-1 5/17/02, 10:25 AM61


Poverty maps provide spatially disaggregated measures of income-based poverty, thus overcoming one of the chief shortcomings of LSMS,HIES, and the other large surveys. These generally rely on samples that aresmall relative to the population, so their results are not amenable topartitioning as required to study poverty in a particular locality. With thepoverty maps and the data generated in their construction, staff gain accessto sufficient data on income-poverty at the household level in a given locality(i.e., the locality where a project is being considered) to characterize andanalyze poverty at the local level. Poverty maps provide useful informationfor geographic targeting and monitoring of poverty reduction projects.

Development of poverty mapping methodology at the World Bank wasone of many poverty assessment initiatives, but application of the output frompoverty maps in PPTA work in ADB's DMCs is largely untested. It is alsouncertain at this point whether it will be cost effective to gather the information,carry out the programming, and gather related resources needed for ADB togenerate poverty maps in its DMCs. Accessing primary data from nationalcensuses may represent another serious barrier to generating poverty mapsthat must be overcome in some DMCs. Accordingly, pilot testing of themethodology as part of an ongoing country poverty analysis or economicand sector work, in consultation with the World Bank, is recommended.

Demographic and Health Surveys

DHS are nationally representative household surveys that provide dataon a wide range of monitoring and impact evaluation indicators in the areasof population, health, and in some cases nutrition. DHS collect informationon assets ownership that can be used as proxies for inferring levels of householdincome and wealth, but does not allow direct income-poverty measurement(see Box 3.3). Calculated wealth indices can be used to classify householdsinto wealth quintiles. However, it is not appropriate to compare such wealthindices to absolute poverty lines, so they are not suitable for povertymeasurement. DHS may be used for producing poverty maps, based on“relative” poverty measure. The permissible level of spatial disaggregationof such maps depends on the sample size of the DHS. Other poverty analysesthat have used DHS data are: poverty risk analysis; generating a poverty profile(i.e., analyzing the relationship of wealth quintile with levels of education,school enrollment, health outcomes, basic service access, basic labor marketcharacteristics); and poverty incidence analysis (i.e., distribution of health,

Untitled-1 5/17/02, 10:25 AM62

1 Introduction 63

education, specific program spending by area, and households’ wealthquintiles). For more information see wwwwwwwwwwwwwww.measur.measur.measur.measur.measuredhs.com.edhs.com.edhs.com.edhs.com.edhs.com.

Core Welfare Indicators Questionnaire Surveys

Developed jointly by the World Bank (Africa Region) in collaborationwith UNDP and UNICEF, the CWIQ is a short and relatively easy to administersurvey that monitors social indicators on an annual basis. The CWIQ is designedto complement rather than replace other more intensive surveys (Box 3.4).The CWIQ does not try to collect data on household income or expenditure.As a result, the CWIQ may not be used for money-metric measurement ofpoverty. Household well-being is assessed using poverty predictors, which

Box 3.3

�� +)�� + �� *��

Using DHS data, the World Bank produced country reports on health, nutrition, population(HNP) and poverty, which provide statistics concerning intra-country differences betweenrich and poor with respect to HNP service and status use. The method has been appliedto Bangladesh, India, Kazakhstan, Kyrgyz Republic, Nepal, Pakistan, Philippines, Uzbekistan,and Viet Nam.

Socioeconomic status is defined in terms of assets or wealth, rather than in termsof income or consumption. Each household asset for which information was collectedthrough the DHS was assigned a weight or factor score generated through principalcomponents analysis. The resulting asset scores were standardized in relation to a standardnormal distribution with a mean of zero and a standard deviation of one. These standardizedscores were then used to create the break points that define wealth quintiles as follows.

Each household was assigned a standardized score for each asset, where the scorediffered depending on whether or not the household owned that asset (or, in the caseof sleeping arrangements, the number of people per room). These scores were summedby household, and individuals were ranked according to the total score of the householdin which they resided. The sample was then divided into population quintiles—five groupswith the same number of individuals in each.

The method has been applied to Bangladesh, India, Kazakhstan, Kyrgyz Republic,Nepal, Pakistan, Philippines, Uzbekistan, and Viet Nam.

The resulting figures describe the HNP status and service use among individualsbelonging to different socioeconomic classes. The figures are intended to provide WorldBank operational staff, the government officials with whom they work, and others withbasic information for use in preparing country analyses and in developing HNP activitiesfor the disadvantaged.Source: -��.//000�0��/� ��/-��-/��/��1�-��


Untitled-1 5/17/02, 10:25 AM63


are variables from the larger survey that are strongly correlated with povertystatus and are easy to collect. For example, analysis of the larger surveymay indicate a strong correlation between the level of education of thehousehold head, tractor ownership, and ease of access to health services,and the household’s poverty status. The CWIQ would then collect informationon these household characteristics.

CWIQ survey data are well suited for monitoring achievements indevelopment objectives defined through leading indicators (e.g., indicatorsthat give advance warning of a future impact such as household access anduse of public services). The CWIQ can be used for estimating the proportionof the poor—as defined by poverty predictors—within the project's intendedclientele, and to identify the likely effects of the project on different beneficiaries.

Box 3.4

��+ �� )��

The poverty predictors are selected from a host of variables constructed from anexisting database. The poverty predictors are constructed under the assumption thatprior to the implementation of the CWIQ, a more comprehensive household survey(e.g., LSMS) has already been carried out in the country. It would also be desirable fora participatory study on poverty analysis to have been constructed in the country toprovide some indication on the determinants of poverty and their variation across regions.

A combination of information from these two different sources—qualitative studiesand more comprehensive household survey—is used to identify the set of explanatoryvariables that are closely correlated to household aggregated total expenditure. A maximumnumber of potential explanatory variables, and the predicted variable of interest (aggregatedtotal household or expenditure) are constructed at this stage of the process.

A regression model will be built with the natural logarithm of the total householdexpenditure (or income) as the dependent variable. The first task is to carry out a multiplecorrelation analysis assessing the magnitude of correlation between the potential regressorsand the dependent variable. The multiple correlation analysis identifies a host of variablesthat are highly correlated with the dependent variable. Further reduction of thedimensionality of the model is attempted using stepwise procedure in the regressionanalysis setting (gradually dropping explanatory variables with less predictive power).

This iterative process of gradual elimination allows one to reduce the number ofpredictors to a core set of variables that can be collected easily. For practical reasons,it is suggested that the core set of poverty predictors could well be limited to 10 variablesas well.

Source: World Bank (August 1999, Annex 4).

Untitled-1 5/17/02, 10:25 AM64

1 Introduction 65

CWIQ surveys can be carried out as national surveys, but are primarilydesigned for use at a more limited geographic scale to collect data neededfor project monitoring and evaluation. It does not attempt to measure theimpact of projects on living standards, but contributes to monitor outcomeindicators (e.g., access, usage, and satisfaction with basic services provided).CWIQ surveys collect indicators of household well-being and indicators ofhousehold access/usage/satisfaction with community and other basic services.Other features include: large sample sizes, a short questionnaire that can beadministered quickly during a single visit to each respondent household, quickdata entry and validation, and simple reporting of results. Additionalinformation about the CWIQ is available on-line at http://www4.worldbank.org/afr/stats/cwiq.cfm.afr/stats/cwiq.cfm.afr/stats/cwiq.cfm.afr/stats/cwiq.cfm.afr/stats/cwiq.cfm.

Multiple Indicator Cluster Surveys

MICS have been developed by UNICEF to be low-cost household surveysthat quickly generate data on key welfare indicators inadequately monitoredin other data collection systems. MICS have been implemented in more than100 countries since 1995. UNICEF, with partner agencies, defined a set ofindicators to guide the assessment process (UNICEF 1995). Indicators includemeasures meant to monitor progress in established health, education andwelfare, and new indicators intended to fill gaps in information concerningthe welfare of children in developing countries (e.g., child labor, birthregistration, disability, orphans/alternative family care, and early childdevelopment). In collaboration with a number of other agencies, UNICEFhas harmonized the MICS with other major survey programs (e.g., USAID-sponsored demographic and health surveys) to generate comparable andcomplementary data across countries. MICS data does not allow measurementof poverty based on income (money-metric), but they can be used to generatepopulation profiles regarding health, education and child labor indicatorsusable as proxy variables of poverty status.

Collecting Project-Specific Primary Data

If after reviewing available data, ADB staff find that additional data arerequired to provide the information necessary for project design, projectimplementation, or the ex-ante or ex-post assessment of a project's effects onpoverty, primary data collection is appropriate. ADB can support datacollection through technical assistance activities.


Untitled-1 5/17/02, 10:25 AM65

ADB development of new instruments and procedures specific for itsprojects should proceed from the established surveys discussed in thisappendix and focus on priority topics. In such efforts, ADB would be welladvised to rely on existing questionnaires (e.g., LSMS) and methodologies(e.g., CWIQ survey development or poverty mapping), as these have beenwidely applied, tested, and validated. Where LSMS or HIES are available,the CWIQ provides a good model for how to develop a well focused and easy-to-collect survey that provides basic poverty measures based on validatedproxies/predictors. Existing surveys can be supplemented with modules toaddress project-specific concerns such as measuring user willingness-to-payfor services, identification of priorities by the population, time use, access toservices, assets ownership, and similar information.

In countries where such data exist, it may be sufficient to collect limitedadditional data, specific to the project's requirement. Well-focused and easyto administer surveys can be conducted to provide baseline and follow-upindicators. Several aid agencies have developed data collection instrumentsspecific to their targeting and monitoring needs: UNICEF developed the MICS,ILO the Statistical Information and Monitoring Programme on Child Labour(SIMPOC), the World Bank the LSMS and the CWIQ, USAID (through MacroInternational Inc.) the DHS and service provision assessment surveys, amongothers.

In countries where no reasonably recent data on poverty are available,implementation of a large LSMS or HIES is likely necessary to generate basicmeasures of income and money-metric poverty in the country. Such informationis essential for setting poverty lines and producing a poverty profile. Theexpense and the commitment of time and resources to carry out such surveyspreclude their implementation for a particular project.


Untitled-1 5/17/02, 10:25 AM66

APPENDIX�4Case Illustrations of Distribution Analysis

The following two cases illustrate the application of distribution analysis.The first from Mongolia is based on an actual report and recommendation ofthe President (RRP). The second from Bangladesh is a research studycommissioned by EDRC and based on the original RRP appraisal. In bothcases the distribution analysis provides a starting point for poverty impactassessments. It should be noted that the purpose of the illustrations here isprimarily to demonstrate that there have been actual cases of application toADB projects. They do not necessarily represent best practices. Project analystsare encouraged to explore improvement on the practice, for example, byincorporating the project financing aspect illustrated in Appendix 5.

Case 1– Mongolia: Energy Conservation Project

The following is a rework of the financial and economic analysispresented in the ADB RRP on a loan to Mongolia for the Energy ConservationProject.1 Only the relevant information on financial and economic analysis isdrawn from the document. To simplify the exposition a few amendments havebeen made but the analysis is basically set out in the RRP. A similar analysiscan be found in the RRP for the Ulaanbaatar Heat Efficiency Project.2

Project Description

Ulaanbaatar is the coldest capital in the world. Because of its harshclimate, heat is a basic human need. However, a serious deterioration in theefficiency and reliability of heat and electricity supply is affecting the people's

Untitled-1 5/17/02, 10:25 AM67


livelihood. The heating system was put into operation in 1953 and much ofthe piping has reached the end of its technical life. Because of leaks in thesystem, there are huge losses of hot water, estimated at 4.1 million tons peryear. Also, following the collapse of the power system in February 1996,frequent electricity outages are adversely affecting industry, commercialenterprises, and also residential consumers who suffer lack of lighting andare unable to use appliances during peak periods. There are only limited backupfacilities installed in hospitals and other essential services. The objectives ofthe Project are to (i) improve district heating reliability and reduce losses byrehabilitating critical sections of the district heating system in Ulaanbaatar;(ii) encourage end-use energy conservation through improved metering andthrough demonstration projects; and (iii) improve district heating systemoperation and maintenance through on-the-job training and technical support.The Project is expected to reduce district heating losses by an average of 11percent (equivalent to 192 GWh/year) and electricity losses by 8 percent(equivalent to 30 GWh/year). The total project cost of $13.19 million will befinanced by ADB ($10 million) and the Government ($3.19 million).

Least-Cost Analysis

There is no practical alternative to the Project given the urgent need formore efficient and reliable electricity and heat supply and the financial constraintson building new Energy Authority (EA) facilities. On an incremental basis, theProject will save both heat and electricity at a lower per unit cost ($6.1 per Gcaland $0.013 per kWh, respectively) compared with generating heat and electricityfrom a new plant ($9.8 per Gcal and $0.046 per kWh, respectively).

Financial and Economic Analysis

All calculations are in US dollars and the assumptions used in thefinancial and economic analysis are as follows.

(i) Project life of 20 years.(ii) An estimated cost (inclusive of physical contingency) of $11.8

million at constant prices.

Untitled-1 5/17/02, 10:25 AM68

1 Introduction 69APPENDIX 4 Case Illustrations of Distribution Analysis 69

(iii) Reduced district heat loss savings due to rehabilitated piping (heatloss reduction in distribution) and metering/demonstration projects(reduced consumer wastage), respectively, and reduced electricitylosses due to more efficient heating pumps (technical lossreduction) and due to transmission and substation metering (non-technical loss reduction). The savings involved are set out in TableA4.1.

(iv) Financial prices of heat and electricity are $7.3 per Gcal and $0.059per kWh, respectively. These are prices current at the time ofplanning the project adjusted for real tariff movements in futureyears.

(v) Income tax of 40 percent is applied to the net positive cash flowsafter allowance of 5 percent depreciation on the capital cost.

(vi) Economic prices of heat and power are determined by estimatedwillingness to pay. Based on an upper bound of $38.4 per Gcalassociated with the observed usage of electric bar radiators anda lower bound of $2.7 per Gcal which is the average price paidfor apartment central heating, and a curvature correction factorof 0.3, the willingness to pay for heat is estimated at $13.4 perGcal. The real heat tariff is 54 percent of this figure. Similarly,based on an upper bound of $0.12 per kWh for diesel generationand a lower bound of $0.033 per kWh which is the present averagepower tariff, and a curvature correction factor of 0.3, willingnessto pay for electricity is estimated at $0.059 per kWh. Reducedtechnical electricity losses are valued at this economic price, butreduced nontechnical losses due to better metering are given novalue in the economic analysis as they represent a transferbetween EA and users who had previously been nonmetered. Itis assumed in the RRP that the real electricity tariff will be increasedto the willingness to pay for electricity.

(vii) A standard conversion factor (SCF) of 1.0 is used for Mongolia.This is almost certainly a simplification, which is not pursued here,as the purpose of this appendix is to illustrate distribution andpoverty analysis. However, use of an SCF of unity has twoimportant implications; first since no economic adjustments aremade to project costs use of resources in constructing the projectcreates no distributional effects and second the analysis is identicalwhether one uses a domestic or a world price numeraire. Hencethese two sources of complication are removed in this case.

Untitled-1 5/17/02, 10:25 AM69


FIRR and EIRR calculations are shown in Table A4.1. FIRR at 16 percentwas considered financially viable at appraisal without the benefit of anestimated financial opportunity cost of capital. The weighted average cost ofProject capital was considered difficult, given the variable economic conditionsprevailing in Mongolia and lack of an indicative set of interest rates and ratesof return at the time of appraisal. EIRR at 38 percent, being comfortably higherthan the benchmark 12 percent, was considered economically viable.

Table A4.1

#�� "�� "�� +� ��

($ million)

!�� % 2��3 % 5 %1

�� #��6

Heat loss reduction 4.69 0.44 0.58 0.58

Consumer heat wastage reduction 5.10 0.47 0.63 0.63

Technical electricity loss reduction 9.87 1.18 1.18 1.18

Nontechnical electricity loss reduction 4.94 0.59 0.59 0.59

�� #��7��6

Capital cost 11.80 11.80 0.00 0.00

Lost consumer heat revenue 5.10 0.47 0.63 0.63

Profits tax 5.18 0.00 0.70 0.70

Net financial flow (9.12) 2.27 2.27

Financial NPV 2.52

Financial IRR (%) 16.00

�� "��!��


Consumer heat wastage reduction 9.36 0.87 1.15 1.15

Technical electricity loss reduction 9.87 1.18 1.18 1.18

Nontechnical electricity loss reduction 0.00 0.00 0.00 0.00

�� "��

Capital costs 11.80 11.80 0.00 0.00

Lost consumer heat revenue 0.00 0.00 0.00 0.00

Profits tax 0.00 0.00 0.00 0.00

��"��!�� (8.94) 3.39 3.39

Economic NPV 16.03

Economic IRR (%) 38.00

Untitled-1 5/17/02, 10:25 AM70

1 Introduction 71

Distribution Analysis

It should be noted that unlike the general case in which the domesticprice numeraire should be used in distribution analysis, the world pricenumeraire in dollar unit was used for this project as the cost estimate wasdone in US dollars. However, as we have noted, the use of an SCF of 1.0 removesthe problem associated with applying distribution analysis to appraisals atworld prices (since by implication the average divergence between worldand domestic prices is zero).

Using the results in Table A4.1 and following the procedure explainedin this paper, we can lay out the distribution of benefits among the basicstakeholders: Energy Authority (EA) as the operating entity, consumers, andthe Government. The result is in Table A4.2 showing how the net economicbenefits accrue to each stakeholder category. A discount rate of 12 percentis used to calculate net present values of both financial and economic flows.EA gains the financial NPV of the project of $2.52 million. This gain arisesfrom the savings in heat and power losses valued at the tariffs minus theinvestment cost of the project and taxes paid on operating profits. Heat wastagesavings to consumers are not a gain to EA and hence in the financial NPV forpresentational purposes—these are shown twice as part of total benefits andas part of costs (in parenthesis) with the two values balancing out. The economicNPV exceeds the financial NPV by $13.51 million. Out of this, $5.18 milliongoes to the Government through profits tax with the remaining $8.34 milliongoing to consumers. The consumers benefit from the fact that they will receiveheat at a tariff that is below their willingness to pay. The tariff is approximately54 percent of willingness to pay, hence for heat loss reduction consumer gainsare roughly 46 percent of the economic value of heat. For reductions inconsumer heat wastage, consumers gain the energy saved valued at theirwillingness to pay, although as they will now buy less from the EA as a resultof these savings the EA will lose the energy involved valued at the heat tariff.Reductions in technical power losses bring no net gain to consumers as byassumption their willingness to pay equals the tariff. For nontechnical powerlosses due to improved metering consumers must pay more while the EAgains this additional revenue, so there is no net economic effect. Part of thistransfer will circulate back to the Government through the profits tax paid bythe EA. No one loses from the Project.

APPENDIX 4 Case Illustrations of Distribution Analysis 71

Untitled-1 5/17/02, 10:25 AM71


Table A4.2

�� -��+�� "�� +� ��

($ million)

PV (1) (2) (2) - (1) Distribution

Financial Economic Difference EA Gov't. Consumers

Incremental benefits


Consumer heat wastage 5.10 9.36 4.26 (5.10) 9.36

Technical electricity loss 9.87 9.87 0.00

Nontechnical electricity loss 4.94 0.00 (4.94) (4.94)

Incremental costs

Capital costs (11.80) (11.80) 0.00

Lost consumer heat revenue (5.10) 0.00 5.10 5.10

Profits tax (5.18) 0.00 5.18 5.18

Net Benefits 2.52 16.03 13.51 0 5.18 8.33

Gains and losses 2.52 5.18 8.33

Notes:

1. EA gains financial NPV.

2. Gov't. gains Profits tax.

3. Consumers gain: — difference between heat loss reduction valued at willingness to pay

($13.4/Gcal) and heat loss reduction valued at heat tariff ($7.3/Gcal)

— saving in consumer heat wastage valued at heat tariff ($7.3/Gcal)

4. Consumers lose: — reduction in nontechnical losses due to improved metering valued at power

tariff ($0.59/kWh).

The result here indicates that the distributional issue does not pose aproblem in selecting the Project as there will be no losers and no need forcompensation for any party. If we were to proceed further to carry out povertyimpact analysis, as explained in the text, further assumptions on the proportionof the poor within each stakeholder category would have to be made. Doingso for the benefits going to the consumers would be relatively straightforwardas data for customer use should be available at the Energy Authority. But forthe benefits accruing to the Government the issue is more problematic.However, the general point is that the original RRP was able to take the firststep in accurately estimating the project's distribution impact. A poverty profileof power consumers would have allowed the consumer category to be splitbetween the poor and nonpoor. An example of an integrated distribution and

Untitled-1 5/17/02, 10:25 AM72

1 Introduction 73

poverty impact analysis for a power sector from a more recent project is givenin Case 1 in Appendix 6.

Simplification in Financial Analysis

The format used in Table A4.1 follows conventional ADB practice infinancial calculations. This involves estimating financial inflows and outflowsand deducting profits tax to give a net-of-tax FIRR. While this approach isoften adequate for most conventional appraisals it is not strictly what isrequired for distribution analysis. This is because it makes no allowance forthe way the project is financed. The FIRR is a net-of-tax return to total capitalrather than a return to the equity funds committed by project owners. If thefinancing plan of a project is known then for purposes of distribution analysisthe financial inflows should include the loan receipts and the financial outflowsshould include loan principal and interest repayments. However, it should benoted that if the loan is at the same real interest rate as the discount rate forthe project (i.e., if there is no subsidy component) then the present value ofthe loan receipts and the present value of the sum of loan repayments shouldbe equal, provided there is no grace period. Hence under this situation theomission of these loan flows, as in Table A4.1, will make no difference to theresult of the distribution analysis. However, if the loan real interest rate isbelow the discount rate, the gain to owners (i.e., the return to equity) will begreater than the return to the project as a whole and there will be transferbetween lenders and the owners of the project, which will be relevant indistribution analysis.

A further complication arises because if the financing arrangementsfor a project are not allowed for in the financial analysis the true liability forProfits tax cannot be established. This is because in the tax regulations ofmost countries interest payments as well as depreciation allowances are coststhat are allowed as deductions against tax. Therefore, omission of interestpayments will lead to an overstatement of the tax liability of owners and henceto an underestimate of their gains from the project. The financial NPV of aproject, as currently calculated in ADB appraisals and illustrated in Table A4.1,may thus not fully reflect the change in income for project owners.


Untitled-1 5/17/02, 10:25 AM73


Case 2– Bangladesh: Jamuna Bridge Project

The following is taken from an unpublished EDRC report. The informationand data used in the paper are based on the ADB Appraisal Report on a loanto Bangladesh for the Jamuna Bridge Project.3

Project Description

The Jamuna, the Meghna and the Padma constitute a system of riverswhich physically divides Bangladesh into East, Southeast, and Northeastregions. The existing transport services such as ferries and boats going acrossthe Jamuna River face numerous problems: a great amount of silt brought byexcessive water flow during the monsoon season; riverbank erosions; oftentoo narrow navigation channels; deteriorating fleet conditions due to the lackof maintenance; and poor parking facilities at jetties. The whole ferry systemhas reached its capacity limits. The Project is designed to provide a permanent,nonmanagement intensive crossing under all-weather conditions for theexisting and potential East-Northwest traffic. It includes the followingcomponents: (i) a 4.8 km-long and 18.5 m-wide bridge to carry four road laneswith sidewalks; (ii) two viaducts, about 128 m-high; (iii) two guide bunds,about 2.2 km each and a flood protection bund on the east bank; (iv) twoapproach roads, about 16 km to the east and 14 km to the west (two-lanesingle carriageway with paved shoulders); (v) measures to mitigate the project'senvironmental impact; (vi) implementation of a resettlement plan; and (vii)technical assistance for project management. The bridge site, 7 km south ofSirajganj, was selected after extensive studies of 10 potential sites. In-depthstudies of the selected site were carried out to determine the optimal designof the bridge and the required embankments. The total project cost of $696million was to be co-financed by ADB, Overseas Economic Cooperation Fund(Japan), International Development Association (World Bank), and theGovernment.

3 Loan 1298-BAN:Jamuna Bridge Project, for $200 million, approved on 8 March 1994.

Untitled-1 5/17/02, 10:25 AM74

1 Introduction 75APPENDIX 4 Case Illustrations of Distribution Analysis 75


The project life is considered to be 50 years after the bridge opens totraffic in 1998 (scheduled at the time of appraisal). On the basis of the pasttraffic growth trends and taking into account population and per capita GDPgrowth rates as well as estimated price and income elasticity of demand fortransport services, the annual traffic growth rates from 1993 to 1998 areestimated at 6.6 percent for buses and trucks, and 8.2 percent for light vehicles.Based on these data, the traffic diverted to the Jamuna bridge in 1998 wasprojected for the three categories of vehicles. The bridge traffic is estimatedto grow at 5 percent per year during 1998–2025 and no increase thereafteras the bridge capacity would be fully exhausted by 2025. The reduction inwaiting time and vehicle operating costs would generate additional passengerand freight traffic on the bridge. The base year (1998) newly generated trafficwas estimated based on the price elasticity of demand for transport services(-1.0 for light vehicles, -1.5 for buses, and -0.6 for trucks, respectively). Theincremental generated traffic for the three types of vehicles is assumed tobuild up gradually from 20 percent of the total volume in 1998 to 40 percent,50 percent, 60 percent, 70 percent, 80 percent, 90 percent, and finally 100percent in 2005; thereafter the growth rate of the incremental traffic is assumedto be the same as the diverted traffic.

Financial revenue for the bridge is based on the toll revenues fromdiverted and newly generated traffic, and electricity interconnector fees. Tolltariffs for the three types of vehicles are set at half the current ferry tariff,respectively. It is assumed that the nominal tariff rates are increased everyyear to remain constant in real terms. Financial costs are the investment costplus operating and maintenance costs of the bridge. The financial NPV ispositive, showing financial viability of the Project. Sensitivity test of the financialNPV shows that it becomes negative if the tariff rate is reduced below 30percent of the current ferry rate, or if the vehicle traffic growth rate goesbelow 2 percent.

Economic benefits of the Project include (i) the savings in vehicleoperating costs (VOCs) gained by the diverted traffic; (ii) the value of timesaved for existing passenger and freight traffic; (iii) the benefits received bythe generated traffic (equal to the gain in consumer surplus plus the financialtoll revenue); (iv) the value of the investment saved by not constructing astand-alone power interconnector; (v) the value of the investment, operation

Untitled-1 5/17/02, 10:25 AM75


and maintenance saved by not improving the current ferry system; (vi) thevalue of the increase in truck waiting time saved from year 2000 onwards bynot operating the current ferry system; (vii) the environmental benefits ofpreventing embankment erosion in areas close to the bridge and increasingagricultural production during the monsoon season.

The estimates for VOC by kind of vehicles are based on the BangladeshRoad Master Plan Study (1992) and the Nalka-Bonpara Road Feasibility Studydeveloped by the China International Engineering Consulting Corporation(1993). The estimates of economic savings on passenger trips are based onthe individuals' hourly income. The analysis followed the practice of the IndianRoad Manual: valuing business travelers' time at 110 percent of their hourlyincome, the time of the remaining economically active passengers at 50 percentof their hourly income, and the time of nonbusiness travelers at 25 percentof their hourly income. The hourly income of each of these three groups wasestimated using the Bangladesh Household Survey for 1985–86 updated to1993 price levels by CPI and the annual growth rate of 1.7 percent in percapita GDP observed in the period of 1968–1990. The time savings for freighttraffic is reflected in the reduction of inventory costs, losses, and damages(especially to perishable agricultural goods). Because of unavailability of reliabledata, these indirect time-related costs are assumed to be equal to 10 percentof VOC. The direct benefits accruing to the diverted traffic are estimated bymultiplying the unit savings in VOC and waiting time costs for the types ofvehicles by their corresponding traffic level. The gain in consumer surplus tothe newly generated traffic is obtained by multiplying half the unit differentialof total cost per type of vehicles by their corresponding incremental traffic.The operating costs saved by type of vehicle due to the Project are thenadjusted by their corresponding conversion factors (weighted averages ofconversion factors of the VOC components) in order to get the real resourcevalue of the savings and increase in consumer surplus. The time savings arenot adjusted as labor is assumed to receive its market price (SWRF = 1).

The real economic values of the investment cost items are adjustedusing appropriate conversion factors. The shadow exchange rate factor of1.304 (based on the 1993 EDRC shadow pricing study for Bangladesh,Indonesia and Philippines (Jenkins 1993) is used for converting the value oftradables to the domestic price numeraire. At the economic discount rate of12.21 percent (based on the same study), the economic net present value ofthe Project is estimated at taka(Tk)7.77 billion, showing economic viability.Sensitivity test of the economic NPV shows that it becomes negative if thetraffic growth rate goes below 3 percent or if the investment cost overrunsby more than 25 percent.

Untitled-1 5/17/02, 10:25 AM76

1 Introduction 77


In the Project's case, the difference between financial and economicflows derives from two factors: (i) some project inputs and outputs havingconversion factors different from unity; and (ii) the Project generates economicbenefits that are not captured as financial benefits. Net stream of both financialand economic flows are discounted at the economic discount rate of 12.21percent. The analysis of allocation of the difference of economic from financialNPV is presented in Table A4.3. Light vehicle owners, bus passengers, truckers/shippers, power company, and the locality gain from the Project, while theGovernment/aid agencies and ferry operators lose.

Light vehicle owners and bus passengers would gain Tk627 millionand Tk1,951 million, respectively. Truckers/shippers would be the largest gainerand realize savings of about Tk31,094 million. The bridge crossing site andits surroundings are estimated to gain Tk457 million of environmental benefitsfrom the increased net incomes from enhanced agricultural production andthe prevention of embankment erosion. The Government/aid agencies wouldlose about Tk27,700 million. This is mainly due to the subsidy on the loan(Tk19,851 million), the government grant (Tk2,455 million), and the premiumlost on the foreign exchange used to purchase tradable components of theinvestment cost of the bridge (Tk5,358 million). The Government also losestax and tariff revenues on the imported components of VOC saved becauseof the Project. The ferry operators would lose Tk1,840 million as the ferryservices are replaced by the bridge crossing.

Poverty Impact

These adjustments allow us to identify the direct income effects createdby the bridge project. By far the most important are the gain to bridge users,which is predominantly in the form of freight traffic, and the loss to theGovernment which meets the cost of the project (and gains the toll revenue).This is equivalent to the distribution analysis of Appendix 25 of the ADBGuidelines (1997). Going beyond this to estimate both the indirect incomeeffects and the project poverty impact requires further information.


Untitled-1 5/17/02, 10:25 AM77


The indirect income effects of the Project arise from the fact that incometo both the freight sector and the Government may ultimately be used in avariety of ways. In terms of savings in trucking costs at one extreme thesemay all stay within the sector as higher profits and at the other they may beall passed on to end users as lower prices. In practice, a combination of thesetwo outcomes is likely, with the share going to consumers varying directlywith the degree of competition in the sector. The issue then becomes one ofhow many consumers affected by the Project are poor and whether higherfreight profits are used in a way that raises the incomes of the poor. Tracingparticular income effects too far back in the productive chain is likely to befraught with difficulty and it is best to avoid linking multiplier effects via higherexpenditure with any individual project on the grounds that if it is simplyhigher expenditure that causes higher incomes then such effects can beobtained from any project, not one in particular. The issue of how additionalgovernment income affects the poor is discussed in the main text.

Where additional or generated traffic is involved, as in this project,economic value is approximated conventionally by half unit VOC savings. Thiswill be valid in a competitive market structure where willingness to pay fortransport services (as proxied by half VOC) equals the net economic gainfrom using those services. For generated traffic therefore consumer surplus(the excess of willingness to pay over actual payment) will be divided betweenproducers, whose output is stimulated by the project and the freight sector,depending upon the pricing structure of the latter. Again the more competitiveis the sector the higher is the proportion of consumer surplus that will accrueto producers. Tracing through these effects will require data on both productionactivity stimulated by the bridge and on the pricing strategy in the freightsector.

As an alternative however one can adopt a set of simplifying assumptionsin order to get a first check on how economic returns relate to poverty impact.These are

(i) that the transport sector is competitive so that ultimately lowerfreight costs are passed on in lower end-user prices. The gain tothe poor from lower VOC on normal and diverted traffic could beallocated in proportion to their share in total income. If the projectinvolved is an expressway then figures can be national ratherthan local;

(ii) that consumer surplus for generated traffic accrues solely toproducers (as consumers of road services) whose production is

Untitled-1 5/17/02, 10:25 AM78

1 Introduction 79

stimulated by the fall in VOC created by the bridge, due to thecompetitive nature of the freight sector. If agricultural goods areinvolved these gains can be allocated to the poor in proportionto their share in land in the area of influence of the bridge. Ifnonagricultural activity is involved the share of the poor could beproportional to their share in nonagricultural employment in thearea of influence; and

(iii) that only the poor are employed in the construction of the bridgeand that they gain the difference between the wage paid andtheir economic wage or opportunity cost.

The aim would be to use the basic data required for an economicappraisal and adjust this to obtain an approximate poverty impact. The finalresults could be refined if better data could be collected. The latest examplesof poverty impact analysis applied to ADB transport projects include TajikistanRoad Rehabilitation Project (Case 5 in Appendix 6). As the above examplewas taken from a study prior to the current ADB Guidelines, it uses a discountrate different from 12 percent. The current ADB practice is to use 12 percent.


4 Loan 1819-TAJ: Road Rehabilitation, for $20 million, approved on 20 December 2000.

Untitled-1 5/17/02, 10:25 AM79


Item Financial Financial PV Economic Economic

PV at 10% at 12.21% PV at 12.21% minus

Cost savings and increase in consumer surplus

Light vehicles 606.2 606.2

Buses 1,904.7 1,904.7

Trucks 27,181.1 27,181.1

Avoided truck waiting time (from 2000) 3,157.3 3,157.3

Toll revenues

Diverted traffic

Light vehicles 105 74.7 (74.7)

Buses 826.8 588.3 (588.3)

Trucks 1,655.3 1,177.8 (1,177.8)

Generated Traffic

Light vehicles 42.5 29.4 29.4 0

Buses 293.7 203.6 203.6 0

Trucks 905.1 627.4 627.4 0

Electricity interconnector fees 461.3 344.1 (344.1)

Savings on stand-alone 2,888.3 2,888.3

power interconnector

Savings from avoided

costs of ferry system

Ferry purchases 327.4 327.4

Dredging and maintenance 460.1 460.1

Environmental benefits 456.9 456.9

Loan 19,630.6 19,851.2 (1,985.1)

Government grant 2,489.1 2,455.4 (2,455.4)

TTTTTotal Benefitsotal Benefitsotal Benefitsotal Benefitsotal Benefits 26,409.426,409.426,409.426,409.426,409.4 25,351.925,351.925,351.925,351.925,351.9 37,842.437,842.437,842.437,842.437,842.4 12,490.512,490.512,490.512,490.512,490.5

Investment costsInvestment costsInvestment costsInvestment costsInvestment costs

Main bridge (9,050.3) (8,898.5) (11,178.7) (2,280.2)

River training works (9,943.6) (9,794.3) (12,396.1) (2,601.8)

Approach roads (2,090.9) (2,056.9) (2,418.3) (361.4)

Technical assistance (919.7) (901.2) (1,016.2) (115.0)

Others (2,048.7) (2,041.2) (2,041.2) 0

Operating and maintenance costs (1,278.9) (1,017.2) (1,017.2) 0

TTTTTotal Costsotal Costsotal Costsotal Costsotal Costs (25,332.1)(25,332.1)(25,332.1)(25,332.1)(25,332.1) (24,709.3)(24,709.3)(24,709.3)(24,709.3)(24,709.3) (30,067.7)(30,067.7)(30,067.7)(30,067.7)(30,067.7) (5,358.4)(5,358.4)(5,358.4)(5,358.4)(5,358.4)

Net BenefitsNet BenefitsNet BenefitsNet BenefitsNet Benefits 1,077.31,077.31,077.31,077.31,077.3 642 .6642 .6642 .6642 .6642 .6 7,774.77,774.77,774.77,774.77,774.7 7,132.17,132.17,132.17,132.17,132.1

�� 89:; �� !��+�� <�� ! ��+� � ��

$�/ ��= %1>8(

Financial

Untitled-1 5/17/02, 10:25 AM80

1 Introduction 81

onomic Light Vehicle Bus Truckers/ Power Government Locality Ferry

inus Passengers Passengers Shippers Company Operators

06.2 627 (20.8)

04.7 1,951.6 (46.9)

81.1 27,936.8 (755.7)

57.3 3,157.3

4.7) (74.7)

8.3) (588.3)

7.8) (1,177.8)

0 0

0 0

0 0

4.1) (344.1)

88.3 2,888.3

27.4 327.4

60.1 460.1

56.9 456.9

5.1) (1,9851)

5.4) (2,455.4)

90.590.590.590.590.5 6 2 76 2 76 2 76 2 76 2 7 1,951.61,951.61,951.61,951.61,951.6 31,094.131,094.131,094.131,094.131,094.1 2,544.22,544.22,544.22,544.22,544.2 (((((2 , 2 3 4 32 , 2 3 4 32 , 2 3 4 32 , 2 3 4 32 , 2 3 4 3 ))))) 456 .9456 .9456 .9456 .9456 .9 -1,840.8-1,840.8-1,840.8-1,840.8-1,840.8

0.2) (2,280.2)

1.8) (2,601.8)

1.4) (361.4)

5.0) (115.0)

0 0

0 0

8.4)8.4)8.4)8.4)8.4) 00000 00000 00000 00000 (5,358.4)(5,358.4)(5,358.4)(5,358.4)(5,358.4) 00000 00000

32.132.132.132.132.1 6 2 76 2 76 2 76 2 76 2 7 1,951.61,951.61,951.61,951.61,951.6 31,094.131,094.131,094.131,094.131,094.1 2,544.22,544.22,544.22,544.22,544.2 (27,701.0)(27,701.0)(27,701.0)(27,701.0)(27,701.0) 456 .9456 .9456 .9456 .9456 .9 (1,840.8)(1,840.8)(1,840.8)(1,840.8)(1,840.8)


ancial

Untitled-1 5/17/02, 10:25 AM81


Untitled-1 5/17/02, 10:25 AM82

APPENDIX�5Incorporating Project Financing in

Distribution and Poverty Impact Analysis

To clarify the link between financial analysis of a project and distributionestimates, this Appendix extends the project example from Appendix 25 ofthe ADB Guidelines (1997) by incorporating the effects of project financingand profits taxation. Both of these aspects were omitted for simplicity in theoriginal, but this simplification may have caused some misinterpretation ofwhat is required.

Before presenting the revised example it is worth restating the differentfinancial calculations that can be carried out for a project. Three of the mostimportant alternatives are

(i) Return to total capital: This is the NPV/IRR for all project resourcesvalued at financial prices, but before any financing arrangementsare incorporated in the analysis. This is the basic financial returnregardless of how the project is financed and before any profitstax is paid. As a test of viability the FIRR to total capital shouldbe compared with the weighted average cost of capital discountrate.

(ii) Return to total capital net of tax: This is the NPV/IRR for all projectresources valued at financial prices, before any financingarrangements but after profits tax has been deducted. Profits taxwill only be estimated approximately at this stage if interestpayments, which are tax deductible, are not known accurately.As a test of viability the FIRR to total capital net of tax should becompared with the weighted average cost of capital discount ratenet of tax.

(iii) Return to equity: This is the NPV/IRR of the project resource flowgoing to the project owners (i.e., those with an equity stake). Hereloans are benefits to equity holders when they are received andcosts when the interest and principal are repaid. Returns to equitymust also be after deduction of profits tax. As a test of viabilitythe FIRR to equity should be compared with the opportunity costof investors' own funds.

Untitled-1 5/17/02, 10:25 AM83


From the viewpoint of distribution analysis financial calculations shouldbe carried out as returns to equity since this allows a disaggregation of financialreturns into: returns to equity, lenders and government (via profits tax). Thereturns to these groups are defined as follows.

(i) Return to equity is the discounted value of profits (either retainedor distributed) minus the discounted value of equity investment.

(ii) Return to lenders is the discounted value of loan repayments(interest plus principal) minus the discounted value of the loan.

(iii) Return to government is the discounted value of the profits taxpayments.

These adjustments can be illustrated taking the telecommunicationsproject example from Appendix 25 of the Guidelines. The numbers there arechanged in three ways here for purposes of illustration. First, we assume aloan of 200, which is 50 percent of the investment cost of 400, so that equityinvestment is also 200. This loan is to be repaid over 5 years at an 8 percentreal interest rate. Second, to show the role of profits tax in the calculation weassume an effective tax rate of 20 percent (This means that after allowing forvarious deductions and allowances of the Taxable Profits, 20 percent isdeducted as tax). Third, to allow enough Taxable Profit we increase revenueto 850.

Table A5.1 gives the full financial and distribution calculation and TableA5.2 gives the supplementary loan calculation. The return to total capital is50 (Revenue minus Investment and Operating Costs), while the return to equityis 84.68. Investors gain at the expense of lenders since the loan interest rateis below the assumed financial discount rate of 12 percent, which implies aninterest rate subsidy. The government gains in the financing calculations fromthe payment of profits tax. Whether the distinction between investors andlenders is critical for distribution analysis partly depends on whether theybelong to different groups. If both are part of the public sector then theirincomes can be included under the general group of government. Howeverone may be in the public and the other in the private sector or one may benational and the other may be foreign.

Untitled-1 5/17/02, 10:25 AM84

1 Introduction 85APPENDIX 5 Incorporating Project Financing in Distribution and Poverty Impact Analysis 85

��?9%;�� "��!��

��

�� 3��4��

Item

FN

PV

CF

EN

PV

(EN

PV-F

NPV)

Cor

pora

tion

Gov

't.La

bor

Con

sum

ers

Lend

ers

5��

Reve

nue

850

185

00

Loan

200

00

(200

)(2

00)

Con

sum

er su

rplu

s0

250

250

250

Inflo

w1,

050

1,10

0

$��

Equi

pmen

t(4

00)

1.3

(520

)(1

20)

(120

)In

stal

latio

n(1

00)

1(1

00)

0La

bor

(100

)0.

9(9

0)10

10O

ther

s(2

00)

1(2

00)

0In

tere

st(3

3.22

)0

033

.22

33.2

2Pr

inci

pal

(128

.74)

00

128.

7412

8.74

Profit

s ta

x(3

.36)

00

3.36

3.36

Out

flow

(965

.32)

(910

)

��46��

84.6

819

084

.68

(116

.64)

1025

0(3

8.04

)to

gov

ernm

ent

3.36

to le

nder

s(3

8.04

)to

tota

l capita

l50

Effe

ctiv

e ta

x ra

te (

%)

20

Untitled-1 5/17/02, 10:25 AM85


Table A5.2

,��

2�� + � � � � �

Balance outstanding 200.0 160.0 120.0 80.0 40.0 0.0Principal 0.0 40.0 40.0 40.0 40.0 40.0Interest 0.0 16.0 12.8 9.6 6.4 3.2��)�� +�+ �"�+ ��* �7�" �"��

��.PV (interest) 33.22 Loan 200PV (principal) 128.74 Repayment period 5 years)�� "��7" Real interest rate 8%Subsidy 38.04 Discount rate 12%

Other aspects of the illustration remain unchanged. The distribution ofdifferences between economic and financial values creates a similar patternof income change as before. Consumers gain the consumer surplus, laborgains the difference between the financial and the economic wages, and thegovernment loses the premium on foreign exchange from the purchase ofimported equipment. The final set of income changes is summarized below.

Corporation 84.68Government (116.64)Labor 10.00Consumers 250.00Lenders (38.04)ENPV 190.00

The important general point is that under some financing arrangementsone would expect the income flows created by financing of a project to bejust as important, if not more so, than those identified from the existence ofdivergences between economic and financial prices for the inputs and outputsof a project. Omission of those flows as a result of basing the analysis on areturn to total capital (before or after tax) can be misleading.

Untitled-1 5/17/02, 10:25 AM86

1 Introduction 87APPENDIX 5 Incorporating Project Financing in Distribution and Poverty Impact Analysis 87

Extension to Poverty Impact Analysis

A question arises as to how losses (or gains) to Lenders explained abovecan be linked with poverty impact analysis. Several possible scenarios canbe identified:

(i) Where the Lender is a private financial institution:

Here in principle the proportion of lenders who are poor can beidentified and this ratio can be applied to their change in income.However, losses (or gains) to the Lender are unlikely to affect thepoor since income changes for a private financial institution willeither impact on shareholder profits or deposit or borrower interestrates. The poor will not be investors in private financial institutionsand will rarely be affected by commercial bank interest rates.Further, it would be rare for a private financial institution tocontinue for long with a policy of subsidized credit, which is theimplication of an income loss to the Lender. Hence this scenariois unlikely to be of relevance in poverty impact analysis.

(ii) Where the Lender is a public financial institution:

This scenario is more relevant but can be dealt with relativelyeasily, for example, by incorporating the Lender as part of theGovernment. In the example in Table A5.1, where the loan comesfrom the government the Lenders column can be merged withthe Government column so that the value of the Loan of 200 isentered with a negative sign and the value of Principal and Interestpayments of 33.22 and 128.74, respectively, are entered withpositive signs in the Government column. The net total for theGovernment column is thus -154.68, which is the sum of the originallosses to Government (-116.64) and Lenders (-38.04). The relevantpoverty ratio for government income can be applied to this figure.In this example if the government had lent to the project at acommercial interest rate, broadly speaking, the loss to the Lenderwould have been zero, but the return to the corporation wouldbe correspondingly lower at 46.64 (84.68-38.04). In the case oflending by aid agencies such as ADB, for practical purposes, wecan assume that the external funds are channeled from theborrower government to a local public financial institution (or a

Untitled-1 5/17/02, 10:25 AM87


private financial institution acting as a nonprofit agent for theborrower government). But this refinement does not affect thebasic structure of the analysis (see Case 5 in Appendix 6 for anillustration).

Untitled-1 5/17/02, 10:25 AM88

1 Introduction 89

APPENDIX�6Case Illustrations of Poverty Impact Analysis

The following cases extend distribution analysis to incorporate a povertyfocus. All are based on actual reports and recommendations to the President(RRPs) to which additional work has been added. Case 1 illustrates the pointthat economic calculations conducted at world price numeraire can be usedas the basis for distribution analysis, although there is an added complicationnot present when the domestic price numeraire is used for economic values.The poverty impact results are speculative in the absence of knowledge ofhow much of the consumer benefits from the project will accrue to poorconsumers. A detailed regional survey would be required to address thispoint accurately and the 25 percent figure used is only an approximation.Case 2 is of an irrigation project from Viet Nam. The original work on this RRPwas supplemented by further EDRC analysis in the field, which allowed arelatively detailed assessment of the project's poverty impact. The case alsoillustrates an important limitation of the poverty impact ratio, which is not anadequate means of selecting alternatives from a set of projects. In this casethe project with the largest net gain to the poor is the one with the lowestPIR. Case 3 on a coastal resource management project from Sri Lanka doesnot contain a distributional and poverty impact analysis, but some indicativecomments on how one might be carried out. It is included here to drawattention to a particularly detailed social impact assessment, which unusuallyprovides detailed data on the incomes of the stakeholders involved. Case 4on a health sector project illustrates a case in which project benefits are notmonetized and cost-effectiveness analysis needs to be extended to povertyimpact analysis using a second-best headcount approach. This is an area inwhich an increasing research effort can enhance the understanding of themagnitude of the differential impact of health intervention on the poor. Case5 on a road rehabilitation project is the most recent example in this Handbook.The analysis was prepared under a separately commissioned study withspecific terms of reference to carry out distribution and poverty impact analysis.A good aspect of this analysis was the importance attached to thecompetitiveness of road transport services as a determinant of the project'spoverty impact.

Untitled-1 5/17/02, 10:25 AM89

It should be noted that the purpose of the illustrations here is todemonstrate that there have been actual cases of application to ADB projectsand they do not necessarily represent best practices. Project analysts areencouraged to explore improvement on the practice, for example, byconsidering the data collection and analysis suggested in Appendix 3 andincorporating the project-financing aspect discussed in Appendix 5.

Case 1– Philippines: Transmission Interconnectionand Reinforcement Project

The following is taken from the RRP.1 The project design is complexwith four components but the procedure and assumptions of economicanalysis are presented well. A reasonable attempt has been made to extendit to distribution and poverty impact analysis. The assumptions and numbersare slightly modified from the original. For the illustrative purpose of thisappendix, sensitivity and risk analyses are omitted.

Project Description

Mindanao is one of the poorer regions in the Philippines. It contains23.2 percent of the country's population. The poverty incidence of the regionis 50 percent as compared with a national average of 36.7 percent. As manyas 62.3 percent of the poor in Mindanao have no access to electricity. TheProject provides high-priority transmission facilities that are required toevacuate power from Luzon to Mindanao which is otherwise expected toface power shortages, and that are essential for the operation of the competitiveelectricity market. The Project consists of about 416 km of high voltage directcurrent transmission lines. Component A (Leyte-Mindanao Interconnection)provides 23 km of 350 kV HVDC submarine cable, cable terminal stations,converter stations, 25 km of electrode lines, 28 km of 138 kV alternating current(AC) transmission lines, and associated substation expansions linking the

1 Draft RRP-PHI: Transmission Interconnection and Reinforcement Project, for the proposed amount

of $100 million, as of May 2000.


Untitled-1 5/17/02, 10:26 AM90

geothermal fields in Leyte to Mindanao. Component B (Luzon TransmissionUpgrading) provides about 117 km of 230 kV AC double circuit transmissionlines between San Manuel, Pangasinan, and Mexico, Pampanga, and 15 kmof 230 kV AC double circuit transmission lines from Biñan, Laguna toDasmariñas, Cavite. Component C (Mindanao Substation Expansion) installsadditional power transformers and circuit breakers at the existing Bislig,Butuan, Davao, Kibawe, Klinan, Sta. Clara, and Tindalo substations inMindanao. Component D (Power Trading and Pooling System) develops thepower trading and pooling system software and provides required computerhardware to control and process transactions and functions associated withthe competitive electricity market. Of the total project cost of $600 million,85 percent is allocated to Component A, 9 percent to Component B, 3 percentto Component C, and 3 percent to Component D. It is to be cofinanced byADB (16.7 percent), Japan Bank for International Cooperation (16.7 percent),commercial banks (55 percent), and the Government (11.7 percent).

Least-Cost Analysis

Using the EGEAS program (developed by the Electric Power ResearchInstitute of USA), the National Power Corporation (NPC) performed a least-cost analysis for the Mindanao grid as part of the preparatory work for the1999 Power Development Program. The first alternative (the Project) allowsfor transfer of power generated at independent power producer (IPP) powerplants in Leyte and Luzon to the growing load centers in Mindanao. Surpluspower in Luzon is planned to be available for export to Leyte from 2004 to2013, peaking at 2,938 GWh in 2009. The weighted average cost of powerfrom the Leyte and Luzon IPP power plants would range between P=2.75/kWh and P=3.26/kWh. The second alternative would be to accelerate theexpansion of the generating capacity in Mindanao using a mixture of generatingplants such as coal, diesel, and hydropower. The equalizing discount rate(EDR) analysis supports the Project over the second alternative. Using constant1999 prices and an SCF of 0.83 (this number is outdated and needs to beupdated for future ADB projects in the Philippines), it was shown that bothinvestment costs and system operation and maintenance (O&M) costs arelower for the Project than the alternative for discount rates up to 18.5 percent.The production costs of the Project are lower because the interconnectionallows more efficient use of resources by utilizing power from two IPPs

APPENDIX 6 Case Illustrations of Poverty Impact Analysis 91

Untitled-1 5/17/02, 10:26 AM91

operating the Leyte Geothermal Plants and the planned Iligan powergenerating plant in Luzon. NPC's contract with these IPPs is on a take-or-pay basis. The interconnection also allows saving in fuel at thermal powerplants in Mindanao.


Approach — Financial and economic analyses are first carried outseparately for Components A, B, and C, respectively, as there is no revenueand benefits quantifiable separately for Component D. Then the analysis forthe overall Project was done by adding the cost of Component D.

The financial analysis for Component A was based on (i) a project lifeof 25 years (after project completion); (ii) an estimated total financial cost ofP=16,259 million covering two converter stations, HVDC overhead transmissionlines, submarine cable and electrode stations, and strengthening of theassociated AC components; (iii) annual O&M cost of 1 percent of capital costs;(iv) annual sales ranging between 1,837 GWh and 3,284 GWh; (v) transmissionlosses of 1 percent in Leyte, 1 percent over the interconnection and 4 percentin Mindanao; (vi) cost of purchasing power in Leyte; and (vii) an averagetariff of P=2.36/kWh. It is assumed that when the competitive markets areintroduced, the tariff throughout the NPC grid will tend to even out. The tariffin Luzon would be reduced over time, while the tariff in Mindanao wouldincrease. To allow for additional benefits from competitive markets throughefficiency gains and improved operations in future, the adopted tariff was setat 90 percent of the weighted average tariff proposed in March 1999 (P=2.96/kWh in Luzon, P=2.65/kWh in Visayas, and P=1.86/kWh in Mindanao, with aweighted average of P=2.62/kWh). Incremental revenue starts to flow in onlyfrom 2008 when the interconnection would meet additional demand in theMindanao system. Before that, the interconnection would displace revenuesfrom the existing power plants. The interconnection will lead to resourcesavings in Leyte, Luzon, and Mindanao. Resource savings in Leyte representthose costs that would otherwise have to be paid by NPC to IPPs in Leyte andLuzon for not meeting their take-or-pay contractual obligations. Resourcesavings in Mindanao result from fuel savings due to deferred construction ofnew plants—backing down diesel-fuelled IPP generating plants totaling 258MW. The FIRR of this component is estimated at 9.7 percent, which is higherthan the estimated weighted average cost of capital (WACC) of 5.2 percentand is satisfactory.


Untitled-1 5/17/02, 10:26 AM92

In the economic analysis for Component A, nonincremental power wasvalued on average at P=4.43/kWh net of distribution losses and costs. The usewillingness-to-pay (WTP) was based on the present-day cost of operatingprivate generator sets, which were purchased by affluent households,commercial establishments and industries to provide power during the 1992/93 power crisis. Given the higher cost of self-generation, it was assumed thatresidential, commercial, and industrial users would only self-generate 27/percent, 35 percent, and 22 percent, respectively, of their normal powerconsumption. The WTP is also based on the present-day cost of using kerosenelamps for lighting among the poorer consumers in rural areas. Incrementalpower demand will mainly come from the connection of rural consumers.The benefits are credited to the Project from 2008 onwards when there is nomore surplus capacity in the system from the generating plants operationalin 2003. The EIRR for this component is estimated at 17 percent and issatisfactory as it exceeds the benchmark 12 percent for economic viability.

The financial analysis for Component B was based on (i) a project lifeof 25 years, (ii) average cost of power in the northern Luzon Grid, (iii) fuelcosts, (iv) annual O&M cost of 2.5 percent of the capital costs of the transmissionfacilities, (v) transmission losses of 3 percent, and (vii) an average tariff ofP=2.36/kWh. The alignment of the transmission line follows an existing right-of-way and thus significant cost savings ensue. The FIRR is estimated at 11.4percent. The economic analysis was based on the average benefit of P=4.42/kWh, net of distribution losses and costs. All economic benefits are considerednonincremental. The EIRR is estimated at 15.4 percent. Component B isconsidered satisfactory in terms of both financial and economic viability.

The financial analysis for Component C was based on (i) a project lifeof 25 years, (ii) annual O&M cost of 2.5 percent of the capital costs of thetransmission facilities, (iii) system losses of 4.5 percent, and (iv) an averagetransmission tariff of P=0.27/kWh. The FIRR is estimated at 21.9 percent. Theeconomic analysis was based on an average benefit of P=4.43/kWh net ofdistribution losses and costs. The EIRR is estimated at 27.2 percent, indicatingboth financial and economic viability of Component C.

The FIRR for the overall Project was calculated by putting together theabove three components and adding the cost of Component D. The FIRR andEIRR are estimated at 10.2 percent and 16.9 percent, respectively, indicatingboth financial and economic viability of the overall Project.


Untitled-1 5/17/02, 10:26 AM93


For the purpose of distribution analysis, the discount rate of 12 percentis used in both flows. The original project economic analysis is at world prices,so that tradables are valued at border prices at the prevailing exchange rate.Nontradables are valued at economic prices, using a mix of SCF and specificconversion factors. The SWRF for unskilled labor is less than one, which ismultiplied with SCF to obtain a conversion factor for unskilled labor at theworld price level. As discussed in Fujimura and Weiss (2000), the use of theworld price numeraire (but in domestic currency units) as in this case canbe applied in distribution analysis, provided both financial flows and economicflows are expressed in the same numeraire. If we use economic prices in theworld price numeraire, financial flows must be adjusted to the same numeraireby multiplying by the SCF. If we use the domestic price numeraire, economicflows must be adjusted by dividing by the SCF. Table A6.1 summarizes thecalculations in both numeraires. Although we illustrate the use of the worldprice numeraire to make the point of its equal applicability, it would havebeen simpler to use the domestic price numeraire for both financial andeconomic flows. See also Curry and Weiss (2000, pp.285–289) for more subtlediscussion on the numeraire issue in distribution analysis.

Table A6.1 shows the distribution of the project effects (economic minusfinancial present values) among the stakeholders: the Government/the restof the economy, consumers of the Project electricity, and labor involved in theProject. As a result of the Project, consumers and labor gain while NPC andthe Government lose (the original analysis mistakenly indicated thatGovernment gains). Since NPC is a state-owned enterprise, the loss to theNPC can be considered the Government’s financial losses from investing inthe Project. The Government’s economic loss is due to inefficient resourceuse associated with domestic price distortion relative to world prices (SERFeffects) and constitutes effective net subsidy to the Project. Consumers gaina consumer surplus, mainly in terms of the difference between the without-Project cost of electricity and the with-Project expenditure on electricityand the value of electricity consumed but not paid for. It is assumed that10 percent of the capital and O&M costs over the project life would comefrom the employment of unskilled labor. This labor gains a net economicbenefit because the Project pays wages in excess of the economicopportunity cost of unskilled labor.


Untitled-1 5/17/02, 10:26 AM94

Poverty Impact Analysis

For the purpose of poverty impact analysis, net economic benefitsaccruing to each stakeholder must be allocated between the poor and nonpoor.

(i) Along the line explained in the main text, the marginal governmentexpenditure required for project cost finance that accrues to thepoor is approximated by the income share of the poor in GDP,which is 14 percent. This ratio is applied to both NPC and theGovernment.

(ii) The existing access to electricity is rather biased in favor of thenonpoor. In Mindanao, 38 percent of the poor has access toelectricity while 82 percent of the nonpoor does. Therefore, theprincipal channel of distribution of consumer surplus attributableto the Project reaching the poor will be the incremental supplyof electricity, particularly due to rural electrification. The poorwill benefit from lower electricity costs from the transmissioninterconnection and reinforcement under the Project, comparedwith the higher cost of thermal energy. Based on the projectedshare of poor consumers, their electricity consumption patternand the number of the poor living below the poverty line, it isassumed that 25 percent of the consumer surplus goes to the poor.

(iii) Based on the projected rural labor employed by the Project, it isassumed that 20 percent of the economic benefits accruing tothe unskilled labor goes to the poor living below the official povertyline.

(iv) Based on the above assumptions, the Project benefits going tothe poor below the poverty line are estimated at P=2,846 million(in domestic price numeraire), or the poverty impact ratio is 34percent, as shown in Table A6.1. In addition to the quantifiedbenefits, there are nonquantifiable economic and social benefitsto the poor in the form of enhanced opportunity of human capitaldevelopment induced by the efficient power supply. The analysishere indicates that a significant portion of the Project benefitsgoes to the poor, perhaps significantly exceeding the current poor'sshare in national income.


Untitled-1 5/17/02, 10:26 AM95

Table A6.1

��

��; � ��

8�� 9�� ,��

&PPPPP=====� ��( :�� :�� 4�� 4��

%�� ' %�� %�� ' :��9��

Benefits 57,067 47,366 59,597 12,231Capital Costs (13,493) (11,199) (12,230) (1,031)Operating Costs (45,095) (37,429) (40,363) (2,934),�� %�� (1,521) (1,262) 7,004 8,266

Gains and LossesSCF 0 .83 Proportion of the Poor (%)SERF 1.204819 Benefits for the Poor

Poverty Impact Ratio (%)

8�� ,��

( million) Financial Financial Economic EconomicPresent Present Present Value minus

Value at 12% Value at 12% at Domestic prices Financial

Benefits 57,067 59,597 71,804 14,737Capital Costs (13,493) (12,230) (14,735) (1,242)Operating Costs (45,095) (40,363) (48,630) (3,535),�� %�� (1,521) 7 ,004 8,439 9,960

Gains and LossesProportion of the Poor (%)Benefits for the Poor

Poverty Impact Ratio (%)

P=


Table A6.1

��

��; � ��

Untitled-1 5/17/02, 10:26 AM96


Table A6.1

��

� ��

�� 3�� 4��, $ ; ��/ $�� <��

4��

12,231(1,157) 126(3,407) 473

Total

(1,262) (4,564) 12,231 600 7,0041 4 1 4 2 5 2 0

(177) (639) 3,058 120 2,362

3 4

Distribution of Project EffectsNPC Government/ Consumers Labor

Economy

14,737(1,394) 152(4,105) 570

Total

(1,521) (5,499) 14,737 722 8,4391 4 1 4 2 5 2 0

(213) (770) 3,684 144 2,846

3 4

Untitled-1 5/17/02, 10:26 AM97

Case 2 – Viet Nam: Second Red River Basin WaterResource Sector Project

The following is based on the RRP2 and the associated EDRC staffwork that contributed to the related supplementary appendix in the RRP.The original economic analysis was limited to the estimation of EIRRs forsubprojects, but in view of the importance of distributional and povertyimpacts associated with the Project, EDRC staff provided additional modestresources to augment the analysis.

Project Description

The Red River basin is home to about one third of Viet Nam's population.Poverty incidence in the upland provinces of the basin is 59 percent,considerably higher than the national average of 37 per cent (using WorldBank poverty line). To tackle the poverty situation in the basin area, the Projectadopts a multidimensional approach through better management of the mostimportant resources—water, and investment in physical infrastructures suchas dams, canals, and extension services. The Project has two components.Component A addresses integrated water resources management andassociated institutional building. These will include capacity building, publicawareness and education programs for water resources management, pilottesting in water licensing and wastewater discharge permit system, waterquality monitoring network, and flood forecasting and warning system.Component B provides infrastructure improvements and the necessaryagricultural support services identified by communities. It will comprise anumber of subcomponents, the most important ones are (i) subprojects toimprove irrigation systems and watershed protection in the uplands; and(ii) subprojects to improve delta irrigation and drainage systems. Of thetotal project cost of $137 million, $22 million is allocated to Component Aand $115 million to Component B. Under the investment Component B theProject Implementation Unit (PIU), in collaboration with local authoritiesand organizations, will identify subprojects for possible rehabilitation andupgrading.

2 Draft RRP-VIE: Proposed Second Red River Basin Water Resource Sector Project, May 2000.


Untitled-1 5/17/02, 10:26 AM98

Economic Analysis

The economic analysis was conducted for three representativesubprojects, two in upland areas (Yen Binh and Ngia Lo districts in Yen BaiProvince, 200 km northwest of Hanoi city) and one in the delta (Gia Thuandistrict in Bac Ninh Province, just outside Hanoi city) to examine the economicviability of indicative subprojects under the Project, which is expected tosupport more than 30 subprojects for rehabilitation and upgrading.

In Yen Bai province, several irrigation schemes were found to beexperiencing rapid sedimentation and will require repair and upgrading ofirrigation infrastructure such as the headworks and related delivery facilities.The Project investment will comprise rehabilitation of weirs, lining of maincanals, and construction of secondary canals of selected irrigation schemes.In Bac Ninh Province, the irrigation schemes are suffering from aging (30years old) and are in need of repair, and experience water shortages in winterand spring and flooding in late summer. Rehabilitation and improvement ofthe irrigation and drainage infrastructure in the Gia Thuan subproject willmainly involve installation of new pump facilities and upgrading of irrigationcanal-related structures.

Increased production in all three irrigation subprojects is projected tobe derived from increased crop yields, higher cropping intensity as well asadditional agricultural land, which were mainly rainfed areas prior to Projectimplementation, converted to irrigated agriculture. It is expected that at fulldevelopment (in Year 6 of Project implementation) all rainfed areas in thethree subproject areas will be fully irrigated. Under Yen Binh and Ngia Losubprojects, 333 ha and 206 ha, respectively, of irrigated land will be coveredfor irrigation upgrading. Under the Gia Thuan subproject, a total of about17,500 ha of irrigated areas will be covered. As a result of converting presentlyrainfed rice and nonrice areas into irrigated diversified agricultural lands,the cropping intensities in the three subproject areas are expected to increasefrom 159, 151, and 202 percent under without-Project situation, to 242, 232and 232 percent under with-Project situation, respectively. Paddy rice yieldsin the subproject areas are expected to increase from current levels of 3.64mt/ha, 3.48 mt/ha, and 5.0 mt/ha, respectively, for spring irrigated rice to 5.2mt/ha. Yields of maize, sweet potato, and vegetable crops likewise are expected


Untitled-1 5/17/02, 10:26 AM99

to increase as a result of improved water availability and delivery. At fulldevelopment in 2006, an aggregate net incremental annual production of2,200 mt, 1,300 mt, and 51,200 mt, respectively, is envisaged to be realizedin the three subproject areas.

Economic pricing was done under the domestic price numeraire. Allsubprojects are assumed to have a 25-year life. Subproject investment andrecurrent costs were first segregated into their foreign and local costcomponents. The foreign exchange component, together with about 20 percentof local currency costs which were assumed to consist of tradable items, wereshadow priced using a shadow exchange rate factor (SERF) of 1.043 (shadowexchange rate of dong(D) 14,599/US$ as opposed to official exchange rateof D14,000/US$). All other local currency costs were expressed in economicvalues following their financial values. Shadow wage rate factor (SWRF) of0.85 was applied to both on-farm labor and the labor components of theinvestment costs. Economic prices of rice and maize as well as agriculturalinputs such as urea, triple superphosphate (TSP) and potassium chloride (KCI)were adapted from actual average international prices of traded commoditiespresented in the World Bank Global Commodity Markets (April 1999). In arrivingat farmgate prices, border prices for international freight and handling chargeswere added and local costs for transportation, bagging, handling, andprocessing were added. Economic prices of sweet potato and vegetablesassumed to be grown in the subproject area were mainly based on theirfarmgate financial prices, adjusted for transportation costs. Based on theestimated incremental economic costs and benefits, the net economicbenefits attributed to agricultural crop production were calculated which,in turn, served as the basis for estimating the EIRRs of the three subprojects.The base case EIRR value was calculated to be 20.6 percent for the Yen Binhsubproject, 12.4 percent for the Ngia subproject, and 25.1 percent for theGia Thuan subproject. This implies that lowland irrigation has a relativelybetter economic return than upland irrigation—a sign of locational trade-offbetween efficiency result of the subproject and targeting of high povertyincidence areas. Sensitivity analysis is omitted here.


Untitled-1 5/17/02, 10:26 AM100


As the three subprojects do not envisage revenue-generating projectentities, the original analysis did not carry out financial analysis for thesubprojects as a whole, which would have gone further than farm budgetanalysis. Here financial prices and financial NPVs corresponding to economicones have been matched one to one in order to examine the distributionalaspects of the subprojects’ impact. Also an extension to poverty impact analysishas been attempted. To carry out these analyses, it was necessary to collectboth primary and secondary data on transfer payments related to projecttransactions and poverty incidence in the subproject areas. The followingmajor groups of data were collected:

(i) project-specific economic and financial data: those on prices,district and provincial market, marketing margins of inputsuppliers and processors, subsidy, taxes, local wage andunemployment rates for each of the three subproject sites; and

(ii) data on commune-level poverty incidence in each of the threesubproject sites as well as national-level poverty incidence.

For distribution analysis, three main stakeholders were identified: (i)farmers, (ii) labor, and (iii) the Government (or the rest of the economy), amongwhich the projected net benefits were distributed. Benefits accruing to on-farm labor have been combined into farmers’ benefits. The domestic pricenumeraire was used in both financial and economic flow calculations and adiscount rate of 12 percent was used to calculate net present values (NPVs)of both flows. A most outstanding feature of the projects such as these, wherethere are no clear revenue-generating (or cost recovery) agents involved, isthat all financial benefits tend to accrue to project beneficiaries (farmers) whileall financial costs accrue to government. One qualification, however, is thatthe original PPTA for this Project did not directly identify the farmgate pricesfacing farmers but instead derived them backward from border prices, eitherdue to unavailability of the direct data in the field or time constraint. If, forexample, the farmgate-to-border middlemen costs have been underestimated,net financial benefits accruing to farmers might be overestimated in thesummary figures presented here. A more accurate assessment and verification


Untitled-1 5/17/02, 10:26 AM101

on the farmers' financial impact would require information on farm operations.Nonetheless, the summary presentation here remains useful for illustratingthe nature of distributional analysis for this category of projects.

Yen Binh Subproject

Table A6.2 suggests that farmers are a significant net gainer while theGovernment (rest of the economy) is a significant net loser from this subproject.Hired labor is a net, though not significant, gainer. The last group's gain isderived from the divergence between financial on-farm wage actually receivedand their economic opportunity cost of labor. Farmers' net gains are derivedfrom the subproject's net financial gains (net of very modest irrigation fees)as well as from the divergence between prevailing financial on-farm wageand opportunity costs of labor. (It is worth noting that while farmers ownlabor, it does not enter as financial costs in farm budget analysis—it doesenter as financial cost either fully or partially in the financial analysis as usedto match economic analysis.) A large part of the Government net loss isattributable to the expenditure to support investment and recurrent costs forthe irrigation upgrading. Once this portion is netted out, the remaining gainsand losses to the Government consist of a number of further transactions.The Government provides a sizeable amount of subsidies to farmers in theregion through controlled farm inputs prices (seed, fertilizer, pesticides) andsubsidized O&M costs of the irrigation system. However, these losses to theGovernment are partially offset by additional tax revenue on purchased materialsfor irrigation investment as well as by the positive effect of increased productionof tradable outputs (SER effect). This disaggregated analysis on theGovernment gains/losses highlights some important issues. First, one needsto consider the extent of fiscal impact of supporting the irrigation investment,and the fiscal sustainability of the recurrent costs of the irrigation system.This aspect tends to be neglected in the analysis where only economicanalysis is done to justify irrigation projects. Ideally, financial sustainabilityanalysis should be included in the economic analysis. Although individualirrigation projects may be marginal to the total public expenditure in theagriculture sector, accumulation of similar projects could have a significantbearing on the fiscal affordability. Second, the analysis here clearlydemonstrates a significant distortion as farmers will be heavily subsidizedin their production activities. One would need to interpret and justify this


Untitled-1 5/17/02, 10:26 AM102

treatment of farmers in relation to arguments such as the peculiar risks involvedin farming such as volatile crop yields and prices and national food securitypolicy, as well as in relation to poverty reduction objectives. During the fieldvisits, discussions with government officials and the local authorities suggestedthat the current policy of subsidizing farm inputs are likely to remain in theforeseeable future.

Ngia Lo Subproject

The gains/losses profile between the three stakeholders and the causesof economic benefits and losses are explained in a similar way. The onlydifference from the above subproject is that the Government subsidy to farmersis in the form of controlled inputs prices only but does not include irrigationrecurrent costs. In the upland provinces, only a fraction of the irrigationschemes are maintained by the provincial agricultural authority and the restare managed by farmers themselves or by farmers' associations. While thedam and main irrigation canals in Yen Binh district are maintained (andsubsidized) through the Department of Agriculture and Rural Development,those in Ngia Lo district are self-run by farmers. Compared with the abovesubproject, the size of farmers’ gain relative to the Government loss is not asgreat (roughly 1 to 1; this ratio for the above subproject is roughly 3 to 2). Thiscomparison is consistent with the above EIRRs for the two subprojects.

Ghia Thuan Subproject

Again, the basic distribution profile is the same as the above twosubprojects. However, the relative magnitudes of farmers gain and Governmentloss is significantly different. The farmers' financial gain is significantly largerthan Government financial loss, making the farmers' economic gains almostfive times as large as the Government economic loss. Also the difference inthe government transfer profile is that the subsidy to the farmers is in theform of O&M costs of the Irrigation and Drainage Management Company, aswell as the subsidies provided through cheap electricity on irrigation pumps.However, there is no subsidy in terms of controlled input prices. In the deltaarea, farmers rely on pump irrigation that is heavily dependent on subsidy.


Untitled-1 5/17/02, 10:26 AM103

Table A6.2

�� @�� ; ��

:��=&��( :�� 4�� 4�� :��

�� ;��/5��/$�� %�� %�� :�� <��

��24,�5!,>��85 �?@4$)

!��5�� - Project net output 13,628 14,543 915!��$�� - Investment (9,278) (8,405) 873

- O&M (312) (550) (238)

,�� %�� 4,038 5,588 1,550 13,628

;��<�� A"+7Proportion of the Poor (%) 66Net Benefits for the Poor 9,642 ��!�� Present Value of Project Economic Cost,��5��-�� 3��$��

5��,;!��<?��85 �?@4$)!��5�� - Project net output 8,714 9,243 529!��$�� - Investment (8,651) (8,549) 102

- O&M (365) (321) 44

,�� %�� (302) 373 675 8,714

;��<�� 7A��Proportion of the Poor(%) 39Net Benefits for the Poor 3,609 ��!�� "�7"Present Value of Project Economic Cost,��5��-�� 3��$��

$��;>!��)>8�,��85 �?@4$)

!��5�� - Project net output 132,431 161,656 29,225!��$�� - Investment (23,881) (23,628) 253

- O&M (13,558) (18,408) (4,850)

,�� %�� 94,992 119,620 24,628 132,431 2

;��<�� A*��Proportion of the Poor (%) 25.5Benefits for the Poor 39,227 ��!�� +��+Present Value of Project Economic Cost 42,036,��5��-�� 3��$��


Gains and Losses 14,609Proportion of the Poor (%) 66Net Benefits for the Poor 9,642Poverty Impact RatioPresent Value of Project Economic CostNet Benefits for the Poor per Project Cost

Gains and Losses 9,254Proportion of the Poor (%) 39Net Benefits for the Poor 3,609Poverty Impact RatioPresent Value of Project Economic CostNet Benefits for the Poor per Project Cost

Gains and Losses 153,833Proportion of the Poor (%) 25.5Net Benefits for the Poor 39,227Poverty Impact RatioPresent Value of Project Economic CostNet Benefits for the Poor per Project Cost

Table A6.2

�� @�� ; ��

!��)��

Untitled-1 5/17/02, 10:26 AM104

Table A6.2

� ��; �� !��)��

�� 3��4��= >�� ; �/4�� 9� �9� �� )�1�� 4� :��/ 4�� 4�� 4�� ;��/<��

981 (1,073) 1,007121 805 (53)37 (275)

8 981 158 (1,348) 805 954 (9,590) Total

��* &7A�#7( 5,58866 12

104 (1,101) 8,645

8,954+�7# (This is different from benefit-cost ratio)

540 -562 550123 32 -5344

4 540 167 -562 32 497 (9,016) Total

�"# &7A+�7( 37239 1265 (1,086) 2,588

8,871+��7 (This is different from benefit-cost ratio)

21402 7,823121 808 -676

1,627 (6,477)

1 21,402 1,748 (6,477) 808 7,147 (37,439) Total

�A#�* &��A7"�( 119,62025.5 12446 (4,315) 35,358

+�*� (This is different from benefit-cost ratio)


158 (9,179) 5,58866 12

104 (1.101) 8,6451.55

8,9540.97 (This is different from benefit-cost ratio)

167 (9,049) 37239 1265 (1,086) 2,588

6.968,8710.29 (This is different from benefit-cost ratio)

1,748 (35,961) 119,62025.5 12446 (4,315) 35,358

0.3042,036

0.84 (This is different from benefit-cost ratio)

Untitled-1 5/17/02, 10:26 AM105


The data analysis started off by compiling project-specific commune-level data on poverty incidence among farmers. Such detailed data were onlyavailable through the district and province offices of the Ministry of Labor,Invalids and Social Affairs (MOLISA). The three subproject areas comprise of4, 4, and 45 communes, respectively, and "hungry and poor households" arerecorded for each commune by MOLISA. These data are aggregated for eachsubproject level (Table A6.3). However, these MOLISA data use a very stringentcriterion equivalent to food poverty as they are designed to target governmentsocial programs. It results in a low national poverty incidence of 17 percent.When the more general poverty line based on food and nonfood basket isused, the national poverty incidence is 37.4 percent (World Bank providedthe technical inputs and documented them in the Government-Donor-NGOjoint report on Attacking Poverty: Vietnam Development Report, 2000). Thelatter data are considered more appropriate for the analysis here but they arebased on sample surveys and do not provide detailed project-specific povertyincidence data. Therefore, as a compromise, the MOLISA data on povertyincidence were converted to the numbers consistent with the general povertyline (labeled WB in Tables A6.3 and A6.4) using a conversion factor based onregional level poverty ratio between the two sets of data. As all threesubprojects belong to the Northern upland region, the conversion factor forthat region, i.e. 2.63, was used to scale up the project-specific poverty dataobtained from MOLISA offices.

Table A6.3

�� +� �� $'(

Subproject MOLISA Conversion Factor WB

Yen Binh 24 .9 2 .63 65.5Ngia Lo 14 .9 2 .63 39.2Ghia Thuan 9.7 2 .63 25.5

,�� # �#��


Untitled-1 5/17/02, 10:26 AM106

Table A6.4

�� #�� 6��)! ��-7,��

Percentage of Poor and Hungry Householdsin Total Population Ratio

Region WB (%)a MOLISA (%)b (conversion factor)

Northern Upland 5 9 22.4 2 .63Red River Delta 2 9 8.4 3 .45North Central 4 8 24 .6 1 .95Central Coast 3 5 17 .8 1 .97Central Highlands 5 2 25.6 2 .03South East 8 4.8 1 .67Mekong Delta 3 7 15.4 2 .40

,�� # ��+ ��#

a The data in this column were obtained from Government-Donor-NGO Joint Report "Attacking Poverty", 1999.b MOLISA data was obtained from national and provincial offices and the HQ of MOLISA.

The poverty incidence data derived using a conversion factor as aboveare used as the proportion of the gains/losses of farmers and labor going tothe poor. The poverty incidence among farmers and off-farm laborers are takento be the same as they are not significantly different in the project area. It isplausible that hired laborers will be mobilized from near the project area. Forthe parameter of incidence of marginal government expenditure/incomebetween the poor and nonpoor, 12 percent derived in Appendix 7 can beapplied.

The results of poverty impact analysis are shown in the lower half ofTable A6.2. A comparison of the three subprojects brings out interestingimplications for the Poverty Reduction Strategy in relation to the use of thepoverty impact ratio (PIR) as a guiding indicator. First, the PIR for the YenBinh subproject is greater than unity (1.14) and that for the Ngia Lo subprojectis significantly higher than unity (6.96). This extreme result is partly attributableto the outstanding feature of the subproject (farmers being a significant gainerand Government being a significant loser) combined with the differentialproportions of the poor applied to these two stakeholders’ gain/loss (66 percentfor farmers and 12 percent for the Government). Also the relative magnitudeof farmers' gain to the Government loss contributed to the extreme result.The interpretation here is that the two subprojects have an "ultra-pro-poor"nature and their benefits accruing to the poor are more than significantly


Untitled-1 5/17/02, 10:26 AM107

greater than their income share in GDP (12 percent), easily qualifying as povertyinterventions. In contrast, the Ghia Thuan subproject has a low PIR (0.30)mainly due to the smaller proportion of the poor among target lowland farmerscompared to the first two subprojects. Nonetheless, this subproject can beconsidered significantly benefiting the poor disproportionately relative to theirincome share in GDP (12 percent).

Second, even when all three subprojects satisfy the minimum criterionof economic viability (EIRR >12 percent) and therefore justified on efficiencyground, the interpretation as to how to rank the three subprojects on povertyimpact aspect is not straightforward. Despite the extremely high PIR for theNgia Lo subproject, its economic NPV going to the poor is only a fraction ofthose for the other two subprojects. This illustrates starkly the inadequacy ofonly looking at the PIR as a guide for poverty reduction impact. In absoluteterms, the Ghia Thuan subproject, despite its modest PIR, has the highesteconomic NPV going to the poor. Therefore, in the context of the sector projectin which the total project fund will be used to finance a series of subprojects,those similar to the Ghia Thuan subproject can be selected without reservation.There is a parallel here with the IRR and NPV comparison, where it is wellknown that in simple choices the NPV criteria should be preferred. However,where there is a fixed budget for irrigation investments, one could rankalternatives on the basis of the ratio of economic NPV going to the poor dividedby the present value of subproject economic costs. This ratio compares thethree subprojects' poverty impact by the same yardstick (but it should not beconfused with a conventional project benefit-cost ratio). The purpose is tocompare the subprojects' efficiency in terms of poverty impact. As shownin the last row of the tables, net benefit for the poor per subproject cost ishighest for the Yen Binh subproject, followed by the Ghia Thuan subproject.The low number for this ratio for the Ngia Lo subproject as well as its smallesttotal NPV to the poor indicates that this subproject is the most inferior amongthe three in terms of both economic efficiency and poverty reduction impact.The Yen Binh subproject is both satisfactory in its pro-poor nature andsuperior to the other two subprojects in terms of efficiency of poverty impact.While subprojects with a benefit distribution profile like the Ghia Thuansubproject are initially a preferred choice, those with a benefit distributionprofile similar to the Yen Binh subproject may become the choice at themargin when the fund gets exhausted.


Untitled-1 5/17/02, 10:26 AM108

Conclusion

Apart from the above comparative interpretation of poverty impactsassociated with the three subprojects, some general implication can be drawnfrom the distribution and poverty impact analysis here. In all three subprojectsfarmers are by far the largest beneficiaries of the Project. Hired laborers willalso benefit in all three subprojects. The Government is a significant loserfrom all subprojects. Especially farming in the upland and mountainous regionsof the country will be effectively heavily subsidized both through the fiscalsupport to the upgrading and maintenance of the irrigation systems andthrough various indirect subsidies to the farming activity. This makes thefinancial sustainability of upland projects to some extent dependent on thefiscal affordability and continuation of the current Government's subsidypolicies. Even though there is an apparent political commitment on the partof the Government to continue supporting poor farmers, there is a need tomake a realistic assessment on its ability to finance such arrangements in themedium to long term.

Resource implication: As this case study was carried out by EDRC staffas a pilot exercise after the PPTA had been done without distribution analysis,the extension of economic analysis to distribution and poverty impactanalysis involved some technical difficulties. Nonetheless, considering themodest resources spent and that the EDRC staff did it hands-on for thefirst time, it is believed the exercise was worth the cost.

Case 3 – Sri Lanka: Coastal Resource Management Project

The following is taken from the RRP3 and the project preparatorytechnical assistance (PPTA) report that formed its basis. The analysis isunusual in that it contains a rigorous rapid social assessment of householdsin the area of project influence. However, in the economic analysis of theRRP there was no attempt to integrate the social information with theProject's economic impact.

3 Loan 1716-SRI: Coastal Resource Management, for $40 million, approved on 7 December 1999.


Untitled-1 5/17/02, 10:26 AM109

Project Description

The Project will cover six districts in the Northwestern, Western andSouthern provinces, targeting about 76,500 households, representing about75 percent of the total households living in the coastal areas covered by thesix districts. The Project will have four components: (i) coastline stabilization,which will address the problem of coastal erosion and develop preventivemanagement schemes; (ii) coastal environmental and resource management,which will address coastal resource degradation and include activities toreduce pollution in lagoons; (iii) fisheries resource management and qualityimprovement, which will aim at sustainable coastal fisheries managementsupported by the construction of harbors/anchorages and ancillary facilitiesto improve fish quality and reduce handling losses; and (iv) institutionalstrengthening for the Ministry of Fisheries and Aquatic Resources Development(MFARD) and other agencies and community organizations concerned. Thetotal project cost of $80 million is to be cofinanced by ADB ($40 million), theNetherlands ($12.8 million), the Government ($27.2 million) and beneficiaries($0.1 million).

Economic Analysis

The benefits of the coastal stabilization component include (i) housesand buildings on the land to be saved; (ii) avoidance of removal andadministration costs, which would have been incurred if the land had eroded;(iii) avoidance of the interruption of people's work that would have been causedby the erosion; and (iv) the opportunity cost of land. In addition, without theProject, coastal erosion will turn away some tourists and reduce theirspending. A conservative estimate is made that both tourist arrivals andper tourist spending would decrease by 1 percent if the beaches were todeteriorate as expected without the Project. The economic internal rate ofreturn (EIRR) for this component is estimated at 17.2 percent. Without thetourism benefits, the EIRR would be only about 5 percent.

The coastal environment and resource management component isdesigned to address the unsustainable and conflicting uses of coastal and


Untitled-1 5/17/02, 10:26 AM110

marine resources in eight selected sites along the southwest coast. Theincremental benefits arise from the difference between resource use withand without the Project. It is assumed that the unsustainable use without theProject will lead to an incremental loss of 1 percent per year of the habitat/resource value. This assumed level is well below the 2.5–3.0 percent rate ofdeforestation in Sri Lanka and about equal to the annual loss of coral coverin the Hikkaduwa area. Only some of the benefits can be quantified and valuedfor inclusion in the analysis: e.g., mangrove vegetation-based products andfisheries products, sustainable recreation uses, and eco-tourism benefitsassociated with the preservation of mangrove, wetland, and reef biodiversity.The overall EIRR for the eight sites is estimated at 20.6 percent.

Under the fisheries management and fish quality improvementcomponent, two fishery harbors and two anchorages will be constructed infour sites to serve fisheries boats. This will result in an increase in fish landings,a higher proportion of multi-day boats, and a greater number of annual tripseach boat will be able to make. In addition, the fisheries management andfish quality improvement activities are expected to lead to an increase in theharvest of fish (as a result of the shift from offshore fishing to deep-sea fishing),reduction of losses during fish handling, transport and marketing, and anincremental increase in the value of fish. The EIRR for this component isestimated at 12.1 percent.

For the Project as a whole, 20-year net incremental benefit streams ofthe three components and the cost of capacity building component arecombined. The overall EIRR is estimated at 15.3 percent. Regarding thesensitivity of project economic viability, either an increase in costs or a decreasein benefits by 20 percent reduces the EIRR to just over the acceptable levelof 12 percent. However, considering the other nonquantifiable benefits suchas institutional strengthening of MFARD, enhanced beneficiary participationin resource management, establishment of strong partnership among localgovernment, the communities and NGOs, etc., the Project is consideredjustified.

Due to the nature of the project, in which there are no revenue-generating project entities, financial analysis was not carried out. However,with more resources devoted to stakeholder identification and thequantification of benefits, it would have been useful if an attempt had beenmade to allocate income changes among the various stakeholders: e.g.,fishery-related operators, distributors/wholesalers, tourism operators,unskilled laborers, etc.


Untitled-1 5/17/02, 10:26 AM111

Use of Social Assessment in Beneficiary-Targeted Project Design

The preparation of the Project provides a good example of how a wellstructured social assessment helps project design for targeting beneficiaries.A rapid social assessment (RSA) was carried out within the identified projectarea to assist the design of the coastal resource management component.The RSA of 750 households employed a combination of structured surveysand stakeholder workshops and focus group discussions (FGDs) amongselected communities within the identified project sites. The results of theRSA and FGDs were analyzed and, together with other pertinent secondaryinformation, discussed with representatives of local institutions and selectedcoastal communities and districts.

As shown in Table A6.5, of the total population of about 100,000households in the coastal areas of the six districts, about 24 percent fall belowthe poverty threshold. Among the people above the poverty threshold, 13percent are considered very poor, 22 percent are poor, 17 percent are lowincome earners, 13 percent are moderate income earners, and 11 percent arehigh income earners. Unfortunately, however, there is no technicaldocumentation in the original PPTA regarding the statistical inference ofthe income poverty measurement except that the households were chosenrandomly. Therefore, it is difficult to tell the reliability of the details of householdincome levels presented. The RSA here is best seen as a device to revealbeneficiary demand qualitatively and to serve project design rather than asubstitute for poverty measurement per se for the project area. Appendix 3addresses the issue of how the use of existing household survey data couldassist PPTA in a cost-effective way of estimating the proportion of the poorbeneficiaries in the project influence area.

The last two brackets in Table A6.5 include landowners, large vesselowners, large processors, multi-day boat owners, prawn farm owners, boatrenters, lime kiln operators, and small mechanized boat owners. Coastalhouseholds will be assisted in identifying and managing livelihood activitiesthat will give them additional income and improve their quality of life. Smallfish processors and small-scale fishers, who rank among the low-incomegroups, will be able to gain access to livelihood credit for the purchase ofinsulated fish boxes, processing equipment, packaging units, and workingcapital. The health and personal well-being of community residents will also


Untitled-1 5/17/02, 10:26 AM112

Tab

le A

6.5

�� *�� =��

"��3��4��&�

'��

�4��5

"6��

��5�

7��5��

8�&�

9��

'��

��

0��

��6�

:��

��'��

+��6�#+�%

:��

> 1

00

,00

01

,38

03

,48

75

90

2,5

90

1,2

56

1,8

91

11

,19

51

1

80

,00

1 –

10

0,0

00

1,6

42

4,7

55

73

82

,82

21

,67

51

,56

01

3,1

92

13

63

,00

1 –

80

,00

02

,01

56

,14

97

58

3,9

59

2,4

05

2,1

43

17

,42

81

7

45

,00

1 –

63

,00

02

,96

76

,43

41

,10

74

,67

53

,22

93

,35

72

1,7

69

22

ab

ove

po

vert

y li

ne

1,8

77

4,1

52

69

42

,73

81

,48

61

,82

81

2,7

74

13

30

,00

1 –

45

,00

0

Be

low

po

vert

y li

ne

1,8

49

3,4

87

55

61

,85

31

,59

42

,11

21

1,4

51

11

18

,00

1 –

30

,00

01

,09

02

,09

23

00

1,4

11

1,1

75

1,4

34

7,5

03

7

9,0

01

– 1

8,0

00

98

01

,14

11

77

1,0

11

68

91

,43

45

,43

26

< 9

,00

0

Tota

l

ho

use

ho

lds

13

,80

03

1,6

97

4,9

19

21

,05

81

3,5

09

15

,75

91

00

,74

21

00

Ho

use

ho

lds

belo

w p

ove

rty

lin

e3

,91

96

,72

01

,03

34

,27

53

,45

84

,98

02

4,3

85

Pove

rty

inci

den

ce (

%)

28

21

21

21

26

32

24

Targ

et

ben

efi

cia

ries

as

perc

en

tag

e

of

tota

l h

ou

seh

old

s7

87

47

37

37

87

87

6


Untitled-1 5/17/02, 10:26 AM113

improve with the provision of clean water, latrines, and sanitary drainagefacilities. Community education and social preparation will raise theawareness of low-income community households.

The RSA revealed that women in the coastal areas are engaged in avariety of tasks aside from housekeeping and child-rearing, as shown in TableA6.6. With the increasing need for women to contribute to household income,they will benefit from the skills training to be provided under the Project.Women in the coastal communities showed a high level of awareness of sharedproblems and in several places, they have unknowingly formed informalgroups composed of relatives and neighbors for income augmentationpurposes. In some areas, women's groups have been purposefully organizedby NGOs for credit assistance, development of income-generating activities,skills training, and provision of social infrastructure. Women are envisionedto be the targets of livelihood development activities such as small-scale fishprocessing, handicraft making, garment-making, and the like and will be ableto access microcredit for these projects. They are also expected to contributesignificantly to the promotion of programs, practices, and services concerninghealth, hygiene, and sanitation. The active and targeted support of the NGOs/CBOs will assist in educating them and opening their eyes to better prospects.

Combining Social and Economic Analysis

As noted above no financial analysis of the Project is carried out sincewith the exception of minor toll charges for use of the harbors, this generatesno income for the project entity. Some economic pricing is done by applyingan SCF and a specific labor CF. However, since the data are only given insummary form in the RRP and its associated appendixes, it is not possibleto conduct a full distribution analysis following the procedures in Appendix25 of the ADB Guidelines (1997). Nonetheless it is possible to get a broadpicture of the beneficiaries from a comparison between project economiccosts (which are all borne by the Government) and project economic benefits(which all go directly to the private sector). These are summarized in TableA6.7, with present values discounted at 12 percent.

The significant point here is that the Project creates three main sets ofbenefits, corresponding to its three separate components, and the share ofthe poor will probably be very low in at least two of these. A poverty impact


Untitled-1 5/17/02, 10:26 AM114

calculation requires estimation of the proportion of the three types of benefitgiven here that accrues to the poor.

The RSA uses a poverty line of SLRs30,000 per year (approximately$1.2 per day) for the very poor and one of SLRs45,000 (approximately $1.8 perday) as threshold for the poor. In the study areas 24 percent of householdsare very poor and 37 percent are poor by these definitions. From the RSA themain activities of the very poor are as part-time fisherfolk, canoe repairmen,casual workers and coir fiber/rope makers. For those with an average incomeof SLRs30,000–45,000 the main activities are listed as net repairs, fish selling,small agriculture, and coral collecting. Although the fishing community istargeted as a major beneficiary both fishing crews and lagoon fishers areshown as above the poverty line and those who own mechanized boats orwho are fish distributors are well above it. With this helpful poverty profilein mind it is clear that a significant share of project benefits will not go directlyto those below the poverty line.

Coast stabilization benefits are described as those that arise fromthe coastal land that will be saved with the stabilization of erosion due tothe project. The most important of these benefits are (i) the house valuessaved, (ii) the opportunity cost of agricultural land [given by the per hectarevalue of copra and paddy], and (iii) the impact on tourism [since tourismarrivals and income are assumed to fall without the project]. Hence if thepoor do not own permanent houses on the coastal area, do not farm significantplots and are not linked with tourism, they will not be expected to gain fromthe net income generated through improved coast stabilization.

Coastal resource management benefits are in the form of the avoideddecline in resource value that would occur in the absence of the project; anannual rate of decline of 1 percent in the without-project case is assumed.Resource value here relates to the sustainable use of coastal products suchas mangrove vegetation based products and fisheries, with some allowancefor biodiversity and tourism benefits. Here the main link with the poor islikely to be through lagoon fishing and some subsistence use of vegetation-based products.

The benefits from the harbor and fishery components are quantifiedby the economic value of the incremental fish catch due to the project. TheRRP stresses that the primary poor beneficiaries will be fisherfolk, who willface greater employment opportunities on boats and higher catches andhigher fish prices, through improved quality. A figure of 1,000 new jobs forfishers is mentioned. However, how far these will be filled by those belowthe existing poverty line is unclear and the main income gains from higher


Untitled-1 5/17/02, 10:26 AM115

�� >��-��!�� &�<��(

375,000350,000

325,000

300,000

275,000250,000

225,000

200,000

175,000

150,000

125,000

115,000

100,000

85,000

75,000

60,000

B��

Fish distributors/wholesalersbusinessmen

Multi-day boat owners/fishers

Prawn farm owners

Trawler owner/fishers

Boat renters

Lime kiln operatorsboat repair operators

Government employeesMechanized small-boatfishers

Prawn farmersSmall fish processors

Traditional/canoe fishersLagoon fishersCoral miners

:��

Large fish processorsTradersTrawler owner/fishers

Small fish plant ownersLandowners/businessmen

Boat renters

Government employees

Table A6.6

� �� +�� +�� + ��

)�� ;��

moderateincomeearners

justabovepoor


Untitled-1 5/17/02, 10:26 AM116

)�� ;��

Resource management measures which will benefit all fishery-related residents in thetarget communities and beyondInfrastructure improvement program to enhance, expand, and upgrade landing placefacilities to improve product quality and provide new processing opportunities.

Environment protection awareness programTraining in milkfish aquaculture

Improvement in fish handling, storage, delivery, and marketing methodsImprovement in fish landing facilitiesTransfer of appropriate fish processing technologiesImprovement of craft design including refrigeration of fish loads

Environment protection awareness programMicroenterprise development and credit assistanceImprovement of craft designSkills training

Microenterprise development and credit assistanceSkills training

Microenterprise development and credit assistanceSkills training

Access to beach.Provision of NGO/CBO support and extensionTraining in milkfish aquacultureImprovement of fish handling, storage, delivery, and marketing methods

Transfer of appropriate fish processing technologiesEnvironment protection awareness programAlternative livelihood assistance and credit


Untitled-1 5/17/02, 10:26 AM117

Table A6.6 .. cont’d.

� �� +�� +�� + ��

50,000

40,000

30,000

20,000

15,000

10,000

Fish vendors

Fishing boat crew

Net repair personsSmall agriculture operatorsWage earners

Mobile fish sellers

Part-time fishers

Canoe repairmen

Casual workers

Subsidy recipients

Lagoon fishers

Fish vendorscoral collectors

Mobile fish sellers

Part-time fishersWage earnersCoir fiber/rope makers

Casual workers

Subsidy recipients

poor

povertythreshold

�� >��-��!�� &�<��(

)�� ;��B�� :��


Untitled-1 5/17/02, 10:26 AM118

Microenterprise development and credit assistanceSkills trainingProvision of NGO/CBO support and extensionImprovement in fish handling, storage, delivery, and marketing methods

Transfer of appropriate fish processing technologiesEnvironment protection awareness programMicroenterprise development and credit assistanceSkills training

Environment protection awareness programMicroenterprise development and credit assistanceSkills trainingProvision of basic social infrastructure, freshwater supply and latrinesProvision of NGO/CBO support and extension


)�� ;��

Untitled-1 5/17/02, 10:26 AM119

revenue from fishing are likely to accrue to boat owners and merchants. Ingeneral a higher level of economic activity in the area, whether through highertourism or fishing income, can be expected to improve the position of thepoor to some extent, through a loose "trickle down" process, but this Projectdoes not appear as a particularly closely targeted poverty intervention.

What would be a useful further input into economic analysis of theproject from the RSA are data on income earnings for poor households fromdifferent sources—such as small farming, fish vending, fishing and othercasual employment. This would allow estimates of how far the incomes ofthe poor would rise with the Project and hence given the number of poorhouseholds would allow estimates of their proportionate share in projectbenefits and the numbers who would move out of poverty.

One can also use the wage cost data on fishing to estimate incomegains to newly employed fishers, if these are taken to come from the poor.However, it should be remembered that logically the income gain to a newworker is not the wage paid but the wage minus the worker's opportunitycost. In this case a labor CF of 0.8 is used, which implies that only 20 percentof the wage paid is actually an income gain to poor workers.

Table A6.7

"�� !�� -��+��

(SLRs million)

Present values at 12%

Total Costs 2 ,191Total Benefits 2 ,850

of which coast stabilization 1,517coastal environment management 630fisheries and harbors 703

4�� , % "�7


Untitled-1 5/17/02, 10:26 AM120

Case 4 – Lao PDR: Primary Health Care Expansion Project

The following is taken from the RRP.4 EDRC staff participated in theproject preparation at the loan fact-finding stage. The economic analysis paidspecial attention to financial sustainability (potential for cost recovery) andfiscal affordability (to cover the project recurrent costs). While in health andeducation projects financial sustainability continues to be a major concern,quantification of economic efficiency (cost-effectiveness) in service deliveryand distributional impact will need more attention in order to demonstratepoverty impact. The following is an attempt in this direction within data andmethodological constraints.

Project Description

The Project will build on the lessons learned from the first ADB-assistedPrimary Health Care (PHC) Project in the Lao People’s Democratic Republic(Lao PDR) and expand its coverage from two to seven provinces in the North,while at the same time improving the quality of services, as well as assistingthe Government in forming a nationwide PHC network. The Project has twocomponents. Component 1 will develop PHC in the northern provinces by (i)increasing access to PHC at health centers and village levels; (ii) improvingthe quality of PHC including training of ethnic minorities staff; (iii) strengtheningmaternal and child health and family planning services; and (iv) supportingvillage health care and promotion. Component 2 will strengthen theinstitutional capacity for PHC nationwide by (v) strengthening PHCcoordination; (vi) standardizing management systems; (vii) supporting staffdevelopment and training for PHC management; and (viii) testing innovativefinancing approaches. The total project cost of $25 million is to be financedby ADB ($20 million) and the Government ($5 million).

Economic Analysis – Cost-Effectiveness Approach

Measurement of economic benefits associated with preventive healthcare could ideally accommodate a diverse set of benefits such as productivity

4 Loan 1749-LAO (SF): Primary Health Care Expansion, for $20 million, approved on 24 August

2000.


Untitled-1 5/17/02, 10:26 AM121

gains due to fewer life years lost to mortality and morbidity, consumptiongains due to a higher quality of life, and increased life expectancy and costsavings on curative treatment. In theory, under certain restrictive assumptions,the willingness of health beneficiaries to pay for health services will be theappropriate measure of the benefits to society from these services. In practice,however, partly due to the technical difficulty and often unjustifiably highresearch costs of estimating unbiased willingness to pay, the most commonlyused approach is to focus on the production side of health effects. The economicanalysis for the first PHC project (appraised in 1995) estimated economicbenefits based on the expected gained lifetime income due to the reducedunder-5 mortality rate (U5MR). This method leaves out the benefits fromreduced morbidity and also the consumption value of improved health status.A fuller discussion of this approach and its limitations is in the ADB Handbook

for the Economic Analysis of Health Sector Projects (2000). For the currentProject, cost-effectiveness analysis has been applied as a well accepted testfor health projects.

Cost-effectiveness of health interventions is often measured as cost perDALY (disability-adjusted life year) saved. A DALY is a measure that combinesmortality, morbidity, and disability weighted for years of life saved at differentages, with social preference in favor of working-age years reflected in theimplicit weights. Despite its limitation, DALYs provide a quantification ofthe burden of disease comparable across alternative health interventionsand are increasingly used in developing countries. For the purpose of theanalysis, DALYs are used to provide an indication of the current burden ofdisease from a selected set of medical conditions addressed by the Project.

For accuracy, DALYs should be calculated from data specific to thecountry. The requisite data for these calculations is not yet available in theLao PDR, however, so a second-best approach for the estimation of DALYshas been taken. A large international database of disease burdens for differentgeographic regions and two large countries (People’s Republic of China andIndia) is available (Murray and Lopez 1996). These data give total DALY lostfor each region for a large number of medical conditions, and DALYs are givenboth in total and for different age groups. It is assumed that the Lao PDR hassimilar epidemiological conditions to the region which also includes Thailand,Cambodia, Viet Nam, and Indonesia (categorized in DALY database underOther Asian and Islands Region). An adjustment was made for the populationsize of the Lao PDR relative to this comparator region to obtain estimates forDALYs lost by age group, sex, and medical condition for the Project provinces


Untitled-1 5/17/02, 10:26 AM122

Table A6.8

��,A� ,��

8��&�� 4��&��&�� &��

Hemorrhage 0.0056 Pertussis 0.0178 Diarrheal diseases 0.0517

Sepsis 0.0119 Poliomyelitis 0.0057 Respiratory Infections 0.0690

Eclampsia 0.0015 Diptheria 0.0006 Malaria 0.0089

Hypertension 0.0019 Measles 0.0583 STDs excluding HIV 0.0047

Obstructed labor 0.0077 Tetanus 0.0295

Abortion 0.0020 Vitamin A deficiency 0.0351

Perinatal conditions 0.1830

;��$ <..�� (�/� �4��&��<.�� (//�$ �� ($/./

Source: Adapted from 1993 World Development Report, augmented by 20% to adjust for time lapse since the research year of 1990.

(again see ADB Handbook for the Economic Analysis of Health Sector Projects

for further information.)The target population in the seven provinces is categorized into three

groups and their health conditions to be addressed by the Projectinterventions are given for each group as in Table A6.8. Per person incidenceof DALY for each health condition is taken from the above source and adjustedto the Project target population. The greatest difficulty in economic analysisof health projects lies in the uncertainties of the reduction of disease burden,or saving in the otherwise lost DALYs every year specifically accruable tothe Project. The incremental DALYs saved due to the Project also dependon the assumed counterfactual scenario without the Project. In some cases,where interventions involve price increases of public services, both consumerresponse and private sector response must be taken into account to projectthe future utilization of the project services and their associated healthimpacts. Because PHC services are targeted primarily to the poor and thenew pricing policy on medical fees and consumables will not significantlyinfluence the target population's utilization of the PHC services. Here, the"before-project" baseline health situation is simply taken as the "without-project" situation and different scenarios (low, moderate, and high) havebeen prepared to test the robustness of cost-effectiveness indicators todifferent effective PHC coverage of the target population. The low scenarioassumes that the percentage of the target population reached and full impactof the PHC intervention on mortality and morbidity realized will gradually


Untitled-1 5/17/02, 10:26 AM123

increase to 10 percent during Project implementation and stay so thereafter.The moderate and high scenarios assume the percentage will graduallyincrease to 15 percent and 20 percent, respectively, during the projectimplementation period and stay so thereafter.

The Project life is assumed to be 15 years, beyond which the projectinvestments will cease to yield health benefits and leave no residual value.Investment costs include construction and upgrading of health units andhospital buildings, medical equipment, drugs and supplies, four-wheel drivevehicles and motorbikes, various consulting services and researchers, andtraining and materials. Recurrent costs include maintenance costs of buildings,equipment and vehicles, and operational costs of supervising health careworkers and utilities and communications for PHC offices. Both financial andeconomic costs per DALY saved are calculated. Financial cost per DALY savedshows cost-effectiveness from the budgetary viewpoint, whereas economiccost per DALY saved shows the impact on the economy in general. In economiccosting, the world price numeraire (US dollar denomination) was used. Thisdoes not prove inconvenient for distribution analysis as the Project benefitswill not be valued. In converting the financial prices of nontradable goodsinto their economic costs, a standard conversion factor (SCF) of 0.9 was used.Estimation of a shadow wage rate factor (SWRF) was problematic as the officialunemployment stands at only 0.9 percent (primarily due to high prevalenceof self-employment in subsistence agriculture). Considering this figure is closeto unity and the uncertainty over the exact unskilled labor content in thelump-sum contract of civil works, SWRF was not used in economic pricingof labor costs. As is obvious in the case of the Project areas in the Lao PDR,the cost of time to the nearest health service unit is borne by the beneficiariesand should be valued as additional economic costs. However, the analysishere did not include such costs mainly due to the lack of detailed data on thequantity and value of the opportunity cost of travel time. Based on theseassumptions, a 3 percent discount rate was used in discounting project costsand output stream (DALY saved), as discussed in the Handbook for the

Economic Analysis of Health Sector Projects (ADB 2000), to enable an

:�� $�� <2� �� 4�� $�� <2� ��

Low Scenario: $49.02/DALY Low Scenario: $45.82/DALY

Moderate Scenario: $30.08/DALY Moderate Scenario: $28.11/DALY

High Scenario: $22.56/DALY High Scenario: $21.09/DALY


Untitled-1 5/17/02, 10:26 AM124

international comparison of different health interventions. The results are asfollows.

ADB's Policy for the Health Sector lists costs per DALY saved for somecomparable health interventions: immunization at $25, family planning at $25-75, malaria prevention using impregnated bednets at $15–20. The high-scenariocost is considered well within the acceptable range compared with thisinternational norm. The low-scenario cost is much higher than the internationalcomparator, which may be explained by the assumptions regarding the costof services, in particular family planning services, which appears quite high,and the geographical condition in the Lao PDR. This would justify a relativelyhigher cost per DALY saved.


Poverty Incidence

The PPTA household survey chose rice shortage (whether the householdhad enough rice during the previous year) as an indicator of poverty. Fieldsurveys found that poverty in rural villages in the Lao PDR in subsistenceproduction is mainly linked with lack of workforce and land shortage. Lackof both combined is best reflected in rice shortage. But this does not allowcomparison between provinces and regions. As of this writing, the Governmentof Lao PDR has not yet decided on an official poverty line. Two alternativelines developed by the World Bank and the Swedish Statistics Office giveconflicting information on poverty incidence. The latest ADB study (Kakwaniet al. 2001) attempted a calculation of a poverty line that is consistent betweenthe 1992/93 Lao Expenditure and Consumption Survey (LECS1) and the 1997/98 update (LECS2), based on which the average total poverty line for the LaoPDR is estimated to be 19,184 kips per person per month in 1997/98 prices(about $9 as opposed to the World Bank criterion of $14). Taking this povertyline for the current analysis, poverty incidence in the target provinces isindicated in Table A6.9.

Out of the total target population, about 1 million people will be underdirect project influence. Using the 1995 census data, about 155,000 (14.9percent) will be children under five years of age and 209,000 (20 percent) willbe women of reproductive age. These two groups will be the major beneficiariesof the reduced burden of diseases due to the Project. By ethnicity, the population


Untitled-1 5/17/02, 10:26 AM125

)��"�7� ��

��77�/7��77#/7*'��$-��,��0��-�B�3��

Bokeo 63.5 37.4 -10.6 0 out of 5Louang Namtha 60.3 57.5 -1.0 2 out of 5Phongsali 68.7 64.2 -1.3 3 out of 7Houaphanh 78.4 74.6 -1.0 5 out of 6Oudomxai 51.1 73.2 7.2 2 out of 7Xieng Khouang* 57.3 34.9 -9.9 1 out of 7Louang Prabang 62.7 49.4 -4.8 2 out of 10Sayaburi 30.1 21.2 -7.0 0 out of 10Northern Region 58.4 52.5 -2.1 14 out of 50 <�� + �*�" C�� *��*

Source: Kakwani et al. (2001). Note: Xieng Khouang province belongs to Central Region.

directly covered by Component 1 of the Project will be distributed as 334,700persons (35.6 percent) to the Tai-Kadai group, 336,600 persons (35.8 percent)to the Mon-Khmer group, 163,400 persons (17.4 percent) to the Hmong-Miengroup, and 106,400 persons (11.3 percent) to the Sino-Tibetan group. Usingthe poverty incidence above for the Northern Region as an approximation,the poor beneficiaries out of the target population will be about 53 percent.In Louang Prabang, the Project will work mainly at the provincial hospitallevel for this component. Although the district level population and belowwill certainly benefit indirectly, these are excluded from the direct beneficiariesas a conservative estimation.

Differential Project Impact

The percentages of the poor in the above table can be considered asthe first order approximation of the poverty impact ratio on a headcountbasis, that is, 53 percent of the project benefits are considered to go to thepoor. But this simple interpretation is based on an unlikely assumption, thatis, neutrality in the burden of diseases falling proportionately on the poorand nonpoor. In reality, it is very likely that the Project will benefit the poormore than this number indicates for the following reasons.

(i) Disease is distributed to the poor more proportionately than tothe nonpoor. A probit analysis using the PPTA household surveydata shows that the poor (defined as those experiencing rice


Untitled-1 5/17/02, 10:26 AM126

shortage) are 28 percent more likely to have suffered from anyillness in the previous month than the nonpoor, and 21 percentmore likely to have suffered from severe illness. Therefore, DALYssaved per unit of Project service will be distributed more thanproportionately to the poor than the headcount indicator suggests.More detailed research on differential health problems for the poorand nonpoor would have allowed a clearer projection of thedistribution of DALYs saved.

(ii) On the other hand, the same analysis indicated no clear differencesbetween poor and nonpoor in health-care-seeking behavior forvarious health providers and health goods expenditure. Thisindicates that the Project services, especially preventive andpromotive services, will be equally utilized by both the poor andnonpoor.

(iii) The analysis found that poverty is not linked with the degree ofremoteness of villages. Every village has its poor households andpoverty is linked to social conditions (addiction, single parents,age profile, relatives) and land availability in the individualhouseholds. Each village is likely to have some households whocannot produce enough and are trapped in poverty and lack offood. An effort to reach the very poor would therefore need totarget individual households, in addition to some specific and smallethnic minorities. Subsidies (or differential user charges) can beused precisely for this purpose.

(iv) Regarding first-level curative care, many patients come toprovincial/district hospitals from far distances simply due to theunavailability of the basic service in their locality and in manycases spend significant economic costs to arrive there. TheProject services at lower level health facilities will create accessfor the poor who cannot currently afford the travel cost and thiseffectively constitutes an economic subsidy targeted at the poor.

There is also an evidence for a differential impact of health interventionsbetween rural and urban areas. A PPTA study for Viet Nam (Dunlop 1999)attempted to quantify the distribution of DALYs lost. It found that burden ofdisease among the poorest rural young children is more than 27 times therate experienced by those in urban areas. For the entire population, the poorrural population has four times as great a disease burden as urban dwellers.


Untitled-1 5/17/02, 10:26 AM127

As this Project is also exclusively targeted to northern rural provinces, thesame differential impact will benefit the poor disproportionately. Due to thelikely underestimation of poor beneficiaries, it could be that their share in tothe total is close to 75 percent, which can be seen as a target for internationalbest practice for poverty-focused interventions.

Case 5 – Tajikistan: Road Rehabilitation Project

The following is based on the RRP5 and the associated PPTA consultantreport on poverty impact analysis. Clear terms of reference for the consultantwere prepared to carry out a separate distribution and poverty impact analysisto extend the preceding conventional economic analysis. A good attempt wasmade to collect primary data through field surveys in order to identify anddisaggregate beneficiaries of road infrastructure. A good judgment wasapplied to the extent of the use of the secondary information from the existinghousehold survey data. A sensitivity analysis was carried out to relate theimportance of the market structure of road transport services to the povertyimpact of the project.

Project Description

Tajikistan is the poorest of Central Asian republics. The breakup ofthe former Soviet Union resulted in a dramatic decline of economic activityand increase in poverty. Rehabilitation of the country's road infrastructureis necessary to sustain the economic recovery now under way. The scopeof the Project includes rehabilitation of approximately 80 km of the mostdeteriorated sections of the Dushanbe-Khulyab road and improvement ofapproximately 150 km of rural roads including grading dirt road surfaces andbuilding appropriate drainage facilities. Of the total project cost of $26.8 million,$20 million is to be financed by ADB, $4 million by OPEC Fund, and $2.8 millionby the Government.

5 Loan 1819-TAJ(SF): Road Rehabilitation, for $20 million, approved on 20 December 2000.


Untitled-1 5/17/02, 10:26 AM128

Economic Analysis

The economic analysis was carried out on the basis of a comparisonof the with- and without-project scenarios. Without the Project, the road wouldreceive only minimal maintenance and continue to deteriorate. With the Project,routine and periodic maintenance will be provided according to internationalstandards. The analysis covers a period of 24 years (2001–2024), includingabout 4 years for project implementation. All benefits and costs were estimatedin constant 2000 prices. With no revenues involved and no financial analysiscarried out, it was decided that economic pricing be done using the worldprice numeraire. It would be normally simpler to use domestic price numeraireif distribution analysis is envisaged, but due to the uncertainty of currencyexchange rates, project cost estimation has been done in US dollars, therefore,this is a case where the choice of world price numeraire is simpler throughoutthe analysis. A standard conversion factor (SCF) of 0.9 was applied to thevaluation of nontradable inputs and a shadow wage rate factor (SWRF) of0.75 was applied to valuation of unskilled labor.

The Project's economic benefits are primarily derived from savings invehicle operating costs (VOCs) in normal traffic, passenger time savings, andbenefits arising from generated traffic (VOC savings estimated as half thevalue for normal traffic). The Project will indirectly improve the access ofthe rural poor to markets and social services, and improve safety conditionson the road. However, sufficient data are not available to quantify thesebenefits and they were not included in the analysis. Unit economic VOCsfor passenger and freight vehicles were estimated using the highway designand maintenance model (HDM-Manager). VOC savings will accrue primarilyfrom improvements in road surface, horizontal and vertical alignment, andincreased average speed in some sections. The estimation result showedsavings in VOCs comprise the largest category of benefits, accounting formore than 85 percent of the total benefits. EIRRs were calculated for thenational road sections and for the overall Project. The EIRR for the overallProject is 15.9 percent while the EIRRs for the national road sections rangefrom 14.2 to 25.4 percent. Sensitivity analysis is omitted here.


Untitled-1 5/17/02, 10:26 AM129


The distribution analysis followed the following steps.

(i) Disaggregation of road user benefits: (VOC and time savings)between passenger and freight vehicles: This was done by goingback to the base data on traffic forecasts and road user unit costsidentified for 4 classes of passenger vehicles and 3 classes of freightvehicles.

(ii) Field surveys to inform road user benefit incidence: In order toinform the distribution analysis and also to assess the structureand performance of the transport market in Khatlon Oblast andthe project area, the consultant designed 3 sets of questionnairesand undertook field surveys for (a) passengers, (b) drivers, and(c) farmers. The drivers survey was designed to learn about thestructure of the bus and trucking markets and the extent of directGovernment involvement in providing transport services. Thepassengers survey was designed to learn about fare structuresfor the various modes of transport and the degree to whichpassengers encounter problems with access to commondestinations such as hospitals and schools. The farmers surveywas designed to learn about freight usage of road infrastructureand its service cost. Income questions were included in all surveysto broadly capture the user profile. These surveys helped determinethe degree to which road user benefits are passed on from vehicleowners to users, and the degree to which the poor and theextremely poor will benefit from the project and what institutionalconstraints bar the poor from receiving a larger share of thebenefits from the Project.Survey results: The market for passenger transport services onthe project road and area is fairly competitive. Small owner-operators dominate 80 percent of the market, making itsubstantially competitive. Drivers' reported incomes and householdexpenditures fit the national pattern and this suggests that theyare not able to extract much, if any, from monopoly rent fromowning and operating vehicles on the project road. For freightservices, farmers are big users of the project road. They prefer to


Untitled-1 5/17/02, 10:26 AM130

sell the output themselves because they are not satisfied withthe prices offered by the middle marketers. Of those reporting tosell locally, 80 percent cited high transport costs (averaging about20 percent of the output price) as the reason for not going to moredistant markets. The surveys revealed that several institutionalbarriers prevent both transport and agricultural markets frombeing more competitive. These barriers will prevent the fullbenefits from accruing to the users of the transport systems. First,Government-owned bus terminals set fares for those buses usingthe terminals. The bus terminals are large and centrally locatedin all the major cities of the project area. Aside from blockingdrivers from passing on cost savings to passengers, this pushesthe transportation markets out of the bus terminals. The mostefficient solution would be for the bus terminals to set rental feesfor the bus bays at a low enough level to attract the buses awayfrom the "informal" high-cost markets to bus operators andpassengers. Second, outside Dushanbe, state-owned truckcompanies are reluctant to lease vehicles. The surveys showedthat outside the capital, there are no leased trucks or buses fromthe Government. Allowing drivers to lease vehicles will increasethe supply of operating vehicles that are free to seek marketbusiness, as opposed to waiting at stations for calls to move freightat Government-set prices. Third, along the project road, the statemotor vehicle inspectorate (GAI) personnel operate informal tollstations. Drivers reported being stopped at an average of 5 GAIcheckpoints while traveling their most frequent route. At eachcheckpoint, they were required to pay an average of Somoni(TJR)517. These tolls do not revert to the Government budget inany form.

(iii) Use of secondary data: The consultant supplemented the aboveinformation with the results from a large survey implemented bythe State Statistical Agency (SSA) in June 1999, known as theTajikistan Living Standards Survey (TLSS). It was designed as astratified random sample of 2,000 households covering 14,000individuals, and is representative of the country. Together withSSA, the consultant computed special tabulations for KhatlonOblast for the purpose of the Project. The sample size for the Oblastis sufficiently large to allow valid statistical inferences in mostcases. The TLSS data allowed estimation of income elasticity of


Untitled-1 5/17/02, 10:26 AM131

demand for transport for Khatlon Oblast. The result shows thatthe demand is income elastic (about 1.2). The implied incomeelasticity of demand for gasoline for transportation is even moreelastic (about 1.6), implying traffic growth will be somewhatdominated by passenger vehicles. The TLSS collected extensiveagricultural data, but the quality of the data is not sufficient todraw inferences, for example, on agricultural price elasticities ofsupply. Moreover, due to the dominance of state-owned farms inthe Oblast and the lack of transparency in the transmission ofrevenue from crops sold and farm workers' incomes, especiallycotton, computation of such elasticities is not meaningful at thistime.

(iv) Assumptions for distribution of road user benefits: Based on theabove steps, the road user benefits that will be passed on to userswere estimated at 20–50 percent depending on the vehicle type.The 50 percent estimate for passenger cars and buses reflectsthe consultant's judgment that these markets are competitive. The20 percent estimate for pickups and articulated trucks reflectsthe paucity of data and information on the use of these vehiclesin the project area and some conservative judgment. Since 1998there has been a significant de facto privatization of the vehiclefleet in Tajikistan. Therefore, the consultant's estimates onGovernment ownership across commercial vehicle classes aremuch lower than the 1998 data supplied by the Ministry ofTransport and Roads. Also, based on the field survey, the consultantestimated the share of Government-owned noncommercialvehicles as small. Passenger cars and pickups remain largely inthe private sector. Leasing of state-owned vehicles is onlypermitted for state fleets in Dushanbe. Therefore, it is estimatedthat a small share of the state's fleet of trucks and buses is leased,and that over half of the benefits from leasing the state vehiclesaccrue to the Government.

Using the above information and assumptions, all economic net benefitsare distributed among the following categories of stakeholders: (i) passengerusers, (ii) freight users, (iii) vehicle owners, (iv) labor, and (v) Government/economy. As some of the Project's capital expenditure and maintenance willbe spent on local labor, unskilled laborers will gain net benefits to the extentthat their paid wages are higher than their opportunity cost of labor (reflected


Untitled-1 5/17/02, 10:26 AM132

in SWRF). As there are no toll roads or other mechanisms for direct cost recoveryfrom road users, the Government (and the rest of the economy) will be thesole bearer of the economic resource costs associated with capital expenditureand maintenance. Because the initial scope of the Project analysis did notinclude rural feeder roads and the traffic forecast and associated VOC datawere generated only for the trunk road, the analysis here is also limited tothat of the trunk road. The summary result of the distribution analysis ispresented in Table A6.10. The total net benefits of about $8 million (in presentvalue discounted at 12 percent) does not include nonquantifiable but indirectbenefits such as increased trade and economic activity that are induced bythe improved road. The size of these indirect benefits depends on how theregional and national economies grow over the Project's time horizon.

As discussed in Case 1 of this appendix, to make the calculationconsistent with the world price numeraire, financial flow data must be multipliedby the SCF (even when the accounting currency unit is US dollars) beforestarting the distribution analysis. In the current case, however, because theoriginal financial data was not included in the consultant report, the financialflow figures in Table A6.10 have been derived by dividing economic flow figuresby the SCF. Therefore, the result is not as accurate as it could have been, butthe errors are unlikely to be important.

The public lender column was added to the original RRP primarily tomake a point that in principle, the gain/loss to the aid agencies can beaccommodated in the distribution table. The numbers associated with lendingarrangements in this example (underlined numbers in Table A6.10) are artificialdue to the lack of data on loan schedules for OPEC lending. The procedureto calculate these numbers would be to discount loan disbursement andprincipal and interest payment flows both at the opportunity cost of capitalof 12 percent. In the example below, the table shows that the public lender(ADB and OPEC) is a net loser and its net loss is matched by the equivalentamount of gain accruing to the project operation represented by the financialpresent value column. The intent of such refinement can be taken to illustratethat aid agencies by definition incur net loss for every loan they lend andpresent this point explicitly in the distribution analysis. However, thisrefinement does not change the aggregate project net benefits, or the povertyimpact ratio. This result arises when we assume that funds from aid agenciesare channeled through the government financing pool for public investment,and therefore, the gain/loss to the aid agencies can be pooled together withthe gain/loss to the government/ economy. Also, if the analyst prefers not toshow the distributional details associated with project financing, the columns


Untitled-1 5/17/02, 10:26 AM133


Tab

le A

6.1

0

�� !�� /��

$B%=333�� %&�� (

Fina

ncia

lEc

onom

icEc

onom

icPa

ssen

ger

Freigh

tVe

hicle

Labo

rPu

blic

Gov

t/

Pres

ent

Pres

ent

Minu

sU

sers

Use

rsO

wne

rsLe

nder

Econ

omy

Value

Value

Fina

ncia

l

"#��

Road

use

r

ben

efits

011

,655

11,6

554,

130

1,94

75,

578

Loan

3,79

70

(3,7

97)

(3,7

97)

��#��

Inve

stmen

t

and

O&

M(3

,797

)(3

,417

)38

038

0

Labo

r(2

60)

(195

)65

65

Intere

st(5

00)

050

050

0

Prin

cipal

(2,0

00)

02,

000

2,00

0

��$��%��

(2,7

60)

8043

10,8

034,

130

1,94

75,

578

65

(1,2

97)

380

Tota

l

Gai

ns a

nd Lo

sses

4,13

01,

947

5,57

865

(3,6

77)

8,04

3

Prop

ortio

n of

the

Poor

(%)

6060

3080

10

Net

Ben

efits

for t

he P

oor

2,47

81,

168

1,67

352

(368

)5,

004

Pove

rty Im

pact

Rat

io0.

62

Note

: Th

e n

et

loss

to P

ub

lic L

end

er

and

Gove

rnm

ent/

Eco

nom

y (-

3,6

77

) is

calc

ula

ted

as

-2,7

60

-1,2

97

+3

80

.

Untitled-1 5/17/02, 10:26 AM134


for the public lender and the government/economy can be collapsed into onecolumn, in which case the underlined numbers disappear but the rest of thetable remains unaffected (this can be easily verified in the table).


In moving from the above distribution analysis to poverty impact analysis,the following assumptions were made.

(i) Passenger users: In the consultant's survey, outside Dushanbe,60 percent of drivers, farmers, and passengers using thetransportation services in the project area are poor, using thepoverty line estimated in the TLSS. Based on the availableinformation, experience from other disadvantaged transitioneconomies, and expert judgment, this number can be taken tobe the percentage of the net benefits for passenger usersaccruing to the poor.

(ii) Freight users: Poor farmers are the most common users of freighttransport in the Oblast and the project area. Again, based on thesurvey result (e.g., farmers lose about 20 percent of their produce'svalue due to low quality road conditions), other availableinformation and expert judgment, the consultant estimated that60 percent of the net benefits for freight users will accrue to thepoor.

(iii) Vehicle owners: According to the consultant's survey, the pooruse taxis for transport to some extent and the TLSS data supportprivate passenger vehicle ownership among the middle threehousehold per capita expenditure quintiles. The consultantestimated that 30 percent of the net benefits to vehicle ownerswill accrue to the poor.

(iv) Labor: Given that the project road will be rehabilitated usingunskilled labor, among which a great majority come from thepoorest to poor households, the consultant estimated that 80percent of the benefits to labor will go to the poor.

(v) Government/Economy: The TLSS documents an extremely highheadcount poverty incidence (over 80 percent) and using theinformation in the TLSS leads to an equally extremely high income

Untitled-1 5/17/02, 10:26 AM135

share of the poor in GDP, compared to other DMCs. Consideringthe uncertain quality of household expenditure data, thefluctuating nature of the economic transformation under way inthe short run, and the long project life by comparison, this numberseems to provide little reliable clue at the moment as to theincidence of the net government/economy loss. Here simply therule of thumb of 10 percent is applied as a base case.

As in Table A6.10, the poverty impact ratio (PIR) based on the aboveassumptions is estimated at 0.62. While the result of the TLSS points to theextremely high poverty incidence nationwide on a headcount basis, it isdifficult to make a sensible judgment as to where a national reference pointshould be to compare with the PIR.

Due to somewhat uncertain assumptions, sensitivity analysis has beenperformed on some key parameters. First, the PIR sensitivity can be tested onthe poor proportion of the government/economy net loss. Under a scenarioin which the extreme poverty situation suggested by the TLSS resultcontinues over a good part of the project life, let us use 0.5 instead of 0.1for this parameter. Then the PIR is 0.44, which is lower than the base case,now due to the assumption that a higher proportion of the government netloss will be borne by the poor outside the project influence area. Second, wecan assume a case where transport services markets are less competitivethan the consultant's base case assumption and stay that way. Instead ofthe 20–50 percent assumed above for the road user benefits passed on fromvehicle owners to users, if we assume 10 percent for these parameters acrossall vehicle types, the PIR is reduced to 0.44. Third, we can also assume acase where transport services markets become much more competitive thanthe consultant's base case assumption and where drivers and vehicle ownersact more like price takers than price setters. If we assume that they passon 85 percent of their cost savings to passengers and freight users in theform of lower fares and freight rates, the PIR is increased to 0.80. Thisillustrates the importance of enhancing competition in the transport servicesmarkets not only to fuel economic growth but also to help reduce povertythrough road infrastructure.


Untitled-1 5/17/02, 10:26 AM136

APPENDIX�7Approximation of Income Share of the Poor

The following summarizes briefly how country-specific poverty surveydata can be used to derive a simple income share estimate for the poor; thatis, the share of current income going to those below the national poverty line.These figures are based on the headcount poverty incidence – HC (share ofpoor in total population) and the poverty gap – PG (average divergence ofincome of the poor from the poverty line) estimates produced by povertysurveys. With this data the income share estimates for the poor are derivedfrom the following formula:

Yp = ((1-PG)*PL*POP*HC) / GDPpc*POP= ((1-PG)*PL*HC) / GDPpc

Where Yp is the income share of the poor,PG is the poverty gap index,PL is the poverty line annual income per capita ,POP is total population,HC is the headcount poverty index, andGDPpc is GDP per capita.

As an example, a calculation for Viet Nam is illustrated as follows. The1999 poverty survey on Viet Nam, "Attacking Poverty," provides the necessarypoverty data. Strictly the report uses expenditure rather than income basisfor poverty, but here we implicitly assume zero savings by the poor, so thatall income is spent and the terms can be used interchangeably. For 1998 theyear of the survey, the national poverty line is set at 1,790,000 dong. This isnot directly comparable with the international $1 per day figure, since it isin nominal, not purchasing power parity prices. This is the higher of thetwo poverty lines referred to in the report, since the lower line covers onlyfood essentials in the basket of goods consumed by the poor. At this povertyline the proportion of the population in poverty is estimated to be 37.4 percent(so that HC = 0.374). On the other hand, the poor tend to be bunched around the

Untitled-1 5/17/02, 10:26 AM137


Table A7.1

�� C� ��

(based on country-specific p

$�� 8�� <�� >��2�� B�-�� 41�� $��

41�� !��1

Azerbaijan 1995 n.a. n.a. 0 .615Bangladesh

Urban 1995 Taka 1,233 14,796 0.263Rural 1995 Taka 658 7,896 0.511

CambodiaPhnom Penh 1996 Riel 1 ,578 575,970 0.111Other Urban 1996 Riel 1 ,267 461,360 0.299Rural 1996 Riel 1 ,117 407,705 0.401

IndiaUrban 1996 Rupee 353 4,236 0.256Rural 1996 Rupee 266 3,192 0.306

IndonesiaUrban 1999 Rupiah 92,496 1,109,952 0.194Rural 1999 Rupiah 74,272 891,264 0.260

Kazakhstan 1998 Tenge 3,716 44,592 0.434Kyrgyz Rep. 1997 Som 4,647 0.510Lao PDR 1997 Kip 19,184 23,028 0.617Mongolia 1995 Tugrik 7,240 86,880 0.365Nepal 1995 Nep Rupee 4,404 0.420Pakistan 1991 Pak Rupee 296 3,552 0.340Philippines 1998 Peso 10,284 0.367PNG 1996 Kina 461 0.375Sri Lanka

Urban 1995 SL Rupee 950 11,400 0.250Rural 1995 SL Rupee 950 11,400 0.410

Tajikistan 1998 Somoni 20,000 240,000 0.826Viet Nam 1998 Dong 1,790,000 0.374

Note: n.a. indicates either not available or judged not applicable.

/day/day/day

Table A7.1

�� C� ��

(based on country-specific poverty lines)

Untitled-1 5/17/02, 10:26 AM138

1 Introduction 139

Table A7.1

��

ountry-specific poverty lines)

�� D�;�� ;� �-�� !��1 &��5� E�� ;� � &'(

!��(

0.243 1997 WB Poverty Assessment n.a.

0 .060 2000 BinayakSen report to ADB 12,720 28.80.141 same as above 12,720 27.2

0.022 1996 WB/1999 WB Pov. Ass. 653,541 9.60.075 same as above 653,541 19.50.097 same as above 653,541 22.6

0.075 2000 ADB/K.R. Mohan, GOI 12,820 n.a.0 .093 same as above 12,820 n.a.

0 .020 2000 ADB Poverty Assessment 5,377,548 n.a.0 .018 5,377,548 n.a.0 .128 2000 UNDP HumanDevReport 114,781 14.70.180 1999 WB Poverty Assessment 6 ,445 30.20.138 1997/98 LECS2/Kakwani2000 455,633 26.90.109 1996 WB Poverty Assessment 239,345 11.80.121 1997 WB Poverty Assessment 10,776 15.10.071 1995 WB Poverty Assessment 9 ,211 12.20.130 1998 PovSurvey/Kakwani2000 35,633 9.20.124 1996HIES/1999WB 1,654 9.2

0.060 2000 ADB Poverty Assessment 36,572 7.30.110 2000 ADB Poverty Assessment 36,572 11.40.358 1999 TLSS 167,669 n.a.0 .095 1998 VLSS 4,790,558 12.6

APPENDIX 7 Sample Terms of Reference and Resource Requirements 139

Untitled-1 5/17/02, 10:26 AM139


poverty line so that the average income shortfall of the poor is roughly 9.5percent below the poverty line (PG = 0.095). In 1998 nominal prices GDPper capita was 4,790,558 dong. Population data are not necessary in thecalculation as they cancel out in the above formula. Putting these valuesinto the formula gives us an estimated income share of the poor ofapproximately 12.6 percent of GDP.

Y p = (1 - 0.095)*1790000*0.374 / 4790558= 0.126

Similar calculations have been carried out for DMCs where data havebeen collected from recent country-specific surveys as shown in Table A7.1.However, note that the results are only indicative as they are based on theexisting poverty surveys and subject to changes depending on the specificpoverty lines to be agreed between DMCs and ADB. Also, there may be adiscrepancy between the expenditure survey data and national account data.For example, it is well known that the survey-based expenditures areunderestimated in India and Indonesia compared to the estimate from thenational account exercise. For these two countries, in particular, it issuggested that the 10 percent rule-of-thumb be used for this parameter. Asthe preliminary calculations here used only the secondary sources indicated,country-specific discrepancy check will require going back to the originalsurvey data. The values for parameters should be adjusted accordingly asraw data become available.

Untitled-1 5/17/02, 10:26 AM140

1 Introduction 141

APPENDIX�8Sample Terms of Reference and Resource Requirements

Sample Terms of Reference for Poverty Impact Analysis under PPTA

Working closely with the Team Leader and in coordination with SocialAnalyst, the Poverty Impact Analyst should undertake the following tasks.(These tasks could be combined into the currently used terms of referencefor project economists. Also, sector-specific considerations could be added(Table A8.1).

Review the questions addressed and data collected during the Initialsocial assessment (ISA). Determine what information and data need to becollected during the PPTA in order to carry out a reasonable poverty impactanalysis. See ADB Handbook for Integrating Poverty Impact Assessment in

the Economic Analysis of Projects (2001) (referred to as the Handbook hereafter)for terms, methods, recommendations and available case illustrations.

Prepare a poverty profile of the project influence area. Clearly definethe poverty line to be used in the analysis. Use the line agreed in the PartnershipAgreement on Poverty Reduction between ADB and the DMC if alreadyavailable. Indicate the poverty incidence and characteristics of poverty andperception of the people in the project influence area. In doing this, collectand use as much available secondary survey data and information as possible(see Appendix 3 of the Handbook).

Provide a profile of the quantity, quality, and prices of project servicesavailable to different socioeconomic groups in the project influence area,particularly to the poor. Identify the potential beneficiary (and/or loser) groupsseparating those below the defined poverty line. Describe the current statusof the target population either in income/consumption or more generalsocioeconomic characteristics. Examine the need or demand for the projectby the target population, and where appropriate, assess how much they arewilling to pay or forego to receive its services.

Provide a statement of the project features relating to poverty reduction,including poverty-reducing measures, such as community participation,

Untitled-1 5/17/02, 10:26 AM141


�3��

Hefei-XianRailway

GuangxiHighwayDevelopment

UzbekistanRailwayModernization

Tajikistan RoadRehabilitationProject

� � )��

Following the Guidelines forthe Economic Analysis ofProjects

Following the Guidelines forthe Economic Analysis ofProjects (for main highwayPLUS provincial and countyroads).

Specific TOR: same asTajikistan plus "status ofseparation of socialservices” and PRA in 5communities

Clear TORs for the DA/PIR

� � � � � � � � ��

January 2000 (variation2 during TA which startedin September 1999)

April 2000 (TOR variationduring contractnegotiations)

June 2000 (variation 3during TA which started inOctober 1999)

June 2000 (variationduring TA which started inSeptember 1999)

�� 29%

� �� D��

��

DA = distribution analysis; PIR = poverty impact ratio; PRA = poverty rapid appraisal; TA = technicalassistance; TOR = terms of reference

Untitled-1 5/17/02, 10:26 AM142

1 Introduction 143

� � � � � $��

1 transport economist(intl.)1 social analyst (intl.)

1 transport economistsocial analyst (survey)

1 social analyst (intl.)1 social analyst (dom.)

1 economist-povertyspecialist (intl.)1 economic/financialspecialist (intl.)1 social analyst (dom.)

� � � � � � ��

2 weeks

2 weeks

Total: 1 month (intl.)

2 weeks1 day

Total: 0.5 months (intl.)

4 weeks6 weeks

Total: 1 month (intl.),1.5 months(dom.)

1 month

1 month

1 month

Total: 2 months (intl.),0.25 months (dom.)

:�� F� ��

US$25,000US$25,000

Total: 23,500(4% of max. contractpayment – US$600,000)

US$20,000

Total: US$10,000

(2% of max. contractpayment – US$535,000)

US$19,000US$1,6000

Total: 21,400


n.a.

n.a.

US$750

Total US$40,000


APPENDIX 8 Sample Terms of Reference and Resource Requirements 143

Untitled-1 5/17/02, 10:26 AM143

considered at project design and implementation stage. Explain the povertyimpact of the various alternatives; this can be quantitative (e.g., number ofjobs) or qualitative where quantitative data are lacking (e.g., access toimproved facilities). Justify the alternative selected on both economicefficiency (e.g., higher economic NPV) and poverty reduction grounds (e.g.,cost per unit of net benefits received by the poor.)

Carry out distribution and poverty impact analysis as outlined in theHandbook and estimate poverty impact ratio (PIR). Spell out the criticalassumptions and data sources required to estimate the PIR. Identify the keyparameters that include uncertainties and apply sensitivity analysis on thePIR, in addition to the currently practiced sensitivity analysis on FIRR/EIRR.Discuss the extent of the reliability of the PIR estimation and qualify the result.

Explain the risks of failure to achieve poverty objectives. Examine scopefor leakage of benefits to nonpoor. Consider the possibility of projectencountering financial difficulties. Identify any complementary activity thatwould reduce this risk.

Identify relevant (i) indicators, (ii) methodology, (iii) plan and timing,(iv) location, and (v) human resource requirements to undertake monitoringand evaluation of poverty impact in the project influence area. Ensure thatthese are introduced in the Project/Program Framework for the proposed projectand in the Project Performance Management System (PPMS).


Untitled-1 5/17/02, 10:26 AM144

Glossary

AAAAAverage incrverage incrverage incrverage incrverage incremental economic cost (AIEC).emental economic cost (AIEC).emental economic cost (AIEC).emental economic cost (AIEC).emental economic cost (AIEC). The present value ofinvestment and operation costs at economic prices, divided by the presentvalue of the quantity of output. Costs and output are calculated from thedifference between the without project and with project situations, and arediscounted at the economic opportunity cost of capital (EOCK).

Benefit-cost ratio (BCR).Benefit-cost ratio (BCR).Benefit-cost ratio (BCR).Benefit-cost ratio (BCR).Benefit-cost ratio (BCR). The ratio of the present value of the economicbenefits stream to the present value of the economic costs stream, eachdiscounted at the economic opportunity cost of capital. The ratio should begreater than 1.0 for a project to be acceptable.

Constant prices.Constant prices.Constant prices.Constant prices.Constant prices. Future price values from which any expected changein the general price level is removed. When applied to all project costs andbenefits over the life of the project, the resulting project statement is in constantprices. Expected significant changes in relative prices, that is, in expectedprice changes for an item compared with the expected change in the generalprice level, should also be incorporated in the valuation of costs and benefitsat constant prices.

Conversion factor (CF).Conversion factor (CF).Conversion factor (CF).Conversion factor (CF).Conversion factor (CF). Ratio between the economic price value andthe financial price value for a project output or input, which can be used toconvert the constant price financial values of project benefits and costs toeconomic values. Conversion factors can also be applied for groups of typicalitems, such as petrochemicals or grains; and for the economy as a whole, asin the standard conversion factor or shadow exchange rate factor. For detailsof CF estimation, see ADB Guidelines for the Economic Analysis of Projects

(1997).

Disability ADisability ADisability ADisability ADisability Adjusted Life Ydjusted Life Ydjusted Life Ydjusted Life Ydjusted Life Year (Dear (Dear (Dear (Dear (DALALALALALY).Y).Y).Y).Y). A composite health impactindicator that combines morbidity and mortality impacts. Its calculationrequires two sets of weights: first by extent of disability and second, byage for premature death and disability effects. For details, see ADB Handbook

for the Economic Analysis of Health Sector Projects, August 2000, or World

Development Report 1993, World Bank.

Untitled-1 5/17/02, 10:26 AM145

Discount rate.Discount rate.Discount rate.Discount rate.Discount rate. A percentage rate representing the rate at which thevalue of equivalent benefits and costs decrease in the future compared tothe present. The rate can be based on the alternative economic return inother uses given up by committing resources to a particular project, or onthe preference for consumption benefits today rather than later. The discountrate is used to determine the present value of future benefit and cost streams.

Economic internal rate of return (EIRR).Economic internal rate of return (EIRR).Economic internal rate of return (EIRR).Economic internal rate of return (EIRR).Economic internal rate of return (EIRR). The rate of return that wouldbe achieved on all project resource costs, where all benefits and costs aremeasured in economic prices. The EIRR is calculated as the rate of discountfor which the present value of the net benefit stream becomes zero, or atwhich the present value of the benefit stream is equal to the present valueof the cost stream. For a project to be acceptable the EIRR should be greaterthan the economic opportunity cost of capital.

Economic opporEconomic opporEconomic opporEconomic opporEconomic opportunity cost of capital.tunity cost of capital.tunity cost of capital.tunity cost of capital.tunity cost of capital. The real rate of return in economicprices on the marginal unit of investment in its best alternative use. This rateof return is estimated as the weighted average of the economic demand andsupply price of capital, and therefore will be equal to the value of the marginalunit of investible funds to both investors and savers.

Equalizing discount rate (EDR).Equalizing discount rate (EDR).Equalizing discount rate (EDR).Equalizing discount rate (EDR).Equalizing discount rate (EDR). The discount rate at which the presentvalues of two project alternatives are equal. It is the same as the internalrate of return on the incremental effects of undertaking an alternative withlarger net costs earlier in the net benefit stream rather than an alternativewith lower early net costs. The EDR is compared with the economicopportunity cost of capital to determine whether the alternative with largernet costs is worthwhile. Also referred to as the crossover discount rate,the discount rate above or below which the preferred alternative changesfrom one to another.

FFFFFactor intensityactor intensityactor intensityactor intensityactor intensity..... It refers to the degree to which a certain productionfactor (e.g., labor, capital, land) is used in production and characterizes thetype of production technology. For example, labor-intensive technology (e.g.,garment industry) refers to a production process in which more labor inputsare required than capital inputs relative to capital-intensive technology (e.g.,machinery industry).


Untitled-1 5/17/02, 10:26 AM146

Financial internal rate of returnFinancial internal rate of returnFinancial internal rate of returnFinancial internal rate of returnFinancial internal rate of return (FIRR).(FIRR).(FIRR).(FIRR).(FIRR). The rate of return that wouldbe achieved on all project costs, where all costs are measured in financialprices and when benefits represent the financial revenues that would accrueto the main project participant. The FIRR is the rate of discount for which thepresent value of the net revenue stream becomes zero, or at which the presentvalue of the revenue stream is equal to the present value of the cost stream.It should be compared with the financial opportunity cost of capital (FOCK),or the weighted average cost of capital (WACC), to assess the financialsustainability of a project.

FFFFFinancial opporinancial opporinancial opporinancial opporinancial opportunity cost of capital. tunity cost of capital. tunity cost of capital. tunity cost of capital. tunity cost of capital. The opportunity cost of usinginvestment resources at market prices in a project. This is often taken as theweighted average borrowing rate of capital used in the project

FFFFFinancial sustainabilityinancial sustainabilityinancial sustainabilityinancial sustainabilityinancial sustainability..... The assessment that a project will havesufficient funds to meet all its resource and financing obligations, whetherthese funds come from user charges or budget sources; will provide sufficientincentive to maintain the participation of all project participants; and will beable to respond to adverse changes in financial conditions.

Money-metric measure.Money-metric measure.Money-metric measure.Money-metric measure.Money-metric measure. The money-metric measure of welfare changeforms a fundamental building block in applied welfare economics. While anindividual’s welfare change is derived from physical consumption of goodsand services, it is necessary to make accounting by putting some money-metric units to the welfare change for any policy application. In a projectcontext, economic prices of commodities, which deviate from observed marketprices, represent the money-metric measure of welfare change when additionalunit of project input or output is added or taken away from the economy. Ina context of policy changes, there are other money-metric measures of welfarechanges such as compensating variation (CV) and equivalent variation (EV).For more details, readers are referred to any graduate level microeconomictextbooks.

Net present value (NPV).Net present value (NPV).Net present value (NPV).Net present value (NPV).Net present value (NPV). The difference between the present valueof the benefit stream and the present value of the cost stream for a project.The net present value calculated at ADB’s discount rate should be greaterthan zero for a project to be acceptable.

GLOSSARY 147

Untitled-1 5/17/02, 10:26 AM147

Numeraire.Numeraire.Numeraire.Numeraire.Numeraire. The common yardstick that measures the objective beingmaximized. In project financial analysis this yardstick is the real incomechange for the project participants valued in domestic market prices. Inproject economic analysis, because the scope of the analysis differs, andbecause domestic market prices do not always reflect the scarcity value ofproject outputs and inputs, this yardstick is the real change in net nationalincome for the project as a whole valued in economic prices. Generally, thereal change in net national income can be measured at two different pricelevels. These are the domestic price level, where all economic prices areexpressed in their equivalent domestic market price level values (the domesticprice numeraire), and the world price level, where all economic prices areexpressed at their equivalent world market price level values (the worldprice numeraire). As long as consistency is maintained in a particularcalculation across all project effects, project decisions will not be affectedby whether the domestic price level or the world price level is used to expressthe numeraire.

OpporOpporOpporOpporOpportunity cost.tunity cost.tunity cost.tunity cost.tunity cost. The benefit foregone from not using a good orresource in its best alternative use. Opportunity cost measured at economicprices is the appropriate value to use in project economic analysis for valuingnonincremental outputs and incremental inputs. For details, see ADBGuidelines for the Economic Analysis of Projects (1997).

PPPPPoveroveroveroveroverty impact ratio (PIR).ty impact ratio (PIR).ty impact ratio (PIR).ty impact ratio (PIR).ty impact ratio (PIR). The proportion of the total net benefits ofa project (NPV) that accrues to the poor. (It is possible that its value exceedsone as in Case 2 in Appendix 6.)

Present value (PV). Present value (PV). Present value (PV). Present value (PV). Present value (PV). The value at present of an amount to be receivedor paid at some time in the future. It is determined by multiplying the futureamount by a discount factor.

Real exchange rate.Real exchange rate.Real exchange rate.Real exchange rate.Real exchange rate. The price of foreign currency in terms of domesticcurrency where the rate of exchange is adjusted for the relative value ofactual or expected domestic and international inflation.

Risk analysis.Risk analysis.Risk analysis.Risk analysis.Risk analysis. The analysis of project risks associated with the valueof key project variables, and therefore the risk associated with the overallproject result. Quantitative risk analysis considers the range of possiblevalues for key variables, and the probability with which they may occur.


Untitled-1 5/17/02, 10:26 AM148

Simultaneous and random variation within these ranges leads to a combinedprobability that the project will be unacceptable. When deciding on aparticular project or a portfolio of projects, decision makers may take intoaccount not only the expected scale of project net benefits but also the riskthat they will not be achieved.

Sensitivity analysis.Sensitivity analysis.Sensitivity analysis.Sensitivity analysis.Sensitivity analysis. The analysis of the possible effects of adversechanges on a project. Values of key variables are changed one at a time, orin combinations, to assess the extent to which the overall project result,measured by the economic net present value, would be affected. Where theproject is shown to be sensitive to the value of a variable that is uncertain,that is, where relatively small and likely changes in a variable affect the overallproject result, mitigating actions at the project, sector, or national level shouldbe considered, or a pilot project implemented.

Shadow exchange rate (SER).Shadow exchange rate (SER).Shadow exchange rate (SER).Shadow exchange rate (SER).Shadow exchange rate (SER). The economic price of foreign currencyused in the economic valuation of goods and services. The shadow exchangerate can be calculated as the weighted average of the demand price and thesupply price for foreign exchange. Alternatively, it can be estimated as theratio of the value of all goods in an economy at domestic market prices tothe value of all goods in an economy at their border price equivalent values.Generally the shadow exchange rate is greater than the official exchangerate, indicating that domestic purchasers place a higher value on foreigncurrency resources than is given by the official exchange rate.

Shadow exchange rate factor (SERF).Shadow exchange rate factor (SERF).Shadow exchange rate factor (SERF).Shadow exchange rate factor (SERF).Shadow exchange rate factor (SERF). The ratio of the economic priceof foreign currency to its market price. Alternatively, the ratio of the shadowto the official exchange rate. This factor will generally be greater than 1. Foreconomic analysis using the domestic price numeraire, the SERF is appliedto all outputs and inputs, including labor and land, that have been valued atborder price equivalent values, with project effects measured at domesticmarket price values left unadjusted. The inverse of the standard conversionfactor.

Shadow wage rate (SWR).Shadow wage rate (SWR).Shadow wage rate (SWR).Shadow wage rate (SWR).Shadow wage rate (SWR). The economic price of labor measured inthe appropriate numeraire as the weighted average of its demand and supplyprice. For labor that is scarce, the SWR is likely to be equal to or greaterthan the project wage. For labor that is not scarce, the SWR is likely to beless than the project wage. Where labor markets for labor that is not scarce

GLOSSARY 149

Untitled-1 5/17/02, 10:26 AM149

are competitive, the SWR can be approximated by a market wage rate forcasual unskilled labor in the relevant location, and adjusted to the appropriatenumeraire.

Shadow wage rate factor (SWRF).Shadow wage rate factor (SWRF).Shadow wage rate factor (SWRF).Shadow wage rate factor (SWRF).Shadow wage rate factor (SWRF). The ratio of the shadow wage rateof a unit of a certain type of labor, measured in the appropriate numeraire,and the project wage for the same category of labor. Alternatively, the ratioof the economic and financial cost of labor. The SWRF can be used to convertthe financial cost of labor into its economic cost.

Standard conversion factor (SCF).Standard conversion factor (SCF).Standard conversion factor (SCF).Standard conversion factor (SCF).Standard conversion factor (SCF). The ratio of the economic price valueof all goods in an economy at their border price equivalent values to theirdomestic market price value. It represents the extent to which border priceequivalent values, in general, are lower than domestic market price values.The SCF will generally be less than one. For economic analysis using theworld price numeraire, it is applied to all project items valued at their domesticmarket price values to convert them to a border price equivalent value, whileitems valued at their border price equivalent value are left unadjusted. TheSCF and SERF are the inverse of each other.

Transfer payment.Transfer payment.Transfer payment.Transfer payment.Transfer payment. A payment made without receiving any good orservice in return. Transfer payments transfer command over resources fromone party to another without reducing or increasing the amount of resourcesavailable as a whole. Taxes, duties, and subsidies are examples of items that,in some circumstances, may be considered to be transfer payments.

WWWWWeighted average cost of capital (Weighted average cost of capital (Weighted average cost of capital (Weighted average cost of capital (Weighted average cost of capital (WAAAAACC).CC).CC).CC).CC). Measured on after-taxincome tax basis, WACC is determined by ascertaining the actual lending (oronlending) rates, together with the cost of equity contributed as a result ofthe project. To obtain the WACC in real terms, the inflation factor is to bededucted from the estimated cost of borrowing and equity capital.

WWWWWillingness to pay (WTP).illingness to pay (WTP).illingness to pay (WTP).illingness to pay (WTP).illingness to pay (WTP). The maximum amount consumers areprepared to pay for a good or service. WTP can be estimated as the total areaunder a demand curve. Changes in WTP can occur when the demand curveitself shifts because of changes in income or in the prices of substitutegoods.


Untitled-1 5/17/02, 10:26 AM150

References

Ali, I. 1990. “An Approach to Estimating the Poverty Alleviation Impactof an Agricultural Project.” EDRC Report Series No.47, Manila.

Anand, P.B. and R. Perman. 1999. “Preferences, Inequity andEntitlements: Some Issues from a CVM Study of Water Supply in Madras, India,”Journal of International Development 11: 27–46.

Asian Development Bank. 1978. “Economic and Financial Appraisal ofBank-Assisted Projects.” Occasional Papers No.11, Economic Office, Manila.

———. May 1994. Handbook for Incorporation of Social Dimensions in

Projects. Manila: ADB.

———. March 1996. Economic Evaluation of Environmental Impacts:

A Workbook. Office of Environment and Social Development. Manila: ADB.

———. February 1997. Guidelines for the Economic Analysis of Projects.

Economic Development and Resource Center (EDRC). Manila: ADB.

———. March 1999. Handbook for the Economic Analysis of Water

Supply Projects. EDRC. Manila: ADB.

———. June 1999. Project Economic Analysis and the Policy Context.

Unpublished draft, EDRC. Manila: ADB.

———. October 1999. Fighting Poverty in Asia and the Pacific: The

Poverty Reduction Strategy of the Asian Development Bank. Manila: ADB.

———. April 2000. Advisory Notes on Poverty Analysis, High Level

Forums, and Partnership Agreements. Strategy and Policy Department (SPD),Manila: ADB.

———. August 2000. Handbook for the Economic Analysis of Health

Sector Projects. EDRC, Manila: ADB.

Untitled-1 5/17/02, 10:26 AM151

———. November 2000. Loan Classification System. SPD, Manila: ADB.

———. 2001. “Integrating the Analysis of Risk into ADB’s EconomicAnalysis of Projects.” A draft methodological note. EDRC, Manila: ADB.

———. November 2002. Guidelines for the Financial Governance andManagement of Investment Projects Financed by Asian Development Bank,EDRC, Manila: ADB.

Baker, J. ed. 2000. Evaluating the Impact of Development Projects on

Poverty: A Handbook for Practitioners. World Bank: Washington, D.C. July2001. Integrating the Analysis of Risk into ADB’s Economic Analysis of Projects,a draft methodological note. EDRC, Manila: ADB.

Bigman, D. and H, Fofack, eds. 2001. Geographical Targeting for Poverty

Alleviation: Methodology and Applications. Washington, D.C: World Bank.

Bolt, R. and M. Fujimura. October 2001. “Policy Based Lending andPoverty Reduction: An Introduction to Processes, Options, and Assessments.”EDRC Working Paper No. 2, Manila: ADB.

Carson, R.T. 2000. “Contingent Valuation: A User’s Guide,”Environmental Science and Technology 34 (8):1413-1418.

Choe, K., D. Whittington and D.T. Lauria. 1996a. “The Economic Benefitsof Surface Water Quality Improvements in Developing Countries: A Case Studyof Davao, Philippines,” Land Economics 72(4): 519-537.

Choe, K., R.C.G. Varley and H.U. Bijlani. 1996b. “Coping with IntermittentWater Supply: Problems and Prospects,” paper prepared for the RegionalHousing and Urban Development Office, New Delhi, Environmental and HealthProject Activity Report No.26, Washington, D.C: USAID.

Consultative Group Meeting for Viet Nam. 1999. Attacking Poverty:

Vietnam Development Report, 2000. Joint Report of the Government of VietNam-Donor-NGO Poverty Working Group.

Curry, Stephen and John Weiss. 2000. Project Analysis in Developing

Countries. Macmillan, revised second edition.


Untitled-1 5/17/02, 10:26 AM152

Demombynes, Gabriel. January 2000. A Manual for the Poverty and

Inequality Mapper Module, Revised Version, World Bank Development ResearchGroup, Washington, D.C.

Dunlop, D. 1999. Economic Analysis of Proposed ADB/MOH Rural Health

Project. PPTA consultant report, Manila: ADB.

Fujimura, M. and J. Weiss. October 2000, Integration of Poverty Impact

in Project Economic Analysis. EDRC Methodology Series No.2. Manila: ADB.

Grosh, M. and P. Glewwe, eds. 2000. Designing Household Survey

Questionnaires for Developing Countries: Lessons from 15 Years of the Living

Standard Measurement Study. World Bank: Washington, D.C.

Hansen, S. May 2000. Framework in Measuring Poverty Impact of

Transport Projects. Consultant draft final report to EDRC. Manila: ADB.

Henninger, N. April 1998. “Mapping and Geographic Analysis of HumanWelfare and Poverty – Review and Assessment,” World Resources Institute,Washington, D.C.

International Labour Organisation. 2000. Child Labour Surveys (Online).Available: wwwwwwwwwwwwwww.ilo.or.ilo.or.ilo.or.ilo.or.ilo.org/public/englisg/standarg/public/englisg/standarg/public/englisg/standarg/public/englisg/standarg/public/englisg/standards/ipec/simpocds/ipec/simpocds/ipec/simpocds/ipec/simpocds/ipec/simpoc

Inter-American Development Bank. 2001. “Social Equity Enhancing(SEQ) and Poverty Targeted Investments (PTI).” (On-line).

Available: http://wwwhttp://wwwhttp://wwwhttp://wwwhttp://www.iadb.or.iadb.or.iadb.or.iadb.or.iadb.org/sds/POg/sds/POg/sds/POg/sds/POg/sds/POV/site_3199_e.htmV/site_3199_e.htmV/site_3199_e.htmV/site_3199_e.htmV/site_3199_e.htm

Jenkins, G. and M. B. El-Hifnawi. 1993. Economic Parameters for the

Appraisal of Investment Projects: Bangladesh, Indonesia and the Philippines.

Consultant’s report prepared for EDRC. Manila: ADB.

Kakwani, N. 2000, “Inequality and Wellbeing in the Philippines withfocus on Mindanao,” mimeo, EDRC. ADB: Manila.

Kakwani, N., B. Sisouphanhthong, P. Souksavath and B. Dark. 2001,“Poverty in Lao PDR,” paper presented at Asia and Pacific Forum on Poverty:

Reforming Policies and Institutions for Poverty Reduction, Manila, 5–9 February2000.

REFERENCES 153

Untitled-1 5/17/02, 10:26 AM153

Khandker, S. 1999. Fighting Poverty with Microcredit: Experience in

Bangladesh. University Press: Dhaka.Lanjouw, P. February 2001.“ Geography of Poverty Alleviation in Pakistan:

the Potential Contribution of a Poverty Map,” note prepared for the Governmentof Pakistan, World Bank Development Research Group, Washington, D.C.

Londero, E. 1995. “Reflections on Estimating Distributional Effects,” inC. Kirkpatrick and J. Weiss, eds., Cost-Benefit Analysis and Project Appraisal

in Developing Countries. Edward Elgar, Cheltenham.

Londero, E. 1999. “Poverty Targeting Classifications and DistributionalEffects,” paper presented at the Development and Project Planning Centre,University of Bradford, 6–7 May 1999.

Macro International Inc. 2000. Measure DHS+: Demographic and Health

Surveys (Online). Available: http://wwwhttp://wwwhttp://wwwhttp://wwwhttp://www.measur.measur.measur.measur.measuredhs.comedhs.comedhs.comedhs.comedhs.com

Minot, N. 2000. “Generating Disaggregated Poverty Maps: AnApplication to Vietnam,” World Development 28(2): 319–331.

Murray, C.J.L., and A.D. Lopez, eds. 1996. The Global Burden of Disease.

Boston, MA: Harvard School of Public Health.

Ozler, B. January 2000. “Poverty Maps: Construction and Use,”PowerPoint presentation, World Bank Development Research Group,Washington, D.C.

Pernia, E.M. ed. 1994. Urban Poverty in Asia: A Survey of Critical Issues.

Hong Kong, China: Oxford University Press for ADB.

Powers, T. 1989. “A Practical Application: The IDB Experience.” The

Impact of Development Projects on Poverty. Pares: OECD.

Quibria, M.G. ed. 1993. Rural Poverty in Asia Priority Issues and Policy

Options. Hong Kong, China. Oxford University Press for ADB.

———. ed. 1994. Rural Poverty in Developing Asia (Country studiesin two volumes). Hong Kong, China. Oxford University Press for ADB.


Untitled-1 5/17/02, 10:26 AM154

Singh, B., R. Ramasubban, R. Bhatla, J. Briscoe, C.C. Griffin, and C.Kim. 1993. “Rural Water Supply in Kerala, India: How to Emerge from a Low-Level Equilibrium Trap,” Water Resources Research 29(7): 1931–1942.

United Nations Children’s Fund (UNICEF). January 1995. Monitoring

Progress Toward the Goals of the World Summit for Children: A Practical

Handbook for Multiple-Indicator Surveys. Planning Office, Evaluation andResearch Office, UNICEF, New York.

US Census Bureau. 2000. Census Dates for Countries and Areas of Asia:

1945-2004 (Online). Available: wwwwwwwwwwwwwww.census.gov/ipc/www/cendates/.census.gov/ipc/www/cendates/.census.gov/ipc/www/cendates/.census.gov/ipc/www/cendates/.census.gov/ipc/www/cendates/cenasia.htmlcenasia.htmlcenasia.htmlcenasia.htmlcenasia.html

van de Walle, D. 1996. “Assessing the Welfare Impacts of Public Spending.Policy Research Working Paper No. 1670. The World Bank, Washington, D.C.

Weiss, J. 1996. “Project failure: The implications of a 25% rule,” inC.Kirkpatrick and J.Weiss, eds. Cost Benefit Analysis and Project Appraisal

in Developing Countries, Edward Elgar.

Whittington, D., D.T. Lauria, A.M. Wright, K. Choe, J.A. Hughes, and V.Swarna. 1993. “Household Demand for Improved Sanitation Services in Kumasi,Ghana: A Contingent Valuation Study,” Water Resources Research 29(6): 1539–1560.

World Bank. 1998. Handbook on Economic Analysis of Investment

Operations. Operational Core Services Network, Learning and LeadershipCenter, Washington, D.C.

______ .1999. Cambodia Poverty Assessment, Report 19858-KH. PovertyReduction and Economic Management Sector Unit and Human DevelopmentSector Unit, Washington, D.C.

______ . August 1999. Core Welfare Indicators Questionnaire Handbook,World Bank Africa Operational Quality and Knowledge Services, Washington,D.C.

______.2000. Core Welfare Indicators Questionnaire (Online). Available:http://www4.worldbank.org/afr/stats/cwiq.cfmhttp://www4.worldbank.org/afr/stats/cwiq.cfmhttp://www4.worldbank.org/afr/stats/cwiq.cfmhttp://www4.worldbank.org/afr/stats/cwiq.cfmhttp://www4.worldbank.org/afr/stats/cwiq.cfm

REFERENCES 155

Untitled-1 5/17/02, 10:26 AM155

———. 2000b. Country Reports on Health, Nutrition, Population, and

Poverty (Online).

———. 2000c. Living Standards Measurement Study of the World Bank

(Online). Available: http://wwwhttp://wwwhttp://wwwhttp://wwwhttp://www.worldbank.or.worldbank.or.worldbank.or.worldbank.or.worldbank.org/html/prg/html/prg/html/prg/html/prg/html/prdph/lsmsdph/lsmsdph/lsmsdph/lsmsdph/lsms

———. 2000d. Poverty Reduction Strategy Sourcebook: Technical Notesto the Public Spending Chapter of the World Bank Poverty Reduction StrategySource Book (Online). Available:http://wwwhttp://wwwhttp://wwwhttp://wwwhttp://www.worldbank.or.worldbank.or.worldbank.or.worldbank.or.worldbank.org/poverg/poverg/poverg/poverg/poverty/ty/ty/ty/ty/strategies/sourcons.htmstrategies/sourcons.htmstrategies/sourcons.htmstrategies/sourcons.htmstrategies/sourcons.htm

———. 2000e. Socio-Economic Surveys: World Bank, PovertyMonitoring Database (Online). Available: wwwwwwwwwwwwwww.worldbank.or.worldbank.or.worldbank.or.worldbank.or.worldbank.org/poverg/poverg/poverg/poverg/poverty/data/ty/data/ty/data/ty/data/ty/data/povmon.htmpovmon.htmpovmon.htmpovmon.htmpovmon.htm

———. 2000f. World Development Report 2000/2001: Attacking Poverty.

Oxford University Press for the World Bank, Washington, DC. Available: http:/http:/http:/http:/http://www/www/www/www/www.worldbank.or.worldbank.or.worldbank.or.worldbank.or.worldbank.org/poverg/poverg/poverg/poverg/pover ty/health/data/index.htmty/health/data/index.htmty/health/data/index.htmty/health/data/index.htmty/health/data/index.htm

———. 2001. “What is the Program on Targeted Interventions?”(Online). Available: http://wwwhttp://wwwhttp://wwwhttp://wwwhttp://www.worldbank.or.worldbank.or.worldbank.or.worldbank.or.worldbank.org/poverg/poverg/poverg/poverg/poverty/wbactivities/pti/ty/wbactivities/pti/ty/wbactivities/pti/ty/wbactivities/pti/ty/wbactivities/pti/index.htmindex.htmindex.htmindex.htmindex.htm


Untitled-1 5/17/02, 10:26 AM156

handbook for integrating poverty impact assessment · handbook for integrating poverty impact...

Documents