Making The New Indonesia

Making the New Indonesia
Work for the Poor
ii
iii
MAKING THE NEW INDONESIA WORK FOR THE POOR
THE WORLD BANK OFFICE JAKARTA

Jakarta Stock Exchange Building Tower II/12th floor
Jl. Jend. Sudirman Kav. 52-53
Jakarta 12910
Tel: (6221) 5299-3000
Fax: (6221) 5299-3111
Website: www.worldbank.or.id
THE WORLD BANK
1818 H Street N.W.
Washington, D.C. 20433, U.S.A.
Tel: (202) 458-1876
Fax: (202) 522-1557/1560
Email: feedback@worldbank.org
Website: www.worldbank.org
Printed in November 2006
Cover and book design: Gradasi Aksara
Cover photos: Copyright
Photographs by Jacqueline Koch Photography
Inside cover photo: Copyright
Jez O Hare Photography
All photographs are from: Copyright Jacqueline Koch Photography, except for photographs on pages 14, 105, 110, 116-7, 118, 121, 127, 145, 147, 152, 165,
167, 176, 176, 187, 198, 204, 220, and 280, they are from Jez O Hare Photography. Photographs in the Portraits from the Regions section are taken by Poriaman
Sitanggang. All rights reserved.
The Making the New Indonesia Work for the Poor report is a product of the staff of the World Bank. The findings, interpretations, and conclusions expressed herein
do not necessarily reflect the views of the Board of Executive Directors of the World Bank or the governments they represent.
The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any
map in this work do not imply any judgment on the part of the World Bank concerning the legal status of any territory or the endorsement or acceptance of such
boundaries.
Please direct comments on this report to Jehan Arulpragasam (jarulpragasam@worldbank.org) and Vivi Alatas (valatas@worldbank.org).
iv
Foreword
Since 2002, a team of leading Indonesian and international analysts under the umbrella of the Indonesia Poverty Analysis
Program (INDOPOV) at the World Bank Office, Jakarta, has been studying the characteristics of poverty in Indonesia. They
have sought to identify what works and what does not in the fight against poverty, and to help clarify options available to
the government and non-governmental organizations in their efforts to raise living standards and the quality of life for the
poor. This report brings together their findings.
This work has been greatly assisted by many Indonesian government institutions, particularly the Coordinating Ministry
for the Economy, the Coordinating Ministry for People s Welfare and Bappenas. Many of the leading academic institutions
and civil society organizations have also contributed to this report. The Economic Analysis Institute, University of Indonesia
(LPEM-UI), the Faculty of Economics, Padjadjaran University (Unpad) and SMERU Research Institute deserve particular
mention for their valuable contributions. The work was made possible by funding support from the United Kingdom s
Department for International Development (DFID), together with funding from the World Bank.
This report seeks to analyze the various dimensions of poverty in today s Indonesia in a fresh light, based on the important
changes that have occurred in the country over the past decade. Earlier Poverty Assessments were undertaken by the
World Bank in 1993 and 2001, but generally not in such depth. This Assessment presents a wealth of shared knowledge
from the World Bank and Indonesian government that we hope will contribute to a lively policy debate, and ultimately lead
to policy and practice changes that will accelerate poverty reduction efforts.
In summarizing its findings, the report presents a concrete policy matrix that suggests how Indonesia might better align
policies and programs to achieve its ambitious poverty reduction objectives. Our hope is that these findings will be useful
for Indonesia in operationalizing its five-year national development strategy (SNPK), and in planning actions to attain the
Millennium Development Goals, and the even more ambitious goals that it has set itself in its SNPK to 2009. These goals
include reducing poverty to 8.2 percent, achieving 98.1 percent junior school enrollments and cutting maternal mortality
rates from 307 deaths per 100,000 live births currently to 226 all by 2009.
Indonesia today is, of course, a very different place from the Indonesia of a decade ago. It should come as no surprise that
the strategies required to reduce poverty have changed just as Indonesia itself has changed. It is our sincere hope that this
poverty assessment will be of help in addressing these challenges. In so doing, we hope to contribute towards the
continuing remarkable transition that is taking place in this country.
Andrew D. Steer
Country Director, Indonesia
East Asia and Pacific Region
The World Bank
Acknowledgments
The Making the New Indonesia Work for the Poor report is an output of the Indonesia Poverty Analysis Program (INDOPOV)
led by Jehan Arulpragasam. INDOPOV is a multi-year poverty analysis program that has been supported by the World
Bank and the DFID Poverty Reduction Partnership Trust Fund.
This report has been prepared by a core team led by Jehan Arulpragasam and Vivi Alatas. The team that drafted the
chapters also included Meltem Aran, Kathy Macpherson, Neil McCulloch, Stefan Nachuk, Truman Packard, Janelle Plummer,
Menno Pradhan and Peter Timmer. Indermit Gill contributed to the drafting of the Overview section of this report.
Extensive excellent contributions were received from Maria Abreu, Tarcisio Castaneda, Leya Cattleya, Jennifer Donohoe,
Giovanna Dore, Luisa Fernandez, Jed Friedman, Djoko Hartono, Yoichiro Ishihara, Anne-Lise Klausen, Ray Pulungan,
Robert Searle, Shobha Shetty, Widya Sutiyo, Ellen Tan, Susan Wong, and the SMERU Research Institute.
Invaluable research and data analysis was provided, in particular, by (Hendro) Hendratno Tuhiman and Lina Marliani.
Additional analytical support was received from Javier Arze, Cut Dian Augustina, Zaki Fahmi, Ahya Ihsan, Melanie Juwono,
Bambang Suharnoko, Ellen Tan and Bastian Zaini.
This report benefited from the valuable comments of Peer Reviewers Bert Hofman (Lead Economist, EASPR), Jeni G.
Klugman (Lead Economist, AFTP2), Kathy A. Lindert (Country Sector Leader, LCSHD), Mohamad Ikhsan (Expert Staff,
Coordinating Ministry of Economy), and Bambang Bintoro Soedjito (Institut Teknologi Bandung (ITB) and former Deputy
for Regional Development and Local Autonomy at Bappenas).
We would like to thank everyone who generously and graciously provided comments throughout the development of the
report. In particular, we would like to thank Javier Arze, Timothy Brown, Stephen Burgess, Sally L. Burningham, Mae Chu
Chang, Soren Davidsen, Giovanna Dore, Wolfgang Fengler, Hongjoo J. Hahm, Pandu Harimukti, Joel Hellman, Peter
Heywood, Yoichiro Ishihara, Anne-Lise Klausen, Ioana Kruse, Josef L. Leitmann, Blane Lewis, Puti Marzoeki, Vicente
Paqueo, Andrew Ragatz, Claudia Rokx, Risyana Sukarma, Michael Warlters, Susan Wong and Elif Yavuz for their valuable
comments.
We would like to thank photographer Poriaman Sitanggang and narrator Rani M. Moerdiarta for their hard work and
commitment in developing the Portraits from the Regions section. Likewise, we would like to thank Scott Guggenheim
(DSF) and our colleagues at the Kecamatan Development Program (KDP) and Community Development Committees
(BKM) for their generous support in identifying families, including: John Odius and Iffah (Lampung), Kuseri and Dasimi
(Jakarta), Sainafur and J. Simbolon (West Kalimantan), Alman Hutabarat and Erna (East Nusa Tenggara), and Leo Koirewo
and Barbara Juliana Sopacua (Papua).
The editor of this report was Peter Milne who also provided important written contributions.
Invaluable project management for this report was led by Widya Sutiyo, with support from Peter Milne and Stefan Nachuk.
Special thanks to them for their dedication and hard work throughout the development of this report. Indispensable
logistical and production support was provided by Deviana Djalil, Niltha Mathias and Juliana Wilson.
vi
Acknowledgments
The development of the report benefited substantially from the Yogyakarta Workshop in January 2006, which was attended
by the core team, Andrew Steer, Wolfgang Fengler, Indermit Gill, Anne-Lise Klausen, Vicente Paqueo and Bill Wallace.
Special thanks to John Adams who facilitated this workshop and has been a team mentor throughout the INDOPOV
program.
This report was produced under the overall guidance of Indermit Gill, Sector Manager for Poverty Reduction (EASPR) and
Bill Wallace, Lead Economist for Indonesia (EASPR) for the World Bank. Strategic guidance and key comments were also
provided by Andrew Steer, Country Director, Indonesia.
This report also benefited from work undertaken under two other key INDOPOV products, Making Services Work for the
Poor in Indonesia and Revitalizing the Rural Economy: An assessment of the investment climate faced by non-farm
enterprises at the District level, and from contributions to those elements of the INDOPOV program.
vii
Glossary of Terms
Abbreviations and Acronyms
viii
ADB
Afta
Amdal
AAHRD
Arisan
Asean
Asabri
Askes
APBD
APBN
Asian Development Bank

Asean Free Trade Area
Environmental Impact Permit (Analisis Mengenai Dampak Lingkungan)
Agency for Agricultural Human Resource Development
Rotating credit scheme at community level
Association of South-East Asian Nations
Pension and health coverage for the Armed Forces and Police
Health insurance provided by PT Askes (Asuransi Kesehatan)
Regional Government Budget (Anggaran Pendapatan Belanja Daerah)
State Budget (Anggaran Pendapatan Belanja Nasional)
Bappeda
Bappenas
Bimas
BIGS
BKD
BKKBN
BKM
BKN
BLK
BOS
BPN
BPOM
BPR
BPS
BRI
Bulog
Development Planning Agency at Sub-National Level (Badan Perencanaan Pembangunan Daerah)

National Development Planning Agency (Badan Perencanaan dan Pembangunan Nasional)
Mass Guidance (or Bimbingan Massal) - government program that provided farmers access to high-yielding rice
varieties, fertilizers and pesticides
Bandung Institute for Governance Studies
Rural Development Bank (Badan Kredit Desa)
National Family Planning Agency (or Badan Koordinasi Keluarga Berencana Nasional)
Student Scholarship Program (Beasiswa Kerja Mahasiswa)
National Civil Service Agency (Badan Kepegawaian Negara)
Vocational training center (Balai Latihan Kerja)
Operational Aid for Schools Program (Bantuan Operasional Sekolah)
National Land Agency (Badan Pertanahan Nasional)
National Agency for Drug and Food Control (Badan Pemeriksa Obat dan Makanan)
People s Credit Bank (Bank Pinjaman Rakyat)
Central Bureau of Statistics (Badan Pusat Statistik)
Bank Rakyat Indonesia, partially state-owned nationwide bank
National Logistics Agency (Badan Urusan Logistik Nasional)
CAF
CBO
CCT
CDD
CEFE
CIFOR
CPI
CSO
Community Achievement Fund

Community-Based Organization
Conditional Cash Transfer
Community Driven Development
Creation of Enterprise through Formation of Entrepreneur
Center for International Forestry Research - Indonesia
Consumer Price Index
Civil Society Organization
DAK
DAU
DBH
DHS
Dinas
DFID
DFID-MFP
DPR
DPRD
Special Allocation Fund (Dana Alokasi Khusus)

General Allocation Fund (Dana Alokasi Umum)
Shared-Revenue Fund (Dana Bagi Hasil)
Demographic and Health Survey
Regional Sector Office
Department for International Development
Department for International Development - Multi-stakeholder Forestry Program
House of Representatives (Dewan Perwakilan Rakyat)
Provincial House of Representatives (Dewan Perwakilan Rakyat Daerah)
FAO
FDI
FE Unpad
FGD
FP Scheme
GDP
Food and Agriculture Organization

Foreign Direct Investment
Faculty of Economics, University of Padjajaran
Focus Group Discussion
Foster-Parent Scheme
Gross Domestic Product
Glossary of terms
GDS
GIAT Project
GIC
GNP
GRDP
Governance and Decentralization Survey

Growth through Investment in Agriculture and Trade Project
Growth Incidence Curve
Gross National Product
Gross Regional Domestic Product
HDI
HH
Human Development Index

Household
IAARD
ICT
ICRAF
IDT
IFAD
IFLS
ILO
IMR
Inpres
IRDA
IRI
Indonesian Agency for Agricultural Research and Development

Information and Communication Technologies
International Center for Research in Agro-forestry - Kenya
Village Infrastructure Development Program (Inpres Desa Tertinggal)
International Fund for Agricultural Development
Indonesian Family Life Survey
International Labor Organization
Infant Mortality Rate
Presidential Instruction (Instruksi Presiden)
Indonesia Rapid Decentralization Appraisal
International Roughness Index
Jamsostek
JKJ
JPS
JPKM
JPS-BK
JSS
A provident fund
Jembrana Health Insurance (Jaminan Kesehatan Jembrana)
Social Safety Net (Jaringan Pengaman Sosial)
Health coverage provided by local goverment
Social Safety Net on Health (Jaring Pengaman Sosial Bidang Kesehatan)
Junior Secondary School
Kcal
KDP
Kepmen
Keppres
KHM
KKB
KLK
Kopinkra
KPK-D
KUD
KUK
Kupedes
Kilo calories
Kecamatan Development Program
Ministerial Decree (Keputusan Mentri)
Presidential Decision (Keputusan Presiden)
Minimum Subsistence Need (Kebutuhan Hidup Minimum)
Fuel Compensation Cards (Kartu Kompensasi Bahan Bakar Minyak-BBM)
Vocational training courses (Kursus Latihan Kerja)
Cooperatives for Small Industries (Koperasi Industri Kerajinan Rakyat)
Regional Poverty Reduction Committee (Komite Penganggulangan Kemiskinan Daerah)
Village Cooperative (Koperasi Unit Desa)
Small-Scale Credit Unit (Kredit Usaha Kecil)
General Rural Credit (Kredit Umum Pedesaan)
LAN
LAP
LE
LMPDP
LOC
LIK
LP3E
National Institute of Administration (Lembaga Administrasi Negara)

Land Administration Project
Large Enterprise
Land Management and Policy Development Project
Land Office Computerization
Small-Scale Industrial Estates (Lingkungan Industri Kecil)
Research institute, Faculty of Economy, University of Padjajaran (Laboratorium Penelitian Pengabdian pada Masyarakat dan
Pengkajian Economi)
MA
Islamic Senior Secondary School (Madrasah Aliyah)
MASS
Microfinance Access and Services Survey
MDGs
Millennium Development Goals
ME
Medium Enterprise
Menpan
State Ministry for State Apparatus Reform (Menteri Negara Pendayagunaan Aparatur Negara)
MI
Islamic Primary School (Madrasah Ibtidaiyah)
MIC
Middle-Income Country
MIE
Micro Enterprises
MMR
Maternal Mortality Rate
MNC
Multi-National Company
MPR
People s Consultative Assembly (Majelis Permusyawarahan Rakyat)
MSS
Minimum Service Standard
MT
Islamic Junior Secondary School (Madrasah Tsanawiyah)
MusrenbangNas Central Government Development Planning Forum
ix
NFE
NGO
NSS
NTB
NTT
OBA
OECD
Opec
OPK
Non-Farm Enterprise
Non-Governmental Organization
Nutrition and Surveillance Survey
West Nusa Tenggara (Nusa Tenggara Barat)
East Nusa Tenggara (Nusa Tenggara Timur)
Output-Based Aid
Organization for Economic Cooperation and Development
Organization of the Petroleum Exporting Countries
Special Market Operation (Operasi Pasar Khusus)
P4K
PAD
PDAM
PDM-DKE
PEPI
PETS
PISA
PKBM
PKK
PKPS-BBM
PKWT
PLN
PMDF
Podes
PoS Cards
Posyandu
PPA
PPP
PRA
Progresa
PSDA
PSIA
PMT
PU
Puskesmas
Pustu
Income Generating Project for Marginal Farmers and Fishermen (Proyek Peningkatan Pendapatan Petani-Nelayan Kecil)
Own-Source Revenue (Pendapatan Asli Daerah)
Municipal Water Companies (Perusahaan Daerah Air Minum)
Regional Empowerment to Overcome the Impact of Economic Crisis (Program Pemberdayaan Daerah Akibat Dampak Krisis
Ekonomi)
National Team to Promote Exports and Investment (Tim Nasional Peningkatan Ekspor dan Peningkatan Investasi)
Public Expenditure Tracking Survey
Program for International Student Assessment
Community Learning Center (Pusat Kegiatan Belajar Masyarakat)
Family Welfare Education (Pendidikan Kesejahteraan Keluarga)
Fuel Subsidy Compensation Program (Program Kompensasi Pengurangan Subsidi BBM)
Temporary Working Agreement (Perjanjian Kerja Waktu Tertentu)
State Electricity Company (Perusahaan Listrik Nasional)
Proportion of Maternal Deaths of Females
Village Potential Statistics (Potensi Desa)
Point of Sale Cards
Integrated Health Services Unit (Pos Pelayanan Terpadu)
Participatory Poverty Assessment
Purchasing Power Parity
Participatory Rural Appraisal
Cash transfer program in Mexico
Water resource management (Pengelola Sumber Daya Air)
Poverty and Social Impact Analysis
Proxy Means Test
Public Works (Pekerjaan Umum)
Community Health Center (Pusat Kesehatan Masyarakat)
Community Health Sub-Center (Puskesmas Pembantu)
RALAS
Raskin
Renja-KL
Renstra
Renstra-KL
Renstra-SKPD
RKA-KL
RKA-SKPD
RKP
RKP-D
RICA
RICS
RPJM
RPJM-D
RPO
RPPG
RPPK
RVA
Reconstruction of Aceh Land Administration Project

Rice for the Poor Program (Beras Miskin)
Ministry Work Plan (Rencana Kerja Kementerian/Lembaga)
Strategic Plan (Rencana Strategis)
Ministry Strategic Plan (Rencana Strategis Kementerian/Lembaga)
Dinas Work Plan Budget (Rencana Strategis Satuan Kerja Perangkat Daerah)
Ministry Work Plan and Budget
Dinas Work Plan Budget
Government Annual Work Plan (Rencana Kerja Pemerintah)
Regional Annual Work Plan (Rencana Kerja Pemerintah - Daerah)
Rural Investment Climate Assessment
Rural Investment Climate Survey
Medium-Term Development Plan (Rencana Pembangunan Jangka Menengah)
Regional Medium-Term Development Plan (Rencana Pembangunan Jangka Menengah Daerah)
Rural Producer Organization
Rate of Pro-Poor Growth
Revitalisasi Pertanian, Perikanan dan Kehutanan (Revitalizing Agriculture, Forestry and Fisheries)
Risk and Vulnerability Assessment
SD
SDI
SE
SIKD
SGP
SLTP
Primary School (Sekolah Dasar)

Surface Distress Index
Small Enterprise
Reginonal Finance Information System (Sistem Informasi Keuangan Daerah)
Scholarship Grant Program
Public Junior Secondary School (Sekolah Lanjutan Tingkat Pertama)
Glossary of terms
SMA
SME
SMK
SMP
SMU
SNPK
SPK-D
SSN
Sakernas
Supas
Susenas
Susi
Senior High School (Sekolah Menengah Atas)

Small to Medium Enterprise
Vocational Schools (Sekolah Menengah Kejuruan)
Junior High School (Sekolah Menengah Pertama)
Senior Secondary School (Sekolah Menengah Umum)
National Strategy for Poverty Reduction (Strategi Nasional Penanggulangan Kemiskinan)
Regional Poverty Reduction Strategy (Strategi Penanggulangan Kemiskinan Daerah)
Social Safety Net
Labor survey (Survei Angkatan Kerja Nasional)
Inter-censal survey, or Survei Penduduk antar Sensus
National Socio-Economic Survey (Survei Sosial Ekonomi Nasional)
Integrated Business Survey of non-registered Micro and Small Enterprises (Survei Usaha Terintegrasi)
Taspen
TBA
TFR
TIMSS
TPC
Pension fund for civil servants

Traditional Birth Attendant
Total Fertility Rate
Trends in Mathematics and Science Study
Targeted Performance Contracting Program
UCT
UNICEF
UPP
UPT
UUPK
Unconditional Cash Transfer

United Nations Children s Fund
Urban Poverty Program
Technical Service Unit (Unit Pelaksana Teknis)
Basic Forestry Law (Undang-Undang Pokok Kehutanan)
Warsi
WDI
WDR
WEF
WHFWP
WHO
WUAs
WSLIC-2
WSS
Conservation Information Kiosk (Warung Informasi Konservasi)

World Development Indicators
World Development Report
World Economic Forum
Women s Health and Family Welfare Project
World Health Organization
Water Users Associations
Second Water and sanitation for Low-Income Communities Project
Water Supply and Sanitation
YPC
Sandalwood Education Foundation (Yayasan Pendidikan Cendana)
Editorial and data note

Indonesian words are italicized (bupati), but where they are acronyms they are capitalized but not italicized (Wartel).
Likewise, where Indonesian words are considered to be used in the sense of proper nouns they are capitalized but
not italicized (Dewan Perwakilan Rakyat, or DPR). Abbreviations and acronyms that are pronounced as words
(not separate letters) tend to have the first letter only capitalized to indicate to the reader they are pronounced as
words (Jamsostek).
Dollar figures are current US dollars, unless otherwise specified. Billion (bn) means 1,000 million (m); trillion
(trn) means 1,000 billion.
xi
Contents
Foreword
Acknowledgments
vi
Glossary of Terms
viii
Overview
xxi
Chapter 1 Introduction
I
Introduction
Chapter 2 A History of Growth and Poverty Reduction

I
Introduction
II
A Troubled History and Chronic Poverty
III
A Period of Growth and Poverty Reduction
IV
A Structural Transformation
12
The Financial Crisis
16
VI
Post-Crisis: Stabilization, Democracy and Decentralization
20
VII
Conclusion: Learning from History to Support Poverty Reduction
22
Chapter 3 Understanding Poverty in Indonesia

I
Salient Facts about Poverty in Indonesia
Progress in poverty reduction: a positive story, with some caveats
The near-poor and vulnerability: significant aspects of the poverty story
Recent Developments
Why did poverty increase from 2005 to 2006?
Counting the poor in Indonesia
Non-income poverty: indicators show persisting problems
Multi-dimensional poverty: a multi-headed beast
Inequality: a positive story, with questions
Lagging regions
Regional diversity and the impact of growth on poverty
xii
25
25
26
29
30
31
34
34
36
38
II
A Profile of the Poor
40
III
The Determinants of Poverty

Correlate 1: Education
Correlate 2: Occupation
48
48
50
Contents
Correlate 3: Gender
Correlate 4: Access to basic services and infrastructure
Correlate 5: Geographic location
51
51
52
IV
The Determinants of Recent Poverty Changes: a Dynamic Analysis

Endowment effect: endowment changes the main contributor to poverty reduction
Price effect: the poverty impact of the return to assets and access was mixed
Occupational-choice effect: changes were poverty reducing but inequality increasing
Shocks and unobservable effects: changes increased poverty and inequality
54
54
56
57
58
Conclusion: Poverty Diagnostics Provide Pointers for Poverty Reduction Efforts
58
Focus on Sumatra
Focus on Java/Bali
Focus on Kalimantan
Focus on Sulawesi
Focus on Nusa Tenggara/Maluku
Focus on Papua
60
62
64
66
68
70
Portraits from the Regions
72
Chapter 4 Making Growth Work for the Poor

I
Introduction
87
II
What Are the Pathways out of Poverty?

What have been the most important pathways out of poverty?
Panel data on individual pathways out of poverty
89
90
92
III
Linking Pathways out of Poverty to Policies for Pro-Poor Growth
94
IV
Maintaining a Stable Macroeconomy
97
Investing in the Capabilities of the Poor

Boosting agricultural capability
Improving education and vocational training
99
99
104
VI
Connecting the Poor to Growth

Connecting the rural poor to urban markets
Connecting the poor to jobs
Connecting the poor to financial services
Approaches to improving access to financial services
109
109
119
119
123
VI
Conclusion: Making Growth more Pro-Poor
125
Spotlight on Gender: Remittances from female migrant workers: A lifeline to communities

Chapter 5 Making Public Spending Work for the Poor
I
Introduction
II
Aggregate Spending Levels and Fiscal Space

Aggregate spending
Sectoral spending
128
131
132
132
134
xiii
III
Human Development Sectors

Overview
Education sector
Levels of spending on education
Priority issues for poverty reduction in education
Recommendations for pro-poor spending on education
Health sector
Levels of spending on health
Priority issues for poverty reduction in health
Recommendations for pro-poor spending on health
Focus on Reducing Maternal Mortality

Focus on Child Malnutrition
154
157
IV
Infrastructure Sectors
Overview
Water and sanitation services sector
Levels of spending on water and sanitation services
Priority issues for poverty reduction in water and sanitation services
Recommendations for pro-poor spending on water and sanitation services
Rural roads sector
Levels of spending and access to rural roads
Priority issues for poverty reduction in rural roads
Recommendations for pro-poor spending on rural roads
158
158
159
159
160
161
164
164
165
166
Conclusion: Achieving Synergies through the Right Development Spending Mix
170
Spotlight on Inefficiencies and Leakages: Corruption in water service delivery: Direct and indirect impacts on the poor
Spotlight on Gender: Aligning village-level spending with the needs of the poor through the participation
and empowerment of women
Chapter 6 Making Social Protection Work for the Poor
I
Introduction
II
III
IV
xiv
136
136
138
138
140
146
147
147
148
153
172
174
177
The Evolution of Social Protection Institutions in Indonesia: from Universal Subsidies to

Targeted Transfer
The crisis-era safety net
The fuel subsidy: a pillar of Indonesia s social protection program in recent years
An opportunity for the future
181
181
183
184
Household Risk Management: Findings from a Risk and Vulnerability Assessment and Implications
for Social Protection Policy
Vulnerability profile
Nature shocks and coping mechanisms
Impact of shocks
Implications for developing a social protection system
187
187
188
190
193
Towards a National Social Protection System that Makes Sense for Indonesia
Conditional cash transfer program
195
197
Contents
Workfare
Health insurance
Rice price stabilization
Social insurance
Targeting
V
Conclusion: Towards a Third-Generation System
Spotlight on Innovation: Poverty maps: A powerful tool in targeting the poor

Spotlight on Inefficiencies and Leakages: Where has all the Raskin gone? Improving the
effectiveness of the subsidized rice distribution program
Chapter 7 Making Government Work for the Poor
I
Introduction
201
204
206
207
209
213
215
217
221
II
Policy, Planning and Budgeting Systems

Recommendations
224
233
III
Institutional Accountability
Clarity of functions between central and sub-national government units
Civil service functions and constraints
Citizen and civil society voice
Recommendations
237
238
241
247
250
IV
Monitoring and Assessment of Poverty and Poverty Reduction Interventions

Recommendations
255
257
Conclusion: Towards Government Services Focused on the Poor
259
Focus on Annual Plan and Budget Preparation for2006

Focus on Elements of Good DAK Design
Spotlight on Innovation: Placing poverty reduction at the center of district annual plans and budgets
Spotlight on Inefficiencies and Leakages: Intensive effort to curb corruption in village infrastructure
development- lessons from Indonesia
Chapter 8 Summary of Recommendations
I
Priorities for Poverty Reduction: An Agenda for Action
II
Poverty Reduction Policy Matrix
261
262
263
264
267
272
List of Boxes
Box 2.1
Indonesias elections and the consolidation of democracy
21
Box 3.1
Box 3.2
Box 3.3
Box 3.4
Box 3.5
Box 3.6
Poverty definitions and measures

Is inequality in Indonesia really so low?
The faces of urban poverty
Unemployment and poverty are not equivalent
Rice-price increases disproportionately hurt the poor
Tobacco damages poor households finances as well as their health
27
35
42
44
46
47
xv
Box 3.7
Box 4.1
There are differences in poverty correlates and constraints across regions

Rationalizing the use of forest lands can help the poor
53
104
Box 5.1
Box 5.2
Box 5.3
Box 5.4
Does the Operational Aid to Schools improve affordability for poor students? Issues and concerns
regarding the BOS program
Local ownership in Lumajang
Scaling up community-driven development through the KDP program
Keeping the C in community-driven development
144
162
168
169
Box 6.1
Box 6.2
Box 6.3
Box 6.4
Box 6.5
Box 6.6
Box 6.7
Comprehensive insurance frameworks and the role of government in social risk management
Indonesia s PKPS-BBM Unconditional Cash Transfer (UCT) program
The emergence of cash transfers as a an instrument of social policy in developing countries
What is a CCT and what is its purpose?
Can public employment programs act as social insurance for the poor and near-poor?
Improving poor people s access to healthcare
Poverty targeting database
180
186
197
199
202
205
212
Box 7.1
Box 7.2
Box 7.3
Box 7.4
Box 7.5
Box 7.6
Box 7.7
Box 7.8
Box 7.9
Box 7.10
Box 7.11
Box 7.12
The challenges of implementing good pro-poor planning and budgeting by sub-national governments
How is the DAU allocated across districts?
High levels of teacher absenteeism
Legislative blockage to local government action
The basics of better staffing
The challenges of implementing local civil service reforms
Informal patronage behavior in the Indonesian civil service
How public-private competition can improve healthcare for poor clients
Developing a high-quality, locality-initiated block grant program in Blitar city
When vouchers work for the poor and when they don t
Increasing the benefits from the private sector
How can incentives change the behavior of frontline service providers?
227
231
237
241
243
244
245
247
249
250
251
253
List of Figures
xvi
Figure 2.1
Figure 2.2
Figure 2.3
Figure 2.4
Periods of sustained growth have brought rapid poverty reduction in Indonesia, 1961-2005
Indonesia experienced a rapid structural transformation in recent decades
Corruption in Indonesia remained a potential source of weakness in the period 1980-98
Rice prices increased substantially during the crisis
7
13
16
18
Figure 3.1
Figure 3.2
Figure 3.3
Figure 3.4
Figure 3.5
Figure 3.6
Figure 3.6b
Figure 3.7
Figure 3.8
Growth has become less pro-poor over time

Almost half of all Indonesians live on less than US$2-a-day
Trends in educational attainment in Indonesia
Under-five mortality since 1960: the fastest decline in the region
But today Indonesia has a high and stagnating child malnutrition rate
Large variations exist in poverty headcount across Indonesia
...and pockets of poverty exist even within rich provinces
Expansion in access to safe water source and improved sanitation, by income group (1994-2004)
Growth is less pro-poor in eastern Indonesia
26
28
32
33
33
36
37
38
39
Contents
Figure 3.9 Households with very small planned areas are much poorer
Figure 3.10 Changes in selected endowments for the poor, 1999-2002
Figure 3.11 Movement towards the informal sector was greater among the poor
41
55
58
Figure 4.1
Figure 4.2
Figure 4.3
87
89
Figure 4.4
Figure 4.5
Figure 4.6
Figure 4.7
Figure 4.8
Figure 4.9
Figure 4.10
Figure 4.11
Figure 4.12
Figure 4.13
Poverty headcount projections depend on the pro-poorness of growth

The pathways out of poverty
Movements along the pathways out of poverty, 1993-2002, when villages can be reclassified as urban
in 2002 from their 1993 status as rural ( published data)
Movements along the pathways out of poverty, 1993-2002, when villages remain in their 1993
classification as urban or rural ( corrected data)
The welfare benefits associated with investing in the poor and connecting the poor
Most important constraints faced by firms in RIC survey
Low land registration in Indonesia
The gap between household consumption associated with high educational endowments has widened between
urban and rural areas for higher levels of schooling (1999-2002)
In rural areas, lack of job availability is a concern even for graduates of senior secondary schools
The ratio of wages for SMK vs. regular senior secondary school graduates have recovered in recent years
Vocational school enrollments are regressive: the poorest 20 percent of population capture
only about 12 percent of benefits from public SMKs
Better infrastructure is associated with higher levels on non-farm enterprises activities in Indonesia
Real minimum wages in Indonesia have been on the rise in recent years
Figure 5.1
Figure 5.2
Figure 5.3
Figure 5.3a
Figure 5.4a
Figure 5.4b
Figure 5.5
Figure 5.6
Figure 5.7
Figure 5.8
Figure 5.9
Figure 5.10
Figure 5.11
Figure 5.12
Figure 5.13
Figure 5.14
Figure 5.15
90
91
95
96
103
106
106
107
107
110
115
Consolidated spending as a percentage of GDP over time

Fuel subsidies ran out of control with increasing oil prices and stagnating refinery capacity
Fuel subsidies and spending in pro-poor development sectors
Fiscal space for development programs and fuel subsidy reductions
Aggregate public education expenditure has increased in decentralized Indonesia
Comparison of public funds available for education
Sources of basic education spending from public and private resources (2003)
Expanded access to secondary school level has made education public spending more pro-poor over time
Learning levels among Indonesian 15-year-olds are low, even among the relatively better off
Non-fee costs of schooling are higher than the cost of school fees for the poor
Aggregate public health expenditure has increased in decentralized Indonesia
Indonesias total health spending from public and private sources still remains low
Sources of healthcare spending in Indonesia (2003)
Over time benefit incidence of public health spending
Health utilization of health providers by income groups (2004)
Development spending on infrastructure
Indonesia is already lagging behind China in the commercial perception of quality in infrastructure services
(ranking: 1 worst, 7 best)
Figure 5.16 Development spending of infrastructure by sub-sector
133
133
133
133
139
139
139
139
141
143
148
148
148
148
150
159
Figure 6.1
Figure 6.2
Figure 6.3
185
206
Figure 6.4
Indonesias universal fuel subsidy: a dominant social policy with a highly regressive impact
Real rice prices were stable before the crisis
The reach of traditional social insurance: participants in the labor force by per capita income,
selected countries circa 1995
The targeting performance of various targeted poverty programs in Indonesia are only slightly pro-poor
159
163
208
209
xvii
Figure 7.1
Figure 7.2
Figure 7.3
Figure 7.4
Figure 7.5
Figure 7.6
Figure 7.7
Figure 7.8
Governance indicators for Indonesia

Three key areas of government action
Less pro-poor distribution of education expenditures after decentralization due to increased inequities
in decentralize spending (2000-03)
Shared natural-resource revenues are distributed to a few provinces only
The education DAK does not correlate with problem areas in terms of SMP enrollment rates
...while the health DAK is better targeted to regions where infant mortality rate is high (2005)
Puskesmas receive most of their revenues in the form of in-kind transfers or earmarked funding
Use of public and private service providers (by income quintile)
224
224
229
230
232
232
238
246
List of Tables
xviii
Table 2.1
Table 2.2
Table 2.3
Table 2.4
Table 2.5
Table 2.6
Long-term patterns of pro-poor growth in Indonesia

Poverty elasticity of growth has varied over time
Employment has moved to non-agricultural sectors and urban areas over time
Indonesia had the longest serving leader of any country impacted by the Asian financial crisis
The crisis caused construction, finance and trade sectors to contract in relative size
Real wages were eroded by inflation during the crisis keeping unemployment surprisingly stable
6
9
12
17
18
19
Table 3.1
Table 3.2
Table 3.3
Table 3.4
Table 3.5
Table 3.6
Poverty has declined to pre-crisis levels in most regions

Multi-dimensional poverty
Rice and tobacco are large spending items in the budgets of the poor
Household expenditure function by urban/rural area, 1999-2002
Decomposition of changes in poverty incidence by endowments, 1999-2002
Post-crisis movement towards informal and agriculture sectors
25
34
44
49
56
57
Table 4.1
Table 4.2
Table 4.3
Table 4.4
Table 4.5
Table 4.6
Table 4.7
Table 4.8
Table 5.1
Table 5.2
True rural-urban migration, 1990-1995

Poverty transition matrix, 1993-2000, from IFLS panel data
Linking policies to pathways out of poverty
Access to infrastructure by region
Condition and surface type of district-level roads, 2003
Maintaining and upgrading the district road network: estimated costs
Financial obstacles facing rural non-farm enterprises
Reasons for not applying to a formal financial institution (firms that need additional funds)
MDGs and national medium-term development targets (RPJMs) for Indonesia
Comparison of health indicators across countries
92
94
96
111
112
113
122
123
131
138
Table 6.1
Table 6.2
Table 6.3
Table 6.4
Table 6.5
Table 6.6
Table 6.7
Movements in and out of poverty, 2003-04

Typology of instruments for household risk management
Indonesia s crisis-era social safety net
Shocks that had a negative impact to family welfare , reported incidence among households, 2004
Households that reported solutions to the shocks, reported recourse among household s reporting shocks, 2004
Risk mitigation and coping instruments captured in Susenas
International comparison of targeting performance for cash transfer programs
177
179
182
191
191
192
212
Contents
Table 7.1
Table 7.2
Table 7.3
Table 7.4
Table 7.5
Table 7.6
What a difference a decade makes

Translating poverty reduction priorities and objectives into results on the ground
Before and after: government efforts to rectify deficiencies in planning and budgeting processes
Suggested functional allocations for service delivery
Locally initiated governance reforms implemented in West Sumatra province
Economic benefits from community-managed basic infrastructure
222
225
226
240
244
248
List of Annexes
Annex II.1
Annex II.2
Annex II.3
Annex II.4
Annex II.5
Annex II.6
Poverty lines, percentage of poor and total numbers of poor in Indonesia

Framework for pro-poor growth
Empirics of structural transformation in Indonesia
Mean share of per capita income from each source (national, rural, urban)
Increase in non-agricultural sector employment was higher than predicted by average employment growth (1982-2002)
Rural and informal economy absorbed workers during the crisis
281
282
283
284
285
286
Annex III.1
Annex III.2
Annex III.3
Annex III.4
Where do the poor who lack access to basic infrastructure services live?
Poverty growth elasticities (supporting table for Figure 3.8)
Differences in poverty correlates and constraints across regions
Microeconomic simulation of changes in poverty
287
288
289
290
Annex V.1
Annex V.2
Annex V.3
Annex V.4
Annex V.5
Inpres development grants

Education spending in Indonesia 1994-2004
Health spending in Indonesia 1994-2004
Infrastructure development expenditures by sector and level of government
Only 18 percent of the urban poor have access to tap water
291
292
293
294
295
Annex VI.1 Parameters and performance of Indonesia s principal targeted social programs
Annex VI.2 Risk and vulnerability analysis table
Appendix Table 6.1 Vulnerability profiles: Life cycle distribution of risks to earnings ability and household income in Indonesia
Appendix Table 6.2 Risk mitigation and coping instruments captured in Susenas
Appendix Table 6.3 Model 1: Impact of reported shocks on household expenditure on food
Appendix Table 6.3 Model 2: Impact of reported shocks on household expenditure on food-including household
and labor market control variables
Appendix Table 6.3 Model 3: Impact of reported shocks on household expenditure on food and the effectiveness
mitigation and coping
Annex VI.3 Indonesia: Likely CCT program impacts on the MDGs in program areas by 2015
Annex VI.4 Budgets for the PKPS-BBM programs for 2006
296
298
299
301
302
References
312
303
304
310
311
xix
Overview
xx
Making the New Indonesia Work for the Poor
Indonesia stands at the threshold of a new era. After the historic economic, political and social upheavals at the
end of the 1990s, Indonesia has started to regain its footing. The country has largely recovered from the economic crisis
that threw millions of its citizens back into poverty in 1998 and saw Indonesia regress to low-income status. Recently, it
has once again become one of the world s emergent middle-income countries. Poverty levels that had increased by over
one-third during the crisis are now back to pre-crisis levels (Figure 1). Meanwhile, Indonesia has undergone some major
social and political transformations, emerging as a vibrant democracy with decentralized government and far greater
social openness and debate.
Figure 1
Poverty in Indonesia fell rapidly until the 1990s, and has declined again since the crisis
45
40.1
40
33.3
35
30
Poverty Headcount (%)
Crisis
28.6
26.9
23.4
25
21.6
18.2
17.6*
20
17.4
17.4
16.7 16.0
15.1
17.8
13.7
15
11.34*
10
5
0
* Revised Method
1976
1978
1980
1981
1984
1987
1990
1993
1996*
1999
2002
2003
2004
2005
2006
The challenge of reducing poverty remains one of the country s most pressing issues. The number of
people living below US$2-a-day in Indonesia comes close to equaling all those living on or below US$2-a-day in all of
the rest of East Asia besides China. The Indonesian government is committed to the objective of reducing poverty in its
medium-term plan (RPJM) for 2005-09 which, in turn, draws on a national poverty reduction strategy (SNPK). In addition
to signing on to the Millennium Development Goals (MDGs) for 2015, in its medium-term plan the government has laid
out its own key poverty reduction objectives for 2009. This includes an ambitious target of reducing the poverty headcount
rate from 18.2 percent in 2002 to 8.2 percent by 2009. While national poverty rates may be close to pre-crisis levels, this
still means that about 40 million people are living below the national poverty line. Moreover, although Indonesia is now a
middle-income country, the share of those living on less than US$2-a-day is similar to that of the region s low-income
countries such as Vietnam.
xxi
Indonesia has a golden opportunity for rapidly reducing poverty. First, given the nature of poverty in Indonesia,
focusing attention on a few priority areas could deliver some quick wins in the fight against poverty and low human
development outcomes. Second, as an oil and gas producing country, Indonesia stands to benefit in the next few years
from increased fiscal resources as much as US$10 billion in 2006 thanks to higher oil prices and reductions in fuel
subsidies. Third, Indonesia can harness still further benefits from its ongoing processes of democratization and
decentralization.
The challenge is how to make the new Indonesia work for the poor . This is the focus of this report, which
aims to contribute to the policy debate and decision-making process in the country by putting forth: (i) new and more
comprehensive analysis of empirical poverty diagnostics; and (ii) suggestions on concrete policies and programs for a
strategic action-plan to achieve Indonesias stated poverty-reduction objectives.
Indonesia can learn from its own economic growth, government policies and social programs. Indonesia
has had remarkable success in reducing poverty since the 1970s. The period from the late 1970s to the mid-1990s is
considered one of the most pro-poor growth episodes in the economic history of any country, with poverty declining by half.
After the spike during the economic crisis, poverty has generally returned to its pre-crisis levels. The poverty rate fell back to
about 16 percent in 2005 following a peak of over 23 percent in 1999 in the immediate wake of the economic crisis.
Macroeconomic stabilization from mid-2001 onwards underpinned this recovery, bringing down the price of goods, such as
rice, that are important to the consumption of the poor. However, despite this steady progress in reducing poverty, more
recently, there has been an unforeseen upturn in the poverty rate. This reversal appears to have been caused primarily by a
sharp increase in the price of rice an estimated 33 percent for rice consumed by the poor between February 2005 and March
2006, which largely accounted for the increase in the poverty headcount rate to 17.75 percent.
Box 1 Why did poverty increase from 2005 to 2006?
In September 2006, BPS announced that the poverty rate in Indonesia had increased during the period February 2005 to
March 2006 from 16.0 percent to 17.75 percent in contrast to steady declines in the poverty rate since the crisis.
The 33 percent increase in rice prices between February 2005 and March 2006 mostly due to the ban on rice imports is the
main reason that poverty rates have increased. Around three-quarters of the additional four million people falling into poverty
during this period did so as a result of the rice price increase and, in addition, recent analysis indicates that the fuel price
increase was not a major factor in the increased poverty rate. The unconditional cash transfer (UCT) program, which provided
cash transfers to 19.2 million poor and near-poor households, more than offset, on average, the negative impact of the fuel
price increase for the poor. In other words, the impact of the combined effects of the fuel price increase and the UCT compensation
point to a net positive income gain, overall, for the poorest 20 percent of the population.
However, with rice prices still going up, and the UCT program drawing to an end, there is a possibility that poverty rates could
rise again next year unless economic growth increases significantly.
Dimensions of Poverty in Indonesia and a Proposed Policy Framework

Poverty in Indonesia has three salient features. First, many households are clustered around the national income
poverty line of about PPP US$1.55-a-day, making even many of the non-poor vulnerable to poverty. Second, the income
poverty measure does not capture the true extent of poverty in Indonesia; many who may not be income poor could be
classified as poor on the basis of their lack of access to basic services and poor human development outcomes. Third,
given the vast size of and varying conditions in the Indonesian archipelago, regional disparities are a fundamental feature
of poverty in the country.
xxii
Overview
A large number of Indonesians are vulnerable to poverty. The national poverty rate masks the large number
of people who live just above the national poverty line. Close to 42 percent of all Indonesians live between the US$1and US$2-a-day poverty lines a remarkable and defining aspect of poverty in Indonesia (see Figure 2). Analysis
indicates that there is little that distinguishes the poor from the near-poor, suggesting that poverty reduction strategies
should focus on improving the welfare of the lowest two quintile groups. This also means that the vulnerability to
falling into poverty is particularly high in Indonesia: while only 16.7 percent of Indonesians surveyed were poor in
2004, more than 59 percent had been poor at some time during the year preceding the survey. Recent data also
indicate a high degree of movement in and out of poverty over time: over 38 percent of poor house holds in 2004
were not poor in 2003.
Figure 2
Forty-two percent of Indonesia s population lives on between US$1- and US$2-a-day
Source: Susenas Panel data, 2006.
Non-income poverty is a more serious problem than income poverty. When one acknowledges all
dimensions of human well-being adequate consumption, reduced vulnerability, education, health and access to
basic infrastructure then almost half of all Indonesians would be considered to have experienced at least one type of
poverty. Nonetheless, Indonesia has made good progress in past years on some human capital outcomes. There
have been notable improvements in educational attainment at the primary school level; basic healthcare coverage
(particularly in birth attendance and immunization); and dramatic reductions in child mortality. But in some MDGrelated indicators Indonesia has failed to make significant progress and lags behind other countries in the region.
Indeed, specific areas that warrant concern are:
Malnutrition rates are high and have even risen in recent years: a quarter of children below the age of five are
malnourished in Indonesia, with malnutrition rates stagnating in recent years despite reductions in poverty.
Maternal health is much worse than comparable countries in the region: Indonesia s maternal mortality rate
(307 deaths in 100,000 births) is three times that of Vietnam and six times that of China and Malaysia; only
about 72 percent of births are accompanied by skilled birth attendants.
xxiii
Education outcomes are weak. Transition rates from primary to secondary school are low, particularly among
the poor: among 16- to 18-year-olds from the poorest quintile, only 55 percent completed junior secondary
school, compared with 89 percent from the richest quintile from the same cohort.
Access to safe water is low, especially among the poor. For the lowest quintile access to safe water in rural areas
is only 48 percent, against 78 percent in urban areas.
Access to sanitation is a crucial problem. Eighty percent of the rural poor and 59 percent of the urban poor do
not have access to septic tanks, while less than 1 percent of all Indonesians have access to piped sewerage
services.
Regional disparities in poverty are considerable. Wide regional differences characterize Indonesia, some of which
are reflected in disparities between rural and urban areas. Rural households account for about 57 percent of the poor
in Indonesia and also frequently lack access to basic infrastructure services: only about 50 percent of the rural poor
have access to an improved source of water, compared with 80 percent for the urban poor. But importantly, across the
vast Indonesian archipelago, it is also reflected in broad swathes of regional poverty, in addition to smaller pockets
of poverty within regions. For example, the poverty rate is 15.7 percent in Java/Bali and 38.7 percent in more remote
Papua. Services are also unequally distributed across regions, with an undersupply of facilities in remote areas. In
Java the average distance of a household to the nearest public health clinic is 4 kilometers, whereas in Papua it is 32
kilometers. While 66 percent of the poorest quintile in Java/Bali have access to improved water, this number is 35
percent for Kalimantan and only 9 percent for Papua. A challenge faced by the government is that although poverty
incidence is far higher in eastern Indonesia and in more remote areas, most of Indonesia s poor live in the densely
populated western regions of the archipelago. For example, while the poverty incidence in Java/Bali is relatively low,
the island is home to 57 percent of Indonesia s total poor, compared with Papua, which only has 3 percent of the
poor.
Figure 3
Poverty incidence and mass vary greatly across the Indonesian archipelago
Percentage of Poor People by Province in Indonesia, 2004
Percentage of Poor People
xxiv
Overview
Number of Poor People by Province in Indonesia, 2004
Number of Poor People (000)
An analysis of poverty and its determinants in Indonesia, as well of Indonesia s history in reducing
poverty to date, points to three ways to fight poverty. The three means for helping people lift themselves out of
poverty are economic growth, social services, and public expenditures. Each of these prongs addresses one or more of
the three defining features of poverty in Indonesia: vulnerability, multidimensionality, and regional disparities (see Table
1 below). In other words, an effective poverty reduction strategy for Indonesia has three components:
Making Economic Growth Work for the Poor. Economic growth has been, and will continue to be, fundamental
to reducing poverty. First, making growth work for the poor is simultaneously key to linking the poor throughout
various parts of the Indonesian archipelago to the growth process whether it is across the rural-urban space or
across the various regional and island groupings. So it is fundamental to addressing the issue of regional disparities.
Second, to address the vulnerability characteristic of poverty associated with the dense concentration of income
distribution in Indonesia, anything that can shift this distribution to the right will rapidly reduce the incidence of and
vulnerability to income poverty.
Making Social Services Work for the Poor. The delivery of social services to the poor whether by the public
or private sector is essential to addressing poverty in Indonesia. First, this is key in addressing the non-income
dimensions of poverty in Indonesia. Lagging human development indicators such as the high maternal mortality rate
must be tackled by improving the quality of services that are made available to the poor. This goes beyond levels of
public spending: it is about improving systems of accountability, service delivery mechanisms, and even government
processes. Second, the nature of regional disparities transcends income disparity and is largely reflected in disparities
in access to services which, in turn, results in disparities in human development outcomes across regions. Thus
making services work for the poor is key to addressing the problem of regional disparities in poverty.
Making Public Expenditure Work for the Poor. Besides economic growth and social services, the government,
by targeting public spending to the poor, can assist them in countering income and non-income poverty. First, public
spending can be used for helping those who are vulnerable to income poverty through a modern system of social
protection that augments their own efforts to deal with economic uncertainty. Second, public spending can be used
xxv
to improve human development outcomes-hence, tackling the non-income multidimensional aspects of poverty.
Making expenditures work for the poor is particularly pertinent now given the increased fiscal space that exists in
today s Indonesia.
Table 1
An approach for addressing Indonesia s poverty problems
Dimensions of Indonesian poverty

Vulnerability
Multidimensionality
Regional disparities
Economic growth
Social services
Public expenditure
Note: Indicates principal link between thematic area and the aspect of poverty; indicates an important linkage.
This Overview highlights the top priority actions to address each of the three salient features of Indonesia s
poverty problem. It is not intended to be a comprehensive summary of this entire report, which provides a broad range
of diagnostics and related policy implications. Likewise, neither does this Overview present the comprehensive set of
policy prescriptions following from this report. Rather, this Overview section highlights the top priority actions under each
of the three thematic prongs laid out above, in line with Table 1 above. It highlights policy priorities under Growth,
Services and Expenditure where it is deemed that action under any of these areas would be most effective in addressing
one of the three salient features of poverty in Indonesia.
Making Economic Growth Work for the Poor

For Indonesia, growth that benefits the poor has been, and will continue to be, the main route to
poverty reduction. From the 1970s through to the late 1990s, growth was rapid and it reached the poor: each percentage
point increase in average expenditure resulted in a 0.3 percent reduction in the poverty headcount. Even since the crisis,
growth has been the primary determinant of poverty reduction. However growth will need to accelerate and benefit the
poor more if the government is to meet its poverty reduction targets. If the current rate and pattern of growth continues,
Indonesia will not meet its poverty reduction target of 8.2 percent by 2009. In fact, if the current pattern of growth continues,
the medium-term poverty reduction target of government will not be met even if growth were accelerated to the projected
rate of 6.2 percent. To meet the government s poverty target, growth must become more pro-poor. For example, if the
incomes of the poor grow at the same rate as those of the rich then the medium-term target can be broadly met.
Making growth work for the poor will require getting the poor onto effective pathways out of poverty.
This will mean harnessing the structural transformation that is ongoing in Indonesia albeit at a significantly slower rate
than pre-crisis. This transformation is characterized by two phenomena. First, there is an ongoing shift from more ruralbased to more urban-based activities. Indonesia has experienced rapid urbanization, with the population of Indonesia s
cities nearly trebling in 25 years. This has stimulated a shift from rural to more urban-based activities, even when households
have not in fact changed location (some 35 to 40 percent of urbanization). Urban markets are thus becoming increasingly
important for both the rural and the urban poor. Second, there has been a marked shift from farm to more non-farm
activities. In rural areas in particular, this has meant substantial growth in the share of employment in rural (or previously
rural) non-farm enterprises (4 percent per year between 1993 and 2002). This transformation suggests two important
pathways that households have taken out of poverty in Indonesia.
xxvi
Overview
The first pathway out of poverty is improved agricultural productivity. This could come from increasing
productivity in small-scale agriculture or an increased shift to commercial farming. Agricultural productivity gains
from the green revolution were one of the main drivers of growth in the three decades commencing with the 1970s.
More recently, high world commodity prices have sustained output growth, while the shift of labor out of the sector
has maintained the growth of labor productivity in agriculture. As a result, recent poverty diagnostics show that
increases in agricultural incomes continue to be an important driver of reductions in poverty. Panel data between
1993 and 2000 show that 40 percent of agricultural workers in rural areas escaped poverty while staying in rural
agriculture.
The second pathway out of poverty is increasing non-farm productivity, whether this be in urban areas or
rapidly urbanizing rural areas. In this regard, the transition through rural non-farm enterprises is an important steppingstone to moving out of poverty, either by connecting rural enterprises to urban growth processes or, importantly, by
these enterprises in the rural fringe being subsumed into urban areas. Between 1993 and 2002, the employment
share of non-poor workers in rural non-farm employment increased by 6.7 percentage points, suggesting that
increasing non-agricultural productivity in rural areas was an important pathway out of poverty. Moreover, many of
these rural areas were urban by the end of the period, showing the complementary roles of urbanization and productivity
enhancements.
A strategy to help the poor take advantage of economic growth has several elements. First, it will be
important to maintain macroeconomic stability: key to this are ensuring low inflation and a stable and competitive exchange
rate. Countries that have had more macroeconomic shocks experience slower growth and poverty reduction than those
with better macroeconomic management (World Bank, 2005a). Indonesia knows better than most countries the dreadful
poverty impact of macroeconomic crises. Second, the poor have to be connected to opportunities for growth. Better
access to roads, telecommunications, credit and formal sector employment are associated with significantly lower poverty.
The benefit of being connected is large, particularly in the case of formal sector employment outside of agriculture. Third,
it will be important to invest in the capabilities of the poor. Part of the strategy for growth must be to invest in the poor so
that they are well prepared to benefit from the opportunities for income growth that present themselves. In both rural and
urban areas, higher levels of education of household heads are associated with higher levels of consumption. Investing in
education for the poor will boost the capability of the poor to participate in growth.
Rural non-farm activities have helped Indonesians leave poverty
Change in share of employment 1993-2002 (%)
Figure 4
Rural Ag
Rural Non-Ag
Urban Ag
Urban Non-Ag
Source: Susenas 1993, 2002, rural/urban classification of 1993.
xxvii
Three priority areas of action stand out:
Revitalize agriculture and increase agricultural productivity. With almost two-thirds of poor household
heads still working in agriculture, boosting agricultural capability remains essential for broad-based poverty reduction.
Analysis shows that households working in informal agriculture are most likely to be poor, with increases in
consumption associated with moving to formal agriculture. Yet agriculture in Indonesia is not doing well. Despite
labor productivity remaining buoyant with the outflow of labor from agriculture, agricultural total factor productivity
growth has been negative since the early 1990s, from annual gains of 2.5 percent in 1968-92 to annual contractions
of 0.1 percent from 1993 to 2000. The government can contribute to increasing agricultural productivity through:
boosting investment in key infrastructure, notably farm-to-market roads and irrigation, while widening local water
management; encouraging and supporting diversification into higher value-added crops; working with the private
sector to ensure that exports meet world standards; boosting expenditure on agricultural research; and redesigning
the decentralized extension service to allow for greater involvement of the private sector and civil society. These
efforts to improve agricultural productivity should also include development of better marketing and information
systems for rural-based businesses. Efforts to speed up land titling and more broadly ensure appropriate forms of
secure tenure throughout the country will also help.
Remove the ban on rice imports. Lowering the rice price and creating greater price stability by removing the
import ban on rice is the fastest way for the government to reduce poverty quickly. Rice is a critically important food
commodity for all Indonesians but especially for the poor, constituting 24.1 percent of their consumption. For the
country as a whole, four out of five households are net consumers of rice, that is, they consume more rice than they
produce. The sharp surge in rice prices during the economic crisis, and then again in 2005-06, increased the total
poverty rate. Indeed, it is estimated that the 33 percent increase in the rice price between February 2005 and March
2006 alone put around an additional 3.1 million people into poverty. This price increase was well beyond the food
inflation rate showing that it was not primarily caused by the rise in the fuel prices (Figure 5). The rice ban could be
replaced by a low tariff. In addition, targeted provision of rural infrastructure and agricultural research and extension
services will help farmers improve rice productivity.
Figure 5
Rice price increases hurt the poor
The domestic rice price overtook international prices

(incl. the tariff) after December 2005 due to shortage
in domestic supply
Source: FAO, Border prices wholesale equivalent for Vietnam

rice 25 percent, Jakarta wholesale price IR 64 III PIBC.
xxviii
Proportion of households that are net consumers of rice (%)

Rice
All
All
farmers
farmers
Indonesians
Urban
27.67
73.7
94.53
Urban poor
25.26
67.32
85.79
Urban non-poor
28.49
75.91
95.51
Rural
26.63
64.19
72.26
Rural poor
33.17
68.1
72.14
Rural non-poor
25.17
63.17
72.28
Total
26.77
65.44
82.74
Poor
31.79
67.98
76.46
Non-poor
25.57
64.75
82.74
Source: Susenas, 2004.
Overview
Launch a rural roads program. Access to infrastructure and roads is shown to be a key correlate of poverty. Having
all-year passable roads is associated with higher expenditure levels in both urban (7.7 percent higher) and rural (3.1 percent
higher) areas. Again, the less well connected regions of eastern Indonesia will see particularly strong gains from improved
connections. The critical importance of infrastructure is also reflected in responses from small rural enterprises. In a firmlevel survey, road access, the cost of transportation and the quality of roads all feature strongly in the top concerns of rural
enterprises surveyed. Analysis shows that improving the quality of roads would be associated with a rise in the average
proportion of income in a village coming from non-farm enterprise income and non-farm salaries and wages by 33 percentage
points. Yet, only 61 percent of poor households have access to all-year passable roads (while 76 percent of non-poor
household access these roads). Currently, around four-fifths of all roads now fall under the responsibility of district governments
and 64 percent of these roads are considered to be in less than good condition. Moreover, the condition of district roads
seems to be deteriorating as less money is allocated to maintenance. Tackling these cross-cutting constraints to investment
could play a major role in connecting the poor to growth. It would be aimed directly at addressing the difficult issue of
regional disparities, by linking regional disparate regions to growth processes. For district level roads, there is a need to
increase financing, particularly for maintenance, through a concerted strategy. One option is a special DAK. These funds
could be targeted (using poverty maps) to areas where access for the poor is worst. The DAK should also leverage and
increase district-level funding for road maintenance. Another possibility is the development of a road fund at the district or
provincial level along with the establishment of a district-level road management system.
Making Services Work for the Poor

Making services work for the poor requires improving institutional accountability systems and introducing
incentives to improve service delivery in order to improve human development outcomes. Currently, poor
service delivery lies at the center of weak human development outcomes, or non-income multidimensional poverty, such
as poor quality of health and educational care. According to survey data, 44 percent of households in the poorest quintile
with children enrolled in school reported difficulties in financing junior secondary education. The poor pay 7.2 percent of
their total expenditure for each enrolled student at junior secondary level. On the demand side, to address this issue the
government should consider programs of targeted transfers, such as scholarships or conditional cash transfers linked to
attendance to junior secondary school (and vocational schools). Junior secondary school capacity in Indonesia provides
learning opportunities on average to only some 84 percent of potential students in the 13 to 15 age group. Likewise, huge
regional disparities in these indicators are reflections of differential access to these services across regions. The government
needs to focus on making services work for the poor to address the multidimensional aspect of poverty and high regional
disparities in these indicators. With regard to addressing the multidimensional aspects of poverty, efforts should be
aimed at improving service delivery especially with a view to improving quality.
The other area of focus is improved access of the poor to services to reduce regional disparities in
human development indicators. The variability in access to services across the country is a fundamental driver of
regional inequalities in poverty-related outcomes. While in some regions, such as Central Java, school capacity exceeds
100 percent, in East Nusa Tenggara and South Sumatra the average coverage of school capacity is below 60 percent of the
number of potential students, indicating a lower degree of access. And the average distance to junior secondary schools
in Java is 1.9 km while in Papua the average distance is 16.6 km (Podes, 2005). Of the junior secondary schools available,
a ministry survey in 2004 found 27.3 percent of their classrooms to be damaged in some way. More junior secondary
classrooms and schools need to be made available, and one way this can be achieved is by converting primary schools
where there is excess supply.
xxix
In education, a key issue is the high drop-out rate among the poor as they transition from primary to
junior secondary level. The main problem is the lack of access among the poor to junior secondary schools and to
vocational schools. Lack of access is both physical and financial. Financial access is limited due to the high level of fees
that creates a barrier to the poor s education at junior secondary level. The share of poor children who complete primary
school is 89 percent, but only 55 percent of poor children complete junior secondary. Yet, diagnostics show that the
returns to education increase with increasing levels of schooling. In 2002, the increase in urban (rural) male wages
resulting from an additional year of schooling for someone who had only one year of schooling already was 8.3 percent
(6.0 percent); after five years of schooling, the return was 10.0 percent (7.6 percent) and after eight years of schooling it
was 11.1 percent (8.8 percent).
Figure 6
Inequality in outcomes remains high
Proportion of current cohort that reaches to each grade
Under-five mortality per 1,000 live births,

by quintile
80
Deaths per 1000 births
70
60
50
40
30
20
10
0
Poorest
quintile
Quintile 2
Quintile 3
Quintile 4
Richest
quintile
Source: Enrollment data are from analysis of data from Susenas 2004, current cohort defined as between 20 and 25 years of age.
Mortality and water and sanitation are from analysis of data from Demographic and Health Survey 2002/2003.
Improving attainment among the poor at the junior secondary school level will require both supply- and
demand-side interventions. On the supply side, teacher management policies that deploy greater numbers to remote
areas are needed. Remote schools have higher student-teacher ratios, and a recent study showed that despite the general
teacher oversupply in Indonesia, 74 percent of remote schools are below entitlement. Likewise, despite the large allocation
to teacher salaries in the budget, teachers monthly salaries are low. While spending on salaries is 50 percent of total
educational spending, teachers are paid 21 percent less than workers with equivalent qualifications. On the demand side,
junior secondary schools and vocational training schools (SMKs) can be made more affordable for the poor by targeting
transfers to poor students through scholarships or conditional cash transfers (CCTs).
Better primary healthcare requires better incentives for both the poor and for providers. Chaudhury et al
(2005) found that absenteeism among health workers in Indonesia is 40 percent, even higher than in Bangladesh and
Uganda. A governance and decentralization survey found that only 30 percent of primary health clinics visited had complete
stocks of medicines. For higher-level healthcare, affordability is an issue and targeted programs would make sense, such
as a health insurance program. It will be important to build on this recently launched program based on recent assessments,
findings point to the need to improve targeting and open up service provision. Progress in reducing maternal mortality
lies in increasing the proportion of births attended by skilled professionals, increasing the proportion of institutional
deliveries and improving access to 24-hour obstetric care. Currently, only 72 percent of births are attended by skilled
personnel in Indonesia nationally, compared with 97 percent in Malaysia and China and 99 percent in Thailand. Increasing
deliveries with skilled staff in attendance in health clinics will require action on four fronts: increasing the availability of
skilled midwives in remote areas; improving the affordability of care by skilled professionals; increasing awareness,
especially among women, of the importance of skilled midwifery at birth; and improving the quality of skilled birth
attendant services.
xxx
Overview
Indonesia needs to address the shortfalls facing the poor in accessing safe water and sanitation. An
estimated 50 million rural poor are not connected to piped water and of that number six million pay higher rates (in excess
of those charged the state water utility company). For rural areas, the existing community management supply model that
has been shown to work should be expanded. This currently covers 25-30 percent of the rural population, but could be
expanded to cover the 50 million people currently without adequate water supply. In urban areas, levels of access to utility
supply are lower in the poorest quintile, but the district water utilities (PDAM) service to all households is also limited. In
practice, the urban poor obtain their water from many sources, primarily non-network water and self-supply. For urban
areas, supply must be strengthened by improving capacity and incentives for water utilities (PDAMs) to plan, deliver and
monitor service delivery. In addition, PDAMs need to be mandated and given incentives to scale up services to areas
inhabited by the poor. Consideration needs to be given to designing appropriate tariff structures for the poor who benefit
from current connections, or who will have future connections. The coverage of sanitation services in Indonesia is the
worst in the region, with less than 1 percent of all Indonesians accessing piped sewerage systems. Survey data show that
80 percent of the rural poor and 59 percent of the urban poor have no access to adequate sanitation. It is estimated that the
cost of poor sanitation is about 2.6 percent of GDP, while public spending on water and sanitation together is less than 0.2
percent of GDP. Two immediate options are the development of a national strategy to increase sanitation financing, and
local government investment in sanitation infrastructure at the neighborhood and city levels, for example through a
specific DAK for sanitation, or by including sanitation services in minimum service standards.
Figure 7
Inequality in services remains high

Main source of drinking water, by quintile
Type of toilet, by quintile
Source: Enrollment data are from analysis of data from Susenas 2004, current cohort defined as between 20 to 25 years of age. Mortality and water and sanitation are from analysis of data from Demographic and
Health Survey 2002/2003.
Three priority areas for action stand out:
Clarify functional responsibilities for the provision of services. The lack of clarity of responsibilities is
paralyzing accountability in service delivery. The financing and provision of services is based on bureaucratic
instructions, providing relatively little actual autonomy to either providers or beneficiaries. A typical government
health clinic has eight sources of cash income and 34 operational budgets, many of which are provided in kind by the
central or local government. The central government should limit its role to policy-making, staffing issues, information
and developing core national service-delivery standards. Provincial level governments should focus on fixing regional
standards, building capacity at the district level and implementing cross-district services, while at the district level
governments should be responsible for planning and budgeting, and ensuring implementation of service delivery.
Providers should be expected to monitor outcomes for the services they provide. Communities should be empowered
to provide feedback to those providers, possibly even managing their own targeted programs and helping to build/
maintain local infrastructure.
xxxi
Improve civil service staffing and management in social sectors. The government needs to review and
provide clarity in the regulatory framework and incentives for organizational and personnel management by instituting
a less rigid employment regime and abolishing the system of structural and functional positions and the rigid
ranking of posts. While civil service reform is not easy, it is a vital component to unsticking service delivery. A
recent study involved making surprise visits on more than 100 primary schools and health centers in Indonesia. The
study found absenteeism rates of 10 percent among teachers and 40 percent among health workers. Indonesia had
the highest health-worker absenteeism rate of all the countries included in this global study. Not only does high
absenteeism reduce quality, but it also reduces the demand for public health services. Creating more robust and
predictable incentives for staff will also help, as has been proven in the case of some frontline service providers (see
Box 2).
Provide stronger incentives for service providers. The provision of clear, predictable rewards and sanctions
is necessary to provide a framework that will systematically promote good behavior and outcomes by service providers.
Incentives can take the form of performance contracts or incentive payments for good results. One specific option is
to pilot the use of service agreements a contract between a public service provider and local government that
specifies the services that will be delivered, and the resources that will be provided to do so. In addition, local
governments can work with private providers as partners in delivering good quality services to the poor. For example,
almost 60 percent of all healthcare visits were to private facilities or providers in 2004. The poor use private services
not only because they are often cheaper, but they are perceived to be of higher quality. Many local governments are
working creatively to provide incentives for improved educational outcomes and healthcare (see Box 2 below). On
the demand side, conditional cash transfers (CCTs) can help where behavioral change is required to enhance demand,
such as child well-being and nutrition status check-ups, and child immunization
Box 2 Better incentives and information can change the behavior of service providers
Numerous experiments using incentives for local service providers have been conducted in Indonesia in recent years. In
some cases, there was a marked change in the behavior of service providers in response to changed incentives. For example:
In Tanah Datar district, West Sumatra, a scheme was launched in 2002 to provide stronger incentives to English teachers and
headmasters by offering them study tours of Australia, Malaysia, and Singapore if they promised to follow up with changed
practices. Teachers returning from study trips had to submit a group report to the bupati, with follow-up observations and
recommendations:
The trips increased the motivation to improve conditions in schools. These included stronger discipline of teachers,
students, and parents; smaller class sizes; providing classes in computer skills and English; changing the teaching
methodology; and, consulting with students.
The trips effected changes in teaching methodology. One English teacher began teaching her classes in English rather
than Indonesian following her trip to Australia. She has also begun using a student agenda, in which students record
their activities in English, as well as what they have learned from them, as an aid in teaching.
The trips increased interest in student performance. Teaching hours increased, due to both school-based management
and the stronger incentives policy; on average students now study about 15 hours more a week. To demonstrate his
commitment to raising student test scores, one headmaster even signed an agreement with his school committee stating
that if scores at his school were not above a certain level he would resign.
xxxii
Overview
In Jembrana district, Bali, health sector reforms created a new health insurance program (Jaminan Kesehatan Jembrana). The
program provided free primary healthcare for all enrolled citizens and free secondary care for all poor members. It also
enhanced client choice by enabling members to choose between private and public healthcare providers, both of whom were
reimbursed under the scheme. In addition to increasing coverage of health services, the scheme directly affected the behavior
of public health staff, who had to compete for clients with private providers in the wake of this reform. As a result, public
providers measurably improved their client orientation by sending mobile clinics and doctors to remote areas at least once a
month (rather than just providing health education in these remote areas, as they previously had); improving medicine
packaging; and providing full smile reception for patients. In addition, the management board of the project supervised
quality control for reimbursements by both creating a clear set of service standards for all providers and investigating cases
of malfeasance.
As part of the Safe Motherhood Program in Pemalang district, Central Java, poor women were issued vouchers that they
could exchange for prenatal care from midwives. The midwives were usually responsible for distributing these vouchers. With
the added incentive of additional fees earned from clients with vouchers, midwives substantially increased the number of poor
women they treated. This had the additional beneficial impact of introducing poor women to the formal health system and
inducing them to seek out healthcare from formal providers more frequently.
Source: Leisher and Nachuk, 2006.
Making Public Expenditure Work for the Poor

By reducing fuel subsidies, Indonesia has taken a big step towards making public expenditure more
pro-poor. The fuel subsidy a universal price subsidy represented the biggest subsidy or transfer to households in
recent years: it was, de facto, the centerpiece of Indonesia s social protection scheme until 2005. By fixing fuel prices at
subsidized levels well below world prices, the government effectively supported a transfer to fuel-consuming households,
protecting them from fluctuations in world prices. Between 1998 and 2005, fuel subsidies averaged three-quarters of the
total subsidies and transfers that constituted Indonesia s social protection system. However, spending on fuel subsidies
benefited mainly middle and higher income groups that consumed more fuel. Figure 8 depicts the regressive incidence of
the fuel subsidy, had the government not changed the domestic price of fuel in 2005. In total, the benefits accruing to the
richest 10 percent from fuel subsidies were more than five times those accruing to the poorest 10 percent
Figure 8
Regressive fuel subsidies have been replaced by progressive cash transfers
Estimated impact of 1 October 2005 fuel price increase

by expenditure decile
Estimated impact of 1 October 2005 fuel price increase by expenditure

decile as a percentage of mean household expenditures in decile
50.000
Rp per capita per month
45.000
35.000
30.000
25.000
20.000
15.000
10.000
5.000
0
Source: Susenas 2004, World Bank staff calculations
Source: Susenas 2004, World Bank staff calculations.
xxxiii
Government can make good use of the increased public resources. To date, public expenditure has not always
been successful at effectively addressing the constraints that still hinder the poor and keep them mired in poverty. At a
time when the government has increased fiscal space following the reallocation of regressive fuel subsidies, it is all the
more important to ensure that this expenditure makes a real impact on the poor. The government now has a window of
opportunity to begin addressing the high level of vulnerability of the poor in Indonesia by channeling spending into social
protection systems that mitigate this vulnerability. An important component of this reallocation of public spending is
through focusing on ways to improve the incomes of the poor, especially in view of the fact that 49 percent of Indonesians
earn less than US$2-a-day. Public spending that can have a direct impact by boosting income levels will also have a
commensurate impact on poverty. One priority that stands out which the government is already moving to address is the
scaling-up of community driven development (CDD).
Public resources can be targeted better. While the CDD approach will allow for a broad focus in addressing
widespread vulnerability, it will also be vital to use public expenditure to target the very poorest, who lag behind in terms
of the non-income, multidimensional aspects of poverty. Only through a more effective targeted approach in public
spending can the government achieve progress in these human development indicators. In particular, the government
should press ahead with the piloting of targeting transfer to the poor. This could be done through CCTs aimed at the
delivery high-quality services to where they are needed most. Public expenditure can also be a sharp instrument for
addressing the issue of regional disparities in both the income and non-income dimensions of poverty. In view of the wide
disparities that continue to exist across the country, there is a need to make the current system of transfers from the center
to the regions more pro-poor, and to build capacity and create incentives within local governments to improve their focus
on pro-poor policy implementation.
Local government capacity is uneven, and is now the binding constraint to poverty reduction. About onethird of total public expenditure is allocated and spent at the district level. While this is a sign that decentralization is
working, the problem is that many local governments are facing difficulties in the planning, budgeting, and execution of
this spending. One indication of these difficulties is the increasing level of local government surpluses witnessed in
recent years. There is therefore a growing and increasingly urgent need for a concerted effort to improve the capacity of
local governments, and the civil servants who work for them, to plan, budget and implement programs for service delivery
and poverty reduction. Once again, this is an issue that could be usefully complimented through incentives at the district
level to encourage local governments to both spend more efficiently and also in a more pro-poor way. For example, the
central government could induce improvements in service delivery with increased funding for those districts achieving
certain targets in service delivery. However, this will require a strong commitment towards building a partnership between
local and central governments and a long-term strategy to address capacity issues.
Three priority areas for action stand out:
xxxiv
Scale up Indonesia s successful community-driven development (CDD) programs. Community-driven

development projects, such as Kecamatan Development Program, or KDP, have a history of success in Indonesia.
The CDD approach has yielded high rates of return on investment; it addresses in an integrated fashion the binding
constraints to poverty reduction at the village level whether these are village roads, water and sanitation systems, or
other constraints to poverty and well-being and it targets poorer areas. Indeed, evidence from community-level
infrastructure projects constructed as part of the KDP project shows that economic internal rates of return are between
22 and 47 percent and were 56 percent cheaper than equivalent roads built through government contracts. The CDD
approach also empowers the poor by giving them a say in how community resources are spent. For example,
community-driven programs have been shown to assist with labor-intensive approaches to the construction and
maintenance of village and kecamatan roads. Moreover, given a choice of investments, villagers in these programs
tend to choose infrastructure programs: some 67 percent of total spending chosen by these communities goes to the
Overview
communities. Indonesia should aggressively scale up its CDD approach to cover the entire nation. It is estimated that
such a national program could be up and running within three years. The governments recent announcement to
launch a community-based development program nationally can be a key component of a vigorous poverty reduction
strategy. While mitigating the vulnerability of the poor, it will address regional disparities in incomes through
employment creation, as well as helping to connect the poor to growth through the development of rural infrastructure.
Pilot demand-side programs that improve service quality and encourage behavioral change. In
particular, CCT programs can be effective in addressing key multi-dimensional poverty goals for poor families, for
instance in terms of preventative basic healthcare and nutrition, and in education. By placing targeted spending in
the hands of poor households, this will motivate poor families to demand the services they require to meet program
conditions and motivate service providers. CCT programs could therefore be targeted to ensure that children and
expectant mothers receive critical healthcare services. Equally important in improving outcomes among the poorest
households will be CCTs aimed at inducing families to enroll their children at school and maintain regular school
attendance. In view of the detailed design and planning that would be necessary for such a program in Indonesia,
CCTs should be piloted for extremely poor, rural households and, if successful, scaled up gradually. Success would
depend on developing: (i) a beneficiary roster with family demographic data; (ii) a policy and procedures to verify
compliance, and to deal with non-compliance; (iii) assessment of the availability of relevant health and education
services to address any anticipated supply-side issues; and (iv) a complaints, resolution and appeals system. While
this approach has proved to be successful in other countries, given that a CCT program has never been tried in
Indonesia the government should assess the effectiveness of such a program before scaling up.
Make both the DAU and the DAK more pro-poor. The General Allocation Fund (DAU) is not structured to
target areas with high poverty incidence indeed, there is no correlation between DAU transfers and poverty rates.
Given that the DAU provides on average about two-thirds of sub-national revenues, it is important that the fiscalgap formula be refined to increase the weighting of the poverty variable already contained in the formula. In addition,
the recent requirement that the DAU must first cover fully the wage bill of all local governments crowds out other
spending, especially at a time when salaries have surged to more than 50 percent of DAU allocations. Meanwhile, the
Special Allocation Fund (DAK) is not being used for its intended purpose the delivery of basic services in lagging
regions. First, the DAK is small, comprising only 3 percent of central government transfers to the sub-national
governments in 2005. Second, the DAK is spread across sectors and regions too thinly, in addition to often being
earmarked for investment in facilities and infrastructure regardless of whether these are constraints to pro-poor
outcomes. Indeed, with the exception of the health sector, poverty indicators are not used to determine DAK allocations.
With the DAK doubling in 2006 and due to rise by a further 25 percent to Rp 14.4 trillion in 2007, this presents an
enormous opportunity to address priority needs in resource-deficient districts and sectors. In particular, the DAK is
a powerful instrument that should be used to support the pathways out of poverty and improved service delivery in
lagging regions. To this end, the DAK should be used to leverage local government resources by revising the matchingfunding requirement, which at present only requires a minimum 10 percent matching funds. The government should
consider introducing performance-oriented incentives to improve the efficiency and effectiveness of DAK funding.
For example, the allocation of an education DAK could be conditional on a pre-defined increase in junior secondary
completion rates. Key areas in which to consider such leverage would be in sanitation, which is in crisis in Indonesia
with only 1 percent of Indonesians having access to piped sewerage systems, and in rural road maintenance, which
has still to recover to its pre-crisis levels.
xxxv
Priorities for Poverty Reduction

Three transformations are taking place in Indonesia, each of which can be more or less pro-poor. This report suggests
policy measures that can make these changes rapidly reduce poverty.
Table 3
First, as it grows, Indonesia s economy is being transformed from one with agriculture as its mainstay to one that will
rely more on services and industry. The priority for making this growth work for the poor is a friendlier rural investment
climate, principally through better rural roads.
Second, as democracy takes hold, the government is being transformed from one where social services were delivered
centrally to one that will rely more on local governments. The priority for making services work for the poor is
stronger capacity of local governments and better incentives for service providers.
Third, as Indonesia integrates internationally, its system of social protection is being modernized so that Indonesia
is both socially equitable and economically competitive. The priority for making public expenditure work for the poor
is to shift from intervening in markets for commodities that the poor consume (such as fuel and rice) to providing
targeted income support to poor households, and using the fiscal space to improve critical services such as education,
health, safe water, and sanitation.
Nine steps towards a poverty-free Indonesia
Key dimensions of Indonesian poverty
Making Growth Work for the Poor
Reducing vulnerability
Reducing non-
Reducing regional
to income poverty
income poverty
disparities in poverty
1. Revise rice policy
3. Rural roads program
2. Revitalize agriculture
Making Services Work for the Poor
1. Improve incentives for

service providers
2. Clarify functional
responsibilities.
3. Improve local capacity
Making Expenditures Work for the Poor
xxxvi
1. Scale up CDD programs
2. Pilot CCTs for the poor
3. Make DAU and DAK pro-poor
xxxvii
Chapter
Introduction
xxxviii
Indonesia stands at the threshold of a new era and at an important juncture of its history. After the historic
economic, political and social upheavals at the end of the 1990s, Indonesia has started to regain its footing. The country
has largely recovered from the economic and financial crisis that threw millions of its citizens back into poverty in 1998
and saw it regress to a low-income status. Recently, it has once again crossed the threshold, making it one of the world s
emergent middle-income countries. Likewise, poverty rates that increased by over one-third during the crisis fell back to
pre-crisis levels in 2005 despite rising somewhat in 2006 largely driven by hefty rice price increases in late 2005 and
early 2006. Meanwhile, politically and socially Indonesia has seen some major transformations: it is now a country with
a vibrant emergent democracy, a newly decentralized government, and far greater social openness and public debate.
Yet as Indonesia faces this new era, the challenge of reducing poverty remains one of the country s
most pressing challenges. While national poverty rates may have largely recovered to pre-crisis levels notwithstanding
the recent increase in the 2006 poverty rate this means that close to 35 million people still live in poverty. This is more than
the total number of poor in all of the rest of East Asia combined, excluding China. Moreover, national poverty rates mask the
large number of near-poor in Indonesia, who lie precipitously close to the poverty line. Some 40 percent of the population,
or close to 90 million people, live between US$1 and US$2-a-day. Indeed, even though Indonesia is now a middle-income
country, the share of those living below US$2-a-day is similar to that of the regions lowest-income countries. This high
vulnerability of the near-poor in Indonesia was once again highlighted by the rice-price driven increase in the poverty rate in
2006, causing poverty to rise from 16.0 percent to 17.7 percent. Indonesia is also seriously lagging on some key non-income
dimensions of poverty. Its maternal mortality rate, junior-secondary enrollment rate, and malnutrition rate, for example, have
not been improving fast enough and are high relative to comparable countries in the region. Indonesia is also characterized
by huge regional disparities and inequalities. There are regions in Indonesia where the poverty levels and characteristics are
more similar to some of the poorest low-income countries in the world, and there are pockets of poverty in even the better-off
regions of the country.
The democratically elected Government of Indonesia recognizes poverty reduction as a primary challenge
and has set itself ambitious short- and medium-term poverty reduction goals. The Indonesian government
is clearly committed to reducing poverty in its medium-term plan (RPJM) for 2004-09 which, in turn, draws from its
national poverty reduction strategy (SNPK). In addition to signing on to the Millennium Development Goals for 2015, in
its medium-term plan the government has laid out its own key poverty reduction objectives for 2009. These include the
ambitious but relevant targets of reducing the poverty headcount rate from 18.2 percent in 2002 to 8.2 percent; increasing
the junior-secondary school enrollment ratio from 79.5 percent in 2002 to 98 percent; and bringing down the maternal
mortality rate from 307 per 100,000 live births in 2002 to 226.
How to achieve its poverty reduction goals and make the new Indonesia work for the poor is one of the
preeminent policy challenges facing the nation. History has much to teach, and Indonesia s record in poverty
reduction since independence and through the crisis stands out as an example for the developing world. The challenges
and lessons of success and failure that are a legacy of this history are important in this regard. But trying to use the
lessons of history alone as a guide for the solutions of the future is not sufficient. This is in many respects a new
Indonesia, with a new democracy, a newly decentralized administration and newly formed (and reforming) institutions.
Consequently, the challenges involved in tackling the issue of poverty in Indonesia today present a very new set of
circumstances to policy-makers.
The purpose of this report is to identify the nature and key constraints to poverty reduction in today s
Indonesia and to provide concrete recommendations on how Indonesia can move forward to achieve its
poverty reduction objectives. It aims to contribute to the policy debate and decision-making process in Indonesia by
putting forth: (i) new and more comprehensive analysis of empirical poverty diagnostics; and (ii) suggestions on concrete
policies and programs for a strategic action-plan to achieve Indonesia s stated poverty-reduction objectives. This report
lays out how Indonesia can better align policies and programs to achieve the key poverty indicators in which Indonesia is
lagging and that are identified by planning documents such as the SNPK and RPJM.
This report is organized around an action-oriented diagnostic and strategic framework that lends itself
to action. Chapter 2 lays out the history of growth and poverty reduction, and analyzes where Indonesia stands today
and how past experiences have influenced the present. It also highlights that the lessons of the past are relevant to the
future. Chapter 3 presents new analysis that enhances our understanding of the nature and causes of poverty in Indonesia,
as well as some of the forces driving the changes in poverty. Based on this understanding, the next chapters lay out pillars
for action.
Making Growth Work for the Poor has been, and will need to continue to be, a fundamental
component to reducing poverty. Chapter 4 analyzes how growth processes have aided people in finding pathways
out of poverty, and lays out strategies to address the constraints that hinder the poor from linking into and contributing
towards growth. Its focus is how to increase the incomes of the poor.
Chapter 1: Introduction
Making Public Spending Work for the Poor focuses on how the government can better use its
resources to achieve its poverty reduction objectives. Chapter 5 analyzes the benefit incidence of spending
in key sectors that matter most for the poor and proposes policies and actions to make budget allocations and sector
spending programs more pro-poor. This issue is particularly pertinent in the context of the fiscal situation in today s
Indonesia. Greater fiscal consolidation combined with added fiscal resources resulting from increased fuel prices
has raised the issue among policy-makers of how best to use resources for poverty reduction. The focus of this
chapter, beyond further reductions in income poverty, is how to reach some of the key targets among the non-income
dimensions of poverty.
Making Social Protection Work for the Poor addresses the high vulnerability that characterizes
poverty in Indonesia. This issue was recently brought to a head by a government decision to reallocate the fuel
subsidy that de facto formed the biggest budgetary share of any such subsidy or transfer program. Chapter 6 provides
original analysis of the actual risks and vulnerabilities faced by the poor and the mechanisms they use to cope with
shocks. It proposes specific programs that might address the nature of shocks that affect the poor in Indonesia. It
looks at mitigating risks and vulnerability as a key component to poverty reduction.
Finally, Making Government Work for the Poor argues that improved governance is a critical
factor in ensuring that growth, spending, and social protection work for the poor. Strengthening key
government systems and accountability mechanisms is thus a key part of the poverty reduction agenda, as without
improved governance and institutions, it will not be possible to get the New Indonesia to Work for the Poor.
Chapter
A History of Growth and

Poverty Reduction
Introduction
Only 50 years ago, Indonesia was one of the poorest countries in Asia. The story of its poverty and poverty
reduction is a story of the political and economic eras that determined the nation s development trajectory: colonial rule
and exploitation; authoritarian rule, coupled with sustained growth, then dramatic collapse; and, most recently, democracy
accompanied by economic flux and tentative stabilization. At independence in 1945, dislocated by Japanese occupation
during World War II after years of colonial rule and commercial exploitation, the vast majority of Indonesia s population
was impoverished. By 1993, however, with poverty reduced to 14 percent of the population and annual economic growth
at more than 7 percent, Indonesia was ranked along with a handful of other East Asian countries as a High-Performing
Asian Economy (World Bank, 1993) and lauded for its astonishing transformation. In order to understand this remarkable
turnaround it is necessary to develop an understanding of the key factors that drove the change in livelihoods of some 100
million Indonesians.1
History has much to teach Indonesia as it struggles to re-establish economic growth and reconnect that
growth to its remaining poor. Because Indonesia has experienced such sharp swings in its development path, there
are numerous successes to examine, just as there are also lessons to learn from the many failures. Drawing on the vast
historical literature in this area (Timmer, 2003; Hofman et al, 2004; Temple, 2001; McIntyre, 2003; and Hill, 1996), this
chapter briefly sets out this history, beginning with the unfavorable starting point, then focusing on the policies of the
Soeharto government that brought about the structural transformation in the livelihoods of the poor, and going on to
reflect on the causes and impacts of the greatest economic crisis in Indonesian history. The story of three decades of
sustained pro-poor growth, juxtaposed with the story of rapid collapse and recovery, provides useful insights for future
policy-making.
The purpose of this chapter is to provide some context to the current poverty scenario in Indonesia by
telling the extraordinary story of poverty reduction and economic growth that has occurred since Indonesia
emerged from colonial rule. To this end, this chapter traces the paths of growth and poverty reduction in Indonesia,
focusing on the policies and structural changes that marked key stages. The chapter first describes the chronic poverty of
the pre-independence period, the post-war muddling through followed by gradual economic deterioration under the
Sukarno regime, and then the later 30-year period of growth and poverty alleviation, and the nature of the structural
transformation that changed the nation s history. It then focuses on the more recent past, describing the economic crisis,
the post-crisis stabilization, and the emergence of democracy and decentralization in the new millennium.
II
A Troubled History and Chronic Poverty
For the duration of the 350-year period of Dutch colonial rule, the trade and tax regime favored colonial
extraction of income with dire consequences for the Indonesian population. Analysis provided by Van der
Eng and interpreted by Timmer enables an examination of growth, the severity of poverty (through a comparison of the
annual food energy intake measured in kcals), and the income elasticity of consumption over the past century. The record
1
This chapter draws on papers from Peter Timmer and others. Timmer s paper draws attention to the global lessons that Indonesia provides: Indonesia is the original home of the dual economy. Boeke s
experience during the Dutch colonial administration of Java led him to identify two types of economic agents rational and traditional with almost entirely separate spheres of economic activity (Boeke,
1946). Lewis (1954) built his Nobel-Prize-winning model of the dual economy with unlimited supplies of labor on the behavior of such agents. Timmer (2005), p.15.
shows marked variability in both the rate of economic growth and how well it was connected to the poor over a series of
political and economic epochs. During the nineteenth century, growth in consumption was negative, estimated at -0.34
kcal, the second-lowest in Indonesia s measured history while the index of pro-poor growth2 was only a fraction of the
long-term average, illustrating the severe disconnect between the poor and the modest economic growth that occurred
during this period. In the mid- to late-nineteenth century, similar to most other regions of Southeast Asia, the historical
record of Indonesia was one of severe poverty (Timmer, 2004).
The collapse of export prices and disastrous economic management in the 1920s resulted in the lowest
growth rate and least pro-poor growth period in Indonesia s recorded history. At the beginning of the twentieth
century, when Dutch public opinion was starting to influence the management of the colonies for the first time, a more
developmental approach, known as the ethical policy , was implemented for a brief period. The policy brought significant
benefits to the economy (growth reached 1.63 percent) and to the poor (growth in food intake reached an annual average
of 1.39 percent per year). But this internal investment in the country lasted for only a brief period. The collapse of world
prices for export commodities in the 1920s, and the abysmal economic management of Indonesia3 during the Great
Depression, resulted in the lowest economic growth rate and rate of pro-poor growth in any period before independence.
While the colonial authorities had built a significant network of irrigation and transport facilities, there was very little
investment in educating the nation s population. Only 3.5 percent of the population was attending school of any kind in
1939, compared with 26.7 percent in 1995 (Susenas, 1995). Poverty increased significantly during the period of World
War II and the subsequent struggle for independence, which only reached closure with final acceptance by the Dutch in
1949. The tumultuous global period spanning the Great Depression, the Pacific War, and the fight for independence
(1925-50), saw a marked deterioration in per capita income growth rates (-2.42 percent) and a negative rate of pro-poor
growth (-2.57 percent). (Table 2.1)
Table 2.1
Long-term patterns of pro-poor growth in Indonesia 4

Avg annual
growth in income
per capita (%)
Avg annual
growth in kcal
consumption
per capita (%)
Dutch colonial exploitation

1880-1905
0.33
-0.34
0.165
0.051
0.05
Ethical Policy under the Dutch

1905-25
1.63
1.39
2.805
0.878
4.57
Depression, Pacific War, and fight for independence

1925-50
-2.42
-0.78
1.064
0.333
-2.57
The Sukarno era, including the Guided Economy

1950-65
1.46
0.68
1.626
0.509
2.37
The New Order regime of Soeharto

1965-90
3.45
2.10
1.901
0.595
6.56
Long-term averages
1880-1990
0.89
0.22
1.000
0.313
0.89
Time Period
Income elasticity
Index of
of consumption pro-poor growth
for kcal
Source: Timmer, 2005.
A crude index of pro-poor growth IPPG shown in Table 2.1 is based on an analytical relationship between the overall incidence of poverty and the observed, average income elasticity of demand. The income
elasticity of food energy for the entire period from 1880 to 1990, estimated to be 0.313, is used as the long-run base, scaled to one. It is multiplied times the long-run growth rate in per capita incomes, 0.89
percent per year, to generate the long-run average index of pro-poor growth (IPPG) of 0.89. The income elasticity for each separate epoch is then scaled relative to the long-run average, and multiplied times
the growth rate in per capita incomes, to generate the IPPG for each epoch. Note that the IPPG incorporates both the growth and the distributional dimensions of pro-poor growth, and this index is thus a
country-specific version of Equation 1 in Concept Paper on Operationalizing Pro-Poor Growth World Bank (2004).
3
The Dutch forced the Netherlands East Indies, Indonesias colonial name, to remain on the Gold Standard well after their regional competitors, including the Japanese, devalued.
4
Details of the regressions are provided in Timmer (2005), along with a full explanation of the analytical relationship between the overall incidence of poverty and the average income elasticity of demand for
food energy.
Chapter 2: A History of Growth and Poverty Reduction
After independence, during the period when the Indonesian nation was being built, weak and inwardlooking policies resulted in an increase in poverty rates after 1960. By the early 1960s, similar to other postindependence states, poverty had fallen in the postwar recovery, and Indonesia was muddling along with modest growth
and weak but quasi-democratic governance. However, after Sukarno imposed guided democracy in 1959, the situation
deteriorated sharply. Adopting an inward-looking development policy and severely neglecting agriculture, Indonesia was
a prime exemplar of the dangerously degenerative consequences of weak governance and a sickly economy (MacIntyre,
2003). Incomes fell dramatically and the hyperinflation of 1965-66 had an adverse impact on the entire population as the
poverty rate increased rapidly and the economy collapsed.5 Probably 70 percent of the population was absolutely poor by
1966 and hunger was widespread (Timmer, 2003). Unsurprisingly, in 1968, with no hint of the future, Gunnar Myrdal
observed in Asian Drama... no economist holds out any hope for Indonesia.
III
A Period of Growth and Poverty Reduction
The trajectory of growth and poverty transformed dramatically under the New Order government. Starting
in 1968, for three remarkable decades, Indonesia s GDP grew on average by 7.4 percent annually. As a result, in 1997
Indonesia s per capita income reached US$906, more than quadruple the 1968 level (World Development Indicators).
When compared with previous periods in Indonesian history, the quarter century from 1965 to 1990 saw an annual growth
of caloric intake of 2.1 percent a year, 50 percent higher than the next best epoch in 1905-25 and almost ten times the
long-term average. The rate of pro-poor growth reached 6.7 percent in the period 1965-90. This was the highest pro-poor
growth rate in Indonesian history, seven times the long-term average and nearly half as high again as the next best epoch
in 1905-25.
Figure 2.1 Periods of sustained growth have brought rapid poverty reduction in Indonesia, 1961-2005
Source: Central Bureau of Statistics (BPS), Hofman et al, 2004.

(See Annex II.1 Poverty lines, percentage of poor and total numbers of poor in Indonesia, for details on poverty numbers over time.)
5
At 2.37, the IPPG is surprisingly high during the Sukarno era, when economic policy is widely regarded to have been a disaster. But a combination of a modest recovery from the quarter century of depression
and wars, with average per capita incomes rising 1.5 percent per year, and large average income elasticity for food energy, suggest that what growth there was actually reached the poor.
The story of Indonesia s poverty reduction is first and foremost a story of sustained pro-poor growth.
From 1970 onwards, large and sustained reductions in poverty in Indonesia have only occurred during periods of sustained
economic growth: the more rapidly the country s economy grew, the more dramatically poverty declined, while slower
growth always resulted in slower reductions in poverty. Figure 2.1 provides an illustration of the close connection between
poverty reduction and growth. Over these three decades, each percentage point of economic growth resulted in a 1.3
percentage reduction in poverty.
The pro-poor performance for these three decades was based on a conscious strategy that combined
rapid economic growth with investment and policies that ensured growth reached the poor. The strategy
integrated the macroeconomy with the household economy by lowering the transactions costs of operating in markets.
This strategy also effectively combined efforts to increase human capabilities and increase demand (conceptualized in
Annex II.2 Framework for pro-poor growth). It was designed and implemented by highly skilled economic planners
outside the political sphere, but at the direct urging of President Soeharto. Huge investments were made in the expansion
of education, family planning and health; transaction costs fell markedly through the construction of roads and other
infrastructure that enabled the poor to be connected to the growth process; and, sound macroeconomic management was
(from the late 1970s) accompanied by a competitive exchange rate.
In the early pre-Opec years of the Soeharto government, policy-making focused on the macroeconomic
stabilization of the economy. During the period 1966-73, as the Soeharto government consolidated political power,
comprehensive changes in economic policy established a first phase of economic liberalization: the restoration of external
viability, the imposing of fiscal constraints, the restoration of the banking system and the liberalization of the investment
regime (Hofman et al, 2004). Major investments were also made to restore agriculture through irrigation rehabilitation,
the introduction of high-yielding varieties of rice,6 fertilizer imports and distribution, and the Bimas7 program of extension
and farm credits. Since median farm size was less than a hectare, rice intensification had widespread benefits (Afiff and
Timmer, 1971)8 and food production and overall food supplies rose sharply. Growth rates sky-rocketed to 12 percent in
1968 and the poverty headcount fell rapidly by 10 percentage points to 60 percent over a seven-year period as the
economy stabilized. Annual growth rates stayed largely within a 7 to 9 percent range during the 1970s (Hofman et al,
2004).
The continuation of sound macroeconomic management was key to the success of the boom years that
followed. From 1973 to 1983, the rapid rise in oil prices created a windfall gain for oil exporters across the world. In
Indonesia, net oil revenues increased seven-fold, from US$0.4 billion in 1973 to US$2.8 billion in 1975, and jumped to
US$4.4 billion in 1979 in response to the turmoil in the wake of the Iranian revolution. This windfall in oil revenues
generated current account surpluses and increased budgetary revenues, enabling a rapid expansion of the economy and
massive public investment in infrastructure, health and education. The first official poverty estimates, based on the 1976
Susenas, indicated a national poverty rate of 40 percent (see Annex II.1).
Sound exchange-rate management enabled Indonesia to stimulate an export-led economy and labor
intensive growth, ensuring the poor were connected into the country s growth. Between 1976 and 1978,
income distribution deteriorated sharply as the real appreciation of the rupiah reduced the profitability of tradable goods
production, especially in agriculture (Warr, 1984). Although the regional and commodity dimensions of poverty masked
its economic roots, during the mid-1970s there was a growing awareness of income inequalities and severe poverty in
rural areas. The technocrats took a highly strategic approach to what was then diagnosed as Dutch Disease and, in
November 1978, devalued the rupiah, to the surprise of financial markets. Tradable goods production rapidly recovered,
6
Obtained from the International Rice Research Institute (IRRI).

Mass Guidance (Bimbingan Massal, or Bimas) is a subsidized government credit program in Indonesia in the 1970s and 1980s to promote the green revolution, i.e. the use of high yielding seeds variety and
pesticides to increase agriculture production.
8
Although larger farmers (those who cultivate about one hectare of land) benefited the most in the early years.
7
particularly in the agriculture sector. After 1978, poverty rates again started to decline assisted by a significant recovery in
the share of income (19.9 percent in rural areas) garnered by the bottom 40 percent of the distribution (see Table 2.2). By
the end of the decade, the poverty headcount had fallen to 28.6 percent. Despite this, the national Gini coefficient of
inequality reached what would be its highest point for the rest of the decade. The government ensured that adjustments
were made in the decade that followed. By 1983, when the international oil boom slowed and commodity prices fell,
Indonesia restructured its economy and policies with devaluation and trade openness. Agriculture continued to grow and
rice prices were kept artificially stable. Simultaneously, the government pursued aggressive exchange rate protection,
devaluing first in 1983 and again in 1986 (Hill, 1996; Thorbecke, 1995).
Table 2.2
Poverty elasticity of growth has varied over time with most rapid poverty reduction in the period 1976-1987
Annual % change
in per capita income
Annual % change
in Poverty Index
Poverty elasticity of growth
1967-76
1976-80
1980-84
1984-87
1987-90
1990-93
1993-96
1996-99
5.48
6.37
4.23
2.69
5.66
5.41
5.23
- 3.25
-6.0
- 8.1
-6.8
- 7.0
- 4.6
- 4.6
- 6.2
9.9
-1.09
-1.27
-1.61
-2.60
-0.81
-0.85
-1.19
-3.05 (+)
1999-2002
2.49
-8.2
-3.29

Note: The Poverty Elasticity of Growth (PEG) is calculated as the ratio of the percentage reduction in the headcount poverty index relative to the percentage change in per
capita incomes (in US$PPP) from the World Bank Data Base on Pro-Poor Growth.
The mid-1980s saw a series of trade reforms being introduced to correct the import bias and establish
a pro-poor trade regime. Indonesia s new trade openness led to a significant increase in the role of the manufacturing
sector. By 1986, the import substitution policies that had stifled the country s growth during the 1970s were replaced by
a strategy focusing on labor-intensive manufactured exports. With broad-based industrial growth underway, the 1980s
saw a determined effort by the government at export promotion. The manufacturing sector contributed 29.2 percent of
GDP growth between 1987 and 1992, a marked increase from the 10 percent contributed during the recovery period of
1967-73 (Hill, 1996). Large-scale and sustained economic deregulation led to better incentives for exports, and these
were matched by incentives for foreign direct investment (FDI). Manufactured exports responded even faster than anticipated
and contributed almost half of all exports by 1992, up dramatically from the 3 percent in 1980. The fortuitous push in FDI
from Japan and the pull from the attractive climate in Indonesia thus allowed the manufacturing sector to become a
significant employer of the poor, increasing wages by the end of the 1980s. This growth in the commercial sectors created
a boom in the non-tradable economy9 where most of the poor were making their living (Timmer, 1997 and 2002). As the
export economy boomed in the late 1980s and early 1990s, overall GDP grew by nearly 7 percent annually, roughly half
of which comprised growth in non-tradable goods and services (Timmer, 2004).10
The reform of trade policy was accompanied by targeted investment-made possible by the massive oil
windfall in sectors that benefited the poor. Sound macroeconomic management was strongly supported by
investment in sectors that benefited the poor education, health, family planning, and infrastructure enabling the poor to
benefit from the oil windfall at the household level. The following paragraphs describe the policy instruments and their
successful outcomes.
9
The two commercial sectors are the engines of growth because of their potential for rapid productivity gains. Connecting them to the non-tradable sector, however, is the key to a high elasticity of
connection between overall economic growth and rapid poverty reduction. Unless demand from rising incomes in the commercial sectors spills over to this non-tradables sector, the poor tend to be left out of
the growth process.
10
Little scope for productivity gains in the NT sector, just need higher wages in T sectors. National income accounts, however, are not kept according to this distinction hence the data are more impressionistic.
In the mid-1970s, Indonesia also began an unprecedented and massive investment in the education of
its people. From 1973 onwards, the government funded the construction of schools for basic education through centrally
administered development programs (Inpres11) and, during the decade that followed, public expenditure more than doubled.
Between 1973-74 and 1978-79, more than 60,000 primary schools were constructed at a cost of over US$500 million
(1990 US$), or 1.5 percent of the Indonesian GDP in 1973 (Duflo, 2001). This was the fastest primary school construction
program ever undertaken in the world. Parallel to the construction of schools, the government also trained and recruited
more teachers. According to Duflo, the Inpres program resulted in an increase of 0.25 to 0.40 in the average years of
education and increased by 12 percent the probability that a child would complete primary school.
Enrollment in primary schools doubled from 13.1 million in 1973 to 26.4 million in 1986, reaching over
90 percent of children of primary level age (Filmer et al, 2002). Unsurprisingly, the better off captured this investment
in education at the outset, but evidence indicates that the poor benefited considerably from the expansion of primary
education between 1978 and 1997, and the expansion of the secondary system became increasingly pro-poor in the
decade 1987-97 (Lanjouw et al, 2001). Furthermore, by the 1990s the educational differentials between females and
males had narrowed and, in many of the best universities, females dominated classes (Hull, 2004). Despite these successes,
the improvements in access to primary education were marred by the low quality of schooling and were not extended
effectively to the secondary level.
Driven by increasingly educated girls and young women, the fertility decline in Indonesia over this
period was substantial (see Figure 2.2 on Structural Transformation in Indonesia). Similar to other countries in the
region, Indonesia experienced a dramatic decline in fertility rates: from 5.6 births per woman in 1970 to 2.6 births in 1990
(WHO, 2003). But this decline also revealed massive regional differences, with Java well below the national average and
rates in Nusa Tenggara and Maluku still at 3.3 and 3.8, respectively (Indonesia Population Census, 2000). Evidence
suggests that the fertility decline was driven largely by levels of, and growing interest in, education. At the time the
national family planning program was established in 1970, only half of women of peak childbearing age (15-30 years of
age) had attended primary school. In the next three decades, these figures rose dramatically so that by the turn of the
century only 2 percent of women in this age group were illiterate (Hull, 2005). Various studies have concluded that the
course of fertility decline in Indonesia was shaped by reductions in the desired family size among the succeeding generation
of mothers (with education as a dual driver) and the opportunity to attain substantial control over fertility through the
government-sponsored provision of efficient forms of birth control.12
From 1978, Indonesia s public investment in primary healthcare led to a substantial increase in access
to health services and a marked improvement in health outcomes. In the late 1970s, Indonesia launched a
primary healthcare approach to health investments.13 Over a 12-year period, the country witnessed the construction and
staffing of over 26,000 public health centers and sub-centers (Puskesmas and Puskesmas Pembantu, or Pustu for short)
and 285 districts and 50 referral hospitals (World Bank, 2000). In 1989, the government began hiring nurse-midwives
and placing them in rural villages.14 The share of communities with a nurse-midwife increased from less than 10 percent
in 1993 to almost 46 percent by 1997.15 By 2002, almost half of all births in rural areas were attended by a village nursemidwife.16 Infant mortality rates dropped from over 94 deaths per thousand births in the mid-1970s to 48 in the 1995.17
11
Also see Annex on Inpres Development Grant in Chapter 5 on Public Spending.

Two drivers were mutually reinforcing the more education the parents had attained, the higher their hopes and expectations for their children. One factor that sharpened this process in Indonesia was the
relative gender equity that the cultural setting supported. While boys may have received more attention than girls, the gap was not as great as those found in Chinese or South Asian cultures. (Hull, 2004, p. 2).
The sustainability of current levels of fertility however is now questioned (see for instance Shoemaker, 2005).
13
The expansion of public basic health facilities was inspired by the Health for All (HFA) initiative of the 1978 UN conference in Alma Ata. Indonesia was one of the first countries to put this primary-care oriented
pro-poor health design into place.
14
By 1994 over 50,000 nurse-midwives had been placed across Indonesia an initiative subsequently formalized as the Village Midwives (bidan di desa) program (Parker and Roestam , 2002).
15
Among communities surveyed in the Indonesian Family Life Survey or IFLS (Frankenberg et al, 2004).
16
Among communities surveyed in the Indonesian Family Life Survey (IFLS) the share of communities with a nurse-midwife increased from less than 10 percent in 1993 to almost 46 percent by 1997. By 2002
almost half of births in rural areas were attended by a village nurse-midwife (BPS and ORC Macro, 2003).
17
Mortality rate, infant (per 1,000 live births): Harmonized estimates of the World Health Organization, Unicef, and the World Bank, based mainly on household surveys, censuses, and vital registration,
supplemented by World Bank estimates based on household surveys and vital registration (World Development Indicators, or WDI).
12
10
Between 1977 and 1997, life expectancy rose from 53 to 65 years18 and malnutrition prevalence fell from 40 percent of
children under five in 1987 to 30 percent in 1998.19 This approach was successful in terms of dramatically improving
access, although the quality of services is a growing concern (see Chapter 5 on Public Spending) and progress towards
the Millennium Development Goals (MDGs) is not satisfactory in malnutrition and maternal mortality.
To enhance tradability and boost agricultural productivity, the government also focused on the
development of physical infrastructure and facilities throughout the country. The windfall also supported the
development of widespread and large-scale investment in infrastructure assets. Up until 1994, the government was spending
30 percent of its development budget on infrastructure,20 equivalent to 2.3 percent of GDP (see Chapter 5 on Public
Spending for more detailed analysis). The expansion of the road network was particularly significant. In 1977, local roads
(mostly farm-to-market roads) totaled 8,500km and rose in a 21-year period to 31,900km by 1998. The transformation of
Indonesia s network of highways took off in the 1970s, expanding by an average 8.3 percent annually between 1970 and
1998. This, together with the massive investment in trucks, radically transformed Indonesian capacity to move goods to
ports and around the country in just two short decades. The impact on transaction costs was significant.
But the policy and institutional framework was far from complete, hindering private-sector investment.
A key gap in the framework for development concerned the legislative and regulatory framework for private-sector investment.
During the 1980s, the private sector expanded without a strong legal system to provide the certainty and structure for
private-sector confidence (Hofman et al, 2004). Apart from a chronically ineffective judiciary (and thus any legal recourse),
the basic legal framework dating back to the colonial period gave inadequate provisions to promote business confidence,
and a lack of cooperation between responsible institutions led to poor sequencing of updated regulations and legal codes
for economic activity. This weak enabling environment had (and continues to have) profound long-term impacts on
investor confidence and the sustainability of private-sector investment in a number of sectors vital to continued growth.
And weak, dysfunctional institutions failed to keep up
with the expanding and complex economy, and vastly
increased levels of government spending. Despite the
strong connections that were established between the poor and
the country s growth, the process was strained by the
dysfunctionality of government institutions, especially in the
sectors and at lower levels of government. In the 1970s, the
government appeared to be a munificent benefactor of the
fledgling service and the problems of under-funding undermined
the quality and integrity of the civil service (Hofman et al, 2004).
The massive growth of the civil service reflected the policy
emphasis on service delivery and infrastructure development
(Rohdewohld, 1995), and the recruitment of teachers, doctors
and engineers increased the overall size of the civil service from
525,000 in 1970 to 2 million in 1990 and over 4 million in 1993.
This represented a five-fold increase from 4.1 to 21.8 staff per
thousand inhabitants between 1960 and 1993 (Hofman, 2004).
The problem was not size but quality.21
18
Life expectancy at birth, total (years): World Bank staff estimates from various sources including census reports, the United Nations Statistics Division s Population and Vital Statistics Report, country statistical
offices, and Demographic and Health Surveys from national sources and Macro International (WDI).
Malnutrition prevalence, weight for age (percentage of children under 5): Prevalence of child malnutrition (weight for age) is the percentage of children under five whose weight for age is more than two
standard deviations below the median reference standard for their age (WDI).
20
See Chapter 5 on Public Spending for more analysis. The infrastructure sectors included in this analysis are water and sanitation, irrigation, roads and electricity only.
21
Although the relative size of the civil service was not dissimilar to that of India or China at its peak in 1993 and thereafter, it had still done little to overcome the bureaucratic deficiencies of the Sukarno regime,
reported some 20 years earlier (World Bank, 2003).
19
11
IV
A Structural Transformation
The policies of the New Order government built upon and accelerated a process of structural
transformation. Macroeconomic stability, investment in human and physical assets, and a reduction in transaction
costs during this period transformed the livelihoods of the Indonesian people from low-value agriculture to higher-value
agriculture, from agriculture to non-agriculture, and from rural- to urban-based activities. This transformation dramatically
affected the pattern of employment and transformed Indonesia s workforce over the two decades from the early 1980s to
the early years of the new millennium. A largely agricultural and rural workforce in 1982 became increasingly off-farm and
urban: in 1982, 54.2 percent and 45.7 percent of workers were employed in agriculture and non-agriculture, respectively.
This ratio was more or less 50:50 by 1993, and tipped to 45 percent agriculture and 55 percent non-agriculture by 2002.
The services sector drove the increase in the share of non-agricultural employment, rising from 30 percent in 1980 to 34
percent in 1990, and to 43 percent in 1995 (Figure 2.2).22 Even more significant was the transformation from rural to
urban, with the urban share of non-agricultural employees more than doubling in this 20-year period (see Table 2.3 and
Annex II. 3).
Table 2.3
Employment has moved to non-agricultural sectors and urban areas over time
1982
( 000)
1993
(%)
( 000)
2002
(%)
( 000)
(%)
Agriculture
30,487
54.24
39,137
49.88
39,035
44.92
Non-agriculture
25,724
45.76
39,329
50.12
47,874
55.08
Rural
15,939
28.36
18,992
24.20
16,785
19.31
Urban
9,785
17.41
20,337
25.92
31,088
35.77
56,211
100
78,466
100
86,909
100
All sectors
Source: Susenas, 1982, 1993, 2002. Data exclude Aceh, Maluku, North Maluku and Papua.
Note: Employment is defined as self employed without help, self employed with help of householders/temporary workers, self employed with help of regular workers,
employees and family workers with age 10 years and over (definition until 1997). The definition of employment since 1998 is the same as above, but for persons 15 years and
above.
The agricultural transformation the shift from low to high productivity agriculture was the primary
driver of poverty reduction in Indonesia. Although the importance of agriculture declined from 1993 to 2002,
agricultural income still represented over 37 percent of the income source for the poor, while by 2002 the richest quintile
was obtaining less than 5 percent income from agriculture. As average land-holdings shrank by 20 percent in 20 years
(see Figure 2.2), particularly on Java,23 households and enterprises responded by improving the productivity of the staple
commodity, rice, and by diversifying out of lower-value food crops into higher-value cash crops. This was achieved both
by farmers transforming their own land, as well as by agricultural workers shifting to more productive enterprises. Between
1982 and 1992, even though the number of agricultural workers increased from 31.8 million to 41.7 million, labor
productivity increased by 26 percent because agricultural output grew much faster (5.1 percent a year). The process
accelerated between 1992 and 1996 with the agricultural workforce declining as output continued to rise, resulting in a
further increase in labor productivity of 33 percent. This rise in labor productivity played an important role in sustaining
poverty reduction over this period (see Annex II.3 Empirics of structural transformation in Indonesia).
First, improvements in the productivity of rice cultivation lay at the heart of the agricultural transformation.
Despite diversification, the planting of rice still dominated agriculture and increases in productivity were marked (Figure
2.2). A substantial increase in the wetland rice area harvested, from around 9 million ha in 1980 to almost 12 million ha
22
23
12
Labor surveys or Survei Tenaga Kerja (Sakernas), 1980, 1990 and 1995.
The average area controlled by land holding households decreased from almost 1 hectare (ha) in 1983 to 0.87 ha in 1993 and 0.81 ha in 2003 (see Figure 2.2).
in 2004, reflected investment in rural infrastructure and, more generally, the strong policy bias towards boosting rice
production. In the late 1960s, the government adopted an agricultural program known as Bimas , which brought the
benefits of modern agricultural technology to millions of farmers. The use of high-yielding seed varieties, particularly
rice, coupled with modern inputs such as fertilizers and pesticides, transformed subsistence rice farmers into productive
commercial suppliers. New programs were introduced in the 1980s that extended the benefits to other food and cash
crops. Rapid agricultural growth led to rapid reduction in total poverty. By 1984, the FAO awarded President Soeharto a
gold medal for the country s achievement in attaining self-sufficiency in rice.
Figure 2.2 Indonesia experienced a rapid structural transformation in recent decades
Land-holding size declined with increases
in the number of farm-holding households.
In the same period, there were significant

productivity gains in agriculture.
Source: Agriculture Census 1983, 1993, 2003.
Source: FAOSTAT data, 2004.

Note: Chart gives productivity gains in agriculture for top ten crops in 2004 in terms of production,
Yield = Production metric tons (Mt) / Total Land area (ha).
Labor moved out of agriculture into services,

except during the crisis.
Increasing urbanization was coupled with lower

fertility.
Source: World Development Indicators (WDI).
Source: WDI.
13
But many farmers diversified out of rice and stepped into the market for high-value cash crops.
Diversification within agriculture to higher-value crops was marked over the period from 1980 to 2004. The growth in
production of the top ten crops over the past 20 years was illustrated by the fifteen-fold rise in the production of oil palm
fruit, from 0.2 million ha in 1980 to over 3 million ha in 2004. Although much smaller harvested areas were involved,
other higher-value crops, such as fresh fruit, vegetables and spices, experienced more than a doubling of area harvested.
Second, the changing opportunities and demands on land and households created by strong economic
growth meant that many households diversified, or partially diversified, their sources of income towards
non-agricultural activities. The growing population and declining land holdings over this period, coupled with the
non-agricultural options that emerged and the large gap between labor productivity in agriculture and other sectors, led to
a change in the share of income recorded from different sources. In 1993, the average share of per capita income from
self-employment in agriculture, forestry or fisheries was 25 percent; by 2002, it has declined to 22 percent, driven by a
strong decline in the average share of income from food crops, as well as an increase in the share of non-agricultural
income. By 2002, for the first time more than half of per capita income came from non-agricultural self-employment or
non-agricultural wages (see Annex II.4).
The third component of the structural transformation was a shift from rural to urban-based activities.
Over the past 25 years, the population of Indonesia s cities has nearly trebled. This rapid increase can be attributed to all
three causes of urban growth: expansion of cities into the rural fringe (35-40 percent), natural growth (35-40 percent) and
rural-urban migration (25-30 percent). Annual urban population growth rates are estimated at 4.6 percent over the
period 1980-2000 and 3.6 percent over the last five years, resulting in an urban population of over 93 million (BPS,
2004). In absolute terms, this means that around 8 to 9 million people were taking up an urban rather than a rural
14
livelihood each year. Between 1982 and 1993, non-agricultural employment grew by 10.5 million people (6.7 million
more than would be predicted by employment growth in the time period) and, between 1993 and 2002, a further 10.7
million joined non-agricultural employment 8.6 million more than would be predicted by employment growth (see Annex
II.5 for detailed analysis).
Many of those who shifted from rural to urban employment did not actually move. Rather, population
growth, the structural shift out of agriculture, and the development of modern facilities in rural neighborhoods resulted in
villages being reclassified as urban. Thus, much of the growth of urban areas is accounted for by the reclassification of
rural areas (see Chapter 4 for further details). However, what is important is not the classification of locations, but the
change in the sources of income from rural to urban activities. Indeed, the most dramatic change by far is the increase in
urban non-agricultural employment. Over the same period, the number of non-poor workers in these jobs rose by 10.5
million, increasing the share of total employment from 22 percent in 1993 to almost 32 percent in 2002 (McCulloch,
Timmer, and Weisbrod, 2006). This long-term evidence suggests that the processes of urbanization (including
reclassification) and migration to urban areas in Indonesia may have been as important for poverty reduction as the
sectoral shift from agriculture to rural non-agriculture.
While growth and poverty reduction in the aggregate was healthy and rising, the regional picture was
more varied and, in the worst lagging regions, somewhat sobering. The bi-modality of poverty in Indonesia
stems from the success of economic growth on Java compared with the lagging growth in eastern Indonesia. In the 1960s
and 1970s, poverty was concentrated on Java in both absolute terms and in incidence. In the mid-1960s, some districts
in Central Java would have been among the most impoverished anywhere in the world (Timmer, 1975). Today, however,
this region is exporting hand-crafted furniture to demanding markets in the West and poverty had fallen to 21 percent in
2004 (Susenas, 2004). But not all of the variance across regions can be explained by lower incomes in regions with high
poverty. There are major differences in income distribution across regions and between urban and rural areas (Friedman
and Levinsohn, 2002). This variance is not so surprising in view of the great diversity seen in Indonesia s local economic
systems although the variance does suggest that factor flows are not as smooth as the absence of formal trade barriers
within the economy might indicate. More surprising is that the actual relationship between economic growth and poverty
reduction seems not to vary substantially across provinces.24 (See Chapter 3 on Understanding Poverty for details on
regional variations in poverty reduction elasticity of growth.)
And corruption was lodged in all transactions at all levels of government. Despite Indonesia s
extraordinary progress, the losses due to corruption at the highest levels of government during the New
Order regime were unprecedented worldwide. Estimates of Soeharto s personal gain amounted to US$15-35
billion,25 but a range of corrupt practices affected all transactions at all levels of government. All basic services health,
education and water on which the poor rely were, and still are, riddled with bribery, embezzlement, fraud and distortions,
radically affecting the services the poor ultimately received.
24
As a result of Friedmans careful analysis of six Susenas data sets from 1984 to 1999, we now have a clear statistical picture of the geographic variation in the relation between levels of poverty and income
and inequality. This statistical picture complements the view from the ground that was developed in 1991 and 1992 by the Harvard-Stanford Poverty Team which reported to Bappenas in September 1992
(Timmer et al, 1992).
25
Transparency International claims Soeharto stole US$15-35 billion between 1967 and 1998. Global Corruption Report (2004).
15
The Financial Crisis
Just at a time when growth was being sustained at around 7.5 percent annually and 1.1 million individuals
were escaping poverty every year,26 the Indonesian economic miracle stalled. The knock-on effects of the
Thai baht crisis in mid-1997, combined with systemic weaknesses and a series of subsequent events,27 resulted in
negative growth of 13.1 percent in 1998, plunging 25 percent of Indonesia s non-poor population back into poverty
(Ravallion and Lokshin, 2005; Strauss et al, 2004). The depreciation of the rupiah floated in July 1997 from its precrisis level of just under Rp 2,400/US$1, to as low as Rp 18,000/US$1 in January 1998, had devastating effects on both
agricultural28 and non-agricultural markets, and the purchasing power of the poor. The country s structural transformation
stopped abruptly; output fell in both industry and services; and growth in agricultural output remained stagnant but
positive as many workers returned to the agricultural sector or were absorbed into the growing informal service sector
(see Annex III.3).
Indonesia experienced such a deep and long-lasting shock as a result of the 1997-98 crisis at least in
part because its institutional development had not kept pace with its economic development. By the mid1990s, the Indonesian state was marked by a highly centralized decision-making process. While this process enabled the
development state to continue to deliver the goods in terms of schools, clinics, and local economic development, the
system was marked by little participation by citizens or organized social/political groups. There was practically no veto
power on the part of organized political groups vis- -vis the presidents power (MacIntyre, 2003). This led to the politicization
of the wider system: civil servants responded to central commands, and not a set of objective rules; the president and his
closest aides had few effective ways of gauging public sentiment; and the emerging middle-class had no legitimate
avenues to participate in civic or political affairs (Schwarz, 1994).
In many other countries in Southeast and East Asia, the 1960s through to the 1990s was not only a
period of rapid economic growth, but also a time during which states strengthened their institutions.
This is partly evidenced by the levels of corruption, as measured by Transparency International surveys over time. Indonesia
did make some progress according to these measures in addressing corruption, but it is notable that in 1980 it had (and
thereafter maintained) the lowest ranking among the Asian Tiger nations in levels of corruption (see Figure 2.3). Progress,
although evident, was painfully slow.
Figure 2.3 Corruption in Indonesia remained a potential source of weakness in the period 1980-98
Source: Transparency International, taken from Haggard, 2000.
26
The number of poor declined from 25.9 million to 22.5 million between 1993 and 1996 using the old methodology for calculating the poverty line (see Annex II.1). This means that about 1.1 million poor were
escaping poverty per year in this time period when growth rates were on average 7.5 percent annually.
27
Structural weaknesses in Indonesia s financial sector, weak institutions, a loss of government credibility and a depreciating currency causing rising US$ debt liabilities all combined in a vicious spiral to spark
capital flight. The result was a sharp decline in real incomes, triggering a breakdown in law and order in the absence of any social safety net. This stoked social unrest and subsequently led to serious ethnic
conflict and political turmoil.
28
Some agricultural producers did very well as the drop in the exchange rate meant the price of tradable goods became more expensive, and those selling tradable goods in remote regions benefited, e.g.
cashews etc.
16
In addition to high levels of corruption, when the crisis hit, the leadership had a limited reservoir of
popular support to call upon. Two factors are notable in Table 2.4. Indonesia had, by far, the longest-serving leader
of any of the countries impacted by the Asian financial crisis, as well as the most centralized decision-making process.
While the other countries impacted by the crisis Thailand, South Korea, and Malaysia had developed at least partially
inclusive representative institutions during the 1980s and 1990s, the Indonesian state continued down an authoritarian
path, with real power continuing to be concentrated in the president and his family, selected cronies, and the military.
While in better times the regime had the power to create and enforce many policies almost single-handedly, when collective
national sacrifice was necessary there were few reservoirs of goodwill to call upon from a disenfranchised citizenry.
Table 2.4
Indonesia had the longest serving leader of any country impacted by the Asian financial crisis
Thailand
South Korea
Malaysia
Indonesia
Prime Minister
President
Prime Minister
President
Chavalit Yongchaiyudh
Kim Young Sam
Mohammad Mahathir
Soeharto
Date started in office
November 1996
February 1993
July 1981
March 1966
Decision-making process
Parliamentary,
six party coalition
Presidential,
unified government
but executive-legislative
and intra-party splits
Parliamentary,
coalition government,
but UNMO dominant
Authoritarian,
highly concentrated
Leader
Source: Haggard, 2000.
Thus, while centralized leadership and rapid growth managed to mask these underlying deficiencies
during the good times, the social compact between citizens and government quickly disintegrated when
placed under severe stress. While other nations had relatively broad coalitions that could, in effect, distribute the
pain across a broad segment of society, in Indonesia a relatively isolated leadership quickly found out that it had no
effective coalition, and that it was impossible to square the circle of demanding sacrifice and compromise from both
ruling elites and ordinary citizens. Citizens, long disenfranchised, had little interest in supporting the regime during
difficult times (Bourchier and Hadiz, eds, 2003), and elites, long accustomed to making and enforcing the rules were
unable to implement the required policies, as these would have directly undermined their financial interests (Bourchier
and Hadiz, eds, 2003; MacIntyre 2003; Haggard, 2000). Thus, the regime found its policy options severely constrained,
and inadequate to the task of promoting shared sacrifice and rapid recovery.
The economic crisis hit the poor through the dramatic increase in the price of rice and the immediate
and widespread loss of jobs. While it is true that the crisis also hit the rich hard because their assets lost their value,
the impact on poor and near-poor households across the nation was devastating because they lost their jobs and suffered
disproportionately more from the inflationary increases in the price of rice and other tradable goods. Most devastating
was the 300 percent increase in the rice price over the year following the start of the Asian financial crisis, peaking in
September 1998 (see Figure 2.4).29 This was especially hard for a population allocating 20-25 percent of household
spending on this one staple. Since rice constitutes up to half the average Indonesian s energy intake, this was to have
massive consequences on levels of nutrition in the months and years that followed (Block et al, 2002).
29
The steepest rise occurred from June to September 1998.
17
Figure 2.4 Rice prices increased substantially during

the crisis...
... directly impacting the poor who are mostly net consumers
of rice, even in rural areas
The increase in the price of rice relative to other products
Proportion of households that are net consumers of rice (%)
Source: BPS.
Many of the poor and non-poor connected to the manufacturing, industrial and construction sectors lost
their jobs. In addition to experiencing this dramatic increase in the cost of staples and other commodities, urban and
non-agricultural workers particularly suffered when the job market contracted. The poverty headcount rose in urban areas
by 43 percent between 1996 and 1999, while in rural areas it increased by 32 percent (Central Bureau of Statistics). The
average gap between the standard of living of poor households and the poverty line also increased dramatically during
the crisis, indicating a widening depth in the incidence of poverty in Indonesia. The poverty severity index rose by 300
percent in urban areas over this period, compared with 84 percent in rural areas (Pritchett, Sumarto, and Suryahadi,
1999).
Table 2.5
The crisis caused construction, finance and trade sectors to contract in relative size
Share of GDP (%)
1996
1997
1998
1999
2000
Agriculture
15.42
14.88
16.90
17.13
16.63
Mining and quarrying
9.12
8.90
9.96
9.72
9.77
Manufacturing
24.71
24.84
25.33
26.11
26.38
Electricity, gas, and water
1.18
1.26
1.50
1.61
1.65
Construction
7.96
8.16
5.97
5.81
5.85
Trade, hotel, and restaurant
16.79
16.97
15.98
15.84
15.95
Transportation and communications
7.18
7.34
7.17
7.06
7.30
Financial, ownership and business
8.79
8.90
7.51
6.92
6.90
Services
8.85
8.76
9.69
9.80
9.56
Total GDP
100.00
100.00
100.00
100.00
100.00
Source: BPS, 1993 index.
The coping strategies adopted by affected households in the immediate post-crisis period resulted in
many moving backwards along the pathways that had led them out of poverty. To lessen the blow of an
urban, formal and modern sector crisis, many households affected by the crisis moved back, from urban to rural, from
formal to informal, and from modern to traditional, non-farm to farm activities a reversal of the structural transformation.
This resulted in a significant change in the shape of the labor market. Detailed data (Annex II.5) show that the rural and
18
informal economy absorbed those retrenched from urban and formal jobs. The crisis resulted in sectors such as agriculture
and manufacturing mushrooming in relative size and others, such as construction, finance and trade shrinking (see Table
2.5). The flexibility of the Indonesian labor market was a mixed blessing: while there was only a small increase in
unemployment (see Table 2.6), high inflation rates led to a drastic fall in real wages of 27 percent.
Table 2.6
Real wages were eroded by inflation during the crisis keeping unemployment surprisingly stable
1996
1997
1998
1999
2000
2001
2002
Real minimum wage (Rp)
311,444
324,204
235,127
225,756
279,416
323,650
372,851
Unemployment rate (%)
5.0
4.7
5.5
6.4
6.1
5.4
5.9
Source: World Bank Database, real wages are provided in constant 2002 prices. Unemployment figures take the Sakernas definition.
Households also cut back their spending, resulting in a significant increase in malnutrition. This had
both long- and short-term impacts on human and household assets: health, education and savings. Part of the long-term
impact of the crisis can be seen in the rising levels of malnutrition among children under five years of age. The malnutrition
rate among these children followed a downward trend until 2000. Malnutrition in Indonesia reached a low of 23 percent in
that year, but subsequently increased to almost 26 percent by 2003 (Block et al, 2002).30 The percentage of children who
were severely malnourished31 has also increased in recent years, although the reversal of the downward trend occurred
later, in 2001 (Abreu, 2005). Increasing malnutrition was also observed at the regional level, although the upward trend
was steeper in some regions than in others (see Focus on Malnutrition in Chapter 5 on Public Spending). However,
Susenas data indicate that while malnutrition continued to rise until 2003, the rate of increase has slowed and appears to
have stagnated. Other data sources32 indicate that levels of malnutrition in rural areas leveled out in 2003, and even started
to decline in urban areas in 2002-03.
During this economic contraction the government developed and extended a number of formal safety
net programs. The social safety net programs (jaringan pengaman social, or JPS), known until then for their patchy
record, were extended to help protect the chronically and transitory poor from the impacts of the crisis. Initially, these
crash programs were directed to urban areas throughout the country but were also intended to reach rural areas where
harvest failures were causing great hardship. The JPS programs had four goals: (i) to ensure the poor could obtain food
at affordable prices; (ii) to create employment; (iii) to preserve access to social services such as health and education; and
(iv) to sustain local economic activity through regional block grants and small-scale credit programs (Sumarto et al,
2001). Evidence highlights the mixed effectiveness of the various programs (SMERU, 2004). While the scholarships
program had a positive impact by keeping children in school, and the health-card program showed improved access of
the poor to public health facilities, the Raskin (subsidized rice for the poor) program saw higher levels of capture by
higher quintiles (see further discussion in Chapter 6 on Social Protection).
30
The data used in the study reported here are taken from the Core, Module and Yodium (survey of iodized salt consumption) components of the Susenas household survey. In order to ensure comparability over
time, the provinces of Aceh, Maluku and Papua, for which a complete time series is unavailable, are excluded from the analysis. See Abreu (2005). The rise in malnutrition levels seems to have been caused by
a combination of several effects, including the impact of changes in the relative prices of rice, cassava and cooking oil (a carryover from the depreciated real exchange rate after the crisis and government policy
toward rice imports), and the presence of a cohort effect resulting from underweight children born during or just after the crisis of 1998. Not all this can be attributed to the crisis. Research has also shown that
the use of infant feeding supplements as a substitute for breastfeeding, the incidence of which has increased by almost 10 percent over the past few years, is associated with a significant increase in malnutrition
levels, while the rate of utilization of Posyandu services, already in decline long before the reversal of the malnutrition trend, has no statistically significant effect on malnutrition levels.
31
Malnourished is defined as children whose weight-for-age is more than three standard deviations lower than the average for their reference group.
32
Data from Helen Keller International/Indonesia.
19
VI
Post-Crisis: Stabilization, Democracy and Decentralization
The key governmental response in Indonesia was not at the micro level, but at the macro level in the
stabilization of the economy. The government s success in reinstating macroeconomic stability and, through the
exchange rate, bringing down the relative price of rice, was critical to Indonesia rapidly reducing the poverty rate from its
crisis spike of 23.4 percent in 1999 to 18.2 percent in 2002. The fall in the relative price of rice (index of rice prices over
all food prices) from 1.43 to 1.08 over the period from September 1998 to September 2000 was a key factor driving the
decline in poverty over that period. While the poverty headcount rates have come back to pre-crisis levels, studies also
suggest there may be some lasting impacts from the crisis. Ravallion and Lokshin (2005) estimate that the headcount
index would have been about half of what it was in 2002 had the
crisis not taken place.
The crisis and recovery showed that the price of rice is
the most important determinant of poverty at the
household level in Indonesia. Macro price stability matters
to the poor (Timmer, 2004). Rice prices are important for poverty
alleviation, not only in terms of their short-term direct benefits
on the poorest quintiles but also because they play a key role in
the structural transformation of the agricultural sector and the
economy as a whole. In agriculture, low rice prices encourage
farmers to diversify crops by reducing the incentives to plant
rice. The result is a move towards crops that give the poor higher
profit margins. In Indonesia, artificially high rice prices have
slowed down the crop diversification process, as well as
discouraging investments in non-farm rural activities (Timmer,
2004).
Indonesia s structural transformation has not yet
restarted. Although output in industry and services bounced back after the crisis, employment growth
since the crisis has been much slower than the pre-crisis rates. The structural transformation by which workers
moved out of agriculture towards more productive employment in industry and services appears to still be on hold, with
the share of workers in agriculture barely changing between 1999 and 2004 (see Annex III.3).
Although the economic crisis was triggered by externalities, it exposed and unraveled the inherent
weaknesses of the institutional and political structures in the country and resulted in structural change
unprecedented in Indonesia s history. In early 1997, although Indonesia had prospered to a level commensurate
with its neighbors and sustained growth rates over 7 percent, many argued that this success was fragile because the
institutions were fragile (Hofman et al, 2004). When the economic crisis hit in 1997, the collapse of the rupiah caused the
withdrawal of foreign investment, the loss of tens of thousands of jobs overnight, and an increase in the poverty headcount
to almost 25 percent. An economic crisis became a political crisis that would transform one of the largest authoritarian
political systems in the world. The 32-year economically powerful but increasingly oppressive rule of Soeharto ended just
eight months later. The enormity of the transition that followed the exchange rate collapse and the subsequent progress in
the eight years of reformasi has been remarkable: the constitutional amendment making way for democratic legislative
elections in 2001, the removal of the military from parliament, and the country s first ever direct presidential election in
2004. Despite inherent weaknesses (see Chapter 7 on Government) by and large the households that emerged from
poverty following the crisis did so in a democratic country with a new openness and freedom for all.
20
Box 2.1 Indonesia s elections and the consolidation of democracy

Before 1999, Indonesians went to the polls every five years to elect members of the national parliament (DPR) and the People s Consultative Assembly
(MPR), the latter of which in turn elected a president and vice-president. This was not accompanied with any expectation that by casting a vote Indonesians
could effect any real change in the way their country was governed. The system was not designed to allow for any real opposition to President Soeharto
or the ruling Golkar party. Opposition parties were interfered with and co-opted by the government; government policies and programs could be criticized
only warily and indirectly; and, the president could not be criticized at all. However, elections did matter to Soeharto: they helped to mobilize the public to
give the appearance of support for government policies, and also created a channel for a limited form of public participation, albeit highly choreographed,
in Indonesia s political process.
The 1999 general election was the first free-and-fair election in Indonesia since the early 1970s. With the euphoria of reformasi and the convening of a
new parliament, Abdurrahman Wahid, a prominent Muslim cleric, was elected president by the MPR in October 1999, with Megawati Soekarnoputri as his
vice-president. President Wahid continued to pursue the path of democratization. However, in 2000, under growing pressure from parliamentarians and
student demonstrators, and against of background of continuing economic strife and violent clashes in Maluku, Poso and Central Kalimantan, President
Wahid issued a presidential decree allowing his vice-president to take over more responsibility in an effort to improve the day-to-day management of the
government. But this failed to mitigate the mounting pressure on the president. Wahid s position was further undermined by allegations of corruption.
Finally, in 2001, parliament used these allegations of corruption as a means to bring about a vote of no confidence in Wahid, engineering Wahid s ouster
and his replacement as president by Megawati in July 2001.
By 2003, the pace of democratization and decentralization quickened with the passage of important electoral and decentralization laws, conceived in the
wake of Soeharto s resignation in 1998. In April 2004, Indonesians participated in their second free-and-fair general election to elect representatives to
a new parliament. As mandated by the decentralization law, local governments would also be elected instead of appointed by the central government. Later,
in July, the world witnessed the largest one-day election ever as Indonesians participated in the country s first direct presidential election. Susilo Bambang
Yudhoyono, a former general and minister for security in Megawati s cabinet, won the presidency by a landslide and was inaugurated in October 2004.
This rapid political change was accompanied by one of the biggest big-bang decentralizations in the
world. By mid-1998, the growing discontent in the regions drew further attention to the imbalance of power between
Jakarta and the thousands of islands off-Java. With secessionist sentiments already strong in several provinces, the new
post-Soeharto leaders feared provincial power would lead to national disintegration. In order to defuse the growing
tensions, the ruling administration decentralized power down to the kabupaten level, empowering 292 districts in 1999.33
Although drafted in haste and lacking important details, the Regional Autonomy and the Fiscal Balancing Laws came into
effect in 2001, and were subsequently amended in 2004.34 Together with the countrys rapid democratization, this devolution
of responsibility for basic services resulted in a much more complex story of economic recovery and the impacts on the
poor than that told in other East Asian states.
Although national policy-makers and local decision-makers are now accountable in a way they have
never been before, decentralization poses a challenge for economic management and poverty reduction.
The decentralization legislation also established new social and political space for civil society, mandating that village
governance support participation and autonomy. This reform has created enormous opportunities for local initiatives in
tackling local problems, and designing development strategies that best meet local needs but it has also led to policy
uncertainty. Many districts lacked capacity and revenue and, consequently, this decentralization in theory political,
fiscal and administrative was in practice fraught with inconsistencies and blockages. Much of the decentralization business
33
34
The number of districts subsequently increased to 336 in 2001 and 434 by 2004. Source: Bastian Zaini, World Bank Decentralization Team.
Regional Autonomy Law No. 22/1999 and the Fiscal Balancing Law No. 25/1999 which was amended by Fiscal Balancing Law No. 33/2004.
21
was left undone: districts inherited responsibility for the delivery of health and education services without clarity on roles
for some of the key functions of government; fiscal decentralization created an imbalance between resource-rich and
resource-poor districts; and the immaturity of Indonesia s civil society and local governments (now responsible for key
poverty reduction functions) pose new challenges for the country.
VII Conclusion: Learning from History to Support Poverty Reduction

Indonesia s history sheds much light on what has worked to reduce poverty in the past. These lessons
can be used when looking to a strategy for poverty reduction going forward. At the same time, it is important
to be highly cognizant of the fact that much has changed in Indonesia, both structurally and institutionally. In particular,
Indonesia now has a deepening and strengthening democratic process. In addition, development processes today must
work in the context of decentralization. Moreover, the structural nature of the Indonesian economy, the global economy
and Indonesia s relationship to it, have all changed. Related, to some extent the nature of poverty is also somewhat
different today than in the past, as will be discussed in the next chapter. Nevertheless, the history of poverty reduction in
Indonesia provides important guidance for poverty reduction efforts going forward and the lessons motivate the framework
of this report.
First, Indonesia s record shows what a driving force for poverty reduction growth can be when it works
for the poor. Indonesia s success is in poverty reduction over the past three decades was because: (i) it engendered so
much growth; and (ii) it made this growth work for the poor . Critical in this regard, as discussed, was a macroeconomic
and liberalized trade and exchange framework that allowed the poor to benefit from world prices while also benefiting from
macrostability (i.e. generally stable prices and low inflation). Furthermore, the poor in Indonesia benefited greatly from
measures that specifically ensured that the sectors that the poor worked in were engines for the growth process. In
agriculture this was enhanced through investment in rural infrastructure and agricultural technology (in the form of more
productive rice varieties) and exchange and trade policies that fostered higher incomes from cash crops. In manufactured
exports this was enhanced again through a competitive exchange rate and
liberal labor market policies that allowed Indonesia to hire labor to
manufacture products that could be competitively exported. An issue of
concern going forward is that Indonesia s overall growth rate has slowed.
Moreover, the extent to which the growth rate is pro-poor has deteriorated
after the crisis. What must be done to reignite growth that benefits the poor
going forward, learning lessons from the past?
Second, Indonesia s record shows that wisely channeling
enhanced spending into efforts and programs that benefit the
poor is key to poverty reduction. History shows that this is one of the
ways through which growth was made more pro-poor: increased public
revenues from growth were channeled into public spending that supported
service delivery, as well as being a handmaiden to the growth process
itself. One important example that Indonesia presents to the rest of the
world is the way in which it used its oil windfall in the 1970s and 1980s to
support service delivery and growth itself by investing in things that mattered
to the poor: rural infrastructure, including roads; and public services,
especially in rural areas, including health and education facilities. Big money
was spent to expand the networks of road, health and education
infrastructure. Now the challenge may shift more towards how to maintain
22
this infrastructure and enhance the software to deliver services. Here, Indonesia faces a challenge. Indicators of service
delivery as evidenced by outcome indicators such as maternal mortality, junior secondary school enrollment, and even
malnutrition are lagging. But Indonesia also is fortuitous in being presented with a fiscal opportunity: the past couple of
years have brought with them another oil windfall. How can Indonesia make enhanced public spending work better for the
poor to foster growth and service delivery?
Third, Indonesia s experience with the shock of the economic crisis shows more than ever the need to
make social protection work for the poor. The poor and near-poor in Indonesia are particularly vulnerable to price
and other shocks given the income distribution profile in Indonesia. (The nature of this vulnerability is explored more fully
in the Chapter 4 on Growth and in Chapter 6 on Social Protection.) Nothing exemplified this vulnerability more than the
economic crisis in 1998. Several lessons are to be learned from that experience. For one, poverty rates in Indonesia were
able to recover much faster than some neighboring Asian countries in large part due to macroeconomic and factor market
policies. Macroeconomic stabilization efforts that were able to reduce inflation and the price of the main food commodity
were critical to protecting the poor. So too were labor markets and wages that were flexible enough at that time to prevent
further unemployment. Specific social protection programs were rapidly put in place after the crisis. Some may argue that
they were too little, too late, although there is evidence that they helped mitigate some of the social costs. Moreover, since
the crisis, many of these programs have shut down. The lesson, if any, is that given the high vulnerability of the poor, it
makes sense for the government not only to put in a well-organized social protection system to protect the poor from the
annual shocks they tend to suffer, but also to be ready for any future crises, whether these be economic, natural disasters,
or otherwise. So a key question is: how should Indonesia go forward in making social protection work for the poor?
Fourth, Indonesia s past experience shows that it has to make government work for the poor. The historical
record here is mixed at best. On one hand, history reinforces the point that a strong central government fully committed
to policies that reduce poverty and with the technical capacity to formulate and implement these policies can be key to
poverty reduction. On the other hand, history also tells the story of how weak accountability and institutional capacity to
deliver services and not investing in such capacity can itself undermine development. Indeed, poor governance at the
very highest levels of government, as well as at lower levels of government,
can cause a collapse in the entire development effort, as witnessed during
the political crisis in 1998. Moreover, weak accountability, capacity, and
institutions at all levels of government make the task of delivering on
poverty reduction whether it be through growth, spending, social
protection-extremely difficult. In this regard, history was not so beneficial
for Indonesia. Far from being left with strong and accountable institutions
and an effective civil service, after the crisis Indonesia has had to contend
with strengthening these systems for poverty reduction. Decentralization
certainly enhances the opportunity to improve governments ability to deliver
on poverty reduction by moving decision-making power closer to citizens
and allowing efforts to be tailored to regional issues. But, at least initially,
decentralization also makes the task more complex and Indonesia is still
working on sorting out the specifics of its decentralization framework. So,
a key question for poverty reduction then is: how can Indonesia better
make government work for the poor?
23
Chapter
Understanding Poverty in
Indonesia
24
Salient Facts about Poverty in Indonesia
Progress in poverty reduction: a positive story, with some caveats

Poverty returned to its pre-crisis level in 2004, but rose again in 2005-06. Despite the huge setback from the
1997 Asian financial crisis, Indonesia has made significant progress in reducing poverty.35 In 1999, at the peak of the
crisis, 23.4 percent of the population had an expenditure level insufficient to support their basic needs. Just five years
later, the incidence of poverty had dropped to 16.7 percent, lifting 7.6 million individuals out of poverty over the period.
This poverty level was even lower than the 1996 pre-crisis level of 17.6 percent. In addition to the improvement in the
poverty headcount figures, since 2002 the poverty gap and the poverty severity index have returned to pre-crisis levels
and even surpassed these in some regions. A historical geographic comparison also shows significant poverty reductions
across all of the six regions used in this assessment, namely Java/Bali, Sumatra, Kalimantan, Nusa Tenggara/Maluku and
Papua. By 2004, all regions returned to, or improved upon, their 1996 poverty levels, the sole exception being Sumatra,
which in 2004 was still 2 percentage points above its 1996 pre-crisis level of 15.5 percent (see Table 3.1). Despite this
progress in reducing poverty, largely as a result of increases in th eprice of rice in 2005-06, the poverty rate subsequently
rose to 17.75 percent in 2006 the first increase since the 1997 crisis.
Table 3.1
Poverty has declined to pre-crisis levels in most regions

(Depth and severity of poverty in Indonesia in 1996 and 2004 by region)
2004
1996
Poverty measure
Sumatra
Java/Bali
Kalimantan
Sulawesi
Headcount
15.5
16.3
15.0
19.2
Poverty Gap
2.5
2.8
2.3
3.7
Poverty Gap Square
0.6
0.7
0.5
Headcount
17.5
15.7
Poverty Gap
3.1
2.7
Poverty Gap Square
0.8
0.7
NT/Maluku
Papua
Indonesia
37.5
42.3
17.5
7.9
11.7
3.1
1.0
2.4
4.4
0.9
11.0
16.7
26.1
38.7
16.7
1.6
2.8
5.8
2.4
2.9
0.4
0.7
2.0
0.2
0.8
Source: Susenas 1996 and 2004.
However, the rate of poverty reducing growth has slowed in recent years. In the immediate aftermath of the
crisis in 1999-2002, economic growth struggled to return to pre-crisis levels, averaging a lackluster 3.8 percent. More
worrying still, the annual rate of pro-poor growth (defined as the mean growth rate for the poor)36 was only 2.7 percent.
Using the relative approach (see Annex III.4 for methodology), the impressive decrease in poverty in 1999-2002 and the
more modest decline in 2002-04 encompass a period of economic growth that cannot be described as pro-poor, because
the expenditure of the poor grew at a slower rate than that of the non-poor (see Figure 3.1).37 Instead, the gap between the
rate of pro-poor growth and rate of growth for the rest (non-poor and poor) widened in 2002-04: the annual average rate
of pro-poor growth was merely 0.3 percent, far below the 4.1 percent growth rate at the mean.38
35
The poor are defined here on a minimal destitution standard, otherwise known as the expenditure poor .
The definition of pro-poor growth comes from Ravallion and Chen (2003), where the authors take the rate of pro-poor growth as the mean growth rate for the poor below the poverty line.
This is true despite the higher poverty elasticity of growth for this period shown in Table 2.2 (in Chapter 2). The higher elasticity results from the clustering of the income distribution around the poverty line,
so that a very small change in mean income results in a significant change in poverty. However, it is still true that the rate of growth of the poor from 1999-2002 is significantly lower than that of the mean.
38
See Chapter 4 on Growth for more detailed discussion.
36
37
25
Data on post-crisis growth and the mean rate of growth for the poor now give cause for concern. First, the
average annual growth rate over the past five years remains lower than in the decades prior to the crisis. In the period
leading up to the crisis in 1980-96, annual GDP growth rates were around 7.0 percent, but these declined to an average
of 4.6 percent in the post-crisis period from 2000-04. This slower growth has clearly resulted in slower poverty reduction.
Second, there is concern over why economic growth, after the crisis recovery period, is no longer as pro-poor as it was
before the crisis. After a low average annual pro-poor growth rate of 2.7 percent in 1999-2002, this declined even further
to a mere 0.3 percent in 2002-04 (see Figure 3.1). Third, rural areas have increasingly lagged behind urban areas over
this period and the mean growth rate for the poor has been considerably higher in urban areas than in rural areas. In 200204, the average annual urban growth rate for the poor was 1.4 percent, fully ten times that experienced in rural areas
(Alatas and Tuhiman, 2005).
Figure 3.1 Growth has become less pro-poor over time39
Source: Susenas.
The near-poor and vulnerability: significant aspects of the poverty story

As many as 42 percent of all Indonesians live between the US$1- and US$2-a-day poverty lines. This is
one of the remarkable and defining aspects of the poverty story in Indonesia. Income distribution (see Figure 3.2) shows
how densely clustered the population is around three poverty lines: US$1-a-day; US$2-a-day; and the national poverty
line (about US$1.55-a-day). (Box 3.1 explains the significance and methodology of these poverty lines, and the poverty
measures derived from them.) The level of extreme poverty (defined as below US$1-a-day) is relatively low, even by
regional standards, at 7.4 percent. However, the proportion of the population earning less than US$2-a-day stands at a
huge 49.0 percent.40
39
Growth incidence curves show how much growth has impacted the poor. The more negatively sloping the growth incidence curve, the more the poor benefit from growth. The more upward sloping the curve,
the more growth accrues to the rich. In the graphs shown in the period 1980-1996, growth was beneficial to the very poor, as well as the very rich. During the crisis, the rich lost out more than the poor.
Worryingly, however, the period 2002-2004 is far more pro-rich in terms of the benefits from growth than any of the previous periods. The line passing through the middle of each graph shows the growth date
of mean per capita expenditures (which are: 6.42 percent in 1980-96, -4.46 percent in 1996-99, 3.8 percent in 1999-2002 and 4.1 percent in 2002-04.
40
The US$2 PPP poverty statistics for Indonesia and other East Asia Pacific countries cited in this section are taken from the April 2005 East Asia Update.
26
Chapter 3: Understanding Poverty in Indonesia
Box 3.1 Poverty definitions and measures

Poverty headcount index (Po): This is the share of the population whose consumption is below the poverty line. The
headcount index, sometimes referred to as the poverty incidence , is the most popular poverty measure. However, this measure
fails to differentiate between sub-groups of the poor and does not indicate the extent of poverty. It remains unchanged even if
a poor person becomes poorer or better off, provided that they remain below the poverty line. Therefore, in order to develop
a comprehensive understanding of poverty, it is important to complement the headcount index with the other two poverty
measures of Foster, Green and Thorbecke (FGT).
Poverty gap index (P1): The mean aggregate consumption shortfall relative to the poverty line across the whole population,
with a zero value assigned to those above the poverty line. The poverty gap can provide an indication of how many resources
would be needed to alleviate poverty through cash transfers perfectly targeted to the poor. This index better describes the
depth of the poverty but does not indicate the severity of poverty. However, it does not change if a transfer is made from a poor
person to someone who is even poorer.
Poverty severity index (P2): This measure gives more weight to the very poor by taking the square of the distance from
poverty line. It is calculated by squaring the relative shortfall of per capita consumption to the poverty line and then averaging
across population while assigning zero values to those above the poverty line. When a transfer is made from a poor person to
someone who is poorer, this registers a decrease in aggregate poverty.
US$1 and US$2 PPP per day poverty measures: To compare poverty across countries, the World Bank uses estimates
of consumption converted into US dollars using purchasing power parity (PPP) rates rather than exchange rates. The PPP
exchange rate shows the numbers of units of a country s currency needed to buy in that country the same amount of goods
and services that US$1 would buy in the US. These exchange rates are computed based on prices and quantities for each
country collected in benchmark surveys, which are usually undertaken every five years. Chen and Ravallion (2001) present an
update on world poverty using a US$1-a-day poverty line. According to their calculations, in 1993 the US$1-a-day PPP
poverty line was equivalent to Rp 20,811-a-month (US$2). The PPP poverty lines are adjusted over time by relative rates of
inflation, using consumer price index (CPI) data. So in 2006, the US$1 PPP poverty line is equivalent to Rp 97,218 per
person per month while the US$2 PPP poverty line is equivalent to Rp 194,439 per person per month.
While Indonesia has recently crossed the threshold to become a middle-income country, the share of
its population under US$2-a-day is closer to lower-income countries in the region. At 49.0 percent, the
population under US$2-a-day is high relative to the comparable average for the East Asia region (including China), at 34
percent. By this measure, the share of poor under US$2-a-day in Indonesia is closer to that found in the lower-income
East Asian countries. The average for Cambodia, Laos, Papua New Guinea and Vietnam stands at 53.8 percent.
Indonesia thus has an extremely high share of near-poor . In the Indonesian context, this important characteristic
of the poverty story is reflected by frequent reference to the near-poor . This category is defined in this report as those
who live above the national poverty line (about US$1.55-a-day) but below the 40th percentile of per capita expenditure.
Analyzing the profile of the near-poor reveals that they have very similar characteristics to the poor.41 This means that
policies relevant to moving the poor above the national poverty line will be just as relevant to moving the near-poor above
the US$2-a-day poverty line.
41
See Section II in this chapter.
27
Figure 3.2 Almost half of all Indonesians live on less than US$2-a-day
(Population density and log per capita expenditure)
Source: Susenas panel data 2006a and WB staff estimates.
The high proportion of near-poor is also reflected in Indonesia s high vulnerability to poverty and
churning . Chaudhuri, Jalan and Suryahadi (2002) show that the ex ante risk of a household falling below the poverty
line, if currently not poor, is high in Indonesia. Using data collected shortly after the 1998 crisis, they found that the
proportion of the population facing a significant risk of poverty was considerably greater than the proportion already
observed to be poor. While only 22 percent of the population was poor in 1998, Chaudhuri et al. estimate that 45 percent
were vulnerable to poverty. Household income and consumption data from more recent years show a very high degree of
churning , or movements of given households in and out of poverty over time. For example, over 38 percent of poor
households in 2004 were not poor in 2003 (see Table 6.1 in Chapter 6 on Social Protection).
28
Recent Developments
Why did poverty increase from 2005 to 2006?
On September 1 2006, the Central Bureau of Statistics (BPS) announced that

the poverty rate in Indonesia had risen to 17.75 percent in March 2006, up
from 16.0 percent in February 2005. In what was the first recorded increase in
poverty incidence since the economic crisis, this meant that an additional 4
million people fell into poverty during the period. How did such an increase
come about? Economic growth during the period was 5.2 percent, not very
different from the previous two years (5.6 percent and 4.4 percent, respectively)
during which poverty fell continuously. If the increase in poverty cannot be
explained through changes in growth patterns, other events must have affected
relative income distribution in such a way as to hurt the poor more than other
groups.
The 2005-06 period witnessed several events likely to have impacted poverty.
The fuel subsidy reduction in October 2005 increased fuel prices by 114 percent
overall and tripled the price of kerosene. Meanwhile, rice prices rose dramatically
(33 percent) between February 2005 and March 2006. While the increase in
fuel prices would have had a limited direct impact on the poor, the large increase
in kerosene and rice prices undoubtedly had a larger negative effect. Conversely,
the unconditional cash transfer (UCT) program, which provided cash targeted
to 19.2 million poor and near-poor households, was introduced to soften the
blow. This is the worlds largest ever cash transfer program targeting some 34
percent of the population well in excess of the number of poor households in
the population with each beneficiary family receiving about US$11 per month.
World Bank analysis indicates the impact of the fuel price increase is unlikely
to have been the main driver behind the poverty increase. It is estimated that in
the absence of the UCT, the fuel price increase would have reduced the welfare
of the poor and near-poor by some 5 percent and increased poverty substantially
because of the tripling of kerosene price and the overall inflationary impact.
However, the cash transfer of around 14 percent of poverty line would have
more than offset the negative impact of the fuel price increase for those among
the poor who received it. The overall impact depends on the targeting
performance of the cash transfer program. Simulations of the combined effect
of the UCT and fuel price increase point to a positive net income gain for the
poorest two deciles (see Figure 6.1d in Chapter 6 on Social Protection). The
preliminary analysis of the UCT targeting (see Figure 6.3) suggests that the
actual targeting was between Scenario 2 (slight mistargeting) and Scenario 3
(greater mistargeting). Both scenarios pointed towards net income gains for
the poor. Even assuming significant mistargeting, the lower deciles were, on
average, more than compensated for the fuel price increases.42
This means, then, that the increase in the rice price by 33 percent between
February 2005 and March 2006 mostly due to shortages as a result of the ban
on rice imports is the main factor behind the increase in poverty rates. Rice
plays an important role, as it makes up about 25 percent of the consumption
basket of the poor. Moreover, three-fourths of the poor are net consumers of
rice. Thus rice price increases hurt far more people than they help. Moreover,
the World Bank estimates that most of the increase in poverty is attributable to
the rice price increase.
With rice prices still on an upward trend and the UCT program drawing to an
end, poverty rates may well increase again next year. The loss of the UCT will
be a major income loss for the poor who receive it, while the conditional cash
transfer (CCT) program to be piloted next year will be too small to offset the
impact of the ending of the UCT. The immediate impact of plans to upscale the
national community empowerment program on poverty will also be limited. In
order to reduce poverty in the short run the government might consider both
(i) continuing with a scaled-down version of the UCT program, and (ii) adopting
a rice policy that promotes stabilized rice prices at international levels.
42
Although on average the scenarios point to a positive net income gain, the actual impact depends on the level of expenditure of the recipients. When Susenas 2006 data become available, we will need to
conduct further studies to fully decompose the change in poverty incidence.
29
Recent Developments
Counting the poor in Indonesia
In estimating poverty incidence, we require data on welfare measures and

poverty line estimates. In Indonesia, the welfare measure used is that of per
capita consumption. Households having a per capita consumption below the
poverty line are considered poor. The poverty line is usually anchored to the
notion of a minimum calorific intake to fulfill nutritional requirement, usually
set at 2,100 calories.
Household consumption is obtained from the Susenas consumption survey.
This module collects data on quantities and expenditure for 218 food items,
and expenditure for 109 non-food items. Previously, official poverty estimates
could only be calculated once every three years, when the consumption module
was administered to around 60,000 households across the country. Since the
start of the Susenas panel survey, which administers the consumption module
to a nationally representative sample of around 10,000 households, national
poverty statistics have been updated yearly.
The most difficult issue in counting the poor is deciding at which level to set
the poverty line, and how to update it over time so that it reflects the same
level of welfare. The poverty line used by BPS has two components: one for
food and one for non-food items. Both the food and non-food components of
the poverty line boil down to costing a bundle of basic minimum needs. This
is done for each province separately, by urban and rural areas, to account for
differences in consumption patterns.
There are two main methods in computing a poverty line in developing
countries: the food energy intake and the cost of basic needs methods. BPS
uses an adaptation of the food energy intake method to obtain the food poverty
line. The chart below summarizes the method. In setting its national poverty
line, BPS proceeded as follows: (i) It defined the reference group as the 20
percent of the population with a per capita consumption just above the first
round estimate of the poverty line (denoted by z in the figure). The first round
estimate of the poverty line is obtained by multiplying last year s poverty line
by the rate of inflation over the previous year. (ii) Average expenditure on, and
calories obtained from, 52 of the most common food items are calculated for
each household in a reference group. Dividing these two and averaging over
all households in the reference group provides us with an average cost per
calorie. This is multiplied by 2,100 to obtain the food poverty line (denoted by
FPL in the chart). (iii) The non-food component of the poverty line is obtained
by multiplying the sum of actual expenditure on 32 non-food items in the
reference group by a correction factor. The factors are unique for each
commodity and remain the same from one year to the next.
Consumption of 2,100 calories per person per day is a standard

norm for caloric needs.
The method used by BPS contains several flaws. First, the bundle changes
over time and across regions, both for the food and non-food poverty line. It is
therefore not guaranteed that the bundle of goods reflects the same welfare
level. Food energy intake does not take into consideration changes in tastes,
activity levels, relative prices or publicly provided goods that can affect the
relationship between food energy intake and total consumption expenditure.
For example, taste differences could result in a higher food poverty line for
regions where the poor tend to consume high cost per calorie foodstuffs (Wodon,
1997; Ravallion and Lokshin, 2003). Second, the bundle of goods that is used
for the poverty line corresponds more to the typical expenditure of those above
the poverty line, not of the poor. It would be better to use a reference group that
reflects the actual consumption patterns of the poor. Poverty lines are quite
sensitive to the choice of reference group (Pradhan et al, 2001) and using
wealthier reference groups generally results in a higher poverty lines. Third,
the sample size of the 10,000 Susenas panel makes setting the poverty line
separately for each province very sensitive to outliers. On average, there are
only around 33 households in the reference group for each region. Since not
every household consumes every item included in the bundle, the poverty line
may be based on actual consumption of just a few households.
BPS could consider adopting a cost of basic needs method to measure poverty.
This method fixes the bundle of goods across regions and time to set the
poverty line. Only the prices are allowed to change to correct for regional price
differences and inflation. The advantage of this method is that it guarantees
that the poverty line is always sufficient to purchase the same bundle of goods
and, thus, is sufficient to attain the same level of welfare. The bundle of goods
can be updated periodically to reflect changes in consumption patterns, but
should remain the same when doing comparisons of poverty over time.
30
Non-income poverty: indicators show persisting problems

Indonesia has achieved remarkable progress in key human development indicators. Chapter 2 described
the history of how Indonesia adopted policies that allowed the country to make remarkable progress in a number of key
aspects of human development. Progress in some key indicators was significant and noteworthy.
Improved educational attainment, particularly in primary schools. A child born in 1940 had only a 60
percent chance of attending any school, a 40 percent chance of completing primary school, and a mere a 15 percent
chance of completing junior secondary school. In contrast, over 90 percent of children born in 1980 completed
primary school and about 60 percent completed junior secondary school. Figure 3.3.a clearly shows younger cohorts
are better educated. The mean educational attainment of the 1980 birth cohort is about 9.5 years of schooling, 2.4
times higher than the 1940 cohort. In 1996, around one-third of the population had not completed primary school.
Eight years on, this proportion has fallen only 24 percent. The average number of years of schooling in the workingage population increased by 23 percent during the same eight-year period, from 6.7 to 8.25 years.
Improved basic healthcare coverage, particularly in birth attendance and immunization. In 1989, the
government began hiring trained midwives and placing them in rural villages an initiative subsequently formalized
as the Village Midwife (bidan di desa) Program. By the end of 1994, over 50,000 midwives had been placed in
villages (Parker and Roestam, 2002). Among communities surveyed in the Indonesian Family Life Survey (IFLS), the
share of communities with a midwife increased from less than 10 percent in 1993 to almost 46 percent by 1997
(Frankenberg et al, 2004). By 2002, almost half of all births in rural areas were attended by a village midwife (Badan
Pusat Statistik and ORC Macro, 2003). A focus on basic healthcare is also reflected in improvements in vaccination
rates. While fewer than 20 percent of 12 to 23 month old children had received their first DPT vaccination in 1980,
close to 90 percent were covered by 2004. However, despite this progress, polio is recurring in some areas.
Dramatic reductions in child mortality. As Figure 3.4 shows, Indonesia s under-five child mortality rate was
extremely high at over 200 per thousand in 1960, more than double the rate in the Philippines and Thailand at that
time. But, by 2005, Indonesia had brought its child mortality rate down to less than 50. While still slightly higher
than its regional peers, this was one of the fastest declines in the region.
However, there are other human development indicators in which Indonesia is seriously lagging and
where special attention is required. Despite solid policy-driven achievements in some areas, Indonesia will need to
focus intense efforts on other indicators in which the country lags behind if it is to achieve progress. These include the
transition rate from primary to secondary school, child malnutrition, maternal mortality, and access to safe water and
sanitation. (See Focus Boxes in Chapter 5 on Public Spending, which discuss several of these poor outcomes in more
detail, including specific policy interventions that could help Indonesia succeed in addressing them.)
43
Transition rates from primary to secondary school are low, particularly among the poor. Since the
economic crisis, improvements in net junior secondary school enrollment rates have slowed. Net enrollment increased
by 16 percent from 1992 to 1997 (Figure 3.3.b), but in the five years following 1997 net enrollment increased by only
4 percent.43 Even worse, many children, particularly poor children, are dropping out after primary school. Among 16
to 18 year olds from the poorest quintile, only 55 percent completed junior secondary school, compared with 89
percent from the richest quintile from the same cohort. A closer look at the data reveals that the rise in junior
secondary school net enrollment since 2001 is due to more children who have completed class level one in junior
Recently, net enrollment has picked up again and an extrapolation of recent trends indicates that it is now growing at about 8 percent for the current five-year period (using data from the 2002-04 period).
31
secondary school continuing on to attend levels two and three. Therefore, the apparent improvement is not in fact
due to an improved transition rate between primary and junior secondary school.
Figure 3.3 Trends in educational attainment in Indonesia
a. Younger cohorts are better educated
b. Net junior secondary school enrollment shows a steady

increase
c. But average educational attainment can hide

differences by region or expenditure group
d. And the poor struggle to make transition to secondary
Proportion of current cohort that reaches to each grade
Malnutrition rates are worryingly high and have only recently stopped rising. Malnutrition is high for a
country with Indonesias overall level of health, with more than one-quarter of all children under the age of five being
malnourished using the weight-for-age measure (see Figure 3.5). By comparison, the malnutrition rate is 18 percent
in Malaysia. While Indonesia s under-five mortality rate is about one-third of Cambodia s, Indonesia s percentage of
malnourished children is only about 20 percent lower than the Cambodian level (Susenas, 2004). Moreover, in view
of the steady improvement in living standards in Indonesia, it is all the more surprising that child malnutrition rates
have increased slightly in recent years. The malnutrition rate among children under the age of five followed a downward
trend similar to that in poverty levels until 2000, reaching a low of 23 percent in that year, but then started to rise.
Although there appears to have since been a levelling-out of the increase, malnutrition rose to almost 26 percent by
2003.44 Data sources other than Susenas indicate that malnutrition may have started to decline once again.
44
The data used in this study are taken from the Core, Module and Yodium (survey of ionized salt consumption) components of the Susenas household survey. In order to ensure comparability over time, the
provinces of Aceh, Maluku and Papua, for which a complete time series is unavailable, are excluded from the analysis.
32
Indonesia s maternal mortality rate is high, much higher than other comparable countries in the
region. The risk of maternal death during, or shortly after, delivery is significant in Indonesia. The national maternal
mortality ratio (MMR) is 307 per 100,000 live births.45 This means that a woman who bears four children has on
average a probability of 1.2 percent of dying as a result of her pregnancies. Maternal deaths account for an estimated
21 percent of all female deaths of reproductive age (Djaja, 2000; National Household Health Survey, 1995).46 This
MMR is far higher than other comparable countries in the region, such as the Philippines, at 172 based on data from
1997, and Malaysia, at only 20 (UNDP and Government of Malaysia, 2005).
Figure 3.4 Under-five mortality since 1960:
the fastest decline in the region
Figure 3.5
But today Indonesia has a high and stagnating child

malnutrition rate
(Trends in child malnutrition in Indonesia)
Source: Making Services Work for the Poor in Indonesia, World Bank 2005.
Under-five mortality data comes from WHO/UNICEF Joint Reporting Forum on
Vaccine Preventable Diseases based on WHO, UNICEF, and IDHS data (www.childinfo.org)
Source: Susenas 2004 and Helen Keller International, Indonesia report different malnutrition figures.
**Note: Child malnutrition is defined as the proportion of children under th age of five who fall
minus two standard deviations below the median weight -for-age of the reference population.
Access to safe water is low, especially among the poor. Access to safe water in rural areas has increased by
17 percent over the past decade, but the poorest quintile still has disturbingly low levels of access. Overall data show
a significant difference between rural and urban access to safe water, access in rural areas being estimated at 69
percent and in urban areas at 90 percent. But for the lowest quintile, access in rural areas is only 48 percent, against
78 percent in urban areas. Data for Susenas 1994 and 2004 indicate that the rate of increase in access has been even
across quintiles in rural areas and is leveling out in urban areas (as would be expected given the already higher rates
of access).
There is a major failure in sanitation adequacy, particularly among the poor. The coverage of sanitation
services in Indonesia is undoubtedly the worst in the region, with less than 1 percent of all Indonesians provided
with piped sewerage services. Susenas data indicate that 80 percent of the rural poor and 59 percent of the urban
poor have no access to adequate sanitation. The costs of inadequate sanitation are substantial and the health and
environmental costs in cities are especially high. Considering those 70 million people who use and share public and
other non-private forms of sanitation facilities, it is estimated that the cost of poor sanitation (just measured by
waiting and walking time) is in the order of Rp 530,000 (US$56) per person per year, or almost 2.6 percent of GDP.
45
Estimate based on the 2002 Indonesian Demographic and Health Survey (DHS) reported deaths over the period 1998 to 2002.
In the 10th revision of the International Classification of Diseases maternal death is defined as the death of a woman while pregnant or within 42 days of termination of pregnancy, irrespective of the duration
and site of the pregnancy, from any cause related to or aggravated by the pregnancy or its management but not from accidental or incidental causes .
46
33
Multi-dimensional poverty: a multi-headed beast

If multi-dimensional poverty is defined as expenditure/income poverty and/or deprivation in terms of
one of the non-income dimensions of poverty discussed above, around half of all Indonesians are poor.
When poverty is expanded to include other dimensions of human well-being-adequate consumption, education, health,
and access to basic infrastructure then poverty remains a major issue in Indonesia. Almost half of all Indonesians are
currently experiencing at least one type of poverty. The true extent of poverty is even more striking in rural areas, with twothirds of the population experiencing at least one type of deprivation, while almost 30 percent of urban households are
deprived of at least one kind of basic need. Although there is a tendency for the poor to experience several forms of
deprivation, it should be noted that other dimensions of deprivation are not perfectly correlated with expenditure poverty.
Around 42 percent of the so-called non-poor households are nonetheless without access to safe water, while around 32
percent of non-poor households have to made do without adequate sanitation.
The variability in the nature of poverty across regions has implications in considering strategies for
poverty reduction in Indonesia. First, it is imperative to recognize these regional differences and disaggregate data
regionally to determine what poverty problem predominates where. Second, differentiated strategies may be needed to
address different poverty outcome indicators. Third, these strategies should take advantage of Indonesia s decentralized
administration, which allows for different strategies to be adopted in different regions.
Table 3.2
Multi-dimensional poverty
Proportion
Births attended by traditional paramedics
Poor (%)
Non-Poor (%)
40.2
22.8
6.0
2.5
Children aged 12-15 not enrolled in junior high school
22.5
10.9
Households without access to sanitation
52.4
32.5
Households without access to safe water
51.8
42.2
Children aged 7-12 not enrolled in primary school
Although the correlation between various dimensions of poverty may not be strong in some regions it is
in others, for instance in eastern Indonesia. In this more remote part of the country, clearly income poverty
coincides with poorer service delivery outreach and poorer non-income poverty outcomes. Areas such as eastern Indonesia
may warrant special regional strategies to address the multi-dimensional nature of poverty.
Inequality: a positive story, with questions

One of the major strengths of Indonesia s historical pro-poor growth story has been its relatively equal
national income distribution, which remained fairly stable for decades. From the early 1980s up to the crisis
in 1998, the Gini ratio for consumption stayed at around 33 percent, while the Gini ratio for income was 11 percentage
points higher at 44 percent. Inequality even decreased slightly during the crisis, reaching a Gini consumption ratio of
31.7 percent in 1999. Indeed, had the distribution of expenditure not changed in an inequality-reducing way in 1997-99
as a result of the impact of the crisis, poverty incidence would have been 28 percent, far higher than the 23.4 percent
actually experienced in 1999.
In recent years, the previously stable Gini ratio has started to register an increase in inequality. The Gini
ratio in 2004 was around 34.8 percent or 3.1 percentage points higher than the ratio in 1999. Decomposition of poverty
34
changes after the crisis into a growth component and a distribution component shows that the two components
both played compensating effects. The poverty-decreasing growth effect in 1999-2004 was far larger than the
poverty-increasing distribution effect, resulting in a dramatic decrease of 6.7 percentage points in the poverty
headcount. Over this period, if inequality had not increased as a result of the change in distribution favoring the
non-poor, poverty incidence would have fallen to only 12.1 percent, significantly lower than the 16.7 percent incidence
actually experienced in 2004. While it is unclear precisely what lies behind this shift, empirical evidence has
documented such a phenomenon in all developing countries. Moreover, given Indonesias historical starting point
of relative equality in distribution of expenditure, poverty reduction would be best driven by growth that benefits the
(large) tail of the income distribution that sits below the US$2-a-day poverty line (see Figure 3.1).
The most compelling income inequality story in Indonesia is in geographic and group disparities
(see previous section). Inequality in urban areas is much higher than in rural areas, with the difference in 2004 of as
much as 9 percentage points in the Gini ratio of income. Java/Bali (47 percent) has a wider inequality ratio than offJava (40 percent). The services sector had the highest income inequality in 2004, while the agriculture sector had
the lowest inequality in both areas.
However, the inequality story in Indonesia raises some questions. Is inequality in Indonesia actually as
low as the Susenas statistics historically indicate? Some anecdotal evidence suggests that this may not be the case (see
Box 3.2 on Inequality). If there is an issue here, it likely relates to a major understatement of the income of Indonesia s
most affluent citizens. Further analysis is required to better understand incomes among this group. Not only would this be
difficult to undertake, but such a study of the most affluent group is outside the scope of this poverty report. However, what
is clear is that the highest incomes among even the top decile alone could significantly change the Gini estimates. While
this might stretch the upper tail of income distribution, leaving the distribution hump where it is, this would still not
change the focus or recommendations of this report in its targeting of the poor.
Box 3.2 Is inequality in Indonesia really so low?
The Susenas household survey data indicate that Indonesia has only a moderate level of inequality. Indeed, Indonesia s Gini coefficient
for inequality of 35 percent makes it comparable to the Netherlands, a country with an extensive social welfare system, and well ahead of comparable
developing countries, such as Brazil (61 percent), China (40) and Malaysia (49). The data also suggest that inequality declined from 2002 to 2004.
But is inequality really so low? The perceptions of many ordinary Indonesians are in sharp contrast to the data and simply strolling around the
streets of Jakarta makes one question the accuracy of the data. Central Jakarta is dotted with high-class shopping malls and extremely affluent neighborhoods,
and it is not unusual to see Jaguars and Mercedes fender-to-fender with the ubiquitous battered Metro Minis that are the main public transport buses in
the center of the city.
One possible explanation for the apparent discrepancy between official statistics and casual observation is that the Susenas
household survey does not adequately capture the expenditures of the rich. For instance, the 2002 Susenas records per capita household
expenditure of the highest 99th percentile as a mere Rp 10 million per month (about US$1,100), a surprising low figure. Meanwhile, casting doubt on this
figure, using 2001 data, McKinsey in Hong Kong estimated that super-rich households, defined as those with assets of at least US$1 million, constituted
only 0.03 percent of the total population, or around some 64,000 individuals. Data on import duties reveal that each year about 1,600 Mercedes Benz with
a price tag of over US$30,000 are imported into the country. Another indication of inequality can be seen in the growing number of self-financing
Indonesians studying at Australian universities. While in pre-crisis 1996 the total number of these students was about 3,000, by 2000, during the depth
of the financial crisis, this number had doubled. All of these observations would indicate that inequality in Indonesia may not be as moderate as depicted
in the data from the household surveys.
In research conducted by the SMERU Research Institute, several drawbacks in the data gathered by Susenas are pinpointed, including the use of expenditures
as a proxy for income, the omission of very affluent households, and the failure to look at inequality within and between rural and urban areas.
Source: Susenas, 2002; Suryadarma, Suryahadi, and Widyanti, 2003; McKinsey, 1997.
35
Lagging regions
Regional disparities and pockets of poverty
Wide regional disparities characterize Indonesia, particularly the so-called lagging regions in eastern
Indonesia. The national poverty headcount index hides wide disparities in poverty incidence rates across the country
(see the six Regional Focus sections at the end of this chapter). For example, poverty incidence ranges from 3.2 percent
in Jakarta to 27.9 percent in East Nusa Tenggara. Based on 2004 data, a Papuan was almost four times more likely to be
poor than someone living in the richest region, Kalimantan, with its relatively low poverty headcount of 11 percent. The
magnitude of these significant differences in the incidence of poverty between different geographical areas is often too
large to be explained by the differences in individual or household characteristics alone. Even after controlling for other
characteristics, eastern Indonesia (the island groupings of Nusa Tenggara and Maluku) still exhibits a higher expenditure
poverty rate (both with regard to poverty incidence and severity), as well as a poorer performance in almost all socioeconomic indicators.
Regional disparities are not only evident in sharp poverty variations across island groupings and
provinces: pockets of poverty are also evident within provinces. Just as with regional variations, so internal
variations in province/district poverty are significant. Within relatively rich Kalimantan (poverty headcount of 11 percent)
there are poor kabupaten such as Landak, where 24.7 percent of the population live below the poverty line, and rich
kabupaten such as the city of Banjarmasin, where only 3.2 percent are poor. Meanwhile, within relatively poor eastern
Indonesia (poverty headcount of 26.1 percent) there are kabupaten with a poverty level even lower than the national
average, such as the city of Ternate, North Maluku. Such variations in poverty headcounts are highlighted by the development
of poverty maps using small-area estimation methods, which dramatically reveal the existence of such geographical
pockets of poverty even within districts (see Box 6.10 on poverty-mapping in Chapter 6 on Social Protection).
Figure 3.6 Large variations exist in poverty headcount across Indonesia...
Percentage of poor people by province in Indonesia, 2004
36
Figure 3.6b
...and pockets of poverty exist even within rich provinces
A rich province with poor pockets
A poor province with rich pockets
(East Kalimantan)
(Maluku)
N
N
W
W
Kab.
Nu nnuka n
Nu nuka
E
S
S
M alu ku Ten ga h
S era m Bag ian Bar at
Kota Tar akan
B uru
Ka
b. B uru
Kab.
Bulongan
Bulongan
Ko t a A m b o n
S e r a m B ag ia n T im u r
Kab.
Ber auBer au
Malinau
Kep ul au an Ar u
Ka
tai T imur
Ku b.
taiKu
T imur
Kab
. Kut
ai Barat
Kut ai
Barat
Mal uku Te ng ga ra
Kab.
K utaiK utai
Kota Bonta ng
Kota Samarinda
Pen ajam Paser Utar a

Pasi r

0- < 5
5-9
10- 14
15 - 19
20+
M al uk u Te ng g a r a B a r a t

0-<5
5-9
10- 14
15 -19
20+
Regional disparities are evident throughout Indonesia even when looking at non-income dimensions of
poverty. These regional disparities in non-income poverty correlate broadly with poverty incidence figures across island
groupings. Some of the provinces with the highest incidence of poverty also exhibit the highest incidents of other
deprivations. Overall, Nusa Tenggara, Bengkulu, West Kalimantan, and Papua lag behind other provinces in combating
multi-dimensional poverty. The Spearman Rho correlation between the poverty headcount and other non-income deprivation
indicators is around 18 to 57 percent.
Regional disparities in educational attainment. While mean educational attainment is less disparate across
regions (with the difference between highest and lowest being only 0.6 years), the difference is larger with regard to
educational attainment among the poor. Educational attainment among the poor is three years less in Papua than in
Sumatra, with Papua also exhibiting a wide disparity in attainment between non-poor and poor (see Figure 3.8). In
addition, there is a striking difference in educational attainment between urban and rural areas, with an average gap
of 2.5 years.
Regional disparities in access to safe water sources. Figure 3.7 shows the 55 percentage-point gap between
the access of the lowest quintile to safe water in Papua compared with their better-served lowest quintile counterparts
in Java/Bali. In Sumatra, Sulawesi and Nusa Tenggara/Maluku access to safe water coverage is in the 50-70 percent
range for all quintiles.
Regional disparities in access to sanitation. Even with low aggregate levels of access to sanitation, there are
striking differences in access across regions. The gap between the lowest and highest quintiles is greatest in Papua,
where 10 percent of the poorest quintile have access to sanitation compared with 70 percent of the highest quintile
(Susenas, 2004). Another dramatic difference is between rural areas, where sanitation access is 44 percent, and
urban areas, where access is 66 percent (see Figure 3.7). This coverage is further differentiated among richer and
poorer households. Only 40 percent of the lowest urban quintile has access to improved sanitation, whereas 80
percent of the highest quintile has access. Only 12 percent of the rural poor have any access to improved sanitation.
While the situation has improved in urban areas, there has been less improvement in rural areas over the past
decade.
37
Regional disparities in child malnutrition. The malnutrition rate, while remaining high in all regions, shows
some regional disparities. In Java/Bali, where the malnutrition levels are the lowest across regions, close to onefourth of children under the age of five are considered malnourished47 (and 6 percent of those are considered
severely malnourished). Malnutrition rates are higher in eastern Indonesia, for example, in Nusa Tenggara close to
40 percent of children under the age of five are considered undernourished, while the level of severe malnourishment
reaches 13 percent.48
Figure 3.7 Expansion in access to safe water source and improved sanitation by income group, 1994-2004
Access to improved water has increased in the past decade
Access to sanitation is still lower than improved
but rural areas still lag behind
water source in both urban and rural areas

Note: Definition of improved water is taken as the percentage of people living in a household that have access to bottled, tap, pumped or protected spring water.
What complicates the problems of dealing with lagging regions in Indonesia is that most of Indonesia s
poor live in regions that have relatively lower poverty incidence, such as Java and Sumatra. This is a
function of the much higher population density on these two western islands. For example, while the poverty incidence in
Java is 15.7 percent, the island is home to 59 percent of all Indonesians and 56.7 percent of Indonesia s total poor. In
contrast, while the poverty incidence in Papua is 38.7 percent, Papua is home to only 1.2 percent of all Indonesians and
only 2.7 percent of Indonesia s total poor. Annex III.2 provides an illustration of where poor people who lack access to
basic infrastructure services in roads, electricity, water and sanitation, live. When looking at the numbers of poor people
who lack access, rural Java/Bali, rural Sumatra consistently show up as areas where most of the poor who lack access to
basic services reside. This poses a difficult dilemma for reducing poverty in Indonesia. Clearly, focusing on the poorest
regions is important for promoting equity, regional integration, balanced growth and political stability. But focusing on
addressing poverty in the most densely populated areas would result in a more rapid reduction of poverty in Indonesia as
a whole and, given population density, would carry lower unit costs for poverty reduction efforts.
Regional diversity and the impact of growth on poverty

There are fundamentally different relationships between poverty and growth depending on the region. But
what drives the difference? The different structures of the economies on- and off-Java are revealed in the highly diverse
responses of poverty to local economic growth. The high degree of heterogeneity across Indonesias regions-agro-climactic
zones, local institutions and other important factors leads to fundamentally different relationships between poverty and
growth. Not every region has the same history of poverty reduction poverty reduction in some regions has been much more
47
Child malnutrition is defined as the percentage of children under the age of five who fall minus two standard deviations from the median weight-for-age for the reference population. Severe malnourishment
is defined as the percentage of children whose weight-for-age is more than three standard deviations lower than the average for their reference group.
48
The data are taken from the Core, Module and Yodium (survey of ionized salt consumption) components of the Susenas household survey.
38
rapid than in others. The question is: are the differences across regions due to differences in the levels of regional growth
alone, or are there region-specific factors that differentially translate the growth experience into poverty reduction?
The answer lies in isolating the effect of growth on poverty change and analyzing elasticities between
poverty and growth. To answer this question, provincial level data on poverty measures and consumption measures
that span the period 1984-2002 are analyzed. Provinces are divided into urban and rural areas, so for most provinces
(except Jakarta) there are two observations. The elasticity between poverty and growth for these provincial and urban/rural
areas is then estimated statistically.49 (See Annex III.2 for results.)
By controlling for initial inequality, it is possible to isolate the effect of growth on poverty change.
Growth in mean consumption is negatively and significantly associated with poverty change, with an estimated elasticity
of -2.1 percent. This number quantifies how much the poverty headcount will change given a 1.0 percent change in mean
consumption. Thus in these data the poverty headcount on a national basis is reduced by 2.1 percent on average for every
1.0 percent increase in mean consumption (which is a proxy for economic growth per capita).
It becomes clear that there are regional disparities in how growth impacts poverty. Next, it is possible to
allow for differences across the six regions used in this report: Sumatra, Java/Bali, Kalimantan, Sulawesi, Nusa Tenggara/
Maluku and Papua. The poverty-growth elasticity for Java/Bali is -2.5 percent, greater in magnitude than for the country
as a whole, thus implying that growth in Java/Bali results on average in faster poverty reduction than for the rest of the
country. Figure 3.8 show that several regions have significantly smaller poverty elasticities than Java/Bali. In particular,
Papua, Nusa Tenggara/Maluku, and possibly Sulawesi, all have lower estimated elasticities. Thus, there is clear evidence
of regional disparities in how growth impacts poverty.
Figure 3.8 Growth is less pro-poor in eastern Indonesia
(Poverty growth elasticities by region, urban rural)
49
This estimation involves regressing the change in local poverty on the change in mean consumption, where this change in mean consumption proxies for the overall growth in the local area. Since the
relationship between growth in local consumption and the change in local poverty is in part a function of the distribution of consumption close to the poverty line, it is necessary to control for the distribution
of consumption in the initial or base period. This is done by also including the base period Gini coefficient and the base period poverty headcount in the analysis.
39
Urban and rural areas also tell very different stories. It is also possible to look separately at urban and rural
areas and these present very different regional stories too. For urban areas, the regional elasticities do not vary much. But
for rural poverty, there is a very different regional pattern. The elasticity estimated for Java/Bali is -2.5 percent, very
similar to the elasticity in urban Java/Bali. However, the elasticities for rural Sulawesi, Nusa Tenggara/Maluku, and Papua
are all significantly smaller in magnitude. For Sulawesi the estimated elasticity is -2.1 percent, while for NT/Maluku it is
-0.9 percent and for Papua -0.5 percent. Thus, rural poverty change in these areas is much more disconnected from
overall growth than rural poverty change in other areas of Indonesia. This suggests a need for much better infrastructure
in these areas if they are to be connected to growth throughout the economy. One last feature of regional diversity stands
out in Figure 3.8. For Sumatra and Java/Bali (which together account for 80 percent of Indonesia s population and 79
percent of its poor), rural growth elasticities of poverty are significantly larger (in absolute terms) than urban growth
elasticities. This finding is consistent with much evidence from Indonesia and other Asian countries that growth in agricultural
productivity is pro-poor. More worrying, and reflecting the lack of appropriate agricultural technology in much of eastern
Indonesia, urban growth has been far more pro-poor than rural growth in Kalimantan, Sulawesi, Nusa Tenggara/Maluku
and Papua.
II
A Profile of the Poor
Of every 100 Indonesians...
But of every 100 poor Indonesians...
57 live in rural areas
69 live in rural areas
44 do no have access to safe water
52 do no have access to safe water
49 do not have access to decent sanitation
73 do not have access to decent sanitation
28 have household size more than five
48 have household size more than five
43 have less than primary education
55 have less than primary education
11 are illiterate
16 are illiterate
44 work in agriculture
64 work in agriculture
60 work in informal sector
75 work in informal sector
16 work as unpaid family worker
22 work as unpaid family worker
42 live in villages without secondary
50 live in villages without secondary
high school
high school
36 live in villages without access to telephone
49 live in villages without access to telephone
Of those aged below five, 25 are malnourished
Of those aged below five, 28 are malnourished
and 32 were delivered by unskilled midwife
and 47 were delivered by unskilled midwife
Most poor households are in rural areas. With rural households accounting for about 57 percent of the poor,
poverty in Indonesia is still largely a rural phenomenon. However, the share of urban poverty is on the rise. In 1976, the
share of poor who lived in urban areas was only 18.5 percent, but by 2004 this had reached 32 percent and continues to
increase. By 2015, the urban share of the total number of households living in poverty is expected to have reached 45
percent and will pass the halfway mark by 2023. Understanding the scope, nature and characteristics of urban poverty, as
well as the complex processes of this rural-to-urban shift, will be critical to policy-makers in the coming years (see Box
3.3 on Urban Poverty).
40
Poor households are concentrated in the agriculture sector. Agriculture employs the majority of household
heads. While agricultural households represent only about 41 percent of the population, almost two-thirds of poor household
heads work in agriculture. For all years, the average per capita expenditure was lowest and the incidence of poverty
highest among formal and informal agriculture-sector workers. With the agricultural household poverty headcount being
25.7 percent, agricultural households are 2.6 times more likely to be poor than non-agricultural households. Agricultural
income accounts for about 40 percent of the total income of the poor, compared with 32 percent of the near-poor and 15.8
percent of the non-poor.
Figure 3.9
Households with very small planned

areas are much poorer
In rural areas, those will little or no land are

particularly poor. The poorest people in rural areas tend
to be the farmhands (buruh tani) or smallholders (petani
gurem) working as laborers on other people s land or
operating extremely small plots. Figure 3.9 shows the
poverty headcount by area of rice land planted. The poverty
rate drops sharply as the area increases up to four hectares,
but then stays roughly constant thereafter. Poverty rates
for households with a very small planted area are
particularly high above 20 percent for households with
less than 0.5 hectares of rice land suggesting even higher
rates for the landless (Susenas, 2004).50 The Agricultural
Census 2003 shows that almost half of all agricultural
households cultivate less than 0.5 hectares overall. Thus,
landlessness and very small land holdings are closely
associated with poverty.
Poverty is closely associated with working in the informal sector. Labor markets are characterized by a
high degree of informality. In 2004, as many as 51.6 percent of household heads worked in the informal sector (an
increase on 47.5 percent in 1996). Return to informal employment differs by sector. Although household heads working
in informal agriculture are 2.1 times more likely to be poor, those working in the informal service sector were 21 percent
less likely to be poor than others.
The natural environment has a strong impact on the nature of poverty. Some of the poor live in depleted and
polluted coastal environments, while others albeit far fewer live in dense forests or in the central mountains. Poor
people living in poor quality or fragile environments will not have the same opportunities as those in better endowed
areas. For example: farmers in Kalimantan have to contend with generally poorer soil quality than that found in Java; the
people of Nusa Tenggara receive less rainfall and have steeper slopes that those in other areas, which constrains their
opportunities, as does their remoteness from major markets; and the costal plain of West and South Sumatra is narrow
and steeply rising, so farmers there are more constrained relative to those in the north and eastern side of the island. Thus,
although the general picture of poverty as predominantly rural and based on the agricultural and informal sectors holds
true, the specific constraints and opportunities open to poor people depend to an important extent on their natural
environment.
50
Asian Development Bank (2000).
41
Poor households have less education. The relative risk of poverty decreases with higher education. Junior secondary
school graduates are 26.7 percent less likely to be poor than primary school graduates, and the likelihood of senior
secondary school and university graduates being poor is even lower. The average gap between the poor and non-poor is
two years of schooling. The net enrollment rate of poor children is substantially below the rates of the near-poor and nonpoor. While the difference is insignificant at the primary level, the gaps in secondary enrollment between non-poor and
near-poor households are 8.1 percent (junior secondary) and 9.1 percent (senior secondary), respectively. A child coming
from a poor family is 20 percent less likely to be enrolled in a junior high school and 44 percent less likely to be enrolled
in a senior high school than a non-poor child.
Box 3.3 The faces of urban poverty
The urban poor include a diverse range of people whose vulnerability and poverty vary widely. These include for instance slum-dwellers, street traders,
street children, informal laborers and sex workers. Below we take a brief look at three of these sub-groups.
Street traders: Perhaps the most visible of the urban poor work in the informal sector as street traders (pedagang kaki lima). These traders can be seen
in abundance working in crowded streets, in public markets and near commercial centers and transportation terminals, typically clustered in groups. In
Jakarta and Bandung, it has been estimated that those working in this informal sector tripled in the aftermath of the financial crisis in 1998.
Key to establishing a street-trader business is modest start-up working capital, normally around Rp 200,000 (US$22). Although street-traders are
generally assumed to be poor, they can still succeed in creating opportunities for their households to better survive and, in some cases, even prosper
above the poverty line. Evidence suggests that many can access health and education services, and often achieve access levels to services that are higher
than average Indonesian poverty indicators. Often traders have completed primary education and detailed surveys suggest that higher education levels
(over 12 years) are associated with street traders running more lucrative businesses. While far from easy, and prone to the environmental hazards from
working long hours in polluted areas, street traders can nonetheless help to support poor households.
Sex workers: The sex industry flourishes in most urban areas and women generally enter the industry because they are bonded by relatives or guardians,
are deceived or abducted, or choose to enter voluntarily. Of those who enter voluntarily, many explain their reason as the need to look after poor or sick
parents, to support children or younger siblings, or to meet important social obligations. Evidence suggests that poor women are highly represented in
the sex industry and studies show a strong correlation between entry and low education levels.
In Indonesia, sex work occurs in: official brothel complexes (lokalisasi) where sex workers both work and live; unofficial brothels (rumah bordil) and
entertainment venues, where women are managed, and often abused, by pimps; and on the street and in open public locations (wanita jalanan). These
traditional forms of sex work and the vulnerabilities that they bring are increasingly being added to by new forms of sex work in the urban environment.
These now include young women who are often educated (and generally not poor) but who choose to accept sex work to pay for schooling or clothes.
Street children: Those described as street children in Indonesia are predominantly part of the urban poor, often (but not always) live outside their homes
and undertake economic activities, such as begging, rubbish-picking, shoe-shining, newspaper-selling, or work as sweatshop workers or petty criminals.
Many are later forced into sex work. An ADB survey of 12 cities conducted in 1999 estimated that there were some 170,000 street children in Indonesia
(ADB, 2000). Typically, domestic violence or sexual abuse, household poverty and seeking economic opportunities were given as reasons that children
become street children. However, some poor households sell or abandon their children to reduce family size.
The lives of street children are characterized by vulnerability and risk. They may work alone, with older street children, or under the control of adults or
gangs. In many cases they are forced to raise a certain sum of money each day for invisible adults who control traffic intersections, footbridges or
markets. The consequences include: detention and beatings by the authorities; prostitution; exclusion; and a loss of self-esteem. Most have no ID cards
or any form of identification, denying them access to formal health services or education.
Source: Suharto , 2002; West, 2003; and Kearney, 2000.
42
Poverty appears to be no higher among households headed by women, although this may be deceptive.
Households headed by men had a marginally higher poverty rate (16.7 percent) than those headed by women (15.9
percent) in 2004. So, using the Susenas data, apparently and contrary to popular belief female-headed households are
slightly less likely to be poor than male-headed households. However, only 9.7 percent of households are female-headed,
marginally more in urban (10.9 percent) than rural (8.7 percent) areas. But when other characteristics, such as demographic
composition, schooling, sectoral work and access to basic infrastructure are controlled for, the story is reversed: maleheaded households do indeed appear less likely to be poor than female-headed household (ADB, 2000).51 (See the
following section on Gender for a more detailed analysis.)
Poverty is no higher among households with an unemployed household head. Among poor households,
heads of household can ill afford to be unemployed (see graph in Box 3.4). Consequently, the relative risk of being
unemployed for poor household heads is negative. Household heads of poor households are 9 percent less likely to be
unemployed than non-household heads from non-poor households. The unemployment rate for the bottom quintile of
household heads is only about 2 percent.52 Thus, the statement that unemployment is a luxury good is indeed true for
poor heads of households, but not the case for other household members. For them, poverty is strongly associated with
unemployment. Non-household-head members of poor households are 8 percent (1.08 times) more likely to be unemployed
than non-household-head members of richer households. While poverty and unemployment are two separate issues,
since 1999 the unemployment rate of the poor has slightly exceeded that of the non-poor (see Box 3.4).
Poor households are larger. The average size of poor households is 5.4 members, compared with only 4.3 for nonpoor households, more than one household member larger. Households of five people or more are 2.7 times more likely
to be poor than smaller households. The difference in household size is largely due to higher fertility rates among the
poor. The average number of children below the age of 18 among poor households is 2.6, while among non-poor households
it is only 1.6 fully one child less.
The urban poor live in the worst locations, in densely occupied settlements, and in inadequate and
crowded housing. A sharp rise in slum city areas was noted in the MDG progress report (Indonesia Progress Report on
the MDGs, 2004). In 1999, some 48,000 ha in Indonesian urban areas were officially classified as slum areas inhabited by
2.3 million slum-dwellers, a significant increase on 38,000 ha in 1996. The average size of dwellings occupied by those
in the poorest quintile is only 10m sq, while for those in the second quintile the average size is double, at 20m sq
(Susenas, 2004).
While the consumption pattern of poor households in Indonesia is broadly typical, two exceptions stand out:
The poor are heavily dependent on rice and therefore very sensitive to increases in the price of
rice: Rice is by far the most important staple in Indonesia. It is consumed more by the poor, for whom it
constitutes almost one-quarter of all expenditures, than the non-poor, for whom it constitutes closer to 10 percent
(see Table 3.3). This means that looking solely at
the consumption side, high rice prices and policies
that increase rice prices will have a severely adverse
impact on the poor. Even taking into account rice
production, the poor are still hurt by rice-price
increases, since three-quarters of the poor are net
rice consumers (see Box 3.5 on Rice).
51
See Section III.

The uses an old definition of unemployment rate that does not include
discouraged workers.
52
43
The poor consume a very high share of their money on tobacco: Tobacco
(including a small amount of betel) constitutes an unduly large share
of the non-food consumption of poor households at 6.3 percent among
the poor and 7.7 percent among the near-poor. The near-poor share is even
higher than the non-poor (6.6 percent). What is remarkable is that this share is
close to what the poor spend on key food items, such as vegetables (6.6 percent),
and six times more than what they spend on eggs and milk. In fact, spending on
tobacco is not much less than the total spent on all goods and services, including
spending on health and education services. This means that any tax that increases
tobacco prices is not progressive (i.e. it will hurt the poor). But consideration also
needs to be given to policies that promote longer-term welfare of households since
tobacco has long-term negative health costs (see Box 3.6 on Tobacco).
Table 3.3
Rice and tobacco are large spending items in the budgets of the poor
Poor
Rice
Near-Poor
Non-Poor
Total
24.1
19.4
9.4
11.8
6.3
7.7
6.6
6.8
6.0
7.5
6.6
6.7
Vegetables
6.6
6.0
4.3
4.7
Eggs and milk
2.5
2.8
3.4
3.3
Tobacco and betel

Of which tobacco
Prepared food and drinks
6.7
8.1
10.3
9.8
Other food and consumables
24.3
24.2
21.3
21.9
Housing and household facilities
14.7
15.1
18.8
18.0
7.3
8.4
13.3
12.2
Good and services

Clothing, footwear and head gear
5.4
5.6
5.1
5.2
Durable goods
1.0
1.5
4.8
4.1
Taxes and insurance
0.3
0.4
0.9
0.8
Festivities and ceremonies
0.7
0.8
1.8
1.6
100.0
100.0
100.0
100.0
Source: Susenas 2002 Consumption module.
Box 3.4 Unemployment and poverty are not equivalent

Unemployment continued to climb and formal job creation lagged following the crisis. The unemployment rate stood at a mere 4.9 percent
in 1996, but jumped to 6.3 percent in 1999 and by 2004 had reached the worrying level of 12.8 percent. Recent trends in youth unemployment are also
alarming: youth unemployment has risen dramatically since the financial crisis, surging from 16.0 percent in 1996 to 25.2 percent in 2004. Lackluster
employment creation due to increasing rigidity in the labor market and continuing setbacks in the investment climate have been the main reasons for
rising unemployment in recent years. On average, employment growth remained slower than labor force growth between 1996 and 2005 (see Chapter 6
for more details). During 1996-2004, while employment grew at an annual average of 1.25 percent,53 the labor force expanded at a rate of 1.97 percent.
As a result, unemployment numbers using the old definition increased by an average 6.28 percent annually. Rates of job creation were significantly
lower than the rates at which new entrants joined the labor force. In some years, job creation rates were even lower than the increase in the population over
15 years of age. Labor force participation rates showed an increasing trend in 1996-2005, although there were some fluctuations.
53
44
Sakernas 1996-2004. It is difficult to measure overall trend using Susenas as some of the years exclude Aceh, Papua and Maluku.
Curiously, increasing unemployment in 1999-2004 occurred against a background of falling poverty rates. As discussed early, poverty
levels declined in Indonesia using all poverty measures in 1999-2004: the headcount came down from 23.4 percent in 1999 to 16.7 percent in 2004. Even
accounting for lag effects, this dichotomy poses a puzzle for policy-makers. Employment creation and connecting the poor to jobs are critical for poverty
reduction and Chapter 6 goes into further detail about constraints in the Indonesian labor market that reduce the access of the poor to jobs (see Chapter
6 section).
Contrary to common perception, unemployment and poverty are not equivalent. While there is a strong dynamic link between poverty
reduction and the poor s connection to the labor market, the characteristics of the poor and the unemployed in Indonesia are not similar. In fact, the
unemployed have a very different profile compared with the poor, and this partly explains the phenomenon of how unemployment can increase while
poverty rates decline.
Educational attainment: While unemployment is mainly a problem of the educated, poverty remains a problem of the less educated. In fact, the higher
the level of education, the higher the risk of unemployment and the lower the risk of poverty (see Table below). Unemployment is highest among senior
high school graduates (18.9 percent) and lowest among those with no primary education (7.9 percent). A senior high school graduate is 1.4 times more
likely to be unemployed than someone who never completed primary education. Meanwhile, the relative risk of a senior high school graduate falling into
poverty is far lower than for those with primary education or less.
Urban/rural area: Another indication that unemployment and poverty are not equivalent is the higher levels of unemployment found in urban areas than
rural areas. Continuing a trend of previous years, in 2004 urban unemployment was 10.1 percent, while rural unemployment was 7.4 percent. Conversely,
poverty is significantly higher in rural areas (20.1 percent) compared with urban areas (12.1 percent).
By age group: Unemployment is also higher among the young, while poverty is only moderately higher. While overall unemployment in the workforce
stands at 18.4 percent, the rate for those aged 15 to 24 is 33.9 percent, making this age group almost five times as likely to be unemployed. Meanwhile,
poverty is only slightly higher among those aged 15 to 24 (18.9 percent) compared with adults (14.5 percent). This indicates that young people are only
moderately more likely to become poor than adults, far lower than risk of young people becoming unemployed.
To conclude, the high level of unemployment among educated Indonesians can be partially explained by the hypothesis that
unemployment is a luxury good . In other words, those who can afford to be unemployed are generally not the poor. This goes some of the way
towards explaining the diverging trends in poverty and unemployment levels. While the profiles of the poor and unemployed help to explain the recent
diverging trends, this analysis does not detract from the importance of labor-market policies that connect the poor to jobs and the growth process. Further
analysis of labor-market policies and the impact on the poor can be found in Chapter 6 of this report.
Unemployment is predominantly a problem of the educated in urban areas,

increases with while poverty remains a rural problem of the less educated.
Characteristics
Educational attainment
Never attended or did not
complete primary
Primary
Junior high school
Senior high school
Diploma I/II/III
Bachelor/Post-graduate
Total
Area
Urban
Rural
Indonesia
Poverty Risk (%)
Unemployment rate of household heads

consumption.
Unemployment
Risk (%)
21.4
8.2
18.0
12.7
6.6
2.5
1.6
10.2
16.8
18.9
12.6
10.9
12.1
20.1
15.9
10.6
16.7
12.8
45
Box 3.5 Rice-price increases disproportionately hurt the poor

Rice is a critically important commodity for all Indonesians but especially for the poor. This is because rice accounts for 24.1 percent of
poor households total consumption, significantly higher than the budget share for rice among the near-poor (19.4 percent) and the non-poor (9.4
percent). Furthermore, for the country as a whole, four out of five households are net consumers (i.e. they consume more rice than they produce).
Although most of the poor live in rural areas and work in agriculture, more than three quarters of the poor (76 percent) are net rice consumers. In urban
areas, 85.6 percent of the poor are net rice consumers, but even in rural areas 72.1 percent of the poor are net consumers. Consequently, any increase in
the price of rice disproportionately hurts the poor.
Sharp rice price increases dramatically increased poverty during the Asian financial crisis in 1997-98. As the rupiah collapsed, so the
domestic price of tradable goods consumed by the poor surged, most notably the price of rice. The urban poor were hit worst: not only were they more
predominantly net rice consumers but they suffered far more from the contraction in employment, both formal and informal. The government was
successful in restoring relative macroeconomic stability by 2001, strengthening the rupiah and bringing down the domestic price of rice as a result. This
helped to bring down the poverty headcount, which fell from 23.4 percent at the peak of the crisis to 16.7 percent by 2004. The experience of the crisis
shows the importance of macroeconomic stability and, in particular, the stability of rice prices in preventing future increases in poverty rates.
These lessons became relevant once again towards the end of 2005 and into 2006, when the price of rice surged by about 30 percent above international
prices, rising 55 percent between June 2005 and June 2006 far ahead of other domestic food price increases.
In this case, the rice price started to run out of control due to shortages in supply caused in part by the government s ban on the
importation of rice. Simulations indicate that the price increase may have moved over 3.1 million Indonesians below the poverty line. Prices subsequently
fell with the arrival of the main harvest but, worryingly, have begun to rise again. The most efficient way for the government to address this issue going
forward despite strong resistance from the rice-producers lobby would be to impose a low tariff and allow general imports of rice. By helping to
maintain rice price stability such a policy would be highly pro-poor. (See Chapter 6 for more on policies for rice price stabilization.)
Rice-price increases hurt the poorest in Indonesia
The domestic rice price overtook international prices

(incl. the tariff) after December 2005 due to shortages in
domestic supply
(Net benefit ratio of increase in rice prices

by expenditure group)
-0.05
-0.1
-0.15
0
20
Poor
40
Non-poor
60
Expenditure percentiles
46
80
100
Source: FAO, Border prices wholesale equivalent for Vietnam rice 25 percent, Jakarta wholesale rice
IR 64 III PIBC.
Box 3.6 Tobacco damages poor households finances as well as their health
Tobacco constitutes a major expenditure for the poor. According to Susenas data (see Table 3.5) the second largest single commodity item
consumed by poor Indonesians after rice is cigarettes. In 2002, 6.3 percent of the poor s expenditure went on tobacco, while at 7.7 percent the figure was
even higher for the near-poor (Susenas, 2002). The poor are also more likely to smoke; men in the lowest three quintiles smoke the most at 62.9 percent,
65.4 percent and 64 percent consecutively, compared with 57.4 percent of men in the highest quintile (Susenas, 2001). Despite the negative impact of
tobacco consumption on family budgets and health, public perceptions of the tobacco industry are largely positive. The government enjoys high revenues
from the tobacco industry, and the industry is perceived as an important source of employment even though its significance is actually in decline.
Expenditure on tobacco takes away money that the poor could spend on food, healthcare and education and increases the burden
on the national health system. It is clear that spending on tobacco, which may be seen as a short-term consumption benefit, has long and short-term
negative health implications for both smokers and passive smokers and inhibits poverty reduction. About 6.5 million Indonesians suffer smoking-related
diseases annually, such as lung cancer, coronary heart disease and strokes. The National Agency for Drug and Food Control (BPOM) reports that 1,100
deaths a day and 57,000 deaths a year (primarily male) are attributable to tobacco use in Indonesia.54 Such high mortality has a cost to household
expenditures as well as to the national health system and economy. At the level of the household, losing a primary breadwinner is shown to be one of the
major shocks that cause poverty (Chapter 6 on Social Protection).
Meanwhile the importance of the tobacco industry is in decline in terms of employment. In 1970, tobacco manufacturing contributed 38
percent of total employment in the manufacturing sector. This figure has dropped significantly to just 5.6 percent in 2000, and only 1.0 percent of
employment in the industrial sector (Ministry of Health, 2004). In addition, the tobacco industry on average pays only two-thirds of the average monthly
wage to its employees. The largest impact of a decrease in tobacco consumption is likely to be felt in the agricultural sector with an estimated 2.3 million
farmers growing tobacco or cloves for domestic production of kretek (clove cigarettes) and cigarettes.55 Other areas that would suffer are trading, transportation
and advertising (Reynolds, 1998).
However, the tobacco industry remains a significant source of government revenues and public perceptions of tobacco and the
industry remain largely positive. In 2004, the industry produced 199 billion cigarettes and contributed Rp 27 trillion (almost US$3 billion) in excise
revenue (Abadi, 2005). Indonesia is now a net importer of tobacco leaves, tobacco-leaf imports exceeding exports by US$44 million. However, it is a net
exporter of tobacco products: net exports were equivalent to US$176 million in 2001. Despite this, tobacco exports contributed only 1.0 percent of total
exports in 2002 (Ministry of Health, 2004). In a 2001 study by the University of Indonesia s Center for Health Research, almost 60 percent of respondents
had a positive view of the industry. The majority of tobacco users perceive it to be an important provider of jobs (39 percent). Only 28 percent acknowledged
that the industry was harmful to health (14 percent) and detrimental to family finances and the environment (14 percent) (Center for Health Research,
2001). Tobacco users have only limited knowledge of the effects of smoking on health: heart ailments (19.8 percent), respiratory ailments (74.1 percent),
cancer (4.3 percent) and impotence (0.2 percent). Knowledge of the harmful effects of smoking fails to prevent teenagers from starting to smoke or
continuing to smoke (Martini and Sulistyowati, 2005), their decision is based on affordability and availability of cigarettes at home and at school.
What would be the most effective way of reducing tobacco consumption among the poor? International experience has shown that the most
effective way of reducing tobacco consumption is by increasing the price of tobacco products through increasing taxes on these products. Research in
Indonesia and other Southeast Asian countries, however, has shown that increased prices alone do not have as significant an impact as in other countries.
A 10 percent price increase in tobacco products is predicted to lead to only a 3.5-6 percent decrease in consumption (Ministry of Health, 2004). Likely to
be more effective, is a widespread advertising and public awareness campaign of the health and economic consequences of smoking and passive
smoking, combined with a comprehensive ban on tobacco advertising (partial bans are ineffective). Increased taxation could be used to ensure government
revenues do not decline as anti-smoking campaigns are put into place, and to finance such campaigns.
What would be the effect of a national anti-smoking campaign on poverty, health and the Indonesian economy? The tobacco issue is
a difficult one to tackle, with strong political and private sector interests determined to maintain a large market, even if this contributes to impoverishment
and disease. A reduction in tobacco consumption will improve health outcomes and improve the welfare of the poor both directly and indirectly. The
reduction in employment in tobacco manufacturing will not likely have a significant impact; however, government will need to create strategies to address
this reduction in employment and in assisting tobacco farmers to switch to different crops. The effect on the economy as a whole is unlikely to be
significant if the government increases taxation on tobacco products in accordance with declining demand.
54
55
World Health Organization (WHO) estimates.

Indonesia: tobacco control policy options (draft) (2005).
47
III
The Determinants of Poverty
This section uses multi-variate analysis to reveal the determinants and the relative importance of key
household characteristics, assets and access as correlates. This analysis gives a more comprehensive picture
than one produced by relying on simple poverty risk descriptions discussed in the previous section. This is because the
high relative risk of certain groups within the population could be attributable to household characteristics within the
group, rather than to the group characteristics. In addition to commenting on the relative roles and magnitude of household
characteristics and endowments in determining poverty status, multi-variate analysis also permits an assessment of the
potential impact that policy changes in these factors are likely to have on poverty, holding all other factors constant.
But it is important to note the limitations of such analysis. First, the analysis is constrained by the data available,
in this case secondary data from Susenas and Podes (Village Potential Survey). Although in some cases proxies can be
used for important determinants, such as access to infrastructure, numerous correlates of poverty are not quantifiable.
Second, while theory also supports the findings that many of the variables in the analysis do contribute to poverty, the
statistical results are more appropriately interpreted as correlates rather than purely statistical causes of poverty, since
causality can run both ways.56
Several key factors matter for poverty and hence matter for reducing poverty. Table 3.4 presents household
expenditure functions in urban and rural areas in 1999 and 2002. The correlates of poverty are analyzed using regression
analysis to see the effect on the household consumption of a specific household, while holding constant all other
characteristics. The regressions estimate the partial correlation coefficient between household consumption, and household
access and asset variables included in the analysis.57
Correlate 1: Education
Poverty is closely associated with inadequate education. Similar to findings in other countries, higher educational
attainment is associated with higher household consumption. In addition, the partial correlation coefficients are generally
higher in urban than in rural areas, both for household heads and members, indicating that urban households benefit
disproportionately more than rural households for every extra year of education.58
Going beyond primary education significantly improves welfare. Correlates of poverty exhibit a convex relationship
to level of schooling, i.e. the gains are even stronger with higher levels of education (see Table 3.4 and Figure 4.8 in Chapter
4). In urban areas, a household head with a senior secondary education is associated with a consumption level that is 33
percent higher than that of an uneducated household head. The consumption increase associated with education is especially
marked for university graduates both in urban and rural areas (72 percent and 45 percent, respectively). The same results
hold true (but to a lesser degree) for other household members, especially for households in urban areas.
56
For example, not only does the lack of a certain critical endowment affect the ability of households to generate income or escape from poverty, but also being poor results in the inability to invest in or have
access to such a critical endowment.
57
In the following discussion, we refer to correlates as determinants and the coefficient associated with a particular factor as its return .
58
In urban households, having a household head with primary education raises consumption by 6 percent (compared with households with uneducated heads working in the same sector). However, in rural
households, having a primary education does not provide additional benefits. On the contrary, having a head of household with a primary education instead of an uneducated household head decreases
household consumption by 2.4 percent.
48
Table 3.4
Household expenditure function by urban/rural area, 1999-2002

1999
Demographic Characteristics of Household Head

Sex
Years of schooling
Square of years of schooling
Experience
Square of experience
Occupation of Household Head
Inactive
Formal agriculture
Industry
Informal
Formal
Services
Informal
Formal
Occupation of Non Household Head with Highest Education
Agriculture
Informal
Formal
Industry
Informal
Formal
Services
Informal
Formal
Other Household Characteristics
Average years of schooling of household
Square of average years of schooling of household
Status of house
Community and Facility Characteristics by District
Population density
Average number of senior high schools
Proportion of vocational skills education
Proportion of existing credit facilities
Proportion of type of widest road that is asphalt
Proportion of existing telephone and wartel
Constant
Observations
R-squared
2002
Urban
Rural
Urban
Rural
0.14290
(15.610)***
-0.02214
(7.610)***
0.00364
(21.629)***
0.02444
(33.790)***
-0.00035
(30.309)***
0.28362
(37.360)***
-0.03169
(12.140)***
0.003674
(20.320)***
0.018353
(31.459)***
-0.000283
(31.379)***
0.15825
(17.170)***
-0.00711
(2.790)***
0.00292
(21.200)***
0.02029
(28.790)***
-0.00029
(26.010)***
0.31072
(38.070)***
-0.02078
(7.710)***
0.00280
(15.150)***
0.01587
(25.799)***
-0.00026
(26.940)***
0.01924
(1.230)
-0.05920
(2.830)***
-0.020381
(2.160)**
-0.060937
(7.610)***
0.10219
(7.460)***
0.03589
(1.870)*
0.00888
(0.8900)
0.03127
(3.530)***
0.01409
(0.570)
0.06052
(3.990)***
-0.018471
(1.140)
0.069911
(8.290)***
0.08382
(4.320)***
0.17636
(13.550)***
0.05413
(3.990)***
0.11729
(12.500)***
0.05909
(3.970)***
0.11654
(8.090)***
0.096282
(11.030)***
0.142563
(17.459)***
0.13452
(10.420)***
0.21392
(17.200)***
0.14269
(14.990)***
0.22667
(25.000)***
0.03277
(2.610)***
-0.01697
(0.980)
0.031849
(8.550)***
0.035793
(5.150)***
0.01016
(0.940)
-0.02273
(1.140)
0.02389
(6.240)***
0.03346
(3.820)***
0.04135
(2.600)***
0.03417
(4.530)***
0.031828
(3.370)***
0.057964
(7.940)***
0.03074
(2.240)**
0.05449
(7.910)***
-0.01333
(1.300)
0.09136
(10.700)***
0.09247
(14.550)***
0.11351
(20.660)***
0.096342
(14.000)***
0.114582
(16.180)***
0.06596
(9.540)***
0.12597
(23.200)***
0.10010
(12.780)***
0.15694
(19.160)***
0.016
(13.920)***
-0.00014
(1.570)
0.13997
(18.020)***
0.030124
(29.540)***
-0.000939
(10.430)***
0.011664
(0.800)
0.01241
(11.770)***
0.00011
(1.390)
0.13657
(19.510)***
0.02913
(27.440)***
-0.00098
(10.240)***
0.02833
(2.820)***
0.00039
(14.630)***
0.01290
(9.130)***
0.06349
(9.770)***
0.00984
(1.550)
0.08027
(6.180)***
0.04678
(4.330)***
12.15499
(546.179)***
24,606
0.4560
-0.000046
(0.300)
-0.001383
(0.360)
0.111026
(8.440)***
0.028603
(5.060)***
0.031031
(6.540)***
0.000919
(0.150)
12.41
(738.450)***
34,034
0.3923
0.00003
(8.230)***
0.00679
(4.330)***
0.07061
(11.370)***
0.05504
(8.750)***
0.07721
(6.950)***
0.01721
(1.300)
12.18094
(584.729)***
27,288
0.4675
-0.00220
(6.010)***
0.03793
(7.220)***
0.00999
(0.7700)
0.01788
(2.920)***
0.03111
(6.370)***
0.00722
(1.3400)
12.42142
(925.669)***
30,712
0.3831

Note: Absolute value of t statistics in parentheses significant at 10 percent; ** significant at 5percent; and *** significant at 1 percent.
49
Improving educational attainment in certain regions/areas is correlated with even greater poverty
reductions. The poverty reduction gains to increasing education levels in Java/Bali are larger than elsewhere, as this
region has the lowest educational attainment and the highest coefficient associated with education in the expenditure
function (see Table 3.6 and Annex III.3). Likewise, increasing educational attainment reduces poverty more dramatically in
urban areas. This highlights one of the major dilemmas facing Indonesian policy-makers: while confronted with higher
levels of poverty and deprivation in rural and remote areas, faster inroads in reducing aggregate national poverty are
possible by investing in the poor in more densely populated areas (such as urban areas and Java/Bali) where economic
activity is also higher. For example, if the education levels of urban poor household heads and members in Java/Bali were
increased to the senior secondary level, this would raise household consumption of the poor by 37 percent. This is a
significantly higher gain than aiming for universal junior secondary school coverage, which would provide only a 13
percent increase.
Correlate 2: Occupation
Working in agriculture is strongly correlated with
poverty. Household heads working in agriculture have
significantly lower consumption levels (and hence are more likely
to be poor) than those working in other sectors. Using household
heads working in the informal agricultural sector as the base, the
correlates of poverty (see Table 3.6) suggest that rural household
heads working in formal agriculture are associated with a 3.1
percent increase in consumption level, while those working in
informal industry can expect a 5.4 percent increase. The gains
are even higher for household heads working in formal industry
(11.7 percent). The highest gains are in services: in informal
services the gain is about 14 percent, while in formal services the gain is about 22 percent, both for urban and rural areas.
Given that the poor have a low share of formal and non-agricultural sector employment and the fact that working in these
more lucrative sectors is correlated to reductions in poverty, so a movement of labor towards the formal agricultural
sector, or the formal and informal non-agricultural sectors, will provide an escape from poverty (see Chapter 4 on Growth).
The vulnerability of the agricultural sector
...usually, if it is the rainy season, we would plant rice paddy and corn. Already the land is not fertile now that the climate
is changing as well. For example, we would put down the fertilizers for chili peppers or plant tobacco after the harvesting
of rice paddy, but then it would rain in the evening. This doesn t happen just once, but many times. As a result, we
experience harvest failures, and then there are also attacks of locusts and rats. So we live on the profit made from selling
cows or chickens... (Bapak Amrillah, participant of the male group of a FGD, Madura, East Java).
Livelihood and seasonal calendar
...paceklik is the period when it is very hot and dry, there are no jobs around, no informal work as laborers, only lots of
thieves and people getting divorced... (Mamiq Dena, participant of a FGD, West Nusa Tenggara)
Kajian Kemiskinan Bersama Komunitas, Kikis, 2003.
50
Correlate 3: Gender
Although poverty appears to be marginally lower among female-headed households, the true picture is
hidden: male-headed households do still have significant advantages over female-headed households. In 1999, holding
other characteristics constant, urban male-headed households had expenditure levels that were 14.4 percent higher than
urban female-headed household. This gender gap is even more striking in rural areas, where there was a 28.4 percent
difference. By 2002, the gender gap had widened to 15.8 percent for urban and 31.1 percent for rural. The seemingly
conflicting results between the correlates (which indicate that female-headed households are significantly worse off) and
the simple descriptive analysis (which indicates that female-headed households are marginally less poor) can only be
explained by unobserved characteristics, such as a higher propensity to shocks and lower accessibility to mitigating and
coping instruments, that may correlate with the gender of the household head. The risk and vulnerability assessment
among different types of household and life-cycle stages indicates that poor female-headed households have a higher risk
of experiencing negative shocks from conflict, health shocks and economic risks.59
Correlate 4: Access to basic services and infrastructure

Poverty is clearly associated with lower levels of access to basic facilities and infrastructure. Several
locality proxies were used to represent these various access levels. These locality variables can be interpreted as proxies
for local endowment and represent the effect of local characteristics not captured by other variables. Almost all of these
accesses to facilities and infrastructure have substantial predictive power with significant and positive coefficients (see
Table 3.6).
Households in rural areas with more access to secondary education are significantly less likely to be
poor. In rural areas, having access to secondary education within reasonable proximity is associated with a 3.8 percent
increase in household consumption.60 The coefficient associated with access to secondary education is the highest in
rural areas of all other access variables considered. Given that the proportion of the rural poor who live within proximity
of secondary school is only 11.7 percent, investing in secondary schools in rural areas with high poverty densities should
yield significant results. Interestingly, the gain from access to secondary education is much less in urban areas, amounting
to only a 0.7 percent increase in consumption level. This could be a consequence of diminishing returns, as urban areas
generally have a higher number of secondary schools. In rural areas, secondary schools also serve as venues for information
and knowledge diffusion, and teachers are often regarded as local activists/key players (tokoh masyarakat) involved in
various development project or village planning activities (so-called Musbangdes). Combined with the earlier findings of
high return to education in urban areas, this indicates that further investment in education in urban areas should place
stronger emphasize on other proximate determinants, such as improving teacher and school management, not on the
construction of additional schools.
Access to informal courses can be a key factor in upward economic mobility, especially in urban areas.
Living in areas with access to informal courses raised household expenditure by 7.1 percent in urban areas.61 In rural
areas, access to informal courses only slightly increased household expenditure, by 1.0 percent. This is not so surprising,
as demand for the technical skills associated with this particular proxy (mechanical and computer skills) is higher in
urban areas.
59
Please see Chapter 6 on Social Protection for further discussion.

Number of secondary high schools within a village/kelurahan is used as a proxy for access to secondary education.
51
Availability of computers or mechanic courses serves as a proxy for access to informal courses.
60
51
Access to a neighborhood credit institution also significantly increases expenditure and reduces the
likelihood that a household will be poor. Rural households with access to a village savings and loans unit (Koperasi
Unit Desa, or KUD) have expenditures that are 2.0 percent higher than those without such access.62 Urban households
with access to a bank have a 5.5 percent increase in expenditures relative to those that do not. The proportion of poor
households with access to a KUD in rural areas is 16.1 percent, whereas the proportion of urban poor households living
within proximity of a bank is 31.2 percent. Although credit programs through commercial banks such as BRI (Bank Rakyat
Indonesia) or community-driven development (CDD) programs such as a KDP (Kecamatan Development Program) or
UPP (Urban Poverty Project) proliferate, there is clearly a strong argument for expanding access to credit for poorer
households as a means of reducing poverty.
Road access is correlated with higher levels of consumption. Having all-year passable asphalt roads is associated
with higher expenditure levels in both urban (7.7 percent higher) and rural (3.1 percent higher) areas.63 This is hardly
surprising as road access is crucial in providing access to opportunities (labor and product markets) and services (health
and education services). Similar to access to other infrastructure and facilities, access is higher in urban areas and rises
with income. Access to roads is limited for the poor. While the proportion of non-poor households with access to all-year
passable asphalt roads is 75.9 percent, only 60.9 percent of poor households have access to all-year passable asphalt
roads. The problem is more acute for the rural poor, where only 53.0 percent of poor households have access to all-year
passable asphalt roads.
Access to telecommunications was insignificantly correlated with consumption at the national level,
but not in some regions. Unlike other access proxies, access to telecommunications has insignificant coefficients.
The only exceptions to this were in urban areas of Sumatra and Papua, and rural areas of Java/Bali and Kalimantan.
Correlate 5: Geographic location

Given these regional disparities, it is not surprising that
geographical location also correlates with poverty. Today, and
despite Indonesia s vast area, it is increasingly possible to make use of
much finer geographic disaggregation techniques to confirm these
disparities and focus poverty-reducing efforts down to the lowest levels.
Indonesia has 33 provinces; 440 districts or municipalities (kabupaten or
kota); 5,850 sub-districts (kecamatan ) and 73,219 villages (desa/
kelurahan). However, for the purpose of this national poverty assessment,
although it is important to capture the disaggregated story as much as
possible, it was decided to focus primarily on the geographic differences
and findings across six broad island groupings: Sumatra, Java/Bali,
Kalimantan, Sulawesi, Nusa Tenggara/Maluku and Papua.64 The six Blue
Regional features focus on each of these island groupings (at the end of
this chapter) provide a sense of these regional variations and an abbreviated
poverty profile for each island grouping. Regarding the correlation between
these geographic groupings and poverty, summary findings from the poverty
correlates of each region are described in Box 3.7 below:
62
Access to credit is defined as having a bank within proximity of a kelurahan in urban areas, or a KUD
within a village in rural areas.
63
Road access is proxied with a variable indicating access to an all-year passable asphalt road.
64
While this gives a flavor of the broad geographical variation, it was also noted that there is high variation
of poverty within these areas. Future poverty reports should focus on more regionalized analysis.
52
Box 3.7 There are differences in poverty correlates and constraints across regions
Specific poverty correlates for each region and the decomposition of the major differences in poverty incidence between Java/Bali and the five other
regions65 (see Annex III.3) reveal the following insights:
Java/Bali: The main advantages that Java/Bali has over the other five regions are its higher return to education, its higher level of access and assets, and
its lower informality rates in employment. Expanding infrastructure, especially in access to roads, as well as access to credit in urban areas, would help
upward mobility strategies in Java/Bali. That said, the main disadvantages of Java/Bali are its higher propensity to shocks than other regions, its lower
return to other household member (i.e. non-household head) occupation, and its lower return to access to secondary education.
Sumatra: The main advantages of Sumatra compared with Java/Bali are its higher return to work experience, its higher return to communication access
and its higher return to occupational choice. Sumatra s access to telecommunications coefficient is the highest of all six regions, correlating to an 8.3
percent increase in consumption levels. On the other hand, Sumatra was at a disadvantage to Java/Bali in terms of its lower return to education, its lower
level of access and assets, and its higher rates of informality in employment. If Sumatra had been able to achieve the same return to education as Java/
Bali s, then poverty incidence in Sumatra would have decreased by an additional 5.6 percentage points.
Kalimantan: The main advantages of Kalimantan compared with Java/Bali include its higher return to occupational choice for non-household heads, its
higher return to roads and also its higher return to work experience. However, Kalimantan was at a disadvantage to Java/Bali in terms of its lower return
to education, its far lower levels of access and assets, and its higher level of informal employment. Had Kalimantan s distribution of assets and access to
basic facilities been similar to Java/Bali s, poverty incidence would have decreased by additional 7.6 percentage points.
Sulawesi: The region of Sulawesi has several advantages over Java/Bali, including its higher return to occupational choice for non-household heads, its
higher return to access to communications, and its higher return to access to secondary education. Conversely, Sulawesi is at a disadvantage to Java/Bali
when it comes to its lower return to education, its lower return to work experience, its lower level of access and assets, and its higher rates of informal
employment. Had Sulawesi s household heads been able to achieve a similar return to work experience as in Java/Bali, poverty would have decreased by
an additional 2.3 percentage points.
Nusa Tenggara/Maluku: Compared with Java/Bali, the region of Nusa Tenggara/Maluku has advantages in its higher return to occupational choice,
and its higher return to work experience. However, the region s disadvantages compared with Java/Bali are stark: Nusa Tenggara/Maluku has a far lower
level of access and assets, lower return to education, far higher levels of informal employment and generally lower return to most accesses. The remoteness
of Nusa Tenggara and the islands of Maluku, and their particular soil and climate conditions, translate into lower levels of access to basic services and
infrastructure. This inhibits the accumulation of human capital by the poor, and curtails access to markets and the diffusion of new technologies. Had the
level of access and assets of Nusa Tenggara/Maluku been similar to Java/Bali s, the incidence of poverty would have decreased by a huge 18.8 percentage
points.
Papua: The region of Papua has several advantages over Java/Bali. These include its higher return to occupational choice, its far higher return to work
experience, its higher return to road access, and its slightly higher return to access to credit. However, Papua is at a major disadvantage to Java/Bali in
many areas, the most important of which include its far lower return to education, its generally lower level of access and assets, and its significantly higher
levels of informal employment. If Papua s labor market structure were similar to Java/Bali s, poverty incidence would have decreased by 4.8 percentage
points.
Source: World Bank staff estimates.
65
See Annex III.1 for a description of the methodology behind this decomposition.
53
IV
The Determinants of Recent Poverty Changes: A Dynamic Analysis
Why did poverty decrease substantially between 1999 and 2002, while inequality increased? This section
focuses on this more recent period of Indonesias history of poverty reduction. In particular, it explores the issues, highlighted
in Chapter 2, flagging the fact that, although important inroads have been made in poverty reduction since the crisis, the
rate of pro-poor growth appears to have slowed down. During this period, poverty declined from 23.4 percent to 18.2
percent, but inequality increased, with the Gini ratio rising from 31.7 percent to 34.0 percent.
This analysis recognizes the complexity of the dynamics of poverty reduction, using a micro-simulation
decomposition method. The method employs microeconomic simulations to decompose changes in poverty over two
different years (1999 and 2002) and across the six different regions into the contributions of four sets of phenomena:66
(i) Endowment effect: changes in the socio-demographic structure of the population, i.e. the assets and personal or
household characteristics in the population (also referred to as the population effect);
(ii) Price effect: changes in the structure of earnings/expenditures, i.e. the returns to those assets and characteristics;
(iii) Occupational choice effect: changes in occupational-choice behavior, i.e. how people use those assets and
characteristics in the labor market; and
(iv) Error term: shocks or unobservable factors.
The analysis provides valuable insights into the determinants of recent changes in poverty and inequality,
with implications for strategic policy going forward. The two decomposition approaches consist of simulating
counterfactual expenditure by changing how markets and households behave, one aspect at a time, and by observing the
effect of the change on the poverty headcount, while holding all other aspects constant. In order to obtain the simulated
price effect, in the 1999 sample the expenditure of households is replaced with the value obtained using the function
estimated for 2002, while keeping the observable and unobservable characteristics of the households constant. A symmetric
definition applies to the occupational-choice effect, while the endowment effect serves as the residual after removing
other effects from the actual changes observed during 1999-2002. Table 3.5 shows the decomposition results using the
two approaches.
Endowment effect: endowment changes were the main contributor to poverty

reduction
An increase in endowments was the key factor in contributing to poverty reduction over the period.
Socio-demographic structures were modified with increases in the average educational level of the population, the
replacement of older cohorts by younger, better educated and more productive cohorts, and the decline in fertility and
family size. Changes in access to basic infrastructures and facilities also impacted endowments of the poor. Figure 3.10
provides the mean change and the incidence of change (by expenditure groups) for various endowments, where a negative
slope in the curve indicates a pro-poor increase in a particular endowment.
The primary endowment effect that reduced poverty in this period was the increase in educational
attainment. The bottom quintile households saw a 2.2 percent increase in the education of household heads and
similar increases for other household members (Figure 3.10a). In addition, the endowment effect related to the reduction
66
Annex III.4 describes this method in greater detail, including the merits of using the micro-simulation approach. It also explains the methodology behind arriving at coefficients using an
absolute approach and a relative approach.
54
in fertility and household size also helped to reduce poverty (Figure 3.10b). Endowments also improved as measured by
increased access to informal courses and telecommunications. The increase in access to informal courses was 1.4 percent
for the poorest quintile and 3.9 percent for the mean income group (Figure 3.10c). The bottom quintile experienced a very
large (10 percent) increase in access to telecommunications which was higher than the increase in access at the mean at
7.9 percent (Figure 3.10d).
The overall endowment effect benefited the rich more than the poor and was a factor in increasing
inequality during the period. There were larger reductions in access to asphalt roads and to credit among poorer
households, and smaller gains in access to informal courses (Figure 3.10c, e and f). The bottom quintile experienced a
decline in access to asphalt roads by 3.0 percent in this period, compared to 0.7 percent decline for the mean income
group. Only the richest three-quarters of the population enjoyed better access to asphalt roads in this time period. The
bottom quintile households also experienced a decline in access to credit (2.8 percent). Meanwhile, although there was
an increase in access to informal courses for the poor (1.4 percent), the increase was larger for richer percentiles, which
consequently also had an inequality-increasing effect.
Figure 3.10 Changes in selected endowments for the poor, 1999-2002
(Growth incidence of various endowment variables on percentiles)

a. Educational attainment increased,
especially for the poorest
b. Fertility declined especially

for the near-poor
c. Access to informal courses (computer and

mechanical) increased, benefiting mostly the rich
d. Access to telecommunications increased,

especially for the poor
e. Access to asphalt roads decreased for

the poor and general population and only
increased for the richest percentiles
f. Access to credit declined for the poor but

increased for the average income group and
the rich
Source: Susenas 1999, 2002 and World Bank staff calculations.

Note: Growth incidence of each variable across expenditure percentiles is shown by the red curve. Growth rate at the mean per capita expenditure is shown by the straight line. The percentiles where the red curve
is above zero increased their endowment in the particular variable in the period 1999-2002.
55
Price effect: the poverty impact of the return to assets and access was mixed
The return to assets and access over the period was poverty reducing in net. The price effect was similar to
the endowment effect in increasing inequality over the period, since the return in net was higher for the non-poor. Despite
these net effects, there was heterogeneity in the impact of the return to different assets, with the significant increase in the
return to education of household heads dominating. It is useful to review how the return to each of these assets tended to
either contribute to, or detract from, poverty reduction and equality over the period.
Table 3.5
A significant increase in the return to education among household heads reduced poverty. Comparing
1999 and 2002, it is apparent that the return to education increased both in urban and rural areas (Table 3.4). While
in 1999 urban heads of household with a junior secondary education could expect a 9.5 percentage-point increase in
their household expenditure relative to non-educated heads of household, by 2002 this increase had risen to 17.2
percentage points. The observed increase in the return to education implies a widening of the consumption gap
between higher-educated household heads and those with lower levels of education. The absolute approach indicates
that the increase in the return to education of household heads contributed to an impressive 2.6 percentage-point
decrease in the poverty headcount. In contrast, the relative approach indicates that changes in the return to householdhead education increased (relative) poverty by 0.4 percentage points, demonstrating the inequality-increasing impact
of this effect (Table 3.5).
The decline in the return to education for other household members tended to increase poverty and
inequality, and decreased both in urban and rural areas. On one hand, as members of poor households
have lower education, the decline in the return to education of non-household heads helped to reduce inequality. On
the other hand, poor households have larger household sizes and, as a result, the total number of household members
affected by the decline in the return to education was higher for poor households. This second effect was dominant
and so the change in the return to education for other household members slightly increased inequality.
Decomposition of changes in poverty incidence by endowments, 1999 to 2002
(Percentage point change)

Initial values (%)
End value (2002) (%)
Change in poverty
Growth effect
Distribution effect
Overall price effect
Gender
Education of household head
Experience
Other member education
Asset: household status
Density
Access to secondary education
Access to informal courses
Access to credit
Access to infrastructure
Access to communication
Occupational choice effect
Household occupation return
Other member occupation return
Endowment of occupational choice (oc)
Constant effect
Non observables variance
Price + residual variance + occupational effect
Population effect
Source: Susenas, 1999 and 2002.
56
Absolute Approach
23.4286
17.6002
-5.8284
-8.2180
2.3900
1.6014
-1.7504
-2.5864
3.5309
1.2774
-0.8124
1.6541
-0.1131
0.0535
-0.4024
0.0636
0.6866
-1.5536
-3.5913
0.0504
1.9874
-1.1401
1.1451
0.0529
-5.8813
Relative Approach
23.4286
17.6002
-5.8284
-8.2180
2.3900
-0.6699
-0.1420
0.4086
-0.2176
0.1630
-0.1951
-0.5493
-0.1344
-0.0592
0.3539
-0.0279
-0.2699
1.3901
0.9034
0.2576
0.2290
0.0683
0.7884
1.6016
The effect of the decrease in the return to experience in the labor market was actually povertyincreasing, while reducing inequality. More experienced household heads are less likely to be poor. This
therefore means that the rich are hurt disproportionately more by a decline in the return to experience and, hence, this
effect reduces inequality.67 The experience price effect increased absolute poverty substantially by 3.5 percentage
points.
The increased return to access to credit caused a decrease in poverty, but was inequality increasing.
While the return to access to credit increased substantially in urban areas, it decreased slightly in rural areas. As the
increase in the return to access to credit in urban areas more than compensated for the effect of a lower return in rural
areas, the overall price-effect of access to credit reduced the incidence of poverty by 0.4 of a percentage point. This
effect was also inequality increasing. This is because, when richer households have higher access to a particular
factor, an increase in the return to this higher access will also increase inequality.
Occupational-choice effect: changes were poverty reducing but inequality

increasing
With the simulated occupational-choice movement, one can assess the effect of changes in occupational
behavior on the expenditure distribution. This is done by using the sample of 1999 and deriving for each household
member what their household expenditure would have been if their occupation behavior had followed the simulated
choices. The results of all these factor changes appear in Table 3.6, which demonstrates the movement towards informal
and agriculture sectors in the post-crisis period.68
Table 3.6
Post-crisis movement towards informal and agriculture sectors

Share (%)
1999
Agriculture
Industry
Services
Net movement (% point)

2002
Simulated
Actual
Simulated
Inactive
Formal
Informal
Formal
Informal
Formal
32.5
6.22
18.58
10.99
1.9
11.97
34.07
4.09
21.19
10.79
2.62
10.4
32.53
4.1
26.5
9.43
2.62
9.68
1.57
-2.13
2.61
-0.2
0.72
-1.57
0.03
-2.12
7.92
-1.56
0.72
-2.29
Informal
17.83
16.85
15.14
-0.98
-2.69
There was a poverty-increasing movement towards the informal sector and agriculture over the period.
The change in occupational-choice behavior between 1999 and 2002 was dramatic. In net terms, about 6.4 percent of the
entire working-age population switched from formal into informal employment, with the major shifts being towards informal
agriculture and informal industry.69 A movement towards the two sectors yielding the lowest occupational returns is
bound to increase poverty. The labor movement simulated an increase in poverty incidence of 1.98 percentage points
from 1999 to 2002.
67
Expenditure tended to be concave functions of years of experience, with the coefficient of experience squared being generally strongly significant and negative. In general, the return to one additional year of
experience is smaller for older workers. This is in contrast to the return to education being higher for those with higher education.
The effect of occupational behavior is analyzed through a multi-nominal logit model. Distinct models are estimated for household heads, spouses and other household members, and for each group within
urban and rural areas. The models include schooling, experience, age and gender composition of the household, and the availability of cultivable land (owned or rented) for all individuals and some
characteristics of the household head as explanatory variables.
69
Although this simulated evolution goes in the direction actually observed, the magnitude of the simulated evolution is far greater. Without changes in the socio-demographic structure of the population, net
movement towards informal agriculture would become 7.92 percent of all working-age, instead of the 2.61 percent switch observed. Interestingly, the model exactly predicted the level of actual movement
towards informal industry from 1.9 percent share in 1999 to 2.62 percent in 2002, and outside formal agriculture with prediction of 2.12 percent net loss.
68
57
This labor movement across sectors hurt the poor more than the non-poor and hence also had an
inequality increasing effect. That this sectoral shift in occupations is inequality-increasing is also not surprising,
since the movement towards the informal sector primarily affected poor households. Indeed, the rate of net entry into
informal agriculture decreases with expenditure (see Figure 3.11). As formal sector jobs experienced larger increases in
return and, at the same time, the increases in urban areas were larger, the non-poor obtained larger gains. Hence, these
changes in return contributed to an increase in inequality.
Figure 3.11 Movement towards the informal sector was greater among the poor
Agriculture
Industry
Source: Susenas, 1999-2002.
Shocks and unobservable effects: changes increased poverty and inequality

Unobservable factors captured in the error term in the expenditure function increased over the period,
simultaneously increasing poverty and inequality. The results of the decomposition suggest that the increase in the
dispersion of unobservable factors substantially increased poverty by 1.14 percentage points. By definition, it is impossible
to identify what was behind this unobservable phenomenon. For example, it could have been an economic factor, such as the
loss of a job and an inability to mitigate or cope with such a shock. The risk and vulnerability analysis in Chapter 6 on Social
Protection goes into further detail about the nature of the shocks faced by the poor and their impact on poverty.
Conclusion: Poverty Diagnostics Provide Pointers for Poverty

Reduction Efforts
The nature and profile of poverty today, and the determinants of poverty reduction over the recent past,
provide pointers for a future strategy for poverty reduction. This chapter has laid out the nature of poverty in
Indonesia, providing profiles of the poor and looking into which factors have determined not only poverty but also poverty
reduction in the more recent past. This analysis reinforces the message for a multi-pronged poverty reduction effort going
forward. It also raises questions, some of which will be taken up in this report, while others are beyond the scope of this
report and set aside for future analysis. In this conclusion, and by way of transition, key lessons learned are highlighted,
as they motivate the four-pillared approach to poverty reduction in Indonesia laid out in the following chapters.
Ensuring that growth benefits the poor must be a key component of any poverty reduction strategy for
Indonesia. Most importantly, given the dense concentration of income distribution in Indonesia (Figure 3.1), it intuitively
follows that anything that can be done to shift the distribution to the right will rapidly reduce poverty, even if growth is
equitably shared by the distribution (and not only if it disproportionately benefits the poor). Clearly, growth that fails to
benefit the lower tail of the distribution will not help the poor. So Indonesia needs growth, and growth that accrues to the
poor. The poverty diagnostics presented in this chapter provide a few key pointers. First is with regards to sectors and
58
occupations: the poor still work predominantly in agriculture and the informal sector. What has been positive for the poor
in terms of the growth experience in recent years is that productivity in these less remunerative sectors has been increasing,
improving the plight of the poor in the process. What has been less positive about the growth experience in recent years
is that the net movement of the poor has been away from, rather than towards, more remunerative employment in the
formal and service sectors. What can be done to make these pathways out of poverty work better for the poor, not only in
densely populated Java/Bali, but across regionally diverse Indonesia? Second, and related, it would appear from these
poverty diagnostics that investing in the endowments of the poor that are particularly important in reducing poverty and
connecting them to factor and product markets will be crucial. The diagnostics in this chapter show that improved educational
endowments and going beyond primary education is a key factor in reducing poverty, and that recent poverty reduction
progress has been dominated by both the endowment and return effects on human capital. Investing in the poor will be
essential. Also key to income growth among the poor is access to infrastructure. Importantly, access to roads and credit
has enabled the poor to link into markets. Connecting the poor to growth will be key.
Ensuring that public spending on services work for the poor will be critical in addressing poverty reduction.
Indonesia is lagging on some key non-income poverty output and outcome indicators, such as access to safe water and
sanitation, secondary school enrollment, maternal mortality and child malnutrition. Clearly, investing appropriately in
health and education services that will allow Indonesia to achieve its goals in these areas will be crucial to the challenge
of spending better for poverty reduction. Moreover, income poverty let alone non-income dimensions of poverty is not
surprisingly critically dependant on access to public services that help improve non-income dimensions of poverty. For
example, the diagnostics presented here show that access to secondary education and informal courses (especially in
urban areas) correlate with poverty. The diagnostics also raise some issues related to the quality of outcomes and service
delivery, with the return to access of certain key public services among the poor (such as to secondary education and
roads) actually declining in recent years. This raises the question of not only which sectors to spend on, but also how best
to spend in these sectors to enhance the return to the poor.
Ensuring that social protection works for the poor must be a key element of any poverty reduction
strategy. The diagnostics presented in this chapter clearly reveal the vulnerability of Indonesia s poor and near-poor to
shocks. As mentioned in the World Development Report 2000-01, vulnerability and insecurity are important dimensions
of poverty in and of itself. Again, this is intuitively clear in looking at the Indonesian income distribution in Figure 3.1 and
the gradient of the distribution at the poverty line. Any decreases to incomes among households anywhere close to the
poverty line will have a major impact on poverty. Moreover, the diagnostics show that there is a high level of churning
and transient poverty, as this dense population around the poverty line, including the near-poor, moves in and out of
poverty from one year to the next, depending on shocks. (The nature of these risks, vulnerabilities and shocks is discussed
further in Chapter 6 on Social Protection). Related to making spending work for the poor is the question of how best to put
together a social protection system that addresses the high vulnerability to poverty among Indonesians. This is a question
that becomes increasingly appropriate to ask and address as Indonesia joins the ranks of middle-income countries and
makes progress in reducing poverty to more manageable levels.
Finally, poverty diagnostics also reaffirm the importance of the government itself being able to work for
the poor and deliver on the above agenda. Putting into place systems that ensure poverty-focused planning and
budgeting in government will be a key part of this economic governance agenda for poverty reduction. But publicspending allocations and policy decisions alone will not automatically lead to poverty reduction. Evidence has shown that
public service delivery is failing the poor (World Bank, 2006k). Programs need to be implemented and policy enacted and
enforced. Enhancing accountability of government, as well as capacity, becomes crucial to this effort. In an archipelago
the size of Indonesia, with weak capacity, this is a major issue. Moreover, given the vast regional disparities noted in this
chapter, the strategies relevant for poverty reduction will invariably need to be fine-tuned from one region to another.
Decentralization, if it works well, offers the best vehicle for achieving this. There is scope for considering how the
decentralization framework could be better used to address the challenge of poverty reduction in its diversity.
59
Focus on Sumatra
Sumatra is a vast island, almost four times the size of neighboring Java, but
with a population of only one third the size of Java s. It contains seven provinces
and numerous ethnicities, and is rich in natural resources. Indeed, its resources
have long been a major factor in driving the Indonesian economy. Medan,
Indonesia s third-largest city and its largest city off Java, is located in North
Sumatra.
Population (m)
Population density (per km2)
Avg life expectancy (yrs)
45.2
91
69.3
Infant mortality (deaths per 1,000 live births)

Avg land-holding size (m2)
Proportion of total poor (%)
31.7
231.0
21.8
Number of poor (person)
7,879,861
Poverty map per district, Sumatra
Local economy: The economy of Sumatra is of considerable significance to

the national economy, rich in oil and gas and an important producer of palm
oil, rubber, coffee and timber. The region s economy is well balanced between
agriculture, the extractive industries and manufacturing, which together account
for about 63 percent of the region s GDP. Sumatra ranks third of the six regions
in per capita revenue, far behind Papua, and slightly behind Kalimantan, with
Rp 784,000 (US$83) per capita revenue in 2004. Of this sum, 56 percent comes
from the governments General Allocation Fund (DAU), while a further 2.4
percent comes from the Special Allocation Fund (DAK). The three largest
components of spending per capita go on government and administration (29
percent), education (28.3 percent) and public works (11.3 percent). Although
the regions largest spending item is government and administration, as a
proportion of total spending this is the lowest for any of the six regions.
The structure of Sumatra s economy, 2004
Where the poor work in Sumatra
Source: Central Bureau of Statistics, National Socio-Economic Survey (Susenas), 2004.
60
Socio-economic indicators: With 20.8 percent of Indonesia s total

population, Sumatra contains a similar proportion of the country s poor, at
21.8 percent, and has a poverty headcount of 17.5 percent, slightly higher than
the national average of 16.7 percent. Notwithstanding Sumatra s relatively
prosperity, the region still contains significant pockets of poverty, particularly
in Aceh, the islands off the west coast of Aceh including Nias, coastal areas of
South Bengkulu, and inland areas of North Lampung. Conversely, the lowest
incidence of poverty is found around the major cities of Medan, Padang and
Palembang, in Pekanbaru and Jambi, and on the island of Bangka.
The regions workforce is concentrated in the agricultural sector, which accounts
for 56.8 percent of all workers. Of this number working on the land, 87 percent
work in the informal sector. Overall, more than 68 percent of the entire workforce
is engaged in informal work, giving rise to insecurity and financial vulnerability.
At almost 12 percent, Sumatra is third to the Java/Bali region and Kalimantan
in the proportion of the workforce employed in industry, with 32 percent involved
in the service sector. While official unemployment is relatively low at 6.5 percent,
the region has the second-highest level of unemployment of the six regions,
second only to Java/Bali with 7.2 percent.
Where the poor work in Sumatra: Three-quarters of the poor in the

workforce are subject to the insecurities of the informal sector, with 72 percent
working in agriculture, 6 percent in services and 4 percent each in the
transportation, construction and industrial sectors.
Non-monetary poverty indicators: Many indicators place Sumatra in third
position, well behind Java/Bali, but close to Kalimantan. Overall, average life
expectancy is almost two years less than in Java/Bali, at 69.3 years. At 97
percent almost everyone in Java/Bali has access to electricity, in Sumatra only
78 percent of the population has electricity, close behind Kalimantan at 80
percent. Clean water is only available to 62 percent, behind Java/Bali at 78
percent and Sulawesi at 63 percent. But in terms of clean water Sumatra is
ahead of Kalimantan, where only 46 percent have access. For child malnutrition,
Sumatra is also third, at 28 percent, higher than Java/Bali (23 percent) and,
surprisingly, Papua (24 percent). But Sumatra does well compare with all the
other regions in education. The region has the highest primary school
enrollment ratio of 93.7 percent, and increases its lead in junior secondary
high-school enrollments (69.3 percent) and senior secondary high-school
enrollments (48.5 percent). Java/Bali is Sumatra s closest rival in school
enrollments, with 93.4 percent, 66.1 percent and 42.4 percent, respectively.
With such positive educational indicators, Sumatra s poor stand a relatively
better chance of escaping from poverty.
Living conditions of every 10 people in Sumatra

Poor
Near-poor
Non-poor
Total
Indonesia avg
comparator
Do not have access safe water

Do not have access to decent sanitation
Do not have electricity
5.7
7.1
3.4
5.1
6.5
2.7
4.0
4.5
1.4
4.6
5.5
2.1
4.4
4.9
1.1
Live in villages without secondary school

Live in villages without a telephone
5.1
6.3
5.1
6.2
4.1
4.5
4.5
5.2
4.2
3.6
Are illiterate
Below the age of five are malnourished
Below the age of five were delivered by a traditional untrained midwife (dukun)
1.0
3.6
3.6
0.9
2.8
3.0
0.6
2.5
1.8
0.8
2.8
2.5
1.1
2.5
3.2
Poverty profile: Close to three-quarters of all poor and near-poor

households live in rural areas. Close to 60 percent of poor households
have more than five members, compared with only 48 percent of poor
households nationally. Looking at non-monetary indicators, almost six
poor households in ten have no access to safe water, while more than
seven in ten have inadequate sanitation. Worse than the national average,
3.4 out of 10 poor households have no electricity. However, somewhat
higher educational standards are reflected in Sumatras poverty profile,
with higher levels of school enrollment leading to less than one poor
person in ten being illiterate. This compares favorably with a national
level of illiteracy among the poor of 1.6 out of 10. Nonetheless, poor households
with malnourished children under the age of five remain higher than the national
average, with 3.6 out of 10 poor children insufficiently fed. This compares
with only 2.8 out of 10 poor malnourished children nationally.
In terms of difficulties experienced by the poor in gaining access, 7.6 percent
have no asphalt road in their village, while 82.6 percent have no bank or credit
facilities in their village. More than six poor households in ten live in villages
with no telephone and half have no secondary school. In Sumatra, the poor
have one paramedic for every 50 households, similar to the national average.
61
Focus on Java and Bali

Home to almost 61 percent of Indonesia s population, the crowded islands of
Java and Bali are the hub of the country s political and economic life, with the
unique characteristics of Bali as a famous tourist destination. Stunning beauty
and an active chain of volcanoes notwithstanding, Java and Bali are Indonesia s
most industrialized and urbanized islands, with two major industrial centers of
Greater Jakarta and Surabaya. But over-population, together with a relative
dearth of natural resources, also means that pockets of poverty exist in Java.
Population (m)
131.7
980
71.1

25.0
177.7
57.3
20,727,966
Poverty map by district, Java and Bali
Local economy: The relative industrialization and sheer size of the economy
of Java/Bali dominates other regions. As a result, the percentage shares of
manufacturing, trade, restaurants and hotels, and financial services are the
highest of the six regions, while Java/Balis agricultural share is the lowest.
Indeed, agriculture comprises less than 12 percent of the region s total GDP,
while manufacturing accounts for over 30 percent. Reflecting its relatively scarce
natural resources, only 2.6 percent of Java/Bali s GDP comes from extractive
industries, compared with 57 percent in Papua. As a result of its high population,

the region has the lowest per capita allocation of the central government s DAU,
at only Rp 268,900 per capita (US$29).70 This sum is tiny compared with the
per capita DAU in Papua, at Rp 1,834,200 (US$195). Java/Bali also receives
the lowest total revenue per capita and per capita government spending of
across all sectors of the six regions. For example, health spending is less than
20 percent per capita of that found in Papua.
The structure of Java/Bali s economy, 2004
Where the poor work in Java/Bali
70
62
Using 2004 exchange rate of Rp 9,400 per US$1.
Where the poor work in Java/Bali: Those from poor households in Java/
Bali work primarily in agriculture (56 percent), with far fewer working in the
next most important sectors of trading and industry, at 14 percent and 11
percent, respectively. Despite the region s relative prosperity, the poor remain
closely tied to the land and rural life. Indicating a high level of vulnerability,
7.2 out of 10 workers from poor households work in the informal sector. Among
workers from poor families, unemployment in Java/Bali is the highest of the
six regions, at 12 percent.
Socio-economic indicators: With 60.6 percent of Indonesia s total

population, Java/Bali has 57.3 percent of the country s poor with a poverty
headcount of 15.7 percent, slightly less than the national average of 16.7
percent. This places the region in the paradoxical position of having the
country s lowest incidence of poverty while containing the largest number of
poor of any of the six regions. While Bali and the Greater Jakarta areas have
the lowest incidence of poverty in the region, poor kabupaten are found in
western Central Java and on the island of Madura, East Java. Additionally, the
region has the lowest workforce proportion engaged in agriculture, at only 36
percent, compared with 83 percent in Papua. Industry absorbs 22 percent of
the workforce, while services employ a further 42 percent. Overall, Java/Bali
has the lowest proportion of workers in the informal sector, at about 52 percent.
Conversely, Papua has the highest overall informal sector ratio, at over 78
percent. With lower levels of informal employment, the region offers higher
job security than elsewhere. However, 7.2 percent of the workforce is
unemployed, the highest level of all six regions.
Non-monetary poverty indicators: Java/Bali measures up comparatively

well with other regions. At 71.1 years, the region has the longest life expectancy,
while at 25 deaths per 1,000 live births it also has the lowest infant mortality
rate. The region has the highest number of households with electricity, at over
97 percent; Kalimantan comes a distant second, with 80 percent. In addition,
the region has the highest levels of access to clean water, at 78 percent, good
sanitation, at almost 40 percent, and the lowest level of child malnutrition, at
23 percent. Of these indicators, the worst are Papua (34 percent), Nusa
Tenggara/Maluku (21 percent) and Nusa Tenggara/Maluku (38 percent),
respectively. Although behind Sumatra, the region has Indonesia s secondhighest school enrollment at all levels of the primary and secondary education
system.
Living conditions of every 10 people in Java and Bali

Are illiterate
Poor
Near-poor
Non-poor
Total
Indonesia avg
comparator
4.8
7.1
0.6
4.9
3.9
1.6
2.5
4.9
4.5
5.9
0.3
4.6
3.5
0.9
2.4
2.2
3.9
3.3
0.1
3.6
2.1
0.9
2.3
2.2
4.2
4.5
0.2
4.0
2.7
1.1
2.3
3.2
4.4
4.9
1.1
4.2
3.6
1.1
2.5
3.2
Poverty profile: The poverty profile indicates 6.3 out of every 10 poor
households are found in rural areas, only slightly less than the national average.
However, only 42 percent of poor households have more than five family
members, less than the 48 percent national average for the poor. Looking at
non-monetary indicators, while 96 percent of poor households have electricity,
almost half of all poor households have no access to safe water, while more
than 7 out of 10 have no decent sanitation. Similar to the national average, 5.4
out of 10 poor households have less than primary school education while 1.6
out of 10 of the poor is illiterate. While the level of literacy in the region is on
a par with the national average, this lags far behind Sumatra, Indonesia s best
performing region. Malnourishment among children from poor households
remains high, at 2.5 out of 10, and half of all births in poor households are
attended by an untrained traditional midwife (dukun).
Access of the poor to markets is important and despite the regions infrastructure,
over 8 percent of poor households still live in villages without asphalt road
and 78 percent have no access to telecommunications. Moreover, access of
the poor to credit is low, with over 75 percent of the poor have no bank or
credit facilities in their village and access to education is limited because
secondary schools are not available in the villages of poor households in 49
percent of cases. However, the region has the best paramedic coverage, with
one paramedic for every 42 poor households.
63
Focus on Kalimantan
Kalimantan, the Indonesian portion of the island of Borneo, covers a huge area
of some 5.4 million km2 but has a small population of less than 12 million.
This makes it one of Indonesias most sparsely populated regions, less sparse
than only Papua. Although Kalimantan has several major commercial and
industrial centers, such as Samarinda, Balikpapan and Banjarmasin, which
are based on timber and the extractive industries, large tracts of the regions
interior remain remote and accessible only by river.
Population (m)
11.9
21
69.3

31.8
308.3
3.6
1,303,586
Poverty map by district, Kalimantan
Local economy: Kalimantan s economic wealth is largely based on timber,

mining and the oil and gas industry, with precious stones also important. The
region s economy has Indonesia s second-smallest agricultural sector,
contributing only 15.6 percent, with the largest contributions to regional GDP
coming from the extractive industries (24.5 percent) and manufacturing (28.6
percent). Kalimantan has Indonesia s two largest coal mines. This means that
as a contributor to the regional economy, manufacturing in Kalimantan is almost
as high as in Java/Bali, where it contributes 30.2 percent. Kalimantan has
Indonesia s second-highest per capita revenue at Rp 1,471,000 (US$156),

second only to Papua. Of this revenue, about 46 percent comes in the form of
the DAU and DAK, a lower proportion than all regions with the exception of
Java/Bali. In terms of government spending, with the exception of Papua,
Kalimantan has the highest per capita spending on public health, education,
housing, agriculture and mining. As a proportion of spending, Kalimantan
devotes more than any other region to public works and transportation.
The structure of Kalimantan s economy, 2004
Where the poor work in Kalimantan
64
Socio-economic indicators: With only 5.4 percent of Indonesia s total

population, Kalimantan contains only 3.6 percent of the country s poor and
has a poverty headcount of 11.0 percent, significantly lower than the national
average of 16.7 percent. Kalimantan has made considerable progress towards
development and, unlike any other region, not one kabupaten has a poverty
incidence of over 30 percent, even in the remote interior of northern East
Kalimantan. Kalimantan s workforce distribution is very similar to Sumatra s,
52.4 percent being concentrated in the agricultural sector. Of these agricultural
workers, a large 90 percent work in the informal sector. Overall, more than 63
percent of the entire workforce is engaged in informal work. At 15.7 percent,

Kalimantan is second only to the Java/Bali region in the proportion of the
workforce employed in industry, with 31.8 percent involved in the service sector.
Official unemployment is relatively low, at 5.2 percent.
Where the poor work in Kalimantan: Seventy-one percent of the poor in
Kalimantan work in agriculture, followed by 8 percent in trading and 6 percent
in industry. Of all those from poor households who do work, three-quarters
are engaged in the informal sector with little job security.
Living conditions of every 10 people in Kalimantan

Poor
Near-poor
Non-poor
Total
Indonesia avg
comparator
6.9
6.6
4.9
5.4
4.4

Are illiterate
7.5
3.3
4.8
5.7
1.4
3.4
5.1
6.8
2.7
4.6
5.7
1.1
3.8
4.4
4.9
1.5
3.8
4.5
0.8
3.3
3.0
5.6
1.9
4.1
4.9
0.9
3.4
3.5
4.9
1.1
4.2
3.6
1.1
2.5
3.2
Non-monetary poverty indicators: Despite this apparent strength, overall

non-monetary indicators of poverty for the region are mixed. Only 46 percent
have access to clean water, well behind all other regions with the exception of
Papua, although 30 percent have access to sanitation, a higher level than in
either Papua, or Nusa Tenggara/Maluku. Indicating a higher level of
industrialization and urbanization, 80 percent of the residents of Kalimantan
have electricity, second only to Java/Bali, at 97 percent. But despite this,
Kalimantans level of child malnutrition is high, at 32.5 percent, with only
Nusa Tenggara/Maluku worse, at 37.5 percent. In terms of education,
Kalimantan is on a par with Java/Bali for primary school enrollments, at 93.4
percent, but as children move up through the school system so enrollments
more closely resemble Sulawesis pattern, with 60.3 percent enrolling in junior
high school and only 38.2 percent in senior high school. As such, Kalimantan
lies in the middle of the six regions in terms of school enrollments.
Poverty profile: Kalimantan s poverty profile indicates that the poor continue
to live primarily in rural areas, where 7.5 out of 10 poor households are to be
found. This is higher than the national average of 6.9 out of 10 poor households.
Another sign of poverty, over 60 percent of poor households have more than
five family members, the highest level of the six regions. Looking at nonmonetary indicators, poor households in the region experience higher levels
of poverty than average poor households across the country. For instance, 6.9
out of 10 poor households have no access to safe water, compared with the
national poor households average of 5.2 percent. Three-quarters of poor
households in Kalimantan have inadequate sanitation, slightly higher than the
national average. While 2.1 out of 10 poor households across Indonesia have
no electricity, in Kalimantan 3.3 out of 10 poor households are without electricity.
Child malnourishment is also rather prevalent among poor households, with
3.4 out of 10 children under five malnourished, compared with 2.8 out of 10
for poor children nationally. Slightly below the national average for the poor,
5.5 out of 10 people from poor households have received less than primary
school education, while illiteracy is close to the average for the poor, at 1.6 out
of 10.
In terms of difficulties experienced by the poor in gaining access to markets,
almost three times as many poor households in Kalimantan live in villages
without an all-weather road as compared with average poor households in
Indonesia, highlighting the difficulties of transportation in the more remote
areas of the region. Over 81 percent of the poor have no bank or credit facilities
in their village, while a similar proportion have no access to
telecommunications. Slightly less than half of all poor households are found
in villages with no secondary school.
65
Focus on Sulawesi
The strangely shaped island region of Sulawesi contains a diverse mixture of
ethnic groups and religions. Close to one-third of the land area of Kalimantan,
Sulawesi has a somewhat larger population, at 15.6 million, and consequently
a far higher density of 80 people per km2, somewhat less than Sumatra. The
region has one major city and port, Makassar in the south, but is otherwise
predominantly rural.
Population (m)
15.6
80
69.4

31.2
196.7
7.2
2,597,716
Poverty map by district, Sulawesi
Local economy: The regional economy of Sulawesi is heavily reliant on

agriculture, which comprises over 34 percent of the region s GDP the highest
agricultural contribution of any of Indonesias six regions. While there is some
mining and manufacturing in Sulawesi, their contribution is modest, together
accounting for less than 18 percent of the region s GDP. Trade, tourism and
services together account for 27.2 percent of the region s economy, with the
services component the second-highest of the six regions, at about 15 percent.
Revenue per capita is closed to the national average, at Rp 889,000 (US$95),
but this accrues largely in the form of the DAU and DAK. Indeed, Sulawesi
receives the second-highest proportion of any region from the DAU and DAK,
at 74 percent, close behind the highest recipient, Nusa Tenggara/Maluku, which
receives 77 percent in government grants. In terms of government spending,
Sulawesi devotes the lowest proportion per capita of the six regions to public
works, but the highest proportion per capita to education.
The structure of Sulawesi s economy, 2004
66
Where the poor work in Sulawesi
Socio-economic indicators: With only 7.2 percent of Indonesia s total

population, Sulawesi also contains exactly 7.2 percent of the country s poor
and has a poverty headcount of 16.7 percent the national average. Pockets of
poverty exist particularly in kabupaten Poso, in Central Sulawesi, exacerbated
by sectarian conflict, and in Gorontalo and Boalemo to the north. The areas
with the lowest poverty incidence center on Pare-Pare, South Sulawesi, and
Minahasa, North Sulawesi, together with the cities of Makassar and Palu. While
Sulawesi s workforce is heavily engaged in agriculture, with almost 57 percent
working in this sector, services are also important, employing almost 34 percent.
Conversely, less than 10 percent of workers are involved in manufacturing.

Seventy percent work in the informal sector, with only Nusa Tenggara/Maluku
and Papua having higher levels of informal employment. Official unemployment
accounts for 6.4 percent of the workforce.
Where the poor work in Sulawesi: Similar to other regions, the poor in
Sulawesi rely heavily on agriculture for employment at 76 percent, followed by
7 percent in trading and 5 percent in industry.
Living conditions of every 10 people in Sulawesi

Poor
Near-poor
Non-poor
Total
Indonesia avg
comparator
5.3
4.8
4.0
4.4
4.4

Are illiterate
7.5
4.5
5.8
7.0
1.8
3.4
5.9
6.3
3.0
5.4
6.2
1.5
3.0
5.3
4.0
1.4
4.5
4.2
1.0
2.4
3.4
5.1
2.3
4.9
5.1
1.2
2.8
4.5
4.9
1.1
4.2
3.6
1.1
2.5
3.2
Non-monetary poverty indicators: Overall, the indicators for the region

are very similar to those of Sumatra s. Seventy six percent have electricity,
while only 34 percent have adequate sanitation and 63 percent have access to
safe water. Malnutrition among the under-fives is also close to Sumatra s level,
at 28.9 percent. However, Sulawesi performs less well in education, with primary
school enrollments at 90.3 percent, only worse in Papua, at 85.3 percent.
Interestingly, enrollments in higher levels of education in Sulawesi improve,
with 39.4 percent of children enrolling for senior secondary school, ranking
third behind Sumatra and Java/Bali.
Poverty profile: Sulawesi s poverty profile reveals the extent to which the
poor are still trapped in agriculture, with 8.7 out of 10 poor households found
in rural areas. This compares with 6.9 out of 10 poor households for the country
as a whole. The size of poor families is large, with 58 percent having more than
five members. In terms of non-monetary indicators, over half of all poor
households have no access to safe water and three-quarters have inadequate
sanitation-levels that are fairly typical nationally. What is less typical is that 4.5
out of 10 poor households in Sulawesi have no electricity, indicating the extent
to which poor households remain disproportionately rural and disconnected

from urban centers. In terms of education, 60 percent of those in poor
households have less than primary education and 1.8 out of 10 remain illiterate.
For children from poor households aged 6 to 14, 7.8 out of 10 are in school,
somewhat lower than the national average for children from poor households
of 8.4 out of 10. Child malnutrition is also rather high, with 3.4 out of 10 poor
children under the age of five malnourished, compared with 2.8 out of 10 poor
children nationally. The likelihood of a child being delivered by an untrained
traditional midwife (dukun) is far higher in a poor household in Sulawesi, at
almost 60 percent, compared with 47 percent for poor households nationally.
In terms of the poor struggling to gain access to markets, 7.9 percent have no
access to an asphalt road. The poor in Sulawesi are the most financially
disconnected of the six regions, with 89.3 percent having no access to credit
or banking facilities, while 70 percent have no access to a telephone in their
village (compared with 49 percent of poor households nationally). Of poor
households, 5.8 out of 10 live in villages that are without a secondary school.
67
Focus on Nusa Tenggara and Maluku

The region of Nusa Tenggara and Maluku (NT/Maluku) covers a vast expanse
of islands and ocean, much of which is remote and only connected to the rest
of the country by limited and infrequent transportation services. Lacking
significant natural resources and major infrastructure, it is not surprising that
this region suffers from high levels of poverty. The region shares some of the
characteristics of Papua in having a relatively small population and low levels
of infrastructure.
Percentage of poor
10.3
62
66.7

41.9
196.2
7.4
2,692,408
Poverty map by district, Nusa Tenggara and Maluku
Local economy: The regional economy of Nusa Tenggara/Maluku is based

primarily on agriculture and fishing, with some tourism until this declined
after the sectarian unrest of 1999 in Maluku and more generally since the 2002
Bali bombing. Maluku is home to some mining and timber industries. As the
fabled spice islands , Maluku also produces cloves and nutmeg, although low
international prices have undermined these two crops as major sources of
revenue. Agriculture contributes 33 percent to the regions economy, with mining
at 15.7 percent and trade and tourism at 15.5 percent. The contribution to the
The structure of Nusa Tenggara/Maluku s economy, 2004
Socio-economic indicators: With only 4.8 percent of Indonesia s population,

Source:
Susenas, 2004. contains 7.4 percent of the country s poor and has a poverty
Nusa Tenggara/Maluku
68
Population (m)
region s GDP from services is the largest of any of the six regions, at 14.8
percent. The region s revenue per capita is about average, at Rp 859,000
(US$91), marginally higher than Sumatra. The region receives the highest
proportion of its revenue in the form of the DAU and DAK, together accounting
for 79 percent of total revenue. In terms of government spending per capita,
Nusa Tenggara/Maluku spends the lowest proportion of all six regions on
transportation, but the highest on health and agriculture, forestry and fisheries.
Where the poor work in Nusa Tenggara/Maluku
headcount of 26.1 percent, well above the national average of 16.7 percent. On
a kabupaten basis, poverty is particularly prevalent in Maluku (North Maluku
fairs slightly better), Sumba, western Flores, West Timor and the small islands
to the east of Wetar. With the sole exception of Ambon, no kabupaten has a
poverty incidence of less than 10 percent. Only in Papua does a higher
proportion of the population work on the land. Over 72 percent of the workforce
in Nusa Tenggara/Maluku works in agriculture (75 percent in Papua), whereas
in Java/Bali only around 36 percent work in agriculture. Compounding this is
the informal nature of employment across the region, with 77 percent of all the
regions workers employed in informal jobs. The level of informal work is even
higher in the agricultural sector, at over 95 percent, higher only in Papua.
Unemployment in the region stands at a low 4.9 percent.
Where the poor work in Nusa Tenggara/Maluku: Up to 75 percent of
the poor work in agriculture, followed by 8 percent in industry and 6 percent in
trading.
Living conditions of every 10 people in Nusa Tenggara and Maluku

Poor
Near-poor
Non-poor
Total
Indonesia avg
comparator

5.1
8.2
5.1
7.3
4.2
5.2
4.7
6.6
4.4
4.9

Are illiterate
5.0
3.8
6.8
1.9
3.4
5.9
4.4
3.9
6.5
1.8
4.0
5.1
2.9
3.7
5.5
1.3
3.2
3.9
3.9
3.8
6.1
1.6
3.5
4.9
1.1
4.2
3.6
1.1
2.5
3.2
Non-monetary poverty indicators: these indicators in Nusa Tenggara/

Maluku exhibit some worrying signs. The region has Indonesias worst life
expectancy, at only 66.7 years (compared with 74 years in Jakarta). The region
has a very high level of child malnutrition, with 35 percent of children under
five being malnourished, far higher than the country s best performing region
of Java/Bali, at 23 percent. Not surprisingly then, infant mortality is the highest
of the six regions, at 41.9 deaths per 1,000 live births, far higher than the
closest level in Papua. While 63 percent have access to clean water, ahead of
Kalimantan and Papua, Nusa Tenggara/Maluku has the country s worst level of
sanitation, at only 20.6 percent, far inferior to even Papua s level of 28 percent.
In education, Nusa Tenggara/Maluku also struggles, with enrollments
throughout the education system behind those of the rest of the country.
Although the region is ahead of Papua, only 85.3 percent of children enroll for
primary school, while this falls to only 48.2 percent for junior secondary school
and 31 percent for senior secondary school. Not surprisingly, almost 20 percent
of the population is illiterate, although this level of illiteracy is still almost
twice as high as in Papua.
Poverty profile: The poverty profile for Nusa Tenggara/Maluku shows that
three-quarters of all poor households live in rural areas. Out of 10 poor
households, 5.8 have more than five family members, compared with only 4.8
out of 10 for poor households nationally. Looking at non-monetary indicators

among poor households, 5.1 out of 10 have no access to safe water, while 8.2
out of 10 have inadequate sanitation facilities. Fully half of all poor households
in Nusa Tenggara/Maluku have no electricity, which compares poorly with the
national average for poor households, where only 2.1 out of 10 have no
electricity. This illustrates the problems and high costs of supplying electricity
to numerous remote islands. Over 60 percent of poor household members
have less than primary school education, with 1.9 out of 10 being illiterate.
Also of concern, only 8.3 out of 10 of poor children aged 6 to 14 are in school,
while half of this age group works as child laborers. Of children below the age
of five, 3.4 out of 10 are malnourished compared with 2.8 out of 10 for poor
children nationally. Almost 60 percent of poor household births in the region
are delivered by a traditional midwife (dukun).
Regarding the difficulties experienced by the poor in gaining access to markets,
more than 10 percent of the poor have no all-weather road in their village. Just
over 8 out of 10 poor families live in villages with no bank or credit facilities.
Similarly, communication is a problem for poor households, with 81 percent
having no access to telecommunications. Thirty-eight percent of poor
households live in villages without a secondary school.
69
Focus on Papua
Indonesias eastern-most provinces, Papua and recently formed West Irian Jaya,
are the country s largest, together covering 22 percent of Indonesia s total
landmass. They are also Indonesia s most remote provinces, being furthest
from Jakarta and the industrial centers of Java and Sumatra. The region is
Indonesia s most sparsely populated, with a population of only 2.5 million
(only 15 percent of Jakartas population). Almost three-quarters of its area
comprise of primary rainforest and Indonesia s highest mountain range.
Population (m)
2.5
6.0
68.4

34.9
156.2
2.7
966,589
Poverty Map by District, Papua and West Irian Jaya
Local economy: Although rich in minerals and other natural resources,

the main economic activities of Papua s residents remain subsistence
agriculture and animal husbandry, together with fishing and hunting. However,
the largest contributor to the region s economy is mining and quarrying, which
accounts for 57 percent of regional GDP. This is followed by agriculture at
just under 20 percent. The region s small population, coupled with high
revenues from industry and central government grants, results in per-capita
revenue for the region that is the highest in Indonesia. Over 58 percent of the
region s revenue comes from the DAU and DAK. As for government spending
per capita, the region has the highest spending across all sectors because of
The structure of Papua s economy, 2004
70
the small population. Curiously, the region allocates the highest proportion of
per capita spending of all six regions to government and administration, at
over 36 percent. Conversely, it spends proportionately the lowest per capita of
all regions on public health and education. The local economy has posted an
impressive average GDP growth of 10 percent over the past 15 years,
outperforming Indonesia as a whole. But fiscal wealth and economic growth
cannot boost development to the extent that this will dramatically benefit the
poor. Despite impressive growth, the region continues to under-perform in
overcoming poverty and improving human development outcomes.
Where the poor work in Papua
Socio-economic indicators: Notwithstanding the economic benefits

stemming from decentralization in 2000 and the 2001 Special Autonomy Law
for Papua aimed at speeding up economic development, the region continues
to register Indonesia s highest headcount poverty level of 38.7 percent. This is
well above the country s second-highest headcount of 26.1 percent in NT/
Maluku and the national average of 16.7 percent. However, only 1.2 percent of
Indonesia s total population and a mere 1.7 percent of the countrys poor reside
in this region. While more than 53 percent of the poor is found in Java, creating
a dilemma for policy-makers. The highest prevalence of poverty is found in
kabupaten Manokwari, together with the costal kabupaten of Nabire and Yapen
Waropen and the highland district of Puncak Jaya. While the lowest incidence
of poverty is in the north and south of Papua and the western half of the bird s
head , west of Manokwari, no kabupaten in this region registers a poverty
incidence of less than 10 percent.
In terms of employment distribution, only 5 percent are involved in industry,
while 73 percent of the population works in agriculture, with the remaining 22
percent working in the services sector. While the official unemployment rate is
low, at 4.8 percent, those working are primarily involved in the informal sector,
which accounts for 78.3 percent of all those in some form of employment.
More worrying, of those working in agriculture, over 97 percent are informally

engaged, making their earnings insecure and vulnerable to shocks.
Where the poor work in Papua: A staggering percentage of the poor in
Papua is involved in agriculture (94 percent). Of the region s poor, only 1
percent work in industry or services. This contrasts with Java/Bali, where 11
percent of the poor work in industry or services.
Non-monetary poverty indicators: Despite the high revenue per capita,
the non-monetary indicators of poverty for Papua are among Indonesia s
highest. Only 34 percent have access to clean water and 28 percent to adequate
sanitation. Only 46 percent have electricity, the lowest level in the country. The
region is the worst performer in education, with only 85 percent of all children
enrolling for primary education, falling dramatically to 48 percent for junior
secondary and 31 percent for senior secondary. Against these alarming
indicators, Papua has the second-lowest level of malnutrition rate for underfives at 24 percent compare with other regions, with Java/Bali at 23 percent.
The continuing prevalence of poverty is illustrated by the fact that 56 percent
have less than primary education and illiteracy stands at 25 percent.
Living conditions of every 10 people in Papua

Are illiterate
Poor
Near-poor
Non-poor
Total
Indonesia avg
comparator
6.9
9.0
8.3
8.4
9.0
3.9
6.1
2.4
5.9
7.4
5.6
6.8
7.3
2.4
1.9
3.0
4.3
4.7
2.1
3.5
3.9
1.3
1.2
1.1
5.7
6.9
5.2
6.1
6.6
2.5
1.8
2.0
4.4
4.9
1.1
4.2
3.6
1.1
2.5
3.2
Poverty profile: The poverty profile of poor households indicates that 9.5
out of 10 poor households live in rural areas, significantly higher than the
average 6.9 households across the archipelago. However, only 4 out of 10
poor households have more than five family members, below the national
average for poor families. Looking at non-monetary indicators, 6.9 out of 10
have no access to safe water, while 9 out of 10 poor households have inadequate
sanitation in their homes and just over 8 out of 10 poor households have no
electricity. Low levels of education prevent many escaping from poverty: almost
4 out of 10 of the poor are illiterate and more than 7 out of 10 have never been
to primary school. Of every ten children aged 6 to 14, only 7 out of 10 are in
school and almost 10 percent already work as child laborers. Over 6 out of 10
poor children under the age of five are malnourished in Papua, compared with
only 2.4 out of 10 in the Java/Bali region. In such circumstances it is extremely
hard for children to escape from the poverty of their parents and far harder than
for many children in poor families in other regions of the country.
The poor in Papua also face difficulties in gaining access to markets. Half of all
poor households are found in villages only accessible by a dirt road, highlighting
the inadequate linkages between the rural poor and the urban economic centers,
while 83.5 percent have no access to bank or credit facilities. Ninety percent of
poor households have no access to a telephone in their village and 8.4 out of
10 live in villages without a secondary school.
71

LACK OF CLEAN WATER CAN HAVE DIRE HEALTH IMPACTS: TULANG BAWANG, LAMPUNG
Tarna s son, Agus, on his way to school. The unsealed road turns to mud in the rainy season.
Family plantation pawned to save the life of a mother

For married couple Tarna and Dasem, their cassava plantation had
been their main source of income and hope for a better future. However,
a little over a year ago, they were forced to pawn the plantation in order
to pay for hospital bills incurred when Dasem experienced
complications during labor. Dasem recalls that during her fifth month
of pregnancy with her second child, she slipped and fell while carrying
a bucket of water home from her neighbors well. Since then she had
experienced pains in her lower back. These became much more intense
during her eighth month of pregnancy, when she also experienced
abnormal bleeding. Soon after, Dasem went into labor prematurely.
72
After withstanding a day of painful labor she finally gave birth, but the
baby was sickly. Dasem s dukun bayi (traditional birth attendant) was
unable to remove her placenta. The one and only midwife in the village
could not offer any help either. But the midwife was worried and advised
Dasem to go to the subdistrict hospital because there were signs of
infection.
It had never occurred to Tarna to put any extra money aside for extra
medical expenses when his wife gave birth. In a panic to save his wife s
life, Tarna made a painful decision: he decided to pawn the family s
prized possession, a half hectare plantation, for just Rp 2 million.
But this modest sum was only enough to cover transportation to and
from the hospital and three days of in-patient care. And, despite making
such a large sacrifice, shortly after Dasem returned home her newborn
baby died. I was totally devastated, recounts Dasem. All I could do
was to hug our family cat everyday to comfort me; without the cat I
wouldn t have known how to go on.
Already having pawned their plantation, the loss of their second child
all but wiped out any hope left in Tarnas family. The financial impact of
that tragic incident fourteen months ago is still being felt by the family
today: besides losing their plantation, they are still indebted to the family
of the village midwife who administered the medication that Dasem
used during labor. On top of that debt, Tarna also has an older debt for
money borrowed to build the family home a loan from a local fishery
businessman, the Fish Boss , to whom Tarna sells his daily catch. And
finally, Tarna still has to find the money to pay the installments on the
loan used to buy his fishing canoe.
Dasem and her son working in the garden collecting worms for Tarna s fishing baits.
Now without his plantation, using his canoe Tarna paddles into the
swamp each morning to catch fish. Tarna needs one full day to set up
his fifty fish traps made of bamboo scattered throughout the swamp.
He inspects each fish trap the following day. One fish trap usually catches
one to three fish, although some traps fail to catch any fish at all. Tarna
on average brings home about three kilograms of varied fish a day. One
kilogram is sold to the Fish Boss for Rp 3,000 to Rp 6,000, the final
price decided by the buyer depending on the type, quality and size of
the fish.
Sometimes Tarna comes home empty-handed. Its because I cannot
predict where in the swamp the fish will feed , Tarna explains. If he
comes home empty-handed and if no rice is available at home, his
family is forced to ask for food from his neighbor. Should one day the
neighbor require assistance (to cut the grass or cultivate the farm),
Tarna and Dasem will offer a helping hand as a way of repaying their
neighbor.
Tarna preparing his canoe to check his fish traps in the marsh.
Tarna holds on to the hope that he will be able to pay off his debts
quickly and not toil in the swamp too much longer. He wishes he could
be back working on his own cassava plantation and hopes one day to
be able to buy the plantation back. Then, who knows, if things work out
maybe Tarna will be able to save a little money, and build up enough
capital to open up his own food stall one day, where Dasem can work.
Hopefully, if Dasem eventually tries for another child there will be a
well closer to the house so there will no need to carry heavy buckets of
water so far. And maybe, too, there will be more affordable and earlier
healthcare during her pregnancy and the family can avoid losing what
little they have if anything goes wrong.
Tarna and his family in front of their home.
73

ACCESS TO WATER AND ELECTRICITY MOVES OUT OF REACH OF THE POOR: TANJUNG PRIOK, JAKARTA
Nur Asmawati bathing her children in the only washing place available at the back of their house.
Trapped by low wages in the informal sector, a hard-working

family cannot make ends meet
Rizal Syam was devastated when, in mid-2002 and at an age when he

felt he was still productive (he was 55 years old), he was suddenly laid
off. He had dedicated 22 years to his job in the service sector working
at a large stateowned port in North Jakarta. Rizal only received Rp 15
million in severance pay after all those years. And the loss of his job
compounded an already difficult situation: Rizal s wife, Nur Asmawati,
had lost her job as a laborer at a manufacturing company back in 2000.
74
Rizal and Nur Asmawati used some of the severance pay for renovating
their parents home, which was falling into disrepair. Some they used
as start up capital for a nasi uduk an Indonesian specialty of rice cooked
in coconut milk) food stall in front of their house, managed by Nur
Asmawati. But the food stall never made a decent profit and after one
year Rizal and his wife were once again unemployed. Rizal was forced
to take the only other opportunity that seemed to be available: working
in sanitation for his local RT (rukun tetangga, or neighborhood
association), where he was made responsible for collecting rubbish
and the cleaning and unblocking of open sewers.
Now three years later, Rizal collects his income directly from the 40 or
so households in his RT at the end of every month. The monthly fees
he collects only come to about Rp 200,000, but that sum does not include
the Rp 20,000 levied by the RT for maintenance of his trash cart.
With such a meager income, it is hardly surprising that after increases in
water and electricity rates Rizal and his family feel increasingly desperate
when the monthly bills arrive. Sometimes, Rizal is able to salvage a few
items that still have some value from the trash he collects, including
plastic water-bottles, carton, steel alloy and used items made of aluminum.
These goods he then sells to a used-goods collector. To further supplement
their income, Nur Asmawati does some baby-sitting, watching over
Adiansyah, and the youngest child of a working mother. For her babysitting
services, Nur Asmawati earns Rp 200,000 a month.
Rizal explains that a representative from the state-owned drinking water
company (PAM) visited the house recently to explain that he had failed to
pay his water bills for over a year. As Rizal has accumulated Rp 1.25
million in fees and penalties for late payment and still could not guarantee
to pay off the arrears, the representative was forced to turn off the water
supply to Rizal s family. Rizal also produces a notification letter from the
state-owned electricity company (PLN). The message is basically the
same: six months have passed since the electricity bill was last paid,
putting his outstanding fees and penalties at Rp 600,000. His land and
property taxes have not been paid for nine years, and now total Rp 600,000.
Rizal s eyes well up with tears as he talks of his circumstances. But he is
quick to add that his are not tears of sadness but rather tears of
compassion. All this time, he feels that poor people such as himself are
never given much thought in Indonesia, let alone helped to deal with
their harsh circumstances. Nevertheless, Rizal does not want to become
a burden to society or the government.
For me, collecting the trash isn t a shameful occupation , says Rizal. I m not
embarrassed, I m proud I m not sitting idly at home or asking for money
The water provider company official turning off the mains supply to Rizals house.
Rizal says he hopes the government will not only pay attention to office
workers. He hopes it will also take into account people such as himself,
who work in the informal sector for small wages-but who are nonetheless
prepared to work. Thinking about the future of his young grandniece and
nephew, Rizal also hopes the government will not only develop high-rise
offices, but also open spaces and parks so that poor children living in
overcrowded neighborhoods can still play in a decent, healthy environment.
In the evenings, when Rizal s large family gathers together, they love to
watch television programs that promise the fulfillment of the dreams of
poor families, such as Rebuild Your Home and Surprise Money . Who
knows, dreams Nur Asmawati, maybe one day it ll be our turn? But
dreaming about TV game shows aside, if the economic climate were to
improve and jobs once again became more abundant, maybe Rizal and
Nur Asmawati could start to look for jobs in the formal sector, and
reconnect to steadier and higher incomes. Then they could start saving
once again, perhaps putting enough aside to have another try at starting
a small food business.
Rizal s family will now have to get used to life without electricity or water.
75

LACK OF OPPORTUNITIES LEADS TO A POOR DIET: SANGGAU, KALIMANTAN
Rosalia preparing instant noodles for her children, five year old Dewi and two year old Amos.
The poorest household in the village
The only house in Entawak with walls still made of bamboo and with a
roof still comprising leaves belongs to the family of Lukas. Most of the
other houses in this village have walls made of wooden boards or stone,
and roofs made from steel alloy. Reflecting this disparity, Lukas s family
is the poorest in the village and Lukas s wife Rosalia often feels belittled
by the talk of her neighbors.
76
The main sources of livelihood for Lukass family come from alternately
harvesting rice on his fathers land and extracting rubber from the trees
on his fathers plantation. Cultivating rice begins in June, when the
land is prepared for planting, and lasts until the harvest at the end of
the year. In the intervening period, from January to May, Lukas s family
members focus their daily efforts on extracting rubber from the rubber
trees on his father s plantation.
The rice harvest is stored for the family s daily consumption and it is
often shared with members of their extended family (father, mother,
siblings) who also participate in cultivating the rice. Meanwhile, the
rubber is sold and the proceeds used to purchase necessities. But this
supplementary income is unreliable given that the extraction process
depends to a large extent on weather conditions.
If the weather is bad, Lukas can often only work three days a week.
Extracting rubber on a normal working day (from 6 am to 2 pm) yields
three kilograms of rubber. Lukas sells one kilogram of rubber to Tauke,
a rubber collector, at a price of Rp 7,000/kg.
Lukas s wife, Rosalia, is thankful that she and her husband rarely fall ill,
but their two children, five-year-old Dewi and two-year-old Amos, often
experience stomach aches. The family diet consists of rice twice a day
(for breakfast they sometimes have coffee, but only occasionally),
flavored with salty or sweet soy sauce. If she feels up to it, Rosalia
searches in the forest nearby to gather vegetables (cassava, bamboo
shoots and ferns). This means that rice and vegetables are the family s
staples, while they seldom consume sources rich in protein, such as
fish, eggs, chicken or pork. Pork is a real treat and only eaten once a
year, when they celebrate the harvest.
The harvest celebration: One of few occasions each year when the family eats meat.
Rosalia is keen to limit her family to two children. To do this she takes
a natural contraceptive made from a mixture of local grass, dried wood
and water. The grass can be gathered from the forest, but only the
traditional healer (dukun) knows the type of wood needed and the correct
way to prepare it.
While buying sources high in protein would be expensive, Rosalia could
always catch fish from the river nearby her house. But she hardly ever
tries to do so. When asked whether she realizes that her two young
children need different kinds of food other than rice and vegetables,
especially protein-rich food, Rosalia says that she is afraid to give her
children fish from the river. This is because according to local beliefs,
children who consume river fish become infected with intestinal worms.
Lukas s packed lunch before he leaves for work in the plantation.
Lukas and Rosalia rarely travel beyond the limits of their village. Their
interaction with outsiders is limited to health officials (at the Puskesmas)
and security officers. Maybe these contacts with officialdom are what
drive their aspirations fortheir children: they hope Dewi will one day
become a nurse and Amos will join the military.
Maybe if there were some more opportunities for work in the village,
perhaps through local development projects, they could make a little
more money, improve their diet and start to renovate their house. A
steady income would make such a difference and allow them to start
constructing a more permanent home one similar to their neighbors
and fit to raise their children.
Lukas with his family. The parents struggle to provide adequate nutrition for their children.
77

LACK OF ACCESS TO CREDIT CONSTRAINS THE POOR: KOLAKA, SOUTHEAST SULAWESI
Mude, an expert boat-builder on a fishing boat that is not his own and he dreams of earning enough capital to buy his own fishing boat.
A traditional boat-builder needs to upgrade

From an early age, Mahmud (nicknamed Mude) has been an expert
boat-builder, a skill passed down from his father. His father also taught
him the skills of fishing, using nets bait and a fishing spear. Now Mude
needs capital to buy a more modern boat, one with a motor.
Mude feels closely connected to the sea. When asked what his most
valued possession is, he responds without thinking: his handmade
boat. Mude would be very happy to make a simple living from building
boats. But with only two to three orders for boats each year, Mude
must find another way to support his family.
78
Mude s five children were born and raised on the coast. As Bajuans,
they are accustomed to a life closely tied to the sea. Most of the childrens
time outside of school is spent playing in the sea. Only two out of his
five children go to school. Mude wanted his children to continue on to
junior secondary school but there is only one primary school in the
village and the only junior secondary school is located in a faraway
subdistrict.
The needs of their five children have driven Mude and his wife Suwarni
to supplement their income, as they have found they can no longer
simply live off the sea as their ancestors used to do. Following the
example of others, they have now also started to grow seaweed
commercially, albeit in a simple way and for modest returns.
Between rearing her five children and looking after the family home,
Suwarni also runs a small food stall in front of the house, where she
sells sugar, coffee, cigarettes, instant noodles, soap and other daily
consumables. Her customers are her neighbors and people from other
communities in the area. There are about 112 heads of households in
the village, with one household comprising around six people. This
small business was set up with some of Mude s savings. Income from
the food stall is enough to cover the snacks for the children.
In contrast to their own parents, Mude and Suwarni s aspirations for
their children are no longer linked to the sea. Instead, as Mude says, he
hopes their children can become office workers , without specifying
what type of work that might mean.
Not quite a dreamboat: Mude s children often play in the water near their house.
From Mude s fishing catch, only the so-called super fish , a type of
grouper, provides a decent income. One kilogram of this elusive fish
can be sold for Rp 40,000. But in one day Mude usually can only catch
about three kilograms. Mude and his wife often dream of buying their
own fishing boat. This would enable them to sail to nearby Selayar
Island, where the fish are plentiful and from where it would be easy to
bring back their catch to the fish market in Pomalaa.
As a Muslim, Mude defines success in his life as the opportunity to
undertake the haj (pilgrimage to Mecca) together with his wife. To this
end, Mude and Suwarni have begun to save little by little, stashing
their savings underneath their mattress as they have no bank account.
If only there were a bank near their village. Then they could save money
safely, and even earn a little interest, not just for the haj, but also for all
the longer-term needs of their family: education of the children, and
unexpected health problems or other unforeseen expenses.
Suwarni working at her food stall to make a little extra money.
Mude and his family sitting proudly in front of their modest house.
79

TRANSITION TO SECONDARY SCHOOL IS BEYOND THE MEANS OF THE POOR: ENUPETU, EAST NUSA TENGGARA
Simon buying goods for his family after selling his coconuts at the market.
A math genius settles for work as a coconut seller

Simon Aprianus Banamtuan is 13 years old. As a student at a SD
(primary school), he was always at the top of his class. His favorite
subject is mathematics. On my report card, I always received the top
grade for mathematics, he says calmly. But his prowess is not limited
to mathematics; he excels in all subjects, as shown by his Final School
Exam (UAS), which placed him second in overall ranking at his school.
But after completing his sixth grade exam in 2004, Simon did not
continue on to SMP (junior secondary school). His parents could not
afford the entrance fee and annual tuition fees. Despite there being a
80
SMP not far from Simon s house and three other SMPs in the local
district, Simons family cannot make use of them.
Instead of going to school, Simon now helps his neighbor by scaling
his coconut trees to collect coconuts. For every ten coconuts he collects,
Simon is given four. Since he usually collects twenty coconuts, this
means he gets to keep eight. After gathering the coconuts, he
accompanies his neighbor to the market to help sell them. If he manages
to sell all eight of his own coconuts at the market, Simon should receive
a total of Rp 4,000, or Rp 500 for each coconut.
Simon s parents, Jonathan and Orance work hard to meet basic

necessities. Jonathan works alternately on cultivating sweet potato and
corn, cleaning weeds out of people s gardens, gathering firewood and
keeping an eye on other people s livestock.
Jonathan Banamtuan and Orance have another child named Sefnat,
who is mentally retarded and suffers from deafness, muteness and
defective vision. The family is still grateful that Sefnat is relatively
independent-he can eat and walk unaided. But even so wherever Sefnat
goes he must be accompanied. However, Jonathan and Orance are proud
to point out that Sefnat helps draw water from a nearby spring. Several
times a day, he carries water to his parents who work in the garden, 600
meters away from the house.
The family diet is limited to corn cooked with papaya or cassava leaves.
Sometimes, they eat rice received from the Raskin program (subsidized
rice for the poor). The Banamtuan family received their health cards in
December 2005, which helps tremendously with their health costs. The
money they save on medicine can now be used to buy food.
Simons mother showing her health voucher for healthcare at the Puskesmas.
The family situation made Simon give up on his dream of continuing

school. Instead, Simon realizes that he has to help his family. Rather
than playing with friends or sitting around idle, Simon works wherever
he can. In this respect he takes after his parents, who never let a day go
by just relaxing.
When asked if there is any financial assistance that could help Simon
to continue his education, Jonathan responds by saying that he is too
embarrassed to ask about it. When asked about his family daily diet,
Jonathan only lowers his head, looks away and remains silent. His wife
also avoids being drawn by the question, indicating only that such
questions make her husband sad.
Simon and his brother collecting water from the familys only water source.
Although Simon is keen to continue his education, the topic of school

is never discussed. But his father is already clearly aware of it: with
obvious pride, Jonathan highlights the awards his son has achieved at
school. He concedes that if Simon were to continue his education, his
future would be brighter. With a hint of bitterness in his voice, Jonathan
says that putting his bright, hardworking son through school is just a
pipe-dream when the reality is that Simon is needed for work, no matter
how modest his contribution.
But the family maintains a sense of dignity: their acceptance of hard
work and difficult circumstances, and their stoic refusal to ask for help
from anyone. Maybe next year Simon s old SD headmaster will
encourage the family to seek financial help from the district government
that might allow Simon to make the transition to junior secondary school.
Who knows where that might lead? Maybe with his prowess at school
Simon could go on to university, find a good job, maybe even as a
mathematics teacher, and then help to support his family, improving all
their lives.
Simon and his family happy at home despite their hardship.
81

BAD RURAL ROADS CUT OFF THE POOR: RANSIKI, WEST IRIAN JAYA
Elieser and his family returning home after collecting produce from their family plantation.
Three generations under the same roof

The road between Manokwari, the provincial capital of the newly created
province of West Irian Jaya, and the district of Ransiki runs for about
200 kilometers and takes four hours to cover. There are more than five
small rivers to cross but many do not have useable bridges forcing
cars to drive in the riverbed. When it rains in the mountainous region,
in many places the road becomes submerged in knee-deep water.
A different challenge awaits travelers on the road between the district
town of Ransiki and the village of Sosmorof: it is often blanketed with
82
mud or sand, making it treacherous and easy to veer off. These poor
conditions are even worse in some places, where precipitous chasms
on one side of the road are shared with high banks prone to mudslides
on the other.
It is understandable, then, that the extended family of Elieser Aiba greatly
appreciates the goods that come from beyond the limits of their village
and district. While not of great monetary value, these non-local goods,
such as cloth, radios, lanterns, local election posters, are highly
esteemed and treated as valuable objects.
Elieser, along with his three daughters, two sons-in-law and six
grandchildren, occupies a house that is elevated over a platform and
has two doors, one at the front and the other at the back, but lacks any
windows. Twelve people, three heads of household, all live in the one
house, without any partitions. Mats spread over the bamboo floor are
the only objects that demarcate one family snspace from another s.
While this living arrangement is a harmonious one, the confined space
seems to be the driving force behind Elieser sndesire to save money
for a steel alloy roof over the existing house and to construct houses
for his children, sons-in-law and grandchildren.
But if the family were to rely on their current livelihoods, Elieser s dream
would appear fanciful: the onions and red beans from the family
plantation that they sell are only enough for their daily necessities,
mainly for cooking oil. Meanwhile, the cocoa seeds that the family
plants in the plantation are only harvested seasonally, with modest
results. Their pigs cannot be relied upon because their fellow residents
only consume pork when there are very important events in the village.
The bustling market in Ransiki is hard to reach from surrounding villages.
Elieser can easily recount the price per kilogram of red beans at
Rp 3,000, onions at Rp 5,000, half-dried cocoa at Rp 8,000 and fully
dried cocoa at Rp 10,000. Meanwhile, the price for one fat pig is
Rp 800,000. But when asked to reveal his age, the age of his children
or grandchildren, the area of his plantation or house, it almost seems
as if Elieser cannot count, or perhaps that such numbers are of no
importance to him. Nor can his children or his grandchildren claim to
be of a particular age with much certainty.
Although he says that family members seldom fall ill, Elieser hopes
that soon there will be a healthcare facility within easy traveling distance.
At present, the sick must be taken far from the village to the district
capital for treatment. On Sunday mornings, most of his family members
walk with an air of excitement to a building on the edge of the village
a building exhibiting a certain splendor, and of which they are clearly
proud. The building is the village church, which a magnificent tower
topped with precisely the kind of alloy roof that Elieser yearns for.
Buah merah fruit has the potential to increase the family s income if only they could get to
market more easily.
Elieser hopes that in the near future the local government will spend
more money on improving the district road. This would make products
in the village cheaper and mean that he could take his own produce to
Ransiki more often, and not spend too much on transport. Then he
could start to save, buy the new roof for his house, install electricity
and a clean water supply. And maybe even start building a new house
for his grandchildren a house with partition or help them to stay on
at school. Such great things could come from small beginnings, such
as a better road to Ransiki.
In Papua pigs are an indication of wealth and only eaten on special occasions.
83
Chapter
Making Growth
Work for the Poor
84
Chapter 4: Making Growth Work for the Poor
Introduction
Growth is the single most important driver behind poverty reduction. This is true for Indonesia (see Chapter
2 on the History). It is also true for many other countries. The World Bank s seminal study on Pro-Poor Growth (World
Bank, 2005c), which provides case studies of growth in 14 countries, gives the same message: growth is essential for
poverty reduction.
But growth can be more or less pro-poor. Growth can create considerable inequality with knock-on impacts on
social cohesion or it can be strongly equalizing, helping to integrate rural and urban communities. As Chapter 2 shows,
Indonesia s experience in the 30 years before the economic crisis was an example of equalizing growth. But the crisis
changed the pattern of growth: since 1998, growth has not only been lower, it has also been less equalizing. The key
policy challenge for growth, therefore, is not only how to raise the rate of growth, but also how to ensure that growth
reverts to the pro-poor pattern that characterized it before the crisis.
The current pattern of growth could mean that the government will fail to hit its poverty headcount
target of 8.3 percent by 2009. Figure below shows three different poverty projections. The base-case projection
assumes that inequality continues to widen at the same rate as it has since the crisis. If this occurs then Indonesia will not
meet the target poverty headcount of 8.2 percent by 2009, set in the RPJM. It would meet the less ambitious Millennium
Development Goals (MDG) target for poverty reduction-but even this target moves out of reach if growth falls from the
projected 6.2 percent to only 4.0 percent per year as shown in the low-case projection. To meet the government s poverty
target, growth must become more pro-poor. The high-case projection shows the impact of sharing growth more equitably.
Even with the high-case projection of 6.2 percent pro-poor growth, by 2009 poverty incidence will only be reduced to
around 12 percent.
Figure 4.1 Poverty headcount projections depend on the pro-poorness of growth
Poverty Projections with different growth scenarios
20%
18%
16%
Poverty Incidence
14%
12%
Low C ase
Base C ase
10%
High C ase
8%
6%
4%
2%
0%
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
Source: World Bank staff calculations.

Note: See below for scenario assumptions.
87
Description of Scenario
Impact on Poverty
Low Case
Inequality as predicted by previous

trend (the poor experience lesser
growth than the rich) but the level of
growths are smaller than the base case
RPJM target is not attainable, nor is the MDG target

of halving the national poverty number.
Base Case
The base case scenario uses the projected

growth figures in the Medium Term
Framework (up to 2007) and then assumes
growth rates to remain at 6.2 percent
until 2016. In this scenario, inequality is as
predicted by previous trend where the poor
experience lesser growth than the rich.
RPJM target of achieving 8.2 percent poverty rate by 2009

is not attainable under this scenario, but the MDG target of
halving the national head count by 2015 is attainable.
High Case
Distribution neutral growth (the poor

experience the same level of growth
as the rich 6.2 percent)
RPJM target of achieving 8.2 percent poverty rate by 2009 is not

attainable under this scenario, but the MDG target of halving the
national head count by 2015 is attainable RPJM and MDG targets
are both attainable.
There are two key pathways through which households and individuals have escaped poverty in
Indonesia s recent past:
improvements in agricultural productivity in rural areas; and

increases in non-agricultural productivity in both urban and rapidly urbanizing rural areas.
Most individuals who escaped poverty in the past ten years followed one of these two paths. However, a significant role
has also been played by Indonesia s structural transformation as people moved from agricultural to non-agricultural
activities as some rural areas rapidly urbanized.
The question is: what policies help the poor onto these pathways? How then can policy help the poor to get
onto, and stay on, a pathway out of poverty? Although the answer clearly depends on sector and location, there are three
cross-cutting types of policies that are needed.
First, it is essential to maintain macroeconomic stability. Indonesia s hard fought recovery from the economic
crisis of 1998 has focused on restoring confidence in macroeconomic management. This was and is the right focus. A
stable macroeconomy is the cornerstone of a successful economy and of poverty reduction. Economic crises not only
dampen long-run growth, but they hit the poor hardest. Sustained reductions in poverty require a stable and conducive
macroeconomic environment.
Second, invest in the capabilities of the poor. The poor, particularly those in urban areas, are poor primarily
because they earn very low incomes (rather than because they are unemployed). Their incomes are low because the
productivity of the activities they do is low, in part because of very low human capital. Improving education for the next
generation and vocational training for the current one are essential in enabling the poor to access better jobs. At the same
time, for poor rural households earning most of their income from agriculture, what matters more is appropriate knowledge
of how to boost agricultural productivity. This requires substantial investment in agricultural research and the re-building
of an effective agricultural extension service.
Third, connect the poor to opportunities. The rural poor do not merely lack capabilities; they lack the means of
connecting to growth. Sometimes the missing connection is physical-for example, access to markets in many rural areas
is hampered by poor quality roads at the district (kabupaten) and sub-district (kecamatan) level. But other forms of
connection matter too: poor households have more difficulty in accessing credit, in part because of their lack of collateral;
and poor people in urban areas find it hard to access the labor market because of labor regulations that discourage the
hiring of less-skilled workers.
88
This chapter proceeds as follows: Section II describes the pathways out of poverty and the evidence
about what constitute the most important pathways. Section III explains the linkages between the pathways
framework and the policy agenda. Section IV describes briefly the policies needed to ensure macroeconomic stability.
Section V discusses policies to build the capabilities of the poor to participate in growth, while Section VI describes
various policies to connect the poor to better opportunities for growth. Section VII concludes.
II
What Are the Pathways out of Poverty?
There are two major groups of the poor that need to be reached. The profile of the poor provided in Chapter 3
on Understanding Poverty shows that there are two main groups of the poor. First, there are poorly or uneducated rural
households whose members are predominately involved in low-productivity agricultural tasks that are mostly disconnected
from the major growth centers. Second, there are the poor who are currently living in close proximity to the major growth
centers, mostly on Java/Bali or the more densely settled parts of Sumatra, but who are struggling to participate in the
economic opportunities in those areas.
Reflecting the challenges of reaching these two groups of the poor, it is possible to construct a simple
framework for thinking about the pathways out of poverty (see Figure 4.2). Obviously, given the variation in the
capabilities of and the opportunities and resources available to the Indonesian population, the routes out of poverty in the
future are likely to show substantial variation. However, by stripping out policy and environmental complexities, and
focusing directly on the two main drivers of poverty reduction, Figure 4.2 allows us to see the basic forces at work.
Figure 4.2 The pathways out of poverty
Urban
Rural
There are two productivity pathways out of poverty. The first is the move from low-productivity, subsistence
farming to commercial farming. This includes intensification through raising productivity of food crops, as well as
diversification into higher-value crops, whether food or non-food. This path of rising agricultural productivity, shown as
Pathway 1 in Figure 4.2, also includes those who exit poverty by gaining better paid employment on such commercial
farms. The second path comes with the increase in productivity and profitability of non-farm enterprises, including the
new jobs and better paid formal employment in such enterprises. This is Pathway 2 in Figure 4.2, and can take place in
both rural and urban areas.
89
Accompanying these two productivity drivers are two transition phases that individuals may go through to
reach the second pathway out of poverty. The first, Transition A, is a sectoral shift from the farm to rural off-farm
employment (although the physical household may remain in the same location). The second, Transition B, is a locational
shift out of rural areas to urban employment, through either seasonal or permanent migration. This can come both from
households currently engaged in subsistence agriculture and from those currently engaged in petty trade, manufacturing and
services that is, from both farm and rural non-farm households. Of course, these transitions should be thought of as continuums
rather than discrete starting and ending points. Indeed, in the densely settled parts of Indonesia, it is often hard to distinguish
between urban and rural ; and households often engage in both farm and non-farm activities. Nonetheless, these distinctions
make it possible to identify the key drivers of poverty reduction in Indonesia in recent years.
What have been the most important pathways out of poverty?

Changing numbers of poor over time indicate the most important pathways. To understand what have been
the most important paths out of poverty it is necessary to know how many people there are in each of the cells of Figure 4.2
and how these numbers have changed over time. Figure 4.3 shows this is a compact way. Each number in a cell represents
the change in the percentage of workers employed in that cell between 1993 and 2002.71
Figure 4.3 Movements along the pathways out of poverty, 1993 to 2002, when villages can be reclassified as urban in 2002 from
their 1993 status as rural ( published data)
Urban
Rural
Non
Poor
Non
Poor
Poor
Poor
The numbers in each cell represent the change between 1993 and 2002 in the percentage allocation of the employed workforce, using data which allow the reclassification of rural areas into urban areas. Using
these officially published data, the share of the workforce employed in low productivity agriculture ( poor ) dropped by 2.53 percentage points between 1993 and 2002, without holding the classification of their
resident villages (as rural versus urban) constant. The share of the workforce that was urban and non-poor rose by 9.88 percentage points when villages could be reclassified from rural to urban.
Note: The size of each rectangle is approximately proportionate to the numbers of workers employed in each category in 1993. See Table 2.3 (in Chapter 2 on History) for the actual numbers. There were 78.5
million workers employed in 1993 and 86.9 million in 2002.
Official data suggest a large increase in the share of employment in non-farm activities in urban areas,
with a corresponding decline in the share of all forms of employment in rural areas. The share of non-poor
workers involved in non-farm activities in urban areas grew by 9.9 percent, while the share of all employment in rural
areas declined by over 12 percent. Of course, in the absence of panel data (which follow individuals or households) it is
71
By focusing on percentages of the workforce rather than absolute numbers of workers we can abstract from population growth and expansion of the workforce over the nine-year period. Using changes in
percentages also means that the negative and positive changes add up to zero.
90
impossible to say exactly who went where, but the movements in Figure 4.3 are substantial and suggest that the way out
of rural poverty from 1993 to 2002 was to leave rural areas altogether.
But urbanization rather than migration is responsible for most of this change. The official published data
from Susenas tell a story of rapid and successful urbanization that seems to be driven by migration from rural areas, both
from agricultural and rural non-agricultural households. However, a substantial number of rural households became
urban households without changing location. This is because the definition of urban is based on village characteristics
that can change over time population density, number of amenities and the share of income from agriculture. Thus,
villages on the periphery of urban areas can themselves become urban areas. The best estimate is that at least 10 percent
of rural villages in 1993 had been reclassified as urban by 2002. Figure 4.4 indicates that this reclassification makes a
tremendous difference in interpreting the significance of the different pathways out of poverty.
Figure 4.4
Movements along the pathways out of poverty, 1993 to 2002, when villages remain in their 1993 classification as urban
or rural ( corrected data)
The numbers in each cell represent the change between 1993 and 2002 in the percentage allocation of the employed workforce, using data corrected for reclassification of rural areas into urban areas. Thus the
share of the workforce employed in low productivity agriculture ( poor) dropped by 1.55 percentage points between 1993 and 2002, holding the classification of their resident villages (as rural versus urban)
constant. The change in the share of the workforce classified as urban and non-poor actually declines by 0.91 percentage points (instead of increasing by 9.88 percentage points in the official published data)
when villages that were rural in 1993 remain in that category (despite some of them having been reclassified as urban in 2002).
Note: The size of each rectangle is about proportionate to the numbers of workers employed in each category in 1993.
See Table 2.3 (in Chapter 2 on History of Growth and Poverty Reduction) for the actual numbers
.
There were 78.5 million workers employed in 1993 and 86.9 million in 2002.
The reclassified data suggest that the growth of the rural non-farm sector has been an important route
out of poverty. Once the official published data are corrected for reclassification of rural areas into urban areas, the
story of rapid urbanization told in Figure 4.3 takes on a very different character, with rather different policy implications. In
Figure 4.4, there is still only one route out of poverty, but now it is through the rural, non-farm economy. Indeed, in terms
of the relative share of the workforce, the original urban areas suffered a loss of about half a million workers from all four
categories. The difference between the pathways in Figures 4.3 and 4.4 is that the successful rural non-farm economies
in Figure 4.4 became urban in the process.
But the success of rural non-farm enterprises in poverty reduction is hidden. Of course, the rapid urbanization
seen in the official data is real in the sense that these are now urban areas, even if they had been rural a decade before.
91
The significance of the difference between Figures 4.3 and 4.4, however, is that the great success of rural non-farm
enterprises in reducing poverty is lost by calling them urban after the fact. Their very success is what made them urban.
Rural-urban migration did also play a role, but a relatively small one. Table 4.1 shows an estimate of true
migration obtained from the inter-censal survey (Supas) in 1995.72 Only 2.8 percent of those individuals in rural villages
in 1990 were living in urban villages in 1995. Moreover, the 3.6 million people who moved to urban villages between
these two years were compensated in part by 1.8 million people who moved from urban to rural villages over the same
period. Overall, around 5 percent of those in urban areas in 1995 had been living in rural areas five years previously. Thus,
although the role of true migration is certainly not trivial the much larger shifts in the shares of urban employment
described above suggest that, overall, urbanization rather than rural-urban migration has been the dominant factor in
changing the nature of activities undertaken by the poor and, therefore, their opportunities for finding a pathway out of
poverty.
Table 4.1
True rural-urban migration, 1990-1995

1995
1990
Rural
Urban
Total
Rural
Urban
Total
122,037,729
97.16
98.53
1,823,701
2.69
1.47
123,861,430
64
3,570,511
2.84
5.13
66,089,601
97.31
94.87
69,660,112
36
125,608,240
100
64.91
67,913,302
100
35.09
193,521,542
100
100
100
100
Source: Supas, 1995.
There are two conclusions from this dynamic. First, urban areas need to be prepared, both economically and
politically, to absorb rapidly growing peri-urban economies. Second, constraints on the growth of these successful rural,
non-farm enterprises need to be relaxed for those not yet absorbed in urban growth poles or, put differently, creating a
better rural investment climate is important for poverty reduction, even if these enterprises end up as urban in the process.
Figures 4.3 and 4.4 are suggestive about the pathways that may be the most important (and in particular they point to a
much larger role for urbanization than might have been thought before), but they are not definitive since all that is shown
is the net position rather than the actual flows themselves. For definitive results on who actually moved and their
characteristics, panel data are needed as discussed below.
Panel data on individual pathways out of poverty

Representative data exist on who moved in and out of poverty. Fortunately, panel data exist in the form of the
Indonesian Family Life Survey (IFLS), which interviewed the same families in 1993, 1997 and 2000.73 The 1993 sample
was representative of about 83 percent of the Indonesian population in 13 provinces. McCulloch, Timmer and Weisbrod
(2006) use the IFLS to track the movements in and out of poverty of a large set of individuals between 1993 and 2000 as
shown in Table 4.2. Although the IFLS data are not entirely representative of the Indonesian population, the 5,308 workers
72
True migration means that individuals actually physically moved village. Because the Supas survey asked about the respondent s address five years ago at the village level it is possible to tell whether they
actually moved village or not. The full Census only asked for this information at the district level making it impossible to tell from Census data whether they moved from rural or urban areas.
73
And hopes to re-interview them again in 2007.
92
tracked in Table 4.2 do represent basic Indonesian poverty patterns reasonably well. Rural poverty made up 74.9 percent
of total poverty in 1993 in the unweighted IFLS data, compared with 74.0 percent in the 1993 Susenas data. Analysis of
how these 5,308 workers fared between 1993 and 2000 should provide very useful insights into the dynamics of poverty
in Indonesia.
There is a high level of mobility in and out of poverty. For example, of the 522 workers who were poor and
working in rural agricultural activities in 1993, only 221 of them were still in that category seven years later; 212 individuals
exited poverty while staying in agricultural activities in rural areas, and a further 71 shifted into rural non-farm activities,
half of them escaping poverty and half not. But mobility was both downwards as well as up: there were 226 individuals
who were non-poor in rural agricultural activities in 1993 who had fallen into poverty in the same sector by 2000. Another
86 people fell from being non-poor in rural non-farm activities into rural agricultural poverty. Overall there is a great deal
of churning in and out of poverty (see Table 3.2 in Chapter 3 on Understanding Poverty).
Improved productivity in agriculture is still an important route out of poverty. Most of the rural agricultural
poor in this sample who exited poverty did so while staying in rural agriculture. Over 80 percent of the poor rural farmers
in the sample were still working in rural agriculture in 2000, but around half of the households from which they came still
managed to exit from poverty.74 Most of the rest moved into non-agricultural activities in rural areas and, more often than
not, out of poverty too. However, moving to urban areas was rare for this group only 3.5 percent of those working as poor
rural farmers in 1993 were working in urban areas in 2000, and more than half of these remained in poverty.
Rural non-agricultural activities can be a stepping-stone out of poverty in rural areas. Individuals from
poor rural households who worked off-farm in 1993 were more mobile than those working in agriculture. Less than onefifth stayed poor in rural non-agricultural activities but another fifth shifted back to agriculture while remaining poor.
Almost half escaped poverty while staying in rural areas a similar rate to those who started poor in agriculture. Urban
migration played a more important role for this group, but still a minor one-less than 10 percent of this group moved to
urban areas, and only half of those that did so exited poverty.75 Still, this migration rate almost 10 percent of poor, rural
non-farm workers migrated to urban areas in just seven years is three times the rural-to-urban migration rate of poor
farm workers. During the same period, 13.6 percent of poor, rural farmers moved to rural non-agricultural jobs and more
than half exited poverty in the process. Thus the panel results confirm the suggestion of the Susenas-based analysis: that
the growth of the non-farm rural economy may be an important stepping-stone out of poverty.
Non-agricultural activities in urban areas are much more stable. As Table 4.2 shows, 76.4 percent of the nonpoor in this category remained non-poor, non-agricultural and in urban areas. At the same time, 46 percent of the urban
non-agricultural poor remained in that category. Both rates of stability are the highest of the four categories. This stability
may arise because earnings are much higher in urban areas, so the poor prefer to stay in the hope of a good job; or it may
be difficult for poor urban workers to move or return to rural areas. The distinction should not be overdrawn, however, as
around 60 percent of the rural, non-poor remained in their original sector of activity. Indeed, when agricultural and nonagricultural activities are combined a common occurrence for households in rural areas 75.1 percent of rural agricultural
households and 76.9 percent of rural non-agricultural households (in 1993) remained stably out of poverty. Nonetheless,
it is noteworthy that almost 90 percent of poor urban non-agricultural workers in 1993 stayed in the same sector and
location, with almost half of them leaving poverty in the process.
74
There is something of a contradiction between the results from the Susenas data shown in Figure 4.4, which emphasized the role of the rural non-farm sector, and those from Table 4.2, which stress the
importance of increases in agricultural incomes. Reconstructing Figure 4.4 from the IFLS data shows a decline in the share of non-agricultural employment in rural areas rather than the large expansion shown
in Figure 4.4 and, conversely, a substantial expansion of the employment share of non-poor workers in rural agriculture rather than the decline shown in Figure 4.4. There are a number of differences between
the two surveys which may have given rise to these differing results, most notably the surprisingly low share of non-farm income in the IFLS survey. Further research is needed to resolve these differences. In
the interim, our approach is to accept the aggregate picture painted by the nationally representative Susenas data, but to draw useful insights about the dynamics of poverty from the IFLS panel.
75
Note that re-classifications are very rare (3.26 percent) in the IFLS panel so that almost all the rural-urban movements observed involve actual physical movement of the person. This adds weight to the
evidence that urbanization rather than rural-urban migration is the principle form of rural-urban transition.
93
Government policy should therefore encourage rural transformation and urbanization. Urbanization has
been rapid, but only a small part is due to actual physical migration of workers and households. In rural areas, there is a
gradual diversification of economic activities taking place, characterized by greater reliance on non-farm sources of
income. This process of rural diversification reflects greater opportunities for growth in a dynamic non-farm economy
than in agriculture per se, although increases in agricultural productivity remain an important pathway out of poverty.
Government policy, rather than inhibiting this rural transformation, should actively encourage it. Moreover, there is an
important role for government policy at both the national and the sub-national level in improving the investment climate
for both agricultural and rural non-agricultural enterprises, as well as in helping connect the urban poor to better jobs.
How policy can build the capacity of the poor and connect them to growth is the subject of the next section.
Table 4.2
Poverty transition matrix, 1993 to 2000, from IFLS panel data (percent of individuals, weighted)
1,243
1,536
1,332
1,270
1,328
139
1,269
Source: Indonesia Family Life Survey (IFLS).

Note: Data are the percentage of individuals in the 1993 category who end up in the 2000 category. The individuals are those who were aged 15-55 in 1993 and were working.
III
Linking Pathways out of Poverty to Policies for Pro-Poor Growth
Having laid out the pathways out of poverty, it is important to elaborate on the policies likely to be most
effective in getting the poor onto the various pathways out of poverty and keeping them there. Of course,
there are myriad policies that have an influence on growth and poverty reduction: macroeconomic policy, which sets the
framework for all economic activities; trade policy, which determines the costs of exchange both internationally and
domestically; educational policy, which influences the capacity of the workforce; financial policy, which determines the
access to credit and financial services; and many more. Amidst this sea of choices, how should the government choose
the priority policies for generating pro-poor growth?
Macroeconomic policy must remain the core of any successful poverty reduction strategy. One priority is
clear: the lessons of the past decade of policy reforms in developing countries show clearly that maintaining sound
macroeconomic policies is central to sustained poverty reduction. Countries that have had more macroeconomic shocks
experience slower growth and poverty reduction than those with better macroeconomic management (World Bank, 2005c)
Indonesia knows better than most countries the dreadful poverty impact of macroeconomic crises (see Chapter 2).
But which microeconomic policies are most likely to reduce poverty? One approach to answering this question
is to estimate the impact of access to education, training, roads, credit and so forth on the level of household consumption.
Figure 4.5 shows the percentage increase in household expenditure associated with improved access for several different
types of assets and services.
94
Figure 4.5 The welfare benefits associated with investing in the poor and connecting the poor
(Percentage increase in household expenditures associated with changes in household characteristics)
Source: Susenas 1999 and 2002 consumption modules and Podes 2003.
Note: The results come from household expenditure regression provided in Table 3.6.
Investing in the capabilities of the poor appears to give a high return. Figure 4.5 shows that in both rural and
urban areas, higher levels of education of the household head are associated with higher levels of consumption. The effect
is particularly strong if the household head already has nine years of education. This suggests that investing in education
for the poor and, in particular, in ensuring transition from primary to junior secondary school and increasing the number
of secondary schools in rural areas will be key to boosting the capability of the poor to participate in growth (see Chapter
5 on Public Spending for more details on education). Improving access to informal courses could also help to improve
the productive capabilities of the poor, particularly in urban areas. Figure 4.5 shows that the return to access to informal
courses and vocational training is generally even higher than the return to general education, particularly in poorer
eastern Indonesia. (Also see Table 3.4 in the section on The Determinants of Poverty in Chapter 3.)
And connecting the poor to opportunities for growth is key. Figure 4.5 also shows the impact on consumption
of access to roads, telecommunications, credit, and formal sector employment. The impact of being connected is high,
particularly for formal sector employment outside of agriculture. Access to asphalt roads is also strongly associated with
higher consumption, as is access to credit in some locations. Again, the less well-connected regions of eastern Indonesia
see particularly strong gains from improved connections.
The Rural Investment Climate Survey (RICS) bears out the importance of connecting the poor to growth.
The RICS surveyed 2,500 predominantly micro and small non-farm enterprises in six districts (World Bank, 2006h).
Figure 4.6 shows which issues these firms identified as their most important constraint. Firms at the district level, which
constitute the vast majority of firms and which employ the majority of the non-farm poor, see their most important
95
constraints as those associated with: low demand for their goods and services; difficulties in accessing credit; and
problems in accessing markets due to poor road, transportation and electricity infrastructure.
Figure 4.6 Most important constraints faced by firms in RIC survey
10,000
20,000
30,000
40,000
50,000
60,000
Source: RICS, 2006.
Based in part on the evidence above, Table 4.3 lays out a framework for linking policies to the pathways out of poverty.
Table 4.3
Linking policies to pathways out of poverty
Farm
Policies
Non-Farm
Rural agricultural poor
Rural Non-Farm poor
Urban Non-Farm poor
Low inflation
Maintaining a Stable
Macroeconomy
Competitive exchange rate

Low prices for staple foods
Investing in the
Capabilities of the
Poor
Connecting the Poor

to Opportunities
Agricultural extension
Education, training and information
Rural roads
Labor markets
Access to credit
The same sound macroeconomic and trade policies are needed regardless of the pathway out of poverty.
Lowering inflation is essential since the poor are often the least able to protect the real value of their incomes against
96
inflation. Keeping a competitive exchange rate encourages export-led growth and boosts competitiveness by encouraging
a focus on the production of tradable goods. And maintaining low prices for staple foods helps the poor, who are
overwhelmingly net consumers of such staples.
But different microeconomic policy priorities are appropriate for different pathways out of poverty.
There are four key microeconomic policies that would support the various pathways out of poverty, as follows:
Boosting the productivity of the rural agricultural poor requires improvements in their capabilities but
these should mostly come from improvements in their access to agricultural knowledge and technology through
rebuilding the agricultural research and extension service. At the same time they need to be connected to growth.
Here the main intervention needed is improved rural roads, although improvements in access to electricity and
irrigation may feature strongly in some locations.
Boosting the productivity of the rural non-farm poor also requires investments in their capabilities. For the
most part, the emphasis here needs to be on better education to enable their access to better jobs, and improved
vocational training. But the non-farm enterprises that employ such poor people also need to be better connected to
urban growth poles, again through better rural roads and electricity.
Boosting the productivity of the urban non-farm poor requires the same sort of emphasis on education and
vocational training as for the rural non-farm poor. But the ways in which the urban poor should be connected to
growth are rather different from those in rural areas. Although infrastructure is still important for poverty reduction in
urban areas (particularly water and sanitation, see Chapter 5), the key need is to connect the urban poor to formal
labor markets.
Improving access to credit helps to connect all three groups of the poor to opportunities. Farmers
require credit to finance inputs; non-farm enterprises in rural areas feel particularly constrained in their access to
credit (see Figure 4.6); and urban enterprises particularly the micro and small businesses in which most of the poor
participate are also credit-constrained. Measures to sustainably boost access to commercial credit are therefore
likely to be an important part of stimulating pro-poor growth.
Our analysis therefore focuses on three key types of policies:

Maintaining a stable macroeconomy
Investing in the capabilities of the poor, and
Connecting the poor to opportunities
The following sections discuss policies options in these areas in more detail.
IV
Maintaining a Stable Macroeconomy
Making growth work for the poor involves two coordinated strategies: one is growth acceleration , the
other is work for the poor . Macroeconomic stability is a prerequisite for both strategies. On the one hand,
macroeconomic stability is a top priority for investors in the absence of stability investment will be adversely affected
(World Bank, 2006e). On the other hand, the poor are the most vulnerable group to the negative consequences of
macroeconomic instability that arise from extreme exchange rate fluctuations.
Restoring low inflation has been of vital importance to the poor. After high levels of inflation in the post-crisis
period, the rate of inflation fell back to 6 to 7 percent in 2003-04. However, the governments bold decision to reduce fuel
97
subsidies in 2005 caused the inflation rate to surge to over 15 percent on a year-on-year basis. Despite this setback, the
secondary impact of the fuel-price increase has been limited. Recent developments suggest that the rate of inflation
should return to single digits by the end of 2006. This return to more benign levels of inflation is in part due to the move
by Bank Indonesia to officially adopt inflation targeting in July 2005. Bank Indonesia currently aims to achieve an inflation
rate of 7 to 9 percent for 2006 and 5 to 7 percent for 2007. This increased emphasis on price stability is a positive
development for macroeconomic stability and hence for the poor.
Crucial to restoring low inflation is ensuring a stable and competitive exchange rate. Exchange rate
fluctuations directly affect the price of tradable goods. The rupiah exchange rate is still susceptible to both domestic and
external shocks, and credible macroeconomic management is key to containing exchange-rate fluctuations. Although the
rupiah exchange rate depreciated to its lowest level in four years during 2005, Bank Indonesia s prompt action to increase
interest rates, as well as the government s decision to reduce fuel subsidies, were both major factors in quickly restoring
market confidence.
Following a sound fiscal policy has helped to create fiscal space . For the first time since the crisis, Indonesias
state budget has enough fiscal space to allow for increases in spending on education, health and infrastructure, without
jeopardizing fiscal sustainability. Fiscal sustainability has therefore been largely achieved. The budget deficit to GDP ratio
improved to 0.5 percent of GDP in 2005, while the government s debt to GDP ratio improved from a peak of 100 percent
in 1999 to less than 50 percent in 2005. The combination of declining debt burdens, the significant reduction in fuel
subsidies and increasing non-oil tax revenues have all helped to expand the governments
fiscal space , creating new opportunities for the implementation of pro-poor spending
policies.
As the regions start to shoulder a higher share of public spending, so they
will have a growing impact on fiscal policy. Since decentralization in 2001, the
role of regional governments has significantly increased. The regional share of total
public spending has grown to over 50 percent. Also, higher commodity prices, especially
for oil and minerals, have contributed towards a swelling of the financial resources of
regional governments. Sound fiscal policy (from both central and regional governments)
is important in fostering macroeconomic stability. Key to achieving this will be the
development of an efficient public financial management mechanism.
Keeping an open trade regime is also important for the poor. Indonesia
already has a very open trade regime with an average tariff rate of 8.5
percent. The government recently announced the medium-term tariff harmonization
program aimed at moving towards a low and uniform tariff rate. The governments
commitments to Afta (the Asean Free Trade Area) and the exemption from duty of certain
goods under various trade facilitation programs are likely to lower the effective average
tariff rate still further. This should benefit the poor by cutting costs of imported goods
and encouraging competition in domestic markets.
But trade restrictions, for example on rice imports, have the potential to
hurt the poor. Nevertheless, some aspects of the current trade policy, such as the
restrictions on the importation of rice, are detrimental to the poor. The significant increase
in domestic rice prices between March 2005 and March 2006 was partly attributable to
the restrictions on rice imports.
98
Improving the general investment climate is also good for the poor. Since the inauguration of the new
administration in October 2004, confidence in macroeconomic policy has greatly improved. One indication of this
improvement can be seen in the three packages recently formulated that are aimed at improving the investment climate,
and opening up the infrastructure and financial sectors. However, effective policy implementation is a key concern. In
terms of improving the investment climate, micro-risks of government failures, such as poor contract enforcement,
corruption and inadequate infrastructure, are binding constraints to higher levels of private investment in Indonesia.
Investing in the Capabilities of the Poor
Boosting agricultural capability

Higher agricultural productivity used to be a key pathway out of poverty. Much of Indonesia s remaining
poverty is caused by low agricultural productivity, so finding a way to raise it would be among the most pro-poor actions
the government could take. Improving the capability of farm households to utilize new and improved agricultural technology
can reduce poverty in two ways. First, by raising farm productivity and household incomes directly, technology adoption
can speed farmers along this pathway out of poverty. Second, higher farm incomes have significant multiplier effects on
the rural, non-farm economy.
But recently productivity growth in the agricultural sector has been low. Agricultural total factor productivity
growth turned negative after the early 1990s, from annual gains of 2.5 percent in 1968-92, to annual contractions of 0.1
percent from 1993 to 2000 (Fuglie, 2004). There are several reasons for this. First, investment in irrigation stalled after
the crisis, with the result that much of the current irrigation infrastructure is in poor condition. Second, agricultural
extension services have suffered a serious decline, particularly since decentralization, with the result that fewer farmers
are actually able to receive such services. Finally, the technical options for improving agriculture are currently limited,
and have been since the early 1990s, especially for rice. There are few immediately applicable new technologies likely to
provide a significant boost to yields.
The government s strategy for rural poverty reduction is therefore right to focus on revitalizing agriculture.
The Medium-Term Development Plan (RPJM) and the Strategy for Revitalizing Agriculture, Forestry and Fisheries (RPPK)
stress the need to boost agricultural productivity and to develop agro-industry. This will require a multi-faceted and
coordinated response from several line ministries (including the Ministries of Public Works, Agriculture, Forestry, Fisheries,
Home Affairs, Health, Transportation and Education, as well as local government agencies), which will deal with both onfarm and off-farm initiatives to improve productivity and empowerment of rural communities. The objectives of the
government s strategy should be threefold:
Higher productivity for rice farmers and the livestock economy. To raise rice productivity, improved
irrigation structures and management will be most critical. There are also opportunities to raise productivity
through the precision management of inputs, the consolidation of rice-farming operations through land rental
or sale markets (and not just for rice land), and production for specialized rice markets to meet more sophisticated
consumer demand. Moreover, if higher rice productivity is to help reduce rural and urban poverty, Indonesia s
domestic rice price cannot remain substantially above import parity, as rice price increases disproportionately
hurt the poor (see Box 3.5 in Chapter 3).
Diversification into higher-valued crops. Diversified farming will be the solution for farmers whose scale
of operations or land quality does not enable them to earn a livelihood from rice farming. Indeed, such rural
diversification has been an important source of higher productivity and a pathway out of poverty in most advanced
99
countries. It is an integral part of a successful structural

transformation. Diversification means switching to high-value
crops, livestock and fish production in response to new types of
consumer demand that are transmitted through modern supply
chains. Many small farmers will need technical assistance from
either the public or private sector if they are to respond
successfully to these new opportunities.
Expansion of export commodities. There is a historic

opportunity to expand the production of export commodities, as
the market values of several of the most important ones,
especially rubber and palm oil, reflect the new reality of scarce
energy supplies. There is also an opportunity to improve the
quality of Indonesia s beverage crops, especially cocoa, coffee
and tea, which would rapidly lead to higher incomes for the small
farmers producing these commodities. Finally, Indonesia would
almost certainly benefit from an aggressive global marketing
campaign to brand its tropical products, perhaps building on
its image as the Spice Islands .
These three objectives can be achieved through action in five different areas:
1. Improving irrigation quality and management

The RPJM recommends increasing public investment in irrigation infrastructure and management.
The systematic underfunding of the maintenance of irrigation systems has led to at least one-third of the 3 million
hectares of government-designed irrigation schemes being rehabilitated twice in the past 25 years (ADB, 2004a).
Growing water scarcity is projected to slow irrigated crop yield growth. The problem of increasing scarcity of water is
heightened by the rising costs of developing new water sources, soil degradation in irrigated areas, groundwater
depletion, water pollution and the degradation of water-related ecosystems, as well as the wasteful use of existing
water supplies. Tackling all these issues will require a cross-sectoral effort between the concerned line ministries.
The RPJM rightly calls for increased participation by water users. In recent years, the government has
developed a localized water-management model that places Water Users Associations (WUAs) at the center of decisionmaking, in close cooperation with local governments. Experience shows that such associations are effective in
enhancing good water use, leading to higher productivity; innovative water uses (crop diversification, fisheries
development, etc); better income generating opportunities; sustained preventive maintenance; and a more positive
partnership between local government, its farmer constituency and national line agencies.76 In addition, securing
protection and equity for existing non-formalized customary rights to water resources is a prerequisite to establishing
orderly, equitable and transparent processes of water re-allocation in order to meet communities changing needs.
This will require strengthening the nascent basin organizations (Balai PSDAs-Pengelola Sumber Daya Air) approach
to water-resource management in order to better manage scarce water resources and to optimize their allocation.
Irrigation investments have a high rate of return if they are accompanied by a more participatory
approach to irrigation management. A cost-benefit analysis conducted for the Water Resources and Irrigation
Sector Management Project (WISMP) showed an economic rate of return (ERR) of 36 percent. Moreover, an enhanced
76
100
See the preparatory studies carried out under the World Bank funded Water Resources and Irrigation Sector Management Program (WISMP).
cost-benefit analysis was also carried out to test the justification of

applying the government s new policy of Participatory Irrigation
Management (PIM) to that of traditional physical rehabilitation. This
showed an ERR of 32 percent for the incremental investment associated
with the implementation of PIM, but only 11 percent when the
implementation of participatory management failed.
2. Diversifying agriculture into higher-valued crop and

livestock systems
Diversification raises productivity. The diversification of
agriculture into higher-valued crop and livestock systems especially
by facilitating access to supermarket supply chains, including
mechanisms for collective action by farmers to reduce transactions
costs is one effective way of raising agricultural productivity.
There is already a strong trend of diversification away
from rice. The RPJM and RPPK recognize the importance of
diversification for improving efficiency and farmer welfare, and the
growing importance of globalization in agricultural competitiveness.
However, measures are called for that go beyond just the RPJM s
focus on diversification in consumption away from rice. The latter
is already happening. Between 1996 and 2002, despite the economic crisis, per capita food consumption in Indonesia
increased by 8 percent in real terms (Susenas, 2005). All of this growth took place in high-value foods, such as
animal products, fruit, vegetables, fish, fats and oils, and prepared foods. Per capita consumption of low-value
grains and tubers actually declined (Susenas, 2005). These changes have induced rapid growth in supermarkets,
which have further influenced the agricultural production structure, including processing, handling and marketing.
Diversification is key in the estimated 24 million hectares of dryland areas77 where measures that encourage livestock,
vegetable intercropping, reforestation of small areas with high-value wood species, and diversification into cashew
or fruit, could all contribute to more stable incomes and poverty reduction.
Partnership between traders, processors and producers in a system of effective self-regulation
will be crucial to further diversification. Much of the future productivity increase in dryland areas from
higher-value commodities (smallholder estate crops, horticulture, livestock and fisheries) will only be possible if
there are effective partnerships among producers, traders, processors and a public sector more focused on regulation
and research. Indonesia s agricultural product regulatory framework is fairly well developed but attention is needed
on capacity building, maintaining the integrity of national systems with decentralization, and focusing on assisting
smallholders to meet trade specification requirements. Private markets depend on an effective and streamlined
regulatory environment, including grades and standards, food safety, bio-safety and environmental regulation, in
order to lower transaction costs. Regulations alone are not enough, however. They must be matched by a partnership
with traders, processors and producers in a system of effective (and transparent) self-regulation. This will be important
not just for domestic consumer protection and safety, but also to gain and maintain access to international markets.
Importing countries are increasingly tightening the quality/safety requirements for food products with measures that
include permit traceability of products all the way back to the farm level. Without attention here, a focus on productivity
gains for farmers could fail to translate into welfare gains for the poor if market outlets are limited.
77
Ministry of Agriculture, Bureau of Planning.
101
3. Boosting agricultural research and redesigning agricultural extension

Rural productivity needs agricultural technology. Broad-based growth in rural productivity needs effective
systems for generating, adapting and disseminating technology relevant to small-scale producers. High-quality
agricultural research and extension systems will be critical to getting productivity onto a higher growth path.
This can be done by boosting research. Indonesian agricultural research expenditure has declined
dramatically since the early 1990s compared with its neighbors. Real expenditure on public agricultural
research in 2001 was no greater than in 1995; presently, it ranks near the bottom compared with other Asian countries
in terms of agricultural research spending relative to agricultural GDP and total government expenditure on agriculture.
Indonesia provides less than 0.1 percent of agricultural GDP to support agricultural research in the country (less than
Bangladesh and well below the recommended level of 1.0 percent) (ADB, 2004a). The immediate challenges for the
agricultural research system are to: (1) increase the overall level of national research expenditure despite closure of
several current research projects; (2) clarify the public funding responsibility for the sub-national adaptation institutes;
(3) counter the effect of decentralization that increases the administrative overhead costs of this sub-national system; (4)
replace the significant proportion of senior researchers nearing retirement; (5) integrate private-sector agricultural research
capacity as part of a national strategy; (6) strengthen capacity in biotechnology research; and (7) reinvigorate rice variety
and system research, while nonetheless rebalancing resources and effort to give more emphasis to non-rice commodities.
And also by redesigning extension. Similar to public sector extension systems in many countries, Indonesia
faces a major challenge in developing an effective institutional mechanism for disseminating technology relevant for
small-scale producers. While the RPJM and RPPK do indicate the need for improving extension, new approaches
will be required in the context of the changed institutional environment. There is growing evidence78 of significant
benefits to decentralized extension systems that involve the private sector and civil society, which need to be replicated
nation-wide and strengthened. Local governments need a paradigm shift: (1) from top-down to participatory
approaches; (2) from input and technology dissemination to dissemination of market and upstream information and
technology; and (3) from centrally managed extension services to decentralized services, and some movement towards
privatization of extension.
Privatized extension services will assume greater importance in the dryland cash
cropping sub-sector in eastern Indonesia. This is because exportable commodity production is
being increasingly supported by the private sector. The present political climate in Indonesia also
provides a more conducive environment for a range of rural producer organizations (RPOs) than was
possible in the past. In all of these initiatives, it is important that measures are put in place to better link
agricultural research and extension; the separation of these functions within the organization of the
Ministry of Agriculture (between the Indonesian Agency for Agricultural Research and Development,
or IAARD, and the Agency for Agricultural Human Resource Development, or AAHRD) has made it
difficult to focus on farmers problems when setting the research agenda, while at the same time achieving
effective dissemination of research results.
4. Developing marketing and information technology for agriculture and rural SMEs.
Weak information systems are a barrier to higher rural productivity. The RPJM identifies
the low status of rural information and communication technologies (ICT) development as a significant
barrier to increasing rural productivity and incomes. Rural communities need up-to-date information
on sources, availability and cost of inputs for production, and also on the potential of different techniques
and technologies used for production, processing and marketing. However, the information that is
78
See for example experience from the World Bank Decentralized Agricultural and Forestry Extension Project (DAFEP) and other similar projects of GTZ
(Gesellschaft fur Technische Zusammenarbeit) and the Asian Development Bank (ADB).
102
often most relevant to improving livelihoods is non-technical, including the role and responsibilities of different
institutions in the provision of key services, such as agricultural extension, credit, health and education, and where
to go and who to ask for more specific information. Rural communities increasingly need information about off-farm
activities, about rural development projects and policies, and how to participate in and influence government processes.
ICT offers an opportunity to make rapid progress. It is important that this information is available in an
appropriate format and language, and that rural communities have the capacity to access, analyze and act on it. With
decentralization and the new political and institutional environment in Indonesia there is an opportunity to use ICT to
support the rural development agenda and improve the delivery of government services in innovative ways (World
Bank, 2005b). Indeed, an example of this can be seen at the National Land Agency (BPN), which is making significant
progress in introducing e-administration services in land management.
5. Improving property rights and the land market, both rental and ownership.
Land conflicts and disputes, concentrated ownership and tenure of land, and lack of legal protection
of poor people s rights over land adversely affect income and opportunities for the poor. Fewer than 25
percent of holders of rural land parcels have formal land certificates, compared with almost universal possession of
land-use certificates by farmers in China and Vietnam and widespread certification in Thailand and Malaysia of close to
90 percent and over 50 percent in the Philippines (see Figure 4.7). The low share of landowners that possess a formal
certificate of ownership is also a constraint on access to credit. Analysis demonstrates that rural recipients of land
certificates borrow more, invest more, and earn more from their land-based economic activities (SMERU, 2002).
Figure 4.7 Low land registration in Indonesia
Source. World Bank staff calculations.
Deforested and degraded land is an asset that can be used to help the poor. Around 70 percent of
Indonesia s total land area is classified as forest area and falls under the jurisdiction of the Forestry Law of 1999.
Ministry of Forestry figures show that 28 percent of this forest area is not actually covered by forests and trees but, in
fact, includes settlements and agricultural areas. Although the 1999 revisions improved many aspects of the legal
framework, there are still uncertainties and conflicts over control and access to land in the forest zone. Rationalization
of allowable land uses and greater tenure certainty in the forest zone, particularly on the degraded and deforested
areas already allocated for economic development purposes, would create opportunities for rural investment and
livelihood improvement that could help the poor (see Box 4.1 and World Bank (2006m).
103
Box 4.1 Rationalizing the use of forest lands can help the poor
Reallocating degraded, deforested land to productive uses by smallholders and the poor is one a way to rationalize land use and reduce poverty (ICRAF,
2005; CIFOR, 2004; DFID-MFP, 2006; World Bank, 2004). Some of the designated forested areas are actually community-managed agro-forests, while
some of the non-forested areas are not simply degraded forests, but agricultural land (ICRAF, 2005).
Rationalizing the use and management of these lands would benefit both the economy and the rural poor, by allowing degraded land to move into more
productive use, and by removing uncertainties that are a barrier to rural investment. A quarter of Indonesians stand to benefit from a policy of rationalizing
the use and allocation of forest land.
Production Forest and Conversion Forest areas, which represent 60 percent of the overall forest zone , also accounts for three-quarters of degraded land
in the country (24.4 million ha). This land is already allocated for productive economic uses. By reclassifying this land, it will allow for more productive
land-use and ownership patterns and could also lead to increased tree cover and protection of the land.79
It has been estimated that with different enabling policies, such as increased land availability, secure access and tenure, or improved productivity (DFIDMFP, 2006), small reallocations of land or increases in security for investment in land productivity can yield high returns, of up to US$1.4 billion a year
in added revenues and possibly 1.6 million more jobs. In other words, it is estimated that this kind of initiative could benefit up to 8 million people, or 25
percent of Indonesia s poor about 80 percent of the poor living in the forest zone. These benefits would materialize over a decade as investments in land
mature and began to reach markets.
Non-forestry land registration needs to happen faster. The government has distributed more than 2.2
million titles to landowners since 1994 through systematic land-titling projects,80 increased capacity at the BPN, and
carried out a comprehensive review of the policy and legal reforms needed to modernize the land system under
democratic, pro-poor principles (Bappenas, 2005). However, titling on non-forest land has been slow. Only about 25
percent of Indonesia s estimated 80 million land parcels have been registered in the 40 years since land registration
began. At the current pace of registration, it will be difficult for land registration to catch up with the growing number
of parcels. In addition, a large share of land off-Java is communal land and private titling of this land may only work
against the poor and increase conflict.
Improving education and vocational training

For those working in non-farm activities in both rural and urban areas, the priority is to boost their
capability to obtain better jobs. Boosting capabilities in agriculture is a key priority for generating pro-poor growth
for the rural agricultural poor. However, for many others, particularly those involved in non-farm activities in both rural
and urban areas, the priority will be boosting their capability to obtain better jobs. As Chapter 3 shows, Indonesia has a
thriving informal sector, so access to work per se is not the major problem. Nor is the amount of work necessarily a major
constraint for the poor: many also work long hours, even if the intensity of work and productivity per hour may be low. The
main constraint for the poor is the lack of access to regular work in which returns are sufficiently high and stable to
support basic family needs.
There are two reasons why the poor fail to access better jobs. First, the national economy is not yet generating
enough employment opportunities, particularly for unskilled workers. We discuss how policy might help to boost the
79
Some 30 percent of these Production Forest and Conversion Forest areas lack tree cover. Sources and accuracy of data are always controversial when it comes to forest land. These figures are based on the
Recalculation of Forest Resources from 2003 that is published with the Statistik Kehutanan in 2004 and is reported on the Ministry of Forestry s web site (www.dephut.go.id).
80
These projects are the Land Administration Project-LAP (1994-2001), the Land Management and Policy Development Projects -LMPDP (2004-now), and also the Reconstruction of Aceh Land Administration
Project (RALAS) and Land Office Computerization-LOC (National Land Agency-BPN supported by the Government of Spain).
104
employment of the unskilled in Section VI. Second, many of the poor simply do not have the skills that are needed to
obtain better jobs. Building the capabilities of the poor to access better jobs through improvements in education and
training should be a major policy priority for the government.
The return to education is higher at higher levels of schooling, so focusing education investments only
on primary education is no longer enough. The benefits of openness, technology and market competition for
economic growth depend greatly on education and skill levels, and so increasingly higher levels of education have
become more important for retaining the benefits of growth. However, in Indonesia the additional earnings that one can
expect to receive from an additional year of education (the return to education) are higher for individuals who already
have more education (see Figure 4.8). In 2002, the increase in urban (rural) male wages resulting from an additional year
of schooling for someone who had only one year of schooling already was 8.3 percent (6.0 percent); after five years of
schooling, the return was 10.0 percent (7.6 percent) and after eight years of schooling it was 11.1 percent (8.8 percent).
This increasing return to schooling can lead to increased inequalities, as the rich have greater access to higher levels of
education. In order to ensure that the poor also benefit from the high return to education a focus on secondary education
will be required primary education is no longer enough.
But market failures are preventing the poor from accessing higher levels of education. When there is a high
return to education, the fact that the poor are dropping out of school at an early age may signal that there are market failures
relating to (i) credit constraints that reduce the ability of the poor to borrow for schooling, and (ii) imperfect information on the
return to schooling. The registration fee for junior secondary schools is a great barrier for enrollment in junior secondary
schools for poor families: the average registration fee for junior secondary schools in Jakarta amounts to 30 percent of a poor
households expenditure.81 There are high externalities to education investments for society as a whole, as well as equity
implications of providing education for the poor given the increasing return to education, so public action to ensure that the
poor also benefit from secondary education is crucial. Also burdensome to the poor are the miscellaneous fees for sitting
examinations, the constantly changing book lists and the variety of uniforms required by students.
There is a widening gap between the return to education in rural and urban areas, so policy needs to
respond to the challenges of urbanization. Figure 4.8 shows that the return to education is lower in rural areas than
in urban areas and that this gap is larger for higher levels of education and has been widening over time. The return to
education in rural Indonesia may be much lower than in urban areas
due to the scarcity of jobs: of the unemployed people with junior (senior)
secondary school degrees in rural areas, 13.4 percent (12.4 percent)
reported their reason for not looking for work as being that it is not
possible to get a job, while in urban areas 6.4 percent (4.7 percent) of
graduates reported such disappointment in the availability of jobs in
the market (see Figure 4.9). This suggests that the priority in rural
areas should be generating jobs through improvements in the rural
investment climate.
The government needs to provide the rural poor with the
skills to escape from poverty. The role of the non-farm rural and
peri-urban sectors in the pathways out of poverty suggests that the
government should focus on providing the rural poor with the skills to
pull themselves out of poverty within an increasingly urban economy.
Indonesia s structural transformation and urbanization require a
81
Calculated from Susenas 2003.
105
comprehensive strategy to balance investment in rural areas with the provision of job information, re-skilling for new
migrants, and basic education services for their children. Otherwise, the growing urban population will give rise to new
forms of poverty and exclusion in urban areas in the years ahead. Within this structural transformation, one of the greatest
challenges will be harnessing the productive potential of urban migrants. In order to achieve this, it is imperative to
provide appropriate education and training to rural populations, as well as to the urban poor.82
Education investments need to focus on the skills and employability of the growing young population in
Indonesia. Youth unemployment is one of the most serious challenges facing the government: in 2004, the total number
of unemployed young people in Indonesia reached 6.4 million, representing 58.5 percent of the unemployed population.83
The country s predominantly young population (29.6 percent under 15 and 37.3 percent under 19 years of age)84 requires
education, while the growing working-age population in Indonesia requires employment opportunities to generate secure
livelihoods for themselves and their dependents. The dependency ratio is declining as a result of more working-age
young people entering the labor force (54.1 percent of population in labor force today compared with 51.2 percent in
1987),85 but this implies there that needs to be enhanced job opportunities for new graduates. In addition to generating
new jobs for unemployed young people, cost-effective educational programs that focus on skills development and enhancing
the employability of young people from disadvantaged households should be central to the government s agenda.
Figure 4.8 The gap between household consumption
associated with high educational endowments has
widened between urban and rural areas for higher
levels of schooling (1999-2002)
Figure 4.9
In rural areas, lack of job availability is a concern

even for graduates of senior secondary schools
Note: the change in household consumption associated with educational attainment of household head is taken from the household expenditure function provided in Chapter 3.
The vocational capabilities of the poor are key. Improving formal education is a key part of building the capabilities
of the poor and ensuring that future generations are able to escape from poverty. But it is equally important to increase the
vocational capabilities of the poor. The government has several instruments available to improve the productivity and
employability of the poor through vocational training.
1. Vocational schools and apprenticeship schemes

It is still unclear how much school-based vocational training should be publicly funded. Currently,
there are over 4,500 vocational schools (sekolah menengah kejuruan, or SMKs) in Indonesia at senior secondary
school level enrolling about 1.4 million students in total, of which about one-third are enrolled in public vocational
82
Core Paper (2006). Asia 2015 Conference on Sustaining Development and Reducing Poverty in Asia, January 2006.
Calculated from Labor Survey (Sakernas) 2004.
84
Calculated from Susenas 2004.
85
Calculated from Sakernas 2004 and 1987.
83
106
schools. In the 1970s and 1980s, the government invested heavily in vocational schools, but over time the provision
of vocational training was increasingly left to the private sector. At the beginning of the 1990s, it was realized that the
wage return to vocational schools was about the same in the labor market as the return to regular senior secondary
schools, while the SMKs were more costly to run. As a result, there was a shift towards less public involvement in
vocational education. By 2002, 83 percent of SMKs were privately run, more than half of all SMKs focused on
business and management training, while about one-third focused their curricula on technology and industry training
(Ministry of Education, 2004). However, today this situation has changed: the wage return in the labor market to SMK
graduates is 23 and 42 percent higher for men and women, respectively, when compared with SMA (senior secondary
school) graduates (Figure 4.10).
Figure 4.10
The ratio of wages for SMK vs. regular senior

secondary school graduates have recovered in
recent years
Source: Sakernas various years.

Note: The analysis takes into account only those students who do not continue
their studies at the tertiary level.
Figure 4.11
Vocational school enrollments are regressive: the

poorest 20 percent of population capture only
about 12 percent of benefits from public SMKs
Source: Susenas, 2004
At present, SMKs are not very effective in helping the poor. Despite their potential advantages, vocational
schools do not enroll many poor students. As a consequence, most additional expenditure on SMKs will not reach
the poor the poorest 20 percent of the population capture only about 12 percent of the benefits from public vocational
schools (Figure 4.11). The return to public investments in vocational schools is related to the quality of teaching, the
ability of schools to follow up with new technologies and the demands of the labor market. Studies on the success of
vocational training need to be carefully updated, as the latest available research dates from the early 1990s. The wage
return to this type of education needs to be assessed against the cost of providing this expensive form of education
and the fact that many people would have found jobs in the absence of training through public funds.
Targeting the poor with vouchers would give them better access to SMKs. In order to make already
existing public spending on SMKs more pro-poor, the government could use a strategy of targeting benefits to poor
students through a voucher/scholarship program similar to the one that exists for senior secondary schools (SMAs).
Alternatively, it could provide block-grant programs to SMKs in exchange for extending services to the community
on weekends and evenings outside of school hours, in order to reach out to poor unemployed adults or young school
drop-outs.
2. Public vocational training centers

Public vocational training centers failed to provide adequate access to the poor in the past. Public
vocational training centers (BLKs and KLKs) under the direction of the Ministry of Manpower were intended to
become short-term (about three-month) training centers for Indonesia s growing manufacturing industry in the 1970s
and 1980s. While these centers received substantial donor assistance at the start-up, they were unsustainable without
public funds, running into problems of low budgets and under-utilization due to low demand towards the end of
107
1980s. The centers were also initially set up to target poorer applicants who had only graduated from primary school
or who had dropped out of secondary school; students were exempt from fees for this reason. By the beginning of
1990s, however, most of the benefits from public vocational training centers were captured by more educated applicants,
since students who had not completed senior secondary school could not qualify in the entrance exams, which
required them to handle technical equipment.
Decentralization now offers an opportunity, but also risks. After decentralization, these vocational centers
have come under the jurisdiction of district governments, whereby some have received more funding while others
have become dilapidated with a lack of funds and modern machinery. The decentralization of these schools to district
governments could mean that the centers will be better able to accommodate local conditions and change their
organizational structures appropriately. However, it may also mean that the funding and quality of these centers could
fall, reducing their effectiveness.
BLKs could take on a new role in targeting pro-poor student selection. Many of the programs offered by
BLKs are already offered by private training centers and BLKs could gradually reduce the direct provision of training
in areas served by the private sector. However, the role of public vocational centers in students selection and placement
could be increased in order to improve the pro-poor targeting of such programs.
3. The dual and apprenticeship (magang) schemes

Apprenticeships act as a bridge between training and the labor market. These apprenticeship schemes
were envisioned as company-based training to connect vocational training schools (SMKs) and BLKs/KLKs with
the labor market. The apprenticeship schemes were implemented by the Ministry of Manpower, while the dual
scheme was carried out by the Ministry of Education. In these schemes, companies select trainees and enter into an
apprenticeship contract with them and devote one production supervisor for training. The cost-sharing of the program
is between the government and the employers.
Incentives could help to overcome employer reluctance to join such schemes. The success of these
types of schemes depends heavily on finding companies that are interested in training through vocational schools
and are prepared to fund supervisors for the training. However, surveys indicate reluctance among employers to
participate in such schemes (World Bank, 1997). In the future, the government might consider providing companies
with incentives for taking on supervision of trainees from public and private vocational training centers, but more
analysis is needed on the potential costs and benefits from the expansion of such schemes.
4. Private training institutions and youth match and link programs

Private training centers serve far more students than public centers. Private training centers in Indonesia
are dynamic market-driven institutions and have been growing in number over the past two decades. In 1994, there
were about 28,000 private training centers registered with the Ministry of Education or the Ministry of Manpower.
These centers were reaching about 4.5 million students annually, compared with only 50,000 students reached by
public vocational training centers (World Bank, 1997). Most of these courses are vocational in nature: of the courses
registered under the Ministry of Education (in 1994) the most popular were sewing, hairdressing, computer courses,
typing and accounting.
Targeting the poor to help them access such training centers is an effective way of raising their
skills and productivity. Recently, the government has implemented a voucher scheme for training centers to
provide economically useful vocational skills to young under-educated, unemployed or under-employed Indonesians.
The scheme identifies and accredits private trainers through local NGO collaboration, and then matches them with
locally unemployed youth aged 16 to 25 through the use of vouchers. The program provides block grants to the
108
community learning centers (PKBM) managed by NGOs in exchange for the PKBM offering services to undereducated youth in this age range. The budget allocation for this project for 2003 and 2004 was about Rp 15 billion,
reaching about 150 learning centers with a grant of about Rp 100 million each. Expanding this type of match-andlink program between poor and under-educated youth and private trainers serves as a cost-effective way of improving
the skill-set of young Indonesians in a market-driven way. Increasing subsidies to this kind of program would be an
effective way for the government to invest more in building the skills and productivity of the poor.
VI
Connecting the Poor to Growth
The poor need to be connected to growth if they are to benefit. Improving the capabilities of the poor is only
part of the solution. Even if their capabilities are enhanced, poor people still need to be connected to opportunities in
order to benefit from growth. But, as noted in Section II, the sorts of connection that matter most depend on which poor
group is being considered.
Rural roads, labor-market reforms and credit are all key. For the rural poor, the top priority should be reducing
the physical costs of connection through improvements in rural infrastructure, notably rural roads. In addition, the physical
and information costs of access to credit are high in rural areas. In urban areas, infrastructure issues are less of a concern
from the perspective of linking the poor to growth (although water and sanitation issues have a direct bearing on poverty.
See Chapter 5.) The key in urban areas is connecting the poor to jobs through reforms in labor-market policies and other
improvements in the investment climate. Again, access to credit for the poor is a major constraint to accessing opportunities,
including problems associated with collateral, complicated procedures and the inability to use a good repayment record
as an asset. We therefore explore how to connect the poor to growth in three areas:
Connecting the rural poor to urban markets

Connecting the poor to finance
Connecting the rural poor to

urban markets
Access to good quality infrastructure is
strongly associated with economic
success. Evidence from Indonesian surveys
matches evidence from the rest of the world,
confirming that access to good quality
infrastructure is strongly associated with lower
poverty often through enabling the success of
non-farm enterprises. This can be demonstrated
through a comparison of the productivity of
enterprises in areas with and without access to
good infrastructure (Willoughby, 2004; Escobal
and Ponce, 2002; Songco, 2002; Lanjouw,
2001).86
86
While the relationship between access to infrastructure and the extent of non-farm enterprises is strong, it should be noted that this does not necessarily imply that improving access to infrastructure alone
will bring rapid growth of non-farm enterprises. Richer villages closer to large, existing markets are far more attractive for telecommunications and energy providers and cheaper to connect to quality
infrastructure. They are also likely to have more non-farm enterprises. Nonetheless, combined with the significance that non-farm enterprise owners attach to improved infrastructure as reported in the RICS
(Rural Investment Climate Survey), the data are strongly supportive of the idea that improving the quality and reach of infrastructure would have a considerable impact on the extent of non-farm activities.
109
All-year sealed roads make a big difference to incomes. For example, estimates from the 1993 and 2000
Indonesia Family Life Survey (Figure 4.12) show that access to improved infrastructure is strongly associated with nonfarm enterprise (NFE) development. Households in villages with predominantly dirt roads earn 39 percent less of their
income from non-farm enterprises than the average across all households. Households in villages where the average
speed of travel to the district capital is faster also see more non-farm income earnings. Villages that had their roads
upgraded between 1993 and 2000 saw significantly faster growth in non-farm enterprise activities than other villages
(Gibson, 2006).
Figure 4.12 Better infrastructure is associated with higher levels of non-farm enterprise activities in Indonesia
Source: Gibson, 2006.

Note: Results are conditional correlations, in the case of telephones on distance and average speed to provincial capital and dirt road, in the case of electricity and blackouts (both in the same regression) also
on distance to provincial capital in the case of roads on distance and average speed to provincial capital.
Indonesia has made tremendous progress in improving rural infrastructure, but major gaps remain.
Overall, at the national level, great advances have been made in improving access to electricity, roads and
telecommunications facilities over the past 25 years. However, the development of this infrastructure has been uneven and
areas outside Java/Bali lag behind in all areas (see Table 4.4). For example, the percentage of villages connected by a
concrete or asphalt road is as low as 46 percent in Central Kalimantan. Similarly, while all provinces in Java/Bali had
electrification rates above 90 percent, in Central Sulawesi only 64 percent of homes are served by PLN (the state electricity
company).
110
Table 4.4
Access to infrastructure by region
Province
Villages with
PLN
Villages with
concrete/asphalt
electricity (%)
road (%)
Villages with
telephone (%)
Infrastructure
investments
Population
density (per km2)
per capita
1994-2002 (US$)
Sumatra
North Sumatra
90
97
75
66
57
63
128
83
158
West Sumatra
Riau
86
87
74
91
55
77
89
287
99
52
Jambi
South Sumatra
82
85
86
81
66
51
157
130
45
74
Bengkulu
Lampung
98
87
93
65
32
50
219
80
79
191
Java/Bali
97
67
65
81
100
93
99
100
97
98
100
68
65
79
63
99
100
64
47
72
82
78
593
36
33
80
42
163
12,635
1,033
959
980
726
559
Nusa Tenggara
West Nusa Tenggara
East Nusa Tenggara
75
88
71
59
79
53
29
42
26
119
108
130
199
83
Kalimantan
West Kalimantan
Central Kalimantan
South Kalimantan
East Kalimantan
90
94
77
94
93
71
72
46
77
88
58
52
45
55
84
301
137
378
147
681
27
12
69
11
Sulawesi
North Sulawesi
Central Sulawesi
South Sulawesi
Southeast Sulawesi
88
89
64
97
91
70
72
82
72
55
41
34
43
55
19
130
189
253
61
230
132
35
129
48
Indonesia
96
74
69
DKI Jakarta
West Java
Central Java
DI Yogyakarta
East Java
Bali
Source: Village Potential Survey (Podes) 2000, Regional infrastructure investments from World Bank calculations based on Regional Financial Statistic (Central Bureau of Statistics-BPS: 1994/1995-1999/
2000); Regional Fiscal Information System (MoF: Regional Government Development Budget-APBD 2000 summary realization, 2001 detail realization, 2002 detail plan); Population from 2000 Population
Census results.
Lack of access to high quality infrastructure is still a major concern, particularly for rural businesses.
Figure 4.6 shows the most important problems faced by non-farm enterprises in the six districts surveyed in the Rural
Investment Climate Survey (RICS) (World Bank, 2006h). Road access, the cost of transportation, the quality of roads, and
the quality of electricity all feature strongly in the top concerns of the 2,500 enterprises surveyed. Moreover, analysis of
these data suggests that reducing the average response across infrastructure variables from somewhat of a problem to
not a problem would be associated with a rise in the average proportion of income in a village coming from non-farm
enterprise income and non-farm salaries and wages by 33 percentage points. A Wartel (telecommunication kiosk) in a
village is associated with a 10 percent increase in non-farm earnings, while in cases where most roads out of the village
are dirt (rather than telford, concrete or asphalt) the share of non-farm enterprise earnings drops by 12 percent.
111
To connect the rural poor to urban markets, improving the quality of district and sub-district roads
should be a priority. Around 290,000km of roads (or about four-fifths of the national total) now fall under the
responsibility of district governments. However, a considerable proportion of these roads are in poor condition (Table
4.5). Furthermore, the condition of district roads seems to be deteriorating rather than improving. As just one example,
the district of Manggarai saw over 60 percent of all its roads classified as being in bad or very bad condition, up
from 48 percent in 1999 (World Bank, 2006c). Looking only at its district-level roads, 86 percent were in bad or very
bad condition, up from 58 percent in 1999. Overall, 37 percent of district-level roads are in damaged or seriously
damaged condition, a much higher percentage than the share of national or provincial roads in such condition.
Table 4.5 Condition and surface type of district-level roads, 2003
Sumatra
(km)
Java
(km)
Bali & Nusa Tenggara

(km)
Kalimantan
(km)
Sulawesi
(km)
Maluku & Papua

(km)
(km)
Total
%
Asphalt
41,814
61,948
12,389
Surface type
9,537
23,718
3,703
153,109
52
Gravel/Stone
Earth
15,580
25,875
10,409
10,099
4,128
8,142
4,417
9,925
7,275
9,411
927
8,917
42,736
72,369
15
25
Other
6,963
1,487
1,041
4,357
4,202
6,510
24,560
Total
90,232
83,943
25,700
28,236
44,606
20,057
292,774
100
Good
Moderate
Damaged
Seriously Damaged
29,779
22,215
21,815
16,423
36,183
22,433
18,283
7,044
9,217
5,456
7,295
3,732
Condition
7,183
5,684
8,818
6,551
18,357
9,557
6,939
9,753
5,179
12,386
903
1,589
105,898
77,731
64,053
45,092
36
27
22
15
Total
90,232
83,943
25,700
28,236
44,606
20,057
292,774
100
Source: BPS, 2003.
Investing in district and sub-district roads can give high rates of return. One concern about investing in
district and sub-district roads is that they tend to be much more expensive per kilometer than national or provincial roads,
while low population densities in more remote locations can mean that the benefits from their improvement are limited.
However, evidence from community level infrastructure projects constructed as part of the Kecamatan Development Project
(KDP) show that this does not necessarily have to be the case. Road projects implemented as part of KDP had an economic
internal rate of return of between 33 and 47 percent (Torrens, 2005) and were 56 percent cheaper than equivalent roads
built through government contracts. It will be important to translate the lessons about low cost infrastructure provision
learned from community level projects such as KDP into effective mechanisms for reducing the costs and maximizing the
benefits of more technically sophisticated district and sub-district roads.
Most of the money for district roads comes from the central government. District governments are responsible
for all road functions not specifically assigned to central or provincial governments, which includes the construction,
upgrading and maintenance of all non-national and all non-provincial roads. Funding for this work comes predominantly
from the central government s general allocation fund (DAU), although funding can also come from the special allocation
fund (DAK), grants from aid agencies, and central government support to rural infrastructure projects financed by savings
from reduced fuel subsidies (PKPS-BBM), as well as district governments own-source revenues.
The poor quality of the road network is a result of the dramatic decline in government expenditure on
roads after the crisis. Analysis in Chapter 5 shows that expenditure from all levels of government on roads dropped
dramatically after the crisis. This prolonged collapse in expenditure was one of the main reasons for the decline in road
quality at all levels, but with a particularly damaging impact on district-level roads.
112
But lack of funding in aggregate is now not the main problem. Road expenditure nearly recovered to its precrisis level in 2002 as a percentage of GDP. In the last couple of years, local governments have seen very large increases
in their DAU allocations, partially as a result of redistribution of the gains from the reduction in the fuel subsidy. As a
result, the aggregate level of funding is not now the principal problem.
The distribution of resources does not always reflect needs. The distribution of resources to the districts
entailed in the current DAU formula does not reflect the road infrastructure needs of different districts very well.
Decentralization led to the abandonment of the Inpres performance-oriented transfer programs, which helped to achieve
national minimum service standards in road infrastructure, with funds granted on the basis of road length, condition,
density and cost. The DAU by contrast is allocated primarily on the basis of population. Therefore, districts with extensive
district road networks but sparse populations tend to receive fewer funds than they need to adequately maintain their road
networks.
The key problem is that not enough money is being allocated to maintenance. At the start of the 1990s, road
maintenance accounted for about 47 percent of the central government grants to districts for roads. By the end of the
decade, this had dropped to 15 percent (Parikesit, 2006). In 2001, routine spending on roads accounted for 8 percent of
infrastructure spending by government. This dropped to only 4.5 percent by 2003 (although total infrastructure spending
did rise over this period) (World Bank, 2006h, Chap 4). Estimating that 5.0 percent of total roads expenditure goes on
maintenance, this suggests that annual government spending on maintenance is about Rp 986 billion. Estimates for the
cost of full routine and periodic maintenance of just the district network are around 17 times this figure (Table 4.6),
suggesting that much more needs to be spent on maintenance.
Table 4.6
Maintaining and upgrading the district road network: estimated costs

Maintenance
Upgrade cost (Rp bn)
cost per year (Rp bn)
Asphalt
Stone/gravel
Earth
9,324
2,513
4,509
14,801
1,282
1,447
Total
16,346
17,530
Source: Based on surface type from Parikesit (2006), condition from World Bank (2004b), upgrade and maintenance cost from World Bank (2006a).
Note: Based on the assumption that asphalt is divided equally between 6m hotmix, 3m hotmix and impact.
International experience shows that road repair costs increase dramatically if maintenance is neglected.
If defects are neglected, an entire road section may fail completely, requiring full reconstruction at three times or more the
cost, on average, of maintenance costs. For example, the South African National Road Agency Ltd. (Sanral) estimates that
repair costs rise to six times maintenance costs after three years of neglect and to 18 times after five years of neglect. To
avoid such escalating costs, Sanral first allocates its available funding resources to ideal maintenance actions and thereafter
to more extensive works.
But in Indonesia road expenditure is biased towards new road construction/upgrading. Around 95 percent
of all road expenditure goes on new road construction, the reconstruction of poorly maintained roads, and upgrading
roads to a higher standard. Moreover, this heavy bias towards construction/upgrading rather than routine maintenance is
perpetuated by low maintenance expenditure itself since, as the roads degrade more quickly, there is greater pressure for
higher expenditure on periodic maintenance and upgrading. Higher routine maintenance expenditure would reduce
substantially the capital costs associated with periodic maintenance and upgrading, enabling these activities to take a less
dominant share of the budget.
113
Why do district governments not spend more on maintenance?
The technical rules for maintenance are not yet in place. One adverse consequence of decentralization
is that it has resulted in the removal of the former centralized standards-setting institutional structure, as well as
the associated financing incentives to meet such standards.
The local political economy demands new roads. Newly directly elected district heads and local legislators
are under considerable local political pressure to deliver new asphalt roads. Each community that does not
currently have access to an asphalt road would clearly prefer this to any other option. But such communities do
not take into account the fact that other communities are asking for the same thing, and that resources are not
available to provide new asphalt roads for all. Local politicians respond to these pressures by biasing expenditure
towards highly visible asphalt roads rather than better maintenance of existing earth or gravel roads (World
Bank, 2006c).
More money can be corrupted from new roads. Sadly, road construction is a major source of corruption
both in Indonesia and elsewhere. Anecdotal evidence suggests that road contractors are major contributors
towards the election campaigns of district heads who, in turn, may influence key local officials to award
construction contracts to the same contractors. Corruption may also account for the choice and type of roads to
be constructed or upgraded in a district. Choosing to construct asphalt rather than Telford roads not only
increases construction costs by 80 percent, but doubles maintenance costs. Despite this, all recent road upgrading
in a case study in the district of Manggarai, East Nusa Tenggara, involved asphalt (World Bank, 2006a). In
general, the larger quantities and more complex procedures involved in asphalt construction allow for greater
opportunities for corruption in construction. Corruption is thus a factor in the preference for expenditure on
road construction (especially asphalt) rather than maintenance.
What should be done?
114
Create more resources for road maintenance. For example, this could be done through the creation of
a district or provincial road fund. Such funds would collect resources from road users (including license and
registration payments) and use these resources to maintain the road network under the oversight of a board
involving both public and private sector representatives.
Introduce an appropriate district level Road Management System. An equivalent of the Indonesia
Road Management System which governs planning and budgeting at the national and provincial levels to
ensure budgeting is adequate and well-targeted is also needed at the district level. This would ensure that new
investments are:
a) Based on verifiable (and sensible) design and selection criteria, which include consideration of spillover
benefits to neighboring districts;
b) Being transparently and competitively bid out for construction; and
c) Being constructed to a high standard based on independent quality audits carried out with inputs from
NGOs and local transport operators.
Greater use of performance-based contracts. For example, some central funds for road investments
could be linked to the production of evidence that the existing road stock is being properly maintained. In
addition, elements of road maintenance could be contracted out to private providers. Third-party verification of
the quality of the maintenance operation might be used by DAK administrators as part of yardstick competition
(comparison of performance between similar districts) to set standards to be met in order to receive investment
funds. Local transport operators could again play an important role in providing inputs into the verification
process, as a significant customer base for the road network.

For the poor to participate in growth there must be jobs for them to do. Connecting the rural poor to urban
markets will be critical to helping them escape from poverty. Most of the urban poor are already connected to such
markets, so the key for them is to obtain better employment within urban areas. Section V discussed how to build the
capabilities of the poor to enable them to participate in growth. However, in order for the poor to participate there must
actually be jobs for them to do. Unfortunately, the rising costs of labor-market legislation and regulations in the postcrisis period, ranging from a strengthening of minimum-wage legislation to highly onerous terms and conditions for
hiring and firing have had a significant impact on employment prospects particularly of the poor. This section explores the
impact of labor market policies on the employment of the poor.
Formal labor market policies affect wages in the informal sector. In an integrated labor market such as
Indonesia s, regulations in the formal sector can have immediate repercussions on employment uptake and productivity
in the informal sector where most of the poor work. For example, restrictive labor regulations that decrease employment
in the formal sector will cause an increase in levels of employment in the informal sector, as displaced workers move from
one sector to the other. In order to absorb this increase in the supply of workers in the informal sector, marginal productivity
and wages will fall as a consequence.
There are three aspects of national and provincial labor legislation that have an important impact on employment growth,
particularly in the formal sector: legislation and regulations on minimum wages; severance pay; and, contract workers
and outsourcing.
Minimum-wage regulations
Minimum wages have risen sharply in recent years. The Indonesian labor market has experienced a significant
change since the early 1990s, with the implementation of regional minimum-wage regulations, which have been updated
annually. SMERU (2001) reports that the government tripled the minimum wage in the first half of the 1990s, and the
nominal wage continued to increase during the latter half of the 1990s. The real value of minimum wages began to taper
off after 1996 and fell significantly in 1998. Since 2000, the economy has recovered and the various governments since
then have vigorously pursued minimum-wage increases (Figure 4.13).
Figure 4.13
Real minimum wages in Indonesia have been on the rise in recent years
Source: World Bank Office Jakarta database.
115
Minimum wages are now under the control of regional

governments. Since the implementation of decentralization
in 2001, power to determine minimum wages was transferred
to the heads of regional governments, i.e. governors, mayors
(walikota) and district heads/regents (bupati). The framework
for setting the minimum wage was established by the national
government and is implemented by provinces and districts. Each
districts minimum wage is anchored around the minimum wage
set by the province. Minimum wages increase each year in line
with minimum basic subsistence needs (KHM).
Minimum wages have been widely, although not
uniformly, adopted by businesses in the modern
formal sector. Large foreign-owned businesses usually do
not see the minimum wage as an important cost burden and it
is not among the most important factors for potential investors
in choosing a location for their investments. It is a problem,
however, for labor-intensive industries facing stiff competition
from similar producers elsewhere in the region, notably Vietnam
and China.
Evidence of the impact of minimum wages on formal
employment in Indonesia is mixed. SMERU (2001)
suggested that increases in the minimum wages had a negative
impact on urban formal sector employment after the crisis. This
negative effect was greatest for those groups that are most
vulnerable to change in labor market conditions, such as
females, young workers and less educated workers. However,
Alatas and Cameron (2003) found different results from a study
of the impact of minimum wages under more favorable labormarket conditions prior to the crisis. Their study looked at the
impact of minimum wages on employment in the clothing,
textiles, footwear and leather industries. Their results suggest
that the increase in minimum wages had no significant
employment effect for large firms, both domestic and foreign.
In contrast, the increase of minimum wages had a significant
negative impact on employment in small domestic firms. An
earlier study by Rama (1996) also found similar results.
Minimum wages may be a more important constraint
on employment growth now than before the crisis. The
different results from studies on the labor market before and
after the crisis suggest that minimum wages may be a more
important constraint upon employment in the formal sector now
than in the more favorable labor-market conditions prior to the
crisis. But it should be noted that the vast majority of micro
and small businesses at the district level do not conform to the
116
minimum wage legislation at all. For example, a case study conducted in the
district of Serang, Banten (World Bank, 2006i) showed that while large and
foreign-owned firms claim to pay their workers at, or even above, the minimum
wage, many small or medium-sized, and even some larger, Indonesian firms
still pay their workers below the minimum wage. Such broad non-compliance
implies that minimum wages are unlikely to be a major constraint to the growth
of most small enterprises.
Severance pay
Severance pay regulations may also restrict mobility in the formal
sector labor market. Dismissal regulations in Indonesia that include
provisions governing severance pay have seen significant changes since 1996,
both in terms of rates and coverage to various groups of workers. While several
countries are reforming their severance-pay systems to reduce dismissal costs,
Indonesia appears to be moving in the opposite direction, with measures that
increase dismissal costs.
Rates of severance pay for workers with long years of service were
increased with the implementation of Law on Manpower No. 13/
2003. The regulations define the rights of workers and rates of severance pay
and long-service payments depending on the cause of separation. Three broad
categories of reasons for separation include:
Voluntary quits;
Dismissal for economic reasons (i.e. downsizing and bankruptcy);
and
Violations (minor and major violations or offences).
For workers dismissed for economic reasons, the law increased the overall
rate of severance pay almost three-fold compared with the 1986 regulations.
But evidence from a case study in the district of Serang suggests
that most small and medium businesses and even many large
labor-intensive firms do not comply with the law (World Bank,
2006b). In the event of economic downsizing, many firms claim that they
could not afford to pay according to the 2003 law, and that they therefore opt
for a compromise settlement through tripartite bargaining. In many cases, the
practice is to pay severance payments of about one to two months of salary,
regardless of legal entitlement.
Despite substantial non-compliance, the law still causes concern
to businesses. Many firms are worried about their ability to stay competitive
in the face of slow business growth and the threat of competition from similar
industries in the East Asia region. There is a tendency for firms to increase the
use of short-term contract workers and outsource their orders, especially for
firms with fluctuating orders, in order to avoid using employees who would be
entitled to higher severance pay.
117
Short-term contract workers and outsourcing

Similarly, regulations governing the employment of contract workers and the outsourcing of labor appear
to be becoming more, rather than less, restrictive. In particular, the Ministerial Decree No. 100/2004 on Temporary
Working Agreements (or PKWT) creates a more restrictive regulatory environment. Among other provisions, this decree
contains the following:
Contract renewals: While the previous decree permitted a once-only renewal of a short-term contract, it is
unclear whether this is possible under the new decree;
Reporting requirements: There are requirements to report the names of all daily and fixed-term contact
workers to the local manpower office (LP3E FE UNPAD-GIAT, 2004);
Limitations on terms of contracts: Under the new decree, employers can only hire workers on limited-term
contracts of no more than two years with a one-year extension allowed for a range of reasons.87 Employers can
hire workers on contract for seasonal work or daily work lasting no more than three months, otherwise they
must become permanent workers; and
Limitation on activities: The new law also limits labor outsourcing of production and services to non-core
activities, such as cleaning services, security and catering.
Restrictions on the choice of employment arrangements have discouraged firms from hiring new
permanent workers in the formal sector. However, these regulations are unlikely to have a direct effect on employment
in informal sector where the regulations are not binding. The impact on firms in the informal sector comes primarily from
the increased supply of labor to the sector as a result of slower employment growth in the formal sector.
Evidence points to the increasing use of contract workers and outsourcing. The case study in the district of
Serang confirms a trend towards the use of contract workers and outsourcing due to fluctuating orders; excess supply in
the labor market; and the impact of the Law No. 13/2003, particularly as it regards severance pay. Firms have largely
shifted to limiting the number of permanent workers hired, or have avoided hiring permanent workers altogether. It is not
uncommon to find firms with fluctuating orders that hire 20 percent of their workers on a daily basis. There are also cases
where firms employ up to 70 percent of their workers as short-term contract workers, paid on a daily basis.
But few contract workers are ever permanently hired. The Serang case study found little evidence of contract
workers being hired on a permanent basis after their two- to three-year contract periods had expired. Firms often recontract the workers for another short-term period, without heeding the required maximum contract period of two to three
years or the 30-day grace period required between completion of one
short-term contract and the start of the next. The discrepancy between
the law and actual practice reflects the labor-market surplus, where many
job-seekers are willing to accept temporary employment without
demanding legal minimum wages and workers benefits.
Making labor policy more pro-poor

A thorough review and revision of Indonesia s labor laws at
both the national and local level is vital. There is ample evidence
that the jobless recovery since the financial crisis is largely a product of
a worsening investment climate for formal-sector firms. Recent labor
87
118
These include a one-off activity or temporary job, a specified job to be completed in a maximum of three years, or a job related to the introduction of new products on trial.
regulations are a significant contributing factor to this. Existing labor regulations are inherently anti-poor , largely because
they are forcing an increasing share of the labor force into the informal, unprotected sector.
A new social contract is needed with respect to minimum wages, severance pay and methods for settling
industrial labor disputes. Indonesia needs to substantially reduce the costs associated with employment, particularly
of younger and female workers who face the greatest barriers to obtaining formal sector jobs. At the same time, the
government must protect and enforce basic labor rights and conditions appropriate to the country s stage of development.
Moreover, these efforts may need to be accompanied by endeavors to socialize the main ideas through
media and advocacy campaigns, if they are to be accepted by key political actors, at the national, provincial and
district levels. Social support policies for the working poor (such as the cash compensation program for those most
affected by the rise in fuel prices) and public works schemes for the poor can be an important complement in building
support for reforms of the regulatory environment for labor.
Connecting the poor to financial services

Good financial systems play a key role in economic development by acting as an efficient intermediary
between savers and borrowers in ways that result in acceptable returns on investment and effective management of
risk. At the microeconomic level, the past three decades have seen a growing appreciation of the potential role of financial
services in reducing poverty, both by supporting income growth among the poor and by allowing them to reduce their
vulnerability to economic shocks.
Indonesia is a world leader in microfinance. Although the Unit System of Bank Rakyat Indonesia (BRI) represents
by far the largest single microfinance institution in Indonesia, it is only one of many financial institutions that have
managed to find ways to serve low-income households and micro-scale enterprises while earning sufficient profit to
survive and expand. Drawing upon this experience, it is possible to characterize three main linkages between financial
services and poverty reduction:
(1) Microfinance for the poor
(2) Savings and other financial services for the poor
(3) Credit for employment-generating enterprises
This section briefly considers each of these in turn, presenting a description of the principal linkages
and some key constraints. The following section suggests how these constraints might be relieved. It is worth noting
that many of the recommendations in this section are also relevant to Chapter on Social Protection in reducing the
vulnerability of the poor and near-poor.
1. Microfinance for the poor

The success of microfinance is a result of growth, as well as a cause. When considering the past and
current success of microfinance in Indonesia, it is important to maintain a degree of perspective. The strong growth
and success of largely rural but mainly non-farm microfinance in the 1980s up to the mid-1990s was preceded and
accompanied by macroeconomic stability, growing agricultural productivity and heavy investment in rural infrastructure,
particularly roads. The combination of growth in agriculture and increased opportunities for trade and services
created by new roads produced an excellent environment for micro- and small-enterprise growth, particularly in
petty trade.
119
But microfinance has its limitations, especially if growth is absent. This experience is strongly indicative
of the role that microenterprises and microfinance can reasonably be expected to play in economic development.
During times of economic growth, microenterprise development can be considered the embodiment of the multipliereffect, taking advantage of the growth opportunities provided by infrastructure improvements or widely disseminated
advances in technology, but rarely acting as an independent engine of growth. Indeed, this is true almost by definition,
since the very few microenterprises capable of creating sustainable growth quickly become small- or medium-scale
enterprises.
However, there is still great potential value in connecting households to the financial system for
the first time. Providing household enterprises with access to credit allows the family to make much more extensive
and efficient use of available family labor and increases the family s ability to take advantage of household investment
opportunities, whether within or outside the enterprise. Moreover, the household increases its ability to smooth
consumption over time, mainly through savings and enterprise diversification.
Reaching even the currently qualifying poor would imply a substantial increase in the borrower
base of existing microfinance institutions. For example, the 2002 Microfinance Access and Services Survey
(MASS) (World Bank, 2006h) concluded that while about two-thirds of households would qualify for credit from a
commercial microfinance institution (an institution with requirements similar to those of Bank Rakyat Indonesia-BRI
Units), a large proportion of non-borrowing households were unaware that they met the requirements for a microscale loan. Just over one-fifth (21.6 percent) of all households were actually borrowing. Moreover, 41 percent of
qualifying non-borrowers (19 percent of total households) indicated that they would like to borrow, while 53 percent
of non-qualifying households (17 percent of total households) said they would like to do so.
This suggests that the potential exists to almost double microfinance lending if ways can be found to
address the constraints faced by households and firms that would qualify for loans and wish to borrow, but do not
currently do so. Many, but not all, of the qualified non-borrowers appeared to lack information on whether and from
where they could successfully obtain a loan.
Furthermore, there are many near-qualifiers who are able to demonstrate their creditworthiness
with some relatively modest initial interventions. Thus, the target market for a special credit initiative can be
broader than just the currently qualifying poor. Based on the experience of the P4K credit program88 and other
programs to date, efforts to build the capacity of potential lenders are almost always better carried out through welldesigned credit programs or NGO-affiliated microfinance institutions than by more commercial microfinance
institutions.
However, few Indonesian institutions wish to actively target the enterprising poor when the number of
potential non-poor (although usually still low-income) borrowers continues to represent a much larger and more
immediately viable market. Therefore, achieving a rapid increase in the number of poor households borrowing from
commercial microfinance institutions requires a special initiative.
2. Financial services for the poor: reducing distance, revamping products

Secure savings mechanisms are invaluable to the poor. For those poor households able to use it productively,
credit continues to dominate other financial services in its potential to increase income and household welfare.
However, a large proportion of poor households are simply in no position at present to use credit effectively. Trying
88
120
P4K (Pembinaan Peningkatan Pendapatan Petani-Nelayan) or Assistance in Income Generation for Marginal Farmers and Fishermen.
to lend to these households runs the risk of either deepening

their impoverishment through debt collection or bankrupting
and discrediting the microfinance institutions through failure
to collect. Savings services carry far less risk of misapplication
(as long as the savings institution is liquid, savers can always
take their money back) and the use of financial savings services
is also more widespread than credit.
The growth impact of savings services alone is far
shallower than that of well-applied credit. Savings
alone, even over a prolonged period, are unlikely to move
families out of poverty. Indeed, financial savings, especially
at the accumulation levels of the poor, are usually not a good
investment returns to small savings lag considerably behind
inflation in Indonesia. Bank savings are, however, safe, liquid and confidential, while maintaining the slight barriers
to convenient access that seem to help many people remain committed to their savings strategies.
But saving can constitute the first stage in asset accumulation or preparation for major expenditure.
Financial savings can also reduce vulnerability to economic shocks (helping to prevent households from dropping
into poverty), and savings accounts can facilitate careful cash management.
The anti-poverty benefits of savings are largely for the next generation, particularly in helping children
continue their schooling a major motivation for household saving. More intensive encouragement of a savingsoriented lifestyle, such as that promoted by many Indonesian credit unions, may of course yield much greater benefits
over time, but is also relatively difficult to propagate outside a strongly supportive social environment.
Payment services also have an important role in facilitating transfers. Credit and savings are still the
two basic financial services, but online payment services (here the term is used to encompass deposit/withdrawal
transactions at multiple locations, electronic transfers, overseas and domestic remittances, ATM access, debit/PoS
cards, and similar services) have become increasingly important even to relatively low-income households in Indonesia
over the past decade. In terms of poverty impact, online access facilitates domestic and overseas remittances, as well
as support for family members, including students or aging relatives living far away from the principal family incomeearners. The introduction of online services to previously unserved rural areas can greatly increase local commerce
in cases where there is potential for substantial trade in agricultural or other commodities. This effect, however, is
likely to be mainly of indirect benefit to poor villagers.
To facilitate electronic payments, by far the single most useful step would be to encourage BRI to
move faster in putting its Unit System and village service posts (pos pelayanan desa) online. Indeed, BRI
is already moving is this direction, committing itself to putting 1,000 units online in 2006 a much higher target than
the 200-300 per year achieved previously. As BRI s majority shareholder, the government could signal its support for
higher levels of investment and technical support by BRI in this area, if it has not already done so.
Complementing this institution-specific approach would be an incentive program to any bank setting
up online service locations in rural or remote areas, with care to ensure that program design does not inadvertently
push a particular technical solution. Peoples Credit Banks (BPRs) would also benefit greatly from improved access to
online services, the current lack of which seriously harms their savings competitiveness.
121
3. Can enterprise credit generate more jobs for the poor?

The indirect impact of credit on generating employment growth may be as important as the direct
impact of microfinance and financial services. Not everyone is an entrepreneur. Just as it is fundamentally
incorrect to design poverty alleviation policy as if the poor are helpless economic actors unable to manage their own
affairs, it is equally wrong to assume that the poor are uniformly entrepreneurial. The real ladder out of poverty for
many poor households is not subsistence-level self-employment but the ability to obtain a steady job for oneself or
one s children.
Would relieving credit constraints generate employment growth among the poor? Given that SMEs
often mention credit (along with marketing or market constraints) as one of their top two business constraints,89 is
there an important financial linkage to employment growth? In other words, could the removal of barriers to credit for
growth-oriented enterprises help encourage employment for the poor? The answer to this question depends on the
answers to four further questions:
Are a substantial number of enterprises credit-constrained?

What are the constraints that they face?
Would relieving the credit constraints for these enterprises increase employment?
Would such employment draw in the poor?
A substantial number of enterprises are credit-constrained. Systemic barriers to accessing lending still
remain for enterprises of all sizes. For large enterprises and conglomerates, the main systemic hindrances to accessing
credit tend to be structural, such as risks related to failures in the legal system. But for small and medium enterprises,
much of the systemic problem is caused by more tractable policy issues. The Rural Investment Climate Survey (RICS)
adds important data and insights on the credit problems faced by micro and small enterprises. Half the surveyed
enterprises said they faced financial obstacles in maintaining operations or expanding (see Table 4.7). Almost a quarter
said that this was their main problem.
Table 4.7
Financial obstacles facing rural non-farm enterprises
Potential finance-related obstacles (N=2,366)
Is a problem
(% firms)
Is the main problem

(% firms)
Possibility to borrow from family, friends or others

Possibility to borrow from formal financial institution
33
46
3
11
Interest rate
Complicated bank loan procedures
47
45
5
2
Fear of not being able to pay installments
45
All financial obstacles
52
23
Source: Calculated from RICS, 2006.
There are many other reasons for enterprises not seeking loans. There are a wide range of constraints that
discourage firms from applying to formal financial institutions for loans. These include lack of collateral, fear of not being
able to repay, a desire not to have debt, complicated procedures and high interest rates.
89
122
See the Survei Terintegrasi (Susi) from BPS 2001-2003 as well as the RIC Survey (www.worldbank.org/id/rica).
Table 4.8
Reasons for not applying to a formal financial institution (firms that need additional funds)
Micro
Small
Medium /
Large
All
Enterprises
856
70
11
961
44
57
36
39
Complicated procedures
Prefer to save
46
45
43
53
27
73
50
0
45
59
Don t want debt 50

Insufficient collateral / Don t meet requirements
62
59
50
50
Prefer to borrow from family or friend

Don t know how to apply
36
32
Don t know where to apply

No FFI nearby
Had previous loan problems %
N - Number of firms in the survey

% of firms which listed the category as a reason for not
applying for a loan (more than one reason allowed)
Afraid of not being able to repay
Interest rates too high
59
48
27
45
45
36
14
27
0
36
30
26
21
11
17
0
0
24
20
18
Source: Calculated from RICS, 2006.
Relieving the credit constraints on these enterprises would probably increase employment. The enterprises
in the RICS indicated that their incomes might grow by 40 to 50 percent if these financial constraints could be removed.
Analysis of loan size in comparison with their working capital suggests that the expansion could be even larger. Given the
labor intensive nature of most micro and small businesses, it is highly likely that such a substantial expansion of activity
would be accompanied by employment growth.
But it is less clear if the poor would be the primary beneficiaries. Unfortunately, we do not know whether the
poor would be among the first beneficiaries of an increase in micro and SME employment, although most workers would
certainly be drawn from low-income households (i.e. the near-poor). Given slow job growth in the economy, poor, lesseducated workers may be at a disadvantage during the initial stages of job growth relative to their slightly better off but still
under-employed peers. Even if SME employment has started to grow, the poorest are still likely to be at the end of the
queue for trade and manufacturing jobs, where median worker education tends to be slightly higher (junior secondary
school level) than in agriculture, construction, or mining (primary school level).
Approaches to improving access to financial services

Three approaches to improving access to financial services might support poverty reduction and pro-poor growth.
1. Support for one or more outreach initiatives aimed at connecting new micro-borrowers to the financial
system
One approach would be to develop a P4K-style outreach program for financing the household
enterprises of rural low-income households. Similar to the Kredit Mini/Midi program that marked the true
beginning of BRI Unit microfinance in the late 1970s, the P4K program is one of the relatively few credit programs
that has worked well. That this program succeeded as well as it did is largely because so many aspects of the
program were already institutional from the beginning. Originally funded by IFAD/ADB and implemented jointly by
the Ministry of Agriculture and BRI, agricultural extension workers carried out group formation and maintenance
while BRI loaned to groups once they were ready. BRI also handled the financial administration. At its peak, the
program was successfully lending to more than 200,000 marginal farming and fishing households.
123
There remains a need to subsidize some of the cost of group formation, supervision and
institutionalization. Although the program had a very good repayment record (higher than 95 percent), especially
compared with other government/donor credit programs, there is a need for an indirect subsidy to cover some of the
cost of group formation; local, regional, and national supervision; and institutionalization, including some technical
assistance. In general, loan risk proved to be concentrated in the first loan; if the borrower repaid their first loan on
time, the risk on subsequent loans dropped to commercially acceptable levels. A successor to this program could be
spread to all rural and semi-urban areas throughout Indonesia, not just in the present provinces. The P4K approach
has the great advantage of scaling down in rural areas better than virtually any other program tried in Indonesia; other
efforts at reaching the same class of borrowers (including small savings and loan cooperatives) require a critical
mass of borrowers (50-100) at the local level in order to support the development of an institution.
To ensure success the early involvement of commercial microfinance institutions is needed. In
order to be a long-term success, such an initiative would need to have as its objective the bringing of borrowers into
the financial system. In the past, this was largely accomplished by institutionalizing the program. A better approach
would be to provide incentives for commercial micro lenders to begin lending to borrowers before the program ends
(rather than attempt end-of-project handovers that often do not work). The key to this is early involvement of interested
commercial microfinance institutions.
2. Make commercial banks (and BPRs) more transparent

Better reporting by lenders would improve transparency. This initiative would require some technical
expertise and a willingness to push key financial institutions to be more transparent in their activities. BRI in particular
has established itself as perhaps the world s best micro-enterprise finance institution, currently lending to more than
3 million micro-borrowers nationwide. However, the way in which reporting is done by BRI, most BPDs90 and some
other banks that claim to focus on micro/SME lending, makes it difficult to determine how much lending has actually
gone to micro/small enterprises, and how much has been consumer lending to individuals.
Banks tend to emphasize total disbursement, not outstanding loans or number of borrowers. For
example, these banks are often reluctant to show micro-enterprise lending separate from micro-scale lending to civil
servants and employees, and they tend to emphasize cumulative disbursement rather than outstanding loan amounts
or the number of borrowers. The point here is not to embarrass banks or to add an additional reporting burden.
Rather it is to require them to provide a more accurate picture of their current microfinance activities. In practice, the
additional information required from banks is fairly modest:
Greater emphasis on reporting the number of accounts/people to whom micro and SME loans have been made,
not just rupiah values;
A more honest classification of loans, distinguishing between enterprise and consumer lending; and
Particularly for state-owned banks, the requirement that the profit and loss of the banks be disaggregated
according to business line.
3. Ensure that all banks have basic no-fee savings products available to the poor (and everyone else too)
Even modest bank fees can erode savings held by the poor. Since the poor are unlikely to earn much
interest income, even a fairly modest monthly or per-transaction fee on deposits and withdrawals can wipe out
interest earnings and even cause their savings balances to decline in nominal terms. Although desirable, a positive
real rate of interest is not a requirement of a good pro-poor savings instrument, but a fee and interest structure that
preserves poor savers nominal balances most certainly is.
90
124
Provincial government commercial banks.
Since the burden of no-fee banking will fall on the banks, an award/reward scheme could act as an
incentive. In implementing this policy (whether through regulation or a voluntary effort coordinated by the central
bank), banks should of course be free to modify features (such as ATM and online services) normally included in a
flagship savings product. There is no need to revert to a Tabanas-like national savings product. The key is to provide
a no-fee option for basic savings services, minimally including unlimited free deposits and withdrawals at one
location. The burden of this policy will fall disproportionately upon the banks with the best networks and fewest
informal barriers to doing business with the poor, and it may be worthwhile to consider recognizing this with a small
award/reward scheme.
VII Conclusion: Making Growth more Pro-Poor

Economic growth is essential for poverty reduction, but making growth pro-poor is also important.
Supported by a growing body of academic research, including the World Bank s seminal study on pro-poor growth (World
Bank, 2005c), it is now widely accepted that economic growth is the primary driver of reducing poverty. This is just as true
in Indonesia as elsewhere and, indeed, Indonesia has been an example of the successful channeling of growth for poverty
reduction goals over a period spanning many years (Timmer, 2004). Making growth work for the poor therefore requires
both making growth and ensuring that this growth is shared by the poor.
125
MAKING THETHE
MAKING
NEWNEW
INDONESIA
INDONESIA
WORK FOR
WORK
THE POOR
FOR THE POOR
Indonesia s successful pro-poor model was dislocated by the financial crisis. Unfortunately for Indonesia s
poor, when the Asian financial crisis struck in 1997, not only was the country s previously robust growth shattered, but in
the aftermath of the crisis Indonesia failed to return to the pro-poor pattern of growth that it had previously enjoyed. Not
only was growth lackluster compared with previous years, but its character was no longer pro-poor. Even today, as growth
finally picks up thanks to returning macroeconomic stability, the poor are not benefiting from the recovery to the same
extent as the non-poor. This worrying state of affairs threatens to undermine Indonesia s own medium-term poverty
reduction objectives and potential even its ability to meet the MDGs by 2015.
Study of Indonesia s pro-poor growth gives clues as to how to make renewed growth once again more
pro-poor. Looking closely at the ways in which the poor moved out of poverty in the past, it is possible to understand the
main pathways that enabled the poor to escape poverty. There are two key productivity pathways followed by individuals
and households: improvements in agricultural productivity in rural areas, and increases in non-agricultural productivity
in both urban and rapidly urbanizing rural areas. The first of these boosts the earnings of the rural poor, while the second
helps to create better jobs (and higher earnings) for those poor who have already moved out of agriculture. In addition,
there are two transitional phases that help the poor out of poverty: there is a sectoral shift from farm to rural non-farm
employment, and there is a locational shift out of rural areas to urban employment, through both seasonal and permanent
migration. In Indonesia s post-crisis renewed-growth environment, the government needs to apply policy measures that
support these pathways and transitions out of poverty in order to make growth pro-poor once again.
Macroeconomic stability is fundamental to growth and therefore to poverty reduction. As Indonesias
experience over the past decade attests, there can be little real growth, and therefore little real progress in poverty reduction,
without a stable macroeconomic environment. Indonesia has struggled hard to recover from the 1997-98 crisis. And in
the past couple of years the government has succeeded in restoring confidence in the economy and bringing about a
return to macroeconomic stability. Despite ongoing risks and the impacts of foreign shocks for example the surge in oil
prices in 2005 the government has managed to maintain this hard won stability. This bodes well for higher levels of
growth going forward.
Macroeconomic stability will also help to make growth pro-poor in several ways. One is restoring low
inflation. This is all the more important after the turbulence of 2005, as rising prices hit the poor hardest. Also important
is keeping the fiscal balance in check, maintaining confidence in the currency and the government s fiscal management.
The government has done well here too, addressing the politically difficult issue of burgeoning fuel subsidies before this
undermined the currency. Finally, maintaining an open trade regime is important and Indonesia has made steady progress
towards this end, although the continuing restrictions on rice imports have the potential to hurt the poor (and throw many
of the near-poor back into poverty) if scarcity of rice is again allowed to drive prices upwards.
To enable the poor to benefit from renewed growth, the government needs to invest in the capabilities
of the poor. Just as important as macroeconomic policies, microeconomic measures are key to building the capabilities
of the poor if they are to be able to move along the pathways out of poverty. Action in two areas would support pro-poor
growth. First, higher agricultural productivity is an important pathway out of poverty and measures to build capabilities
here are crucial. These include revitalizing agriculture through improved irrigation quality and management, promoting
diversification into higher-value crops and livestock, boosting agricultural research, redesigning agricultural extension
systems, developing marketing and IT for agriculture and rural SMEs, and improving property rights and the land market.
Second, providing targeted education and vocational training to build the capabilities of the poor is key. Here there is
126
scope for helping the poor gain better access to vocational schools (SMKs) and vocational training centers (BLKs and
KLKs), for example through targeted voucher schemes. The BLKs and KLKs should also be used as a means of improving
pro-poor targeting of training programs, while companies should be provided with incentives to train apprentices from
poor households.
The government also needs to ensure that the poor are connected to the opportunities generated by
renewed growth. Even with their capabilities enhanced, the poor still need to be connected to opportunities if they are
to benefit from growth. First, the government needs to invest in rural infrastructure to ensure that the rural poor are
physically connected to urban markets. In particular, rural roads have deteriorated markedly since the crisis and additional
resources now need to be channeled into the repair and routine maintenance of rural roads, supported by a new districtlevel road management system. The increased use of performance-based contracts for road maintenance work would also
be beneficial in ensuring quality standards of maintenance work. Second, for the poor to escape poverty they must be
connected to employment opportunities. Rigidities in the labor market are causing jobs to move from the formal to the
informal sector, hurting the poor. As the government addresses the issue of labor policy and considers revisions to the
current labor regime, its emphasis should be on making Indonesia s labor policy more pro-poor. This could be done by
constructing a new social contract regarding minimum wages, severance pay and methods for settling industrial labor
disputes. Third, the poor need to be better connected to financial services. This can be achieved through expanding
microfinance programs, making commercial banks more transparent and ensuring that all banks have no-fee savings and
other pro-poor products.
127
Spotlight on Gender
Remittances from female migrant workers: A lifeline to communities
More than 80 percent of the 400,000 registered overseas workers
that remittances return to those selected regions. Several districts in Indonesia
in 2004 were women. These women are known as TKW in Indonesia, or
receive large volumes of remittances, for example: (i) in Sukabumi, West Java,
tenaga kerja wanita. Given the high number of informal or undocumented
remittances (compared with regional own source revenues in brackets) were
workers, the total could be as much as twice as high as those officially registered.
Rp 480 billion (Rp 23.9 billion) in 2001, Rp 285 billion (Rp 31.2 billion) in
Women as a percentage of the total number of migrants have been increasing
2002, and Rp 237 billion (Rp 34.7 billion) in 2003;92 and (ii) in West Nusa
since the government began promoting labor exports in the 1980s. Their general
Tenggara remittances average Rp 111 million per month and constitute 3.8
low level of educational attainment means that the majority work in unskilled
percent of the region s gross domestic product (Ananta and Arifin, 2004).
occupations usually in the informal sector as domestic workers, but also in

factories and the entertainment industry. The vast majority of female migrant
It is difficult to measure the value of remittances reliably and data
workers are from rural areas in selected regions and provinces. The main
are not sex disaggregated. Official balance of payments or central bank
destination countries are Saudi Arabia, Malaysia, Kuwait, Singapore, Hong Kong
data on remittances are compiled using data on wire transfer flows officially
and Taiwan.
reported by financial-sector institutions. Such data significantly underestimate
91
actual remittance flows, in particular because they do not include funds

Limited job opportunities and low salaries in rural areas, together
repatriated informally by both formal and informal migrants through mail
with a need to repay debt and/or a desire to emulate other successful
services as well as intermediaries. They also fail to capture non-monetary goods
returnee migrants, provide the stimulus for women to seek work
that are brought back.
overseas. Recruiters or middlemen (calo), who actively recruit from certain

villages, play an important role in influencing women to work overseas, providing
Costs of sending remittances are often unnecessarily high given
information and funds that can be borrowed by migrants to cover the various
the relatively small transfers made by female migrants. If sent through
costs of migration.
banks or transfer service provides, fees vary by country. Often family members
do not have ready access to banking services and so money is sent through
Placement costs are stipulated by the Ministry of Manpower and
brokers or agents accounts. Brokers tend to charge relatively high transaction
Transmigration, but they only represent part of the total cost, as the
fees, sometimes taking advantage of the financial illiteracy of migrants and/or
actual costs are often far higher. Placement costs charged by brokers or
their families.
agents vary from one month up to one year of the salary of migrant workers.
Often the higher the wages migrant workers will potentially receive the higher
Salary level is the key factor influencing the amount of remittances
the placement cost. The placement costs may be paid in cash upfront or in the
sent. There is a significant variation in domestic workers wages paid to female
form of a loan that is deducted from migrants wages. In addition to placement
migrant workers and therefore in how much is remitted. In a recent World Bank
costs, formal migrants also have other expenses charged to them, such as
study, the average salary was US$70/month for domestic workers in Malaysia,
accommodation in the training center before departure, living expenses in the
compared with a much higher salary of US$390/month in Hong Kong and
host country and the cost of returning home.
Taiwan. How much is remitted, as well as how much is received, is also

influenced by the various fees that have to be paid as part of the recruitment
Total official remittances from Indonesian migrant workers were
process, the level of a migrants education, ease of money transfers, the number
US$1.35 billion in 2004 and estimated to be US$2.5 billion in 2005.
of dependents in the household and the family relationship.
Given that migrant workers come from selected regions, it is not surprisingly
91
There are no comprehensive data on places of origin, although West Java, Central Java, Yogyakarta, East Java, East Nusa Tenggara, West Nusa Tenggara, South Sulawesi and Lampung have been documented
as major sources of female migrant workers.
92
Female Migrant Workers Research Team, 2006. Impacts of Migration and Usage of Remittances: seeking Ways towards Improving Female Migrant Workers Welfare, World Bank, Draft Final Report.
128
Research from other countries suggests that women migrants are
Reduce the costs of transferring remittances. An international study
more reliable remitters than men. Some researchers have posited that
of remittances suggests that greater competition among remittance service-
this is because female migrants are more closely attached to their families
providers could reduce remittance costs (World Bank, 2006). While this is
and/or there is more pressure upon them to fulfill familial obligations.
largely up to the remittance-source countries, Indonesia could facilitate

the establishment of payment networks that are shared by banks and
Studies show that the majority of remittances received are used to
microfinance institutions. Another alternative could be to facilitate
pay off debts and for consumptive needs. The World Bank study found
establishment of cooperatives of female migrants to pool and remit savings.
that the prominence of unproductive expenditures stems in part from the lack
of investment opportunities at the village level. Remittances spurred a housing
Facilitate remittance inflows through formal means using
construction boom in the four districts that were part of the World Bank study.
microfinance mechanisms that migrant women and their families are
Remittances were also occasionally used to buy land or cattle. A study of
familiar with. Small cooperatives and rural credit unions are springing up
returned migrants in West Java also found only a small proportion spent their
to receive transfers, hold savings and make small loans to develop small
money on productive uses, such as buying agricultural land or starting
businesses. The legal framework should be adjusted to permit these
businesses.
institutions to receive remittances.
There are no reliable data in Indonesia comparing how male and
Provide supports to facilitate the use of remittances towards
female migrants spend their remittances. However, a worldwide UNFPA
investment. This could include the creation of social funds, savings and
study documented that male and female immigrants spend remittance income
credit plans to finance small business start-ups, as well as promotion of
differently. Men tend to purchase conspicuous consumer goods, such as cars
financial literacy. Providing advisory services to households with migrant
and televisions, while women are far more likely to invest remittance income
workers to start up microenterprises could also be explored. Banks should
in food, schooling and healthcare services. The report tentatively concluded
be encouraged to provide financial services, such as remittance
that female spending patterns may raise families out of poverty in the long
management, savings and investments. The Unit System of Bank Rakyat
run.
Indonesia (BRI), for example, lends to migrant workers families using the
future remittances as collateral. They also offer credit to migrant workers
Ways forward in maximizing the benefits from female migrant
at the time of departure, usually at more favorable rates than placement
workers
agencies.
A comprehensive integrated migrant policy and management system would

include measures to facilitate transmittal of remittances to boost both their
contribution to economic growth, as well as provide benefits to women and
their families, such as:
Review and improve the quality of data and methodologies used

to estimate remittances. Basic sex-disaggregated data on many
aspects of remittances need to be collected and made available on a regular
basis. Currently, many of the remittances that enter Indonesia are not
recorded and, where they are, the data do not indicate the sex of the sender.
This would allow more in-depth analysis to determine, for example, whether
there are differences between men and women in terms of the share of
remittances that is remitted and how the money is spent. A better
understanding of the gendered patterns of migration would contribute
towards the development of more appropriate policies.
129
Chapter
Making Public Spending

Work for the Poor
130
Chapter 5: Making Public Spending Work for the Poor
Introduction
The next decade poses some very specific challenges, but also a tremendous window of opportunity for
Indonesia to use its public resources to make a dramatic leap in reducing poverty in all its dimensions.
Indonesia recently surpassed an international threshold, moving out of the range of low-income countries in 2003. Even
with current growth rates which well exceed population growth Indonesia is expected to consolidate its place as a
middle-income country by the end of this decade. This, coupled with a continued stable macroeconomy and a sound
fiscal position, should continue to provide the government with the fiscal resources it needs to achieve its development
objectives and to address the main poverty issues that still plague the country (see Chapter 3 on Understanding Poverty).
The fiscal scope for making progress on the poverty reduction agenda has also been augmented by the significant
increase in state revenues from oil and gas exports, as well as the concomitant steps taken to reduce the significant
budget share absorbed by regressive fuel subsidies. The public resources are now available to address poverty in Indonesia
in the remainder of this decade as the country moves to take its place among other middle-income countries.
The government has expressed a strong commitment towards achieving the Millennium Development
Goals (MDGs) by 2015, and to making concrete progress against those and other Indonesia specific
objectives during the current administration. The government has set itself ambitious targets to address poverty
in its multiple dimensions. Indeed, the current administration has stated its intention to halve income poverty over the
five-year period up to the end of this decade. Critical to this effort will be putting Indonesia s growing public financial
resources to the most effective use in reducing poverty and achieving Indonesia s development objectives.
Table 5.1
MDGs and national medium-term development targets (RPJMs) for Indonesia
Indicator/Target
Poverty Reduction
Population below US$1 a day (%)
Poverty head-count ratio (%)
Population below US$2 a day (%)
Health
2000
2002
2004
RPJM 2009
MDGs 2015
Actual
Actual
Actual
Target
Target
7.2
7.5
4.8
55.4
18.2
52.4
16.7
45.4
Under 5 mortality rate (per 1,000 live births)

Infant mortality rate (per 1,000 live births)
Maternal mortality ratio (per 100,000 live births)
Education
48.0
36.0
Net enrollment rate in primary education (%)
93.9
Gross enrollment rate at junior level (%)

Literacy rate of 15-24 years of age (%)
Water & Sanitation
Population with access to improved water (%)
Rural Development
Agricultural sector growth (%)
38.4
29.6
307
10.3
8.2
7.5
33
26
226
95.3
94.3
99.6
79.5
82.2
98.7
98.1
105
100
100
77.0
80
3.5
Source: Indonesia Progress Report on the MDGs 2001, World Development Indicators, Medium-Term Development Plan and Country Assistance Strategy, 2004.
131
The effective and efficient use of public funds to target sectors that are most beneficial to the poor is
crucial for poverty alleviation. How those funds are allocated across sectors and also within sectors in programs
better targeted towards the poor is of considerable importance to the government. There are three broad areas through
which government spending can help to reduce poverty: (i) through social services in the health and education sectors
that increase the human capital of the poor and hence improve their productivity; (ii) through infrastructure investments
that increase the income opportunities and market access of the poor; and (iii) through social transfers and safety nets that
help to supplement the income of the poor and near-poor in the short term (considered in more detail in Chapter 6 on
Social Protection). If Indonesia is to succeed in achieving its objectives, it needs to spend its resources more effectively
on service provision, particularly in these sectors. Indonesia needs to spend enough (level), on the right things (sector
allocation), in the right places (geography). This chapter takes stock of the poverty focus of Indonesia s spending today
and points to some concrete and strategic actions that would help to improve spending in ways that could accelerate
Indonesia s development and reduction of poverty by the end of this decade. The chapter starts with a discussion of
aggregate spending levels in Indonesia and fiscal space in the years going forward, and then considers sectoral spending
in the education, health, water and sanitation, and roads sectors.
Success in achieving Indonesia s development goals does not only depend on increased spending in
pro-poor sectors but also on the ability to translate these resources into improved services to the poor.
Weak performance incentives, a lack of accountability mechanisms and imperfect monitoring systems can all undermine
the delivery of services to the poor even when the availability of resources is not a constraint (World Bank, 2004d). While
this chapter looks at how spending can be better allocated to address some of Indonesia s most pressing poverty issues,
Chapter 7 on Government focuses on issues of accountability, governance and transparency, which are just as vital if
these services are to better serve the poor.
II
Aggregate Spending Levels and Fiscal Space
Aggregate spending
In the period since the oil boom in the 1970s, Indonesia has made exemplary progress in harnessing its
growth to reduce poverty, primarily through its public investment program.93 With the reduction in fuel
prices in 1986-88, Indonesia experienced an external revenue shock, which resulted in reductions in capital spending.
The levels of spending recovered towards the end of the 1980s and remained stable in real terms until the Asian financial
crisis in the late 1990s.
In the period leading up to the financial crisis and also following it, aggregate fiscal spending was
restrained, with implications for the public investment program. As a percentage of GDP, central expenditures
were already beginning to fall after 1994. After the financial crisis, there was a peak in percentage of GDP expenditures
mostly due to the reduction in the denominator with shrinking economic activity. Indonesia has successfully achieved
fiscal consolidation in the years following the crisis. The budget deficit was reduced to about 1.0 percent of GDP in 2004
and 0.5 percent of GDP in 2005, while the central government s debt outstanding contracted from about 100 percent in
1999 to 46.8 percent in 2005 (PER, World Bank, forthcoming). Such a conservative fiscal policy has contributed to
macroeconomic stability in this period and Indonesia s fiscal situation has significantly improved over the past five years.
On the flip side, fiscal consolidation after the crisis has squeezed out spending in the development sectors.
93
Through the Inpres program Indonesia increase its supply of basic healthcare facilities (Puskesmas) and also increased the availability of primary schools even in remote areas of the country. Investments in
provincial roads also improved the productive use of labor and helped connect the poor into the growth process by linking them to labor and product markets (see Chapter 2 on History and Annex V.1 on Inpres
grants).
132
Fuel subsidies played a major role in determining levels of central spending after the crisis and
fluctuations in world oil prices have had a direct impact on Indonesia s fiscal spending as a result of
price control policies. Although real consolidated spending increased in Indonesia following the crisis, the changes
in spending were largely driven by changes in the fuel subsidy which, in turn, were closely related to fluctuations in world
oil prices. The fuel subsidy reached Rp 69 trillion in 2004 and continued to surge to Rp 96 trillion in 2005, despite
reductions in the subsidy in that year.94 The increase in fuel subsidies was mainly attributed to the fixed fuel-price
mechanism introduced in January 2003 and subsequent international oil-price increases (see Figures 5.1 and 5.2).
Concurrently, domestic fuel consumption increased over these two years and, in order to meet this increasing demand,
the government was forced to import expensive fuel products. While in the 1970s and 1980s Indonesia was able to ride
the wave of increased world oil prices as a net oil exporter, after the crisis, with reduced production capacity and continuing
price controls, Indonesia began to subsidize domestic oil consumption as a net importer of refined oil products (PER,
2005). As a result, the large chunk of funding allocated to fuel subsidies took away resources from important development
sectors (see Figure 5.3). But while fuel subsidies were the major transfer program in the country and constituted a large
part of the government budget, they failed to respond to the needs of the poor and were highly regressive in their targeting.95
Figure 5.1 Consolidated spending as a percentage
of GDP over time
Source: Ministry of Finance (MoF) data, World Bank calculations.
Figure 5.3 Fuel subsidies and spending in

pro-poor development sectors
Source: MoF, SIKD.
94
95
Figure 5.2
Fuel subsidies ran out of control with increasing

oil prices and stagnating refinery capacity
Source: MoF, EIA 2006.

Note: International oil prices taken as refiner acquisition cost of crude oil, composite (US$/barrel).
Figure 5.3a Fiscal space for development programs and fuel

subsidy reductions
Source: MoF data, World Bank PER 2006 calculations.

Note: International oil price assumption for 2006 is taken as US$60 for this calculation.
The actual fuel subsidy in 2004 was Rp 72 trillion. However, the government only paid Rp 69 trillion in 2004 and, as such, the remaining Rp 3 trillion is arrears (PER files, World Bank, 2006).
See Chapter 6 on Social Protection for details on the regressivity of the fuel subsidy.
133
Another important development in the period after the crisis was the decentralization of fiscal spending
to the districts after 2001, allowing spending by the regions to constitute a growing share of total public
spending. Whereas in 2000 the central government share of spending was 87 percent and the regional governments
share 13 percent, by 2004 the share of the central government had declined to 69.3 percent, while the regional governments
share had grown to 30.7 percent (PER, World Bank, forthcoming). (See Figure 5.1.) Indonesia now operates within a
decentralized fiscal setting and policies going forward need to take into account incentives by local governments to
prioritize pro-poor development sectors in order to achieve national poverty reduction priorities. This topic will be further
discussed in the Chapter 7 on Government.
Aggregate development expenditure is close to the pre-crisis level mostly as a result of increasing
district-level spending. Aggregate development expenditure as a share of GDP started to recover, reaching 5.4 percent
in 2004, although this was still lower than the pre-crisis high of 7.5 percent in 1994. Although for a period of time the
transition to decentralization may have adversely affected the disbursement of development expenditure, the subsequent
recovery in development spending was largely driven by district governments. Development expenditure by district
governments increased from 1.1 percent of GDP in 1994 to 1.8 percent in 2004, while development expenditure at the
central government level decreased from 5.6 percent in 1994 to 2.7 percent in 2004.96
The current fiscal situation and fiscal prospects compared with the past decade are much more promising.
The government s decision to reduce fuel subsidies is a major contributor to this outcome. Initially raising fuel prices in
March 2005 by a weighted average of 29 percent, the government followed this increase with a far more dramatic increase
in October 2005, when prices were raised by a weighted average of an additional 114 percent.97 Current projections
foresee the government continuing to focus on fiscal consolidation, with deficits projected at no more than 1.0 percent of
GDP. The medium-term fiscal framework shows substantial scope for increasing spending levels in the medium term in
order to achieve development targets. Fuel-subsidy reductions in 2005 provided a gross saving of US$4.6 billion for the
budget (from the March and October price increases) and an additional US$1.0 billion from the July/August adjustment
in industrial prices (see Figure 5.3a). The government reallocated about US$1.7 billion of these savings into development
programs in education, health and rural infrastructure, as well as putting into place a cash transfer program designed to
mitigate the impact of the October fuel-price increases, leaving the rest of the savings for increased deficit financing. For
2006, the savings from the October price increase are estimated to be around US$10.1 billion. In the medium term, in
order for fiscal policy to play an active role in supporting economic growth, fiscal consolidation must be balanced with
rising public investment in key social and infrastructure
sectors. With this level of savings available from the
reduction in the fuel subsidy, the governments challenge
will be to spend the available funding wisely in poverty
reducing sectors that will also support future economic
growth.
Sectoral spending
The focus of the sections that follow is on
government spending on services that are
central to the needs of the poor. In particular, these
focus on public spending aimed at poverty reduction in
96
GDP figures using 2000 GDP index definition.

Prices changes in October 2005: the price of premium (subsidized) gasoline was increased from Rp 2,400 to Rp 4,500 per liter; diesel fuel was raised from Rp 2,100 to Rp 4,300 per liter and kerosene nearly
tripled in price, rising from Rp 700 to Rp 2,000 per liter.
97
134
key social service sectors, namely in the human development sectors of (i) education and (ii) health, and in the infrastructure
sectors of (iii) water and sanitation systems, and (iv) rural roads. (Spending on social protection and transfer programs is
discussed in greater depth in Chapter 6.) These sectors are prioritized under the current government medium-term plan
(RPJM) and already make up a significant portion of total spending.
As an extension of the discussion of health sector spending there is also a focus on two non-income
dimensions of poverty, namely (i) reducing maternal mortality, and (ii) reducing malnutrition. These
areas have been prioritized by the government in its national development targets and are also components of the MDGs
(see Table 5.1). The Focuses on Maternal Mortality and on Child Malnutrition at the end of the chapter provide diagnostics
on the constraints that limit Indonesia s achievement of better outcomes in these priority areas and suggest specific
vertical interventions for overcoming these constraints.
With the exception of infrastructure development spending, in the period between 1994 and 2004
spending in all these sectors increased, although there was a temporary dip in all sectors following the
crisis. Post crisis, education and health expenditures picked up (both in real terms and as a percentage of GDP) with
increases in decentralized spending by district governments and continued development and routine spending by the
central government (see Figures 5.1 and 5.3). Infrastructure development spending has been slower to recover than other
sectoral spending, but by 2004 it was close to returning to pre-crisis levels.
The reduction of fuel subsides in 2005 has further freed up resources for more pro-poor spending and
increased allocations in these sectors. The government reallocated a total of US$1.7 billion in 2005 away from the
regressive fuel subsidy to pro-poor sectors and programs with better-founded development objectives (the PKPS-BBM98
programs). The government has taken a politically difficult but economically rational and pro-poor step in allocating
resources more effectively in sectors that matter to future growth and prosperity. The boxes on each of these four programs,
in this chapter and in Chapter 6 on Social Protection, provide information on the design, budget and targeting of these
programs, as well as preliminary results from the recently undertaken qualitative evaluations of the programs.
98
99
The PKPS-BBM unconditional cash transfer (UCT) program was designed to serve as a social safety net,
protecting the poor against the short-term impact of the fuel-price increases. (For details on budget, targeting and
preliminary evaluation results see Box 6.3 in Chapter 6).
In the education sector, the compensation package included an operational aid program to primary and junior
secondary schools (BOS99) aimed at canceling school fees, as well as providing targeted scholarships for poor
senior secondary school students (see Box 5.1).
In healthcare, the fuel-subsidy savings were reallocated in a program that supports basic healthcare at the
Puskesmas level and also offers targeted health insurance coverage to the poor for in-patient care in third-class
hospital beds (see Box 6.7 in Chapter 6).
In rural infrastructure, the government financed direct grants to 12,800 villages with the objective of improving
village infrastructure and generating labor-intensive employment through a bottom-up decision-making process
(see Box 5.4).
PKPS-BBM: Program Kompensasi Pengurangan Subsidi Bahan Bakar Minyak (Compensation for Fuel Subsidy Reductions Program).
BOS Program: Bantuan Operasional Sekolah (Operational Aid to Schools).
135
III
Human Development Sectors
Overview
Inadequate education is a powerful determinant of poverty and unequal access to education is a strong
correlate of income inequality. Providing basic education for all has long been a poverty reduction strategy for
Indonesia. Since the 1970s, Indonesia has invested vastly in infrastructure for primary education through the Inpres
program (see Annex V.1 Inpres development grants), and has achieved high enrollment rates for most of its population at
the primary school level. Basic education or literacy training has been crucial in equipping the poor with the means to
contribute to, and benefit from, economic growth. Furthermore, improving the educational attainment of Indonesians is
critical to their obtaining better jobs and working their way out of poverty, while also improving Indonesias competitiveness
in the global economy (see Chapter 4 on Growth for linkages between education investments and growth). Indonesia s
human capital policies have played a critical role in creating and supporting rapid economic growth from the 1970s
through to the late 1990s. The education policies in Indonesia focused on primary and secondary schooling rather than
higher levels of schooling that would have benefited only a select few. Meanwhile, limited public funding for postsecondary education focused on technical skills and vocational training; higher education at university and college levels
was largely met by self-financed private systems.
In Indonesia, the return to education are increasingly higher for higher levels of schooling, so focusing
education investments only on primary education is no longer sufficient. The benefits of openness, technology
and market competition for economic growth depend greatly on education and skill levels, and increasingly higher levels
of education become more important for retaining the benefits of growth. Alatas and Bourguignon (2005) find that in
Indonesia, the earning function for male wage workers is convex , which implies that an additional year of schooling
means an increasingly higher relative gain of income for those who have high initial levels of schooling.100 The increasing
return to each additional year of schooling in the earnings function can lead to increased inequalities, as the rich have
more access to higher levels of education. Poverty diagnostics also show that the gap in the return to educational assets
has also increased in Indonesia and educational assets have proved to be a means for escaping poverty after the crisis for
many of the transitory poor (see Chapter 3 on Understanding Poverty). The implication of these findings is that the
returns to human capital are rising and the gap between the educated and non-educated is widening, with an increasing
return for each additional year of schooling, especially in urban areas. While in previous decades providing primary
education to the poor was sufficient, the government has to increasingly think of ways to integrate the poor into these
labor markets and to improve their ability to tap into the benefits of growth.
Indonesia has set itself a medium-term development plan target of reaching 100 percent enrollment at
the primary school level and 98 percent enrollment at the junior secondary school level by 2009.
National Education System Law No. 20/2003 states that every citizen aged 7 to 15 must attend basic education, implying
that the government needs to provide free educational services to all pupils at these levels of schooling. Achieving such
ambitious enrollment targets, coupled with investments in improving the quality of education, will be central to Indonesia s
poverty reduction strategy, as well as its competitiveness in the region and its rate of growth in the years to come (see
Table 5.1 MDGs and national medium-term development targets (RPJMs) for Indonesia).
Indonesia s past enrollment expansion closed the enrollment gap across income groups at the primary
education level, but striking inequalities remain at the junior secondary and senior secondary levels.
In 2004, primary school enrollment rates were 107 percent gross and 93 percent net in Indonesia.101 Problems with
100
In 2002, the returns to urban (rural) male wages for an additional year of schooling after 1 year of schooling was 8.3 percent (6.0 percent), after five years of schooling was 10.0 percent (7.6 percent) and
after 8 years of schooling it was 11.1 percent (8.8 percent). See Heckman Selection Model Wage Earnings function in Chapter 3 on Understanding Poverty.
101
Gross enrollment rate in education: A gross enrollment ratio is the total enrollment at that level, regardless of age, as a percentage of the official school age population for that level. The ideal is 100 percent,
but ratios greater than 100 percent can occur when there are a high number of overage students in that level. A high (greater than 100) gross enrollment ratio can be indicative of inefficiencies in the educational
system. Net enrollment ratio gives the number of students that are of the required age group and are enrolled in school divided by the total number of students in that age group.
136
access become more significant at the junior

secondary school level. Using figures from 2004,
there is a significant break in enrollment rates at
the junior secondary level, where the gross
enrollment rate is 82 percent but the net enrollment
rate declines to 65 percent. Officially, basic
education (through grades from 1 to 9) is
compulsory for children aged 7 to 15, although
this is not strictly enforced. While access to
primary schooling may still be a problem in remote
areas, today for most of the poor in Indonesia the
most pressing issue in terms of access to
education concerns the transition to junior
secondary schooling.
Equally important, health is one of the key
goals of development and lack of access
to healthcare is in itself a key dimension
of poverty. Out of seven Millennium
Development Goals (MDGs), four relate to
health102 including reductions in maternal, infant
and under-five mortality, disease control for malaria and HIV/AIDS, as well as the elimination of extreme poverty which,
broadly defined, is also related to access to affordable healthcare by all households. Indonesia itself has set ambitious
targets for the remainder of this decade: the government s medium-term development plan (RPJM) outlines its health
sector goals as being: (i) the increase in life expectancy from 66.2 years to 70.6 years; (ii) the reduction in infant mortality
rate from 35 to 26 per 1,000 live births; (iii) the reduction in maternal mortality rate from 307 to 226 per 100,000 births;
and (iv) the reduction in the prevalence of infant nutritional deficiency from 25.8 percent to 20 percent (see Table 5.1
MDGs and national medium-term development targets (RPJMs) for Indonesia).
Indonesia has made significant efforts in recent decades to expand basic healthcare on an equitable
basis. The 1970s and 1980s witnessed a significant expansion of public facilities, including hospitals, community
health centers (Puskesmas), 103 supporting health posts (Puskesmas Pembantu, known as Pustu) and regional
pharmaceutical depots. The focus on expanding healthcare facilities has helped to improve access to healthcare even in
remote areas (Knowles and Marzolf, 2003). In parallel with this expansion, partially subsidized health services were
provided by public health facilities with the aim of increasing the affordability of healthcare, especially for the poor.
Modern public health services have since become more accessible to most of the population, with health outcomes
showing marked improvement during the 1980s and 1990s, until 1997 when the economic crisis first began (Lieberman
and Marzoeki, 2000). The infant mortality rate (IMR) fell from over 80 deaths per 1,000 births in the late 1970s to under
50 in the mid-1990s. Usage of contraception in this period rose by over 60 percent, while the total fertility rate fell from
4.7 births to 2.8 (Lieberman and Marzoeki, 2000). Interventions through family planning and fertility reduction programs
have reduced the risk of maternal death over a lifetime, but the risk of death facing a pregnant woman remains high
(Knowles and Marzolf, 2003).
102
103
See Indonesia Progress Report on the Millennium Development Goals (2004).

Publicly funded community health centers are called Pusat Kesehatan Masyarakat, or Puskesmas for short.
137
Notwithstanding the progress made in expanding the public healthcare system, access and quality of
healthcare remain low and Indonesians, including the poor, rely heavily on private sector provision.
Despite the expansion of healthcare facilities in Indonesia, the utilization of public health facilities has remained low. In
the 1970s and early 1980s, although the community health situation improved slightly and the number of health centers
rose dramatically, increasing the percentage of Indonesians receiving medical help from trained staff, the reliance on
public provision of community services remained low. One reason for this was the persistently low government spending
on healthcare (Knowles and Marzolf, 2003). Although public services were widely available, even among the poor, these
highly subsidized public health centers were the providers of choice only half the time, especially in the 1990s when a
separate private health sector first started to emerge in Indonesia. For instance, the poor still rely heavily on the services
of traditional midwives (dukun) for child deliveries: more than half of all children born to the poorest quintile in Indonesia
are delivered by traditional midwives. With no formal training, traditional midwives are unable to foresee possible
complications that may arise during childbirth (see Focus on Maternal Mortality at the end of this section). The utilization
of the traditional midwives with no formal training, as well as lack of access to 24-hour obstetric care (even in situations
when there is a skilled birth attendant), contributes to Indonesia s high maternal mortality rates.
Indonesia lags behind in several other health outcomes, particularly in infant mortality and malnutrition.
Trends towards improving health outcomes in these areas are only moderate. The current Indonesian infant mortality rate
(IMR) of roughly 30 infant deaths per 1,000 births is down from the 70 infant deaths level in 1985, but remains well above
rates in other East Asian countries (World Development Indicators, or WDI). Over the same period, several other lowincome countries achieved faster IMR reductions. Moreover, trends in malnutrition in Indonesia have shown only moderate
improvement despite systematic primary care interventions. In the coming years, effective policies and programs will
need to be put in place if Indonesia is to meet the MDGs on reducing infant mortality (see Focus on Child Malnutrition for
recommendations on possible vertical interventions).
Table 5.2
Comparison of health indicators across countries

Births attended by
skilled health staff
(% of total)
Improved water source

(% of population
with access)
Infant mortality rate

(per 1,000 live births)
2004
Maternal mortality rate

(per 100,000 births)
Thailand
Malaysia
99
97
85
95
18
10
36
50
China
Vietnam
97
90
77
73
26
17
50
95
Indonesia
72
77
30
307
Source: World Development Indicators, MMR and IMR: Unicef, 2004. Births attended by skilled health staff, World Development Indicators (2000-06).
Education sector
Levels of spending on education
Education expenditure has been increasing. In recent years real per capita spending on education in Indonesia
increased by 49 percent between 2000 and 2003, reaching Rp 63.6 trillion in 2003. In 2004, there was a slight decline in
real spending levels compared with the previous year, but spending was Rp 62.6 trillion in 2004 still higher than precrisis levels in real terms104 (see Figure 5.4a). In 2004, education expenditure constituted 2.8 percent of GDP (up from 2.5
percent of GDP in pre-crisis year of 1996-97). As a share of total budget, in 2004 the education sector took up 14.0
percent of total government expenditure at central, province and district levels (PER, World Bank forthcoming). With its
104
Education expenditures in this calculation include all education sectoral spending by Ministry of National Education and Ministry of Religious Affairs, all management costs for the Ministry of National
Education, and the APBD I and APBD II decentralized spending on education coming from SIKD data.
138
2004 level of spending, Indonesia ranks slightly below average compared with countries in the region (see Figure 5.4b).
Both the Constitution and the Law on National Education System stipulate that a minimum of 20 percent of the central
budget (APBN) exclusive of salary costs, and 20 percent of the districts budgets (APBD), also exclusive of salaries, must
be allocated to education.105 In 2003, it is estimated that about 57 percent of education spending went towards financing
salaries. This means that in order to attain the 20 percent quota set by the Constitution (exclusive of salaries), an additional
14 percent of government spending would have to be switched from other sectors into the education sector.106 However,
rather than dramatically increasing resources to the sector in terms of salary spending, the solution to many of the sector s
problems is a more effective use of resources. (For more analysis refer to Education PER, World Bank, forthcoming.)
Figure 5.4a Aggregate public education expenditure has
Figure 5.4b Comparison of public funds available for education 107
increased in decentralized Indonesia (% of GDP)
Source: World Development Indicators, MoF, SIKD.
Source: MoF, Ministry of Education, SIKD.

Note: For details see table in Annex V.2.
Figure 5.5
Sources of basic education spending

from public and private resources (2003)
Figure 5.6 Expanded access to secondary school level has made

education public spending more pro-poor over time
(% indicates percentage of spending from private resources)
Source: Total spending in primary and junior secondary levels for 2003. MoF, SIKD, Susenas
Education Module 2003 All spending data come from 2003.
Source: World Bank, 2006k.
105
The Indonesian constitution stipulates that the state should allocate a minimum of 20 percent from the APBN budget and APBD budget to education expenditures, in order to respond to national education
needs. In 2002, the constitution was amended to specify that The state prioritizes a budget for education of at least 20 percent from the national budget and regional budgets to fulfill the needs of providing
national education. National Education Law No. 20/2003 redefines this benchmark by narrowing the range of spending items that should reach this earmarking target by excluding the salary of educators and
service education. (Education Public Expenditure Review, World Bank 2006.)
106
The calculation is as follows: In 2004, aggregate government spending was Rp 447 trillion; 20 percent of this amount is Rp 89 trillion. In 2004, non-salary education expenditures were estimated at 43 percent
of the education expenditures: Rp 26.9 trillion. An additional Rp 62 trillion needs to be added to education expenditure to make up the difference, which is equivalent to 13.9 percent of total government spending
in 2004.
107
The WDI definition of total public expenditure on education as a percentage of GDP is the current and capital expenditures on education by local, regional and national governments, including municipalities
(household contributions are excluded), expressed as a percentage of GDP.
139
The major reason for increased education spending was the increase in decentralized routine expenditure
and the persisting levels of routine and development expenditure of the central government. Despite
decentralization, the central government continued to spend on district functions: in 2004, 62 percent of education
development expenditure in the regions came through the central budget. While Indonesia decentralized functions and
spending for education services it maintained central spending on education, with the central government continuing to
pursue district government roles for development spending in the regions. Out-of-pocket spending has remained relatively
stable in real terms since 1998 except for the richest quintile, which has experienced increases in education spending in
real terms. This is partly due to a shift away from public to private schools on the part of the richer quintiles.
In the past, Indonesia s education spending was mostly pro-poor as the country successfully channeled
funding into primary school education. The benefit incidence of primary schooling is pro-poor since the poor have
more children and benefit more from funding in basic education. At the junior secondary level the poor capture about the
same amount of benefits as the other quintiles, so increased spending on junior secondary schooling will also benefit
poor quintiles in terms of increasing access. The change in benefit incidence of education spending over time also
suggests this conclusion. From 1998 to 2003, the capture of the poor from benefits in primary schooling increased only
slightly, with enrollment rates being close to universal at the primary school level. At junior secondary school level,
however, far greater strides have been made in terms of the poor s capture of benefits (see Figure 5.7). This suggests that
while the average benefit incidence of education expenditure is pro-poor at the primary school level (at least in terms of
enrollments), the poor will benefit more from expansion of the supply of schooling at the junior secondary level. That
said, it is also important to note that any improvement in the quality of schooling at the primary school level will be propoor given the current distribution of average benefits (Akhmadi and Suryadarma, 2004).108
Priority issues for poverty reduction in education

The poor are adversely affected by three main issues in the education sector: (i) overall quality, (ii) affordability, and (iii)
supply-side constraints in terms of schools and teachers in remote areas.
Quality
Improving learning quality remains a challenge for Indonesia s education system. Given the already high
average enrollment rates, improving quality, especially in primary school education, could have a significantly beneficial
impact on the poor. Standardized test scores are an objective indicator of performance in the education sector. Internationally
comparable test scores reveal that the standard of education among 15-year-olds in Indonesia is slightly lower than some
comparable countries. Indonesia participated in the Program for International Student Assessment (PISA) study109 over
two consecutive rounds, in 2000 and 2003. The fact that the Indonesian government participated in this study provides a
useful opportunity to compare student performance in Indonesia over time, as well as compare levels of learning with
other countries. The study defined three areas of literacy in reading, science and mathematics, and was given to young
adults aged 15 who are approaching the end of nine years of basic education. While Indonesian students improved their
performance in reading and math skills over this period, they remained behind other comparable countries in the sample
(see Figure 5.7). Low quality of schooling raises questions concerning the adequacy of the school system in delivering
returns and improving employability and income prospects. This is especially an issue for poor rural migrants to urban
areas, for whom unemployment is a more serious risk factor. While there is a balance to be struck between the allocation
of resources towards improving enrollments and the raising of quality levels in education, investments in teaching quality
are necessary in tandem with funding for raising enrollment rates if the return to education for the poor is to be increased.
108
The benefit incidence calculation also assumes uniform expenditures on each student not allowing for regional quality differences or variations in absenteeism of teachers. In Indonesia, where the
absenteeism of teachers is estimated to be 19 percent on average, the benefits that accrue to the poor are expected to be lower by at least a similar amount. The data on teacher absenteeism was obtained from
two unannounced visits to 147 sample schools across the country (Akhmadi and Suryadarma, 2004). The data were obtained from two unannounced visits to 147 sample schools across the country, and
revealed that on average 19 percent of teachers were absent for one reason or another.
109
OECD/Unesco-UIS 2003. Overview of PISA: the Program for International Student Assessment (PISA).
140
Figure 5.7 Learning levels among Indonesian 15-year-olds are low, even among the relatively better off
Distribution of PISA 2003 test scores in Indonesia
and comparison countries
Distribution of PISA 2003 test scores by economic status

quintiles in Indonesia
Teacher quality is one determinant of school effectiveness. Indonesia has one of the lowest pupil-to-teacher
ratios in the world at the primary school level at just under 20 pupils per teacher.110 However, this can hardly be seen as a
quality indicator in Indonesia: despite the abundant supply of teachers, the average class size in Indonesia is not so small.
For instance, at the junior secondary level the pupil-to-teacher ratio is on average 14 to 1, although the class size is 37
students on average. This is due to the fact that Indonesia has an oversupply of teachers along with low salary levels and
short working hours per week.111 In fact, Indonesia s low pupil-to-teacher ratio signals inefficiencies in personnel
management in basic education. A ministerial decree (World Bank, 2005d) allocating a minimum of nine teachers to each
school is poorly adapted to the needs of schools. Sixty-five percent of schools in Indonesia are over-staffed, while at the
same time there is an inequitable allocation of teacher entitlements, which works to the strong disadvantage of schools in
remote rural areas. School staffing levels would be far better addressed by a teacher quota per school determined according
to the number of pupils.
Teacher salaries are a large line item in educational finance in Indonesia, although teachers monthly
salaries are low. Indonesia s education spending on salaries is estimated to be about 60 percent of total education
spending. Although the total spending on salaries is high, teachers are paid 21 percent less than other workers with
equivalent qualifications (Education PER, World Bank,forthcoming). This is not the case, however, when hourly wages are
compared: hourly earnings shows that teachers earn more per hour relative to other workers since they work less hours.
In order to improve teaching quality and spending effectiveness, the government could consider reducing the number of
teachers while increasing the salaries of those teachers retained, such that the overall impact on the teacher salary bill is
not too burdensome, but so that quality gains are made in the process.
In order to encourage teachers to take up postings in remote areas, the government is currently
implementing a teacher deployment and incentives package. The incentives package will double (or in some
places even quadruple) teacher salaries depending on the remoteness of the teaching post and the qualifications of the
teacher. As it is not clear that doubling teacher salaries will necessarily improve the quality of teaching or reduce absenteeism,
so it is important that this initiative is coupled with further mechanisms that improve the accountability of teachers to
provide better quality outcomes. In order to improve teacher quality, Indonesia should invest more resources in teacher
networks for career development. Along these lines, more resources could be put into arranging meetings for teachers
working in similar fields to enable them to share their experience and lesson plans, while also providing them with new
teaching materials.
110
Although there are significant variations across districts, this average ratio is even lower than the ratio in the US and many European countries, as well as far below the level set by national policy (19.97 vs.
40). Source: WDI, 2002.
111
Education Public Expenditure Review. World Bank. Forthcoming.
141
Investing in textbooks and in-class teaching materials may be a cost-effective way of improving quality
in Indonesia. Making textbooks freely available produces one of the highest learning returns to investment (McMahon,
2001), since even in the absence of teachers students can continue to learn if they have access to quality textbooks.
Unfortunately, about 40 percent of students in grades 1 to 6 in Indonesia report having no access to textbooks,112 while
more than 50 percent of children at the primary school level in the lowest quintile report difficulties in financing books and
school equipment (Susenas, 2003). In Indonesia, the resources devoted to covering salaries may be crowding out
investments in teaching materials and textbooks. As an alternative, the governments of both South Korea and Singapore
try to maintain an average class size of more than 40 in basic education, enabling resources to be assigned to other inputs
such as books, teaching materials and computers. Educational research across a wide range of countries supports the
view of South Korea and Singapore that this trade-off is cost-effective (World Bank, 2002c).
Affordability
The poor and near-poor are not benefiting sufficiently from secondary education, undermining their
chances of obtaining higher paying jobs in urban areas. The private share of expenditure is small for primary
schooling, but much higher for junior secondary schools, indicating the need to further subsidize the junior secondary
level of schooling for the poor. While most of the resources in the education sector are allocated to primary schools, the
binding constraints for increasing enrollments for nine years of free basic education are mostly on the supply side for
junior secondary schooling, especially in rural areas.
In Indonesia, school fees create barriers to the poor s access to education, particularly at the junior
secondary school level. According to Susenas data, 34 percent of households in the poorest quintile with children
enrolled in school reported difficulties in financing school fees at primary school level and 44 percent reported difficulties
at the junior secondary school level (Susenas Education Module, 2003).113 While education expenditure at the primary
school level is better subsidized for the poor (constituting only 2.8 percent of per capita expenditure for the poorest
quintile compared with 3.3 percent for the richest quintile), at the junior secondary level education expenditure becomes
far more difficult for the poor to finance. Reflecting this financial strain, the poor pay a larger proportion of their expenditure
per capita on education than the richest quintile (7.2 percent vs. 6.1 percent) at junior secondary school level.
General discussion of school fees focuses specifically on tuition fees. In reality, there are numerous other
expenses that are onerous for the poor. Education demand estimates reveal that the indirect costs of education also play
an important role in enrollment decisions (Paqueo and Sparrow, 2005). Where possible, it is important for the government
to implement programs that reduce such costs in order to improve access to schooling for poverty alleviation purposes.
The extra costs of schooling, such as textbooks, uniforms and transportation, make up a larger share of costs for the poor
than school fees at all levels of schooling. For instance, for the poorest two quintiles, 55 percent of education spending at
the primary school level goes towards textbooks, stationary, uniforms etc. while only about 18 percent of education
spending goes towards tuition fees at this level of schooling (Susenas Education Module, 2003). Uniform costs are
strikingly high in Indonesia with multiple school uniforms being required by schools even at the primary school level: for
the poorest two quintiles in Indonesia the spending on uniforms is more than double the amount spent on school books
at the primary school level (see Figure 5.8).
Another barrier to education is the opportunity cost of schooling. Instead of attending school, children can choose
to work, either for a wage or in the household. Higher local labor wages have the effect of reducing demand for schooling by
raising the opportunity cost of attending school. Controlling for regional and household characteristics, every Rp 1,000
increase in the local child-wage rate reduces demand for education at the junior secondary school level by 0.4 percent
112
According to UNICEF, 40 percent of students in grades 1-6 do not have books, 43 percent in grades 1-3 and 38 percent of those in grades 4-6. http://www.right-to-education.org/content/unreports/
unreport8prt1.html#contents
113
Since this finding only includes households with children in school, there may even be a downward bias in these estimates.
142
(Paqueo and Sparrow, 2005). A regional analysis of the average-odds ratio114 reveals that the poor in Java/Bali are less likely
to continue on to junior and senior secondary school compared with the national average for the same income groups. The
determinant of dropping out in this region could be the opportunity cost of schooling, with more job options available to
primary school graduates. In this case, spending on junior secondary schools needs to be better targeted towards the poor
and needs to cover not only the extra costs associated with schooling but also the opportunity cost.
The Operational Aid to Schools (BOS) program is a valuable attempt to improve affordability of education
(see Box 5.1 on the BOS program in this chapter). However, while the BOS program offers value in terms of improving
quality of schooling, it may not be as effective as a demand-side transfer in improving school affordability and raising
enrollment rates. As such, additional block funding to primary schools may not be the most effective way of using
precious resources to increase affordability for poor students or improve teaching quality. The need for increased funding
for front-line providers could be addressed by better allocation of resources within the sector. International experience
indicates that if additional funds are allocated to address the issues of quality and affordability of education for the poor,
additional targeted subsidies for the poor and/or conditional cash transfers (CCTs) are both possible options.
Targeted CCT or scholarship programs could have a greater impact on school enrollments. Developing a
targeted program, possibly a conditional cash transfer (CCT) program, may be key in addressing this issue from the
demand side. There is also a clear need for a program to address a supply-side shortage of secondary schools. Here too
the issue of teacher management and training is crucial, as it will be necessary to ensure that adequately qualified teachers
are available to staff new schools and classes. The return to secondary school education also needs to be improved if
there is to be demand. From the education sector perspective, this also entails a careful examination of the secondary
school curriculum to establish whether it adequately prepares young Indonesians for the demands of the job market.
Figure 5.8 Non-fee costs of schooling are higher than the cost of school fees for the poor
(Private costs of education for the poorest 40 percent)
Primary school level
Junior secondary school level
Source: Susenas Education Module, 2003.

Note: The data are provided initially for the July-December 2002 six-month period and depending on the frequency of spending on certain items some are multiplied by 2 in order to obtain annual expenditure
amount.
114
The average odds-ratio of participation is given by the ratio of the quintile-specific average participation rate to the overall average.
143
Box 5.1 Does the Operational Aid to Schools improve affordability for poor students? Issues and concerns regarding
the BOS program
The Operational Aid to Schools (BOS) program was put in place in mid-2005, as the government moved to raise fuel prices and
divert substantial funding away from this regressive commodity subsidy into the education sector. The program provides block grants
to participating schools, in return for their reducing or eliminating school fees up to the amount of the grant. All public and private schools are eligible for
the program. School grant size is about US$25/year/student at the primary school level and US$35/year/student at the junior secondary level. The status
of schools within the program varies according to their initial reported budgets: if school were able to raise funds from parents prior to the program in
excess of the funding allocation through the program, such schools can continue to charge tuition fees but should cancel fees for poor students and lower
fees for other students where possible. Schools with previous incomes less than the grant are required to cancel tuition fees altogether. The budget of the
program is generous: Rp 5.3 trillion in funding for June to December 2005 and Rp 11 trillion for 2006. Annually, this amounts to about 15 percent of
Indonesia s education budget in 2004.
In previous block grant programs, schools received the same amounts of funding independent of how many students they served.
Providing grants based on the number of students rather than a set amount per school has the benefit of targeting fund allocation and providing incentives
to schools to expand the coverage area of each school. A rapid assessment of the program found, however, that schools with small numbers of students,
many poor students, or located in isolated areas were disadvantaged by this allocation mechanism. A review of the allocation mechanism with some
emphasis on equalization for disadvantaged schools could reduce this problem.
Another major concern with allocation by student numbers is that school administrators may be tempted to inflate artificially the
number of students at their schools. Program monitoring needs to include verification of student attendance throughout the year (for example,
through spot checks). In addition, program transparency could be expanded into the community through various mechanisms such as public (school
committee) participation in allocating BOS funds at the school level, and publication of the BOS program budget and expenditures at each school.
The program has been successful in improving school finance, especially since the funding goes directly to the school level and
thus school administrators can calculate precisely how much they should receive from the program based on the size of their
student bodies. The simple formulation for funding allocations renders the program highly transparent, at least down to the school management level.
There are strong concerns, however, regarding the targeting of poor students, which is not being expressly carried out in most schools according to a
recent assessment of the program. This is occurring for several reasons. The objective of providing free tuition for poor students, and other forms of
support such as subsidized textbooks, uniforms and transport, has not been communicated well at the local level. In addition, school administrators are
under pressure to reduce or eliminate fees for all students, regardless of their economic status, due to misconceptions that the program is intended to
provide free education for all. The current design contains no accountability or feedback mechanism to ensure that tuition fees are lowered or eliminated
for poor students. Given its supply-driven design, targeting the poor at the school level is crucial for achieving the program s poverty alleviation outcomes.
The program should be complemented with a targeted scholarship component, and clarification and enforcement of the rules for
targeting poor students. Currently, there are clauses in the program design that enable school administrators to provide stipends to poor students to
cover transportation and extra school costs over and above tuition fees. However, the definition of these transfers is very weak and no quota exists for
targeting poor students. In further revisions of the program, the definitions of such transfers to poor students need to be clarified and their implementation
monitored in a more rigorous fashion.
Finally, while recentralizing school finance is, on one level, an advantage of the program, as it makes the flow of funding more
predictable down to the school level, it is also a disadvantage in that district governments may see this large highly visible
program as an opportunity to reduce their own operational spending on schools. In the end, this could result in the crowding out of local
government spending, impairing coordination in school financing across levels of government. The Department of National Education stated recently that
it plans to create a mechanism for local governments to provide matching funds for the program from 2007. It is hoped that this will overcome some of
these issues.
144
Supply-side constraints
There is a general lack of access to secondary schooling due to inadequate supply. This creates a bottleneck for school progression in general, but one that particularly hinders the educational attainment of the poor. The
availability of schools at the junior secondary level is much lower than for primary schools with a 7 to 1 ratio, making it
more difficult to access basic education for all children aged 12 to 15.115 Given that the government s goal is to bring about
universal access to nine years of basic education, there is a need to expand resources for the junior secondary level, since
this is the level at which the access of the poor breaks down.
Junior secondary school capacity in Indonesia provides learning opportunities on average to only 84
percent of potential students in the 13 to 15 age group (Hartono, 2005).116 In Central Java and Yogyakarta
provinces, the average school capacity exceeds 100 percent. This means that the available classrooms are under-utilized.
Conversely, in East Nusa Tenggara and South Sumatra provinces, the average coverage of school capacity is below 60
percent of potential students, indicating a lower degree of access. In terms of the condition of classrooms, a ministry
survey conducted in 2004 revealed that 57.2 percent of primary school and 27.3 percent of junior secondary school
(general and madrasah) classrooms are damaged in some way (Ministry of National Education, 2004). Renovation of
classrooms and the conversion of some primary schools (of which there is an over-abundance in most provinces) into
junior secondary schools should be prioritized to reduce constraints on the supply of classrooms.
Some gains can still be made in primary school access in remote regions. By looking at potential enrollment
rates of the poor in a region (the number of children in appropriate age groups) and comparing these to the existing
distribution of benefits across quintiles, one finds that there are still potential gains to be made by expanding primary
schooling in the two regions of Nusa Tenggara/Maluku and in Papua. For the other regions, the benefits to the poorest
quintile for primary schooling appear to be saturated (the benefits accruing to the poor are at least proportional to the
population of children) in terms of enrollment rates, although quality gains can probably still be made in schools attended
by predominantly poor children.
The low quality and absenteeism of teachers in remote and rural areas also disproportionately impact
the poor. In Indonesia, remote schools have higher student-teacher ratios (25:1 for remote schools vs. 20:1 average).
According to the Teacher Employment and Deployment Study, despite general teacher oversupply in Indonesia, 74 percent
of remote schools are below entitlement (World Bank, 2005d). There is also an unequal distribution of honorary teachers
with predominance in remote schools (78 percent have honorary teachers), thus reducing the quality of teaching in
remote areas. A wide range of incentives for teachers exists, coming both from local governments and the schools
themselves, which creates a high incentive differential between teaching in urban Jakarta compared with in remote rural
areas. Salaries often depend upon a district s or a school s capacity to pay. Thirty-six percent of teachers receive local
government incentives, while 14 percent receive school incentives.
Teacher incentives and management are issues that require
urgent attention, especially given that such a high proportion
of education sector resources go towards teacher salaries.
With the passage of the Law on Teachers and Lecturers in December
2005, the government is trying to couple a teacher certification and quality
enhancement scheme with improved teacher incentives. The law
stipulates that all teachers must be certified within 10 years and that,
115
Total number of primary schools in Indonesia is 147,793, while the number of junior secondary
schools is 22,274. Source: Indonesia Educational Statistics in Brief, 2004-05.
116
Calculated based on Ministry of National Education data (Proyeksi Pendidikan Tahun 2003/2004
dan 2009/2010) based on the assumption of 38 students /junior secondary school classroom.
145
upon certification, they will receive a professional allowance equivalent to their base salary. A special area allowance is
specified in this law, which will be given to teachers in conflict, natural disaster, remote, and other hardship areas. This is
expected to increase the number of teachers available for teaching in remote areas but the impact of this expensive
reform on absenteeism and quality of teaching depends on the implementation of reform.
Recommendations for pro-poor spending on education

This analysis points to some key priorities and strategic directions for improving spending in education to make it more
pro-poor and help address Indonesia s key poverty-related development goals.
1. Ensure affordability, continued high enrollment and improved quality of education for primary
schools. Primary school enrollment outcomes are high and an appropriately high level of funds is allocated to this
sector. The benefits of these funds can be improved further for the poor with particular emphasis on maintaining
affordability. The BOS program is a worthy attempt to address this issue. However, it is questionable whether additional
block funding to schools is the most effective way to increase affordability to poor students and improve teaching
quality. The need for increased funding for front-line providers could be addressed by better allocation of resources
within the sector. International experience indicates that if additional funds are allocated so as to address the issues
of affordability and quality of education for the poor, the government should consider either further targeted scholarships
for the poor and/or conditional cash transfers (CCTs). Reducing the number of uniforms demanded by schools will
be also be a pro-poor reform since the poor spend twice as much money purchasing various uniforms as they spend
on textbooks.
2. Balance class sizes in primary schools to improve quality of education. There is an oversupply of
primary school teachers, coupled with inefficient and inequitable allocation of teacher entitlements to schools. Given
that 65 percent of schools in Indonesia are over-staffed based on the current policy (World Bank, 2005d); the
government should change the law requiring a minimum allocation of nine teachers to each school. This could
involve a system whereby school staffing levels are calculated on a teacher-quota-per-school basis, determined
according to the number of pupils. The government should also encourage teachers to take further training and
certification to move to junior secondary teaching and its higher levels of compensation.
3. Improve incentives to front-line providers and teachers, including through better teacher
management. Improving incentives is key to improving the quality of educational outcomes, including the quality
of education itself. The government should enhance incentives both to local governments and schools in order to
enhance education capacities. It should consider the use of conditional grants contingent on improved educational
outcomes to enhance desired outcomes, such as improved enrollment rates for the poor. Given the large proportion
of education sector resources that go towards teacher salaries, there is also a clear need for the government to
address weaknesses in teacher management. It is important for the government to introduce incentives to attract
teachers to remote schools and to devise a simplified remuneration structure that is more equitable and includes
national incentives for teaching in remote schools.
4. Redouble efforts on primary to junior secondary school transition. The primary challenge to meeting
Indonesia s education targets is to reduce the drop-out rate in the transition to secondary school-a problem that
applies particularly to the poor. The government should consider developing a targeted program, possibly a conditional
cash transfer (CCT) program, to address this issue from the demand side. On the supply side, the government could
consider addressing the shortage of secondary schools through the conversion of some primary schools to secondary
schools, the construction of new schools, or both. In the construction of new junior secondary schools there are
opportunities to mobilize community labor, thereby reducing construction costs. Opportunities also exist to convert
146
primary school buildings in some areas into junior secondary schools by considering demographic changes and the
distribution of primary and junior secondary school aged children, forecasting declining future demand for primary
schools by district and mapping this with existing school data. In densely populated areas more schools can be
turned into double-shift schools at the primary school level, freeing up additional school facilities for use by the
junior secondary school level. At the same time, the government will need to ensure that adequately qualified teachers
are available to staff these new schools and classes.
5. Focus on improving learning quality. Research indicates that investing in in-class teaching materials is one of
the most cost-effective ways of improving learning quality. Making textbooks freely available also produces one of
the highest returns to investment (McMahon, 2001). Very high numbers of students in Indonesia report having no
access to basic textbooks. The largest resource component covers salaries at the school level leaving only limited
resources for learning materials. In terms of improving teacher quality, the government should also invest more
resources in teacher networks for career development. Along these lines, more resources can be put into arranging
meetings for same-field teachers in order to share experience and lesson plans, while also providing them with new
learning materials.
Health sector
Levels of spending on health
Public health expenditure in Indonesia has been increasing
but is still well below international allocations, even among
comparable countries in the region. Indonesia has seen some
increases in aggregate public health spending from 2000-03 and a slight
decline in 2004 (see Figure 5.9). Despite this, real per capita public health
spending in 2004 was 47.8 percent higher than in 2000. In 2004, public
health spending was Rp 16.7 trillion, with 50 percent of development
spending coming from the central government. The level of public health
spending still remains low by international standards. In 2004, as a share
of total budget the health sector constituted only about 3.8 percent of
total government expenditure. Despite the recent increase in healthcare spending, Indonesia still spends less on health as
a percentage of GDP than comparable countries in the region (see Figure 5.10).
The determinants for this increase in public spending are higher decentralized spending by district
governments and continuing deconcentrated development spending from the center being spent in the
regions. It is still too early in the stages of decentralization to say whether the trend of increasing health expenditure will
continue. Currently, it seems that the increase in spending on health is a consequence of the central government holding
onto health spending while local governments also increase their own health spending (in addition to the other sectors
that now fall under their responsibility). In the future, public health spending needs to be better coordinated across levels
of government, as well as increased. It also needs to be channeled more carefully towards the areas of greatest need.
Currently, public health spending generally benefits the richer quintiles more than the poor through
regressive subsidies for secondary care, as well as spending being mainly channeled to richer districts.
As a consequence, there is a need to better target health spending, not only towards the type of facilities and programs that
are funded, but also in the geographical distribution of the spending, in order to better reach the poor. Currently, the
benefit incidence of public spending on primary healthcare is not pro-poor but neutrally distributed among quintiles,
147
while spending on secondary healthcare is certainly not pro-poor, with most of the benefits accruing to the richer quintiles.
While the public health services most utilized by the poor in Indonesia are basic healthcare facilities, Indonesia spends
about 40 percent of public healthcare resources on regressively targeted subsidies to public hospitals (see Figure 5.11).
The poor have very little access to public hospitals and, hence, do not make use of the vast majority of the spending that
goes into secondary healthcare. Of the funding that goes into hospital care, the benefits that accrue to the poorest quintile
of the population are about 9.5 percent, while those that accrue to the richest quintile are about 40 percent. Spending on
secondary healthcare is a highly regressive way of allocating limited resources at a time when Indonesia is struggling to
meet its medium-term development targets in health.
While increasing marginally in the period since the crisis, the increase in public health expenditure
has not been pro-poor. Most of the increase went into secondary healthcare, despite the regressive nature of supplyside secondary care subsidies: whereas in 1995-96 the proportion of health expenditure going into secondary care was
about 25 percent, this ratio increased to 40 percent of health spending in 2003. Overall, the poor s utilization of health
services and their capture of health spending have not increased significantly since 1998 (Figure 5.12). The PKPS-BBM
healthcare program is aimed at both increasing access to basic and secondary healthcare for the poor in a targeted way.
This program, if effectively targeted and implemented, could be key in expanding health services for the poor (see Box 6.7
in Chapter 6).
Figure 5.9
Aggregate public health expenditure has

increased in decentralized Indonesia (% of GDP)
Note: See Annex V.3 Health Spending in Indonesia 1994-2004 for details.
Figure 5.10 Indonesia s total health spending (from public and

private sources still remains low
Source: World Development Indicators.
Figure 5.11 Sources of healthcare spending in Indonesia (2003) Figure 5.12 Over time benefit incidence of public health spending
(% indicates percentage of spending from private resources)
Source: Susenas 2003, MoF and SIKD.
148
Priority issues for poverty reduction in health

The poor have a high vulnerability to health shocks that have a negative impact on family welfare. For
instance, across all groups loss of income from health factors is more frequently reported than loss of income from
unemployment (see risk and vulnerability analysis in Chapter 6 on Social Protection). The urban poor male-headed
households and rural poor (and non-poor) female-headed households have a particularly high vulnerability to health
shocks. Also, among urban households headed by women, loss of income due to health factors is the second most
commonly reported shock across all income groups, whereas for other groups loss of income from health factors is more
likely to be the third most frequently reported shock. Although increasing spending in healthcare alone is insufficient for
achieving health sector goals in itself (without the correct institutional set-up and accountability mechanisms), the right
level and mix of pro-poor spending in the sector, as well as improving the efficiency of that spending, matter tremendously
for reaching health and poverty reduction goals in the medium term.
Among the priority tasks in healthcare are: (i) improving quality of healthcare provided in basic community
healthcare clinics (Puskesmas); (ii) investing in the training of private (and public) paramedics to benefit the poor, especially
in rural areas where the poor have little access to other services; and (iii) investing in demand-side activities that increase
the access of the poor to in-patient care. Beyond these priorities, many of the Indonesian health sector s problems are
related to systemic problems and institutional constraints in service delivery. In addition to providing increased health
financing through supply and demand-side schemes, in the medium term there are systemic problems within the sector
that need to be addressed, including management of the health sector workforce and determining the function of provincial
governments. These issues are discussed further in Chapter 7 on Government in which the institutional problems of the
sector are explored.
Quality of basic healthcare

In order to align public expenditure on health with the needs of the poor, a higher proportion of subsidies
for basic healthcare services through public or private provision is required. While less cost-effective services
in secondary in-patient healthcare should continue, these should be less dependent on public subsidies as they benefit
mainly the wealthy. Instead, Indonesia needs to redirect current public spending for health, as well as any increases in
expenditure, in order to better meet the needs for public healthcare with the aim of providing a minimum package of
essential clinical services that are accessible to the poor.
Contributions from patients, with a rate determined at around US$0.25/visit, make up only about 2
percent of Puskesmas revenues in Indonesia.117 Having low levels of user fees at the Puskesmas is not necessarily
pro-poor. User fees may in fact improve the welfare of the poor. If fees are used to improve quality and if, as a result of the
quality improvement, demand for basic healthcare by the poor increases, then the welfare of the poor will also increases
(Bitran and Giedion, 2003). The Puskesmas need to be able to charge higher fees from those who can afford to pay for
services. In the current set-up, the richer quintiles benefit from subsidies to the Puskesmas just as much as the poor and
they utilize these services almost as frequently. Finding the optimal level of user fees for the richer quintiles and being
able to better target the poor, who will be exempt from these fees, are both key to increasing the resources available at the
Puskesmas and for improving quality of services. In Tanah Datar and Purbalingga districts, where the Puskesmas were
allowed to charge higher user fees for services, there has been an increase in the utilization of health services by the poor
due to better perceived quality of services.118
Absenteeism of health personnel is a major problem in ensuring quality of healthcare in public health
clinics. A recent study involved making surprise visits on more than 100 primary schools and health centers in Indonesia
117
118
GDS 1+ findings Puskesmas finance data.

Dinas Kesehatan Tanah Datar and field observation in Purbalingga.
149
(Chaudhury et al, 2006). The study found absentee rates of 19 percent among teachers and 40 percent among health
workers. Indonesia had the highest health-worker absentee rate of all the countries included in this global study, worse
than Peru, at 25 percent, Bangladesh at 35 percent, and Uganda at 37 percent. Not only does high absenteeism reduce
quality, but it also reduces the demand for public health services, since people are reluctant to travel long distances when
there is a significant chance that health personnel might be absent. High absentee rates may be a consequence of many
Puskesmas doctors running their own private practices: the GDS1+ data collected from over 120 Puskesmas showed that
three-fourths of Puskesmas heads also had private practices of their own, which has implications on the time they are able
to allocate to public health centers.
Investing in private provision of healthcare

Supporting private sector provision of healthcare will also be important for the poor. Since the early 1990s,
private healthcare has started to play an increasingly important role in the financing and delivery of services in Indonesia.
Indonesia s approach to private providers expanded the opportunities for private doctors and nurses to practice and added
to the availability of health services to both poor and non-poor households. In the 1980s, when the low salaries of
government health workers made it difficult for them to keep practicing their profession, the government-rather than
restricting levels of employment and raising salaries allowed its staff to maintain private practices outside of their normal
working hours (World Bank, 2003). Today, the majority of healthcare professionals in Indonesia engage in the delivery of
both public and private services. While this dual position of public health providers created perverse incentives, lowered
the quality of services in the public health system (mainly due to the reduced number of hours these doctors put into
public practices, as well as the usage of drugs and other public facilities in their private practices), it also allowed the
private provision of services to develop and the average number of hours served by trained physicians and paramedics to
increase. Arguably, one of the reasons why Indonesia is probably not doing too badly in most health outcomes despite
low levels of public spending is that private provision services have filled the service provision gap in areas where public
provision has been inadequate in supply or quality. In this situation, private providers are very much part of health service
delivery in Indonesia, and training and the contracting and monitoring of services need to be an integral part of government
health policy.
Figure 5.13 Health utilization of health providers by income groups (2004)
Public hospital
Of the private providers, the poor make most use of private paramedics and doctors (see Figure 5.13). With
increasing income there is a switch away from paramedics towards doctors. The average-odds ratio of participation the
ratio of the quintile-specific average participation rate to the overall average provides a useful tool for understanding the
current utilization of services and highlighting those quintiles the services are likely to benefit most. The average-odds
150
ratio of participation is highest for the poor in public Puskesmas and private paramedics (nurses, midwives etc), which
means that investments in these areas, if participation rates remain the same across quintiles, will be more likely to benefit
the poor than the richer quintiles. In contrast, investments in private polyclinics, and public and private hospitals are
among the most pro-rich investments in Indonesia given the underlying utilization rates for health services.
Investing in demand side activities that increase the accessof the poor to in-patient care
When the binding constraints on the poor s access to services are on the demand side due to reasons
that limit their access to these services, it may be more appropriate to stimulate the demand for these
services through voucher or cash transfer schemes. A demand-based financing scheme that is pro-poor targeted
gives more power to the client to select preferred providers. A well-targeted demand-based program also has the potential
to provide incentives for improving the quality of healthcare for the poor. In order for demand-side programs to succeed,
two conditions need to be in place: (i) the supply of services should be already available in the area covered by the
program. If not, demand-side programs will be ineffective and transfers to households may result in wasting significant
resources with little impact on outcomes; (ii) the targeting of the poor, which has been so problematic in Indonesia, needs
to be improved so that program benefits reach the poor. Targeting remains a central issue for reaching the poor with
demand-side programs. The distribution of voucher benefits is of critical importance in the design and targeting of any
demand-side intervention scheme. The objective can be achieved by geographic targeting, which strengthens provision
of services through the Puskesmas and other available private providers in the area, especially those in districts and subdistricts with a high proportion of poor families. The targeting function can be contracted out to NGOs that then ensure the
distribution, registration and recording of vouchers is pro-poor.
Indonesia has experienced a series of targeted insurance programs since the mid-1990s, starting with
the health card (kartu sehat) program in 1994. In the initial program, poor families were distributed health cards
by their village heads, which allowed them to access the nearest Puskesmas and referral care in third-class in-patient
wards in district hospitals. Providers were reimbursed on the basis of fixed fees varying for out-patient or in-patient visits.
However, healthcare staff were not compensated for the services they provided to card holders and the program was not
fully funded. By 1998, the program had become largely inoperative and the social safety net (JPS) program that was
designed after the crisis, while based on a similar system, did not continue the transfers to the same beneficiaries (World
Bank, 2003; Knowles and Marzolf, 2003).
In 1998, the government developed several targeted programs to protect the poor in the face of the
crisis. These programs, collectively referred to as jaringan pengaman sosial (JPS) or social safety net programs, and
included workfare, subsidized rice sales, village block grants and targeted scholarships, as well as subsidized health
services. The JPS-BK program provided block grants to health providers through district post offices in a way that was
proportional to the number of poor families residing in an area. The objectives of the JPS-BK program were to assist poor
families to cover the costs of basic health services and referrals, and to provide nutritious supplement foods for children
and pregnant mothers of poor families. The targeting of the JPS-BK health cards was done on the basis of the BKKBN119
Family Welfare (KS) targeting system. The health card entitled the families to free curative and preventive care at public
health facilities, and private providers were not included in the scheme. The program also included in-patient care on
referral in third-class in-patient wards of public hospitals, as well as contraception and mother-and-child healthcare
(MCH) from village midwives. In 2001, the government introduced additional subsidies aimed at healthcare services for
the poor through the BBM subsidies program. In 2002, this program was renamed the fuel-subsidy compensation program ,
or PKPS-BBM. However, the nature of the program remained the same (Arifianto, Tan et al, 2005).
119
Badan Koordinasi Keluarga Berencana Nasional, or BKKBN (National Family Planning Coordination Board).
151
Pro-poor financing schemes need to be based on the real costs of providing public health services to
the poor. Otherwise the services will either not be provided or be of low quality. The JPS-BK program used capitationbased grants in order to provide supply-side funding to health providers. The scheme distributed funds to service providers
on the basis of the number of poor residing within their jurisdiction. Throughout this program, providers received funding
regardless of their services to the poor. In this regard, although the program targeted the poor and was meant as a targeted
demand-side program, in reality it worked as a supply-driven program only subsidizing the utilization of the poor (Pradhan
et al, forthcoming). If the funding scheme had been based on a fee-for-service or prospective payments system, there may
have been more incentives on the part of providers to reach out to the poor. The health-card program, while intended to
operate as a demand-side health insurance program, in practice fails to function in a demand-driven way since the flow of
money is not attached to final service delivery.
Although the lack of accountability of providers to the poor was a problem in itself, the more significant
concern with the health-card scheme was the ineffective targeting of the program. The program benefits
were almost homogenously distributed across quintiles (see Chapter 6 for targeting analysis). In 2004, only about 22
percent of the households in the poorest quintile were covered by the health-card program and only about 31 percent of
the cards were given to the poorest 20 percent, while 12 percent of the cards found their way into the households of the
richest quintile. Health cards therefore had a limited impact because of socialization problems that resulted in poor
targeting. For the JPS-BK program, targeting was slightly pro-poor but the degree of targeting effectiveness fell short of
what has been achieved in some other countries. In the design of the program, there was little incentive for providers to
socialize the program.
120
Posyandu are public health posts run by volunteers with support from sub-district health services. First started in 1984, the posts were established to weigh infants and provide nutritional, immunization,
mother-child care and family planning advice to mothers.
152
Regardless of the poor targeting performance of the JPS-BK and PKPS-BBM health-card programs, in
general they have provided benefits to the beneficiary groups in terms of the utilization of public services.
There has been a significant increase in use of health facilities, such as the Puskesmas and Posyandu,120 and supplementary
feeding to the poor was provided through the program (children under five years of age and pregnant mothers) (SMERU,
2005). Conditional on being sick, the poor s likelihood of using public facilities was higher for those who had access to
the health card when compared with peers in the same quintile (Pradhan et al, forthcoming). For all quintiles, those who
had access to health cards were less likely to only self-treat themselves than those who did not have cards. Card holders
were also slightly more likely to use public services than private service providers. As expected in any demand-side
intervention, the benefits resulting from the health-card program have generally been found to be directly proportional to
the particular geographical aspects and available infrastructure of a certain area. The benefits of the JPS-BK program in
rural areas, particularly in remote areas, tended to be less evident than the benefits evident in urban areas.
In 2005, the government once again undertook a remarkably large scheme of health financing aimed at
increasing access and health service quality to all people, in particular to the poor, so as to achieve
better health outcomes. The program is very similar to the predecessor JPS-BK (1998-2001) and PKPS-BBM (200105) programs in that it also proposes to provide: (i) free-of-charge health services at Puskesmas; and (ii) in-patient
treatment at third-class hospital beds for the poor. The only difference in the new program is that the in-patient treatment
part of the program is run by PT Askes, whereby health cards are distributed by this insurance company and the hospitals
are reimbursed for their services on a fee-for-service basis. (Further details of the program can be found in the PKPSBBM Box in Chapter 6 on Social Protection.)
Recommendations for pro-poor spending on health

While the overall allocation of public spending on health is low, augmenting spending needs to be done
strategically if it is to be effective, especially if it is to benefit the poor. The benefit incidence provides three
major insights into how the government can improve health services for the poor:
1. Increasing spending on primary public healthcare services for the poor and focusing on interventions
that improve the quality of services. Investing in basic healthcare facilities on the supply side can be done
through public or private provision. There is great merit in centralized campaigns to address major communicable
diseases. Much can be accomplished through major communication campaigns on basic health information when it
comes to tackling communicable diseases in Indonesia today.
2. Investing in the training of private paramedics benefits the poor, especially in rural areas where
the poor have little access to other services. Their use by the poor should be subsidized through demandside voucher schemes allowing for the poor to claim benefits when using private providers. Investing in improvements
in the quality of the private-sector providers giving healthcare to the poor will in effect be pro-poor. De facto, this is
an important part of the healthcare system and needs to be reinforced.
3. Investing in demand-side activities that increase the access of the poor to secondary in-patient
care. In the current utilization pattern, the poor benefit very little from subsidies going to hospitals and the only propoor financing for hospital care would be through targeted vouchers (health cards) that allow free care for the poor on
a fee-for-service basis. In order to improve the pro-poorness of health financing, all other subsidies to secondary
care facilities should be channeled into primary care. There may be special merit to subsidizing ambulatory care,
especially in remote regions. The current BBM-PKPS program is well-intentioned in terms of increasing the poor s
access to primary and secondary in-patient care. The key is to make sure it works, and to assess it and improve it
incrementally (see Box 6.6 on the PKPS-BBM program in healthcare in Chapter 6).
153
Focus on Maternal Mortality121
The risk of death during childbirth or shortly after delivery is
the proportion of institutional deliveries and improving access to
significant in Indonesia. The Indonesian maternal mortality ratio (MMR)
24-hour obstetric care.. Previously, the government s strategy to reduce
122
is 307 per 100,000 live births, implying that a woman who decides to have
maternal mortality focused on increasing assisted deliveries and access to birth
four children has a probability of 1.23 percent of dying as a result of her
control. The bidan di desa (Village Midwives) program aimed to increase access
pregnancies. Maternal deaths account for an estimated 21 percent (Djaja, 2000)
to midwife services. During the 1990s, some 54,000 nurses underwent a one-
of all female deaths of women of reproductive age (National Household Health
year course to become village midwives and were placed in underserved villages.
123
Indonesia fares poorly compared with most of its regional
Studies do indicate that the arrival of a midwife was linked to a significant
peers: the Philippines has an MMR of 172 based on data from 1997 (UNDP,
increase in infant birth weight (Frankenberg and Thomas 2001) and subsequent
2003), Malaysia s MMR is a mere 20 (UNDP and Government of Malaysia
nutritional health of children (Frankenberg, Suriastini et al, 2005). These are
2005). Only Cambodia has a worse ratio than Indonesia s, at 437 based on the
signs of healthier pregnancies, which also help to reduce the likelihood of
DHS survey from 2000 (Kingdom of Cambodia, 2003).
maternal mortality. Despite some improvements, currently only 72 percent of
Survey, 1995).
births are attended by skilled personnel in Indonesia nationally, compared with

Main medical causes leading to maternal mortality are
97 percent in Malaysia and China, and 99 percent in Thailand (WDI, 2006).
hemorrhaging (bleeding), infection and eclampsia. Over 60 percent
Most deliveries, 59 percent, are at still at home and only 9 percent of deliveries
of maternal deaths are estimated to be caused by these direct obstetric
are at public sector health facilities. Most institutional deliveries (30 percent of
complications. Hemorrhages occur most frequently and account for between
all deliveries) are in private practices, often in the homes of midwives (BPS and
25 and 45 percent124 of maternal deaths, followed by eclampsia (13 percent)
ORC Macro, 2003). Increasing deliveries with skilled staff in attendance at health
(JHPIEGO, 2004) and post-natal infection (10 percent) (Supratikto et al, 2002).
clinics will require action on several fronts.
The majority of deaths caused by hemorrhaging are reported to be a result of a

retained placenta a strong indication of inadequate management of the third
Five steps towards reducing maternal mortality in Indonesia
125
stage of labor. Death due to infection is an indicator of poor prevention and

management of infections. Unsafe abortions and a lack of post-abortion care
also contribute to a significant percentage of maternal deaths, with government
estimates based on the Indonesian Demographic and Health Survey (IDHS)
indicating that abortions comprise 11 percent of all maternal mortality (Ministry
of Health, 2003). The pattern of maternal mortality highlights the importance of
skilled midwifery during labor, as the risk of most complications can be
significantly reduced by the presence of a skilled birth attendant (bidan) and
the availability of a referral system to 24-hour obstetric care if complications
arise. Preliminary estimates (Graham, Bell et al, 2001) suggest that only 16 to
33 percent of maternal deaths due to obstructed labor, eclampsia, infection and
1.
Increasing availability of skilled midwives in remote areas.

Nationally, there is one midwife per 4,000 of population, which is actually
better than the internationally recommended level of one per 5,000.126
However, remote areas certainly do lack skilled professionals for delivery.
There is huge variance in provision: in Java/Bali the average client has to
travel 1.5km to see a midwife, while on many of the outer islands
(Kalimantan, Maluku, Nusa Tenggara and Papua) average travel distance
varies from 12km to 30km. Thirty percent of rural villages have no midwife
present, while 57 percent have no maternity house where children can be
safely delivered (Podes, 2003).
hemorrhaging could have been avoided by the presence of a skilled midwife at
Keeping skilled midwives in remote areas is a challenge and past efforts
delivery. A well-functioning referral system is needed to reduce MMR further.
to locate midwives in remote villages have proven unsustainable. Midwives
Progress in reducing maternal mortality lies in increasing the

proportion of births attended by skilled professionals, increasing
121
are less likely to live in remote areas, as conditions are difficult and there
are fewer paying clients. Of the bidan placed in underserved villages under
Adapted from Tan, Ellen (2005). Determinants of Maternal Mortality in Indonesia. Policy Note. World Bank Office Jakarta.
The estimate comes from the 2002 Indonesian Demographic and Health Survey (IDHS) and is based on reported deaths over the period 1998 to 2002.
123
In the Tenth revision of the International Classification of Diseases maternal death is defined as the death of a woman while pregnant or within 42 days of termination of pregnancy, irrespective of the duration
and site of the pregnancy, from any cause related to or aggravated by the pregnancy or its management but not from accidental or incidental causes
124
IDHS 2002/2003 PPH attributed to 28 percent of maternal mortality.
125
Ministry of Health (2001).
126
Set by the International Confederation of Midwives (ICM) and the International Federation of Gynecologist and Obstetricians (FIGO). Koblinsky et al (2003).
122
154
the Village Midwives program in the 1990s, only 40 percent elected to
an improvement in birth attendance by skilled professionals (CCTs).
stay in the villages where they were placed once their government contracts
Vouchers to poor pregnant women, similar to the TPC program but not
has expired. However, potential solutions are being explored: Unicef and
distributed by midwives, would also be a more direct way of promoting
the government are supporting a new project, called Improving Maternal
access for poor women.
Health in Eastern Indonesia (JHPIEGO), which aims to provide enhanced

support from district health centers for village-based midwives. Taking a
3.
Increasing awareness especially among women of the

importance of skilled midwifery at birth. The poor often fail to
fundamentally different approach, JHPIEGO attempts to identify daughters
seek care from trained professionals, because birth is considered a natural
of TBAs in remote areas and offer them scholarships to enable them to
process that does not require professional skills. A recent Voices of the
study to become skilled midwives.127
Poor study at eight sites in Indonesia (Mukherjee, 2006) suggests that,

2.
Improving affordability of care by skilled professionals. Only
as long as no complications are expected, the traditional birth attendant
48 percent of women in the poorest quintile have assistance form a
(TBA) is the favored care provider among the poor. The TBA remains the
qualified midwife or doctor during pregnancy and childbirth, while 90
principal choice of childbirth assistance provider more so according to
The average fee for a
women than men. Although the most obvious reason seems to be the
midwife-assisted delivery is Rp 200,000, close to monthly per capita
cost savings of using a TBA compared with midwives/Puskesmas, several
consumption in the poorest quintile. One way to address the problem of
other factors favor the TBA. In five out of eight sites, the satisfaction ratings
affordability is through providing targeted price subsidies for poor women
of the TBA were higher than those of local clinics or midwives.
percent of women in the richest quintile do.
128
to access skilled healthcare services. To date, there have been two

A comprehensive approach is needed to address these issues and one
fundamentally different approaches. The first is the national health-card
program, called Siaga129 (Be Alert) and initiated in 1998, has demonstrated
scheme, which was set up in response to the financial crisis of 1997 and
good results in addressing underlying issues of empowerment and
has more recently been funded largely from savings on fuel subsidies.
community structures to reduce maternal mortality in Nusa Tenggara.130
Poor households are provided with health cards that entitle household
This community-driven approach has led both to practical changes in
members to free basic health services, including access to midwives, by
behavior and measurable results (Sood, Chandra et al, 2004). A recent
public providers. Providers are directly compensated for the services
evaluation by Maternal and Neonatal Health found that Siaga villages had
they deliver to health-card holders. However, the impact of the health-
measurably higher levels of reported assistance to women during childbirth
card program on increasing access to skilled delivery appears to be
than non-Siaga villages, as well as an increased awareness among women
minimal. The other approach is an experimental scheme, the Targeted
that bleeding was a danger-sign during childbirth. Siaga villagers have
Performance Contracting (TPC) program, which was launched in 10
also taken practical steps to improve maternal health, including reviewing
districts in Central and East Java, but was later abandoned when financial
maternal deaths with local health providers in order to evaluate what could
support came to an end. Booklets containing pre-paid vouchers for free
have been done better, creating rapid response vehicles to rush pregnant
midwifery services were distributed to poor pregnant women. TPC-
women to hospital, and villagers working together to improve the quality
contracted midwives received a basic monthly salary, but with an option
of local water and sanitation facilities.
of topping up their salaries based on number of services delivered.

The government should consider greater use of demand-side financing
to help the poor overcome the financial barriers preventing them from
accessing skilled providers. Greater impact might be possible if funding
were channeled to the poor through alternative mechanisms (rather than
through providers). One possibility is to develop contracts with
communities, or cash transfers to poor households, both conditional on
4.
Improving quality of skilled birth attendant services: Besides

increasing access to professionally assisted deliveries, government policy
should focus on improving the quality of services delivered. Low quality
service providers harm poor women disproportionately, as these women
have the least choice in selecting providers. Most pregnant women,
especially poor pregnant women, seek maternal health services in their
127
Direct training of traditional birth attendances has been an approach abandoned in the nineties as studies have found the impact of trained TBAs on maternal mortality is low and could be counterproductive
by reducing resources available for the training of medium level providers such as midwives. Although Indonesia specific evidence is lacking it is now widely recognized that the success of TBA training
programs is limited. Bergstrom and Goodburn (2001).
128
Authors calculation based on Susenas 2003.
129
Originally funded UNFPA and more recently folded into USAID Maternal and Neonatal Health Program. It is currently implemented in 50 villages in Nusa Tenggara province.
130
Siaga was first implemented in 1999 in East Java, South Sumatra and South Sulawesi.
155
villages, but the quality of services provided by village midwives is often
majority of deaths caused by hemorrhaging are reported to be due to retained
of low quality due to poor training (Parker and Roestam, 2002). Reasons
placenta.132 A comparative effectiveness trial conducted by the WHO
for this include an overly theoretical curriculum, the lack of follow-up
Collaborating Center in Bandung and JHPIEGOs Maternal and Health Program
supervision by more experienced midwives and, more recently, uneven
in Indonesia assessed the impact of the drug. The study concluded that a
quality of training as a result of decentralization. For instance, many
community-based approach using Misoprosol is a safe, acceptable, feasible
districts have opened new midwifery schools but lack experienced teachers
and effective approach where skilled care is not available (Maternal and
to train staff.
Neonatal, 2004). Women in the intervention area were 24 percent less likely to
suffer excessive bleeding and 31 percent less likely to need emergency referral.
Support and supervision by more experienced midwives, as well as peer review
However, the drug can cause abortions if taken during pregnancy. In view of
and support, can reinforce theoretical knowledge and improve midwifery skills
the nature of the drug and the need to train community volunteers and regional
by imparting practical knowledge. This finding is supported by an analysis of
coordinators who would be responsible for supplying Misoprostol, the drugs
communication training (APN basic obstetric training) for district health center
use should be phased in over time (Sanghvi et al, 2004).
midwives: while formal training resulted in new midwives increasingly using

newly acquired skills, training coupled with follow-up self-assessment and
peer reviews helped maintain these skills (Abdallah, 2002). In one recent
evaluation, the skills of about 70 percent of new midwives declined three months
after they had completed their training if they received little or no peer review
or support (Abdallah, 2002).
Promising approaches include a stronger focus on practical skills, supervision
and peer review, as well as accreditation. Both UNICEF and Project Concern,
in cooperation with provincial governments, have promoted more rigorous
peer review and support systems, with the longer-running Project Concern
approach demonstrating that peer reviews can result in improved skills
(Robinson, Burkhalter et al, 2001). The newest approach to this effort is the
Bidan Delima program, initiated in 2003 in 160 districts to promote higherquality private midwives. The program, supported by the Indonesian Midwife
Association, enables midwives to conduct self-assessment, and then to study
to improve their skills. Finally, when they reach an objective standard, they are
certified as high-quality providers. Midwives are strongly motivated by this
opportunity because certification enables them to garner more clients.
Quality will also improve with the availability of medical equipment and drugs
to support deliveries. For this reason, increasing the availability and use of
well equipped clinics for deliveries is needed. For instance, in remote areas,
the government can reduce the risk of maternal mortality by increasing access
to medicines such as Misoprostol. Misoprostol131 is a prostaglandin, causing
muscles to contract and thereby speeding up the release of the placenta. The
131
Misoprostol is a second best to Oxytocin, which needs to be injected by a skilled provider and requires refrigerated storage to maintain its effectiveness (Maternal and Neonatal, 2004). Misoprostol is not
only much cheaper (around Rp 5,000 to Rp 10,000 per pregnancy), but can be administered in pill form and does not require careful storage or a skilled provider. Provision of Misoprostol for all births in one
year would cost about US$4.5 million, equivalent to a 0.26 percent increase of total public health expenditure in 2003. Currently, the drug is only available in a limited number of provinces, including West Java,
Banten, Lampung and South Sumatra.
132
Ministry of Health (2001).
156
Focus on Child Malnutrition133
Despite reductions in Indonesia s poverty rate and economic
from 1998 to 2003 have been associated with considerable increases in the
recovery in recent years, in the period 2000-03 there was no
child malnutrition, as parents could ill afford to maintain their children s usual
improvement in the child malnutrition rate. The Nutrition and
caloric intake. While the government needs to implement policies to mitigate
Surveillance Survey (NSS) provides continuous data on trends in selected
pressures that are likely to drive up the price of staple foods, such as lifting the
urban and rural areas.134 The NSS shows no change in trends after 2000,
import ban on rice, it can also fortify foods with multi-micronutrients and
giving the impression that malnutrition has been stagnant since then. Child
subsidize iodized salt (see Box 3.5 in Chapter 3 on Understanding Poverty for
malnutrition, defined in terms of weight-for-age for children under the age of
more details on rice).
five, has even increased slightly, mostly due to an increase in malnutrition in

urban areas. For several decades before 2000, child malnutrition in Indonesia
Contrary to common perceptions, the decline in utilization of
had declined steadily and, although it remained substantial, was successfully
Posyandu units does not have a statistically significant effect on
brought down to levels close to those in other countries in the region. In
malnutrition. In fact, the increase in malnutrition levels took place in 2003,
1999, for instance, child malnutrition was 25.9 percent in Indonesia, while
at a time when there was a higher availability of Posyandu units per 1,000 of
it was 18 percent in Malaysia, 28 percent in the Philippines and 33.8 percent
population. Any effect from the increase in utilization and coverage of Posyandu
in Vietnam. The reasons behind the stagnation in improving malnutrition levels
centers in reversing the downward trend of child malnutrition appears negligible
are three-fold:
and, in some areas, even irrelevant. First, the closure of Posyandu centers
135
postdates the rise in child malnutrition. Second, some areas have high Posyandu
First, changing infant feeding practices and the use of infant food
utilization rates while also exhibiting high child malnutrition. In Nusa Tenggara,
supplements as substitutes for breastfeeding during the first year
for example, an increased utilization of Posyandu centers has had no effect on
of life have a strong negative effect on nutrition. The use of
child malnutrition rates, suggesting that any observed effect in other provinces
supplements increased over the period from 1998 to 2003 by almost 10 percent.
may be specific to that province. Nonetheless, while the decline of Posyandu
Indonesian mothers are shifting from exclusive breastfeeding during the first
centers appears to have had little measurable effect on child malnutrition, the
six months to greater use of supplementary food and drink products. From
reassertion of a well-implemented health campaign that vastly improves
1999 to 2003, exclusive breastfeeding declined from 54 to 43 percent in rural
awareness in communities of good complementary infant feeding practices, as
households, and from 51 to 21 percent in urban households (De Pee et al,
well as the benefits of exclusive breastfeeding, will be beneficial in reducing
2002). Concurrently, use of supplementary food and drink products increased
malnutrition going forward.
from 65 percent in 1999 to 74 percent in 2002 (De Pee et al, 2002). Regression
results suggest that breastfeeding at some point during an infant s life is
associated with lower levels of child malnutrition. Relative to exclusive
breastfeeding, the use of supplementary food and drink results in significantly
lower values of weight-for-age (higher malnutrition levels), after controlling
for other individual factors such as gender and age, and household
characteristics. This finding suggests that the government should invest in
putting in place information campaigns that relay the clear benefits of
breastfeeding for child nutrition.
Second, sharp increases in the price of staple foods, in particular
rice, had an adverse effect on child malnutrition that reverberated
in both rural and urban households. Significant increases in rice prices
133
Adapted from Maria Abreu (2005).

The NSS covers 40,000 households in nine rural provinces including Banten, West Java, Central Java, East Java, West Sumatra, Lampung, Lombok, and South Sulawesi, and urban poor targeted in slum
areas in Jakarta, Makassar, Semarang, and Surabaya.
135
Defined as the proportion of children under the age of five who fall minus two standard deviations from the median weight-for-age of the reference population.
134
157
IV
Infrastructure Sectors
Overview
Infrastructure development matters both for income and non-income dimensions of poverty. Investment
in infrastructure services contributes to growth by reducing transaction costs and facilitating trade flows within and
across boundaries. Infrastructure investment can enable economic actors to respond to new types of demand in different
places, lowers the costs of inputs used in the production of almost all goods and services, opens up new economic
opportunities and makes business more profitable. However, going beyond its impact on growth and the income dimensions
of poverty, it is also important to understand the synergies between infrastructure and non-income dimensions of poverty.
Global estimates suggest that differences in access to safe water could explain about 25 percent of the difference in infant
mortality rates between the poorest and richest quintiles, and 37 percent of the difference in child mortality.136 In Indonesia,
diarrhoeal disease is the second largest cause of mortality among children. One half of all child deaths are associated with
water-borne, faecal-borne or respiratory disease a result of inadequate water and sanitation services, drainage and
detrimental fuels.
Public investment in infrastructure in Indonesia has fallen in the past decade and has not recovered to
pre-crisis levels. In the growth period of the late 1980s and early 1990s, infrastructure was allocated the lion s share of
central government revenues and looked set to continue as the primary and most favored sector for development spending.
In 1994, development spending in infrastructure sub-sectors137 was close to 2.3 percent of GDP, when three-quarters of
this spending came from central government development spending (Figure 5.14). After the 1994 fiscal adjustment,
policies encouraging more private-sector investment, together with decision-making processes that allocated expenditure
away from infrastructure resulted in a declining trend in infrastructure sector investment. The economic collapse in 1998
exacerbated this situation as the government tightened expenditure and focused on stabilizing the economy.
All infrastructure sectors have suffered as a result and regional competitiveness has been lost. Power
outages are imminent in Java, while the outer regions already suffer regular black-outs. Limitations to internet and data
access are affecting business. Road congestion has increased and is pushing up costs for exporters. Maintenance of
existing roads is neglected: almost half the district roads are in poor or bad condition, leading to increases in transport
costs and lower prices for farm produce. The water and sanitation sectors are in crisis, registering some of the lowest
access rates in the region and undermining health outcomes.
This downturn in spending on infrastructure is markedly different from spending patterns found elsewhere
in the region. Data comparing gross fixed capital investment across a range of East Asian countries show that while
Indonesia was investing close to 30 percent of GDP in capital investments in 1990 (ranking just below Thailand and
Malaysia in the region), by 2004 Indonesia s capital investments were lower at around 21 percent and even Cambodia,
India, Bangladesh, Sri Lanka and China had higher levels of fixed capital investment as a percentage of GDP (WDI). This
has serious implications for future economic growth. Several business surveys have identified poor infrastructure as a
key barrier to investment in Indonesia and the country now ranks below most of its neighbors on key infrastructure
indicators. By 2002, China for instance, previously behind Indonesia in competitiveness rankings had overtaken Indonesia
in terms of quality of infrastructure.
136
Child mortality refers to child mortality risk and is defined as the probability of dying before the age of five and expressed as deaths per 1,000 live births. Infant mortality refers to the probability of dying
between birth and the exact age of one and is expressed as deaths per 1,000 live births.
137
Development spending on infrastructure as described in this chapter includes water and sanitation, irrigation, rural roads and electricity sectors.
158
There are signs that the tide is slowly turning, but the levels of spending needed to revive the sector
will take years to achieve and need to be better targeted. The gradual upward trend that is now emerging in
infrastructure spending continues to be biased towards growth-oriented investment and an infrastructure gap is evident in
household-level services. The primary challenge for Indonesia is in the selection (and implementation) of a pro-poor
development spending mix.
Figure 5.14
Development spending on infrastructure (% GDP)
Figure 5.15
Source: MoF, SIKD data.

Note: Includes only water and sanitation, irrigation, electricity and roads sectors. See Annex V.4 for details.
Indonesia is already lagging behind China in the

commercial perception of quality in infrastructure
services (ranking: 1 worst, 7 best)
Source: Estache and Goicoechea, 2005.
Water and sanitation services sector

Levels of spending on water and sanitation services
Despite the linkages to poverty reduction and human development, government expenditure on water
and sanitation is low compared with other sectors, and other comparable countries. Overall spending on
water and sanitation services (WSS) in 2002 amounted to Rp 2.57 trillion, equivalent to only 0.14 percent of GDP.138
Figure 5.16 provides an illustration of the trends in the water sector over the past decade, including the decline in total
expenditure as a percentage of GDP.
Decentralization has not created significant change in aggregate terms, but there is inconsistency in
the proportions of central, provincial and district government WSS spending in different regions. This
indicates differing levels of effective fiscal decentralization in the sector. Whereas in Java/Bali, 60 percent of WSS expenditure
was central government spending, in Sumatra over 60 percent was district spending, while in Papua over 70 percent was
provincial spending. This lack of consistency in sector spending levels can be explained by project spending.
Access levels to water services are low and self-provision characterizes the sector. As of 2004, WDI
figures indicate that 77 percent of the population of Indonesia had access to what was termed an improved water source .
Official Susenas data categorize access to water according to ownership: whether ownership is private or public, individual
or shared. These data indicate that of those households that have access, 52 percent have individual privately-owned
sources of drinking water (e.g. pump or well), and 25 percent have a jointly-owned source (e.g. communal taps). Most of
the water consumed in Indonesia is not delivered. Instead, the majority of Indonesians obtain water by drawing on the
ground water table, either by accessing water on their own land or in the locality. In urban areas, only 33 percent of
population live in households with access to piped water and, of the poorest quintile, only 18.3 percent have access to tap
138
Infrastructure development spending data at the sub-sector level (including water and sanitation spending data) are only available up to FY2002, since after this year the classification of certain expenditure
items for the sector have been changed. In order to come up with figures that are comparable over time expenditure data need to be reclassified according to the pre-2002 system.
159
water (Susenas, 2004). Where possible people in Indonesia rely on water collected from rainwater, lakes, rivers, and
hand-pumped wells, springs or communal wells, and these people are probably not using enough water for basic hygiene
standards to be met (see Annex V.5 for details on access to water for the poor).
The real problem the type of supply and the associated convenience, cost and quality of water is
hidden in access data. Access figures suggest Indonesia is on par with other countries in terms of water service
delivery, but most of the water consumed in Indonesia is not delivered. The high rates of self-provisioning, mostly at the
individual but also at the community level, make Indonesia a country with one of the highest rates of self-supplied water.
Water-supply strategies must be extremely cognizant that: (i) this self-provisioning is hard to convert to utility supply
once in place and functioning effectively at the individual household level; (ii) there is a huge variation in potential
sustainability, which is very dependent on location; and (iii) self-supply is probably a water-supply lifeline in the short
term. Few water sector strategies have recognized the importance of self-supply, or worked out what the implications are
for devising a water-supply strategy.
Access to sanitation services in terms of piped sewage disposal is less than 1 percent of the total
population in Indonesia. This is the lowest level in the region and among comparable countries.139 The
lack of sanitation in Indonesia is longstanding, and has resulted in low demand and low expectations. There is also low
awareness of the adverse impacts of inadequacy and the potentially huge benefits of improving sanitation systems. In
Indonesia, policy dictates that basic sanitation is the responsibility of private households, even though the costs of
inadequate sanitation are substantial. The health and environmental costs of inadequate sanitation in cities are especially
marked. Any improvement in this situation will require far higher public awareness of the negative impacts of inadequate
sanitation on health, economic growth and the environment, and political will to address the problems (see Chapter 3 on
Understanding Poverty for more details on access levels).
Priority issues for poverty reduction in water and sanitation services

The supply of water to the urban poor is complex. The existing situation reveals significant variation across cities
and suggests a mix of solutions for the future. In urban areas, levels of access to utility supply are lower in the poorest
quintile but the district water utility (Perusahaan Daerah Air Minum, or PDAM) service is also limited to all households.
This analysis, together with analysis of tariff and connection fees, suggests that if there is a piped service available it is
likely that the poor can access it. In practice, however, the urban poor obtain their water from a number of different
suppliers. Collignon has documented by volume and value the proportions of network and non-network water consumed
in poor areas surveyed in selected cities. This shows not only the high reliance on self-supply in all cities except Jakarta,
but the predominance of water terminals in Palembang, private networks in Bandung, both public and private networks in
Makassar, and a mixed pattern in Subang (Hydroconseil, 2004). It is highly likely that the household
decision on connection is closely linked to alternative supply options available in the locality.
Poor households connected to utility networks pay subsidized fees and tariffs, but
only a small percentage of households are connected (see Annex V.5). Subsidies are
provided through a rising block tariff (RBT) and are therefore not targeted as well as they might
be. (The rising block tariff exclusively benefits those households that are connected, disadvantages
poor households that often share connections or have large consumption, and provides the benefit
of subsidized water in the first block to all consumers.) High connection fees should be discouraged
for pro-poor expansion of services even though utilities currently insist on applying these fees as
they increase instant revenue.140
139
140
160
While sewerage system coverage is very low, community-based sanitation models do exist in Indonesia although supporting data on their status are lacking.
The government is currently working on a regulation for revising the tariff structure.
Targeting and subsidy mechanisms are flawed and need to be rectified to increase access for the poor.
Pro-poor expenditure basically means making affordable services available to the poor. A variety of subsidy policy options
should be considered to help increase access of these services to the poor. Tariffs and subsidies need to be better targeted
to ensure that they are not captured by the non-poor. Appropriate pricing is the most controversial aspect of sector reform.
The restructuring of sector finances necessary to create viable service delivery institutions is likely to have some negative
impact on the poor in the short term. Mechanisms need to be put in place to specifically target those who need the
services most. Past subsidization through the rising block tariff has benefited poor and non-poor alike, creating inefficiencies
in institutional performance and hindering the capacity of PDAMs to expand coverage to include more poor areas. As
things stand, the low level of household connection means that the tariff invariably subsidizes the non-poor.
The rising block tariff is, despite the best intentions, a regressive tariff that needs to be replaced and/or supplemented
with other instruments. Experience suggests that the most effective means of targeting the poor will be highly dependent
on local conditions, especially where supply options vary (McGranahan and Satterthwaite, 2004).
A number of alternative, non-quantity-based subsidies, such as geographic targeting (where the poor are
spatially clustered), means-testing or even (temporary) lower service-level subsidies (e.g. promoting access to
standpipes) that also encourage incremental service improvement, might be considered in different situations.
Connection subsidies would need to be very carefully considered in viable contexts and combined with
complementary non-price mechanisms (such as security of tenure) to make utility services accessible and affordable
to poor households.
In both urban and rural contexts, low sanitation coverage has significant health and environmental
(and therefore economic) impacts, but demand for sanitation remains weak. Sanitation is a public good,
cost recovery is difficult to establish at the outset and in Indonesia there is little demand/pressure coming from consumers.
The sanitation component of the WSS sector needs detailed and separate consideration if new developments are to reach
the poor. In the absence of such careful forethought, and given the very low current levels of sanitation service provision,
direct improvements are likely to be captured by the non-poor for many years to come.
Recommendations for pro-poor spending on water and sanitation services

The linkage between inadequate water and sanitation services and poor human development outcomes
is clear and public spending in water and sanitation is widely proven to be poverty-reducing. But propoor expenditure in water and sanitation services requires understanding of blending, targeting, community-driven
development, multi-stakeholder delivery and viable financing. The following recommendations aim to achieve this and
also suggest ways of making spending on water and sanitation services more efficient and effective:
1. Blend water and sanitation expenditure with health and education spending to provide a mix of
services specifically designed for poor areas/quintiles. A mix of services would provide synergies and
cumulative benefits, and more so for the poorest quintiles. A targeted strategy is needed for higher spending on WSS
in districts and provinces where under-five mortality is high. This should then be blended with a package of better
preventive healthcare, maternal education, and efforts to monitor the progress of the lowest quintile in these targeted
areas to achieve optimal pro-poor WSS expenditure.
2. Design and allocate resources and programs for WSS at the regional level to respond directly to
local and institutional constraints and variations, other supply options and household needs.
a.
Account for household and regional needs: map water and sanitation access and type against infant mortality
rates, and then target water and sanitation DAKs to needy areas.
161
b.
Account for supply options: the vast range of services currently available and the very different profile of supply
options across the country suggest that a one-size-fits-all solution is unlikely to be appropriate, particularly in
urban and peri-urban areas.
c.
Account for existing services: the high level of self-provisioning of WSS in Indonesia is a dilemma that needs to
be taken into account in local level water and sanitation strategies.
d.
Account for the capacity of the delivery agencies: allocate to the stakeholder (LG, NGO, CBO) most able to
deliver. Irrespective of context, awareness of the capacity of the delivery agency to delivery WSS and specifically
to deliver WSS to the poor is a critical factor in the design and allocation of resources.
3. In terms of water provision in rural areas, establish a time-bound national action plan and budget,
linked to the MDGs to replicate the community-management supply model to the 50 million people
without adequate water supply (see Box 5.2). Household provision is currently estimated at 55-60 percent and
community level provision (village-level managed and operated systems) has increased to 25-30 percent, while
PDAMs and small-scale providers supply an estimated 5-8 percent of households at the rural-urban interface (World
Bank, 2004a). The predominance of rural self-provision and a country-wide recognition of the success of rural
community water-supply initiatives mean that an approach for rural areas is at least already known. Policy has
developed over the past decade to build on these opportunities, and implementation and roll-out is planned, but
greater government spending is needed in order to achieve a broader impact. Community-based solutions can also
benefit from the lessons emerging elsewhere. More attention is needed to the development of sustainable management
systems, alternative technologies, cost recovery, economies of scale and a delivery system able to improve the rate
and scale of progress of service improvements. The financing of policy into an implementation framework should be
identified and incorporated into national budgets. Efforts to secure financing and plan the roll-out of successful
initiatives to lagging regions are urgently needed if the problem of rural water is to be tackled.
Box 5.2 Local ownership in Lumajang
The Second Water and Sanitation for Low-Income Communities Project (WSLIC-2) started in Indonesia in 2000. The project aims to improve the health
status, productivity and quality of life of poor rural communities. Project activities focus on improving health and health behavior, providing better access
to water and sanitation, and community participation. In Lumajang district, East Java, where a WSLIC-2 project began in 2002, 75 percent of targeted
villagers in villages visited by the research team now have access to project water supply (although the tank-to-household ratio of between 25 and 43 to
1 is rather high). Distances traveled by water collectors have shortened and water quality is better.
WSLIC-2 pays up to Rp 200 million for a village water supply system; for larger villages with higher project costs, the consent of the Management Unit
is required. Villagers are required to contribute 20 percent of the total (4 percent must be in cash and 16 percent in kind and labor). Because of Lumajang s
geography, only piped water systems-the most expensive type-are possible, so the project has paid the maximum Rp 200 million for each village system
in this district. Each village has therefore contributed Rp 8 million (US$900) in cash and Rp 32 million (US$3,600) in kind and labor, demonstrating
ownership of the systems from the beginning of construction.
Villagers in one site said: We are happy and proud of [the water system]; we celebrated the birth of the water system on the evening of the fifth of May with
bull-races, Together with the facilitators we have created the water infrastructure, and, WSLIC-2 is the first time the people have been trusted. The
head of one village explained that WSLIC-2 woke up sleeping people; it gave them the incentive to... take matters into their own hands.
Source: Innovations in Pro-Poor Service Delivery: Nine Case Studies from Indonesia.
162
4. In terms of water provision in urban areas, strengthen capacity and establish incentive structures
for PDAMs for planning, delivery and monitoring of service delivery. The failure of water institutions
generally, and PDAMs specifically, has been a serious blockage hindering sector development, but signs of success
in a number of PDAMs over recent years provide models and hope for change. The chronic inefficiency of PDAMs
the interference of local government owners, the payment of dividends annually to local governments irrespective of
profit, the need for expansion, compromised hiring practices, the lack of transparency, and inappropriate connection
charges and tariff structures affects the poor and non-poor alike. A pro-poor effort is needed to support or supplement
systemic reforms (World Bank, 2004). But PDAMs need to be mandated, and given incentives to scale up services to
marginal areas inhabited by the poor. The low utility coverage requires effort by utilities to institutionalize pro-poor
approaches, reconsider service levels, resolve the issue of subsidies and the pricing of water, develop pro-poor
financing strategies and affordable cost recovery solutions, and target the needs of the poorest households in the
development of solutions. A pro-poor strategy needs to accept that at present the vast majority of the poor do not
have access to network water. Changing tariff structures may indirectly support these groups, but other instruments
will need to be used if the poor are to be served with adequate and affordable water. While connecting poor households
to utility water at an affordable cost is undoubtedly optimal, not all improvement strategies will focus on connections
in the foreseeable future. And for those currently connected, or targeted for future connection, a number of instruments
(appropriate billing and connection, service options, poverty selection) are needed to ensure sustainability of service.
5. Sanitation for the poor requires increasing public awareness and political will. Stronger institutions
and increased financing will not come about without stronger political will and far greater public awareness of the
adverse impacts of inadequate sanitation on health, the environment and economic growth. The lack of sanitation in
Indonesia over many decades and political eras has resulted in low demand, low expectation and low awareness of
both the impacts of inadequacy and the potential impacts of improvement. This low demand from constituents has
meant weak political support for sanitation. Expenditure on hygiene and sanitation awareness-building might be
stimulated in the private sector if government used public finance to launch and leverage the effort, and provided
strong political leadership to a sanitation clean-up campaign.
Figure 5.16
Development spending on infrastructure by sub-sector (% GDP)

Road Infrastructure
Water and Sanitation
Source: MoF and SIKD. Annex V.4 for details on spending levels in nominal rupiah terms.
163
Rural roads sector

Levels of spending and access to rural roads
Roads and transport contribute to both growth and poverty reduction. This is because roads and transport
significantly affect the remoteness of rural households. It therefore comes as no surprise that investment in roads has an
important impact on income and non-income dimensions of poverty. Better transport increases the efficiency of resource
allocation, improving the performance of markets and fostering economic growth, and economic efficiency lowers costs
and enhances economic opportunities (see Chapter 4, Section VI on Connecting the rural poor to urban markets). In terms
of non-income poverty dimensions, improved roads and transportation have a positive impact by enhancing access to
health and education services, while in emergency situations access to roads can even be a defining factor in reducing
maternal mortality.
Roads and transport significantly affect the remoteness of rural households and lack of road access is
highly correlated with poverty. Poverty diagnostics show that having access to an asphalt road is one of the key
variables associated with increases in household expenditure in Indonesia (see Chapter 4 on Growth). Controlling for
other household and regional characteristics, having access to an asphalt road is associated with an 8 percent increase in
household expenditure in urban areas of Indonesia.141 But it is not just distance that matters. The quality of road matters
greatly too. Better quality roads, as measured by the proxy of a higher average speed traveled between village and provincial
capital, raise real per capita expenditure, with an elasticity of 0.06.
Road infrastructure expenditure and access reflect overall infrastructure trends: a dramatic increase in the
1980s and early 1990s and then a dramatic decline since 1994. Road sector development has been marked by highs and lows
for a 30-year period. The length of the road network in Indonesia increased by an average 8.3 percent annually between 1970
and 1998, and included a surge of construction of local farm-to-market roads (from 8,500km in 1977 to 31,900km in 1998).
In 1990, the total road network was 1.8km per 1,000 people, and today still remains higher in these terms than China and
Vietnam. The roads that were constructed widely throughout the country prior to the crisis supported the growing economy,
particularly the agricultural and manufacturing sectors. While the allocation was not optimal and investment in urban areas
still dominated, the absolute levels of spending meant that vast road networks were provided in rural areas as well. By 1993,
the overall coverage of the network was only slightly below the regional Southeast Asia average density in relation to population
(and arable land area) and 92 percent of the population had access to passable roads all year round.
Spending levels are comparatively low and unpredictable, and a lack of maintenance is the primary
problem characterizing the sector. The decline in real investment following 1994 was stark: over the seven-year
period from 1994 to 2000, spending halved as a percentage of the national development budget from a sound 1.47
percent of GDP in 1994, to only 0.48 percent in 2000 (see Figure 5.16). By 2002, and despite widespread agreement in
the government and donor community that the roads sector was vital to economic growth and poverty reduction, this
figure was still only 0.69 percent of GDP. Unfortunately, it is currently not possible to retrieve spending data after 2002 at
the sub-sector level due to a reclassification of infrastructure spending data at the decentralized level.
Spending in recent years has been biased towards the national networks and has not met development,
improvement and preservation needs. While the national arterial road network is in good condition, with 95 percent
paved and 88 percent in good-to-fair condition (in 2003) and the provincial network is also predominately good-to-fair,
the district and urban road networks are of much poorer quality: 44 percent are paved and 50 percent are in poor-to-bad
condition (World Bank, 2004b). While budgets for national-level roads are sufficient, as one moves out from the center
141
164
See household expenditure function regressions in Annex of Chapter 3 on Understanding Poverty.
budgets and expenditure become increasingly insufficient to cover the costs of maintaining roads. The result is that
district and village roads (i.e. those roads the rural poor use most) are deteriorating badly, some are dysfunctional and
urgent efforts are needed for their rehabilitation.
Decentralization has created disincentives for the allocation of resources for road investments or
maintenance with benefits that go beyond the kabupaten level. It is clear provincial governments have very
little say over the flow of funds regarding the construction or maintenance of kabupaten and village roads. Meanwhile,
national funds have not been used to ensure that optimal investment decisions are made not just at the kabupaten but also
the regional and national levels (in the case of roads near kabupaten borders, for example, where much of the benefit of
construction or maintenance may flow to a neighboring authority). While the DAK is in part designed to provide resources
precisely to support investments with spillover benefits, in practice it has rarely been used to encourage such projects
(and may, regardless, be too small to achieve such objectives).142
Priority issues for poverty reduction in rural roads

The roads sector is in need of increased and more predictable investment. It is not only the decline in
recent years that has created a problem for the sector. The high levels of expenditure in the 1980s and early
1990s created a large stock of road assets that then needed routine and periodic maintenance. This rapid rise in investment
and asset-building in the sector followed by a steep fall in expenditure has meant that the sustainability of previous
investments is now in question.
Lack of maintenance is the primary problem characterizing the sector. By 2000, expenditure on maintenance
had reached its lowest point as a proportion of development, dropping from 30 percent to 10 percent of overall expenditure
(on national and provincial roads) over the past two decades. The
result has been a progressive decline in the quality of, and access to,
roads. The lack of maintenance of roads is a cause for significant
concern in that it has significantly reduced the quality and efficiency
of the road stock, and reduced the corresponding quality and efficiency
of services. Spending on road maintenance has three times the rate
of return as spending on new roads, but its economic and social
importance is often under-estimated. Despite the high return to
maintenance expenditure, data in Indonesia show a significant decline
in routine, periodic and emergency maintenance in rural areas
(Infrastructure PER, World Bank, 2004).
There is little focus on transport services, especially in
remote regions. Indonesia s poor are also transport poor:
their livelihoods are sensitive to the transport they access. Disparities
in access and mobility in Indonesia are very marked. The rural poor
have limited mobility beyond their immediate settlements due to
geographic isolation and the high cost of motorized transport.
Transport is a service, but it is also a significant employment
opportunity for the rural poor. Typically, transport services provision
has a low entry cost and is highly labor intensive, with a vast range of
142
Adapted from World Bank, 2006d.
165
technologies. Transport services are not only a service but one of the key non-agricultural livelihood opportunities
available to rural people. Most agree that the best government can do to support this transport marketplace is to lighten
the regulatory and taxation burden that impedes the development of competitive transport services, and keep the roads in
a condition that makes it cost effective for transport operators to provide services that utilize them.
Recommendations for pro-poor spending on rural roads143

In Indonesia, commitment to infrastructure at the policy level has not been matched with a financing
strategy commensurate with government goals. Instead, the sector is plagued by variability and unpredictability.
The sector is essential to economic growth and poverty reduction, and a strategy for making infrastructure expenditure
work for the poor must carefully consider the role of roads. Recommendations to improve the quality and accessibility of
rural roads in Indonesia are as follows:
1. Roads at the local level, including both district and village roads, should be a primary focus of
investment and increasing the scale of the DAK might be a first step towards reconstructing a
system of incentives for better road management. The funds should be more carefully conditioned so that
they act, as originally intended, as an incentive for projects with spillover benefits between districts and regions, and
also as an incentive for good planning and maintenance. Reforms to improve the quality and reach of rural roads
would center around providing additional funds to poorer district governments. Reforms would also involve efforts
to maximize returns to targeted expenditure through more efficient procedures. However, just as importantly, better
targeting of those funds is required to ensure: (i) increased expenditure on roads rather than other expenditure, (ii)
increased focus on maintenance rather than construction, and (iii) increased focus on appropriate quality rather
than highest quality .
2. Construction and maintenance of village and kecamatan roads can be financed through communitydriven approaches similar to the Kecamatan Development Project (KDP). Indonesia s communitydriven development model through KDP is internationally recognized for its success in delivering results to
communities through better infrastructure service, as well as community empowerment and decision-making (see
Box 5.3 on the KDP program). Labor-based construction provides employment to the poor. Involving the poor in the
building of infrastructure creates a win-win situation by producing assets needed for growth and productivity, while
also providing income for poor households to keep children in school (with its longer-term gain) (Timmer, 2004).
The government is already planning to scale up the KDP program to become a nationwide poverty reduction and
infrastructure program. The government has also been channeling about Rp 3.3 trillion into the village infrastructure
PKPS-BBM program as a result of fuel subsidy reallocations since late 2005, funds that are being distributed to
some 12,800 villages (see Box 5.4 on the PKPS-BBM program on village infrastructure in this chapter). In scaling
up CDD projects in rural infrastructure, there is scope for Indonesia to be more systematic in managing these
community-driven infrastructure programs and also providing technical supervision to achieve a higher quality of
construction.
3. While there is strong evidence that a participatory approach works at the village level in the case
of road construction and design, evidence is less clear that it does so at the kabupaten level. An
equivalent of the Indonesia Road Management System, which governs planning and budgeting at the national and
provincial level to ensure budgeting is adequate and well-targeted, is needed at the district level. Funds for new
investment projects in the transport sector will only flow if there is evidence that the existing road stock is being
properly maintained, and that new investments are: (i) based on verifiable (and sensible) design and selection
143
166
Adapted from Chapter 4, World Bank, 2006d.
criteria which included consideration of spillover benefits; (ii) being transparently and competitively bid out for
construction; and (iii) being constructed to a high standard based on independent quality audits carried out with
inputs from NGOs and local transport operators. An earmarked priority for investments could be the provision of
basic all-weather roads to all villages currently lacking such access.
4. Efforts are needed to link road user charges to resources for maintenance. One approach would be to
create a provincial or district fund to collect resources from road users (including license and registration payments)
and use these resources to maintain the road network under the oversight of a board involving both public and
private sector representatives. The fund could contract with private providers to ensure road maintenance. Thirdparty verification of the quality of the maintenance operation might be used by DAK administrators as part of yardstick
competition (comparison of performance between similar kabupaten) to set standards to be met in order to receive
investment funds. Local transport operators could again play an important role in providing inputs into the verification
process, as a significant customer base for the road network.
5. It is essential that efforts are made to ensure that the poor can access transport services and
benefit from improved road infrastructure. Better understanding is needed of the transport poverty of different
poor groups. The government can explore targeted subsidies in urban areas and support for non-motorized transport,
and also underpin SME transport activity in more remote areas.
167
Box 5.3 Scaling up community-driven development through the KDP program: A strategy for building rural infrastructure
while reducing poverty144
Indonesia s Kecamatan Development Project (KDP) is one of the largest community-driven development projects in the world. KDP
is a government program first started in 1998 that aimed to alleviate poverty, strengthen local government and community institutions, and improve local
governance. The KDP program supports participatory planning at the village and kecamatan levels, leading to block grant financing of Rp 500 million to
Rp 1.5 billion to each kecamatan depending on population size. Funds are used for productive economic assets, such as rural infrastructure, economic
activities, and education and health. To date, KDP has covered some 34,200 villages, equivalent to 48 percent of the poorest villages in the country. The
KDP program has achieved significant successes to date:
First, community-driven use of funding ensures high rates of return on investment, as project design is adapted to address the most
binding constraints faced by local communities. KDP projects have improved the poors access to markets, town centers, education and health
facilities, as well as providing clean water supplies and sanitation. KDP has funded some 104,000 infrastructure, economic and social activities across the
country since 1998, including new bridges, reconstructed irrigation systems, and clean water supply and sanitation units. The program has also contributed
to the construction and renovation of schools and community health clinics, as well as the provision of school equipment and targeted scholarships for poor
students. Torrens (2005) calculated the economic internal rate of return for four provinces covering both KDP cycle 1 and 2, and estimated expected internal
rate of return (EIRR) at 38.6 percent (irrigation), 51.8 percent (road) and 67.6 percent (water supply). These high benefits mostly resulted from tapping
suppressed or latent production capacity. For instance, new road infrastructure made it possible to access previously isolated villages where, prior to KDP, all
produce had to be hand carried or carried in small bundles by motorcycle. Similarly, irrigation projects doubled agricultural production in some areas, by
channeling water from local springs during the dry season. Water projects also helped to free community women from the time-consuming burden of water
collection.
Second, the program is successful in the targeting of poor kecamatan and has had demonstrable impact on poverty. Alatas (2005)
evaluated the targeting performance of the program at the kecamatan level, as well as at the household level, and found that the geographic targeting of
KDP was pro-poor: the incidence of poverty was 24.7 percent in KDP kecamatan vs. 13.93 percent in non-KDP kecamatan in the first stage of the program.
However, as Alatas notes, many of the poor also live outside of the areas covered by the KDP. By comparing KDP targeting with simulations from poverty
maps recently developed by BPS, Alatas found that by using poverty maps the poverty targeting of the program could be further improved to cover 52
percent of the poor in Indonesia instead of the 30 percent of the poor covered by KDP I kecamatan (Alatas, et al, 2006). The poverty reduction impact of
the program was significant according to the same study: by 2003, rural households residing in the cycle 1 KDP kecamatan were able to increase their
annual expenditure by 5.0 percent of initial mean expenditure in 1998. For cycle 2 the impact was around 3.1 percent of the mean expenditure in 1998.
The estimated annual impact was larger for kecamatan that received KDP over a longer period of time.
Third, the program was cost-effective, generated employment opportunities locally and suffered only low levels of corruption. For
instance, the costs for infrastructure projects were estimated to be 20 to 25 percent less than other government-sponsored public
works of the same quality. Labor-based construction is not only cost-effective but also provides employment to the poor. As a result of KDP projects,
37 million workdays of employment were generated in labor-intensive infrastructure projects. The application of labor-based approaches to rural road
construction contributes substantially to their poverty alleviating impact (Lebo and Schelling, 2001). Utilizing local labor not only allows the local
community to earn wages and transfers skills, but it also creates ownership and empowers communities.
Building on the knowledge base of this program, Indonesia now has a unique opportunity to launch an effective national poverty
program one that will have immediate results on the ground. In developing this process, it is important to build upon and expand the
successful KDP model. Establishing a strong policy and management structure to guide the program, and factoring in costs/time for program facilitation
and socialization in new kecamatan, will be extremely important if such a program is to succeed. Also, securing financing at the start for technical
assistance to projects, and monitoring and evaluating programs (at around 20 percent of project costs) are crucial to ensure program effectiveness and
continuity.
144
168
Adapted from Strategy Note for Government of Indonesia: Expansion of a National Poverty Reduction Program. World Bank. May, 2006.
Box 5.4 Keeping the C in community-driven development: Concerns with the PKPS-BBM Village
Infrastructure Program
The PKPS-BBM Village Infrastructure Program was developed using savings accruing from the reduction in fuel subsidies in
March 2005. The program aimed to generate labor-intensive employment and improve village infrastructure through a bottom-up decision-making
process. The program was designed with similar characteristics to other community-driven development (CDD) programs. The government financed
direct grants to villages with an annual program budget of about US$345 million distributed to some 12,800 villages (at Rp 250 million per village)
through the Ministry of Public Works (PU). Prior to deciding on the longer-term future of the program, the government commissioned a rapid
assessment study to understand whether the program was being successfully implemented and had the potential to achieve the objectives outlined.
The preliminary findings of the rapid assessments revealed many problems with the implementation of the program due to (i) the time pressure for
disbursement of funding, (ii) the lack of initial socialization of the program, and (iii) poor quality of project outputs.
First, the overall timeframe was extremely limited as the instructions were such that the funds should all be expended by the
end of 2006. Consequently, there was insufficient time for proper and thorough consultation and socialization of the program with communities,
and planning for much needed infrastructure. Preliminary results from the field suggest that the program was implemented far more slowly than
initially planned. About half of all sites randomly selected for assessment had not yet completed their projects. Issues that contributed to the delays
included that the program was started at the beginning of the wet season when it is difficult (or at times impossible) to build infrastructure, and that
the technical facilitators were deployed to the projects late.
Second, the socialization of the program was limited and details on implementation and community involvement were not
clear to participating villagers. Sixty-seven percent of villagers interviewed knew that their village would receive funding from the fuel-subsidy
compensation scheme, and 56 percent knew that the community should be involved in the project. But very few respondents knew what the funds
should be used for or what their rights were in terms of participation in decision-making and implementation. In half of all the villages assessed, the
decision on which project to carry out was made outside of the designated community meeting. Those involved in community meetings were
generally village elites and government representatives, and very few community representatives or poor citizens involved. This failing was reflected
in low satisfaction levels with the program among community members, with a 100 percent dissatisfaction level (stated as less than satisfied or
unsatisfied ) in the location or type of infrastructure project that was selected.
Third, the quality of infrastructure produced by the program gives cause for concern. Of all the roads built using project funding, 47
percent were ranked of poor quality (unlikely to last more than one year), and for all projects, only 50 percent were ranked of good quality. Only 20
percent of all villages surveyed had established committees to maintain the infrastructure built through the program. Cost-effectiveness was also
questionable, with one-third of all projects seen as not cost effective.
Finally, some additional concerns did not show up in the community survey. The choice of using contractors rather than communities
to implement projects was a highly risky facet of project design and was a potential source of corruption. The size of the grant at Rp 250 million/
village was also considered to be too large for the type of community projects under consideration.
If the government decides to continue the implementation of this program, it is crucial that time and funding be allocated for
a process that engages communities in project selection and implementation. In order to achieve cost-effective and quality results from
such a program, it is essential to provide communities with the necessary technical assistance and involve them in determining not just the choice
of project but also decisions about whether to complete the project with community labor/resources or to contract out to third parties. (See also Box
5.3 Scaling up community-driven development: Towards a national poverty reduction program.)
169
Conclusion: Achieving Synergies through the Right Development

Spending Mix
To make public spending work for the poor, attention should be focused on those sectors and programs
that will be of most benefit to the poor, particularly where the poor experience affordability constraints
in accessing services. This chapter takes stock of levels of government spending and the access of the poor to
services, as well as constraints to improved access and affordability, in four sectors that are of particular concern to
poverty reduction: education, health, and the infrastructure sectors of water and sanitation services, and roads.
In the health and education sectors, Indonesia faces mainly secondary level problems. These problems
need to be addressed through measures that improve quality and induce demand for services. In health and education
services, Indonesia made impressive progress in the three decades prior to the economic crisis in 1998, through centrallyled expansion of public services: large-scale programs for the construction of schools and basic health facilities, and the
hiring of civil servants to staff these nationwide facilities. Evidence suggests that this approach was successful in terms of
dramatically improving access, although service quality failed to improve at the same pace as access. Today, Indonesia is
faced with the secondary level problems of improving quality
in basic services in these sectors (i.e. the quality of services
provided by primary schools and Puskesmas), and
expanding access in secondary level services, such as
improving transition rates from primary to junior secondary
schools, as well as improving the access of the poor to health
insurance and in-patient hospital care. Supply-side issues
still remain in terms of improving quality of services and
hiring qualified staff, including the management and
deployment of teachers and health personnel. In the
education sector, the low availability of secondary schools
is also a constraint in expanding services to the poor.
However, in order to fine tune the delivery of services in
these sectors, demand-side measures come into their own
at this stage of developing service delivery. Such demandside measures include interventions such as voucher
schemes, as in the case of maternity healthcare vouchers
for poor pregnant women, as well as scholarships or
conditional cash transfer (CCT) programs to improve the
access of the poor to junior secondary schooling.
Public spending in health and education has
increased following decentralization, with
substantial contributions now coming from district
governments. Overall, public spending in health and
education has increased following decentralization, with the
combination of increasing district government spending
coupled with central government spending that has still to
be reduced from pre-decentralization levels. As district
governments now cover most salaries of personnel, their
spending on health and education is dominated by these
transfers and, as a result of continued deconcentrated
170
spending by the central government, decentralization has resulted in an increase of resources flowing into these sectors.
However, particularly in the health sector, the increases in public spending have not necessarily benefited the poor. This
is because the districts have focused more attention on the supply of hospitals and secondary level healthcare services
that do not primarily benefit the poor. The deteriorating benefit incidence of public spending in the health sector is of
concern to the central government. Its response to the worry that spending is not as pro-poor after decentralization has
been to increase central development spending in the regions. While this improves the incidence of spending on the poor
in the short run, the central government risks crowding out the attention and fiscal space provided by district governments
to these services, harming the decentralization process in the medium term.
The central government needs to be sensitive in managing central development spending in these
sectors in order to synergize with the resources that come from district governments. Centrally run programs
with large visible budgets, such as the BOS and the PKPS-BBM health insurance program, run this type of risk. As a
result, they need to be carefully integrated into the vision of district governments in budgetary and service delivery terms
to avoiding being detrimental to service delivery in the long run. Matched grants from the center to the districts in the form
of DAKs can help mitigate this problem and need to be carefully considered and implemented going forward. (Further
discussion of these issues is provided in Chapter 7 on Government.)
Infrastructure spending has followed a more unfortunate pattern in the post-crisis and decentralization
period. Development spending on roads and water and sanitation services has still not recovered to pre-crisis levels.
Currently, the incentives operating in infrastructure sectors do not encourage the provision of quality infrastructure services.
In water provision, government regulatory and subsidy policies are largely designed to reduce the cost of infrastructure to
existing consumers rather than to improve access and quality. In roads, expenditure priorities are skewed in favor of new
construction of over-engineered roads rather than towards maintenance of the existing stock and construction of appropriate
new infrastructure. Spending data reveal that district governments respond less well to the need for infrastructure
development. Consequently, matched grants from the center, as well as an increase in funding for community driven
projects for infrastructure development, will both be important steps in these sectors in the near future.
The synergies and complementarities between the infrastructure and human development sectors should
be recognized, with a comprehensive strategy and budget for poverty reduction. Although the link between
infrastructure and poverty reduction goals seems more indirect, infrastructure sectors are key in attaining improved
outcomes in both the income and non-income dimensions of poverty. Striking the correct spending balance between
infrastructure sectors and human development sectors is vital in creating the necessary synergies needed to overcome
constraints in achieving the MDGs and national poverty reduction goals. The expansion of quality water supply services
and the maintenance of rural roads are particularly important for those outcomes that most impact the poor. Sanitation
services, which clearly lag in Indonesia and have detrimental impacts on outcomes such as infant and child mortality,
need to be expanded-this despite the fact that the sector faces demand-side constraints that also need addressing.
With high international oil prices and the recent reductions in domestic fuel subsidies, Indonesia stands
today at a point where it can make significant strides in poverty reduction through targeted spending in pro-poor sectors
and by combining the right mix of spending on infrastructure and human development services. This chapter makes
specific recommendations on how to improve access to services for the poor in these sectors, and on how to improve
affordability while also focusing on the quality of services already accessible to the poor. It also recognizes that the
problems in the sectors analyzed are not only problems of levels of spending. A number of structural blockages need to
be addressed in order to make public spending in these sectors work more effectively and deliver better results to the poor.
These institutional blockages regarding accountability mechanisms in service delivery and incentives to providers are
considered in more detail in Chapter 7 on Government.
171
Spotlight on Inefficiencies and Leakages
Corruption in water service delivery: direct and indirect impacts on the poor
Corruption has a direct impact on water poverty by decreasing access to, and
quality of, water assets, management and services, and by increasing costs.
Understanding the differentiated impacts on poor households and the types of
corrupt interactions they experience is essential for the development of effective
pro-poor strategies in Indonesia. When a poor householder pays a bribe or
indirectly suffers from the diversion of resources, one or more of their assets
changes.145
It is therefore critical for anti-corruption strategies to measure all effects of
corruption on the poor. Many poor citizens live and function predominately in
the informal economy, have much lower access to formal basic services, are
subject to much greater intimidation, and have less access to justice than other
citizens. Their marginalization from formal processes is therefore a key descriptor
and determining factor of their interactions. However, we do not know how the
different poor groups perceive this or what their views are on tackling corruption
and changing their service options.
The very poorest probably do not bribe because they do not have the money or
assets to do so, but their marginalization is the greatest and the impacts of poor
redistribution most marked. But what of the other poor groups nearer the
Indonesian poverty line? How does water corruption affect these groups? How
does corruption function as a part of their coping strategies?
Of course, the poor may be involved in forms of corruption other than bribery.
There are strong incentives for the poor to pay the bribes or defraud projects,
and thereby strengthen the structure of corruption at the local level (Jenkins and
Goetz, 2004). Indeed, poor people often re-elect corrupt officials or leaders
because they return something to them through an unofficial economy.
Conversely, there is often a cost to the poor for not entering into corrupt
transactions and whistleblowers are sometimes punished. Understanding these
conflicting views helps us to understand why civil society can be so divided on
the issue of curbing corruption and why some key actors are apparently so
unwilling to confront it.
Developing pro-poor anti-corruption strategies
Development policies in Indonesia today are focused on user pay models, which
mean that the poor are meant to pay for capital infrastructure and recurrent costs
more than ever before. When they cost-share, they pay for leakages (such as
those occurring in procurement and elite capture) from scarce household
resources.
An analysis of the poor s interaction with water services and resources helps
point towards some of the areas where corruption might be concentrated and
result in greater impact on the livelihoods of the poor. A simplified diagram
indicating a typical flow of funds in the water sector is provided below. This
value chain suggests three key areas of concern in a pro-poor strategy: fund
transfers, procurement and the local capturing of assets intended for the poor.
145
172
Corruption-in-water typically refers to three different types of money/assets:

public, private or citizens . Each of these has a different impact on the poor.
Public money is diverted, leaked and misallocated, leading to underinvestment
in the sector and in assets that benefit the poor. Private money which mostly
changes hands during procurement and construction (apart from state
capture) results in lower level and quality of interventions. Corrupt water
practices that directly involve the poor are mostly the poor allocating their
limited household money to the bribery of officials. They do this for market
reasons: entry (to access a service or asset), quality (to ensure its continuation
at a certain level), and cost (to ensure they are paying the right price for it).
Furthermore, in practice the flows shown below are associated with a broad
spectrum of water services and resources. The way the poor use/interact with
the water sector, or the way water services and resources comes to them, is
also a determining factor in understanding the nature and scope of their corrupt
transactions.
In Indonesia water supply, about 15 percent of the population obtain network
drinking water from utilities, about 20 percent (and many of these will be
poor) use utility water (obtained illegally or through other providers) and 65
percent use non-utility water (obtained from private suppliers, communitymanaged systems or provided by themselves). In this case, focusing anticorruption efforts on improved management and efficiency of the utility would
miss the majority of the poor (benefits for the poor may result, but are not
guaranteed). Cleaning up the illegal band (essential for the viability of utilities)
probably hurts the poor in the short term if alternatives are not provided.
However, a strategy that tackles the corruption in informal systems, communitymanaged systems and in the construction of wells and on-site sources would
change the corruption field in which many poor live.
Whereas private-sector stakeholders interact with procurement officials and
supervising engineers (or at the higher levels with senior bureaucrats and
politicians), in their corrupt transactions the poor interact with the lowest level
of officials. Many of these officials will be poor or near-poor themselves, using
their public role to meet their own household needs. The poor also operate in
a context where the most indigenous and powerful stakeholders (most
committed to the status quo) generally determine who gets what. In situations
where money needs to change hands for social leaders, it may be a poor
household that acts as middleman. The interaction with the village elder, tribal
leader or elected community leader is an important relationship in the
corruption field of the poor. These figures extort or accept bribes from
community members to facilitate action that will improve their household
situation (be it bribing to have latrines emptied or to divert water to irrigate
crops). In rural areas, the poor also interact with quasi-government officials,
such as village elites/leaders entrusted with public roles, over water resource
management or the management and decision-making over water supply. In
CDD or CDR water projects, poor households will also interact with facilitators
and project officials (Woodhouse, 2002).
Drawing on a livelihoods framework (University of East Anglia (UEA), 2004), the poors livelihoods can be described in terms of social, human, physical, natural and financial assets.
Simplified flow of funds in the provision of water services and resources to the poor
Where does the money go?
Source: Adapted from Plummer, J. Making Anti-corruption Approaches Work for the Poor: Issues for consideration in the development of pro-poor anti-corruption strategies. Stockholm International Water
Institute (SIWI), Stockholm. Forthcoming.
173
Spotlight on Gender
Aligning village-level spending with the needs of the poor through the participation and
empowerment of women
What is a participatory approach in village-level spending?
Participatory community budget formulation requires an inclusive dialogue of
the poor, women and other social groups whose interests may be overlooked
in traditional approaches to planning and budget formulation. A shift in local
government attitudes towards building capacity in order to work more closely
Wonosobo, that the gender sensitivity of facilitators was integral to enhancing

women s participation and ensuring equitable development.146 Conversely,
facilitators who are unconcerned about process, power relationships and
differences can easily reinforce forms of developmental practice that do little
to address inequalities.
with the community is crucial if participatory budgeting exercises are to succeed.

The importance of women in the process: Women area vital part of this
process: involving them in needs assessments and proposal selection benefits
What do women want?
not only the women but also the wider community, while failing to involve them
Women prefer community resources to be spent on health, education water
can lead to project failure. The participation of women has also been shown to
and sanitation facilities. It is women, for example, who largely shoulder the
reduce corruption.
burden of water collection. Time spent by women fetching water prevents them
from availing themselves of educational opportunities and participating in
The lack of knowledge of facilitators: Many facilitators do not understand
income-generating activities, including paid labor.
gender issues well, and even when women s participation has been actively
sought, development practitioners have generally assumed that the presence
Proposals submitted by female groups for support by KDP differ from those of
of women at a meeting is sufficient for them to be represented . Thus, better
male groups. In a six-month period from end-2004 to mid-2005, 39 percent of
methods have had to be developed that enable womens participation.
proposals from women s groups were for water and sanitation, education and
health-sector related activities. Conversely, 59 percent of proposals submitted
Examples of success: However, some programs, like KDP, have succeeded
by men and mixed groups were for transportation infrastructure.
in making women s participation meaningful. During a four- to six-month

facilitated socialization and planning process, male and female villagers
participate at the hamlet, village and sub-district levels. In order to ensure women
Proposal submitted by special women s group and

male/mixed group, by major activities (%)
(end-2004 to mid-2005)
actively participate, they are also active in dissemination activities and the
selection and training of village facilitators. Special women s meetings are held
to prepare funding proposals and women are facilitated to participate in
preparing, implementing and monitoring the resulting projects. Participatory
auditing within KDP has successfully involved women and men in monitoring
both the quality of service delivery and actual budgetary spending and, as a
result, helped to reduce corruption of project funds.
The Asian Development Bank (ADB)-assisted Gender Equity in Policy and
Program Planning program implemented by the Womens Empowerment Bureau
in four districts is another example of actively supporting gender-balanced
development. In this program, gender-sensitive facilitators introduced the
concept of power relations into the community, while also encouraging womens
participation in development planning. It was found in at least one district,
146
174
Interview with Head of Women s Empowerment Bureau, Wonosobo District, Maria Susiawati.
Roads and bridges (%)

Water and sanitation (%)
Irrigation (%) 3.7
Health (%)
Education (%) 20.9
Women s
group
Male/Mixed
group
22.3
9.4
52.7
7.2
7.6
8.4
7.6
1.8
SME (%)
28.9
Others (market, training, electricity) %

Total (%)
6.4
100
20.1
100
Source: KDP, October 2005
Participatory auditing of budgets in Aruan village, North Tapanuli

Upon reading an information board in front of the UPK office, some women in Aruan reacted angrily to a financial report posted there. The total budget for
building the village road was far higher than estimated. The women brought the issue to Renta, the KDP facilitator. Renta called for a sub-district
accountability meeting (Musyawarah Pertanggungjawaban) to discuss this issue. During the meeting, it was agreed that nine women and men from three
hamlets would visit local shops to check prices of material for building the road. This revealed that there should have been Rp 14 million over budget.
Both men and women from the community asked the village head who was responsible for purchasing the road development material to return the money.
It took nine meetings to persuade him but finally, with a sense of shame, the village head eventually handed over Rp 14 million. Community members
clapped their hands in joy. One old woman hugged the KDP monitoring officer and said: For over 29 years he used our money. This time we caught him.
Some other women whispered: No wonder! Now we know why he just bought a piece of land and a motor cycle.
Source: Enurlaela Hasanah, KDP, 2004.
The women of Sukalaksana win a struggle for piped water

For as long as anyone could remember in Sukalaksana, clean water had been hard to come by. Sukalaksana is a village located in a mountainous part of
Garut kabupaten, West Java, where the nearest source of clean water is a kilometer downhill. Previous efforts to dig wells usually never struck water but
when they did it was dirty and yellow. The people of Sukalaksana were forced to drink, bathe and wash using waste water flowing from Parakan, a village
upstream, which had the color and consistency of milky coffee and caused outbreaks of dysentery and skin disease. When KDP was introduced to
Sukalaksana, local women hoped they would finally have access to clean water in their village.
But the women of Sukalaksana still faced many obstacles. At the Special Women s Meeting, the proposal to pipe clean water from a source located further
downhill faced stiff competition from a road infrastructure plan, unofficially sponsored by the village head (lurah) and advocated in the womens meeting
by a power bloc of village elite women, including the lurah s wife. When it came to a vote, the water supply proposal won the day, but when the KDP
verification team conducted a survey of the site, they found that water would never reach the village due to difference in altitude between the village and the
spring s position downhill.
Debate ensued as to whether the women should be allowed to alter the specifications of the proposal, with the lurah s group agitating for the water supply
to be dropped altogether. Finally, the village decided to hold the Special Womens Meeting again and abide by the final results. Women were allotted
seeds and asked to vote by placing their seeds in envelopes representing the various proposals. The result was resounding support for an altered watersupply system proposal, this time utilizing bore water and an electronic pump. As one woman said, Just by putting our seeds into the clean water
envelope, we were able to defeat the lurah s desire to build a road. Flushed with victory and in the face of the lurah s displeasure, the women hinted that
if the lurah was opposed the clean water project, then he need not have the facility connected to his own house!
Still, difficulties were not yet over for the village women. During the second visit of the verification team, concerns were raised that the bore water they
sought to tap was of poor quality, and that the installation of the electric pump could be a long-term financial burden on the community. A third solution
was proposed that would see water piped from a spring 3.5 km away. With the plans finalized, and KDP funds secured and dispersed, villagers set to work
on building the pipeline.
As the midday call to prayer sounded on Monday 19 April 2004, clean water flowed into Sukalaksana. It was greeted by the cheers of villagers and the
shrieks of children, who raced into the washhouse to play together in the water. Village elders gathered to douse themselves in prayerful thanks. Not to be
outdone, the women, who had worked so hard to see their dreams realized, arrived en masse with buckets at the ready. The realization of their long-held
dream saw some women moved to tears. After so much initial opposition, Sukalaksana women had defeated their detractors and brought water to their
village. Today, the water flows as a symbol of their triumph over those who sought to dominate the decision-making process and deny women the
expression of their true needs.
Source: Various KDP reports, World Bank. 2003. Enhancing Womens Participation. Learning from Field Experience. KDP Project; PRA report for Wonosobo district. ADB s TA on Gender Sensitive
Public Policy and Planning. 2003.
175
Chapter
Making Social Protection

Work for the Poor
176
Imagine several mountain climbers scaling a cliff face who want protection from falling. One way to protect them would be to place
a net at the bottom of the cliff to catch any climber just before he hits the ground. Another would be to provide a rope, and a set of
movable devices that can be attached to the cliff; as the climbers scale the cliff, they attach the rope at higher levels, so that if a
climber falls, he falls only by the length of the rope...the safety net guarantees against a fall past an absolute level; the safety rope
guarantees against a fall of more than a given distance.
- Sumarto, Suryahadi, and Pritchett (2001)
Introduction
Why understanding vulnerability is important to fighting poverty in Indonesia

Although income poverty in Indonesia is relatively low, vulnerability to income poverty is high. As
discussed in the Chapter 3 on Understanding Poverty, while only 16.7 percent of the population fell below the national
poverty line in 2006,147 a much greater share of households (49.0 percent) were living on or below US$2-a-day. This
means that some 32 percent of the population lives between the national poverty line and the US$2-a-day line, a much
higher share than any other country in the region.
In any given year, the risk that a large portion of households will fall below the poverty line, even if
currently not poor, is high in Indonesia. The clustering around the poverty line largely explains the high risk that,
in any given year, a significant portion of households that were previously non-poor or near-poor can fall below the
poverty line (Chaudhuri, Jalan and Suryahadi, 2002; Pritchett, Sumarto and Suryahadi, 2001 and 2002). Panel data from
the Village Potential Survey (Podes) reveal that a substantial percentage of households move in and out of poverty. In the
period directly after the crisis, from August 1998 to October 1999, 58 percent of respondents were poor at least once, even
while the average poverty rate over the period was only 37 percent and only 17 percent of respondents were chronically
poor (Pritchett et al, 2001). More recent household income and consumption tracking also shows the same degree of
movement in and out of poverty. Over 38 percent of poor households in 2004 were not poor in 2003 (see Table 6.1).
Income and expenditure shocks, such as job loss, the failure of a business, a poor harvest, lost income from ill health and
the costs of treating sickness, or a rise in basic commodity prices such as rice, can lead to these sudden changes in a
household s poverty level.
Table 6.1
Movements in and out of poverty, 2003-04
2003
2004
Non-poor household
Poor household
Total
Non-poor HH
(Row percent)
156,377,528
92.27
13,097,915
7.73
169,475,443
100.00
(Col percent)
Poor household
(Row percent)
91.57
14,396,615
40.51
38.26
21,138,941
59.49
82.67
35,535,556
100.00
(Col percent)
Total
(Row percent)
(Col percent)
8.43
61.74
17.33
170,774,143
83.30
34,236,856
16.70
205,010,999
100.00
100.00
100.00
100.00
Source: Indonesia Poverty Assessment team, using Susenas panel data, 2003-04.
147
Susenas panel data 2006.
177
The real concern for policy-makers is how households cope with income shocks that throw them into
poverty (or deeper into poverty). Without the right tools and assistance, households seeking to manage these
shocks may take actions that lower the likelihood of emerging from poverty later. This sort of bad coping typically
consists of actions that diminish a household s stock of human capital such as cutting spending on food and health, or
pulling children out of school and, as a result, perpetuate a cycle of low earnings, chronic disadvantage and vulnerability
to future shocks. Although closely related and part of a long-term poverty elimination agenda, the policy concern and
recommendations to tackle vulnerability can have a separate, distinct rationale.
Today there is a real opportunity to seize the moment and set in place a well-designed social protection
system that can address the risks and vulnerabilities faced by Indonesia s poorer and more vulnerable
households. A confluence of increased fiscal space and an increasing recognition that, even after recovery from crisis,
there continues to be a critical need to protect the poor from high vulnerability, provides an opportunity today to put into
place a well-designed social protection system. This chapter provides an overview of what the objectives of such a social
protection policy might be, in the context of overall social risk management. It goes on to discuss the evolution and
experience of social protection in Indonesia to date, leading to the recent changes being made by the government. It
juxtaposes this against an empirical household risk and vulnerability assessment undertaken for this report. This assessment,
of household needs in addressing actual risk and vulnerabilities shows a mismatch between needs and many existing
social protection programs. It leads to implications for a restructured social protection policy. Section VI of the chapter
therefore makes some recommendations on how to move towards a possible national social protection system that would
make sense in the Indonesian context.
Given the well-established links between human capital and economic well-being, social protection is
a core part of development policies to fight poverty and increase equity. Although social contracts will differ
widely between countries, social protection policies are generally intended to safeguard households investment in human
capital. Social protection systems (comprising policy interventions, public institutions, and sometimes the regulation of
private institutions) aim at preventing poverty and helping families manage shocks to income that can threaten their
human capital.
Social protection to safeguard human capital is also important for promoting and sustaining efficiency
and growth. Where households do not have a sufficiently wide array of tools with which to insure against risks/losses,
they are condemned to spend precious resources recovering from adverse shocks that could otherwise be put to productive
use. Coping with certain shocks can impair households human capital or hinder further investment. When few options for
mitigating the losses from shocks exist, households may be deterred from taking appropriate risks in their investment
decisions in other areas, fearing the costs of failure of adopting a new, more productive technology, for example, or
experimenting with a new crop. Since there are many risks to which private insurance markets cannot or will not respond,
there is a clear role for public policy in augmenting the set of mitigation instruments available to households.
Social protection can be seen as social risk management. Individuals and societies can respond in a variety of
ways when faced with the prospect of economic losses from shocks, such as natural disasters, sickness, sudden death,
disability and unemployment. Social risk management is a concept encompassing three broad categories of responses to
shocks: ex-ante prevention, mitigation, and ex-post coping. The challenge for households or societies is how, in the case of
a bad or shock state, income can be transferred from a good or normal state period to the bad or shocked state period to
allow for consumption smoothing. There are effectively three ways in which this can be done: (i) risk-pooling (or market
insurance), which pools risks across individuals and allows for the redistribution of consumption opportunities towards the
bad or shocked states at a price; (ii) saving (or self-insurance), which, while it does not involve risk-pooling and has no
explicit price, does have an imputed cost of forgone consumption; and (iii) prevention, which lowers the probability of the bad
state occurring but does not reduce the size of loss should one occur. Households that are unable (or choose not) to take
178
Chapter 6: Making Social Protection Work for the Poor
preventative measures, or to insure themselves by saving or through risk-sharing structures, are forced to cope with the full
cost of losses in the wake of shocks. According to the framework, individuals or households smooth consumption over good
and bad states. Where risk-pooling is missing the individual is forced to smooth consumption using only savings and
prevention. An array of instruments exists that households can make use of, to pool, save or prevent, in order to address risks.
Table 6.2 provides examples of the array of instruments households have at their disposal, arranging these into pooling,
saving and prevention . (See also Box 6.1 on a comprehensive insurance framework for theoretical perspectives on how the
government can approach social risk management depending on size of loss, frequency and level of externalities involved.)
Table 6.2
Typology of instruments for household risk management

Informal
Inter-house hold or
inter-family
Prevention (lowers the

likelihood of a shock)
Less risky
production
Migration
Proper hygiene
Proper nutrition
Formal
Community-network
based
Saving (transfers
resources from good
times to bad times,
but without pooling
risks)
Money under the

mattress or in
the household
Investment in
human capital
Provided by the
market
Community
disaster
preparation
strategies, plans
and structures
Community level
works to lower
likelihood of
disasters
Unregulated
community level
banks
Intra-household
transfers
Coping (measures
taken to deal with a
shock both savings
and pooling but that
were not originally
intended for this use)
Pulling schoolage kids out of

school to work
Selling household
assets
Cutting spending
in food or education
(spending that
builds human capital)
Risk-pooling (similarly
transfers resources
from good times to
bad times, but
compensates for
diffrences in
vulnerablility between
individuals)
Privately
provided
education and
primary/prevent
services
Training
Provided by
government
Investment in
social capital
(rituals and
reciprocal gift
giving)
Funeral
societies
Unregulated
rotating credit
schemes
Religious or
charity
Charity from
neighbors
Borrowing from
unregulated
money lenders
Saving accounts
in regulated
commercial banks
Regulated microfinance
Financial assets
Private insurance
Severance (job
protection)
schemes)
Borrowing from
commercial
banks
Public education
Public nutrition
programs
Public
primary/preventative
health care
Work-place
safety regulation
Pollution
controls
Mandated
savings
Savings
instruments
provided by public
banks
Public pension
Public health
insurance
Unemployment
insurance
Other social
security
Healthcare
provided by public
hospitals directly
Public work safety
net programs
Emergency
disaster relief
Budget
reallocations
Source: Based on Holzmann and Jorgensen, 2001, and Gill and Ilahi, 2000.
179
Box 6.1 Comprehensive insurance frameworks and the role of government in social risk management: A conceptual
framework for policy-making
The comprehensive insurance framework offers a tool for determining which mitigation instruments and preventive measures will
be most effective given the size, frequency, and the extent of externality of a range of possible economic losses and can be used
to identify when coping is the most efficient course of action, as well as distinguish effective from ineffective or even damaging
forms of coping. From a financial protection perspective, it is more efficient for individuals to cope with, rather than try to insure against, small rarely
occurring losses. However, as prospective losses become more frequent, it is relatively more efficient to engage in prevention and savings to mitigate the
loss (that is, to lower the probability and cover the costs of the loss). As a prospective loss becomes less frequent but increases in size, it is more efficient
to engage in risk-pooling. For many of these large, rare losses households will have incentives to engage in prevention to lower the probability that they
will occur even further
Go
Ex gre
ter e
nt na of
er li
ve ty
nt
La
io
n)
rg
F
sa acil
vin ita
gs te
De
Small Medium Large
Size of
Loss
Impoverishing
a
v. PrevSav nda
Su en ing to
Pr b ti
r
R om si on y Po
ol
In egu ot dy
in
su la e a
g
ra te nd
or
nc
e
In the context of social risk management, a key question is what is the role and justification for public sector intervention into the
arena of social protection. Indeed, there can be a role for public-sector intervention to ensure that households can attain optimal
levels of risk-pooling, saving and prevention. Much of this justification lies in the theoretical instances of market failures for example, when there
is asymmetrical distribution of information or, importantly, when there are important externalities that warrant public-sector intervention. The simplistic
yet powerful prescriptions of the framework become
dramatically clear: the size and frequency of the
prospective loss should determine whether these are best
mitigated with risk-pooling, individual saving, and/or
prevention (and the relative weight each instrument
should have). But, regardless of the instrument, along
Pooling and
Savings and
Prevention
Prevention
the third dimension, as we move away from the origin
and the externality posed by the prospective loss grows,
Pooling, some
Savings and
the justification for intervention by the public sector to
savings and
Prevention
prevention
ensure appropriate measures are undertaken even if these
Savings and
measures
are purely private increases.
Some savings
Prevention
Do Nothing
Some Savings
(G
ov
.I
The rationale for government policy intervention

arises when individuals fail to attain optimal
levels of risk-pooling, saving and prevention. This
could be either because one or more of the instruments is
not available to the individual, or if all three instruments
Source : Baeza and Packard, 2005.
are available, because market inefficiencies (information
problems and other market failures) prevent individuals from employing or utilizing each instrument optimally. Gill and Ilahi (2000) draw four clear
arguments for policy intervention from the comprehensive insurance framework.
Frequency
Frequent
No
ne
Rare
Government should provide (or help provide) the instruments that the market cannot (or will not) provide. Risk-pooling to cover certain
losses (such as the lost earnings from the risk of becoming unemployed; the risk of poverty; disasters; and certain frequently occurring or pre-existing
health events with catastrophic costs) does not exist in many contexts due to information problems. Government can step in to correct market failures
by providing risk-pooling instruments.
Government should provide (or help provide) superior instruments, where only inferior instruments are available. For risks best
covered with individual savings, private agents may turn to bad saving instruments (for example, using cattle, land or other non-liquid assets as a
medium of precautionary saving) because good instruments (such as diversified financial assets, safe reliable and competitively priced forms of liquid
savings, or credit) are not available. Furthermore, poorer households may simply not have the margin to save. Government can intervene to foster the
development of more efficient instruments for saving through the prudential regulation of capital and credit markets, as well as provide direct
subsidies for households that are too poor to hold savings or debt.
Government should help households build and protect their human capital. Investing in human capital education, hygiene, and primary and
preventive healthcare can be an effective and powerful means of prevention. Better-educated individuals are more likely to invest in preventative activities,
180
such as exercise, as well as to seek preventative healthcare (vaccinations). Healthier individuals are less likely to be unable to work, and more educated
or better-trained workers may be less likely to suffer long-term unemployment. However, where credit is constrained, individuals may choose lower-thanoptimal holdings of human capital in favor of assets with greater collateral value. To prevent individuals and households from tilting their portfolios away
from human capital, government can subsidize its acquisition through spending on education and health.
It is usually better to help households to mitigate losses than to cope with them. The instruments for individuals and governments to pool
risks and save are not always available. The resources to take preventative measures are often scarce. Where individuals and governments are
constrained, bad coping in the short run can result. Some prevention and insurance (risk-pooling and saving) is always desirable. Effective policy
should place greater emphasis on enabling individuals to insure against losses through risk-pooling and individual saving, and lower the probability
of losses through prevention, rather than coping with losses after a shock.
In addition to helping households manage their risks, government also has an important role to play in the management of risks
at the macro level. Governments can pool the risks of a limited (but growing) range of possible losses through private market insurance, as well as
through international, multi-lateral risk-pooling structures; save by accumulating surpluses in good times to spend on social programs during bad
times (using ear-marking, stabilization funds, counter-cyclical spending policies); and prevent by practicing prudent monetary and fiscal policy,
engaging in reforms that increase the efficiency and safety of factor markets thus lowering the probability of crises and by investing in increasing their
administrative capacity.
II
The Evolution of Social Protection Institutions: From Universal

Subsidies to Targeted Transfers
Similar to many lower-income countries, Indonesia lacks a structured and coherent social protection
system. East Asia in general and Indonesia in particular rely less on social protection instruments relative to other
comparable countries. From the crisis to 2005, Indonesia s social protection system was characterized primarily by (i)
crisis-era safety net programs; and (ii) large commodity price subsidies and transfers, particularly through fuel products.
Universal commodity price subsidies may be considered a first generation social protection approach, having been the
instrument of choice used by many developing countries in the 1960s and 1970s. The targeted programs undertaken
during the crisis period may be considered as an attempt at second generation type safety net programs. In Section III,
this chapter will argue that Indonesia may be ready to move to a third generation system that is more comprehensive and
matched to the actual risks and vulnerabilities faced by its more susceptible citizens.
The crisis-era safety net

The targeted poverty programs put into place in Indonesia in the wake of the 1998 financial crisis were
designed as a social protection response to the crisis. Prior to the crisis, the emphasis had been on broadbased poverty reduction programs. The events of 1998 led Indonesia to establish a broad social safety net (jaringan
pengaman sosial, or JPS) set up to help poor households impoverished by the crisis (see Table 6.3). The JPS programs
were designed more as temporary short-term safety net interventions than a long-term social protection system and
aimed to ensure the continued access for the poor to affordable food, as well as health and education, through the crisis
period. Many of these programs have remained entrenched in social policy since (a detailed JPS program review is
provided in Annex VI.1). Initially funded primarily by donors, starting in 2001, the budgets for the existing programs were
first complemented and later fully replaced by government funding.
The continued social safety net programs address a diversity of issues, but are low in coverage,
institutionally fragmented and are not managed as one umbrella system. In the years prior to 2005, specific
programs covered only a small number of the poor. They had been designed and implemented by a wide array of government
181
agencies, with limited budgetary resources and little coordination. In 2003, the budgets allocated to the three principal
targeted programs, namely Raskin (subsidized rice for the poor), educational scholarships and health cards, were around
Rp 4.8 trillion, Rp 2.7 trillion and Rp 941 billion, respectively. Together these largest programs constituted only about 2.2
percent of total government expenditure. Moreover, in 2003 the scholarships program covered only 12.1 percent and
health cards only 22.2 percent of households in the poorest quintile. Clearly, the JPS programs were failing to operate as
an effective safety net for most of the poor.
Table 6.3
Indonesia s crisis-era social safety net
Safety Net Objective
Specific Program
Food security
Special Market Operation (operasi pasar khusus, or OPK) program: sales of subsidized rice to targeted households,
currently known as Raskin.
Employment creation
Padat Karya : an uncoordinated collection of labor-intensive programs executed by various government

departments
Education
Scholarships to elementary, secondary and upper secondary school students and block grants to selected schools
Health
Funding for:
Medical services
Operational support for health centers
Medicine and imported medical equipment
Family planning services
Nutrition
Midwife services
Community empowerment
Regional Empowerment to Overcome the Impact of Economic Crisis (PDM-DKE): block grants directly to villages
for either public works or subsidized credit
Source: Sumarto, Suryahadi and Widyanti, 2004.
Assessments of individual programs point to some positive aspects, but raise questions over the ability
of previous safety net programs to respond to household needs from the standpoints of both effectiveness
and efficiency. Pritchett et al (2001) found that the self-targeting labor-intensive public works programs succeeded in
keeping many households out of poverty, but the fragmentation and efficiency of these schemes has been called into
question. A more recent safety net programs evaluation (Sumarto, Suryahadi and Widyanti, 2004) was generally positive,
but found that impact varied across programs. Earlier evaluations of the scholarships program found that although these
reduced the chances of students dropping out of school during the crisis period, they were not effective in increasing the
likelihood of transition to the junior secondary school level (Filmer et al, 2002). Raskin program evaluations found that
the benefits accruing to an average beneficiary household were very low given the high cost of running the program
(SMERU, 2005). The total subsidy received by all Raskin beneficiaries ranged between Rp 2.0 and Rp 2.7 trillion. This
corresponded to only 42 percent to 56 percent of the program budget in APBN 2003, indicating high institutional costs for
relatively low benefits. Health-card program evaluations have been relatively more positive, showing that the scheme led
to increased utilization of public health centers in the period after the crisis (Pradhan et al, forthcoming).
The targeting performance of these safety net programs has also been inadequate. While households in
the poorest quintile are the main objective group for these programs, they have access to only 31 percent of health cards,
39 percent of scholarships and about 29 percent of Raskin benefits (Susenas, 2004) (see Targeting in Section IV). Evaluations
have found that, although participation in these programs is greater among lower income groups, participation of households
in higher income groups is still substantial. There are two main reasons for this poor targeting performance typically cited
in recent studies. First, the programs each had their own registries of the poor, which are created and run by separate
institutions. The targeting of scholarships was usually handled at the provider level by school principals and school
committees. The targeting of health cards was carried out by district health offices through community identification of the
182
poor. Second, there were different targeting criteria employed in the programs. For instance,
in the distribution of health cards, chronic illness was often used as a criterion for eligibility,
which does not necessarily correlate with income poverty. Third, and importantly, the
clustering of households (especially in rural areas) around the poverty line, together with
cultural norms about sharing, challenged efforts to target the poor. In the case of Raskin
distribution, for example, it was commonly found that the subsidized rice was distributed
evenly in communities due to: (i) community expectations that benefits would be shared
equally (bagi rata); (ii) lack of cash resources in poor households to cover the cost of rice
(even highly subsidized rice); and (iii) a perception at the local level that the official list of
the poor coming from higher government authorities does not accurately reflect the current
status of the poor at the village level.
The fuel subsidy: a pillar of Indonesia s social protection

program in recent years
The fuel subsidy a universal price subsidy represented the biggest subsidy
or transfer to households in recent years; it was, de facto, the centerpiece of
Indonesia s social protection scheme until 2005. By fixing fuel prices at subsidized
levels well below world prices, the government effectively supported a transfer to fuelconsuming households, protecting them from fluctuations in world prices. Between 1998 and 2005, fuel subsidies averaged
three-quarters of the total subsidies and transfers that constituted Indonesia s social protection system148 (see Figure
6.1a). Indeed, in the decade prior to 2005, the portion of Indonesia s budget spent on universal fuel subsidies was similar
in absolute and relative size (that is, compared with spending on education, health, targeted programs and infrastructure)
to what many governments in middle-income countries spend on welfare and social insurance.149 Seen in this light, the
fuel subsidy could be historically considered Indonesia s first-generation social protection mechanism.
However, the fuel subsidy was regressive in nature, benefiting richer households that consumed more
fuel far more than it did poorer households. As with commodity price subsidies in other countries, spending on
fuel subsidies benefited mainly middle and higher income groups that consumed more fuel. Figure 6.1 depicts the
regressive incidence of the fuel subsidy had the government not changed the domestic price of fuel in 2005. In total, the
benefits accruing to the richest 10 percent from fuel subsidies were more than five times those accruing to the poorest 10
percent. The top 40 percent captured 60 percent of the subsidy.150 In 2005, an individual in the top decile was expected to
receive Rp 551,000 annually in fuel subsidies on premium gasoline, kerosene and diesel. In contrast, an individual in the
bottom decile would have received less than one-fifth of this amount at only about Rp 98,000. The total incidence of fuel
subsidies on the poorest quintile was estimated to be only around Rp 5 trillion out of the total Rp 103 trillion spent on fuel
subsidies in 2005. In addition, although the poor clearly benefited from the subsidy on kerosene, which is widely used as
a cooking fuel, even the incidence of the kerosene subsidy was regressive. An equivalent direct transfer to the poor could
have been accomplished at a fraction of the cost.
148
With the increase in world fuel prices in 2005, the subsidy level rose to 75 percent of total subsidies and transfers, 24.8 percent of total government expenditures and 5.1 percent of GDP.
In 2004, the fuel subsidy totaled Rp 69 trillion (equivalent to total development expenditures that year) and in 2005, spending on the subsidy increased even further reaching up to Rp 103 trillion (US$10.3
billion). The spending on fuel subsidies in 2005 was about 5 times the spending on health care, 2.6 times the development spending on infrastructure and 1.5 times the spending on education in 2003.
150
A lot of the subsidy also accrued to smugglers that opportunistically made use of domestic and international price differentials-which do not show up in this graph.
149
183
In 2005, the government took a bold move to reduce regressive fuel subsidies. The move was precipitated
by the escalation in world oil prices along with a concomitant and ongoing reduction in domestic production capacity.
The medium-term fiscal outlook certainly pointed to lack of fuel subsidy policy sustainability.151 The government raised
fuel prices in March 2005 by a weighted average of 29 percent, followed by a more dramatic increase in October 2005,
when prices rose by an additional 114 percent. While in March 2005, the price of kerosene was left unchanged, in the
October price rise the kerosene price almost tripled.152 The annualized budgetary savings from the 2005 October fuel price
increase (as they apply to the 2006 budget) are estimated to be equivalent to some US$10.1 billion. Currently, the
budgeted amount for fuel subsidies is down to some US$5.6 billion153 (in contrast to the US$15.7 billion it would have
been in 2005 without the October price increase). The October price increase impact, by decile, is also shown in Figure
6.1.154 The total short-term fuel price increase impact made up about 5.1 percent of per capita expenditure155 for the
poorest decile and 6.2 percent of per capita expenditure for the richest decile. In the absence of any compensatory
measures, it is estimated that the October price increases would have led to a 5.6 percentage point increase in the poverty
headcount index; other factors being constant (see Chapter 5 on Spending for more details).
The impact of the reduction in the fuel subsidy in March 2005 was mitigated by the introduction of three
programs. The compensation policy enacted shortly after the first fuel price increase in March involved reallocation of
Rp 17 trillion towards programs targeting the poor in education, health and rural infrastructure (these three PKPS-BBM
programs discussed in detail in Chapter 5). With the second wave of price increases in October 2005, the government
took a markedly different step to compensate households. Anticipating a relatively greater negative impact on the poor
and near-poor with the inclusion of kerosene in the October price liberalization (excluded in the first wave in March), the
government opted for the introduction of a cash transfer, (subsidi langsung tunai, or SLT). With 19.2 million beneficiary
households in 2006, this program is the largest ever cash transfer program in coverage and total volume of transfers.
Each beneficiary household receives a Rp 300,000 transfer distributed quarterly (see Box 6.3).
An opportunity for the future

The establishment of a targeted cash transfer program signifies a shift from the generally ineffective
and inefficient blunt instrument, that is, a general subsidy benefiting primarily the non-poor to a social
protection regime that is leaner and more focused on targeting poor and vulnerable households. The
Unconditional Cash Transfer (UCT) program was designed as a temporary, one-year cash transfer program aimed at
mitigating the impacts of the fuel subsidy. It makes sense to use the UCT program, as well as the health insurance
component of the PKPS-BBM program, as stepping-stones to developing a social protection program for the future. As
Indonesia moves forward to develop a comprehensive and coherent social protection system, it is important to ensure that
government programs (which together, on an annual basis, currently account for the equivalent of about US$2.4 billion
per year in reallocated funds from the fuel subsidy. See Annex V.5) are focused on addressing the actual risks and
vulnerabilities faced by poorer Indonesians. Section III undertakes focused empirical analysis to ascertain exactly what
these risks and vulnerabilities are, with a view to pointing to the implications for developing a social protection system for
Indonesia that makes sense.
151
Interestingly, albeit for different reasons, the fuel price increase parallels the way in which the demographic transition is similarly making social welfare systems difficult to financially sustain in many other
countries.
152
In October 2005, the price of premium (subsidized) gasoline was increased from Rp 2,400 to Rp 4,500 per liter; diesel fuel was raised from Rp 2,100 to Rp 4,300 per liter and kerosene, nearly tripled in price
reaching from Rp 700 to Rp 2,000 per liter.
153
The fuel subsidy calculation for 2006 assumes that that oil prices stay at US$60/bbl in 2006.
154
This is a static price impact analysis, which assumes no mitigating substitution effects. The total impact is disaggregated into the impacts on household expenditure of fuel prices; public transport prices
(assuming 25 percent pass-through of the diesel price); and the general impact of residual inflation on the rest of the consumption bundle. The overall incremental inflation (due to the fuel price increase)
assumption is based on time-series analysis of the overall elasticity of inflation to fuel price increases (0.06 percent for 10 percent increase in fuel prices). The residual (non-fuel; non-transport) incremental
inflation rate is computed at about 1.9 percent for the October fuel price increases.
155
The figure for per capita expenditures is derived using Susenas 2004 adjusted to (June) 2005 prices.
184
c. Estimated impact of 1 October 2005 fuel price increase by expenditure decile
Source: Ministry of Finance.
a. As of 2005, fuel subsidies were the major source of transfers in Indonesia and make
up a much larger portion of the budget than any other social safety net or transfer
programs
d. Estimated impact of 1 October 2005 fuel price increase by expenditure decile as a

percentage of mean household expenditures in decile
b. Incidence of subsidy transfer by household decile groups
Figure 6.1 Indonesia s universal fuel subsidy: a dominant social policy with a highly regressive impact
185
Box 6.2 Indonesia s PKPS-BBM Unconditional Cash Transfer (UCT) program: Problems encountered and ways forward
In August 2005, the government decided to put into place a targeted Unconditional Cash Transfer (subsidi langsung tunai, or SLT)
program before undertaking a major increase in kerosene prices in October. Given the concentration of people just above the poverty line,
the UCT program was initially targeted to 15.5 million poor and near-poor households (some 28 percent of the population, well in excess of the poverty
rate of 16 percent). This over coverage was also intended to mitigate the risks acknowledged by the government of rapid rollout a political requirement
given the need to launch the program in conjunction with the large reduction in the kerosene subsidies. Under the program, each beneficiary family
receives about US$10 per household per month transferred quarterly in four tranches over the course of one year. Transfers are distributed directly to
beneficiaries via the post office system. The government provided about US$600 million in transfer funds in October 2005, the first tranche of the
program.
For poor recipients, the cash transfer more than compensates for the losses incurred due to the fuel price increase (see Figure 6.1d).
Analysis of three alternative scenarios with regard to the targeting show that the lower deciles were on average more than compensated for the impact of
the fuel price increase by the cash transfer. Even under assumptions of moderate mis-targeting (e.g. with cash benefits randomly distributed to the bottom
40 percent instead of the targeted bottom 28 percent), the lower deciles were on average more than compensated for the immediate price impacts.156
In order to improve the program before the second tranche the government commissioned an early assessment of problems
encountered during first tranche disbursement. Considering that the nationwide program was prepared in a matter of months, it was found to
have performed better than expected. Regional targeting and funds transfer mechanisms worked and were timely. The sharp rise in fuel prices, and in
particular the inclusion of kerosene, also passed without major public protests.
The assessment also noted several problems, particularly in household targeting. The government relied on local authorities to identify
households as program beneficiaries in order to create a poor household directory . Comprehensive data verification was not carried out prior to first
tranche disbursement and the assessments found that incorrect data led to under-coverage (poor households being left off the list), and over-coverage
(households outside the beneficiary range entering the list). Mis-targeting of this kind was due to various issues including the limited timeframe, weak
capacity of enumerators, failure of local authorities to follow data collection procedures, and local interests influencing the selection process (see also
Box 6.7).
Based on this information, the government allowed more people to apply for the transfer after the first tranche. After investigating
these claims, 3.7 million additional households were added as beneficiaries, giving a total of 19.2 million households as of 2006. With three tranches in
2006, the total annual budget for the program is estimated to be close to about US$2.4 billion.
The rapid rollout of the program also caused some problems with implementation, including lack of planning and crowd control for
cash disbursement. Program socialization was carried out sporadically and generally only after problems occurred or when beneficiaries received their
beneficiary card. In most cases, socialization to beneficiaries only included the location and date of collection. As a result, program objectives and criteria for
beneficiaries were not conveyed to the public satisfactorily. No unit was specifically prepared to receive complaints regarding the program. In most cases,
complaints were brought to local authorities and resolved locally. However, a lack of outlets for complaints led to public frustration in some areas.
The government has acted on recommendations to improve these early implementation problems. It has worked to improve distribution
logistics at post offices, and socialization, dissemination and complaints resolution mechanisms. It has also integrated lessons learned into the design of
the conditional cash transfer program to be piloted in 2007.
156
Currently, the independent and largely unrelated increase in the price of rice (brought about largely because of a rice import ban in conjunction with a poor harvest) is more likely to drive an increase in poverty
rates for 2006.
186
III
Household Risk Management: Findings from a Risk and Vulnerability

Assessment and Implications for Social Protection Policy
A first step in developing a coherent social protection system in Indonesia is to understand the risks
and vulnerabilities faced by households. That is the purpose of this section, which summarizes findings of a
household risk and vulnerability assessment (presented fully in Annex VI.2). Such an analysis has previously not been
undertaken in this country. The assessment consists of three parts. In the first, we present indicators of vulnerability to
lower income and earnings capacity organized over the life cycle a vulnerability profile for different household groups
of interest. In the second, we analyze the most important shocks reported by households and how they managed and/or
coped with these shocks. In the last, we examine the availability of risk-pooling, savings and prevention instruments for
mitigating losses, and whether these make a difference to how households fare, proxied by observed changes in spending
on food in the wake of shocks (Packard, Alatas and Marliani, 2005).
Vulnerability profile
Risks to earnings capacity and income are usually significantly greater among poor and near-poor
households than among the non-poor households in the three highest quintiles.157 The main differences
between poorer households during working life are labor market risk factors that increase vulnerability to shocks, namely
informality, non-professional self-employment, and underemployment. Thus, the risk to incomes is one of the most
significant factors differentiating them from the non-poor (see Annex VI.2, Appendix Table 6.1.V).
Moreover, factors that pose risks during infancy, early childhood and among school-aged children are
significantly more frequently observed among the poor than among the non-poor. These risks malnutrition,
lack of immunizations and dropping out from school at the earliest stage of the life-cycle can increase the likelihood of,
and perpetuate, chronic disadvantage. Higher risk factors among poorer households persist into the latter teens and early
working life, and in adulthood manifest in higher rates of incomplete education, underemployment, informal employment,
non-professional self-employment and reported disability.
The poor and near-poor have similar profiles of risk directly related to the labor market and employment.
There are no significant differences in the inactivity and unemployment rates between the poor and
near-poor. The only significant differences between poor and near-poor households are in the risks faced by children.
Poor children have less pre-natal healthcare, a higher propensity
to malnutrition and higher drop-out rates than near-poor children.
In contrast (but similar to the poor), the near-poor generally have a
markedly different risk profile to the non-poor: the only similar
risk profile of the near-poor and non-poor is the self-employment
rate.
Rural households show a greater degree of vulnerability
than urban households do at almost every stage of the
life cycle. Rural households appear more vulnerable: risk factors
at various points in the life cycle are higher among rural households
than among urban households. Rural households have higher rates
of unattended birth, higher drop-out rates, and higher incidence of
157
Significant in this section indicates statistically significant differences, using simple t-tests, at the 5 percent confidence level.
187
child labor. Some important exceptions to this general pattern are inactivity among 16 to 25 year-olds (neither working
nor studying), unemployment among this age group and unemployment among adults aged 26 to 55, which is consistently
higher among urban households. This may reflect the luxury good nature of unemployment in Indonesia, particularly
among urban groups, found by others (Alatas et al, 2006; Alisjahbana and Manning, 2005).
Gender is also an important determinant of household vulnerability. There are again significant differences in
risk factors during the earliest stage of the life cycle between households headed by men and those headed by women.
The direction (i.e. whether male-headed households fare better than female-headed households, and vice versa) and
magnitude of significant differences in risk vary by indicator. However, some general patterns in the vulnerability profiles
emerge: there is a higher incidence of mothers failing to
receive prenatal care in households headed by women;
school attendance is lower; and the incidence of child
labor is higher in households headed by women. There
are also significant differences between households
headed by men and women of the same income group.
For instance, among poor and near-poor households
school attendance at the primary and secondary level is
lower in households headed by women. On the other hand,
among poor households, young children are less likely
to be sent to preschool and the incidence of unattended
births is higher among households headed by men.
However, the elderly in poorer households do not
appear to be significantly more vulnerable than
those in non-poor households. The risk factors that
typically provoke concern among policy-makers, such as
the elderly living alone, labor-market participation,
unemployment and disability, are not significantly greater
among the elderly poor and near-poor. The main indicator
of vulnerability in old age (over 65), the elderly living
alone (or accompanied only by their spouse), actually
increases among higher income groups. Fewer elderly in
lower income groups the near-poor and the poor live
alone. Unemployment among the elderly is also highest
among non-poor households. An exception to this general
finding is underemployment among the elderly, which is
higher among poorer households. There are no significant
differences in the incidence of permanent disability among the elderly between any of the household groups. Even among
those aged 55 to 64, there are no significant differences in risk factors between poor and near-poor households except for
underemployment.
Nature of shocks and coping mechanisms

Three types of reported shocks stand out among all households in Indonesia: the ranking of reported
shocks (i.e. from most frequently reported to least frequently reported) is strikingly similar across all
households (see Table 6.4 copied from Annex VI.3 Model 2). The most frequently reported are systemic or covariate
188
losses from changes in government regulations (such as increases in the price of subsidized goods, such as fuel and
rice). The second most commonly reported shocks are economic risks or losses in earnings that could be covariate or
idiosyncratic to the household. More typically, idiosyncratic shocks are reported third, with losses from adverse health
events ranked highest and uniformly more frequently reported than even unemployment.
Indonesian households are more likely to use different forms of saving to solve the problems caused by
these shocks than other forms of risk management. In particular, the rural and urban poor and near-poor households
are more likely to use saving mechanisms, such as borrowing money and reducing household expenses over pooling
mechanisms, such as seeking help from family and asking another family member to work. Two-thirds of urban maleheaded poor households used saving mechanisms.
Among different forms of saving , recourse to credit/
borrowing and actual financial savings are more
frequently reported than the selling or pawning of goods
or cuts in household spending (see Table 6.4).
However, selling household goods and cutting
spending are more frequently reported among
poorer groups, raising concern that poorer
households may be forced into bad coping (see
Table 6. 5). Around 20 percent of rural poor femaleheaded households claimed that they reduce their
household expenses in coping with shocks. Notably, the
poor and near-poor are more likely to report asking
household members who were not previously working
to enter the labor market. Around 10 percent of urban
poor households stated that they asked household
members to work, while the corresponding figure for
non-poor is just 7 percent. Similarly, in rural areas the
gap between poor households and non-poor households
is also around 3 percent, with 11 percent of poor
households relying on bad coping mechanisms, such
as asking members, most notably children, to work, and
8 percent of non-poor households relying on bad coping
mechanisms.
The high recourse to saving rather than riskpooling mechanisms is probably due to the lack
of access to risk-pooling instruments, either formal or informal, among households. The lack of riskpooling at households disposal may be driving the dominant recourse to saving instruments and worryingly, the recourse
to spending cuts that is uniformly reported by poor and near-poor households with greater frequency. Cuts in expenditure
or increases in household labor supply after a shock are to be expected. The concern from a social protection perspective
is that these cuts may be in investment in human capital, and that those being deployed into the labor market are children
who would otherwise be in school.
The various forms of coping and instruments used to cope across poor, near-poor and non-poor
households differ, but do so in an expected fashion. Table 6.6 catalogues the mitigation and coping instruments
captured in the Susenas surveys, using the risk-pooling, savings and prevention typology presented in the conceptual
189
framework section. Forms of prevention, such as safe water and sanitation, increase with income, as do the coverage of
traditional social insurance and the actual receipt of pensions. The poor and near-poor are less likely than the non-poor
to hold savings in financial institutions but, as discussed previously, are observed to borrow money, and sell and pawn
household assets, more frequently than the non-poor.
Impact of shocks
Reported shocks have a significantly negative impact on household food spending and the magnitude
of this impact varies greatly across income groups. This finding is the result of regression analysis (Appendix
Table 6.3 Model 2 and Appendix Table 6.3 presented in Annex VI.2), which examines the impact of reported shocks on
investment in human capital (proxied by changes in household spending on food). Food spending by poor households is
more negatively affected by a wider range of shocks, such as economic and health losses, than for near-poor and nonpoor households. However, near-poor households experienced cuts in food spending of the greatest magnitude.
The severest cuts in food spending among poor households are among those who reported losses from
conflict, natural disaster and economic risks (grouped together), followed by poor households facing
economic shocks, and then by those facing health shocks. The largest cuts in food spending were experienced
among near-poor households that reported this set of losses (see Annex VI.2). These were followed in severity by cuts in
food spending experienced in poor households reporting economic losses alone, and then those reporting losses from
health factors alone. Again, households that reported losses in all income groups experienced significant cuts. To put the
magnitude of these cuts into perspective, while poor households experienced falls in food-spending of up to 26 percent
(depending on the type of shock they experienced), and near-poor households experienced a decrease of 52 percent,
food-spending of non-poor households that reported shocks fell by only 5 percent.
Withdrawal of financial savings, such as the sale of household assets, does not prove adequate: it fails
to prevent a decline in food spending in the face of shock. Recourse to forms of saving , such as withdrawal of
actual financial savings and pawning, was in the case of some shocks found to be positive, but in the case of others
associated with a decrease in food spending. The sale of household assets is unambiguously associated with a decrease
in food spending, notably so among near-poor households. The withdrawal of financial savings is to be expected in the
wake of a shock, and the sale of household assets could be deemed bad coping depending on what these assets are. In
Indonesia, recourse to the sale of assets appears to be at the very least ineffective coping , particularly for the near-poor
but also for the poor. Such a release of savings does not succeed in preventing cuts to spending on food following the
shock.
Formal forms of risk-pooling, such as social safety net programs, have not been successful in mitigating
shocks, while informal risk-pooling instruments, such as informal household transfers and rotating
credit schemes, have had more success. The effect of social safety net programs (i.e. the Raskin subsidized rice
program, education scholarships and health-card coverage) and public healthcare, where significant, were consistently
associated with cuts in spending. Pensions were found to be significant only among non-poor households. The effects of
informal forms of risk-pooling were also found to be significant, and the nature of their impact was mixed. However, on
balance money transfers, food transfers and payments from Arisan a community-level, rotating credit scheme were
associated with a positive change in food consumption.
190
Rural households
Urban households
Rural households
Households that reported solutions to the shocks, reported recourse ( percent) among households reporting shocks, 2004
Reported Solutions
Table 6.5
Urban households
Shocks that had a negative impact to family welfare , reported incidence (percent) among households 2004
Negative experience which had

an impact on family welfare
Table 6.4
191
Table 6.6
Risk mitigation and coping instruments captured in Susenas
Instruments (percent observed in each group)
2003
Poor
Near-
Non-
poor
poor
Poor
2004
Near-
Non-
poor
poor
I. Mitigation (ex ante a shock)

I.I.
Prevention (actions that lower likelihood of losses)
1.
Privately provided education
11.31
12.97
15.15
10.41
12.97
15.75
2.
Safe water
45.46
51.92
58.64
49.08
49.82
60.29
3.
Sanitation
20.01
32.41
57.51
19.13
32.51
60.83
31.05
31.44
27.92
30.63
32.20
29.40
I.II. Saving (inter-temporal transfer without compensation for risk differential)

4.
Investment in education
5.
Savings in financial institutions
11.18
15.75
27.51
14.57
16.65
30.01
6.
Other financial assets
0.04
0.10
0.25
0.29
0.35
0.21
7.
Jamsostek (provident fund for retirement)
1.07
2.71
7.18
1.04
2.87
7.73
I.III. Risk-pooling (inter-temporal transfer with compensation for risk differential)

8.
Private insurance policies
1.02
2.04
6.08
1.36
2.36
7.71
9.
JPKM (health coverage provided by local government)
0.94
1.18
0.47
1.52
0.75
0.70
10.
Health Fund
0.46
0.29
0.73
0.47
0.38
0.75
11.
Health Card
20.56
14.86
7.68
15.98
13.11
6.77
12.
Askes Health Insurance (SI for civil servants)
0.83
1.77
12.14
0.89
3.34
12.16
13.
Payments to Arisan (rotating credit scheme)
20.81
26.75
29.64
20.13
23.07
31.39
1.71
1.27
II. Coping (ex post a shock)

II.I. Risk-pooling
14.
Payment from insurance policy
0.80
1.20
1.49
1.02
15.
Payment from Arisan (rotating credit scheme)
13.22
18.97
19.99
13.40
14.96
20.98
16.
Use of public health services (at local clinics)
0.67
1.26
1.28
0.65
1.12
1.50
17.
Benefits from social safety net program
71.33
63.38
38.91
62.84
54.98
32.99
18.
Scholarships (other than safety net scholarships)
4.83
2.61
1.69
3.97
2.76
1.75
19.
Money transfers (usually from family)
36.76
37.91
35.77
38.18
34.66
36.30
20.
Food transfers (usually from family)
62.36
57.38
52.40
60.14
57.72
51.91
11.20
14.11
18.68
10.74
11.22
19.75
II.II. Saving
21.
Withdrawal of financial savings
22.
Payments from other financial asset
0.59
0.26
0.25
0.44
0.36
0.27
23.
Pensions
0.97
1.51
4.74
0.48
1.59
4.68
24.
Inherited assets
0.66
1.14
1.63
0.55
0.91
1.68
25.
Repayment of money lent to others
3.73
5.16
5.62
3.20
4.50
5.77
26.
Borrowing
17.82
18.67
15.87
16.41
17.35
16.32
27.
Equity release payments
0.11
0.12
0.31
0.06
0.12
0.32
28.
Pawning goods
1.52
1.35
1.35
1.47
1.65
1.24
29.
Sale of HH assets
11.37
8.93
7.97
10.83
9.35
6.94
30.
Cuts in spending in food/education
26.36
22.90
16.47
18.91
14.28
9.87
192
Implications for developing a social protection system

Several implications can be drawn from our assessment of household vulnerability and risk management that are useful
in designing effective social protection policies for Indonesia.
First, given the similarities between the poor and near-poor, it is important to recognize that in addressing
broad risks and vulnerabilities, both these groups should be considered. Indeed, this section showed that
while poor households experience cuts in food spending from a wider range of shocks, the severest cuts (of greatest
magnitude) in the wake of shocks are observed among near-poor households. Consistent with these results, Chapter 3 on
Understanding Poverty revealed that inability to manage shocks explains up to 1.4 percentage points of the poverty rate.
Clearly, shocks are a problem not only for the poor but also for the near-poor, and could be leading to coping mechanisms
that have longer-term consequences than mere movements in and out of poverty.
Second, both the vulnerability profile and assessment of the nature of shocks point to four major sources
of vulnerability that may need to be the focus of any social protection scheme in Indonesia:
Risks to earnings capacity and income are significantly greater among poorer (poor and near-poor)
households. Labor-market risk factors, such as reliance on informal sector, non-professional self-employment,
and underemployment increase vulnerability to shocks among the poor. The risk to income shocks is one of the key
factors differentiating the poor and near-poor from the non-poor. Programs may need to be considered that address
idiosyncratic or covariate shocks to household incomes of these poorest households
There is evidence of relatively higher risks to children s future earnings ability and welfare following
shocks among poor and near-poor households. This may indicate that the opportunity costs of undertaking
preventative healthcare, and sending and keeping children in school, could be high. In contrast to acute shocks,
however, the potential losses wrought by these risk factors are likely to manifest themselves gradually in the form of
chronic disadvantage. Programs intended to assure at least basic healthcare for mothers and children; minimum
nutrition requirements; and, preventing poorer households from pulling their children from school, could help address
this dimension of vulnerability.
Health shocks are a particularly important source of vulnerability among poorer households in
Indonesia. Health shocks are uniformly identified as one of the major idiosyncratic risks and shocks that households
face. A comprehensive social protection system would need to consider how to address this issue.
A further major self-reported shock reported by households themselves is losses from government
regulations. Further analysis would be required to disentangle what is meant by this. However, in the broadest
terms it would follow that the maintenance of the terms of trade (or cost of living) for poor households would be a key
factor. Hence, maintenance of macroeconomic stability and price stability of key commodities consumed by the poor
(such as rice) would be important, while dramatic increases in prices would be expected to have an immediate
negative impact. Likewise, the uncertainty of government regulations might also be considered a risk factor by
households. Policy stability is usually highly valued in terms of household security. These factors may lie beyond the
remit of a social protection program per se but are important aspects in the broader strategy for social protection and
mitigation of risks and vulnerabilities, as per the framework put forward earlier.
Third, there is little evidence that the risk factors that arise with ageing contribute to household
vulnerability. The rate of reported disability in old age is relatively uniform across all income groups; labor force
participation and unemployment among the elderly is highest among the non-poor; and most poor and near-poor elderly
live in households with working-age family members. In light of this finding, the emphasis given to the risks associated
with ageing in recent efforts to create a national traditional social security system seems misplaced.
193
Fourth, greater access to safe, competitively provided financial forms of saving, and formal risk-pooling
is likely to augment household risk management considerably. (See section on: Connecting the Poor to Financial
Services in Chapter 4 on Growth for more detailed recommendations on this sector.) The incidence of shocks to households,
while widespread, leads to the most severe cuts in food spending among poor and near-poor households, whose
preponderant reliance on savings still does not permit adequate mitigation of shocks. Immediate, acute losses, both
systemic (regulatory changes, sudden economic shocks, unemployment) and idiosyncratic (again unemployment, other
earnings losses, adverse health events) are widespread and reported by all types of households. However, as discussed
above, these shocks lead to more severe cuts in food spending in poor and near-poor households. Households rely
heavily on forms of saving in the wake of shocks. Recourse to financial savings is widespread, but the poor and near-poor
rely more heavily on non-financial forms of savings, such as pawning and the sale of household assets. While these
actions may or may not prove to be bad coping , they are not fully effective in preventing cuts in food spending. Formal
risk-pooling instruments are found to be effective, but only for the non-poor. This reflects the very low coverage of
traditional social insurance and private insurance policies among the poor and near-poor. However, where available,
informal risk-pooling mechanisms help households to manage these losses better. Households could clearly benefit from
better instruments to augment risk management. However, to be really effective, publicly-provided formal risk-pooling
would have to take an institutional form that is very different from the current highly structured, institutionally intensive,
pay-roll-tax financed models, which do very little for Indonesia s most vulnerable.
194
IV
Towards a National Social Protection System that Makes Sense for

Indonesia
There is substantial interest among policy-makers in the creation of a national social protection system
and now is a prime opportunity to develop such a system. Interest in social protection is driven by at least two
recent factors touched on to some extent in earlier sections of this chapter. These are: (i) the reduction in the universal fuel
subsidies that generated enormous budget savings for the government and a desire to channel these savings towards
more effective poverty reduction programs; and (ii) concern over signs of a return to an ad hoc, fragmented approach to
the design and implementation of poverty programs in the post-crisis recovery period. There are also some concerned
about the Social Security Law (Law No. 40/2004) that nominally extends coverage of payroll-tax-financed social insurance
nationally, with no consideration of institutional capacity, little regard for the losses that pose the greatest risks to Indonesian
households, and potentially unsustainable fiscal costs.
The government is right in wanting to establish a coherent and coordinated social protection system
that effectively augments the set of options households have to manage shocks and helps in the broader
policy drive to eliminate poverty. The readiness to revise the social protection system acknowledges important facts
about current social protection policies and programs, including targeted social assistance, traditional social insurance
programs, and the once dominant universal fuel subsidies. Namely, that these forms of
social protection have been to varying degrees expensive and regressive, have done little
to address vulnerability, and could do much more to bring the poor out of poverty as part
of a broader poverty reduction strategy. Indeed, it is important that such a system is relevant
to the shocks that most threaten the poor and near-poor, and is institutionally appropriate
for Indonesia. For this purpose, Section III assessed vulnerability, shocks and risk
management at the household level. This provides pointers to policy-makers on the type
of social protection system that would be most beneficial to Indonesian households given
the risks they face, the losses they most frequently report, the set of instruments within
their reach, and the instruments that would be beneficial but are currently out of reach,
particularly of poor and near-poor households.
It is important to clarify the objectives of various social protection policies
and programs as distinct but not necessarily separate from the wider set of
policies for reducing poverty. As stated earlier, the objectives of Indonesia s crisisera social safety net (JPS) are generally well aligned with internationally recognized goals
of social protection policy: (i) to alleviate the poverty of the current poor, by raising their
consumption level; (ii) to prevent the near-poor from falling into poverty; and (iii) to help
households both near-poor and non-poor mitigate the fall in income in the wake of shocks.
The three objectives are sound and in line with what social protection systems are commonly
expected to achieve. However, the current structure of the safety net fails to differentiate
these objectives. From purely the standpoint of helping households manage risks to their
human capital, with regard to the objectives, (ii) and (iii) are thus distinct from (i), which
is a more generic objective of Indonesia s overall poverty reduction strategy. This has led
to a mis-alignment between each objective and the instruments the government deploys
to achieve each objective. The performance of past social safety net programs has been
hindered by a lack of clarity and separation of these objectives of a social protection
system, and where these may at times even diverge from purely poverty-elimination
objectives. For example, while the structures of some of the component programs of the
safety net are best suited to raising the consumption of the current poor, they do not
195
perform well in keeping the near-poor out of poverty. Looking ahead, as policy-makers explore options for a new social
protection system, although the three objectives listed above are adequate, greater clarity on how each is best pursued is
required.
Different objectives are better pursued with different instruments. The programs that are best suited to tackling
the determinants of (and cover) losses that manifest over a longer period (such as the damage done by child malnutrition,
lack of immunization, etc) are very different from the programs that will most effectively cover sudden, acute losses (such
as unemployment, disability or losses from health events). The specific instrument deployed (risk-pooling, saving or
prevention) to cover a given prospective loss (job-loss, sickness, etc) should reflect the nature of the loss (frequent or
rare; small or large; purely individual or wider external impact). Moreover, the logistical design of these instruments
(targeting method, etc) should also match the losses that they are intended to cover.
There are several critical areas on which policy-makers can immediately focus their efforts towards
establishing a national social protection system that makes sense for Indonesia today. First, there is
clearly a need for more effective, publicly provided risk-pooling instruments to augment household risk management
options, given the limited reach of private formal risk-pooling and the nature of the risks (under-employment, unemployment,
health losses) that should be covered. Second, there is also evidence that households would benefit from more efficient
provision of privately provided risk-pooling and savings instruments, particularly sophisticated, safe and competitively
priced financial instruments. Third, to a great extent the chronic disadvantage that makes poor and near-poor households
more vulnerable to shocks could be addressed with more extensive and effective prevention efforts. In the remainder of
this section, we discuss the policies and programs the government should consider to address the first and third areas
directly.158
In particular, given the risks and vulnerabilities outlined in the previous section, there is merit to
Indonesia considering a four-pronged approach to a social protection strategy, which would build where
possible on existing programs and address the actual risks and vulnerabilities faced by the poor. The
risk and vulnerability assessment (RVA) outlined in this chapter shows that households are most affected by catastrophic
(and covariate) shocks best covered by simply designed, broad, national-level risk-pooling instruments, rather than
sophisticated closed forms of traditional social insurance. Indeed, the RVA pointed to three broad risk and vulnerability
factors that Indonesia should consider when developing its social protection system. The large-scale social insurance
system envisioned by the National Social Security Bill would be less of a priority from this standpoint.
158
A conditional cash transfer program ( cash for education and health ). The objective of such a program
would specifically be to address the findings on the risk factors that appear to be hindering human development
outcomes among the poor due to their bad coping mechanisms for shocks, and particularly those that hinder childrens
development (e.g. through cuts to healthcare and education spending).
A workfare program ( cash for work ). One of the biggest risk factors to poor and near-poor households alike
was shocks to their incomes-especially given their vulnerabilities to the particularities of informal sector employment,
self-employment, etc. A workfare program would have as an objective the reduction of this risk factor.
Health insurance. One of the most important risk factors to be consistently reported by poor and near-poor
households is health-related shocks, such as reduced spending on regular healthcare for children (as opposed to
actual health shocks, such as illness of a family member). Thought is required on how best to address this risk factor
with an appropriately efficient and targeted program.
The second area, while critical to effective household risk management, is more closely linked to policies that promote financial sector development and the provision of safe, competitively priced savings
and credit instruments. See Chapter 4 on Growth for details.
196
Rice price stabilization. Maintaining stable rice prices is very important for the poor, but the current system for
preventing large increases in the price is seriously flawed. The most effective way to prevent large increases in rice
prices would be to allow the general importation of rice subject to a low tariff.
Related to this, in developing a new system that focuses on actual risks and vulnerabilities, the
government should also consider which existing social protection or safety net programs it may
need to cut. It will be important to consider refining existing programs to support the development of the above programs
or to address the risks and vulnerabilities identified in another manner. A possible phasing-out of some existing programs
may be required in order to ensure a streamlined social protection system. From the fiscal standpoint, a consolidated and
comprehensive social protection system with a few major programs can also improve institutional coordination, costeffectiveness and program efficiency.
Conditional cash transfer program

Indonesia s move to a monetized cash transfer system is a step in the right direction away from regressive
and inefficient commodity subsidies. A shift to cash transfers, such as the UCT program recently instituted in
Indonesia, is important because it allows households to simultaneously address their immediate income needs (admittedly,
an objective and feature of risk-pooling) and because cash-in-hand is fungible spend in areas that can strengthen their
human capital (whether these be investments in improved nutrition, more primary health attention, or keeping children in
school) and thus lower their vulnerability to future shocks (i.e. prevention). Cash transfers have been shown to be more
efficient and effective than transfers in kind (see Box 6.3). Indeed, even after reductions in the fuel subsidy, Indonesia can
go even further in this direction by monetizing other commodity price subsidies, such as the current subsidies on fertilizer
and rice for the poor, and reducing distortions in market prices.
Box 6.3 The emergence of cash transfers as an instrument of social policy in developing countries
Countries at all levels of development and income have used a variety of instruments to compensate households for price changes,
to help them to manage economic shocks, and to alleviate the plight of poor households. In developed countries, the use of cash has been
more prevalent in temporary or permanent assistance programs and unemployment subsidies.
In developing countries, cash transfers are less prevalent and the preferred means are broad price subsidies on the products that
the poor consume, or for in kind (mainly food) transfers. The rationale for preferring these mechanisms over cash transfers is somewhat
paternalistic, and rest on a fundamental distrust of household consumption decisions, particularly the capacity of the poorest households to make sound
decisions that will lead to eventual welfare improvements. For example, since cash is fungible, in kind food transfers are believed to guarantee food
consumption and to deter frivolous spending on goods or spending behavior that society deems undesirable, such as alcohol, tobacco and gambling.
However, none of these claims has been supported by strong evidence. In fact, quite to the contrary, the more traditional methods used by
governments concerned with helping households manage risks and poverty can lead to worse outcomes. Universal price subsidies are typically captured
by non-poor groups and can result in regressive transfers. In kind food assistance may compete with local food production and consumption and may
have deleterious effects on local food production.
Recently, the governments of many developing countries have started to prefer cash transfers which, among other things, are
cheaper to deliver than in kind transfers and present similar or better benefits that many of the in kind transfers. With cash transfers,
families are able to buy the foods they prefer in local markets and can buy in small quantities to rationalize consumption and storage.
The fundamental shift for policy-makers is away from meddling in markets whether directly in the form of price subsidies or
indirectly through the distortions caused by in kind transfers and towards an approach that places greater trust in mother s decisionmaking. This shift has been pushed by evidence showing that, although some frivolous spending can indeed exist, the overwhelming majority of
mothers make sensible, welfare-enhancing decisions.
Source: Adapted from Castaneda, 2005; and Castaneda and Lindert, 2005.
197
CCT programs can go further than just providing a cash transfer to poor households, by making cash
transfers conditional on certain human capital behaviors by households, CCT programs go further than
just reducing consumption poverty in the short term. As an anti-poverty instrument, a cash transfer alone addresses
one important dimension of poverty, but leaves other dimensions unaffected or affected only indirectly. Conditional cash
transfer (CCT) programs, on the other hand, allow the simultaneous transfer of cash to the poor, while also leveraging an
increase in demand by these households for health and education services, thus promoting accumulation of human
capital. CCT programs provide cash transfers to the poor, conditional on families receiving health check-ups, vaccinations
and other preventive healthcare for their small children, and keeping school-aged children enrolled in school (see Box
6.4). From a risk-management perspective, conditioning cash is a way for governments to augment household prevention
measures that address chronic vulnerability by lowering the likelihood of losses from shocks. The CCT model presents a
number of advantages. It is flexible (adaptable to a country s specific needs), scalable (works for pilots and can be
expanded to large numbers), and fiscally sustainable (cost effectiveness is typically greater than that of other programs).
Impact evaluations of CCTs show impressive results in terms of increased school enrollment and usage
of preventive health services by poor households. A study by Rawlings and Rubio (2003) shows that CCT
programs are effective in promoting human-capital accumulation among poor households. For instance, CCTs were
found to: improve child nutrition and increase vaccination, thus, preventing the spread of communicable diseases and the
advent of under-nutrition; improve maternal health by increasing pre-natal check-ups; and, increase transition rates from
primary to secondary education. Higher enrollment and transition rates also reduced child labor. In Mexico, the poverty
gap was reduced by 30 percent and the severity of poverty by 45 percent among program beneficiaries (Skoufias, 2005).
In Colombia, the incidence of extreme poverty was reduced by 6 percent after about one year of program operation
(Attanasio et al, 2004). In most cases, over 40 percent of program benefits went to the poorest quintile of the population.
In Nicaragua, average enrollment rates in treatment areas increased nearly 22 percentage points as a result of the program,
from a low starting point of 68.5 percent. In Mexico, the CCT program reduced the probability of working among children
aged 8 to 17 by 10 to 14 percent relative to the level observed prior to the program (Parker and Skoufias, 2000). The
Progresa evaluation shows a significant increase in nutrition monitoring and immunization rates. Infants under the age of
three participating in Progresa increased their growth monitoring visits by between 30 and 60 percent, and child beneficiaries
aged under five had a 12 percent lower incidence of illness compared with non-Progresa children (Gertler et al, 2006).
Several countries also report that the program improved the status of women within the family and/or community.
198
Box 6.4 What is a CCT and what is its purpose?

A CCT program provides a cash transfer to selected and often extremely poor families once they have met certain program conditions
or responsibilities typically related to obtaining preventive basic health and nutrition services, and education. This unleashes two
powerful forces. First is the motivation of those families who want the cash transfers and demand the services to meet program conditions. The second is
the motivation of those service providers that need to provide the services-if services are unavailable families will not be able to obtain the cash transfer.
In most countries, the transfer is provided to the female head of household.
What are the most common CCT components?
1.
Nutrition and health components. These components aim at assuring that children aged 0 to 5 and pregnant, lactating women receive critical
preventive and basic primary healthcare services that would prevent communicable diseases, under-nutrition and ill health. These services include
regular health checks, such as growth monitoring of children, vaccinations and pregnancy controls, and vitamins and iron supplementation,
among others. They also can include a strong nutrition and health education component. Households with children aged 0 to 5 often receive a
grant usually a lump-sum amount independent of the number children under the condition that the household assures that children are taken to
health posts or centers to receive the services. To make sure that the supply of services and the supplements is available on demand, the program
sometimes also provides funds to improve the supply side of health and nutrition services at the local level.
2.
Education components. These components aim at inducing families with school-aged children-generally aged 6 to 17-to enroll and maintain
regular school attendance of these children. The payment usually depends on the number, grade or gender of children. This is in order to influence
enrollment in certain grades, for instance when children pass from primary to secondary education, or to provide more payments for older children
whose opportunity cost of working is higher, or to try to induce enrollment of women if they are lagging behind (e.g. in Mexico). The cash transfer
will help families to pay for the direct education costs (books and other materials) and the earnings children forego when studying rather than
working. In many countries, poor children do not attend school (particularly secondary) because they need to work to help support their families.
In this case, supply-side interventions alone (more investments in schools, teachers, etc.) will not be sufficient to induce higher enrollment and
school attendance of the poor. The CCT program can also induce improvements in education quality by requiring minimum standards for program
participation (e.g. Colombia) and providing additional funds for improving the supply side.
In some longer established programs, formal linkages are established between CCT beneficiaries and other complementary
services. Chile s Bridge Program, for example, provides beneficiaries with the support of a social worker for two years, to help integrate them into the
wider network of services and programs. Similarly, Bolsa Familia in Brazil links beneficiaries to literacy training, micro credit and local business development
programs. These efforts often involve engaging civil society, through participation in consultative councils (in Argentina, Brazil and Chile) or through
elected beneficiaries who serve as conduits between the program and their communities (in Mexico and Colombia).
The CCT model can be adapted to Indonesia to address its key multi-dimensional poverty goals. A CCT
program could be specifically designed to address Indonesia s high maternal and child mortality rates, under-nutrition,
low progression to junior secondary education of children from poor households, or deficient access to water and sanitation.
Based on the experience of other countries, a CCT program in Indonesia could produce a decrease in maternal mortality
rates by improving pregnancy check-ups early on, providing vitamins and other supplements, and dispensing health
education. In addition, it could lower child mortality rates by increasing vaccination coverage and provide vitamins and
supplements (iron, etc.) together with nutritional education. However, it would be critical to ensure that health conditions
could be met by utilizing the services of non-governmental/private providers. In education, the poorest children could
achieve higher progression rates to junior secondary school and there would be more secondary school graduates. These
would all be significant contributions to achieving the Millennium Development Goals (MDGs).
Given its long tradition of community driven development, Indonesia could experiment with a conditional
transfer to communities. One possible innovation of the CCT model not yet tried elsewhere might be to provide
grants to local communities on condition that the community reaches the same health and education indicators that form
the conditionalities of a household CCT. This could be most effective where there are problems that are best dealt with
collectively. As with other community empowerment programs, trained facilitators would likely be required to help
communities collectively diagnosis the specific problems in health and education in their village, and identify steps they
could take to resolve them.
199
The community grant, just like a household transfer, could be used to address the identified problems
or to meet other community needs, as long as the conditions were met. Communities could, for example,
help address small-scale issues with the supply of services: (i) in education, provide subsidized transport to ensure
children are able to attend a distant junior secondary school, provide textbooks or other school materials, support teacher
professionalism, or improve school infrastructure; and (ii) in health, provide subsidized transport to the health clinic to
ensure pregnant women receive appropriate prenatal examinations and/or cover costs of midwives to visit the community
for prenatal, delivery and antenatal services, provide nutrition supplements, or sponsor public health education.
Communities could also undertake activities to increase demand for services, for example, by helping parents understand
the relevance and benefits of education for their children or by building community understanding and interest in ensuring
pregnant women and new mothers receive the health services they need.
In the Indonesian context of decentralization, a CCT needs to be designed to foster stronger service
provision at the local level. With decentralization, district/municipal governments are the focal point for health and
education services, so for a program to work it needs to assure their interest and commitment. Given that the impact of a
household CCT could be limited if there is not sufficient supply to meet the increased demand for services, addressing the
supply side would be an important role for district/municipal governments.159 Their involvement in the actual delivery of
the program is also consistent with decentralization. If jointly implemented it would be important to outline clearly the
roles and responsibilities of the different levels of government in a program implementation agreement. A substantial
investment in institutional and individual capacity building would be required.
The effectiveness of a household conditional cash transfer on health, nutrition, and education outcomes
will depend, in part, on the quality of the targeting and the human capital conditions. The conditions need
to be kept simple so they can be easily understood by beneficiaries and service providers alike, while also addressing
what are the key constraints for the poor in achieving better health and education outcomes. Evidence from several impact
evaluations suggests that CCT effects on enrollment, for example, are largest among children with lowest probability of
enrollment, and for transition grades with high drop-out rates (Schady, 2006; de Janvry, 2006). Targeting of poor under
CCTs in other countries has generally been through some combination of geographic and household targeting. But
although the targeting performance of CCTs has generally been better than other social assistance programs, they still
exclude some of the poor. Given Indonesia s poverty profile (described earlier), targeting poorer households will be
particularly challenging (see Targeting Section and Box 6.7).
Given the detailed design planning that would be necessary, and the operational and administrative
complexity involved, a CCT program would need to be introduced in phases. In Latin American middleincome countries, CCTs often began as pilot programs for extremely poor, rural and indigenous families, and were scaled
up gradually. For example, the largest program (Bolsa Familia, covering 11.1 million families in Brazil) was created
through the merger of four pre-existing cash transfer programs.
The design of the key program design features would be a complex task requiring sufficient time and
resources. While the experience of the UCT suggests that cash can be successfully transferred via post offices to people
throughout the country, making the receipt of a cash transfer dependent on meeting health and education conditions
would significantly increase the complexity of the program and introduce a very different dynamic. For example, a household
CCT would require: (i) a beneficiary roster with family demographic information relevant to the conditions; (ii) a policy
and procedures to verify compliance with conditions, as well as to deal with non-compliance; (iii) assessment of the
159
Given continued central level spending on health and education, it will be a challenge for the organization managing a CCT program to coordinate complementary interventions on the supply side to meet
increased demand.
200
availability of relevant health and education services, and a follow-up plan to address anticipated supply-side issues; and,
(iv) a complaints resolution and appeals system. Similarly, a community CCT would also require significant administrative
capacity, although it could build on the experiences of community driven development programs in facilitating communities
to plan and implement infrastructure projects. Both approaches would also require a monitoring (including MIS) system,
a socialization strategy targeted toward beneficiary households or communities and other stakeholders, and lastly, a set of
accountability mechanisms.
Any new program will come with implementation challenges, not all of which can be anticipated. What
would be key is whether there are mechanisms in place to identify and resolve issues early rather than waiting for them to
fester or deteriorate. This would be particularly important if the program were introduced with joint central-district
responsibilities. In addition to ongoing monitoring of program implementation, conducting quality control reviews and
independent spot checks periodically would also be important.
As a CCT program has never been tried in Indonesia, it would be important that government rigorously
assess the impacts of a CCT program before scaling it up. There have been high quality evaluations of CCT
programs in other countries that provide evidence of their
impacts. The success of CCTs in other countries is, however,
no guarantee that success will be reproduced with similar
results in Indonesia. It would be important for Indonesia to
use a similarly high-quality approach to assess the
effectiveness of program targeting and coverage, and to
determine whether there is increased utilization of health
and education services, and resulting higher education
attainment and improved health and nutrition status.
Process evaluation would also help policy-makers to
understand more about what works, what does not, and
why.
Workfare
One promising publicly provided option for
addressing the shocks to household incomes
shown to be a particular vulnerability of poor and
near-poor households is the deployment of
workfare or cash-for-work programs. Such
programs have worked well in low- and middle-income
countries with large informal sectors. They have been
instituted in several countries to respond automatically in
economic downturns, acting as a source of income when
reduced hours, reduced earnings, outright job loss and the
failure of small businesses become widespread. In this way,
cash-for-work programs can act as social insurance or
earning insurance in countries where highly structured,
institutionally intensive forms of traditional social insurance
are not viable (see Box 6.5).
201
Box 6.5 Can public employment programs act as social insurance for the poor and near-poor?
The risk of losses from unemployment is generally not considered insurable by private markets, since it can be highly systemicthat is, when unemployment strikes, say in an economic recession, a large number of individuals in the risk pool are affected.
Since there are typically not enough winners to compensate all the losers from the shock, it becomes too expensive for private insurers to cover losses.
However, household earnings protection is important, both for social protection and efficiency reasons: earnings protection can help households avoid
bad coping but also help improve matching of jobs with job-seekers by removing an element of urgency from the job search. For this reason, governments
help correct this market failure to provide or mandate earnings insurance instruments: from risk-pooling at the firm level in the form of severance
programs, to pooling across the working population in pay-as-you-go systems of unemployment insurance, and even systems based on individual
savings accounts where these are backed with minimum benefit guarantees financed by pooled funds.
However, in economies where large segments of the labor force are self-employed or work informally, providing earnings insurance
through any of these traditional devices is institutionally difficult and can even lead to regressive subsidies to relatively well off
formal workers. What is more, financing social insurance structures from pay-roll contributions draws a sharp distinction between the protections
enjoyed by workers with a legal contract, and those without, including the self-employed.
To surmount this problem and extend coverage of earnings protection, governments in many middle- and lower-income countries
offer public employment or cash-for-work programs. Since these are financed directly from general revenues and typically do not discriminate
according to the type of employment individuals have lost (i.e. whether legal employment, informal or self-employment), these structures can effectively
function as a social insurance instrument (i.e. a public risk-pooling intervention). Indeed, if correctly structured, these programs can be a form of social
insurance that is most likely to reach workers who lose employment in the informal sector or the self-employed whose businesses fail in a downturn (see
de Ferranti et al, 2000 and Galasso and Ravallion, 2001).
The critical feature of public employment programs that directly determines whether they succeed in acting as earnings insurance
for the informal sector is the program wage. Correctly setting the program wage is critical in ensuring that protection reaches those who need it
most, is readily available in economic downturns, and does not introduce damaging distortions when labor markets improve.
In India s new Guaranteed Employment Program , beneficiaries are offered benefits equivalent to the legal minimum wage.
However, the number of hours and workdays under the program, together with the number of beneficiaries in each district, has to be strictly limited to
contain costs. In effect, with regard to public employment programs, India has foregone unrationed, self-targeting forms of earnings protection, for
relatively expensive (in terms of above-market wage costs) forms of employment assistance that are strictly rationed and thus of limited coverage.
Indonesia has had experience in deploying public employment programs in the past that can provide
invaluable lessons to policy-makers. Several of the labor-intensive programs deployed as part of the JPS safety
net package of interventions remain to this day. Early evaluations of the JPS programs point to the public employment
programs as the best example of a safety-rope , i.e. a self-targeting program that was effective not only in responding to
the plight of the poorest, but also to the shock-management needs of the near-poor who benefited most from the
consumption-smoothing mechanisms these programs provided in the wake of the crisis (see Pritchett et al, 2001). However,
more recent evaluations of the labor-intensive programs show that these have significantly degenerated, focus on lowproductive work, and that their administration has fragmented among several public institutions that then misuse them for
political patronage. Many of the programs may have repeated the mistakes of workfare committed so often in other
countries, for instance by paying benefit wages greater than market wages. Conversely, some experiences, most notably
in Aceh, demonstrate how a public works intervention can succeed in simultaneously addressing the need to provide an
income to households, as well as rebuilding urgently needed infrastructure.
Key to the success and sustainability of such programs is the setting of the program wage at an
appropriately low relevant safety net wage for unskilled manual labor. The most effective way to ensure
public employment programs succeed in providing reliable and sustainable earnings protection to households that are at
greatest risk is to pay wages below the market wage and ensure that the work is relatively undesirable to the individual.
This way the program will be self-targeted to those who really need temporary income support. However, many governments
202
make the mistake of setting program wages at the legal minimum wage, well above the market wage for unskilled manual
labor. Offering above-market wages imposes three separate economic costs: (i) they attract more workers to public
employment programs; (ii) they pay each more than they would otherwise accept; and (iii) they crowd out private
employment. As many governments have discovered, if public employment programs offer (or subsidize) above-market
wages, the fiscal costs of these programs can increase unless protection is rationed (see Box 6.5). However, this places
government in the uncomfortable position of working through quantities rather than through prices, and hinders the
program s effectiveness at protecting households.
The main objective of the cash-for-work is to provide earnings insurance for the poor, not to tackle
general unemployment. As discussed in previous chapters, poverty and unemployment are not equivalent. Given that
the heads-of-household unemployment rate is merely 2.5 percent,160 this strongly supports the luxury good hypothesis
of unemployment. Indeed, poor heads of household simply cannot afford to be unemployed. As previously discussed,
poverty and unemployment are not equivalent (see Box 3.4 in Chapter 3). Instead, unemployment is predominantly a
problem of the educated, while poverty is mostly a problem of the uneducated. Unemployment is higher in urban areas,
while poverty higher in rural areas. Similarly, the unemployment rate is higher for women, while the relative risk of poverty
is somewhat gender neutral. As a consequence, the design of programs to provide social or earnings insurance for the
poor through cash-for-work is completely different to programs whose objective is to lower the unemployment rate: the
target groups are different, their main problems are different and the jobs provided should also be different.
The unemployment problem in Indonesia is mostly a youth unemployment problem and specific actions,
such as cash-for-training , are needed to address this problem. The number of unemployed youth has increased
steadily over the past decade, attaining 13 million in 2004. This indicates a youth unemployment share of 61 percent, far
higher than its population share of 18.4 percent. Programs to combat youth unemployment need specific actions. The
Joven programs that have been applied in many Latin American countries are good examples of cash-for-training for
youth. These are training programs for young adults aged 16 to 29 that provide practical training and financial support to
unemployed low-income youth in order to prepare them to compete in the labor market. The programs are demand-driven
and consist of a three- to five-month training course, followed by a three-month internship. Technical training and internship
experiences with employers are combined with basic life skills and other support services to ensure social integration and
job readiness. Private and public institutions contracted through public bidding mechanisms provide the training services
and are responsible for organizing the internships. The programs served a vast number of disadvantaged young adults:
Proyecto Joven in Argentina reached about 116,000 young adults from 1993 to 2000, and Chile Joven served nearly
165,000 beneficiaries between 1991 and 2001. The programs increased the probability of beneficiaries finding employment
upon graduation. The impact on employment measured against a control group was statistically significant, increasing
the probability of employment by about 10 percentage points in Argentina (Aedo and Nunez, 2001), 21 percentage points
in Chile (Aedo and Pizarro, 2004), and 7.5 percentage points in Peru (Deutsch et al, 2001; and IADB, 2005).
Workfare public employment programs can be seen as just another form of CCT where, instead of
cash-for-immunization or cash-for-school attendance, the government gives households cash-for-work.
Indeed, the new risk-pooling and prevention measures discussed above can be viewed as a package (albeit each with
specific implementation requirements as discussed below). The innovations on the CCT program discussed above that
would build on Indonesia s long tradition of community-led development efforts, might also include a cash-for-work
component where communities in which the supply of local services was found to be deficient could quickly build the
roads, schools or clinics necessary to provide the basic services that beneficiaries of the other transfers would eventually
demand.
160
Based on the old definition of those who do not have job and are looking for a job .
203
Health insurance
The one pervasive risk factor discussed previously in the risk and vulnerability assessment that would
not be directly addressed by the cash transfer programs above is health shocks. Among typically idiosyncratic
losses reported by households, the financial losses from adverse health events are the most frequently reported in Indonesia.
Among households headed by women, these are more frequently reported than economic risks or earnings losses. A
health component of a conditional cash transfer could help lower the likelihood of these adverse health events. However,
households are very likely to still face sickness and accidents that lead not only to lost days of labor and earnings, but can
also impose substantial treatment costs that most households have to pay out-of-pocket (that is, through financial and
non-financial savings instruments). The cost of care paid out-of-pocket can be impoverishing in and of itself in Indonesia
(Wagstaff and Pradhan, 2003) and elsewhere (see Baeza and Packard, 2005).
These findings would argue for the establishment of a well-functioning and efficient social protection
program in the form of health insurance for the poor. On this issue, Indonesian policy-makers have been correct
in identifying an appropriate area for intervention. Key to moving forward will be the need to build on emerging lessons
from practice to date. While past evaluations of the health-card scheme, in which the government extends entitlement to
treatment to the poor, found a positive impact on utilization (Pradhan et al, forthcoming), it is not yet known whether the
new social health insurance program introduced following the first partial repeal of the fuel subsidy in March 2005 will
have the same effect. Through the new program, about 60 million households were given health insurance cards entitling
them to receive free healthcare at local public health clinics, as well as in third-class public hospital beds (see Box 6.6 on
the PKPS-BBM program for healthcare). If this extension of coverage were to increase utilization, the government may
have succeeded in extending vital risk-pooling against the financial losses from adverse health events to the vulnerable.
(See also Chapter 5 on Public Spending for analysis on demand-side programs for improving the poor s access to health
services.)
It will be critical to review the impacts of the health insurance program. Early assessments point towards
scope to improve the design of this program in numerous areas. One curious feature of this expansion in health coverage
was that financing for the measure was channeled through a traditional social insurance institution PT Askes, a healthinsurance provider for Indonesian civil servants. While the overall impact of extending publicly provided risk-pooling for
health losses may prove positive in time, the measure could probably have been done in a more efficient way. Channeling
such a social protection program of risk-pooling for the poorest through a traditional social insurance institution such as
PT Askes departs dramatically from international experience. It will also be key to determine whether increased access
actually leads to an increase in utilization by the poor,
and whether increased use of public facilities leads
to increased utilization overall or just decreased use
of non-government private providers. Addressing
these issues in improving the program going forward
will be critical to ensure its efficacy as a cost-efficient
social protection for the poor and near-poor in the
future. (See Chapter 5 on Public Spending and Box
6.6 for further analysis.)
204
Box 6.6 Improving poor people s access to healthcare: Concerns with the PKPS-BBM health insurance scheme
In 2005, the government introduced a healthcare scheme aimed at increasing access and service quality to all Indonesians, and
particularly the poor. The program provides: (i) free-of-charge healthcare services at Puskesmas (public health centers) and (ii) in-patient treatment
in third-class hospital beds for the poor. The Puskesmas component of the program is very similar to its predecessors, the JPS-BK (1998-2001) and
PKPS-BBM (2001-05), in that it provides supply-side block grants to Puskesmas for operational requirements, equipment and medicines, and free
access those poor who have health cards. The difference, however, is that the in-patient treatment component of the program (60 percent of funding) is
now run by PT Askes, whereby individual health cards are distributed by this insurance company and hospitals are reimbursed for their services on a feefor-service basis. The program is budgeted at around US$400 million across 2005 and 2006, equivalent to about one-fifth of Indonesia s public health
spending in 2004.
However, the program does have many problems that need to be addressed to ensure the poor benefit. Some of the issues and concerns
highlighted by a preliminary assessment of the program include:
(1)
The targeting mechanism of the program is problematic. The insurer, PT Askes, was given responsibility for targeting the poor and for
socializing the program, but given how they are funded (see point 2 below), it was not in their financial interest to reach poor households. There
have also been difficulties due to the changes from household-based cards used in previous programs to individual Askeskin cards. Results of field
assessments show that in some places local implementers have questioned central poverty data and, as a result, have postponed health-card
distribution, being afraid of repercussions from disgruntled community members. Self-targeting has been allowed through a SKTM (a letter from
the neighborhood head stating that a person is poor), which has rendered the costing of the program unpredictable. It is also unclear what
indicators local officials use to determine who should receive the SKTM letter, and therefore whether they are indeed poor.
(2)
The program is, in essence, supply-driven whereby funds flow through the providers and the poor are not empowered to
question the quality of services they receive. Receiving services is a favor rather than a right for the poor under this scheme. There are no
incentives for Puskesmas staff to improve service quality for the poor, since they are not compensated on a fee-per-service basis. Even PT Askes,
which compensates hospitals on a fee-per-visit basis, receives funding from government in block grant form regardless of the quantity or quality
of services provided to the poor through the hospital system. Given this supply-side flow of funding, the system depends solely on the altruistic
behavior of providers and insurer to improve quality of services to the poor.
(3)
The program does not enable the poor to utilize the services of non-government or private providers, such as midwives and
paramedics in villages services that are in high demand by the poor. Although the intention of the program was also to cover the
services of private providers in rural areas, the mechanism through which this would take place is not outlined in the program; private providers are
generally not included in the scheme. With the majority of poor people in Indonesia preferring to use private providers, this has serious consequences
in terms of increasing access to and utilization of healthcare services by the poor.
(4)
The program does not take into account regional variations. The prevalence of certain illnesses and difficulties in accessing health
facilities in isolated areas is not taken into account. In addition, the design fails to accommodate the role of local government in planning,
implementation, monitoring and evaluation. Worse still, the program undermines some of the existing health insurance schemes previously started
by some forward-looking local governments.
Going forward, greater impact may be possible if program design takes into account incentive problems and program limitations.
For instance, private providers need to be included in the scheme and this would be possible if funding were channeled to the poor through alternative
mechanisms. Even a voucher system similar to the health card that operates on a fee-for-service basis would imply a change in incentives for the service
provider. The funding that goes through PT Askes also needs to be carefully monitored, as too much discretion is left with the company in terms of
targeting the poor, as well as providing reimbursements to hospitals.
* Regulations for the PKPS-BBM health program changed again in 2006 reverting to the first half of CY05 where almost all funds, including for Puskesmas, are channeled through PT Askes. Certain
out-patient and other services (such as ambulatory care, birthing and midwifery services) are also reimbursable (whether from a Puskesmas or hospital).
205
Rice price stabilization

Maintaining stable rice prices is critically important for the poor. As noted in Chapter 2, high rice prices were
very harmful to the poor during the crisis in 1997/98 (see Figure 2.4). More recently, the 33 percent increase in the price
of rice between February 2005 and March 2006 was primarily responsible for the increase in the poverty headcount in
2006 (see Chapter 3 Box 3.5 for more information on the impact on the poor). Policies to stabilize rice prices are therefore
an important part of social protection.
Before the economic crisis, Indonesia had one of the best records for stabilizing rice prices in East
Asia. Figure 6.2 shows that the real price of rice in Indonesia was held roughly constant for a period of 20 years prior to
the crisis. Aside from the turbulence of the early 1970s, Bulog was successful in maintaining the real price of rice in
Indonesia at or near the world price for a period of two decades. During this period Bulog had an explicit mandate to
ensure price stability and was given monopoly import rights to enable it to achieve this. Growing corruption within Bulog,
together with the economic crisis, led to the collapse of the rice price stabilization system. There followed a brief period
of free trade in rice between January 1999 and December 2003, initially with no tariff and then with a specific tariff of Rp
430/kg from January 2000. Finally, in January 2004, the government announced a seasonal import ban, which has been
repeatedly extended so that it has effectively become a permanent ban.161 Meanwhile, Bulog has become a state-owned
enterprise and claims to no longer have a mandate to stabilize prices. It does, however, have responsibility for purchases
and sales from Bulog stocks (currently the only mechanism of price stabilization), as well as implementing the rice for the
poor (Raskin) program (see Red Spotlight on Inefficiencies and Leakages: Where has all the Raskin gone?).
Figure 6.2 Real rice prices were stable before the crisis
Source : Palacios and Pallares, 2000.
The current system for preventing falls in the price faced by farmers is not very effective. The government
continues to set an official government purchase price for rice (HPP), which dictates the price that Bulog pays for paddy
and rice. Bulog purchases around 2 million tonnes each year, primarily for the Raskin program. Since this constitutes
161
Decree No. 9/MPP/Kep/2004 allows the approval of one-off imports to ensure food security. This has been used on a number of occasions, but the overall level of imports since 2004 has been much lower
than in previous years, in part because of the ban.
206
around 6 percent of supply, government purchases could, in principle, have a significant impact upon prices. Whether
such purchases do so or not depends on what the private sector would do in the absence of government purchases. If it
simply increased purchases by the same amount, then government purchases would have no effect on prices. If private
purchases remain unchanged in the absence of government purchases, then GKG prices could fall by up to 20 percent.
However, the experience of 2003 and 2004, when the market price for GKG was frequently below the HPP, suggests that
government purchases are not very effective at defending the floor price.
And the current system for preventing large increases in the price is seriously flawed. The rice import ban
prevents imports acting as an automatic price stabilization mechanism. In its place, the government has a mechanism to
release rice from its stockpile in locations when rice prices are rising too quickly (operasi pasar, or OP). However, the OP
mechanism currently suffers from two drawbacks. First, it offers little relief from high prices since the quantities dispersed
are rather small. Second, and more seriously, the formula used to determine whether an OP is necessary is seriously
flawed. The formula recommends OP when rice prices rise by more than 25 percent above the average price of the
previous three months. This in theory could allow rice prices to rise by 533 percent in the course of one year without
triggering OP!162 If the government wishes to use the OP mechanism to stabilize prices then a more appropriate formula to
prevent sudden increases in rice prices would be to set a rice price ceiling at a certain percentage above the world price for
rice and to allow imports for a set period when the price of rice exceeds this ceiling.
The most effective way to prevent large increases in rice prices would be to allow the general importation
of rice subject to a low tariff. This would prevent domestic prices from rising significantly above the tariff-inclusive
price. It would also be more efficient since it would allow competition between different importers. During the brief period
after the crisis when Indonesia enjoyed free importation of rice, domestic rice price fluctuations were just as stable as the
period during which Bulog managed rice prices. Of course, it is still possible that domestic rice prices could increase
sharply if a fall in local production coincided with a sharp increase in world prices or a major depreciation of the exchange
rate. However, the world rice market is significantly larger and more stable than it was even 10 years ago, and the
macroeconomy is significantly more stable too. Even if such an event did occur, a revised system of operasi pasar could
compensate for temporarily high world prices.
Social insurance
As part of Indonesia s social protection agenda much effort has recently focused on social insurance.
Recent legislation may imply fundamental changes to traditional (i.e. pay-roll tax financed) social insurance in Indonesia.
However, these changes are expected to be slowly phased in over the next five years. Indonesia s current formal social
insurance programs covering the cost of healthcare, old-age income needs, the risk of disability, death of income earners
and job loss are fragmented. Coverage is low and deeply segments the formal labor force (Martineau and Geurard, 2005;
Arifianto, 2004). Currently, there are four main providers of traditional social insurance: Asabri, which provides pensions
and health cover for the armed forces and police; Taspen, which provides pensions to civil servants; PT Askes, which
provides health cover to civil servants and has recently expanded its activities to include subsidized healthcare to the
poor; and Jamsostek, which primarily provides old-age income support through a provident fund plan but also makes
disability, survivor and health insurance available to private sector workers. These plans are complemented, and in many
cases substituted, by a voluntary private pension sector providing a number of defined-benefit and defined-contribution
162
Peraturan Menteri 22 / M-DAG / PER /10/ 2005 specifies that OP will take place when the price rises for a week by more than 25 percent above the normal price, where normal is the average price of medium
quality rice in the three months before. Even if rice prices rose steadily by 15.5 percent per month, this rule would not be triggered.
207
products to workers through their employers. Furthermore, Indonesia s labor law requires complying employers to pay
large lump-sum termination benefits.
However, it must be recognized that in Indonesia s current economy traditional social insurance will have
limited coverage and be of little relevance to poorer households. The vast majority of Indonesias labor force as
many as 60 million workers, and possibly more work beyond the reach of government-mandated social insurance. Many of
these workers either cannot or choose not to rely on voluntary formal pension plans. The results of the risk and vulnerability
analysis show that institutionally intensive forms of pay-roll-tax-financed social insurance do not reach the poor or the nearpoor, and make little difference in mitigating the impact of shocks. Moreover, the risk factors in old age that typically concern
policy-makers, and are one focus of the social security bill, do not seem to be contributing to household vulnerability in
Indonesia. As shown in Figure 6.3, which presents data on the reach of pay-roll-tax-financed social insurance in a wide
selection of countries, these results should come as little surprise given Indonesias level of development.
Figure 6.3 The reach of traditional social insurance: participants in the labor force by per capita income, selected countries circa 1995
100
100
Netherlands
United States
90
90
80
80
Contributors / Labor Force (%)
Contributors / Labor Force (%)
70
70
60
60
South Korea
50
50
Malaysia
40
40
Turkey
Sri Lanka
30
30
Philippines
20
20
China
Indonesia
Indonesia
India
Vietnam
Pakistan
10
10
00
0
10
10
15
15
20
20
25
25
30
30
Income per capita (PPP US$ 000)
Source: Palacios and Pallares, 2000.
This does not mean that Indonesia should not proceed with caution in developing a social insurance
scheme for the future. Immediately addressing the weaknesses of current traditional social insurance structures,
which cover such a small segment of Indonesian households yet pose risks for the broader economy, would be prudent.163
However, an effective social protection strategy for today s Indonesia, one aligned with the main risks
and vulnerabilities faced by the population and consistent with administrative capacity, would suggest
a relatively modest role for traditional social insurance. Keeping the above points in mind, it is primarily
163
Indeed, addressing the issue of traditional social insurance should not be ignored altogether. The new law is a fact on the ground that has to be examined. A critical place to start this examination is with an
actuarial and institutional assessment of the existing social insurance providers, Taspen, Jamsostek, PT Askes and Asabri. Even without the nominal extension of coverage promised in the law, these institutions
may pose a fiscal risk for the government that needs to be quantified. Doing so will involve a long, and necessary program of analysis that in addition to providing a valuation of current contingent liabilities and
those implied in the new law, would enormously increase the governments capacity to craft policy in social protection.
208
important for policy-makers to focus on covering the losses that are currently the most likely to impoverish households
and threaten their human capital, as well as those that deepen the poverty of the current poor or prevent them from
overcoming their poverty. The analysis of risk and vulnerability showed that households are most affected by catastrophic
(and covariate) shocks best covered by simply designed, broad, national-level risk-pooling instruments, rather than the
sophisticated, closed forms of traditional social insurance found in middle-income and OECD countries. In the medium
term, a social protection strategy for Indonesia should thus focus primarily on the more generic three prongs of a social
protection program laid out above rather than focus excessively on the development of highly structured social insurancetype instruments, such as contemplated in the recently passed National Social Security Law. Current low administrative
capacity makes such instruments impractical and may even impose distortions on the broader economy.
Targeting
A keystone to targeted social protection programs is a quality targeting system. Given its budget
limitations, the government needs to ensure that poverty programs are well targeted. Countries have
applied a variety of targeting instruments for social programs, including household (or individual) assessment mechanisms,
broad categorical eligibility and self-targeting. Many programs adopt a combination of these three mechanisms. Health
cards and Raskin are examples of programs using household assessment, where each household or individual is assessed
for eligibility. IDT, KDP and UPP164 are examples of programs using category targeting, in this case the geographical
location of poor communities. Workfare programs with low wages and price subsidies for low-quality items are examples
of programs using self-targeting, as the benefits offered are only likely to be attractive to the poor.
Figure 6.4 The targeting performance of various targeted poverty programs in Indonesia are only slightly pro-poor
Scholarships Program (2004)
Health Cards (2004)
Raskin Program (2004)
UCT Program (2005)
Source: Susenas 2004, reflecting the distribution in 2003 for Raskin, scholarships and health cards.
Note: For UCT targeting, Susenas 2005 panel data are used for preliminary findings.
164
IDT or Inpres Desa Tertinggal is a village infrastructure development program; KDP or Kecamatan Development Program is a community-based development program in rural areas; and UPP or Urban
Poverty Project is community-based development program in urban areas.
209
Historically, program targeting in Indonesia has not been particularly strong. As discussed above, evidence
from various targeted programs (Figure 6.4) shows that the targeting performance of various safety net and poverty
programs was low. The under-coverage error-the proportion of the first quintile of households that are not included in the
program is generally over 50 percent. Similarly, the leakage error the proportion of beneficiaries who are classified as
non-poor or above the bottom quintile-is around 50 to 70 percent. Several elements of safety net program design explain
the reasons for less-than-efficient targeting of the poor.
First, this weak performance is linked to problems with the use of the National Family Planning
Coordination Board (BKKBN) system of classifying households as prosperous families and preprosperous families . This classification system is in principle a proxy indicator based on a range of variables
(food consumption, material of the floor, type of healthcare services, ownership of clothes, religious practices, etc).
Unfortunately, the BKKBN indicators were chosen not based on a proper statistical exercise to find a set of variables
and a scoring system that could discriminate between the poor and non-poor, but solely on their being the only data
available up to the village level. Most of these indicators are not easily observable and could be easily manipulated. The
indicators also include non-economic criteria, such as the ability to meet religious obligations. Moreover, the data were
compiled by relatively poorly trained volunteers. As a result, the BKKBN indicators have only a weak correlation to the
poverty expenditure indicator, with 43 percent of pre-welfare households not among the expenditure poor.
Second, as discussed, weak targeting has also been induced to some extent by cultural norms,
which foster sharing (bagi rata) or spreading program benefits at the community level. Such practices
have contributed, for example, to the benefits of the Raskin subsidized rice program being spread broadly and thinly.
Given Indonesia s geographical diversity, there is scope to build
on both geographic targeting of areas, as well as systems for
household targeting. Geographic targeting has become easier to implement
in recent years, as a result of the widespread availability of household survey
and census data, and has overcome many of the data problems that still hinder
the implementation of other forms of indicator targeting. By combining census
with household survey data, estimates of poverty down to the sub-district
level, and in some cases, even at the village level can be made (Elbers, et al
2002). A study by Alatas (2004) shows that the coverage of the poor
(headcount) by the KDP I program could have been improved from 30 percent
to 52 percent if such small area estimation maps had been used for determining
its first-stage allocation. Indonesia has now completed small area estimation
or poverty maps for the entire country by combining the information from
Susenas, the Village Potential Survey (Podes), and the population census,
and processing this information based on an econometric model (see Green
Spotlight on Innovation at the end of this chapter). These poverty maps are
able to provide reliable data on the number of poor people down to the village
(kelurahan) level for districts in densely populated Java and to the sub-district
(kecamatan) level for off-Java districts. The existence of these maps opens
up huge potential for using such maps for targeting and placement of public
investment.
Involving the community can improve targeting effectiveness.
Current and past Indonesian welfare programs use a mixture of geographic
targeting, household assessment and community participation to distribute
210
benefits. Rather than targeting the poor through administrative measures executed by program administrators, these
programs make the community central in delivering the project, including having a role in selecting the beneficiaries.
Communities may have good information about the poor and the kinds of life shocks they experience. The community can
also play an active role in monitoring programs since they live where services are delivered and can even be involved in
delivering the benefits. However, the community is not a single entity. Different stakeholders have different concepts of
what constitutes poverty. Selecting beneficiaries becomes a collective-action problem and compromises the position of
key community members. At the same time, connections through kinship and networks play a role in community decisionmaking (Alatas, Pradhan and Rao, 2006). In such a set-up, it is important to balance the gains from utilizing local
information with the possible costs of local rent-seeking and elite capture of benefits. Conning and Kevane (2002) point
out that community targeting can be more effective than outside agency when they have egalitarian preferences, open and
transparent systems of decision-making, and clear rules for determining who the poor are. The best system should
include hybrid mechanisms whereby the center defines and monitors targeting instead of solely relying on the community
for control and selection.
The quality of household targeting systems depends on the quality of the management information
system. Several issues need to be taken care of in designing a successful household targeting system. It is important to
have a consolidated national database to avoid duplications and track beneficiaries. This database should be designed
flexibly enough to serve multiple social programs. In addition, proper identification of individuals/households, preferably
through unique social identification numbers, is crucial in linking registry information and beneficiaries with other systems
and programs. While the actual design and implementation of household targeting systems vary significantly by country,
most systems involve the following steps: (i) collecting data on specific (potentially eligible) households via interviews
(and sometimes home visits) using pre-designed questionnaires (which depend on the type of household assessment
mechanism); (ii) entering these data into a unified household information registry (with varying degrees of verification
and consistency checks); (iii) comparing household characteristics with pre-established eligibility criteria (programspecific); and (iv) establishing program-specific beneficiary lists (sub-registries) for the purposes of program
implementation and benefit payment (Castaneda and Lindert, 2005).
Indonesia took a big step forward recently in attempting to put together such a unified targeting registry
with the compilation of a household database for the UCT program. In order to target the UCT program,
Indonesia rapidly put together a database of poor and near-poor households using a combination of geographic, communitybased and household targeting techniques. Initially, 15.5 million households were distributed cards, but later around
600,000 cards were cancelled following verification of eligibility. Upon opening up the registry for additional household
applicants, an additional 3.7 million households of a total 12.2 million applicants were entered into the registry and will
be beneficiaries, bringing the total registry up to 19.2 million households.
Indonesia s rapid development of a targeted household registry has imparted some key lessons and an
assessment shows scope for improvement. To some extent, the necessarily rapid roll-out of the targeted registry
led to some inadequacies in the process of developing the household registry and to some shortcomings in the resulting
database (see Box 6.8 on the poverty-targeting database). In particular, analysis of recent data as part of the assessment
of the UCT program shows that there is substantial room for improving targeting. Both the under-coverage error and
leakages error are high. Whereas the UCT program was targeted to the lowest 30 percent of households, survey data
analysis indicates that only 45 percent of this group of households received the benefits, translating into 55 percent of
under-coverage error. At the same time, the leakages error 50 percent of the upper 70 percent of the population received
benefits. By international standards, this would indicate room for improvement (see Table 6.7). Nevertheless, the targeting
performance of the UCT program is better than that for health cards. While health cards can potentially reach 100 percent
of the poorest decile, in reality only around 22 percent of them actually received cards.
211
Table 6.7
International comparison of targeting performance for cash transfer programs

Share of transfers going to :
Country
Name of program
Poorest
10 percent
(%)
Poorest
20 percent
(%)
Poorest
40 percent
(%)
Source: Coady et al, 2004. Data for Indonesia are extracted from Susenas 2005 panel data and represent the percent of poorest households. The data for Brazil are taken from World Bank, 2006.
Some concrete steps can be taken to improve the UCT program database to make it the basis of a
unified household targeting system. Close analysis of the database and the methodology used to derive the targeting
mechanisms, while pointing to some of the weaknesses discussed, also point to some clear solutions: (i) devise a better
scoring system, a better questionnaire and weights; (ii) combine proxy means test (PMT) targeting with geographic
targeting by using geographic quotas based on estimates from poverty mapping; (iii) re-verify the eligibility of current
beneficiaries based on the new scoring system; (iv) do an additional survey sweep in sub-districts with high poverty
densities and high exclusion rates; and (v) resolve the ID problem. Coming up with technological solutions, such as
fingerprinting, may be necessary in order to resolve problems of identification.
Box 6.7 Poverty targeting database

Given the imperatives of rolling out the UCT program in conjunction with the fuel price increases of October 2005, the government
worked rapidly with the BPS to compile a poverty-targeting database. This database was compiled using the following steps. First, survey
rosters were developed through community-level identification of poorer households (usually by the administrative head of neighborhood). Second, a
proxy means test of 14 questions was administered to these identified households. Third, these data were centrally analyzed and scored. Fourth, a cut-off
point was drawn that identified beneficiaries. Fifth, in distributing beneficiary cards and vouchers to these households, enumerators were to verify that
households met the criteria.
While all agencies involved are to be commended for the rapid roll-out of the UCT program, the expedited schedule did result in
some shortcomings, both technical and institutional. The BPS put a commendable amount of thought into the task, including selecting the
variables using discriminant/logit models and taking into account regional disparities by doing the analysis separately for each district. However, due to
time constraints, some sacrifices were made and resulted in discrepancies in comparisons with the gold standard of proxy means testing. The weights
were not determined based on the results of the models but instead based on the standardized value of the mean of the selected 14 variables for the poor
households. Due to time limitations in conducting the survey, the BPS also rationed the length of the questionnaire into a one-page questionnaire,
limiting the number of variables and sacrificing accuracy in the process. Moreover, there was often not time for enumerators to complete the secondary
verification. The high under-coverage of the poor also indicates that the premise that community leaders know better the local conditions of community
members and are prepared to convey accurate information needs to be re-evaluated.
After the resulting database is verified, updated, and refined, it is important that it is maintained as an
organic list that is frequently updated. Transitions in and out of poverty are common in Indonesia, as shown in
previous sections. Verification, updates and re-certifications are important for tracking fraud, and allowing for turnover in
beneficiaries. If the database is to be used more widely for sector programs targeting specific demographic target groups
(e.g. infants, the disabled, children and the elderly), it will also be important to collect basic information about household
212
members in the form of a household roster. In the future, this organic database and ranking can also provide the government
with a continuous ordering of the poor and near-poor with various income and non-income dimensions of poverty to
serve multiple social programs with differing thresholds for eligibility. Database management should be designed to
allow flexible responses to changing policies and updates with the pre-testing of systems, well-designed manuals and
adequate training for users.
Clarity on the objectives of different social protection instruments is critical in determining what type of
targeting is appropriate and will lead to greater effectiveness. For example, for interventions aimed at helping
households manage the acute losses from economic risks such as labor-intensive and cash-for-work, self-targeting is the
most effective way to ensure that benefits reach those most in need. This type of risk-pooling mechanism will fail in
meeting its consumption-smoothing objective (objectives (ii) and (iii) in Section IV) if it relies heavily on any type of
administrative targeting, no matter how efficient. In sharp contrast, for those elements of the package aimed at preventing
the longer-term, chronic disadvantage and vulnerability of households (objective (i) in the list in Section IV), administrative
targeting is highly effective.
Conclusion: Towards a Third-Generation System
Almost half the population of Indonesia is either poor or vulnerable to poverty. With 49.0 percent of the
population living on or below the US$2-a-day level, many Indonesians are vulnerable to shocks that could either make
their poverty deeper, or drag them from near-poverty into poverty. While this in itself is of major concern, policy-makers
in Indonesia also worry that shocks to poor or near-poor households can force them into using bad coping strategies that
erode their human capital and make it harder for them to escape from poverty subsequently. Therefore, protecting the poor
and near-poor in Indonesia from shocks is an important component in any comprehensive poverty reduction strategy.
Having experienced a highly turbulent decade triggered by the 1997/98 financial crisis, there is now growing interest
among policy-makers in Indonesia in establishing a national social protection system to protect the vulnerable from
future shocks and thereby contribute towards Indonesian poverty reduction goals.
Indonesia now has a window of opportunity to establish a well-designed social protection system. The
combination of the governments decision to reallocate fiscal resources away from regressive universal commodity subsidies,
together with the growing recognition even in the wake of the crisis that there continues to be a critical need to protect
the poor and near-poor from risks and lower their vulnerability, provides Indonesia with a unique opportunity. It is now
possible to start to put into place a well-designed and effective social protection system that addresses the specific
problem of vulnerability in Indonesia.
The time is right for Indonesia to move on to a third-generation social protection system. Similar to many
developing countries, Indonesia has traditionally chosen to use universal commodity price subsidies, particularly through
fuel products, as its major social protection strategy. Later, from the crisis through to 2005, the government employed a
social protection system that was a mixture of universal subsidies and targeted safety net programs developed in the wake
of the crisis. These short-term temporary targeted programs, such as Raskin (subsidized rice for the poor), educational
scholarships and health cards, can be considered an attempt to establish a second-generation social protection system
better geared to the needs and vulnerabilities of the poor. While undoubtedly an advance on universal subsidies, evaluation
of these second-generation programs indicates that they require considerable improvement if they are to continue, as well
as an improved performance in targeting. However, with the move away from burgeoning fuel subsidies in 2005, Indonesia
is now ready to move on towards a more effective third-generation social protection system that is more comprehensive
and better matched to the actual risks and vulnerabilities faced by Indonesia s poor and near-poor.
213
The first step in developing a coherent social protection system is to understand the risks and
vulnerabilities faced by poor and near-poor households. Section III of this chapter undertakes a risk and
vulnerability assessment (RVA) for Indonesia. The implications of this RVA are important in pointing to ways of designing
social protection policies for Indonesia. There are four primary findings: (1) given the similarities between the poor and
the near-poor, both of these groups can be considered as one in addressing broad risks and vulnerabilities; (2) both the
vulnerability profile and an assessment of the nature of shocks point to four primary sources of vulnerability that should
be the focus of any social protection system in Indonesia. First, the risks to earning capacity and income are significantly
higher among poor households. Second, there is evidence of relatively higher risks to children s future earnings ability
and welfare following shocks among poor and near-poor households. Third, health shocks are a particularly important
source of vulnerability among poorer households. Finally, a further significant shock reported by households comes from
losses stemming from changes in government regulations; (3) there is little evidence that the risk factors that arise with
ageing are a major factor in contributing to household vulnerability; and (4) greater access to safe, competitively provided
financial forms of saving, and formal risk-pooling is likely to aid household risk management considerably. This is
because currently shocks among poor and near-poor households lead to the most severe cuts in food spending (i.e. bad
coping strategies).
There needs to be a closer connection between social protection objectives and programs. The objectives
of Indonesia s crisis-era social safety net (JPS) are generally well-aligned to: (i) alleviate poverty by raising the consumption
levels of the poor; (ii) prevent the near-poor from falling into poverty; and (iii) help poor and near-poor households to
mitigate losses of income in the wake of shocks. However, the structure of the JPS muddles these objectives. Consequently,
there is currently a mismatch between needs and many existing social protection programs. Looking forward, while the
three objectives are sound, greater clarity is required on how each is best pursued.
Towards a national social protection system that makes sense for Indonesia. There is a clear need for more
effective, public risk-pooling mechanisms to augment household risk management options, given the limited reach of
private formal risking-pooling. Second, there is evidence that households would benefit from more efficient provision of
privately provided risk-pooling and savings instruments, in particular safe but sophisticated and competitively priced
financial instruments. Third, the chronic disadvantages that make the poor and near-poor most vulnerable call for more
extensive and effective prevention efforts. The first and the third of these three areas point towards the implementation of
a three-pronged social protection strategy, as follows.
A four-pronged approach would form the core of a national social protection strategy. These four prongs
would be: (i) a conditional cash transfer (CCT) program ( cash for education and health ); (ii) a workfare program ( cash
for work ); (iii) a health insurance program; and (iv) rice price stabilization. The first addresses the bad coping that can
hinder human development outcomes. The second protects the poor against earnings and income shocks that are all too
common for those working in informal and self-employed jobs, The third protects from one of the most frequently reported
shocks by the poor, namely health-related shocks. The fourth prevents large increases in rice prices that are so harmful to
the poor.
The government will also need to consider which existing social protection and social safety net programs
it may need to cut. In order to develop the above three programs it will also be important for the government to
streamline Indonesia s existing programs in order to better address the most important risks and vulnerabilities that affect
the poor and near-poor. Programs that do not address the major risk and vulnerability issues, such as some small safety
net, subsidy and transfer programs, should be reviewed and possibly phased out.
214
Spotlight on Innovation
Poverty maps: A powerful tool in targeting the poor
What are poverty maps?

Poverty maps are a visual presentation of the spatial incidence of poverty that
can facilitate more finely focused pro-poor interventions and guide the allocation
of public spending to reduce poverty. A poverty map will be most useful if it
can provide information at a fine level of geographic disaggregation. The
innovation of the small-area estimation method enables the creation of
statistically accurate maps at much smaller administrative units than those
given by household surveys or composite indices, such as the Human
Development Index (HDI) or other basic needs indicators.
Small-area estimation maps
Small-area estimation maps create detailed maps that that are accurate to the
sub-district level. The unique aspect of this small-area estimation method is
that it enables the creation of statistically accurate maps at much smaller
administrative units than first generation poverty maps, which relied upon
composite indices such as the HDI or other basic needs indicators. However,
small-area estimation is technically more complex to undertake than mapping
using composite indices.
The technique involves combining census data with household survey data,
such as the Susenas. A model that identifies a relationship between Susenas
and census variables is created and then applied to the census data, since they
cover many more households than the Susenas. These household data can
then be aggregated into sub-district findings and applied to a mapping program
to show spatial distributions of poverty among all sub-districts in the country.
In Indonesia, poverty maps reliable down to the kecamatan level and even the
kelurahan level for Java have been developed by a small number of
organizations, including the SMERU Research Institute and the Central Bureau
of Statistics (BPS). While their development requires skills in the areas of
geographic mapping and statistical analysis, the most severe constraint to date
may be lack of funding and lack of time. Poverty maps can take up to six
months to create and are expensive, and both SMERU and BPS received external
funding.
However, poverty maps have limitations. While they are effective at identifying
poverty correlates or geographical targeting of programs, they do not identify
the causes of poverty, and are therefore less useful for project design.
Percentage of poor people by district in Indonesia, 2004
Poverty Headcount (%)

<5
6 - 10
11 - 15
16 - 2 0
20 +
215
How are poverty maps used ?
Future uses of poverty maps
Internationally: Poverty maps can be used to promote greater transparency

and accountability. In Brazil, Panama and Nicaragua widespread
dissemination of poverty maps to libraries, government agencies and research
institutes has sparked debate regarding spending and poverty. They can
also be used to respond to emergencies, such as epidemics. In KwaZulu
Natal province, South Africa, a cholera epidemic was successfully contained
in 2001 by combining poverty maps with information on sanitation and safe
water availability. Poverty map analysis showed that the epidemic was
spreading through the river floodplain and this provided the basis for a
public health campaign that resulted in a fatality rate of 0.22 percent-among
the lowest ever recorded. In Mexico, poverty maps have proven to be a
powerful tool in analyzing and helping improve the accuracy of central
government transfers to localities throughout the country. The use of poverty
maps and income maps has replaced the use of variables from the census
data and this has demonstrated that financial transfers to localities bear
little relationship to their poverty levels. The use of poverty maps illustrated
this anomaly for the first time.
As discussed, poverty maps have been used to improve the unconditional

cash transfer beneficiary database. Poverty maps were used to allocate a quota
of beneficiaries at the kecamatan or kelurahan levels, to ensure that benefits
among the kecamatan are accurately distributed.
Indonesia: By including data from the BPS poverty map with data on the
environment and infrastructure, the Center for Local Government Innovations
(CLGI) has generated a poverty map of Sulawesi that assists the Luwu Utara
district government to plan and implement better-targeted poverty alleviation
interventions by linking program spending to local poverty rates.
Similarly, the local government in Riau province, with technical assistance
from BPS, has created an impressive composite poverty map that provides
geographical, land use and poverty information for all sub-districts in the
province. Information is based on BPS 2004 data, making it the most current
map in the country. The provincial government paid for these maps, as it is
both relatively wealthy and open to trying new techniques. These maps are
used by local government to promote improved poverty targeting and service
delivery.
Poverty maps could also be used for project implementation. For example,
the recently announced national community empowerment program that will
provide small block grants to villages throughout the country could benefit
from the use of poverty maps to help identify the poorest sub-districts in the
country in a more systematic manner. Recent studies indicate that while the
allocation of resources under the existing Kecamatan Development Program
(KDP) is pro-poor, the targeting of resources could have been improved further
through an improved regional targeting strategy using poverty maps (Alatas,
2005).
Local governments and NGOs could use poverty mapping to target their
activities more precisely in the future. Many local governments recently
interviewed had not heard of poverty mapping techniques. Their routine use
could enable institutions to more precisely and cost-effectively target the poor.
Sources:
(a)
(b)
(c)
(d)
(e)
(f)
(g)
(h)
(i)
(j)
(k)
(h)
(i)
216
Alatas, Vivi. (2005). An Evaluation of Kecamatan Development Project. Mimeo. World Bank
Office Jakarta.
Ahmad, Yusuf, and Goh, Chorching. (2006). Indonesia s Poverty Maps: Impacts and Lessons
(draft mimeo). Washington, DC: World Bank. May 2006.
Coudouel, Aline. 2006. Poverty Maps for Policy-making: Beyond the Obvious Targeting
Applications (presentation). Washington, DC: World Bank.
Henninger, Norbert and Mathilde, Snel. Where are the Poor: Experiences with the Development
and Use of Poverty Maps. World Resources Institute, Washington, DC. 2002.
Interview with Dipayan Bhattacharyya. World Food Programme. 5 December, 2005.
Interview with Bina Desa. 25 November, 2005.
Interview with Eko Susi Rosdianasari. Information Specialist, Centre for Local Government
Innovations. 10 December, 2005.
Interview with Dedi Walujadi. Central Bureau for Statistics (BPS). 12 December, 2005.
Interview with Mario Boccucci. World Bank,.15 November, 2005.
Snel Mathilde, Ballance Anna. Workshop on the Impacts of Poverty Maps: Past Experiences
and New Applications. 2004.
Szekely, Miguel. 2006. Income/Consumption Maps: A Powerful Tool for Increasing Policy
Effectiveness (presentation). Washington, DC: World Bank.
Vishwanath, Tara. Poverty Maps: Uses and Caveats (presentation). Washington, DC: World
Bank.
World Resources Institute. 2005. www.features@wri.org

Where has all the Raskin gone?
Improving the effectiveness of the subsidized rice distribution program
With coverage of 57 percent of the poor, the Raskin (beras miskin) program
has been one of the most important components of the government s social
Distribution of the Raskin subsidy to the poor, the non-poor and

Bulog
protection system.165 Introduced as the operasi pasar (OP) program in mid1998 at the peak of the crisis, Raskin delivers subsidized rice to poorer
households. However, it has been criticized for weak targeting and excessive
Operating Costs
and Bulog Profit
28%
Subsidy received
by the poor
20%
leakages.
Under the Raskin program, the National Logistics Agency (Bulog) planned to
distribute about 2 million tonnes of subsidized rice to some 8.3 million
households in 2005. Each selected household is supposed to have received
20 kg of rice per month at the subsidized price of Rp 1,000 per kg. The rice is
Subsidy received
by the non-poor
52%
purchased by Bulog from wholesalers at a price determined annually by

presidential decree (Inpres)166 using a budget allocation from the government
to Bulog to cover the cost of the subsidy.167 The price paid to Bulog for each kg
Source: World Bank staff calculations.
of rice in 2005 was Rp 3,351 per kg, equivalent to a 2005 budget allocation of
Rp 4.7 trillion.168 The compensation program associated with the removal of
the fuel subsidy in 2005 added a further Rp 765 billion to the 2005 Raskin
budget.
Leakages in the Raskin program

The three types of leakage in Raskin are lost rice, lost money and high markup prices.
The poverty targeting of the Raskin program

Raskin reaches more than half of the poorer households in the country, making
it an important instrument of social protection. However, it delivers rice to far
more non-poor households than poor households, such that only around a
quarter of all recipient households are classified as poor. The pie chart below
shows that, of the Rp 4.7 trillion allocated to the program in 2005, only about
one-fifth was actually received by the poor. The bulk (52 percent) went to the
non-poor and more than a quarter of the money went to Bulog to cover
operational running costs.
Lost rice: Bulog distributes rice to distribution points in villages and subdistricts throughout the country, and leakages of rice occur at various levels in
the distribution process. Comparing official administrative data on amount of
rice distributed with the data from household surveys on the amount actually
received by households, Olken (2005) finds that about 18 percent of the rice
went missing.169 Olken estimates that the welfare losses from corruption may
be so large that they offset the potential welfare gains of the program. An updated
estimate of leakages calculated in 2006 using the same methodology it was
found that more than 30 percent of rice went missing between Bulog s
165
Data source: Susenas 2004.

For 2005, these purchases are regulated by Inpres No. 2/200,5 which specifies that Bulog should pay Rp 1,765/kg for gabah kering giling (GKG) purchased from traders who bring to its warehouses or Rp
1,740/kg for purchases of GKG at the rice mills. The vast majority of Bulog s purchases are GKG if it buys GKP (gabah kering panen, i.e. gabah that has not been dried) at the mill it should pay Rp 1,330/kg;
if it buys rice which has already been milled it should pay Rp 2,790/kg. Purchases for 2006 are regulated by Inpres No. 13/2005.
167
This is calculated as the total quantity to be distributed times the difference between the price that Bulog is paid by the government per kg of rice and the Rp 1,000/kg selling price.
168
Bulog explains that the Bulog rice purchase price has to be higher than the actual regulated purchase pricing of Rp 2,790/kg because Bulog has to borrow from commercial banks in order to fund the
purchases and pay commercial interest rates on this loan.
169
For this, the author uses the conservative assumption that all beneficiary households received the 20kg of rice allotted to them by the program.
166
217
warehouses and the households receiving it from November 2003 to January
heads collected money from poor households and never paid Bulog for the
2004. This loss is equivalent to 164,000 tonnes of rice valued at Rp 433
rice received. However, a lack of transparency in the amounts of rice allocated
billion (US$43 million). However, there is a wide variation in leakage by
to sub-districts or villages undermines the ability of poor communities to
province (see map below): the greatest leakage is in Banten, where more
hold Bulog and village officials accountable.
than 75 percent of the Raskin was lost, and in another six provinces the
leakage rates were over 50 percent. However, in six other provinces less
than 10 percent of the rice went missing.
Missing rice by province (%)
In Madiun district, East Java, Ngawi village had the highest debt amount
to Bulog for the rice distributed with their debt arrears at Rp 19.2 million.
While the money is collected from villagers at the point of distribution
by the village officials, Bulog has not received the money and is currently
halting further allocations of subsidized rice to the village. (Media
Indonesia, 25 January 2006.)172
The poor pay more: In many cases, the poor were asked to pay more than
In addition, rice never reaches the poor in some areas because of the
the Rp 1,000 per kg amount mandated by the program in 2005. On average,
allocation mechanism used at the village level. In theory, rice is only
Raskin recipients had to paid an extra 19 percent operational costs or some
supposed to be given to pre-welfare and welfare households (according to
Rp 30.5 billion (US$3 million) between November 2003 and January 2004
the BKKPN s classification of households). However, in some places
(Susenas, 2004). The highest average mark-up was observed in Jakarta, where
allocations are given out to all village members, or by alternative non-
recipients paid Rp 2,900 for Raskin rice almost three times the official price.
transparent allocation mechanisms. As a result, the amount received by

households is typically much less than 20 kg. Surveys suggest that
In many cases, mark-ups are far higher than justified by the costs of
sometimes it can be as little as 5-6 kg, greatly diluting the benefit to poor
transportation from distribution points to households. For instance, the
households.
170
estimated transportation costs to selected villages in Bengkulu were Rp 20Rp 150 per kg, while most recipients in Bengkulu paid Rp 150 higher than
Lost money: Distribution of the rice from the village (desa) or outside
even the highest estimate of transportation costs.
Java the sub-district (kecamatan) level down to households is conducted

by teams of village officials.171 The media have reported cases where village
170
For 2006 the planned allocation has been reduced to 15kg per household for 10 months of the year.
In principal these teams are supposed to direct the rice allocation only to households defined as pre-welfare and welfare 1 according to the BKKBN listing of household welfare.
172
Media Indonesia. Jawa Timur - Kabupaten Madiun dan Ngawi Penunggak Raskin Tertinggi , January 25, 2006.
171
218
Reforms are needed to make Raskin work for the poor
3.
Independent monitoring should be encouraged. In this context,

the government could undertake a study to identify the districts in which
It would be wrong to abolish Raskin, but some critical reforms should be
most leakages occur and try to pinpoint the biggest holes in the system.
considered.
Understanding the incentives of officials involved in the program would

also be useful in tackling the underlying causes.
1.
Identifying clear lines of accountability within Bulog is a

key first step in stemming losses. It is crucial to clarify
4.
Involving the private sector in the distribution process. Most
responsibilities and penalties associated with corruption in program
villages in Indonesia well served by private rice traders. It therefore makes
implementation. The creation of an investigative team at the highest
little sense for Bulog to play the same role as the private sector. If
level could send a message that corruption will be seriously
distribution costs were benchmarked by locality, then the private sector
punished by Bulog s senior management.
could compete for this business through a transparent bidding process.

Thus, Bulog could devote its resources to managing and monitoring the
2.
Improving transparency regarding rice allocations and

benefit amounts would increase community participation in the
distribution process and reduce leakages at the village level. In
addition, formalizing the rules regarding the operational costs for
performance of private operators rather than implementing the program

itself. Only in remote areas where no private operators were willing to
undertake distribution for a reasonable cost should Bulog continue to
deliver rice directly.
distributing the rice would give certainty to the Raskin selling price,
avoiding misuse and mark-ups.
219
Chapter
Making Government
Work for the Poor
220
Chapter 7: Making Government Work for the Poor
Introduction
With the launch of one of the biggest decentralization efforts in the world in 2001, it was inevitable that
Indonesia would experience some systemic challenges. Not all local governments were fully prepared for the
new responsibilities that they had to shoulder, and some have had to focus on improving their capacity in order to fully
take up their new obligations to provide basic service delivery. Five years on, the roles and responsibilities of different
levels of government remain unclear in some cases. The system of unconditional and conditional block grants to district
and municipal governments is still in its infancy, while resource flows to districts and municipalities (and ultimately to
frontline service providers) remain fragmented, making it more difficult for beneficiaries to demand the accountability that
should accompany decentralized decision-making. Furthermore, accountability of service providers to higher levels of
government has also become less clear.
Together with decentralization, political democratization has recently been introduced. The highly centralized
and authoritarian approach of the Soeharto era brought with it a centralized approach to service provision that is now
undergoing fundamental change. Following the economic and political crises of the late 1990s, the political process has
become more democratic all the way down to the district level. There has been a move towards multi-party politics, with
the direct election of the president, and at the district/municipal level, with the election of legislative assemblies (DPRD I
and II) and the direct election of district leaders (bupati in districts and walikota in municipalities). These transitions have
had two impacts. First, they have allowed regions to articulate their own preferences and priorities for public investments.
Second, they have created new ways of holding policy-makers accountable. Local-level democratization has also resulted
in increased popular participation in some aspects of service provision. This is clearest in the case of education, where
parent and community participation has been strengthened in newly reinvigorated school committees in some localities
(World Bank, 2006k).
Indonesia has come a long way in the past ten years in institutionalizing democratic norms. Not only are
there direct elections for both the executive and legislative branches at the national, provincial and district levels, but
these elections are now contested by many political parties or, in the case of executive positions, by coalitions between
parties. In 1996, as had been the case since 1975, only three political parties were allowed to legally function Golkar, in
effect the New Order regime s political vehicle; the United Development Party (PPP); and, the Indonesian Democratic Party
(PDI). In the last general election of the Soeharto era in early 1997, Golkar won almost three-quarters of the popular vote
but the ruling party effectively had even greater control in the DPR, because 15 percent of the seats (75 seats) were set
aside for the military. In contrast, in the most recent legislative elections, over 20 parties competed for seats and 17 parties
won at least one seat in the DPR, with seven parties winning 45 seats or more (KPU, 2005).173 The DPR today is no longer
dominated by one party coalitions have to be formed in order to draft legislation and the incumbent president s party has
only about 10 percent of the total 550 seats. As a result, the DPR is no longer a rubber stamp for executive decisions
laws, plans, and budgets are all now effectively negotiated between the executive and legislative branches.
173
Indonesia National Election Commission (2005).
221
What a difference a decade makes174
Table 7.1
Issue
1996
2006
Presidential selection
Selected indirectly by the People s Consultative

Assembly (MPR) every five years
Directly elected through universal suffrage every five years
National parliament
Unicameral legislature 500 members, with 20 percent

of seats reserved for the military (reduced to 15 percent
in 1995). Dominated by Golkar.
Bicameral legislature (DPR and DPD).

DPR members (550) elected directly in multi-member constituencies
through proportional representation.
Each province elects four members to the national DPD, for total of
128 members.
Political parties
Only three parties legally allowed to contest elections Golkar,

the United Development Party (PPP), and Indonesian
Democratic Party (PDI).
More than 20 parties competed in the 2004 general election, and five
parties put up serious parties for the presidency.
Presidential-legislative relations
De facto, a rubber stamp body for the presidents policy decisions.
Role of the military
Reserved seats at all three levels of parliament.

About 6,000 military staff seconded to government positions
(as of 1995).
Territorial command system enables military self-financing.
The presidents party (the Democrat Party, PD) holds only 56 of 550
seats in DPR.
DPR is now a serious check on presidential authority.
No reserved seats in parliament
Territorial command system remains intact.
Provincial and district executives
Appointed by Ministry of Home Affairs (under tight supervision

of the president).
Directly elected.
Provincial and district legislatures
Only three parties legally allowed to contest elections.

15 percent of seats reserved for the military. Dominated
by Golkar.
Directly elected in multi-member constituencies in a proportional

representation system.
Civil society
Tight restrictions on the press and NGOs.
Free press and proliferation of NGOs.
The role of sub-national governments has also changed dramatically. Ten years ago, provincial and district
governments largely carried out policies that were mandated from the center, while today they play a major role in developing
local policies and budgets, and implementing these policies on the ground. Today, sub-national governments have
significant control over financial resources, and develop their own plans and budgets.175 This is in marked contrast to the
situation that prevailed during the New Order period, when sub-national governments had little or no direct budgetary or
planning authority (Baker et al, 1999; Schwarz, 1994). Sub-national governments, while not yet agents of the electorate ,
are no longer exclusively agents of the state .
And at both the national and sub-national levels, the legislative branch has begun to assert its role as
a check upon the executive. During the New Order period, the national and regional parliaments rarely proposed
legislation.176 Policies and budgets are now negotiated, not mandated by the executive (NDI, 2006). While many observers
have criticized regional legislatures for being relatively inactive and/or corrupt, their members have only been directly
elected through open lists in the last election cycle (ICG, 2003). Thus, while legislatures still have a long way to go in
terms of acting in the best interests of their constituents, they are no longer simply passive players on the political scene.
And the military has neither reserved seats in legislatures, nor a commanding role within the state
apparatus. A decade ago, 20 percent of seats in national, provincial, and district legislatures were reserved for the
military (Lowry, 1996) this is no longer the case. In addition, active military staff are no longer seconded to civilian
174
Information for this table and following paragraphs was garnered from a number of sources, including: Emmerson, Donald K. (1999); Lloyd, Grayson and Smith, Shannon. eds. (2001); Mietzner, Marcus.
(2006); The National Democratic Institute (2005); IFES, (December, 1999); International Crisis Group, (December, 2003); Lowry, Robert. (1996); Robison,Richard and Hadiz, Vedi. eds. (2004); Schwarz,
Adam. (1994); and World Bank (October, 2005).
175
However, one significant problem is the lack of clarity regarding functional allocations between the center and sub-national governments (see following section).
176
In the period 1966-95, of 340 laws passed parliament initiated only seven of them. See Baker et al (1999).
222
positions, and the dual function (dwifungsi) doctrine was revoked in 2000 (Mietzner, 2006). Thus, the first steps have
been taken in terms of separating unelected military officials from the state apparatus. This is important in building a
democracy in which the government responds to the wishes of its citizens, as well as a first step in promoting security
sector reform within the military. However, the military does retain its territorial control system, which means that it still
has a high degree of autonomy from civilian control and continues to engage in income generating activities that compromise
its professionalism (Mietzner, 2006).
However, having shifted from being a development state to a democratic state , Indonesia faces
many challenges that are common to transitional societies. While a large majority of Indonesians polled
recently said that they were in favor of the current democratic system,177 there is also widespread frustration with ongoing
corruption, weak formal sector job growth and a sense that the possibilities inherent in a democratic transition have not
been fully realized (Robison and Hadiz, 2004). While citizens are free to demonstrate and the press is largely free to write
what they want, many citizens are still subject to arbitrary decisions and/or rent-seeking by government officials, the
police and the legal system (World Bank, 2004). It is still common for citizens to be required to pay bribes in order to
acquire permits and settle legal disputes, for example.
Decentralization has resulted in less accountability of front-line service providers to central ministries,
but this has not yet been effectively substituted for by enhanced accountability to locally-elected officials
or citizens. Money politics continues to be a serious problem at both local and national levels (IFES, 1999), and many
analysts note that, although electoral reforms were introduced in 2003 to make parliamentarians more accountable to
their constituents,178 they remain beholden to party bosses who place them on party electoral lists (ICG, 2004). Thus,
while the promise of democratization and decentralization was that elected officials and civil servants would become more
directly accountable towards the citizens they serve it is at best debatable whether this has yet happened in the majority of
localities in the country.
This weak accountability to citizens is exacerbated by a lack of clarity regarding functional allocations
between central and sub-national governments, making it difficult to hold elected officials and civil
servants accountable for their actions. While decentralization has placed significant administrative and fiscal
authority in the hands of sub-national governments, the division of responsibilities and authority among central, provincial
and local governments (not to mention service providers and communities), remains unclear in many cases (see Section
III for more details). For example, both central ministries and local governments currently provide or facilitate services to
clients. This creates significant accountability problems, as service providers are responsible to both central and local
governments, but in many cases held accountable by neither.
And, in any case, Indonesia has moved from first to second generation poverty problems that are
harder to address through the type of top-down planning and budgeting system that prevailed during
the New Order period. While the New Order government was very successful at dramatically increasing the quantity of
services in the form of roads, schools, or health centers these services were often neither of high quality, nor tailored to
the specific needs of individual regions. For example, while there was a vast expansion in the number of teachers and
healthcare workers, many providers were poorly trained, often did not (and still do not) regularly show up for work (World
Bank, 2006k). These second generation challenges mean fixing problems that have to do with the software of government,
namely training civil servants, and motivating them and supervising them appropriately. Ensuring that a well-trained,
motivated teacher is in the classroom every day is in many respects more difficult than building the classroom.
177
178
Jakarta Post, September 15, 2006.

Law on General Elections No. 12/ 2003.
223
Measuring the performance of government in this process of change is far from easy or straightforward.
Figure 7.1 illustrates that there was no clear trend in governance for Indonesia over the period 1996-2004. It illustrates
governance trends through (selected) indicators in voice and accountability, political stability and regulatory quality.179
Each dimension of governance affects the environment and setting for both demand- and supply-side poverty interventions.
However, several of the recent changes are notable: improvements in voice and accountability reflect the impacts of the
post-crisis upheaval, with the subsequent emergence of a democratic, decentralized Indonesia and the return of political
stability following a free-and-fair election process. Regulatory quality, however, remains weak. This situation directly
affects investment and perceived competitiveness, and ultimately the overall performance of government at all levels in
serving the citizens of Indonesia.
Figure 7.1 Governance indicators for Indonesia
Figure 7.2
Three key areas of government action
Better align policy, planning and

budgeting systems
Strengthten institutional accountability

Enhance assessment and monitoring of
poverty reduction
Source: WBI Governance Indicators, 2004
Good governance is a critical ingredient for successful poverty reduction. It is not just the level of spending
or the focus of that spending (described in Chapter 5 on Public Spending and Chapter 6 on Social Protection) that
determines impact, but also the way government spends that matters: how decisions are made, how effectively the funds
move, how delivery processes are aligned and how well programs are monitored. There are both successes and blockages
in the institutions, systems and procedures adopted by national and sub-national governments for poverty reduction.
This chapter considers the key ingredients of the systems and procedures necessary for the government to implement
poverty reduction objectives and recommends three areas of action for government. Figure 7.2 sets out these areas of
action and highlights the four recurring themes of the chapter.
II
Policy, Planning and Budgeting Systems
The alignment of policy, planning and budgeting systems has a major impact upon poverty reduction
outcomes. The systems for planning and budgeting at the national level have faced challenges, most importantly in
ensuring that the priorities as outlined in medium-term and annual plans are properly aligned with the budget. Moreover,
since decentralization, it has become more complex to coordinate both these processes among different tiers of government.
The result has been a less predictable environment for sector and local spending characterized by ad hoc decisionmaking and both overlaps and gaps in efforts to alleviate poverty.
179
224
Governance Indicators. World Bank Institute (2005).
At the national level, until recently, systems have been characterized by a series of structural challenges.
These challenges include: unique and sometimes overlapping sector-based budget classifications; separate planning and
budgeting systems managed by separate agencies (Ministry of Finance and Bappenas), which sometimes leads to
overlapping routine and development expenditures; and a focus only on inputs.80 All these structural difficulties accumulate
to make it difficult to align ministry spending with government poverty reduction priorities. These factors have been
exacerbated by the one-year budgeting time horizon, making it all but impossible to translate the most well-intentioned
efforts into longer-term responses to address poverty.
Table 7.2
Translating poverty reduction priorities and objectives into results on the ground
Step
Issue
1.
Broad poverty reduction objectives are agreed upon by the

president and the cabinet.
There may not be full agreement regarding all objectives or how best to achieve them.
2.
Each year, these objectives are translated into an overall plan

with priorities by Bappenas (RKP) and into a sector-specific
plan by each ministry (Renja-KL) based on indicative budget
ceilings (pagu indikatif).
Ministries may not have any incentive to revise their sectoral priorities to reflect the poverty
reduction objectives. Ministries are not required to include either program monitoring data
or evaluation results, so ineffective programs can continue.
3.
RKP priorities with fiscal policy and indicative ceilings are discussed
with parliament and revised. Based on the revised temporary ceiling
(pagu sementara), ministries prepare a work plan and budget (RKA-KL)
and the MoF prepares the draft budget.
The transition from the RKP to Budget and Renja-KL to RKA-KL preparation is not seamless,
so poverty-related programs and associated budget proposals may change (Bappenas
prepares guidelines for Renja-KL and reviews the substance of ministry plans while the MoF
prepares guidelines for RKA-KL and reviews to ensure adherence to budget ceilings and
national price standards. Within most ministries, the planning unit coordinates preparation
of the Renja-KL, while the finance unit coordinates preparation of the RKA-KL).
4.
This budget is submitted to parliament, where it is reviewed and revised

in the Budget Committee, as well as individual commissions.
Discussions with parliament may lead to budget revisions not consistent with the governments
poverty reduction priorities and objectives.
5.
Following parliament s approval of the budget, a presidential regulation

(Perpres) formally appropriates budget resources (pagu definitif). Ministries
prepare concept DIPA (the authority to spend allocated funds) and the Ministry
of Finance approves.
The transition from the RKA-KL to DIPA is not seamless, so poverty-related programs and
corresponding budget allocation may change (DG Treasury has separate manuals, formats
and computer applications). Delays in receiving DIPAs delay program implementation,
potentially affecting effectiveness.
6.
Programs/activities are implemented by central sectoral ministries

and by sub-national governments.
a. Sub-national governments have their own priorities and plans, which may be at variance
with national objectives (although this is a legitimate exercise of local planning and
budgeting authority in a decentralized system by elected local leaders and parliaments).
b. Both central ministries and local governments implement, or facilitate implementation
of, programs simultaneously in the same localities. They often fail to coordinate effectively
together, creating gaps and overlaps in services to the poor.
There is a complex process of translating broad policy objectives into concrete results on the ground.
Unlike the centralized policy and program implementation system that prevailed during the New Order period, today many
actors play a role in devising policies and implementing them. Not surprisingly, with so many steps involved, broad
priorities and policy objectives are translated in an uneven fashion into results on the ground.
Efforts to remove the structural constraints have been facilitated by both changes in legislation and
strengthened government leadership. The State Finance Law No. 17/2003 and accompanying regulations foresee
significant reforms in the rules governing budget development (see Table 7.3). Progress has been made in developing a
unified budget using internationally consistent economic and functional budget classifications. Starting in 2005, Bappenas
was instructed by the cabinet to prepare a prioritized annual government work plan (RKP) for 2006, and ministries were
asked to develop sector plans (Renja-KL) within indicative ceilings jointly set by Ministry of Finance (MoF) and Bappenas.
Moreover, ministry budget proposals now include multi-year costs and indicators by program. The higher levels of
collaboration between Bappenas and the MoF are serving to ensure that allocations to ministries reflect the RKP priorities,
including those relating to poverty reduction. These are important first steps towards greater accountability in planning
and budgeting.
180
In a shift towards greater performance of foreign financed projects, in 1996 Bappenas introduced the requirement for periodic performance reporting and evaluation of development projects.
However, these tended to focus on project management and disbursement information rather than outputs and outcomes.
225
Table 7.3
Before and after: government efforts to rectify deficiencies in planning and budgeting processes
Pre-reform
Post-reform
Budget classification
Idiosyncratic sector budget classification.
Modified governance finance statistics (GFS) classification.
Budget categories
Dual budget system with

overlapping routine and development
expenditures.
Unified budget, a work in progress, as well as a requirement in law

that the budget should be linked to the government s priorities
through the work plan (RKP).
Approach
Input-focused approach.
Results or performance-based approach not yet started in the

budget, but government work plan has results oriented targets and
indicators.
Implementing regulations stipulate that ministries should use
previous years results as input for preparation of the subsequent
year s budget proposals.
Planning outputs
Plans did not set priorities and were

normative wish lists without indication
of costs.
The annual government work plan (RKP), as a plan with select (albeit
broad) priorities and priority programs, as reference for ministries
to prepare their own ministry plans, also with indicative ceilings for
each ministry and its programs.
Time frame
Plans developed for one and five years,

but development and routine spending
based on a one-year time horizon.
For 2007 budget, ministries are expected to incorporate estimates

for 2007 and the following two years.
Evaluation
No formal requirement for evaluation.

Development projects monitored for
disbursements.
The State Finance Law requires that every program is to be evaluated

every five years and that ministry financial reports explain results
achieved.
The current government made poverty reduction a priority in its 2004-09 medium-term plan (RPJM),
but this does not necessarily translate into poverty focused sector plans and budgets. The poverty focus of
the medium-term plan was reinforced in the 2006 and 2007 RKPs and reflected in the ministries budget ceilings. Sector
ministries are responsible for proposing policies and spending programs to meet the priority objectives and targets, but
there is a built-in practice that results in the same or similar programs being proposed each year. While the legal framework
now provides a basis for results-oriented planning and budgeting, and there are clearly articulated poverty reduction
priorities, the institutional systems/procedures and sanctions are not yet in place. The result is that there is no mechanism
to ensure that the sectors really do focus plans and budgets on interventions that benefit the poor.181
Results are not always the prime consideration in allocating resources across programs within sectoral
ministries. In practice, funds are sometimes allocated to programs that are formulated without the full benefit of poverty
data or results of program monitoring (see Section III), or have too many and/or unclear objectives, and only input indicators.
This makes it difficult to monitor and report on whether progress is being made towards achieving the program objectives.
Given that 40 percent of total public spending is now the responsibility of sub-national governments,
district and municipal spending patterns and processes are critical. The pro-poor focus of public spending in
Indonesia is contingent on the performance of lower levels of local government, and experience since decentralization
provides some sobering lessons. To date, planning and budgeting at the provincial and district levels have not been as
pro-poor as they could have been, although total spending on health and education in richer provinces increased in the
post-decentralization period of 2000-03. Evidence from the education sector, for example, suggests that increased inequality
in spending stems from sub-national spending becoming more unequal.182
181
For example, ministries are not required to report, in anything other than a general way, on demonstrated progress against the implementation of previous years plans, nor include monitoring/
assessments of performance of programs or analysis of alternative ways of achieving the priority objectives and targets.
182
World Bank (2006k) and Chapter 5 on Public Spending.
226
The constraints at the sub-national level are similar to those at the national level, only magnified due to
lower capacity and variable levels of local political commitment. To the extent that decentralized spending is
expected to more directly address the needs of local communities, the planning and budgeting process at the local level is all
the more important. The initiation of direct elections for regional government heads in 2005 is intended to improve political
accountability which, when combined with incentives (such as conditional matching grants), may work to enhance the
poverty focus of local government spending. But structural problems with district/municipal level planning and budgeting
systems are compounded by the weak capacity in local government to manage these processes.183 Plans and budgets are not
usually supported by local-level poverty diagnostics and, where diagnostics do exist, these are not used for decision-making.184
Preparation of medium-term plans at the local level have been frequently contracted out to universities or other organizations
with the result that they are completed for official purposes but have little ownership by government officials.185 The disconnect
between plans (prepared by Bappeda)186 and budgets (prepared by finance units) is equally problematic at the local level.
Central government ministries exacerbate the inefficiencies and incapacity of local government by continuing to spend on
local-level services, such as health and education, without coordinating effectively with the local governments mandated with
the service delivery responsibility (see Chapter 5 on Public Spending). Performance-based budgeting was introduced in
2002 at the sub-national level, but has met with only limited success. There was neither sufficient socialization of the principles
and concepts, nor sufficient training or support for their practical application.187
Box 7.1 The challenges of implementing good pro-poor planning and budgeting by sub-national governments
The legal and regulatory framework for sub-national planning and budgeting largely mirrors that at the national level. The National Development Planning
Law (No. 25/2004) and the State Finances Law (No. 17/2003) and relevant regulations (Government Regulations [PP] No. 20 and No. 21) provide a
somewhat coherent framework for sub-national planning and budgeting. The provisions of the new laws on regional autonomy (Regional Governance Law
No. 33/2004) and Fiscal Decentralization (Law No. 33/2004) are generally consistent with the above laws.188 In practice, however, many sub-national
governments face significant challenges to develop coherent plans and budgets. Reasons for this include:
Although there is a long history of formal bottom-up planning exercises, these were normally ineffective in involving the poor to influence plans or
budgets.
Many planning documents are not of high quality, in part because sub-national governments had only a minor role in planning and budgeting prior
to decentralization.
Planning and budgeting arrangements are less unified than at the central level.
Many local governments do not know the total amount of central transfers that they will receive and these transfers are often not disbursed in a
timely manner.
Direct spending by central ministries in localities makes it difficult to coordinate local and central plans and budgets.
There are no institutionalized mechanisms for citizen participation in the budget process.
Performance-based budgeting has been introduced, but there has not been adequate training and support for sub-national governments to be able
to use it.
Sub-national governments have very limited monitoring systems to assess impact.
183
See, for example, Norio Usui and Armida Alisjahbana (2003). Local Development Planning and Budgeting in Decentralized Indonesia: Key Issues. January 2003. Also Sukabumi Public Expenditure Review,
North Sumatra Public Expenditure Review and East Lombok Public Expenditure Review. (2002). World Bank (mimeo).
184
Ibid.
185
Ibid.
186
Regional Development Planning Agency.
187
The USAID-funded program, Building Institutions for Good Governance (BIGG), made technical assistance and training available to a small number of governments to assist them in introducing performancebased budgeting. Traditional training sessions were abandoned when they did not work at all, however, on-site in-depth technical assistance and coaching was found to be effective. They also found that there
was a need for basic critical thinking and problem solving skills before work on budgeting could start. See Checchi and Company Consulting, Inc., BIGG-PERFORM: End of Project Evaluation, Final Report ,
April 2005.
188
There are minor contradictions, such as whether further regulation of the development planning process will be via local regulation (Perda), as per Law No. 25/2004, or by Government Regulation (PP), as
per Law No. 33/2004.
189
One notable findings from the Making Services Work Case Studies (2005-6) undertaken by the World Bank was the inability of even very motivated and risk-taking local leaders to obtain accurate, real-time
information regarding the impact of their innovations in service delivery.
227
Capacity for translating priority objectives and targets into relevant and effective programs and activities
has varied by locale. Evidence from various districts/municipalities suggests that even where the medium-term and/
or annual plans set down results-oriented targets, service units (Dinas) have had difficulty in specifying how proposed
programs and activities would contribute to the achievement of the targets (Kamelus, Ludwig, and Suhirman, 2004). An
emphasis on good documentation of inputs and standard unit costs remains, with few sub-national government programs
or projects moving towards a measurement of outcomes.189
At present, the geographic allocation of revenue to regions in Indonesia is not pro-poor. Although about
one-third of central government spending is allocated geographically across Indonesia through transfer payments to subnational governments, this funding is not effectively used for poverty reduction (see Chapter 5 on Public Spending). The
unequal distribution of the two largest transfers is the cause of the increasing inequality in this sub-national spending.
Controlling for the amount of revenues received at the local level (aggregated at the provincial level), analysis shows that
there is no relationship between health and education spending, and per capita expenditures by province. This suggests
that low spending by sub-national governments post-decentralization is not as much determined by local decisions as by
varying levels of revenue being channeled through established transfer mechanisms.
Evidence from the education and health sectors, for example, suggests that the increased inequality in
spending in these sectors at the provincial level stems from sub-national spending becoming more
unequal, rather than the central government s deconcentrated spending in the regions. In other words,
while the increase in the central government s deconcentrated spending has been neither strongly pro-poor nor pro-rich
in terms of the provinces it has targeted in the period 2000-03, the richer provinces have posted greater increases in
education and health spending than poorer provinces because of more province- and district-level spending. In Figure
7.3, a more positive slope indicates that more spending (in per capita terms) is allocated to richer provinces. The slope is
more positive for decentralized spending in 2003 for both health and education compared with 2000.190 This means that
decentralized spending became less pro-poor during this period, while deconcentrated spending become more propoor.191
190
The change in the slope for central spending in the regions is not significantly different from zero, hence this analysis does not imply that central spending is more pro-poor. It just means that decentralized
spending is becoming less so.
191
Differences in levels of shared natural resource revenues and the DAU across provinces are enough to explain differences in education and health spending. Controlling for mean welfare level in a province
and all categories of fiscal revenue to the local government, a Rp 100 increase in shared natural resource (dana bagi hasil) revenues per capita is associated with a Rp 14.6 increase in education and Rp 5.3
increase in health spending. A Rp 100 increase in DAU per capita is associated with a Rp 6.8 increase in education and Rp 2.1 increase in health spending (see results in Annex VII.1).
228
Decon + local health spending is higher after decentralization

but may be less pro-poor in terms of its targeting of provinces
Deconcentralized health spending (from district and province governments) is

much less pro-poor in terms of its targeting of provinces after decentralization
Central development health spending (deconcentrated) is slightly pro-poor in

terms of its targeting of provinces after decentralization
Deconcentrated + local education spending has increased but is less pro-poor

after decentralization in terms of its targeting of provinces
Deconcentralized education spending (from district and province governments)

is less pro-poor in terms of its targeting of provinces after decentralization
Central development education spending (deconcentrated) is slightly more

pro-poor after decentralization
Figure 7.3 Less pro-poor distribution of education expenditures after decentralization due to increased inequities in decentralize spending (2000-03)
229
Local governments depend heavily on resource transfers from central government in order to finance
the gap between their own-source revenues and expenditure levels. As of 2004, about 90 percent of subnational government revenues came in the form of transfers from the central government. Transfers consisted mostly of:
Shared Revenue Fund: this fund consists of transfers from property-based and personal income taxes, as well as
from natural-resource-based taxes (average 15 percent of sub-national revenues in 2004).
General Allocation Fund (DAU): a general purpose grant intended to be based on the size of the fiscal gap
between expenditure needs and fiscal capacities (average 66 percent of sub-national revenues in 2004).
Special Allocation Fund (DAK): specific purpose, matching grants (average almost 4 percent of sub-national
revenues in 2004).192
Policy towards natural-resource (NR) revenues results in a variance between resource-rich and resourcepoor regions. This apparent imbalance in allocations across Indonesia is the consequence of a political settlement in a
nation with significant diversity in natural resources. The policy, legislated in 1999,193 ensures that 15 percent of revenues
derived from oil and 30 percent of revenues derived from gas are returned to the originating province. In the autonomous
provinces of Aceh and Papua (and also presumably the newly created province of West Irian Jaya), this ratio is even
higher. As a result, about 75 percent of the NR revenue is distributed to the provinces of Aceh, East Kalimantan, Papua and
Riau, and the districts/municipalities within them. Moreover, 10 percent of districts/municipalities receive 80 percent of
shared NR revenues (Lewis, 2005). Being a significant proportion of the revenue of some provinces (accounting for more
than 50 percent in East Kalimantan in 2003, for instance), this establishes very different fiscal capacity scenarios for
resource-rich and resource-poor provinces all without any consideration of their respective levels of poverty.
Provinces ranked according to per capita personal

expenditures (Susenas data 2004)
Figure 7.4 Shared natural-resource revenues are distributed to a few provinces only
Per capita revenues of districts (Rp million) aggregated at the province level (2003)
Source: SIKD data and Susenas 2004.
Likewise, the DAU which is the primary instrument for equalizing sub-national fiscal resources is not
structured to target areas with a high incidence of poverty.194 While there is a formula for the allocation of the
DAU across districts that includes a poverty variable (or proxy in the formula), this fiscal-gap formula only partially
192
Data from the Regional Finance Information System, SIKD.

Law No. 25/1999.
194
It is important to note that the DAU s primary objective is to equalize fiscal capacity across sub-national governments.
193
230
determines how much districts receive in DAU funds. Administrative costs, specifically civil-service salaries, have dominated
the determination of DAU allocations and, together with the hold-harmless provisions, have crowded out the fiscal-gapformula portion of the DAU. As a result, there is little or no correlation between DAU transfers and poverty rates. In 2005,
a one-percentage-point increase in the poverty headcount of a district was associated with an increase in DAU per capita
of only 0.6 percent.195
In addition to its weak targeting of high poverty areas, a second constraint of the DAU is that its funds
must first be allocated to cover the wage bills of local governments and only the residual amount can
be allocated on the basis of the fiscal gap formula. The 2004
decentralization revisions and the passage of Decentralization Law
Box 7.2 How is the DAU allocated across districts?
No. 32/2004 require that the DAU cover 100 percent of the civil service
wage bill at the district level (see Box 7.2 on DAU allocations). This
The DAU is allocated across regions according to a formula
bill absorbed about 50 percent of all DAU funds in 2006. It thus reduces
that is meant to address disparities in local expenditure needs
the amount of the DAU allocated using the fiscal-gap formula, in
and fiscal resources. The allocation levels are decided as follows in
turn reducing the significance of the poverty proxy. Wage costs are
2006: the overall size of the DAU pie is calculated as 26 percent of net
escalating rapidly and have increased from 40 percent to 50 percent
national revenues (after including shared revenues); the districts are
of the DAU in the past year.196 More worrying still, in 2006, local
given 90 percent of this total amount while the remainder is distributed
wage costs would have absorbed a much higher percentage of DAU
to provinces. Then the fiscal gap formula is used to allocate resources
funds had it not been for a rapid increase in the total DAU pool as a
across districts, taking into account relative expenditure needs and a
result of record high oil prices. In practice, the 2004 requirement
proxy for fiscal capacity. Until 2005, the central government used
creates significant perverse incentives for sub-national governments
population, area, poverty, and a cost proxy to measure expenditure needs.
to maintain large district-level civil service corps.
The poverty indicator was replaced by the inverse of the Human
With the dramatic 65 percent increase in the DAU in 2006,

there is concern over the capacity of some regions to spend
these new resources effectively. Transfers to sub-national
governments increased sharply after the central government adjusted
its oil price assumptions upwards and increased its revenue
projections. So far, the regions have had difficulty spending these
large windfall inflows and, as a consequence, have accumulated
sizeable savings from these funds (Wallace et al, 2006).
Development Index (HDI) and GRDP per capita, effective from 2006.
The 2004 revisions of the decentralization laws (implemented
in 2006) took a revised approach to the allocation of DAU.
The DAU is first used to fund 100 percent of the existing wage costs in
the districts and the residual is then allocated based on the fiscal gap
formula. The hold-harmless condition which guarantees each district
will receive at least the same amount of revenues as it obtained in the
previous year is planned to be phased out by 2008.
There is wide variation across districts regarding the

amount of increases in transfers. While some of the poorest
regions of eastern Indonesia, such as Papua, have received significant increases (in Papua more than half of the local
governments received increases of 100 percent or more), not all areas have been so fortunate. For example, some already
very poor areas in NTB, NTT and parts of Sulawesi failed to benefit from the transfer increases to anything like the same
extent as Papua.
195
Simulations show that even by eliminating the hold harmless provision and the 100 percent coverage of the civil service wage bill, the relationship between poverty incidence and the DAU still remains weak
(at around 0.9 age increase in DAU/capita associated with a 1 percentage point increase in the headcount).
196
World Bank (2006). Evaluation Fiscal Equalization in Indonesia. Policy Research Working Paper 3911. May 2006.
231
The second fiscal instrument that the government can use to transfer funds to tackle poverty is the DAK,
a special purpose grant mechanism. The explicit focus of this special purpose grant is to encourage regional
expenditure on the provision of basic services and to direct resources to regions lagging in health, education and other
services. The DAK has the capacity to ensure that funding is targeted by sector and by region. To date, the government has
failed to use it to any meaningful degree. In 2005, the total DAK was Rp 4.7 trillion,197 making up only about 3 percent of
central government transfers to sub-national governments.198
The relatively small pool of DAK funding available for sub-national governments is also spread thinly
across sectors and regions. As of 2005, the DAK was primarily used to provide grants for education (25 percent),
health (20 percent), and infrastructure (33 percent). Other components include public administration infrastructure, fisheries
and the environment, which means that the DAK budget is spread over a large number of sectors and activities. In addition,
it is allocated to almost all regional governments. Every year, and despite the provisions of the Fiscal Decentralization Law (a
law that aims to ensure that allocations are based on objective criteria), many regions become eligible after deals are made
with parliament. In addition, the matching funding contribution from the recipient government has not been used strategically
to leverage sub-national spending. All regional governments have been asked for the same (and very low) 10 percent minimum,
irrespective of their respective fiscal capacity or other criteria.
In the past, DAK funds were earmarked for investment in facilities and infrastructure, whether or not
this was a constraint to pro-poor sector outcomes. Although the revised decentralization law leaves it open,199 the
precedent set by the 1999 legislation that DAK funds be used for physical infrastructure and facilities has not been
unraveled either in practice or in the implementing regulations. Since alternative sources of funding have been fungible,
this restriction has been managed at the district level through the flexibility of other revenue sources (allowing the DAK to
be used for infrastructure, and the DAU or locally raised revenue [pendapatan asli daerah, or PAD] used to fund noninfrastructure activities). However, this restriction ultimately limits and counters the principle intention that the DAK be
used as a pro-poor instrument.200
Poverty indicators are not used to determine DAK allocations. With the exception of the health sector, none of
the sector DAK components has used poverty as a criterion in determining allocations to local governments. In education,
for instance (see Figure 7.5) there is no relationship between DAK spending and enrollment rates in junior high schools
(SMP). Conversely, where health indicators are used in the allocation calculation there is a much stronger correlation
between the health DAK and infant mortality rates (see Figure 7.6).
Figure 7.5 The education DAK does not correlate with problem
areas in terms of SMP enrollment rates...
Figure 7.6
... while the health DAK is better targeted to

regions where infant mortality rate is high (2005)
Source: BPS, MoF and SIKD data, 2005.

197
This includes about Rp 700 billion rupiah for the DAK reforestation fund.
Data from SIKD.
199
Law No. 32/2004.
200
The conditions for the use of the DAK are also technically prescriptive. For each sectoral DAK component there is a Technical Manual providing details regarding what activities can be funded, limiting local
government scope to design interventions most appropriate to local conditions.
198
232
Recommendations
There are four priority areas of action for strengthening policy, planning and budgeting systems. First,
improved systems for linking poverty reduction priorities with sectoral plans and budgets would help to translate political
priorities into budgetary allocations that more closely reflect these priorities. Second, strengthened capacity and incentives
for pro-poor planning and budgeting could be developed through improved technical guidance from Bappenas and the
MoF. Third, district and municipal planning and budgeting systems can be made more effective in terms of poverty
reduction by improving poverty analysis, capacity building, incentives, and the use of participatory processes. Finally, the
DAU and DAK transfers the main sources of local government revenues can be used more effectively to address poverty
in the poorest areas of the country, as well as to more effectively reward good local government performance.
1. Linking broad poverty reduction priorities with sectoral plans and budgets
Linking broad poverty reduction priorities with sectoral plans and budgets in sector spending and
program design will be the key to progress. To ensure that government commitment to reducing poverty is
carried through into expenditure allocations, efforts are necessary in two key areas: strengthening the link between stated
national priorities (articulated through medium-term and annual plans), budget allocations, and line ministry programs
and activities; and improving sectoral decision-making.
Ensure that the cabinet s stated objectives are reflected in spending choices. This can be achieved by
two key changes. When ministry ceilings are proposed to the cabinet at the start of the process (March), options
should be put forward that include reallocations within and across ministries to correspond with stated priorities (as
opposed to using priorities to allocate incremental resources only, as occurs currently). While structural rigidities may
prevent the cabinet from selecting those options, cabinet members would at least understand how closely the ministry
ceilings reflect their priorities. Second, the cabinet should dedicate sufficient time to reviewing the final budget in order to
understand what changes occurred during line item budget negotiations with parliament and how these changes may
affect (possibly adversely) the achieving of poverty reduction priorities.
Bappenas and the MoF should be given a mandate to ensure established poverty reduction priorities
are incorporated into sector plans and budgets. Building on the trilateral meetings process that began in 2006
for the 2007 budget year, Bappenas should rigorously analyze ministry work plans (Renja-KL) in a formal but transparent
and inclusive review process, and provide constructive feedback as appropriate to line ministries. This analysis and a
record of the results of the review process should be provided to the cabinet as necessary during the approval of the final
RKP, and also shared with the Directorate General for Budget for use during the ministry planning and budgeting (RKAKL) approval process.
2. Invest in capacity and develop incentives for pro-poor planning and budgeting
Bappenas and the MoF should strengthen guidance and technical assistance. While the onus is on ministries
to justify their expenditure requests explaining how proposed plans and programs have achieved, or are expected to
contribute to achieving, the governments priorities it is also necessary to ensure that ministries receive consistent technical
guidance on the preparation of planning and budgeting proposals. To this end, the MoF and Bappenas are already
collaborating on substance, as well as on the procedures and software used by sector ministries. Guidance for the RenjaKLs needs to specify the types of poverty-related data to be included (e.g. data by income quintiles, urban/rural areas,
administrative regions, and gender, as appropriate to the particular program), as well as demonstrated evidence of program
performance.
233
Line ministries should review and adapt organizational arrangements to ensure that these are consistent
with aligning budgets with plans. The practice whereby separate units within ministries prepare plans and budgets
needs review. At a minimum, closer and more formally defined coordination would be beneficial. Planning/Finance Bureaus
also need to strengthen analytical skills and be given the mandate to assess the poverty reducing or growth components,
and their impacts of the programs proposed by directorates.
All ministries should introduce a results-orientation focused upon poverty reduction objectives. While
theoretically a results-oriented system is now in place, this can be improved by focusing efforts towards outcomes
associated with poverty reduction. At present, the performance targets identified in annual plans are measured by a
mixture of specific, measurable indicators and more general, ill-defined or unclear objectives needing refinement.
3. Strengthen local-level performance budgeting systems that

are evaluated against results on the ground
At the district and municipal government level, linking analysis
with spending allocations, improving incentives and capacity, as well as
making processes more inclusive and results-oriented, will strengthen
the pro-poor focus of local budgets and implementation.
Link poverty diagnosis to policy and program decision-making.
To improve the poverty focus of regional and local government spending,
poverty reduction planning and budgeting should be based on an
understanding of the poor and their needs. Connecting the process of
poverty assessment and program monitoring (discussed in Section IV) to
the process of planning and budgeting is crucial.201 The experience of
poverty reduction strategies being developed at the district level has shown
how this can be done.202 (See Spotlight on Innovation at the end of this
chapter for further details.)
Establish financial incentives for pro-poor spending. Additional
incentives for local governments to address poverty reduction can be
provided by central government funding approaches that are tied to propoor outcomes and matching funds (e.g. in the form of DAK grants, see
Section II).
Develop local-level capacity for participatory, pro-poor planning and budgeting. Embark on a concerted
and sustained capacity-building effort to ensure staff are skilled and the organizational framework is established for improved
local-level planning and budgeting. Efforts have been made to enhance linkages between planning and budgeting processes,
and to introduce an explicit poverty focus, both in the development of new initiatives and within existing projects. This has
been done by ensuring that processes are participatory and transparent, understanding local poverty determinants, and
focusing on relevant pathways out of poverty. It is likely that a results orientation is best introduced gradually as capacity
is built. This could start with: (i) defining overall poverty reduction priorities with measurable targets and performance
indicators in the RKP-Ds (regional government annual work plans); and (ii) ensuring that programs have clear objectives
and measurable indicators.
201
Given capacity constraints, however, diagnostics at the local level are likely to be based on basic qualitative approaches and descriptive statistics. There is a role for central government and the national
statistics bureau (BPS) to provide standardized sub-national statistics on key poverty and social data to regional stakeholders for this purpose.
202
World Bank (2004). Initiatives for Local Governance Reform Report. Jakarta. July 2004.
234
Pilot performance-based subsidies to complement or replace user fees. Output-Based Aid (OBA) is the use of
explicit, performance-based subsidies to complement or replace user fees. It involves the contracting out of basic services to
a third party (be they private companies, NGOs, CBOs, or a public service provider) and the payment of a subsidy to that
provider tied to the delivery of specified outputs (e.g. the number of connections or kilometers of roads constructed) (Brook
and Smith, 2001). In contrast to traditional approaches of contracting out, OBA seeks to define objectives and performance
in terms of outputs rather than inputs. It tries to mobilize commercial financing of service provision and differs from many
private infrastructure schemes by complementing user fees with carefully targeted subsidy payments. These characteristics
increase the potential for mobilizing private funding for public services, while ensuring a high level of accountability for the
use of public funds. Indonesia could benefit from looking at successes in Cambodia, where subsidies for water services for
low-income households are paid to providers when a household has received the service; in Peru, where telecommunications
companies compete to expand in rural areas on the basis of the smallest
subsidy required; and in Argentina, where contractors responsible for the
maintenance and rehabilitation of rural roads are paid when they have
achieved an agreed performance standard.
4.
Revise the DAU transfer to be more pro-poor and use
the DAK to attack poverty more effectively
The DAU, the main source of revenue for local governments,
is not designed to address poverty. The DAK, a special purpose
grant to local governments, has recently doubled in size, and there are
tremendous opportunities to use it to provide both more direct assistance
to the poorest governments, as well as to create incentives for better
results on the ground.
More pro-poor allocations from the central government could
be achieved by revising the DAU to include a stronger poverty
component within the fiscal gap formula. Ideally, the formula for
allocating the DAU203 should place greater weight upon local poverty rates.
This could be achieved by increasing the weighting of the poverty variable
in the formula. Currently, the inverse of the HDI index is used as the
poverty variable in the DAU formula. This has two problems: it is only
partially measurable at the district level (meaning that it does not accurately measure data for all districts) and, in addition,
it incorporates GRDP per capita, which is already included in the DAU formula (see footnote). A better indicator may be
the poverty gap , which measures the depth of poverty and is reliable and measurable at the district level. This would not
only ensure a bias towards regions with a greater percentage of poor but would provide a more transparent and easily
computable indicator.
Use the increased DAK funding to prioritize poverty reduction. The more than doubling of the DAK budget to Rp
11 trillion in 2006, and its further increase to Rp 14.4 trillion in 2007, present an enormous opportunity to address priority
needs in resource-deficient districts and sectors. The DAK is a powerful instrument that should be used to focus subnational spending towards national poverty reduction objectives. For instance, the key interventions discussed in Chapter
5 on Public Spending and Chapter 6 on Social Protection could be supported by more clearly defined DAK allocations to
support the pathways out of poverty and better service delivery.
203
The current formula has five components: (i) total population, (ii) area of locality, (iii) construction cost index, (iv) gross regional domestic product (GRDP) per capita, and (v) human development index
(HDI).
235
Although the development and implementation of an effective DAK is urgent, a consultative process for
its revision will ensure it is better targeted and tailored to the capacities of each local government.
Involving sub-national governments and other stakeholders in decision-making over key design features, such as conditions,
allocation criteria and administrative arrangements will be paramount for the impending revision of the DAK. Given the
large diversity in capacity, asymmetrical arrangements may be appropriate. More tightly earmarked grants with ex ante
controls could be provided to weaker-capacity governments, while stronger-capacity governments would be given less
rigid conditions with greater emphasis on ex post sanctioning.
Use the allocation criteria in a manner that promotes pro-poor geographical incidence. A pro-poor DAK
could explicitly have allocation criteria that promote the targeting of funds (geographically) to districts based on the
relevant dimension of poverty (e.g. education poverty or infrastructure poverty). As noted elsewhere, for example, the
supply of junior secondary school classrooms is sufficient in most areas of the country, but not in eastern Indonesia.
Also, the poorest regions have the oldest sub-district and village roads in the poorest condition, as well as the lowest allyear access.
Leverage local government resources by revising matching-funding requirements. District/municipal
governments are required to match a minimum of 10 percent of central DAK funding, unless they have limited fiscal
capacity. For more strategic effect, the matching amount could be made higher in order to leverage more sub-national
spending, especially where it is expected that sub-national governments should be very responsive to achieving the
national objective(s); and/or the matching requirement could be varied with fiscal capacity.204
Link the DAK to sub-national capacity building. Irrespective of a well-designed DAK, unless institutional capacity
(and bottom-up and top-down accountability) is strengthened, service delivery performance to the poor may not significantly
improve. Existing efforts by the government to strengthen sub-national government
capacity should be linked to the transfer system.
Consider how the DAK allocations can be used as an incentive to improve
performance. Apart from revising and applying new criteria for the DAK, in order to
improve the commitment and accountability of sub-national governments in achieving
national-priority objectives and meeting their decentralized responsibilities, local
governments also need incentives. Introducing a results orientation is vital to
improvements in efficiency and effectiveness of funding, and is applicable to funding at
other stages. This can be achieved through efforts to benchmark or measure the
comparative performance of sub-national governments. For example, allocation of an
education DAK could be conditional on a pre-defined age increase of junior secondary
school completion rates. Or, allocation of a health DAK could be conditional on specific
improvements in the accessibility and quality of health services, or a decline in the
maternal mortality rate. Such a results orientation would help to measure both
improvements for the total population and for the poor/near-poor). Efforts would be
needed to pilot approaches to performance-linked grants to establish the basics (e.g.
the conditions for the availability and size of grants, objective and transparent performance
measurement), and to develop mechanisms to ensure that weaker-capacity governments
those that need the support most were not further penalized for their lack of resources
and capacity (e.g. support to prepare plans).
204
To ensure that the matching contribution becomes a substitute for own source revenue that a government would have allocated to the activity, the central government could include a maintenance of effort
condition, and then match additional effort to the DAK funding.
236
III
Institutional Accountability
The incomplete and dynamic nature of decentralization has created a challenging environment for
service delivery and poverty reduction programs at the local level. The confusion over financing and
responsibilities lies at the heart of the problem of effective and coordinated service delivery (Wallace et al, 2006). In
addition, however, local governments have only just started to consider the best ways to deliver the services for which
they are responsible. The following discussion highlights the key areas that require improvements and provides
recommendations for action originating from both national and local levels.
At present the delivery and implementation of poverty-related programs by local governments are
characterized by uneven efficiency and effectiveness. A high proportion of the money that gets through to
implementing agencies is used for salaries and other administrative costs. The lack of capacity, and the low level of
success in delivery and implementation, can be attributed to both long- and short-term blockages in government systems,
processes and organizational arrangements and are not simply an impact of decentralization.
Blockages in service delivery occur on both the demand and the supply sides: the government and
delegated providers do not always perform their functions, and the users often do not demand
accountability from them. The problems of basic services in Indonesia and a range of policy options to address
blockages are the subject of a publication, Making Services Work for the Poor In Indonesia (World Bank, 2006k). This
work highlights the relatively low levels of accountability in Indonesia of service providers to their clients and provides a
set of recommendations for addressing this concern. There are a number of factors characterizing service delivery over a
range of sectors in Indonesia: basic services (health, education and water, for instance) are predominately public; they are
poorly coordinated across a range of agencies and tiers of government; and, these public agencies are neither accountable
for the quality of the service they deliver, nor for the outcomes of their interventions.
Box 7.3 High levels of teacher absenteeism
A recent study of teacher absenteeism in Indonesian primary schools found that, on average, 19 percent of teachers were absent for one reason or another
each day. This is significantly higher than the figure for other developing countries. Of eight countries in the study, Indonesia was third behind Uganda (39
percent) and India (25 percent).
In Indonesia, 45 percent of absent teachers had no clear excuse for their absence, 36 percent were either sick or on leave, and the remaining 19 percent
gave the explanation that they were performing official duties outside the school, such as attending meetings or participating in training sessions.
Teacher absenteeism has a disproportionately negative effect on students learning in rural (i.e. usually poorer) areas, where substitute teachers are often
unavailable. This usually means that two classes have to be merged or one teacher has to teach two classes in different rooms. In some cases, a senior
student substitutes for the absent teacher, but in many cases students are simply sent home.
Source: Usman, Akhmadi and Suryadarma, 2004.
Although there are, in practice, a range of service providers in all the basic services, the government
still identifies with its role and promotes a public system of service delivery. Chapter 5 on Public Spending
provides a detailed description of public expenditure in health, education, water and sanitation services, and rural roads.
In many sectors, however, this public funding may have hindered the effective delivery of services by deterring the
development of alternatives. In the health sector, for example, while public spending has steadily increased, the rate of
improvement in maternal mortality rate (MMR) has slowed, preventive healthcare has declined, the health gap between
rich and poor provinces has increased and, consequently, there has been a shift towards demand for private services
(World Bank, 2006k). While it is impossible to prove that government spending has crowded out private spending, it is
notable that public spending alone appears not to correlate with improved outcomes in this case.
237
Clarity of functions between central and sub-national government units

Service providers are hindered by coordination difficulties, some outside their control. Many of the current
arrangements governing public service provision are unclear on what providers are to deliver and how much they are to
receive for doing so. The financing and provision of services is based on bureaucratic instructions, providing relatively
little autonomy to providers or beneficiaries. A typical government health clinic (Puskesmas) has eight sources of cash
income and 34 operational budgets, many of which are provided in kind by the central or local government (World Bank,
2005b). The system is characterized by overlaps local government proposing interventions that central government has
already programmed from Jakarta and gaps, and it is inevitably the poorest who fall through such a complex system.
Lack of clarity regarding central and local functions impacts directly upon the provision of education
and health services. Teacher managemen in the form of decisions regarding recruitment, deployment, performance
evaluation, and pay are made at different levels of government. For example, hiring decisions are now decentralized, but
compensation standards are still centralized. Similarly, it is not clear who has the authority to open and close health
facilities. A 1986 ministerial decree gives this power to the Ministry of Health, and it remains unclear if and how this might
change under decentralization (World Bank, 2006k).
However, the national government plays too limited a role in water and sanitation. Although district
governments now have authority over water maintenance, Article 9 allows higher levels of government to take decisions
whenever a cross-district maintenance issue arises. This is leading to confusion, as in this case too much power may
have devolved to district governments. According to the Water Resources Law No. 7/2004, the management principle is
one basin, one management, and basins may cross district or provincial boundaries . For sanitation, no ministry at the
national level is responsible at present. A key factor in Indonesia s very low rate of urban sewerage coverage is the lack of
institutional responsibility for making policy that would identify and set out a legal and regulatory framework, as well as
define a strategy for involving households, communities, and the private sector in sanitation. Local governments require
greater clarity and incentives regarding their roles (World Bank, 2006k).
And providers do not have the autonomy to optimize service delivery. This is particularly true of the
uncoordinated systems of delivery seen in the health sector and described above. Currently, government health clinics
(Puskesmas) receive almost half of their revenues in the form of earmarked funding and in kind transfers (Figure 7.7). Of
the cash revenue received at the Puskesmas about 80 percent goes to pay staff salaries. Most of the remaining funding
comes from deconcentrated budgets, usually earmarked to assist the poor to access Puskesmas facilities. The nonearmarked funding, which could be used for improving the quality of services at the Puskesmas and expanding Puskesmas
activities for information dissemination and immunization programs, is only about 2 percent of total Puskesmas revenues
(World Bank, 2006k).
Figure 7.7 Puskesmas receive most of their revenues in the form of in kind transfers or earmarked funding
Source: GDS 1+ Financial Module
238
The decentralization process is the basic framework for government to effectively work towards poverty
reduction and is characterized by both enormous potential and challenges. The decentralization reform
implemented in 2001 is possibly the largest reform ever undertaken by the government. Political reform was accompanied
by administrative and fiscal reform on an unprecedented scale (World Bank, 2003d). Five years on, the decentralization
process is far from complete. While decentralization is not considered to have been regressive in terms of service delivery
for Indonesia s poor,205 some major gaps are pertinent and need to be addressed. This section outlines the key problems
that have emerged, and the structural changes needed to develop more accountable institutional frameworks, efficient
organizations and staff systems capable of working together for the poor.206 One of the most critical tasks ahead is to
reform the civil service and to improve the organizational and human capacity of government in support of poverty
reduction processes and programs.
The continued lack of clarity in roles and responsibilities is a primary institutional constraint to improved
service delivery. Service delivery arrangements continue to be hindered by overlaps and gaps among various agencies
and between levels of government.207 There are three issues. First, the central government does not focus as much as it
could in terms of making good policies, setting standards, and/or oversight and monitoring roles, all of which would
improve the quality of service delivery. Second, overlaps and lack of coordination (uncoordinated spending) between the
tiers of government208 in service delivery are common, making accountability ambiguous. Third, local governments and
service delivery agencies often plan and budget in an unpredictable environment.
This lack of clarity results in sub-optimal spending arrangements. In some sectors, the government has yet to
decide who is funding and implementing which service. In many situations, there is weak coordination between centrally and
locally funded programs, and how this coordination should be linked to local governments legally mandated service delivery
obligations. Local governments plan and budget for services or improvements only to find that the central government is
providing them directly. Ramifications at the delivery level are manifold. In education this is typified by schools receiving
funding from four sources at various unknown times of the year. In hospitals, it results in the interruption of supplies, the
delivery of unnecessary supplies (medicines, equipment), and duplication (World Bank, 2006k). Districts and provinces can
go a long way in sharing information and cooperating so that programs create synergies instead of overlaps. However, such
arrangements are voluntary and require local leadership commitment-not a long-term solution to what is a structural problem.
Clear functional divisions will promote greater clarity, more accountability, and higher quality and
more efficient implementation of plans. Table 7.4 below offers a detailed set of ideas regarding how authority and
responsibilities could sensibly be allocated among different levels of government. The guiding principles are that overlapping
authority and responsibilities among different levels of government should be avoided, and that the central government
should focus on setting the rules of the game and holding lower levels of government accountable, but not on implementing
policy directly. Thus, each level could take on the following roles:
National:
Provincial:
District:
Providers:
Communities:
Policy-making, staffing, information, core standards.

Fix regional targets and standards, build capacity, and implement cross-district services.
Develop local plans and budgets, implement services.
Deliver services and monitor outcomes.
Provide feedback to service providers, manage targeted programs, and build/maintain local
infrastructure.
205
Study on Governance and Decentralization Survey, GDS 1+, Center for Population and Policy Studies, Gadjah Madah University, Yogyakarta, 2004.
Ibid.
207
Institutions have overlapping and unclear roles and responsibilities. Several agencies share responsibility for civil service management:: National Civil Service Agency (BKN); State Ministry for State
Apparatus Reforms; Ministry of Home Affairs; National Institute of Administration; Ministry of Finance; sectoral ministries; and regional governments. All have responsibilities for oversight and/or regulation,
policy-making, line management and technical assistance and training.
208
Overlaps and lack of coordination also occur horizontally between different departments.
206
239
Table 7.4
Suggested functional allocations for service delivery209
Provider
National
government
Function/role
Related actions
Make policy.
Finalize detailed function assignment for three tiers of government.

Develop criteria for asymmetric decentralization, since not all regional governments are ready to take
on new powers.
Develop regulatory framework for using private sector and NGOs in service delivery.
Establish systems whereby regional governments can collaborate to take advantage of scale economies.
Organize and staff central ministries according

to central functions.
Re-educate central agency staff about their roles and functions in a decentralized setting.
Audit systematic management and governance skills, against assigned governance and management
functions.
Set core service standards.
Develop a small number of core standards for each sector.

Disseminate information about standards through training programs for government officials and
service providers.
Develop systems for monitoring core standards (for example, national achievement tests for
education, national surveillance system for health).
Improve management and governance systems to reduce corruption.
Select key programs to fund at the national level

using deconcentrated budget.
Reduce cross-regional inequalities
Establish national mechanism to improve equity across regions.
Collect and disseminate information about

innovations and good practiceacross regions.
Provincial
government
Develop detailed rules and regulations for

implementing them.
Develop regional service targets.
Hold provincial and district officials accountable for their responsibilities.
Set service standards in line withnational core

standards.
Disseminate information about standards through training programs for government officials and
service providers.
Build district implementation capacity.
Monitor service standards throughout the province.

Inventory the financial management, personnel, and procurement systems in all districts; address gaps
and strengthen weak systems.
Develop training programs on financial management, personnel management, and procurement
procedures.
Require all districts to submit expenditure plans against performance targets.
Obtain support for targets from provincial and district levels.

Develop scorecard systems for rating or ranking districts in their performance of functions.
Fund and manage services for which scale

economies are high (forexample, training
programs fordistricts), but do not compete
withdistricts.
District government
Frontline providers
(public or private)
Communities
209
240
Ibid.
Develop local plans and budgets, with specific

performance targets.
Inventory skills at the district level and undertake appropriate training to address skill gaps.
Monitor performance against local targets.
Report up the management line to bupati/walikota and horizontally to the Dewan Perwakilan Rakyat
Daerah or DPRD I (Provincial House of Representatives).
Manage services to meet local targets, applying

national andregional service standards.
Monitor and assess service standards in the district.

Report on performance against standards.
Manage decentralized financial, personnel, procurement, and information systems.
Inform providers and user groups about service targets and standards.
Ensure coverage of remote and disadvantaged groups.
Deliver services in accordance with service

agreement.
Increase use of service agreements that focus on outputs and services, and provide autonomy to providers
in achieving these targets.
For public providers: monitor and report on

outcomes in service area.
Budget block grants and human resources to providers based on what isneeded to achieve targets.
Reduce in kind financing.
Make civil servants working for providers more accountable to providers.
Incorporate user assessments in performance evaluations.
Maintain and build village infrastructure.
Ensure democratic processes within the community.
Manage targeted programs for the poor within

the community.
Move project-financed, community-driven development programs on budget.
Provide feedback on service provision by

providers andachievement of outcomes.
Strengthen and create mechanisms for user feedback, such as the school committee.
A number of legislative and administrative blockages are thought to be the cause of this confusion. The
first decentralization law (Law No. 22/1999) stipulated only the sectors that would be devolved to local governments, not
the tasks to be performed, leaving it unclear as to who would and should perform the various functions within sectors.
This has created confusion between district governments, provinces and line ministries. In principle, the latter relinquished
the implementation and personnel management functions and their main role changed to policy-making, standard-setting
and monitoring.
Box 7.4 Legislative blockage to local government action
Decentralization has brought more ambiguity to civil service management. There is uncertainty about the roles and functions
of the central agencies, the sectoral ministries (whose laws have not been revised in light of decentralization) and the regional
governments. The problem is further compounded by the Law on the Civil Service No. 43/1999, which assumes the existence
of a national civil service and mandates the creation of a new Civil Service Commission.
The recent revised decentralization Law No. 32/2004 places a question-mark over the entire essence of decentralization
because it removes the regions right to manage their own public servants, partially negating the intent of the original
decentralization law, namely Law No. 22/1999, which stipulated that so-called obligatory sectors are the responsibility of
local governments. For line ministries, this law required them to relinquish in principle their personnel management function,
changing their main role mainly to that of policy-making, standard-setting and monitoring.
However, this shift was never defined in law and the recent Law No.32/2004 has thus muddied the waters further. The specific
functions of local governments within these so-called obligatory sectors remain unclear. Some local governments have
interpreted the new law to mean that they are responsible for all tasks within the sectors.
Civil service functions and constraints

The strengths and weaknesses of the civil service in Indonesia have been extensively analyzed, and
there is broad agreement that significant challenges exist in reforming the civil service. To build upon the
broad literature and knowledge on the subject, this section addresses the roles, opportunities and constraints of the civil
service to engage in, and deliver on, poverty reduction objectives. The framework below outlines three priority areas that
will drive reforms: improving rules and restraints, enhancing voice, and stimulating competitive pressures. This section
takes up rules and restraints, as well as stimulating competitive pressures. The following section addresses the use of
voice.
Drivers of Public Sector Reform
-Watchdog bodies
-Audits
Me
rit-
ba
s
pro ed r
mo ecru
tio itm
n
en
t/
Rules and Restraints
Decentral
ization
Competitive Pressures
-Competition among
service providers
Voice
Client Surveys
-Community monitoring
and action
Source: World Bank, 2003. Reforming Public Institutions and Strengthening Governance.
241
In present day Indonesia, the civil service is both a potential

asset for, as well as a significant hurdle in, the formulation
and implementation of major government policies and programs
to reduce poverty. The civil service is inefficient in part because of an
overlapping regulatory framework, and inflexible organizational, procedural
and management systems.210 Achieving high-quality service-delivery
results and being responsive to citizens appear in many cases to be
relatively less important than following the lead of superiors. As with many
other civil services, the Indonesian civil service is inclined towards rules
rather than results. The poor, who lack voice and exit options, are more
negatively impacted by this than are others.211 Indeed, the relative weakness
of the civil service has led a major education sector review to recommend
the establishment of a separate teaching service one that would encourage the introduction of merit- and performancebased personnel management and continuing professional development.212
Rules and restraints

Weak incentives and a rigid personnel system make successful program implementation more difficult.
Managers are in many cases not accountable for the outputs associated with service delivery.213 The lack of accountability
is exacerbated by a closed promotion system and the lack of a performance-linked salary structure. Promotions are based
on seniority, there is no lateral entry (promoting competition) and, as a consequence, there are few incentives to focus on
results.214 This lack of accountability creates few incentives to, for example, implement a cabinet poverty reduction mandate,
or to mobilize staff to engage in poverty reduction initiatives.
Staff have little incentive to perform effectively. Few within the civil service see the benefits of striving for results
in terms of poverty reduction. Incentives to perform depend on rewards (pay and allowances), as well as informal incentives
that come with a meritocracy (such as recognition from supervisors and peers) and punishment for poor performance or
for breaking the rules. In reality, however, rewards for integrity and punishments for poor performance of any kind are rare.
Indonesian newspapers commonly report that civil servants accused of involvement in corrupt activities are simply
transferred to other positions, but rarely sacked.
And the rigidity of the staffing system hinders the flexible hiring, transfer and promotion of good people.
Following decentralization, overstaffing has become most common at the provincial level, but many districts are also
overstaffed, with an imbalance favoring generalists over more needed technical staff.215 Transfers are the exception rather
than the rule and civil servants with ill-adapted skills for their positions can stay put for years. The inflexible civil service
structure (with structural and functional positions, see Box 7.5 on staffing) and employment terms, such as constraints on
hiring and firing, and the lack of opportunities for early retirement, re-training, re-deployment and attrition, are significant
stumbling blocks in developing capacity and improving personnel management. In some regional governments, for
instance in West Sumatra (the provincial administration, the city of Solok, and the kabupaten of Solok and Tapanuli
World Bank (2006). Opportunities and Constraints for Civil Service Reform in Indonesia.World Bank and Partnership for Governance Return. Jakarta 2006.
Ibid.
212
World Bank. Education Sector Review.
213
World Bank (2006). Opportunities and Constraints for Civil Service Reform in Indonesia.
214
A recent study presents a critique of the system: the management of civil servants has been undermined by rampant abuse of the rules and procedures for personnel actions, as well as by the opaqueness
of the remuneration system. Widespread patronage and rent-seeking, combined with little enforcement of sanctions, has seriously undermined the accountability of staff and managers . Quoted from: Working
Paper: Civil Service Reforms at the Regional level: Opportunities and Constraints. Institutional, regulatory and financial issues facing reforming regions in Indonesia. The Word Bank (July, 2005).
215
Creating additional problems by requiring the hiring of contract staff to meet their needs. Provinces have fewer functions and are often overstaffed and they are better off financially than districts.
210
211
242
Selatan), innovative staff management practices have been applied, but the centrally governed set of rules is a straitjacket
that prevents serious personnel management reform in the regions.216
The intergovernmental transfer system now encourages local governments not to reduce the number of
civil servants. DAU funds are allocated to pay for staff and the current formula means that if regional governments
reduce their staff numbers, their DAU grants will also be reduced, resulting in a strong disincentive for these governments
to cut their staff. Overall, the structural deficiency of local-level governments to effectively deliver the services and perform
the functions allocated to them is a major concern.217
Box 7.5 The basics of better staffing
The current staffing system can be strengthened by doing the following:
Overstaffing in frontline institutions can partially be dealt with by moving excess personnel temporarily to a pool of staff in
waiting . Pool staff can re-apply for positions once a re-organization has been completed and clear job descriptions are in place. If skill-sets match
and staff are trainable, employees can return to the mainstream of the organization. Those without potential for re-deployment can be left in the pool
or encouraged to seek opportunities outside government. If voluntary departures do not occur, it is still better to keep such staff in a pool .
Ensure that recruitment and promotion are open and transparent, and based on clear job descriptions. Although candidates put
forward for promotion still have to fulfill certain rank criteria, the government can use specialized companies to carry out tests and assessments of
candidates for recruitment and promotion to raise professionalism. The drafting of clear job descriptions for those recruited or promoted is also
important because it clarifies and specifies their tasks, enables performance evaluation and introduces performance-based incentive schemes.
Set proper terms and conditions for contract staff. Contracting is at present used to hire a large number of low-skilled staff in support
functions, but if used wisely it can also be used to improve the skills of the civil service. This will ease the constraints posed by the rigid civil-service
rules. Contract employment can also introduce an element of flexibility within the local human-resources planning system.
Plan the use of human resources wisely. Building a database for human-resource planning and management is the basis for planning of
organizational and personnel reforms. A modern database is a transparent and alternative human-resource planning and management tool and it will
enable regions to do more independent and tailored planning of personnel actions.
Develop predictable incentives for good performance and sanctions for bad performance. While schemes to enforce attendance and
discipline abound in frontline institutions, government should add performance-based incentives either on a group or individual basis. Revamping
policies so that the provision of training opportunities and monetary rewards promote good performance and a regime of sanctions deter bad
performance would go a long way towards providing better services. Such a system should be transparent and monitored closely.
In the post-decentralization context, there is a great need for leadership at central and local levels to
bring about change. Leadership is needed to complete the most painful part of the decentralization process: to streamline
the functions between different levels of government, to determine the right size and right tasks for staff, and to get the
budget process on track. Although there is a dearth of leaders, there are some very successful walikota (municipal mayors)
and bupati (district heads) who have seized the incentive to lead. Yet while these individuals have been noted by the
central level, they have neither been used in the public policy debate, nor up-scaled through legal and regulatory changes.
Political ownership of effective decentralization is critical to the delivery of effective poverty reduction programs.
And reformers are emerging throughout the country. A recent review of nine case studies of service delivery
innovations found that high quality local leadership was the most important factor in determining the relative success of
reforms. While in some cases, such as a community block grant program in Blitar city (see Box 7.9), or healthcare reforms
in Jembrana district (see Box 7.8), local reforms have been successfully executed, there are also cases where wellintentioned reformers have found it difficult to make headway in implementing changes at the local level (see Box 7.6).
216
Some regions take advantage of the age structure of the civil service and leave non-essential positions vacant when staff retires. In the education Dinas in the Province of West Sumatra there used to be 700
staff. Through natural retirement the staff number has now been reduced to 480, and the administration hopes to see the number naturally reduced to 200 over the next five years. This is a second best solution
because the reduction in staff has not resulted in better match between post and employees.
217
Indonesia is not different to most other developing countries in this regard. Even in the most successful, such as South Africa, the capacity deficiency of local government following decentralization remains
one of the greatest problems and challenges to poverty reduction.
243
Box 7.6 The challenges of implementing local civil service reforms

An example of a local attempt at top-down civil service reform to improve transparency and accountability can be found in the case study of the district of
Boalemo district, Gorontalo province. Since 2002, the bupati has led a drive to improve civil servant performance and accountability. Reforms have
included a district regulation on transparency, the introduction of competitive bidding for procurement, and a new system of salaries for civil servants,
together with fines and other sanctions to improve civil servant discipline. But the success of such reforms has been limited so far.
While salaries of most civil servants in the district have increased by as much as 80 percent, performance improvements were only detectable in the
limited form of better attendance. No fines were imposed for poor quality services, probably because quality related violations were never investigated.
The removal of so-called operational allowances used mainly to pay for fieldtrip expenses had a negative impact on staff performance, creating
financial hardship for some staff and making them financially dependent on those who they were supposed to be inspecting. Transparency in bidding for
lower-budget projects has improved but collusion remains in the larger projects. These are usually conducted by non-local firms and represent a
substantial portion of all construction project money being spent in the district. But the reforms do seem to have made civil servants more accountable to
the public. Citizens living near the capital have more opportunities to lodge complaints, and seem to do so more often: at least one such complaint has
resulted in the firing of a corrupt official.
However, the limited success of the reforms has been largely due to the commitment and leadership of the bupati, and to the financial promises and
technical support offered by the World Bank-funded Initiatives for Local Governance Reform (ILGR) Project. But many factors have conspired to limit the
reforms effectiveness to date, including poorly designed incentives reform (the lack of any salary fines for service quality, the impracticality of the
monitoring system for salary fines, and the short-sighted decision to remove operational allowances), over-dependence on the bupati as an individual,
the marginalization of citizens living in distant and poor parts of the district, a very weak civil society and unfulfilled financial commitments from the ILGR
project.
Source: World Bank, 2006. Innovations in Pro-Poor Service Delivery: Nine Case Studies from Indonesia.
However, some sub-national governments have already successfully undertaken significant public sector
reforms. In West Sumatra, a range of reforms have been initiated since decentralization that have attempted to
professionalize the civil service, introduce more rigor into procurement procedures, and introduce greater participation
into planning and budgeting procedures. While the outcomes of these reforms have yet to be seen, the new governor is a
well-know reformer who was most recently the bupati of Solok district, indicating that local leaders can prosper and
therefore continue to promote an ambitious reform agenda in the medium term.
Table 7.5
Locally initiated governance reforms implemented in West Sumatra province
Reform
West Sumatra
Kabupaten Solok
Kota Solok
(province)
(district)
(district)
Performance contracts for Echelon II officials
More equitable distribution of incentive payments

Proper tests for Echelon II or Echelon III-IV officials
External assessment for promotion
Reorganization under PP 08 (regulation on organization)
Pakta Integritas (integrity pact) for transactions with the public and the private sector
Giro (clearing account) to giro: financial transaction by bank without interference
Performance-based budgeting (Kepmendagri Ministry of Home Affairs Decree No. 29/2002)
Participatory planning and budgeting for service delivery

Strengthened procurement processes (Kepres Presidential Decree No. 80/2003)
Pakta integritas (integrity pact) for procurement

Source: World Bank, 2005e.
244
Box 7.7 Informal patronage behavior in the Indonesian civil service

Opaque arrangements for development budget-financed allowances appear to conceal an informal patronage network through which corrupt practices
may occur. Through what many acknowledge to be a semi-formalized system, significant discretionary allowances are distributed by top management in
individual agencies to their subordinates in exchange for loyalty and, frequently, collusion in malfeasance. Membership in such personal loyalty networks
is reputed to be pervasive, ensuring that officials can accept bribes and kickbacks without fear of reprisal, since their colleagues are likely to be engaged
in the same practice. Bribery is, reportedly, closely linked to the purchase of key positions. Indeed, official positions with access to the development
budget and control over the discretionary allowances it funds are reportedly a traded good, with a market value based on the estimated returns that might
accrue from the access to rents provided by the position. The loyalty network thus makes the job purchase possible, and the job purchase in turn creates
the demand for corruption since the position purchaser must recoup the purchase costs. In sum, the existence of discretionary allowances locks staff into
a loyalty network that enables extra-budgetary transactions to be conducted and shared under protected conditions.
This system of patronage also appears to allow civil servants to reap extra-budgetary rewards from unofficial payments, bribery, graft and returns from
rent-seeking. The types of transactions have been outlined in assessments of leakages from World Bank-financed projects. Case studies suggest that the
system allows management in a wet agency (i.e. an agency with access to the development budget and the capacity to purchase access to a project
listing, and hence eventual access to donor and counterpart funds) to reap informal benefits through the following representative transactions:
Payments from contractors and groups of contractors in exchange for selection (this may be recoverable subsequently through the project budget).
Payments from staff in exchange for hiring on projects.
Loan accounts structured so that interest earned on cash in hand is retained by the agency; this may be in collusion with commercial banks for a
share of the returns from the accumulated interest.
Provision of ghost services and/or inflated invoicing in collusion with contractors.
This system appears to be enabled by the significant contribution of the development (non-recurrent) budget to civil service rewards. Requirements for its
continued operation include a continued commitment to project financing by donors across a wide spectrum of agencies, and the current dysfunctional
split between the routine and development budgets, which allows for inadequate project supervision.
Source: World Bank, Indonesia: Priorities for Civil Service Reform, 2001. This box is adapted from Pay and Patronage in the Core Civil Service in Indonesia, Nick Manning, PRMPS, World Bank,
March 2000, pp. 33-36. Also published in Combating Corruption: Enhancing Accountability for Development. World Bank, 2004.
However, these ambitious reforms remain the exception at present. Many civil servants remain part of patronage
networks in which earning income and satisfying superiors is more important than achieving work output targets (see Box
7.7). These networks have existed for years and became more deeply entrenched during the New Order period. It is not
clear that they have changed significantly in the past five years. Thus, today, reformers are in a battle to change the status
quo by attacking entrenched formal and informal rules of the game that continue to prevail throughout much of the civil
service.
Competitive pressures
The private sector can both provide competition to the public sector, and also deliver some services
more efficiently and effectively than public service providers. Much of the dialogue regarding government
services and administration focuses upon how to improve efficiency and effectiveness through internal reforms. However,
equal attention should be paid to use of the private sector. The private sector is not only more cost effective in delivering
services in some cases, but the mere existence of private sector competition can, under the right conditions, stimulate the
public sector to deliver higher quality services to citizens (see Box 7.8 on healthcare in Jembrana district). In addition,
there are some tasks that the public sector can subcontract to the private sector, as these tasks do not fall within the core
responsibilities and/or competencies of the public sector.
245
Figure 7.8 Use of public and private service providers (by income quintile)
Source of drinking water
Primary education
Out-patient visits
Junior secondary education
Borrowing
Source: Drinking water, outpatient visits, and education data are from Susenas (2003); borrowing data are from the Indonesian Family Life Survey, 2000.
The private sector provides both high- and low-end services to clients. Not surprisingly, the private sector
provides high-quality services to wealthy clients in the form of private schools and hospitals. Perhaps more surprising, in
almost every sector, in addition to private providers of high-quality services, there are low-end private sector providers
that serve the poor (Figure 7.8). This opposite end of the spectrum exists for poor clients either because these services
are closely tailored to their needs, or because they are the only ones available. However, the poor are often charged
substantially more for these private services than by public facilities. A recent Voices of the Poor study found that prices
charged by private water vendors in urban slums in Simukerto and Antasari were 15-30 percent higher than those charged
by public water utilities (Mukherjee, 2006). Private religious schools provide education for the poorest, but their quality
fails to measure up to that of public schools (Newhouse and Beegle, 2005).
Key factors that impact upon the private sector s ability to provide services to the poor include:
accreditation; knowledge of service standards; provision of demand subsidies; and selective subcontracting to the private sector (World Bank, 2006k). If citizens have greater knowledge of the quality of private
service providers, they can make more intelligent choices. For example, accreditation of midwives through a governmentapproved testing process can provide clients with better information, especially regarding a topic on which a layperson
may have difficulty understanding. Demand subsidies, in the form of reimbursement to private providers by governmentmanaged healthcare schemes, have already been piloted for healthcare in both Jembrana and East Sumba (see Box 7.8).
This can be accompanied by an accreditation system for private providers. And finally, it is possible to contract out
selected services to the private sector, especially in remote areas where the public sector does not deliver high-quality
services. For example, there is a private market for mosquito spraying. Arguably, the private sector could arguably be subcontracted to do this job as it is a task that the private sector can do more effectively than the public sector.
246
Citizen and civil society voice

Decentralization and democratization offer an opportunity to
bring providers of services and poverty reduction programs
closer to the poor. Involving end-users and poor citizens in
development efforts is critical, as it improves accountability and the
likelihood that interventions will be designed to meet demand. The role
of communities in poverty reduction programs can vary both in terms
of the degree of their involvement and the stage at which they become
involved in the process (project identification, planning, budgeting and
programming, monitoring, delivery and implementation). Local
governments can support users to become involved in service delivery
by promoting their participation in the actual decision-making over their
participation, 218 and by appropriately assigning resources to
communities or to partnerships between providers and communities.
Box 7.8 How public-private competition can improve

healthcare for poor clients
In Jembrana district, Bali, health sector reforms created a new health
care program, the Jaminan Kesehatan Jembrana (Jembrana Health
Insurance). The program provided free primary healthcare for all enrolled
citizens and free secondary healthcare for all poor members. It also
enhanced client choice by enabling members to choose between private
and public healthcare providers, both of whom were reimbursed under
the scheme. In addition to increasing coverage of health services, the
scheme directly affected the behavior of public health staff, who had to
compete for clients with private providers in the wake of this reform. As
a result, public providers measurably improved their client orientation
by sending mobile clinics and doctors to remote areas at least once a
month (rather than just providing health education in these remote areas,
as had previously been the case); improving medicine packaging; and
providing full smile reception for patients. In addition, the management
board of the project supervised quality control for reimbursements by
both creating a clear set of service standards for all providers and
investigating cases of malfeasance.
Accountability relationships in Indonesia are still being

developed and the bureaucracy and the executive remain
powerful. Currently, the paradigm of an authoritarian state with upward
accountability remains very strong: citizens have a voice, but lack the
Source: Kuznezov and Ginting, 2005.
power to enforce the accountability of law-makers, or tackle the
bureaucracy. Oversight by an active civil society and parliamentarians
is essential, but is currently missing in the triangle of policy-maker-provider-citizen accountability.
Throughout Indonesia there are examples of how democratic checks and balances are starting to function,
but capacity remains a problem. Pro-poor leaders have tackled both the inefficiencies of local administrations and
encouraged civil society to play a role in scrutinizing the management of public office and funds. However, at present, both
sets of stakeholders governmental and non-governmental have limited experience and capacity. Indonesian organizations
are unfamiliar with the inputs necessary for planning or analyzing detailed budgets, and government is inexperienced in
structuring consultations effectively to receive input. In Boalemo district, for example, an ambitious civil-service reform
program was partially derailed because of capacity and institutional weaknesses in the district (see Box 7.6).
But to date, the failure of civil-society organizations to effectively assert pressure and control over
those who serve them is a key problem in the development of an effective framework for accountability.
Civil-society organizations have developed considerably since decentralization, and corruption watchdogs and consumer
associations are emerging that will increasingly act as a check on government behavior. However, a great deal of scope
exists for further empowering civil society and the public. There is a lack of transparency about rules, the fees and charges
that the public are required to pay for services, and the standards it can expect to receive from the civil service. The
experience with the Kecamatan Development Program (KDP) suggests that empowering communities to monitor
development performance and facilitating this process through civil society and media coverage have helped moderate
corruption (see Spotlight on Inefficiencies and Leakages at the end of this chapter).
Three factors impact upon the ability of citizens to effectively exercise their voice. First, the degree to
which citizens are involved in and influence planning and budgeting by their local governments; second, giving communities
the power to deliver some services themselves (for example, small-scale infrastructure); and third, the provision of vouchers
to the poor so that they are financially empowered to make choices.
218
Not all the poor want to be involved in all the processes, participation should be the decision of an empowered group.
247
The government has had very mixed success in enabling and fostering the participation in planning that
is needed to bring about the structural change in process. In developing its national strategy for poverty reduction
(SNPK), the government made proactive efforts to elicit participation from non-government stakeholders, although their
contributions to the preliminary draft varied. The process of revising the SNPK was a far more participatory one than
previous national planning efforts, albeit for a small group of stakeholders. There have also been modest changes to the
structure of national-level consultations with sub-national governments on the annual work plan (RKP).
The budget formulation process is relatively closed to non-government stakeholders. There are several
formal opportunities for parliamentarians to provide input to government: initially in discussions between the Budget
Committee and central agencies on macroeconomic assumptions, fiscal policy, and government priorities; and subsequently,
during sessions set aside for central agencies and the Budget Committee to discuss the draft budget. To date, however, it
is only the sectoral commissions219 that have engaged in any consultation with the public. Generally, this process has
been ad hoc and conducted by invitation to a limited number of individuals or organizations rather than through any
proactive, systematic or structured process. However, budget documents do meet a best-practice standard in terms of
public availability of information.220
The central government has already issued laws recognizing the roles users can play in planning and
monitoring service providers. Governments at all levels should now develop practical strategies for implementing
these laws to promote the wider involvement of user groups in service delivery decision-making and monitoring.
Indonesia has enjoyed impressive results from some community-based approaches to basic
infrastructure. Evidence on realized cost savings from four large community development projects shows that
communities were able to build infrastructure for as much 66 percent less than contractors (Table 7.6). Savings were
evident in both single-sector (water supply) and multi-sector programs. Using community-based approaches on largescale projects has the potential to save hundreds of millions of dollars. Cost savings achieved by community methods are
often particularly high for upland and isolated communities, where the costs of mobilizing contractors can be higher than
the cost of the construction itself.
Table 7.6
Economic benefits from community-managed basic infrastructure

Project
Total value
(US$ million)
Community contribution
(%)
Percentage savings of using community

rather than contractors (%)
Source: Bappenas, 2005. Findings of Post-Construction Economic Impact Analysis . Jakarta.
219
These commissions have hearings with the relevant ministries who fall under their area of responsibility. External experts are invited to these hearings, but not always. In addition, these hearings are not open
to the public.
220
IMF (2006). Draft Report on the Observance of Standards and Codes (ROSC): Fiscal Transparency Module. March 1, 2006.
248
At the local level, Indonesia has long engaged in the rhetoric of bottom-up planning.221 However, until
recently it has been usually described as a ritualized performance . Many studies have concluded that the
participation of the poor and near-poor is not a genuinely participatory process.222 While the formal process elicits village
proposals mostly from formal village leaders or village elite the results rarely connect to district-level planning processes,
sector or otherwise.223 At the same time, there is no institutionalized mechanism for citizen participation in the budgeting
process, and the process as a whole lacks transparency.224
Results from five rounds of the Indonesia Rapid
Box 7.9 Developing a high-quality, locally-initiated block grant
Decentralization Appraisal (IRDA) during the period 2002program in Blitar city
04 found that participation of civil-society organizations and
The Blitar community block grant program, implemented since 2002, provides
citizens in decision-making processes remains limited, and
grants for development projects in the 20 urban villages (kelurahan) of Blitar city,
most consultations are in fact a token effort too socialize
East Java province. Community participation is a key element. Communities are
the government s pre-determined positions and plans. The
very involved in selecting projects, mainly through annual city-led, governmentstudy found that bottom-up planning is still regarded as a
mandated meetings (Musrenbang), which elicit citizen input into city planning.
ritualized performance .225 A study of three districts in
Blitar also holds pre-Musrenbang , basically Musrenbang at the local level. At
eastern Indonesia found that only 20 to 25 percent of the
both Musrenbang and pre-Musrenbang, the active participation of attendees is
encouraged something that is rare in Indonesia.
budgeted activities originated from the bottom-up
226
participatory development planning process.
A number of institutional and attitudinal blockages
hinder participatory processes being adopted by
government. One of the underlying reasons why
community participation has not been fully realized is that
Indonesia s implementing organizations and staff adhere to
top-down norms and traditional attitudes. There has been a
lack of leadership commitment and little organizational effort
to create the structural change that is needed. Given the
already limited capacity at the local level to deliver services,
there is even less capacity to take on participatory processes
with which civil servants are neither familiar nor committed.
However, localized block grant programs have the
potential to both empower the poor and to provide
better outcomes for local citizens. While many donor
programs and localities have rolled out block grant programs
in recent years, targeting and quality have been consistent
problems. Administering a block grant program requires
local civil servants to work in a more participatory manner,
and to learn by doing adjusting program rules as lessons
are learned regarding those approaches that produce the
best results.
Communities financial and in kind contributions have ranged from 13 percent

to 22 percent of the total program budget annually (which itself increased from
Rp 3.62 billion, or about US$380,000, in 2002, to Rp 6.14 billion, or about
US$646,000, in 2004). Funded projects reflect communities preference for
infrastructure, but city government has been encouraging intangible but (in its
view) more efficient projects, such as training: beginning in 2005, no more than 60
percent of program funds may be spent on infrastructure. The Village Community
Empowerment Institutions (Lembaga Pemberdayaan Masyarakat Kelurahan, or
LPMKs) play an important role in mobilizing communities.
Although targeting the poor was not originally an explicit goal, there has been a propoor funding bias in two of Blitars three subdistricts, likely because of a new (2003)
project selection criterion on the number of poor beneficiaries. Program funding for
renovating slum housing has provided the most direct benefits to the poor. Recognizing
this, the city government mandated that, starting in 2005, 13 percent of program
funds would have to be spent on this activity.
The community block grant program appears to be technically, institutionally,
financially, and socially sustainable. It is also very low cost, representing less than
2 percent of the city budget. The mayors willingness to let communities make
mistakes has increased the likelihood of sustainability, not only because the program
is able to improve based on lessons learned, but also because communities are
learning how to implement programs over which they then have ownership.
Source: Kuznezov and Ginting, 2005.
221
The P5D (Pedoman Penyusunan Perencanaan dan Pengendalian Pembangunan di Daerah or Guidelines for Preparing Planning and Supervision of Development in Regions) from 1982 laid out a standardized
process for regional development planning.
222
See, for example, Norio Usui and Armida Alisjahbana (2003). Local Development Planning and Budgeting in Decentralized Indonesia: Key Issues and Catur Sugiyanto and Norio Usui (2003) Development
Planning, Budgeting, and Service Delivery: A Case of Lombok Tengah. Also Sukabumi Public Expenditure Review, North Sumatra Public Expenditure Review, and East Lombok Public Expenditure Review,
World Bank, (mimeo), 2002.
223
World Bank (2005). Local Legislative Institutions Research: Regional Planning and Budgeting, Observation Phase. July 2005.
224
International City/County Management Association (ICMA), Research Report: Changes in the Local Government Unit Structure for the Budgeting Process , mimeo, November 2003.
225
Asia Foundation. Indonesia Rapid Decentralization Appraisal. 1st-5th Reports. 2002-2004.
226
Deno Kamelus, Jessica Ludwig and Suhirman (2004). Study on the Efficiency and Effectiveness of the Planning and Budgeting Process in Selected Districts in NTB/NTT. GTZ PROMIS-NT. June 2004.
249
Vouchers can increase end-user choice and promote competition among providers. Vouchers allow
communities or users to periodically decide which provider offers the best service. Both vouchers and formula-based
programs implemented in order to promote competition among providers presume that users face a choice of providers
and are able to accurately monitor local choices and respond accordingly. These conditions rarely exist in isolated areas.
In urban areas and areas of high population density, however, both vouchers and formula-based funding strategies have
considerable potential to improve provider performance (see Box 7.10).
Box 7.10 When vouchers work for the poor-and when they don t
Distributing vouchers through providers limits the power of the client, because it does not allow them to choose their provider. Two Indonesian experiences,
the social safety net (jaringan pengaman sosial or JPS) scholarship program and the midwife voucher program, illustrate the problem. The examples also
show that strong incentives for providers to seek out new clients can increase effectiveness.
The JPS scholarship program provided scholarships to junior secondary school, but distributed through junior secondary schools. While the program
was intended to increase the number of students who went on to junior secondary school, most scholarships were given to students who were already
enrolled rather than to those who had dropped out. The implementation of the JPS program was not always consistent with a focus on the client. While
the program started as a direct transfer (provided by the post office), it soon effectively became a block transfer to service providers: scholarships were
frequently collected on behalf of parents by school principals or treasurers. Allowing providers to select the recipients, and failing to provide rewards to
school to seek out new clients (drop-outs), limited the program s effectiveness.
As part of the Safe Motherhood Program, vouchers were distributed to poor women in Pemalang district, Central Java, for the purchase of midwife
services during pregnancy. As a result of the program, midwife use by the poor rose from almost nothing in the late 1990s to 1,164 in 2000. Poor women
reported that they started using midwives for the first time only after the Targeted Performance Contracting for Midwives program placed midwives in their
villages. The salary of the midwives consisted of a low base, topped up with the money they received through the vouchers. This provided them with a
powerful direct incentive to seek out new clients.
Source: Ridao-Cano and Filmer, 2004; Tan, Kusharto, and Budiyati, 2005.
Recommendations
To improve accountability within government and between the government and citizens, three priority
areas of action are desirable. First, greater clarity of functions between and among central and local government
units, and the private sector, is necessary. Second an enhanced focus upon capacity building and incentives within the
civil service would be helpful. Finally, mechanisms for strengthening client voice regarding the performance of civil
servants should be improved.
1. Clarify functions between central and local governments, and within government units
Clarity of functions is an important prerequisite of good governance. The much-discussed need for better
allocation of functions for better governance in Indonesia requires action at both the national and local levels. The lack of
clarity in roles and responsibilities of government in sectors related to poverty reduction, and the obvious overlaps and
gaps that result, highlights the need to implement a rigorous policy of separating the regulatory, financing and delivery
functions. This requires a shift at the national level among agencies accustomed to having power over service delivery
budgets, and at the local level in understanding that not all services need to be delivered by public agencies. Moreover,
poverty reduction is spread across a number of sectors that function independently of each other, and greater coordination
will result in a better blending of poverty alleviation activities. Analysis of functions between the tiers of government, and
within each tier, is also an important step forward, and one that will clarify overlaps and gaps between tiers and service
units within the same administration. Work agreements between provinces and districts, such as those seen in West
Sumatra, including information-sharing on activities and resources, would facilitate a better operational environment for
service delivery.227
227
250
Both in West Sumatra province and in Yogyakarta province, district and provincial governments have tried to design systems of cooperation.
Focus on areas of institutional reform that would closely and directly

support poverty reduction. Notwithstanding the importance of basic civil-service
reform, there is a need to put in place organizational frameworks that promote
accountability of the various tiers of government in service delivery and other povertyfocused programs. While it is true that these are broad-based reforms, it is also possible
to sequence these reforms in part to address the uncoordinated areas that make the
most difference to the poor. To this extent, it is recommended that the government focus
on sectors driving the pathways out of poverty (for instance, education, agriculture,
rural roads) by developing accountability at the sector level through a series of linked
institutional changes. These changes should be focused on optimizing spending on
services that promote the pathways out of poverty.
Promote and develop the role of the private sector. Within the context of a
more regulated environment, and recognizing that one size does not fit all, efforts are
needed to develop and optimize the role of the private sector. The private sector already
provides services in health, education and water to different stakeholder groups. The
standard of service may be better or worse, or similar, to the government service. Price
varies too and can be a major concern. The key is that there should be some structure to the delivery of basic services to
the poor. There should be regulations that enable competition, and maintain the interest of the private sector, but nevertheless
control the quality of basic services that are so important to the poor.
Box 7.11 Increasing the benefits from the private sector
The private sector already delivers many services to the poor, but they are often of poor quality. The government can improve the services the poor receive
from the private sector by:
Informing the public of service standards of private sector providers. Consumers can make better choices about the use of private service
delivery if they are aware of the quality of the service they buy.
Extending training and recognition to qualified private providers that serve the poor. A combination of training and certification of providers
allows the poor to recognize quality through certification and enables private providers to improve the standards of services they deliver.
Making qualified private providers eligible for demand subsidies. Government could promote choice by making private sector providers of
recognized quality eligible to receive voucher-type subsidies. Districts such as Jembrana and East Sumba have already allowed private health providers
to be reimbursed under a district managed health insurance program (Gaduh and Kuznezov, 2006; and Arifianto et al, 2005).
Selectively contracting out service delivery to the private sector. In the short run, most contracting out to the private sector will be for services
that the public sector does not deliver. But the private sector can also assist the public service provider in delivering on its service agreements. Spraying
against mosquitoes, for instance, is usually carried out by public clinics. It would arguably be more efficient to use private providers, as there is a welldeveloped private market for this service.
There is already legislation that states the government s intention of providing more clarity in functional
allocations. It states that the central government should relinquish, in all agreed sectors, the role of provider and
concentrate efforts on creating an effective and conducive enabling environment for poverty reduction. As established in
PP No. 25/2000,228 the central government should focus on: (i) making and clarifying policy and legislation; (ii) mobilizing
funding for basic services, both through better inter-governmental systems (described above) and through good policy
for leveraging private and community funding; (iii) supporting local government to play its role both through increased
228
A revision to this presidential decree is in the final stages of promulgation, having already passed through the Ministry of Legal Affairs, but has not been circulated publicly.
251
clarity and a capacity development function; (iv) setting the framework for, and monitoring service standards for, local
governments to develop in the local regulatory context; and (v) supporting lagging regions through cross-regional initiatives
and dissemination of information from good performers to poor performers. To this end, line ministries structure and
staffing need to be rethought: the framework of functions, budgets and staffing of central ministries needs to be aligned,
starting with those sectors hindered by a lack of clarity and directly affecting the lives of the poor.
2. Focus on civil service reform by addressing hiring practices, capacity-building, sanctions and
rewards
While not easy, reform will be vital. While civil service reform is not easy, in the medium term it is a vital component
to unsticking other reform processes.
Review and provide clarity in the regulatory framework and incentives for organizational and personnel
management by instituting a less rigid employment regime and abolishing the system of structural and functional
positions and the rigid ranking of posts. A more open and competitive recruitment system, combined with easier transfer
mechanisms, early retirement and other schemes, could help right size and strengthen the civil service in many regions.
Review of the budget for a formula for intergovernmental transfers, such that regions are rewarded for rationalizing their
personnel establishments and can keep the savings they make on personnel expenses, will provide them with appropriate
incentives to have the right number of staff with the right skills. Finally, strengthening controls over, and rationalizing
arrangements for, civil-service employment and compensation, and linking personnel information systems to payroll,
could create cost savings.
Create more robust and predictable incentives for staff. Indonesia has pockets of success in developing appropriate
incentive structures for frontline service providers. In the best cases, incentives that drive service-provider behavior are
robust, timely, applied consistently, and linked to improved outputs and/or outcomes for clients. Examples in Box 7.12
below show that incentives have already been piloted successfully in selected localities.
252
Box 7.12 How can incentives change the behavior of frontline service providers?
Numerous experiments using incentives for local service providers have been conducted in Indonesia in recent years. In all cases, there was a marked
change in the behavior of service providers in response to a changed incentive framework.
In Tanah Datar district, West Sumatra, a scheme was launched in 2002 to provide stronger incentives to English teachers and headmasters by offering them
tours of Australia, Malaysia, and Singapore to study English and observe teaching methods. These trips motivated behavioral change in several ways:
The trips increased the motivation to do better work. Teachers returning from study trips submitted a group report to the mayor, with follow-up
observations and recommendations. Their observations included the need for stronger discipline of teachers, students, and parents; smaller class
sizes; and improvement in the quality of education by providing classes in computer skills and English, changing the teaching methodology, and
consulting with students.
The trips effected changes in teaching methodology by some English teachers, including those who went abroad and some who did not but associated
with those who did (as friends or colleagues). One English teacher began teaching her classes in English rather than Indonesian following her trip to
Australia. She has also begun using a student agenda, in which students record their activities in English, as well as what they have learned from
them, as an aid in teaching.
Interest in student performance and teaching hours increased, due to both school-based management and the stronger incentives policy. On average
students now study about 15 hours more a week. To demonstrate his commitment to raising student test scores, one headmaster actually signed an
agreement with his school committee stating that if scores at his school were not above a certain level he would resign.
In Jembrana district, Bali, health sector reforms created a new health care program, the Jembrana Health Insurance program (Jaminan Kesehatan Jembrana).
The program provided free primary healthcare for all enrolled citizens and free secondary care for all poor members. It also enhanced client choice by
enabling members to choose between private and public healthcare providers, both of whom were reimbursed under the scheme. In addition to increasing
coverage of health services, the scheme directly affected the behavior of public health staff, who had to compete for clients with private providers in the
wake of this reform. As a result, public providers measurably improved their client orientation by sending mobile clinics and doctors to remote areas at
least once a month (rather than just providing health education in these remote areas, as they previously had); improving medicine packaging; and
providing full smile reception for patients. In addition, the management board of the project supervised quality control for reimbursements by both
creating a clear set of service standards for all providers and investigating cases of malfeasance.
As part of the Safe Motherhood Program in Pemalang district, Central Java, poor women were issued vouchers that they could exchange for prenatal care
from midwives. The midwives were usually responsible for distributing these vouchers. With the added incentive of additional fees earned from clients
with vouchers, midwives substantially increased the number of poor women they treated. This had the additional beneficial impact of introducing poor
women to the formal health system and inducing them to seek out healthcare from formal providers more frequently.
Source: Leisher and Nachuk, 2006.
Thus, developing skilled, organized and accountable institutions committed to poverty reduction requires
the introduction of a focus on increased skills in, and knowledge of, poverty and poverty reduction. Steps
towards more poverty-concerned and capable local-level delivery institutions would include: (i) defining what skills are needed
for addressing poverty functions; (ii) assessing what skills exist to take on these tasks; and (iii) developing a staffing strategy that
matches staff to functions, and develops pro-poor skills and tools (e.g. community participation through the citizen report card).
3. Strengthen citizen and civil society voice at both the national and local levels
At the national level, enhance and institutionalize the consultative approaches developed to date. This
process should occur at the overall and sector levels with efforts to inculcate the processes into systems and procedures.
At the district and municipal level, pro-active efforts are needed to formalize and open up consultation processes.
Bappenas and the MoF should develop formal, regularized mechanisms for public input into the annual
planning and budgeting process. Formal consultations with academia, professional associations, labor unions,
farmers associations, service-provider NGOs, and business organizations should be held as a regular part of the process.
To create demand and develop capacity, it may be prudent to incrementally increase the level of participation. Although
formal consultations would by definition favor citizens who are presented by formal and established organizations over
those who do not enjoy such representation, this is still a step forward from the current situation.
253
Encourage sector ministries to establish consultative

processes on a more regular, ongoing basis to inform
priorities, policies and program budgets. Annual consultations
regarding overall government priorities and expenditure allocations are
not sufficient to develop effective inputs on the vast range of government
programs and activities. Consultation strategies need to be articulated
in terms of stakeholders, issues and processes. Irrespective of whether
central government undertakes consultation directly with communities
themselves or works with local governments, independent organizations
and/or citizen groups, the process involves reaching out into the
community and openly consulting the poor, and then effectively
synthesizing citizen views on, inter alia, access, adequacy and quality
of services and other supporting programs.
Integrate all forms of stakeholder provision public, private
and community in sector delivery strategies. Local governments
should develop strategies that bring together the benefits of all
stakeholders (public, private and civil society) to perform the roles most
suited to them. Effort is needed to develop a delivery environment in
which a range of public, private and CSO agents has an incentive to
deliver to the poor. To achieve this it is necessary to develop a receptive
operating context and identify strengths and weaknesses in the context
of effective strategies. Building the capacity of actors on both the supply and demand sides is critical to ensuring the most
effective contribution to the overall sector marketplace. Coordinating any continuing inputs from other government agencies
may also be important in the short term.
At the local level, improve access to meaningful information. Increasing transparency of government systems
and decision-making processes by providing clear information at the right time is key to enhancing accountability and
enabling participation in local-level planning and budgeting processes. With better knowledge of processes and better
public access to plans, budgets, and supporting information, the bargaining power of citizens and civil-society groups
will be radically improved. A range of low-cost approaches has been tested to improve the quality and quantity of information
flows and thereby ensure citizens have at least a basic understanding of local government priorities and expenditures. To
this end, district/municipal governments should be given incentives to establish minimum levels of transparency. These
might include: availability/publication of statistical data, key priorities and targets in the overall government plan, as well
as sectoral plans, and summary budget information (planned and actual). The central government could promote this
process by providing sub-national governments with transparency/information guidelines based on appropriate best
practice established in Indonesia and similar contexts.
Encourage multi-stakeholder consultations during the formal annual planning and budgeting processes,
and beyond. Opening up the planning and budgeting processes, and conducting consultations around key issues and
sectors at the district/municipal level may enable organizations speaking on behalf of the poor and near-poor to participate,
advocate, and possibly influence government priorities and spending. But, conversely, it is also possible that opening up
the process will increase the influence of individuals and groups representing the elite. District/municipal governments
can be encouraged to mitigate against this potential capture by integrating pro-poor processes in the consultations.
Engaging appropriate civil-society groups to facilitate focus group discussions on specific issues and focusing on specific
target groups (such as slum-dwellers, women, or youth) can also be beneficial.
254
Allocate block grants to villages and/or sub-districts to directly

assess needs, and develop and implement community-based
activities, using examples of best practice in community
participation.229 Indonesia has wide experience of community-driven
programs from which it can draw valuable lessons. Lessons learned point
to the need to match responsibilities to communities comparative
advantages, work with existing local institutions, invest in facilitation to
ensure informed decision-making and encourage non-elites to participate,
and monitor performance. Therefore, to make most effective use of the
existing system of village block grants, district/municipal governments
need to (i) explicitly define the responsibilities of village governments,
(ii) ensure that grants are of a meaningful amount, (iii) support a
participatory project proposal and selection process, (iv) promote
community contributions, either in kind or in cash, to bolster local
ownership, and (v) monitor both usage and results of the grants. In addition
to using trained facilitators and manuals from the various communitydriven development (CDD) projects, districts/municipalities can draw on
the expertise of organizations such as The Forum for Village Reform.230
Blitar city is one example of a very successful, locally-initiated block
grant program.
Utilize the strengths of communities to manage and deliver
services that target the poor. In many situations where incentives are low for government and the private sector to
engage in service delivery, there is strong evidence that well facilitated community interventions, in the Indonesian context,
yield more efficient construction and maintenance of village infrastructure.231 Communities also can play an important
role in targeting the poor, as providers have little incentives to reach out to those who do not demand services. For
instance, in the Kecamatan Development Program (KDP) communities have been instrumental in the delivery and
implementation process of a wide range of activities (village roads and basic infrastructure). Ex post evaluations indicate
lower costs and lower levels of corruption than in similar programs delivered by district administrations.
IV
Monitoring and Assessment of Poverty and Poverty Reduction

Interventions
To reduce poverty more effectively, it is necessary to build more knowledge and information on the
nature and determinants of poverty and then develop policies and programs that respond to empirical
evidence. The key starting point and basis for better understanding is through the (continued) development of good
data (both monetary and non-monetary indicators) that reflect the multi-dimensional nature of poverty and lead to better
understanding of poverty determinants across the diverse regions of Indonesia. Regular assessment of the key dimensions
of poverty at the regional level and better understanding of the differentiated needs of the poor are vital for Indonesia s
efforts to reach the MDGs.
229
Guidance from the ministry suggests that each district/municipality develop its own policy in collaboration with the DPRD and other stakeholders. It further suggests that district/municipal governments
provide villages with a flat amount (consistent with past practice where villages received the same fixed amount for administrative expenses) as well as an amount to be allocated using variables related to
poverty levels, basic education levels, and health status (although it is not suggested that the variables should be linked to the objectives of the grant).
230
Forum Pengembangan Pembaruan Desa (FPPD) and Forum Pengembangan Partisipasi Masyarakat (FPPM), Membangun Tanggung-Gugat Tata Pemerintahan Desa , August 2004. FPPD has, for example,
developed a training handbook with several modules on participatory budget formulation and monitoring at the village level.
231
Bappenas (2005). Findings of Post-Construction Economic Impact Analysis Jakarta. Also see CGI Brief Investing Growth and Recovery. June 14, 2006. Jakarta.
255
To date, the government has assessed poverty through annual household

surveys that have provided very sound quantitative data at the national
level. Indonesia has been successful in collecting good quality monetary indicators,
and has also developed some useful non-monetary indicators of poverty across the
regions through the Susenas232 household survey. The National Socio-Economic
Survey (Survei Sosial Ekonomi Nasional, or Susenas) was initiated in 1963-64 and
has been fielded every year or two since then. Few developing countries have
established such large-scale surveys that allow households welfare to be assessed.
Indonesia has systematically stepped up poverty monitoring as a key
aspect of its development program, but these systems have deteriorated
since decentralization. Since the elimination of the deconcentrated central
government agencies,233 many ministries indicate that there has been a significant
decline in the reporting of administrative or routine data. Moreover, while the Central
Bureau of Statistics (BPS) had established a solid reputation in terms of the number
and quality of its statistical products, since the financial crisis it has been subject to
budget constraints and a freeze on staffing levels. This has raised questions about
the quality of current technical and operational capabilities.
Notwithstanding the commendable efforts in Indonesia in relation to data collection, there is still a lack
of information at the regional and district level. Given fiscal and functional decentralization, this is where information
is now urgently required. Many local governments do not have real-time access to information from many national
surveys (see below), and thus are not as well equipped as they could be to make planning and spending decisions.
I.
For some important non-monetary indicators, such as infant mortality, maternal mortality and nutritional
status, data compiled by the Indonesian Demographic and Health Survey (DHS) are not collected on an annual
basis and in most cases are only aggregated at the national level.
II.
Regional coverage of data is lacking. Unfortunately, most household surveys are only reliable at the
regional/provincial level, but program design and budget allocations (including the DAK) require a more precise
definition (below the kabupaten level) of where the poor reside.
III. Quantitative data are not supported by systematic qualitative assessment. Indonesia has some
experience in participatory poverty assessment. However, this is far from widespread and is not mainstreamed
at the local levels of government or in national approaches to poverty assessment.
IV. Routine and administrative data at the district/municipal level are lacking. The regional branches
(Kanwil/Kandep) of the central government that previously collected and reported these data no longer exist.
While some districts now collect data, these are not necessarily fed into national-level information systems.
V.
232
Data collection efforts are not well coordinated across departments. There is no system to coordinate
the collection of data by the BPS and line ministries. Incentives tend to foster independent data collection efforts
(for example, data associated with individual programs or projects) and there are gaps in the development of the
sorts of poverty diagnostics that are needed.
Since 1993, Susenas surveys cover a nationally representative sample typically composed of 200,000 households. Each survey contains a core questionnaire which consists of a household roster listing the
sex, age, marital status, and educational attainment of all household members, supplemented by modules covering about 60,000 households that are rotated over time to collect additional information such as
healthcare and nutrition, household income and expenditure, and labor force experience.
233
Kanwil at the provincial level and Kandep at the district/municipal level.
256
But the most critical problem is that the information available is not effectively used for decisionmaking on policies and budget allocations, inevitably affecting outcomes. Quite irrespective of the limitations
of the information being compiled, too often the data are not used to confirm suspected problems or issues, to develop
policy and design programs, or to monitor implementation either at the national or the local level. BPS data are collected,
published and used effectively by academics, but policy-makers rarely use them to make decisions on poverty reduction
interventions. Line ministry data are not regularly published or disseminated (even within line departments) and also go
unused in the design of sector policies and programs.
Monitoring is a key aspect of poverty reduction and making progress towards the MDGs. It takes two
forms: monitoring of poverty (as carried out by the government through Susenas, etc, and described above); and monitoring
of the implementation of plans, programs and expenditures. Strengthening monitoring of interventions is particularly
important if the central government is to adopt more results-oriented processes that increase incentives. Currently, most
programs are monitored to some degree, but with ill-defined objectives and inconsistent indicators; and the data collected
may not be fed back to make program adjustments or used to take other necessary actions.
Recommendations
1. Make better use of existing and emerging data sources to both better understand poverty and program
impact
The national data system could be improved. While Indonesia already has a good national data gathering system,
the existing system could be improved to provide data in a more timely manner to a wider range of stakeholders. While data
gathered is already comprehensive, the household survey could benefit from including a facilities survey as well, which
would capture the characteristics of key local service-provision organizations.
Strengthen systems to both understand and analyze poverty in the decentralized context. This means
creating and analyzing data sets faster, so that they can be used by planners to revise programs and budgets in the light of
new findings. In addition, it means making information more widely available to local planning bodies (Bappedas) so that
they can use the information for local planning.
Use small area poverty maps that are accurate to the sub-district level more systematically. Small-area
mapping is a relatively new technique that holds great promise in illustrating key poverty variables, and is accurate to very
local levels. It enables a much more precise understanding of local poverty, and can be a powerful tool in devising locallyappropriate solutions to these problems (see Spotlight on Innovation at the end of Chapter 6 on Social Protection for
more information on this type of poverty mapping).
Although the potential for better poverty diagnosis exists, Indonesia needs to meet a number of challenges
to benefit from these new technologies and methods. These challenges include ensuring technical and financial
sustainability by deepening governments technical skills, utilizing and updating information more frequently, and blending
qualitative and quantitative methods to understand better why people are poor,234 and promoting better program planning.
A good first step would be to focus upon the technical and analytical skills of the Bappedas, which are directly responsible
for understanding and using information provided by these surveys.
234
For an example of the relative merits of qualitative and quantitative methods, see: Kanbur, Ravi. ed. (2003). Q-Squared: Qualitative and Quantitative Methods of Poverty Appraisal. Permanent Black, Delhi.
or Eckardt, Sebastian and Stefan Nachuk (2003). Measuring Governance: Art or Science? World Bank. October 2003 (draft mimeo).
257
2. Link monitoring findings more effectively with feedback to, and design of, sectoral programs and
for use by local governments
Monitoring findings should be made use of. While the generation of accurate and useful findings is important, it is
also vital to use these findings to improve budgeting allocations and program design. The central government also has a
key role to play in ensuring that these findings are easily available to local governments in a timely manner.
It would be desirable to strengthen the role of the central government to enable better monitoring of
regional poverty and links to poverty reduction expenditures. In order to enable (i) the expansion of the Susenas
to cover specific remote and disadvantaged regions, and/or facilities (e.g. schools, health clinics); and to link this nationally
to (ii) the development of a set of public expenditure tracking surveys (PETS), which tracks the flow of funds and determines
the extent to which target groups are being reached, it may be necessary to enhance the role of the BPS and allocate it
additional finances. In addition, at the sector level, line ministries need to work with regional sector offices (Dinas) to
identify and agree on what information needs to be shared, in what way, and what incentives need to be provided to
regional governments.
Develop approaches to feed the information collected more systematically into program design. The
actual use of program results in determining ceilings or ultimately budget allocations is still some years away. Indeed,
while many countries include integrated performance targets in budget processes, few use this information to determine
the levels of allocation.235 As an immediate measure, however, ministries should document the findings of program
assessments/evaluations as part of their annual work plans and budgets, and these should be proactively and widely
disseminated, through ministry websites, workshops, legislative assemblies (the DPR, the DPRDs and the DPD), regional
governments and their associations, universities, think-tanks and other CSOs.
Establish a multi-year program to proactively assess the poverty impacts of selected programs. Due to
limited funding and capacity, it is inevitable that only a few programs will be rigorously evaluated. Therefore, the government
(central agencies with selected ministries) needs to strategically select major programs to evaluate for their poverty
impacts. The lessons learned from these evaluations should then be used to inform decisions on the expansion, modification
or elimination of the portfolio of programs.236 For key poverty-related initiatives, such as social protection, employmentcreation, micro-credit, SME support, community-driven development and rural infrastructure all central to the pathways
out of poverty program assessments should be undertaken. There are, for example, some 127 SME assistance programs
operated by 13 government institutions.237 It seems appropriate to ask if their objectives are still relevant and if they are all
meeting their objectives.
Develop systematic approaches to elicit and utilize citizen reporting and feedback at central and regional
levels. Any stakeholder, be they national and sub-national governments, citizen groups, or communities can solicit and
compile citizen views on poverty reduction programs. Also known as citizen scorecards or report cards , this form of
participatory reporting can be used to monitor access, adequacy, quality and/or general citizen satisfaction, and is a
powerful tool to obtain feedback from the poor and (socially and regionally) marginalized groups. The recent rapid appraisal
235
While many OECD countries have integrated performance targets into their budget processes, only a few use the performance information in making decisions on the level of funding provided. See Public
Sector Modernisation: Governing for Performance. OECD Observer. October 2004.
Strengthening monitoring processes is particularly important for the central government to make greater use of specific purpose conditional transfers (DAK), especially if they were to be performance based.
To the extent that the central government can start a process of increasing incentives and moving to resource allocation through block grants that are conditional on performance, it will require the associated
monitoring mechanisms. Related, the development of any conditional cash transfer program will also increase the necessity of efforts to improve monitoring at the district/municipal level and lower.
237
SMERU Research Institute (2004). Mapping Assistance Programs to Strengthen Microbusinesses. SMERU News. No. 10. April-June.
236
258
of the unconditional cash transfer (UCT) program represents

an example of such an effort, in which the views of
beneficiaries and other stakeholders were sought in order to
respond to policy-makers need for information. Best practice
in Bangalore238 illustrates the benefits of report cards in
improving service delivery to the poor. Findings can, indeed
should, be used in the preparation of plans and budget
proposals.239 Given the time and resources involved in their
development, the focus initially could be on frontline services
(e.g. health clinics, schools) and programs targeted
specifically to the poor.
A number of effective initiatives provide good
practice for program monitoring and feedback.
Following the rapid design and roll-out of the PKPS-BBM
programs (the three main programs funded by reallocated fuel-subsidy funds, together with the UCT) the government
instigated a process to monitor program implementation. The recently conducted qualitative assessments of the
implementation of these major fuel compensation programs (in health, education, and village infrastructure, and the
UCT), are examples of good practice in conducting program assessments to improve design and implementation.
Build on the institutional successes of Indonesia s rural development program. The Kecamatan Development
Program (KDP) has achieved considerable success in providing basic services to the rural poor, connecting villages into
the country s growth and vastly improving their access to basic services. The scaling-up of this program now a primary
government goal will require government to mimic the skills and organizational arrangements that have made the program
so successful.
VI
Conclusion: Towards Government Services Focused on the Poor
The public sector has the potential to work much more effectively for the poor in Indonesia. While
Indonesia may have made significant strides in reducing poverty over the past 30 years, the public sector has not always
played its full role in contributing to this success. This situation must change going forward if Indonesia is to achieve the
MDGs by 2015 and its own more ambitious medium-term poverty reduction goals. This chapter attempts to analyse the
reasons behind the less-than-optimal performance of the public sector and to suggest ways of gearing in the policy and
implementation framework in such a way as to better serve poverty reduction efforts. This exercise is also conducted in
the context of decentralization, which now, in the post-crisis environment, offers Indonesia a unique window of opportunity
for far-reaching changes and improvements in government performance.
Three key areas of government action need to be addressed if the public sector is to start to work more
effectively for the poor. First, there is a need to improve policy alignment, and also augment planning and budgeting
systems. Second, there is a need to strengthen institutional accountability. Third, it will be necessary to enhance the
assessment and monitoring of poverty reduction efforts by government if progress is to be measured and assured.
238
For more detail regarding methods and results in Bangalore, see: Ravindra, Adikeshavalu (2004). An Assessment of the Impact of Citizen Report Cards on the Performance of Public Agencies. World Bank,
Washington, DC.
239
Surveys should also specifically include questions on the treatment the poor receive in their interactions with program officials. The citizen/community perception information collected could be combined
with independent assessments by research institutes and/or self-assessments by program officials and service providers. Results should be disseminated to the public, and used to make improvements/
refinements in program design and implementation.
259
Achieving better policy alignment, together with improved planning, budgeting and allocation systems.
There are four priority areas requiring action in order to strengthen policy, planning and budgeting systems. First, improved
systems for linking poverty reduction priorities with sectoral plans and budgets would help to translate political priorities
into budgetary allocations that more closely reflect these priorities. Second, strengthened capacity and incentives for propoor planning and budgeting should be developed through improved technical guidance from Bappenas and the Ministry
of Finance. Third, district and municipal planning and budgeting systems could be made more effective in terms of
poverty reduction by improving poverty analysis, and by encouraging capacity building, and the use of incentives and
participatory processes. Finally, the DAU and DAK transfers the main sources of local government revenues can be used
more effectively both to address poverty in the poorest areas of the country, as well as more effectively reward good local
government performance.
Strengthening accountability both within government, as well as between government and citizens. In
order to improve accountability within government, and between the government and Indonesian citizens, there are three
priority action areas: first, there is a need for greater clarity of functions between and among central and local government
units, and the private sector; second, more focus should be placed upon capacity building and the use of incentives within
the civil service; and, third, there is a need to strengthen client voice regarding the performance of civil servants.
Improving monitoring and evaluation systems, and using the results more systematically in budgeting
and design. First, it is important that the government makes better use of existing and emerging data sources to both
better understand poverty and program impact. While Indonesia already has a good national data-gathering system, the
existing system could be improved to provide information more frequently and to a wider range of stakeholders. While the
data gathered are already comprehensive, the household survey could benefit from also including a survey of facilities.
This would capture the characteristics of key local service-provision organizations. Second, there is a need to link monitoring
findings more effectively with feedback to and design of sectoral programs, and for the use of these findings by local
governments. While generating accurate and useful findings is important, it is also vital to use these findings to improve
budgeting allocations and program design. The national government also has a key role to play in ensuring that these
findings are easily and frequently available to local governments.
260
Focus on Annual Plan and Budget Preparation for 2006
The government s fiscal year begins on January 1, and preparation for each
annual work plan and associated budget begins about one year beforehand.
January-April 2005
Bappenas prepares preliminary government work plan (RKP).
Bappenas prepared a preliminary draft of the annual government work plan
(RKP), which established priority policy areas based on Bappenas interpretation
of how the three overarching development agendas in the medium-term plan
(RPJM) translated on an annual basis (incorporating current realities, notably
the need to respond to the natural disaster in Aceh and North Sumatra). The
priorities were discussed in cabinet and revised based on its input. The most
significant change was the introduction of strengthening defense capability as
a priority, in response to a controversial border dispute with Malaysia that was
ongoing at that time.
Fiscal policy, indicative ceilings and final government work plan
approved by cabinet. Parallel to the above process, the Ministry of Finance
developed the fiscal policy and, together with Bappenas, the indicative ceilings
for each ministry. Finance apportioned the non-discretionary spending across
ministries, while Bappenas allotted the discretionary , based on the cabinetapproved priorities. Allocations for the ministries of health, education and public
works, as well as the army and police, were accordingly increased. Cabinet
reviewed the proposed aggregate expenditure budget and sectoral ceilings
together with the proposed RKP priorities (as above). A joint circular letter was
sent to line ministries outlining the overall government priorities and the
indicative ceilings as the basis for preparing their preliminary draft annual
work plans (Renja-KL).
Ministries prepare work plans. The internal ministry processes varied,
but generally the Planning Bureaus were responsible for coordination. For
2006, the process was somewhat complicated as ministries were preparing
their medium-term plans (Renstra-KL) at the same time. Bappenas sector
Directorates informally liaised with ministries as they prepared their Renja-KL,
and used ministry input to finalize the RKP. Ministries discussed their
preliminary draft Renja-KL with the relevant parliamentary committee at the
same time as the Ministry of Finance and Bappenas discussed fiscal policy
and the overall government plan with the Budget Committee (see below).
Discussions tended to focus on details of deconcentrated spending in regions,
with many committee members lobbying for their respective constituencies.
May-August 2005
Discussions with parliament begin, and ministries prepare work
plans and budgets (RKA-KLs). The Ministry of Finance presented the fiscal
policy and Bappenas the government priorities to the parliamentary Budget
Committee. Based on the results of these discussions, the Ministry of Finance
revised the ceilings that were used by ministries together with input from the
parliamentary discussions to convert their Renja-KL into work plans and budget
submissions (RKA-KLs). RKA-KLs included budget details at the program and
activity level, as well as performance targets and indicators (which were of
mixed quality). In many cases, the 2007 estimates are guesstimates, and it was
widely acknowledged that the process of estimation will need to be improved
upon in future years. It was also recognized that the budget is not yet truly
integrated. For example, some ministries simply took their salary budget and
put it against one program. The draft RKA-KLs were subsequently reviewed by
Bappenas for congruence with the RKP priorities and the Ministry of Finance
for consistency with the ceilings. This was a fairly cursory exercise due to the
limited time available, as were the discussions with parliamentary committees
on the draft RKA-KLs. The draft state budget was approved by cabinet before
being submitted to the parliament by the president.
September-December 2005
Finalization and parliamentary approval of state budget and RKAKLs. During September and October, iterative discussions between line
ministries and the Ministry of Finance took place, focused on the detailed input
norms for the proposed program spending. Ministries had further discussions
with their respective parliamentary committee, while the Ministry of Finance
and Bappenas met again with the Budget Committee. Due to the significant
reduction in fuel subsidies announced in September (effective October 1), the
Ministry of Finance and Bappenas agreed on a plan for reallocating the resources
which was presented to a limited cabinet meeting (coordinating ministers and
selected line ministers) for approval. A significant amount was reallocated to a
new program of unconditional cash transfers to the poor and near-poor to
mitigate the inflationary impacts of the rise in kerosene and other fuel prices,
as well as to doubling the size of the special allocation grant (DAK). Other
savings were allocated to programs intended to increase access to quality health
and education services for the poor. The revised state budget was approved by
the parliament at the end of October. Ministries then have to apply for, and
receive, spending authorization from the Ministry of Finance.
261
Focus on Elements of Good DAK Design
The DAK is a fiscal transfer to sub-national localities that has the potential to
more effectively target poor localities and lagging sectors and thus have a
stronger impact against poverty.
Sector, objective and expected results: The overall purpose of the DAK
in law is twofold: (i) to address basic services that have not meet certain
standards; and (ii) to support areas requiring accelerated development. At
this level of generality, these are consistent with the government s poverty
reduction goals. Selection of the specific DAK objective(s) should consider
national poverty reduction objectives and targets, and spending trends. The
capacity of the sector ministry could also be a factor.
Conditions: Closely linked to the decision above are the conditions to be
applied. These could relate to how services are delivered, which services are
delivered, or to which users. Consideration also has to be given to sanctions
for breach of conditions, as well as incentives to meet the conditions (e.g. to
build capacity of recipient governments to effectively spend the grant).
The law requires that district/municipal governments match a minimum of 10
percent of the DAK. If it is anticipated that local governments will be very
responsive to achieving the national objective, then the matching amount could
be higher. Since fiscal disparities remain high, the matching requirement could
vary inversely with fiscal capacity. It could also vary inversely with per capita
costs of providing the service(s), although this might be difficult if the objective
and conditions were results-oriented. Since a matching requirement may only
become a substitute for the expenditure that the sub-national governments
were funding, the government can include a maintenance of effort condition,
and then match additional effort to the DAK funding.
Open- or close-ended: The DAK can be open-ended grants, that is, regions
receive however much they are prepared to match (and spend). This could
provide an incentive for greater regional spending on national priorities and,
therefore, potentially improved results (assuming regions have the matching
resources and there is a strong monitoring of adherence to conditions). This
makes it difficult, however, for the central government to predict what its
expenditure will be. Moreover, there is a risk that if regional use of the grant
was oversubscribed, the central government will need to stop disbursing,
leaving a very messy situation. The more typical situation is to use closedended grants in which the central government sets the total pool, as well as
individual district/municipal shares, upfront. This ensures that governments
with greater fiscal capacity are not able to access a greater proportion than
needs-based criteria would suggest simply because they are better able to
provide the matching funds.
Timeframe: Developing and allocating a DAK on an annual basis could be
useful for experimenting . A multi-year timeframe, however, would enable
district/municipal governments to more effectively plan-both for designing
programs/services that will meet the grant objective and conditions, as well
as for raising their matching share (assuming it is additional funding). A
multi-year timeframe is also consistent with the move towards medium-term
budgeting.
Allocation criteria: The allocation of the grant between districts/
municipalities is the most political aspect of grant design and implementation.
262
Good practice suggests that allocation should be based on objective, simple

and transparent criteria, linked to the objective and conditions. In addition, the
capacity to collect and analyze the relevant data needs to be taken into
consideration. Data for each of the three types of criteria [(i) general criteria
(defined as fiscal capacity); (ii) special criteria (defined as regional
characteristics, specifically Aceh, Papua and eastern Indonesia, as well as
regions near borders, underdeveloped areas, coastal areas, and post-conflict
regions); and, (iii) technical criteria as determined by sector ministries (and
implicitly related to grant objective and conditions)] should be statistically
sound, regularly updated, and preferably come from an independent source
that cannot be manipulated and is respected by all stakeholders.
Size of the funding pool: Individual DAKs need to be of a meaningful size,
both to warrant the time and resources necessary to design and manage its
implementation properly, and to ensure regional governments active interest
in the process. The central government should have an idea of what it expects
regions to achieve with the funds and how these might be allocated across
regions before setting the size of the pool.
Implementation: A range of administrative matters need to be determined
before a grant is implemented, including timing. Ideally, there should be
sufficient time between the dates by which the central government and subnational budgets are due to allow the central budget to be approved and subnational governments to be advised of their grant before the sub-national
budgets are due.
A key aspect of managing the DAK is the reporting, monitoring and evaluation
arrangements. Sub-national governments are already expected to report on a
regular basis to the MoF and relevant sector ministry. It is important that
reporting requirements include data allowing an assessment of whether
conditions have been met and the effectiveness of the DAK. The central
government needs to both verify that the grants are used according to their
intended use (including compliance with conditions), as well as analyze the
information that is reported, to adjust as necessary grant conditions and criteria.
Consultation: At a minimum, sub-national governments should be consulted
on the various design features of a DAK, in particular on the conditions, criteria
to be used in allocating across regions, and reporting and monitoring
arrangements. Such consultation could be difficult to structure given the number
of sub-national governments, but it is critical to ensuring that there is common
understanding.
Legal Instrument: Currently, description and allocation of DAK grants are
legally enshrined annually in a Ministerial Decision (KMK) of the MoF, and
sector ministries prepare and distribute technical guidance or a manual for
their respective DAK. An alternative that could be considered is to have a formal
grant agreement between the two levels of government. This would: (i) ensure
there is the necessary degree of specificity and clarity (particularly if there are
asymmetrical arrangements); (ii) ensure that both levels of government are
clear on their respective rights and obligations; and, (iii) provide a solid basis
for reporting and monitoring, including by the community. The signing of a
formal agreement can also be a visible indicator of the governments intentions,
and can be the basis for socialization.
Spotlight on Innovation
Placing poverty reduction at the center of district annual plans and budgets
Lessons learned from the experience of developing district poverty reduction

strategies and action plans (PRSAPs) under the Initiatives for Local Governance
Reform (ILGR) Project are useful for improving the poverty orientation of the
district annual plans and budgets. Under the project, multi-stakeholder groups
worked to identify and agree on critical local poverty issues and then formulate
a strategy and action plan to be implemented in order to reduce poverty.
(1) Consultation and participation. In all ILGR districts, multiple
stakeholders were involved at all stages in the formulation of PRSAPs,
demonstrating that they have the capacity to formulate plans/strategies.
Moreover, it shows that stakeholders are willing to become involved
voluntarily given the right facilitation and support. It needs to be
recognized that participatory plan and policy formulation takes time, and
usually requires building capacity of both government and nongovernment stakeholders. Moreover, specific efforts need to be made to
include the views of the poor, including using participatory poverty
assessment (PPA) tools and methodology.
(2) Poverty focus. Key to developing the PRSAPs was poverty diagnostics
using district-level data, as well as information gathered directly from
poor households. Where there was a well-rounded poverty diagnosis,
that information was used as input in setting priorities and proposing
new policy directions and/or programs. It was sometimes difficult,
however, to integrate the district level (often quantitative) data with the
qualitative findings from the community level-requiring changes to the
PPA tools. Whether it is quantitative or qualitative, poverty diagnostics
should be used in the annual planning and budgeting processes by Dinas
in considering the targets they set, as well as the programs/activities, but
also in the setting of inter-sectoral priorities and budget allocations. BPS
poverty maps, for example, can be used by districts to improve geographic
targeting of their spending.
(3) Results orientation. While an explicit orientation towards the
attainment of results was not emphasized in the PRSAP process, it
underpins the new national and sub-national planning and budgeting
regulatory framework. A results orientation would need to gradually be
integrated into the annual process of establishing plans and budgets and
just as importantly, into implementation. At a minimum, Dinas can start
to define priorities in a results-oriented way and ensure targets are
measurable. This would allow for implementation monitoring and

provide useful feedback if used appropriately. This will, however, require
consistency and coherence between the annual plan and the budget.
As much as there are technical capacity issues associated with integrating
poverty into plans and budgets, and in developing plans and budgets in a
participatory manner, the politics matter just as much. The above assumes
that there is a commitment from the head of government and the district
parliament to poverty reduction, including funds to support a consultative
process that includes the poor, and a willingness to allow the results of such
a process to influence the decision-making process.
Formulating and institutionalizing regional poverty reduction
strategies in Bulukumba district
Bulukumba was one of several districts that participated in ILGR. The bupati
established a multi-stakeholder group (Pokja), dominated by civil society,
that worked together to formula a strategy and action plan to reduce poverty.
The process was perceived as being participatory, not just consultative. The
PRSAP was developed by all the stakeholders in the Pokja, and the public
was involved in different ways throughout the process. While the Pokja found
that it was difficult to get the involvement of the poor in district-level
consultation meetings, women were active participants in the working group
and at all steps in the formulation of the PRSAP. This success was, at least in
part, due to previous experience in using participatory approaches and the
willingness of civil servants to be responsive to civil society and the public,
including the poor. Based on this experience, the formulation team for the
medium-term development plan (RPJMD) was multi-stakeholder, and several
focus group discussions and public consultations were held as part of plan
formulation.
The bupati made the PRSAP a formal reference for all Dinas, and several
proposals from the PRSAP were adopted: (i) introduction of cross subsidies
and scholarships for poor students; (ii) establishment of a voucher system to
increase access for the poor; (iii) formulation of a mobile training unit to
provide skills training for microenterprises; and, (iv) clean-up of polluted
areas.
263
Intensive efforts to curb corruption in village infrastructure development: Lessons from Indonesia
All projects in Indonesia operate in a high-risk environment when faced with

issues of leakage and corruption. Because of this, new World Bank projects in
Indonesia are now required to include an anti-corruption action strategy as
part of their project designs. One project that has seriously focused on fighting
corruption is the Kecamatan Development Program (KDP). KDP is a US$1.3
billion World Bank community-driven development project that was piloted in
25 villages in 1997, just before the East Asia financial crisis, and has since
scaled up to fund small-scale development activities in some 34,200 villages
nationwide. Its approach to combating corruption is based on an analysis of
the political economy of corruption in Indonesian villages and on mapping
potential points of leakage. The case is useful to understand what works to
limit corruption in a large, rural development project in a country with endemic
corruption, a weak legal system and a history of top-down political control by
a powerful state bureaucracy.
How does the KDP approach attack corruption?
KDP limits corrupt practices by: (i) significantly reducing the complexity of
project design and procedures; (ii) ensuring that all information and transactions
are transparent; and (iii) responding quickly to complaints.
Some of the concrete steps to reduce corruption in KDP include:
264
Simplifying financial formats so that they can be understood easily by

villagers;
Transferring funds directly into collective village bank accounts;
Insisting that all financial transactions have at least three signatures and
that at least three quotations are found for the procurement of goods, to
be shared publicly at village meetings;
Insisting that basic project information and details of all financial

transactions are posted on village notice-boards;
Requiring that regular village meetings are held to account for project
funds at which villagers have the right to suspend further disbursements
of funds if irregularities are found;
Providing village-level sources of information and channels for complaints

independent of local government;
Applying sanctions or suspending projects, when necessary;
Ensuring intensive field-level supervision by elected village facilitators

and sub-district level project facilitators;
Maintaining an internal, roving financial supervision and training team

who provide financial training to project stakeholders as well as conduct
spot-checks of project financial records; and
Optimizing independent monitoring of the project by provincial NGOs

and local journalists.
While it is difficult to reduce corruption in its entirety, even with the

aforementioned measures, KDP has achieved some concrete evidence of
success:
Over the past several years, World Bank-commissioned independent

audits by Price Waterhouse and Moore Rowlands reported no significant
qualifications. Fund irregularities were less than one percent of total project
funds.
Audits by the official government audit agency (BPKP) also found few
leakages and no major abnormalities.
KDP s internal financial supervision and training team audits an average

of 18 percent of KDP kecamatan throughout the country. Less than one
percent of total grant funds that were checked indicated abnormalities.
Several independent evaluations show that village infrastructure built

through KDP methods costs significantly less-on average 56 percent lessthan works of equivalent quality built through the Ministry of Public Works
or through local government contracts.
There is evidence also of some governance spillovers from KDP, illustrated

by examples of villagers using their experience from KDP as a precedent
for protesting against corruption in other projects.
Experiments in measuring corruption

KDP has always attempted to learn from its experience and add to its anticorruption arsenal. In 2002, the project commissioned field research to
examine innovative approaches to reducing corruption. The study
conducted a randomized, controlled analysis of corruption in 600 KDP
road-building projects in East and Central Java during the period 2002 to
2004. The research examined three different approaches:
Normal KDP projects.
Projects in which participation was deepened by inviting many

villagers to public meetings, and/or by encouraging them to write
anonymous comment cards regarding the local project.
Projects in which villagers were told in advance that their project

would definitely be audited, raising the chance of an audit from the
normal 4 percent to 100 percent for these villages. Audit findings
were then announced in public village meetings.
The study found that:
The intervention of announcing an increased probability of a

government audit, combined with reporting the audit results directly
to a public village forum, was more effective at reducing corruption
than increasing villagers participation in the monitoring process.
Second, the absolute levels of corruption are heavily dependent upon

calibration assumptions. Some of the missing amounts could be attributed
to corruption but most of it could just as rationally be attributed to other
factors such as standard engineering loss rates and sheer inefficiency.
Factors such as weather, erosion, compacting of materials, transport losses,
difficulties of field measurement, and incompetence are all difficult-tomeasure ingredients in the missing amounts or loss ratios .
Third, even in the audit intervention, community participation was critical

because the plan to conduct audits was announced and disseminated to
villagers beforehand, and audit results were read out publicly in village
meetings.
The study further concludes that grassroots monitoring may be more effective
in certain contexts than in others. For projects where villagers inherently have
good information about potential corruption and a strong personal stake in
minimizing theft of funds, such as subsidies for food, healthcare, education, or
potentially microcredit projects, grassroots monitoring is likely to be effective.
For projects where information gathering is difficult and benefits from reducing
corruption are more diffuse, such as infrastructure projects, grassroots
monitoring alone is unlikely to be sufficient.
Learning from this study, KDP is in the process of increasing its audit sample
in project sites.
The increased probability of an audit reduced the percent of missing

funds, i.e., funds that could not be accounted for by an engineering
inspection of the completed infrastructure project-by about 8
percentage points from about 28 percent in control villages to 20
percent in villages receiving the audits. There was a significant
decrease in missing funds in the audit intervention.
(2)
(3)
Three points however should be considered in interpreting these figures.
(4)
First, the projects estimate of the change in the percent missing, the 8
percentage-point reduction, is substantially more reliable than the
estimated level of the percent missing-28 percent in control villages.
The numbers are much more useful for comparisons between villages
than absolute levels per se.
(5)
Sources:
(1)
Alatas, Vivi. 2005. Evaluation of the Kecamatan Development Program. World Bank, Jakarta.
(draft mimeo).
The Economist, Digging for Dirt. London, May 16, 2006.
Guggenheim, Scott et al. 2004. Indonesia s Kecamatan Development Program: A LargeScale Use of Community Development to Reduce Poverty. World Bank, Jakarta.
Olken, Ben. 2004. Monitoring Corruption: Evidence from a Field Experiment in Indonesia.
NBER Working Paper. Washington, D.C. November 2004.
Woodhouse, Andrea. 2002. Village Corruption in Indonesia: Fighting Corruption in the World
Bank s Kecamatan Development Program. World Bank, Jakarta. June 2002.
265
Chapter
Summary of
Recommendations
266
Chapter 8: Summary of Recommendations
Priorities for Poverty Reduction: An Agenda for Action
To conclude, Indonesia s persisting but specific poverty problems, coupled with the government s
priorities and its fiscal resources to address them, now positions the country to make significant headway
in reducing poverty. The question is: where to start in addressing such a broad, multidisciplinary and interwoven
issue as poverty reduction on a national scale? Action is urgently required in several areas in order to address the four
poverty-reducing imperatives in Indonesia of (i) reducing income poverty through growth, (ii) strengthening human
capabilities, and (iii) reducing vulnerabilities and risks among poorer households, while also (iv) strengthening the
institutional framework to do so and to make public action more pro-poor. In the view of the authors of this report, and
with these four areas in mind, the following 16 actions should be viewed as priorities to be addressed at the soonest
opportunity going forward.
1. Remove the ban on rice imports. Lowering the rice price and creating greater price stability by removing the
import ban on rice is the fastest way for the government to reduce poverty quickly. The ban should be replaced by a
low specific tariff. Allowing general importation subject to a tariff will be a far more effective mechanism for stabilizing
rice prices and preventing the large increases in the rice price that are so damaging to the poor. In addition, targeted
provision of rural infrastructure, and agricultural research and extension services will help marginal farmers improve
productivity or diversify into other higher-margin crops.
2. Invest in education with a focus on improving access and affordability of secondary schools and
vocational training among the poor, while improving the quality and efficiency of primary schools.
To improve attainment among the poor at the junior secondary school level will require supply- and demand-side
interventions. On the supply side, more junior secondary classrooms and schools need to be made available. This
can be done by converting primary schools where there is excess supply. On the demand side, junior secondary
schools and vocational training schools (SMKs) can be made more affordable for the poor by targeting transfers to
poor students through scholarships or conditional cash transfers (CCTs). To improve the quality of primary education,
a priority action will be to undertake a program to improve teacher management so that fewer but better-quality
teachers are present in schools and are also deployed in greater numbers to remote areas.
3. Invest in health with a focus of improving the quality of primary healthcare public and private and
access to higher level healthcare. The issue of public service delivery quality continues to require concerted
efforts to improve provider accountability and civil service staffing. The poor are using the private sector as an
alternative for primary healthcare, and efforts should be made to improve the quality of the private sector through
regulatory and training programs. On the demand side, CCTs can help specifically in interventions where behavioral
change is required to enhance demand, such as child well-being and nutrition status check-ups, and child
immunization. For higher-level healthcare, affordability is an issue and targeted programs would make sense, such
as a health insurance program. It will be important to build on and improve the recently launched program based on
recent assessments, which points to the need to improve targeting and open up service provision.
4. A focused effort is required to address Indonesia s shockingly high maternal mortality rate. The key
to reducing maternal mortality is to increase the proportion of births attended by skilled professionals. Traditional
birth attendants remain the principal choice of childbirth assistance provider for the poor in Indonesia as a result of
both demand and supply constraints. First, in order to increase demand for skilled professionals there is a need for
267
a national campaign informing communities of the benefits of professionally assisted birth delivery. To improve
affordability of skilled professionals, the government needs to increase demand-side financing through voucher or
health-card schemes. The health insurance system could also allow for a transport stipend for delivery at health
clinics for prenatal visits and delivery. Second, on the supply side, formal and in-service training can be expanded
for village midwives who are often the first line of defense, especially in more remote areas. To increase the availability
of skilled midwives in remote areas a goal that will not be easy to achieve one approach may be to offer formal
courses to train new skilled midwives.
5. Improve the quality of water accessed by the poor using separate strategies for rural and urban
areas. For rural areas, the existing community management supply model that has been shown to work should be
expanded. This currently covers 25-30 percent of the rural population, but could be expanded to cover the 50 million
people currently without adequate water supply. For urban areas, supply must be strengthened by improving capacity
and incentives for water utilities (PDAMs) to plan, deliver and monitor service delivery. In addition, PDAMs need to
be mandated and given incentives to scale up services to marginal areas inhabited by the poor. Consideration needs
to be given to designing appropriate tariff structures for the poor who benefit from current or will have future connections.
Consideration also needs to be given to enabling the poor to benefit if they do not have connections, such as
geographic targeting or lower service level subsidies (e.g. promoting access to standpipes), which would also
encourage incremental service improvements at lower cost to the poor.
6. Address the sanitation crisis facing Indonesia and its poor. Improving sanitation requires a twopronged approach: stimulating demand while improving service delivery on the supply side. On the demand side,
there is a lack of recognition of the widespread benefits of good sanitation. The government should undertake a
simple national public awareness campaign aimed at improving sanitation practices by all and stimulating demand
and pressure for change a relatively low cost measure with potentially high returns. On the supply side, service
delivery must be improved. A critical starting point is to finance increased but sustainable investment in sanitation.
Two immediate options are: development of a national strategy to increase sanitation financing among all players;
and, local government investment in sanitation infrastructure at the neighborhood and city-wide levels, for example
through a specific DAK for sanitation, or including sanitation services in minimum service standards.
7. Launch a large-scale program to invest in rural and village-level roads. For district level roads, there is
a need to increase financing, particularly for maintenance, through a concerted strategy. One option is a special DAK.
These funds could be targeted (using poverty maps) to areas where access for the poor is worst. The DAK should
leverage and increase district-level funding for road maintenance. Another possibility is the development of a road
fund at the district or provincial level along with the establishment of a district-level road management system.
Community-driven programs have been shown to assist with labor-intensive approaches to the construction and
maintenance of village and kecamatan roads. Scaling up such approaches will be useful to improving the access of
the rural poor to markets.
8. Scale up to national level Indonesia s successful community-driven development (CDD) approach.
Community-driven development projects (such as KDP) have had a history of success in Indonesia. The CDD
approach has been shown to have high rates of return on investment; it addresses in an integrated fashion binding
constraints to poverty reduction at the village level (whether it be village roads, water and sanitation, or other constraints
to poverty and well-being); and it targets poorer areas and has had a demonstrable impact on poverty reduction.
While doing so, the CDD approach also empowers the poor to have a say in how community resources are spent.
Importantly, the approach is easily replicable. Indonesia should quickly scale up its CDD approach to cover the
entire nation. It is estimated that such a national program could be up and running within three years.
268
9. Complete development of a comprehensive social protection system that addresses the risks and
vulnerabilities faced by the poor and near-poor. A new social protection program for Indonesia can be made
to address the four main risks and vulnerabilities faced by the poor in the country. Four programs could be considered
to constitute a social protection system. First, a conditional cash transfer program could be used to target cash to
poorer households while making this cash conditional on households or communities achieving certain human
development conditions. These conditions can be those that are related to high priority outcomes for Indonesia and
that mitigate against the tendency of households to cope with shocks by reducing spending on health and education,
especially of children. Second, a workfare program could be considered to mitigate against unforeseen shocks to
household income. For this to be successful it would need to self-target at a low relevant safety net wage below the
current minimum wage. Third, a health insurance program targeted to the poor and near-poor can help these households
deal with health shocks one of the most prevalent household level shocks that throw families into poverty. It would
make sense to build this program around improvements of the newly launched health for the poor scheme. Fourth,
policies to ensure low prices of staple commodities, such as rice, will be key to a comprehensive social protection
policy. In developing such a social protection scheme, several complimentary actions will be key. There is scope for
improving the current household targeting database. This should be done as a matter of priority, using a combination
of geographic targeting and revised household proxy means-testing. While developing more effective and bettertargeted social protection schemes will be key, so too will be the phasing-out of a number of programs that have been
shown to be ineffective, either from a targeting, costeffectiveness, or impact stand-point.
10. Revitalize agriculture through investment in
infrastructure and rebuilding research and extension
services. With almost two-thirds of poor household heads
still working in agriculture, boosting agricultural capability
remains essential for broad-based poverty reduction. The
government can contribute to this through: boosting
investment in key infrastructure, notably farm-to-market roads
and irrigation, while widening local water management;
encouraging and supporting diversification into higher valueadded crops; working with the private sector to ensure that
exports meet world standards; boosting expenditure on
agricultural research; and redesigning the decentralized
extension service to allow for greater involvement of the private
sector and civil society. These efforts to improve agricultural
productivity should also include development of better
marketing and information systems for rural-based
businesses.
11. Accelerate land titling and reallocate degraded and
deforested land to productive uses. Only about 25
percent of Indonesia s estimated 80 million land parcels have
been registered in the 40 years since land registration began.
At the current pace of registration, it will be difficult for land
registration to catch up with the growing number of parcels.
The government needs to redouble its efforts to speed up
land titling and more broadly ensure secure appropriate forms
269
of tenure throughout the country. In addition, reallocating degraded, deforested

land to productive uses by smallholders and the poor is one way of rationalizing
land use and reducing poverty. Moreover, reclassifying Production Forest and
Conversion Forest areas that are already allocated for productive economic uses
will allow for more productive land-use and ownership patterns and could also lead
to increased tree cover and protection of the land.
12. Make labor regulations more flexible. Stimulating faster growth of formal
sector employment is key to reducing poverty because of the impact that this has
on wages in the informal labor market. But there is ample evidence that the jobless
recovery since the financial crisis is largely a product of a poor investment climate
for formal-sector firms. The National Manpower Law, Law No. 13/2003 is a significant
contributing factor to this. Existing labor regulations are inherently anti-poor , largely
because they discourage the employment of younger and unskilled workers, forcing
them into the informal, unprotected sector. The government is right to have
commissioned a thorough review of Indonesia s labor laws. It should strive to build
a new social contract with respect to minimum wages, severance pay and methods
for settling industrial labor disputes. Such a social contract should be developed
with a view to substantially reducing the costs associated with employment,
particularly of younger and female workers, while protecting and enforcing basic
labor rights and conditions appropriate to Indonesias stage of development.
13. Extend the reach of financial services to the poor and boost access to
commercial credit for micro and small businesses. Surveys suggest that
potential exists to almost double microfinance lending if ways can be found to
address the constraints faced by households and firms that would qualify for loans
and wish to borrow, but do not currently do so. One approach to doing this would
be to develop a P4K-style outreach program for financing the household enterprises
of rural low-income households. In addition, the government could encourage the
development of free online savings products for poor households in underserved
areas as a way of supporting capital accumulation and commerce. To stimulate
local-level growth, the government should help small businesses gain access to
commercial credit by improving the debtor information system. Developing this
system to include repayment records both for loans, and ultimately other forms of
regular payments too, would help small businesses access credit by using their
good repayment history as an asset.
14. Improve the poverty focus of national planning and budgeting for service
delivery. Several steps can be taken at the national level. First, with regard to
planning and budgeting, a concerted effort is needed to ensure that sector workprograms line up with national planning priorities. Cabinet and central ministries
should sign off on the final budget after parliamentary deliberations to ensure that it
is in line with government priorities. Second, it is imperative for poverty reduction
and service delivery that there is better clarity in functional responsibilities for the
provision of specific services. This lack of clarity is paralyzing accountability in
service delivery and must be addressed without delay. Third, a civil service reform
process needs to be started that allows for the aligning of staff and incentives to
270
match with functional responsibilities, and increases the quality of staff by moving
to a merit-based system. Fourth, the intergovernmental fiscal framework can be
better used to attain national poverty reduction objectives. Most importantly,
government should design and pilot several performance-based DAK instruments,
targeted to achieve key outcomes and possibly targeted to poorer areas. This
should leverage increased local government funds for national priorities where
there are perceived externalities. Key areas on which to start may be sanitation
and rural road maintenance.
15. Undertake a major capacity-building initiative to strengthen local
government capacity to plan, budget, and implement programs for
poverty reduction. After decentralization, about one-third of total public
expenditures are allocated and implemented at the district level. Yet local
governments have weak capacity to plan, budget, and execute spending of these
funds, particularly in a pro-poor fashion. In recent years, this is reflected in
increased surpluses run by local governments. For more effective and pro-poor
spending to take place, there needs to be a substantial improvement in the capacity
of local governments, and the civil servants who work for them, to plan, budget
and implement programs for service delivery and poverty reduction. It may be
useful to complement this with actions aimed at changing incentives at the district
level, for example by publicizing service performance and budget availability.
Performance-based transfers from the central government could also induce
improved service delivery. This is no quick-fix action, but it is a necessary condition
for decentralization to work for the poor. The central government will need to work
in partnership with local governments to develop a systematic, long-term strategy
and program aimed at addressing these capacity issues. There is substantial scope
to leverage support from international partners to address this broad, challenging
and long-term agenda.
16. Strengthen poverty monitoring and assessments of poverty programs.
Poverty monitoring needs to be strengthened on several fronts. The capacity and
budget of the Central Bureau of Statistics (BPS) needs to be enhanced following
cuts during the crisis in order to improve coverage and quality of data. Poverty
monitoring can be enhanced through more systematic use of qualitative measures
and through improved monitoring over the course of the year by tracking market
prices. A strategy is needed to reinstate regular, quality monitoring of administrative
data (e.g. in health, education) in the context of decentralization by allocating
responsibility and budgets for this task. Likewise, consideration is needed of what
minimum poverty data requirements are necessary for decision-makers at the district
level and how these data are to be collected. For decentralized decision-making,
more use can and should be made of small-area poverty maps. The government
needs to assess and evaluate poverty programs more systematically with a view to
either improve them, scale them up, or eliminate them. Process and impact evaluations
are both important. Assessments underway of the four large PKPS-BBM fuel subsidy
reallocation-funded poverty programs are an important precedent in this regard.
These assessments should be used to make program decisions and similar good
practice needs to be applied more systematically to other programs.
271
Policy
Area
Growthh /macro
Gro
/m ro stabi
stability
lity
Agricult
Agr
cultural prod
oductivit
ivity
Rural roads
Rur
ads
272
Po
Poverty
verty Reduction
ction Policy
cy Matrix
ix
Introduce a district-level Road Management System to govern planning and

budgeting, and ensure that budgeting is adequate and well-targeted.
Scale up community-based approaches for the construction and maintenance of

village and kecamatan roads paying special attention to technical assistance and
facilitation.
Lack of maintenance is the primary problem characterizing the sector

with recent spending being biased towards national networks and
new road construction.
Launch a large-scale program to invest in rural- and village-level roads.

Improve the quality of district and sub-district roads and create new resources
for road maintenance in order to connect the rural poor to urban markets.
*****
***
*****
ST
ST
ST
MT
***
Decentralization has created disincentives for the allocation of

resources towards road investments/maintenance because benefits
are captured beyond the district responsible for spending.
Access to good quality roads is one of the strongest correlates of

household welfare in Indonesia. After the crisis, the poor's access to
roads has deteriorated as a result of lower investment in road
infrastructure.
MT
***
Support expansion of export commodities through improved quality of beverage

crops and global campaigns.
Promote diversification into higher-value crops with improved technical
assistance to small farmers.
MT
****
Accelerate land titling and reallocate degraded and reforested land to productive
uses.
MT
****
Revitalize agriculture through investments in infrastructure and the rebuilding of

research and extension services.
Improve productivity for rice farmers and the livestock economy through better
irrigation and management of inputs. Promote partnerships between traders,
processors and producers in a system of effective self-regulation.
Higher agricultural productivity is a key pathway out of poverty (with

two-thirds of poor household heads still working in agriculture).
However, in recent years, productivity growth in the agricultural
sector has been too low.
ST
MT
*****
Remove the import ban on rice to reduce price fluctuations when domestic
supply is low.
Trade restrictions, especially the rice import ban, hurt the poor.
Due to the current import ban, rice prices increase whenever the
domestic supply of rice is insufficient to meet national demand.
ST
Action
tion
time
me
MT
****
*****
Ensure a stable and competitive exchange rate.
Macro price stability and low inflation matter to the poor.

Export competitiveness is important for maintaining growth.
****
Priorit
Priority
Implement the required package of reforms to improve the investment climate

and increase employment-generating growth.
A ctions
Specific Actions
Growth that benefits the poor is the single most effective means of
achieving poverty reduction.
ey IIssue
uess and Con
onstrai
traints
Key
ts
Making Growt
Mak
Growthh Work
Wor for
or the
t Poor: Connecting the Poor to
t Opport
Opportunities
II
MT
MT/LT
MT
MT
MT
MT
ST
ST
ST
MT
Impact
pact
Im
time
me
MT
Labor
Lab
or markets
m
Extend the reach of financial services to the poor and boost access to
commercial credit for micro and small enterprises.
Support for one or more outreach initiatives aimed at connecting new microborrowers to the financial system.
Possibly subsidize some of the cost of group formation, supervision and
institutionalization.
Promote all banks to have basic no-fee savings products available.
Microfinance:
Enterprise credit: Many enterprises are credit-constrained. SMEs

often mention credit as one of their top two business constraints.
poor. The anti-poverty benefits of savings are largely for the next
generation,
Financial services: Secure savings mechanisms are valuable to the
LT: Long-term
M T: M edium -term
Encourage more transparency among commercial banks (and BPRs) in order to

better determine how much lending has actually gone towards micro/small
enterprises.
Review and revise of Indonesia s labor laws/regulations at both the national and
local level.
Introduce a new social contract with respect to minimum wages, severance pay
and methods for settling industrial labor disputes.
Severance pay regulations may also restrict mobility in the formal

sector labor market.
S T: S hort-term
Financial services
Fin
there is still great potential value in connecting

households to the financial system for the first time. Few Indonesian
institutions wish to actively target the enterprising poor.
Implement further study on the impact of minimum wages on the formal and
informal employment of the poor.
High minimum wages may reduce the ability of the poor to find jobs
in both formal and informal sectors.
Real minimum wages have risen sharply in recent years.
**
ST
ST
ST
ST
**
**
ST
ST
****
**
ST
ST
****
***
ST
MT
MT
MT
MT
MT
MT
ST
273
Educa
ucation
Health
274
Policy
Area
Continue to convert some primary school buildings with low student

numbers into junior secondary schools, while selectively building new
secondary schools.
Invest in health with a focus of improving quality of primary health
facilities public and private and access to higher level healthcare.
Improve quality of care provided in basic community healthcare clinics

(Puskesmas).
Focus on systemic problems in health personnel management and review
incentive structures within the sector.
Invest in the training of private (as well as public) paramedics and
midwives particularly those that serve in remote areas
Supply-side constraints exist especially above the junior secondary

school level and at the primary school level in remote areas.
Improving affordable access to quality healthcare is among the priorities

of a poverty reduction strategy. In recent years, there have been
incremental increases in health spending. However, most of the funding
has gone to secondary care at hospitals. Most health spending continues
to come from private sources in Indonesia.
The quality of basic healthcare remains low with very high absenteeism
rates among public health professionals. The demand for public health
services is low as a result.
The poor do not benefit from the subsidies that go to secondary and
tertiary healthcare at hospitals, since their access to these facilities is
limited. At the same time, health shocks are a particularly important
source of vulnerability among poorer households in Indonesia.
Facilitate the linking up of these providers to the health insurance scheme

for the poor so that the poor can continue to use the providers of their
choice in a more affordable way.
Invest in demand-side activities that increase the access of the poor to inpatient care so that the poor can benefit from spending on hospitals.
Make sure PT Askes has the right incentive structure to socialize the new
heath insurance scheme and to target the poor.
Focus on teacher quality and investing in textbooks and in-class teaching

materials for improving teaching quality.
Teaching quality is low and any improvement in the quality of schooling,

especially at the primary school level, is pro-poor given the current
distribution of average benefits
The poor heavily utilize private providers, especially in rural areas where
they have little access to other services.
Implement targeted scholarship or cash transfer programs to improve

school enrollments.
Reduce schools' requirements on uniforms and extra costs.
Improve access of the poor to secondary schools, vocational training

schools and private training centers through targeted programs.
The return to education is higher for higher levels of schooling, so

focusing educational investment only on primary education is no longer
enough.
Affordability: in Indonesia, school fees and other expenses create barriers

to the poor s access to education, particularly at the junior secondary
school level.
Invest in education with a focus on improving access and affordability of

secondary schools and vocational training among the poor, while
improving quality and efficiency of primary schools.
Specific Actions
Educational investment needs to focus on the skills and employability of

the growing young population in Indonesia. For those working in nonfarm activities in both rural and urban areas, the priority is boosting their
capability to obtain better jobs.
Key issues and Constraints
Making Spending Workk for

Mak
for the
t Poor: Invest
esting in the
t Assets
Asset andd Capabil
Capabilitiess of the Poor
ST
ST
ST
ST
*****
****
****
ST
MT
****
***
MT
ST
****
**
ST
****
****
ST
MT
Action
time
****
****
Priority
MT
MT
ST/MT
MT
MT
MT
MT
MT
ST/MT
ST/MT
MT/LT
Impact
time
Specific he
Spe
health issue
issues
Water
ater
Improve quality of water accessed by the poor by using separate strategies

for rural and urban areas:
In rural areas, replicate the community-management supply model to

reach those people without adequate water supply.
Devote more attention to the development of sustainable management
systems, alternative technologies, cost recovery and economies of scale.
In urban areas, strengthen capacity and establish incentive structures for
water utilities (PDAMs) for planning, delivery and monitoring of service
delivery.
Give mandate and incentives to PDAMs for scaling-up services to
marginal areas inhabited by the urban poor.
Most water is not delivered, especially in rural areas, and household and
community provision dominate the sector.
The urban water sector is in a state of crisis in terms of delivery to the

poor. More than 80 percent of the urban poorest quintile does not have
access to piped water. Subsidy mechanisms do not target the poor,
because they are not usually connected to utility networks.
ST
ST
****
ST
ST
ST
ST
ST
ST
MT
ST
ST
****
*****
**
Subsidize fortified foods with multi-micronutrients and iodized salt.
Access levels to public water services are low and self-provision

characterizes the sector.
Government expenditures on water and sanitation services are low.
***
Invest in information campaigns about benefits of exclusive breastfeeding

for child nutrition.
Despite reductions in poverty in recent years, the malnutrition rate has

remained stagnant.
***
Carry out further research into the causes of high malnutrition in

Indonesia.
alnu
nutr
tritio
ition: One-fourth of all children under the five are malnourished in
Mal
Indonesia.
***
***
Implement a wide-scale awareness campaign on the importance of skilled

midwifery.
Increase the proportion of institutional deliveries, as well as access to 24hour emergency obstetric care by improving the availability of facilities,
affordability of these services and the condition of infrastructure (such as
roads) around these facilities.
The poor still rely heavily on the services of traditional midwives (dukun)
for child deliveries due to lack of awareness, as well as difficulties in
access and affordability of skilled midwives.
***
***
Increase the availability and quality of skilled birth attendants in remote

areas.
Maternal mor
mortality: Indonesia has very high maternal mortality rate at 307
per 100,000 live births.
****
Improve affordability of care by skilled professionals by linking them into

the health insurance schemes (such as through PT Askes ) implemented
by government.
Implement a widespread anti-smoking campaign specifically targeting the

poor.
Tobacco constitutes a major expenditure item for the poor and damages
their health and finances.
Despite this high adverse impact, public perceptions of tobacco and the
industry remain largely positive.
MT
MT
MT
ST
MT
ST
MT
MT
MT
MT
MT
275
276
Address the sanitation crisis facing Indonesia and its poor.
Stimulate demand for sanitation through an awareness-building campaign

and leverage the private sector s expenditure on hygiene through stronger
political leadership.
Finance increased but sustainable investment in sanitation by developing a
national strategy to increase sanitation financing among all players, for
example through a DAK for sanitation.
The coverage of sanitation services in terms of piped sewage disposal is

very low and there is a major failure in sanitation adequacy.
Less than 1 percent of Indonesians have access to piped sewage disposal

and 60 percent of the urban poor do not have access to a septic tank.
Low sanitation coverage has significant health and environmental

impacts, especially for the poor, but demand for sanitation remains weak.
*****
*****
*****
ST
ST
ST
MT
MT
MT
Sanitat
anitation
Targeting
Targ
Rice poli
olicy
Addressing the risks

isks and
nd
vulnerabi
lnerabilities of the poor
Policy
Area
Indonesia recently put together a unified targeting registry for the UCT
program, which recent evaluations show can be improved.
Cultural norms, which foster sharing program benefits (bagi rata) at the
community level, have also contributed to weak targeting.
Rice is the most important commodity for the poor: the poor spend almost a
quarter of their budget just on rice.
Over three-quarters of the poor are net consumers of rice.
The price of rice is one of the most important determinants of poverty at the
household level in Indonesia.
Targeting is a particular challenge in Indonesia, with significant clustering
around the poverty line, large numbers of poor moving in and out of poverty
and low inequality. This makes it difficult to identify the poor, especially in
rural areas.
2. Factors that pose risks during infancy, early childhood and among schoolaged children are significantly more frequently observed among the poor. In
the absence of coping mechanisms, poorer households may be forced into
bad coping.
3. Health shocks are a particularly important source of vulnerability among
poorer households in Indonesia.
Involve communities in targeting the poor in order to tap into local

information and improve targeting effectiveness.
Socialize the target group well in order to reduce the sharing of program
benefits.
Verify, update and refine the existing poverty database. Keep this
database as an organic registry of the poor.
Further develop the current targeting system.

Build on both geographic targeting of areas, as well as systems for
household targeting.
Vary the type of targeting used (community vs household level) by
program type and levels of poverty/inequality in a particular setting.
ST/MT
MT
***
***
MT
ST
***
***
ST
ST
ST
***
***
*****
Replace the rice import ban with an appropriate level of tariff that allows
for the free trade of rice.
ST
ST
***
*****
ST
MT
*
***
MT
ST
ST
Action
time
Remove the ban on rice imports.
Bring together workfare programs and build a national but modest in size
system of workfare with wages set below the minimum-wage, in order to
help the poor cope with transitory shocks to income caused by the loss of
a job.
Ensure that in the setting-up of such a program wage levels are not
manipulated by political pressure.
Develop targeted voucher or cash transfer schemes (such as CCTs) that
enhance the ability of the poor to cope with shocks and allow them to
smooth consumption without reducing investment in the human capital of
future generations.
Build on and improving the targeting of the existing health insurance
schemes for the poor to increase the access of the poor to in-patient
hospital care, as well as subsidize the poor s use of selected private
providers.
*****
Move away from price subsidies on fuel towards smaller and more
targeted social protection schemes based on the needs of the poor.
The existing social protection system is mostly dominated by regressive

price subsidies for fuel and does not respond to the risks and vulnerabilities
faced by the poor.
1. Risks to earnings capacity and income are significantly greater among

poorer households.
***
Priority
Complete development of a comprehensive social protection system that

addresses the risks and vulnerabilities faced by the poor and near-poor.
Specific Actions
Although poverty in Indonesia is relatively low, vulnerability to poverty

remains high. There is a high level of churning around the poverty line,
causing many of the near-poor to move in and out of poverty.
Key
ey Issues and Constraints
Making
Mak
ing Social Prot
Protection Work for the Poor: Reducing
g Vulnerabilities
Vulner
of the Poor
MT
MT
MT
ST
ST
ST
ST
ST
MT
MT
MT
MT
MT/LT
MT/LT
Impact
time
277
Policy, planning and budgeting systems
Institutional accountability
278
Policy
Area
Develop local-level capacity for pro-poor planning and budgeting.

Revise the DAU transfer to make it more pro-poor by including a stronger
poverty component within the fiscal gap formula.
Use the recently doubled DAK to prioritize poverty reduction, and leverage local
government funding through matched grants. In this way, also target poor areas
or areas with low outcome indicators in the specific program area.
Clarify functions between central and local governments and within government
units, whereby the responsibilities are distributed in the following way:
Central government: Has core functions of policy-making, staffing,
information and setting standards.
Provincial government: Fixes regional targets and standards, builds
capacity, and implements cross-district services.
District government: Develops local plans and budgets, implements
services.
Focus on civil service reform by addressing hiring practices, capacity building,
sanctions and rewards.
Strengthen citizen and civil society voice at both the national and local levels
The transfers to districts are not necessarily distributed to target poorer

areas.
The central government currently does not leverage local government

resources and provide incentives for spending at the local level
towards national priorities.
Lacking clarity of functions between central and sub-national

governments hinders the ability of all levels of government and service
providers to serve the poor.
Significant challenges exist in reforming the civil service. Staff have

little incentive to perform effectively and the rigidity of the staffing
system hinders the flexible hiring, transfer and promotion of good
people.
Citizen and civil voice are critical in improving institutional

accountability but a number of blockages hinder the new participatory
process being adopted by government.
Provide a mandate for Bappenas and the MoF to ensure established poverty
reduction priorities are incorporated into sector plans and budgets.
Improve the poverty focus of national planning and budgeting for service
delivery.
Ensure that the cabinet s stated objectives are reflected in budgets. Introduce a
results-orientation focused upon poverty reduction objectives for all ministries.
Specific Actions
More than one-third of all government spending is now at the local

government level as of 2006.
Poverty reduction priorities are currently not linked with sectoral plans
and budgets, and there is low capacity and incentives for pro-poor
planning and budgeting.
ey Issues and Constraints

Key
Making Government Work for the Poor: Improving Service Delivery
***
***
****
****
**
****
MT/LT
MT/LT
ST
ST
ST
MT
ST
ST
****
****
ST
Action
time
****
Priority
MT/LT
MT/LT
MT/LT
MT
MT
MT/LT
ST
ST
ST
Impact
mpact
time
In order to monitor progress towards poverty reduction goals, data

collection and use needs to be improved.
Link monitoring findings more effectively with feedback to, and design of,
sectoral programs, and for use by local governments. Develop approaches to
better feed information collected into program design.
Make better use of existing and emerging data sources to both better
understand poverty and program impact.
Strengthen systems to both understand and analyze poverty in the decentralized
context.
Use poverty maps in program targeting.
Strengthen poverty monitoring and assessments of poverty programs.
ST
***
ST
ST
***
***
ST
ST
***
***
MT
MT
MT
MT
MT
279
Assessment and monitoring of poverty

reduction
Annexes
280
Annexes
Annex II. 1 Poverty lines, percentage of poor and total numbers of poor in Indonesia
By urban-rural areas, 1976-2005
Year
Poverty line
% of poor people
Number of poor people
(Rp monthly per capita expenditure)
(headcount index)
(million)
Urban
Rural
Urban
Rural
Urban + Rural
Urban
Rural
Urban + Rural
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
1976
4,522
2,849
38.79
40.37
40.08
10.0
44.2
54.2
1978
4,969
2,981
30.84
33.38
33.31
8.3
38.9
47.2
1980
6,831
4,449
29.04
28.42
28.56
9.5
32.8
42.3
1981
9,777
5,877
28.06
26.49
26.85
9.3
31.3
40.6
1984
13,731
7,746
23.14
21.18
21.64
9.3
25.7
35.0
1987
17,381
10,294
20.14
16.14
17.42
9.7
20.3
30.0
1990
20,614
13,925
16.75
14.33
15.08
9.4
17.8
27.2
1993
27,905
18,244
13.45
13.79
13.67
8.7
17.2
25.9
1996
38,426
27,413
9.71
12.30
11.34
7.2
15.3
22.5
1996*
42,032
31,366
13.62
19.77
17.55
9.6
24.6
34.2
1999
84,773
67,382
19.41
26.03
23.43
15.6
32.3
48.0
2002
130,449
96,512
14.46
21.10
18.20
13.3
25.1
38.4
2003
138,803
105,888
13.57
20.23
17.42
12.3
25.1
37.3
2004
143,455
108,725
12.13
20.11
16.70
11.4
24.7
36.1
2005
15.97
35.1
Source: Central Bureau of Statistics (BPS).

Note: * Using Standard 1998.
281
Annex II. 2 Framework for pro-poor growth
Political Will and Commitment
Empowerment
Education
& Health
Capabilities
Agricultural
Technology
Rural
Non-tradables
Infrastructure
Regulations
Transaction
Costs
Reducing
Corruption
Technology
Costs of Food
Staples
Income
Distribution
Demand
Exports,
Exchange
Rate and
Trade
Macro Economy and

Rapid Growth
Rapid Propoor Growth

282
Annexes
Annex II. 3 Empirics of structural transformation in Indonesia

1982
Output (Rp trillion)
Share of GDP (%)
Number of workers (million workers)
Labor productivity per worker (Rp million)
Share of workers in total (%)
1992
Share of GDP (%)
1996
Share of GDP (%)
1999
Share of GDP (%)
2002
Share of GDP (%)
Number of workers (million workers) [6]
2002 [2]
Share of GDP (%)
Number of workers (million workers) [6]
2004 [2]
Share of GDP (%)
Industry
58.2
36.0
8.7
6.7
15.0
Services
69.5
43.0
17.3
4.0
30.0
Agriculture
32.3
20.0
31.8
1.0
55.0
TOTAL
162
99.0
58
2.8
100.0
119.2
40.0
12.0
10.0
15.0
125.2
42.0
26.3
4.8
32.9
53.7
18.0
41.7
1.3
52.2
298
100.0
80
3.7
100.0
171.2
42.0
14.1
12.2
17.1
171.8
42.2
30.4
5.6
37.0
64.4
15.8
37.7
1.7
45.9
407.3
100.0
82.1
5.0
100.0
151.3
39.9
14.7
10.3
16.4
162.6
42.9
34.5
4.7
38.6
65.4
17.2
40.1
1.6
44.9
379.2
100.0
89.3
4.2
100.0
179.1
40.8
15.7
11.4
17.9
188.2
42.9
32.4
5.8
37.1
71.3
16.3
39.2
1.8
44.9
438.6
100.0
87.3
5.0
100.0
635.7
43.3
15.7
40.6
17.9
594.5
40.5
32.4
18.3
37.1
239.3
16.3
39.2
6.1
44.9
1,469.5
100.0
87.3
16.8
100.0
677.4
42.1
16.2
41.8
18.0
674.7
42.0
33.7
20.0
37.4
255.2
15.9
40.1
6.4
44.6
1,607
100.0
90.0
17.9
100.0
% growth of output
1982-1992
1992-1996
1996-1999
1999-2002
2002-2004
7.2
9.0
-4.1
5.6
3.2
5.9
7.9
-1.8
4.9
6.3
5.1
4.5
0.5
2.9
3.2
6.1
7.8
-2.4
4.8
4.5
% growth of employment
1982-1992
1992-1996
1996-1999
1999-2002
2002-2004
3.2
4.0
1.4
2.2
1.7
4.2
3.6
4.2
-2.1
1.9
2.7
-2.6
2.1
-0.7
1.1
3.2
0.7
2.8
-0.7
1.5
% growth of labor productivity
1982-1992
1992-1996
1996-1999
1999-2002
2002-2004
4.0
5.0
-5.6
3.4
1.5
1.7
4.3
-6.0
6.9
4.4
2.3
7.1
-1.6
3.6
2.1
2.9
7.1
-5.2
5.6
2.9
Source: National Account data for GDP and Susenas data for employment.
Notes:
1. Real GDP based on 1993: 1982, 1992, 1996, 1999 and 2002.
2. Real GDP based on 2000: 2000 and 2004.
3. Industry consists of mining and quarrying, manufacturing, electricity, gas, water and construction.
4. Services include trade, restaurants, hotels, transportation, storage, communication, finance, insurance, real estate, business and public services.
5. Employment is defined as self-employment without help, self-employment with help of householders, self-employed with help of regular workers, employees and family workers with age 10
years and older (in order to be consistent with 1982 data).
6. Excludes Aceh, Maluku and Papua.
7. Labor productivity is defined as GDP per worker
283
Annex II. 4 Mean share of per capita income from each source (national, rural, urban)
National
Urban
1993
2002
1993
2002
1993
2002
Wages
36.05
36.29
28.97
27.55
50.12
46.74
Agriculture
25.00
22.11
36.29
36.03
4.11
5.46
Non agriculture
19.93
21.11
16.79
16.86
26.16
26.19
Ownership
10.20
7.81
10.48
7.66
9.65
8.00
Transfer
2.13
5.12
1.26
4.2
3.85
6.22
Receipts
3.24
3.88
3.15
3.64
3.42
4.18
Other
2.94
3.57
3.06
3.94
2.69
3.12
Source: Susenas.
284
Rural
Annexes
Annex II. 5 Increase in non-agricultural sector employment was higher than predicted by average
employment growth (1982-2002)
Agriculture
Actual
Predicted
Actual
Difference
1982
1993
1993
( 000)
( 000)
( 000)
( 000)
30,487
42,557
39,137
-3,420
Non-agriculture
25,724
35,909
39,329
3,420
Rural
15,939
22,250
18,992
-3,258
Urban
9,785
13,659
20,337
6,678
All sectors
56,211
78,466
78,466
Actual
Predicted
Actual
Difference
1993
2002
2002
( 000)
( 000)
( 000)
( 000)
Agriculture
39,137
43,348
39,035
-4,313
Non-agriculture
39,329
43,561
47,874
4,313
(Based on 3.03 percent annual average employment growth)
Rural
18,992
21,035
16,785
-4,250
Urban
20,337
22,526
31,088
8,562
All sectors
78,466
86,909
86,909
Actual
Predicted
Actual
Difference
1982
2002
2002
(Based on 1.14 percent annual average employment growth)
Agriculture
( 000)
( 000)
( 000)
( 000)
30,487
47,137
39,035
-8,102
Non-agriculture
25,724
39,772
47,874
8,102
Rural
15,939
24,644
16,785
-7,859
Urban
9,785
15,128
31,088
15,960
All sectors
56,211
86,909
86,909
(Based on 2.18 percent annual average employment growth.)

Source: McCulloch, Timmer, and Weisbrod, 2006.
Note: The predicted column provides the number of workers that would have been employed in a sector if all sectors grew at the same annual average employment growth.
285
Annex II. 6 Rural and informal economy absorbed workers during the crisis
(Workers by sector, formal/informal 1996-2004)
Formal
Agriculture
Informal
Industry
Manufacturing
Total
Agriculture
Industry
Total
Manufacturing
Total
Number of workers ( 000)

1996
5,233
11,833
19,765
36,831
31,268
3,433
12,369
47,070
83,900
1997
5,207
12,792
20,308
38,306
29,583
3,511
14,005
47,099
85,406
33,695
3,279
14,260
51,233
87,672
31,940
3,704
14,763
50,407
88,817
89,838
1998
5,720
10,999
19,720
36,439
1999
6,438
12,141
19,831
38,410
34,602
4,464
15,685
54,751
36,028
4,888
14,891
55,806
90,807
91,647
2000
6,079
11,198
17,811
35,087
2001
3,716
12,127
19,158
35,001
37,321
5,646
15,009
57,976
39,888
5,222
14,901
60,011
92,811
93,722
2002
3,313
11,548
18,810
33,671
2003
3,154
11,213
18,432
32,799
11,386
19,990
34,509
37,475
5,488
16,251
59,213
2004
3,134
Share in employment (%)

1996
6.2
14.1
23.6
43.9
37.3
4.1
14.7
56.1
100.0
1997
6.1
15.0
23.8
44.9
34.6
4.1
16.4
55.1
100.0
1998
6.5
12.5
22.5
41.6
38.4
3.7
16.3
58.4
100.0
1999
7.2
13.7
22.3
43.2
36.0
4.2
16.6
56.8
100.0
2000
6.8
12.5
19.8
39.1
38.5
5.0
17.5
60.9
100.0
2001
4.1
13.4
21.1
38.5
39.7
5.4
16.4
61.5
100.0
2002
3.6
12.6
20.5
36.7
40.7
6.2
16.4
63.3
100.0
2003
3.4
12.1
19.9
35.3
43.0
5.6
16.1
64.7
100.0
2004
3.3
12.1
21.3
36.8
40.0
5.9
17.3
63.2
100.0
Source: Sakernas.
286
(The size of the bubble indicates the number of poor in region (urban/rural) who lack the specific type of infrastructure service)
Annex III.1 Where do the poor who lack access to basic infrastructure services live?
Annexes
287
Annex III.2 Poverty growth elasticities (supporting table for Figure 3.8)
Dependent variable: Poverty headc
ount (Po)
headcount
Ln Real Mean PCE (rmexp)

Ln Gini Ratio (ginibase)
Specification
(1)
-2.18
0.110***
-0.377
0.082***
(2)
-2.102
0.110***
-0.494
0.088***
-0.038
0.020*
(3)
-2.09
0.109***
-0.646
0.103***
-0.092
0.025***
-0.075
0.032**
-0.075
0.040*
-0.049
0.03
0.108
0.038***
-0.115
0.131
-0.574
0.105***
304
52
-0.787
0.120***
304
52
-1.011
0.141***
304
52
Base Period FGT0 (p0ubase)

Sumatra
Kalimantan
Sulawesi
NT/Maluku
Papua
Ln Gini Ratio (ginibase)* Sumatra
Ln Gini Ratio (ginibase)* Kalimantan
Ln Gini Ratio (ginibase)* Sulawesi
Ln Gini Ratio (ginibase)* NT/Maluku
Ln Gini Ratio (ginibase)* Papua
Ln Real Mean PCE (rmexp)* Sumatra
Ln Real Mean PCE (rmexp)* Kalimantan
Ln Real Mean PCE (rmexp)* Sulawesi
Ln Real Mean PCE (rmexp)* NT/Maluku
Ln Real Mean PCE (rmexp)* Papua
Constant
Observations
Number of group(prov rural)
Standard errors in brackets
* significant at 10%; ** significant at 5%; *** significant at 1%
Source: Friedman, 2006.
288
(4)
-2.483
0.123
-0.645
0.195***
-0.083
0.023***
-0.073
0.335
-0.922
0.537*
-0.614
0.405
0.484
0.282*
-0.294
0.576
0.011
0.252
-0.625
0.398
-0.45
0.318
0.347
0.216
-0.23
0.477
-0.025
0.138
-0.247
0.242
0.385
0.197
1.36
0.119
1.142
0.283
-0.926
0.257***
304
52
Urban
(4)
-2.612
0.265
-2.069
0.661***
-0.071
0.061
1.344
0.881
-3.982
1.717**
-1.116
1.616
2.519
0.905***
0.831
0.864
1.282
0.752*
-2.927
1.384**
-0.691
1.343
2.163
0.792***
0.925
0.752
-0.032
0.264
-1.367
0.434
-0.548
0.696
0.565
0.428
-0.316
0.385
-2.645
0.749***
153
26
Rural
(4)
-2.519
0.144
-0.557
0.241**
-0.048
0.033
0.24
0.438
-0.244
0.591
-0.373
0.49
0.036
0.328
1.255
0.622**
0.22
0.317
-0.116
0.422
-0.28
0.368
0.012
0.246
0.934
0.516*
0.031
0.161
0.239
0.233
0.359
0.154
1.631
0.137
1.996
0.187
-0.692
0.320**
151
26
Annexes
Annex III.3 Differences in poverty correlates and constraints across regions

(Supporting Table for Box 3.7)
(Decomposition of changes in poverty incidence: the absolute approach)
Sumatra
to Java
Kalimantan
to Java
Sulawesi
to Java
NT/Maluku
to Java
Papua
to Java
Initial values
17.6538
12.3500
18.5400
29.2100
End value (2002)
17.2192
17.2192
17.2192
17.2192
23.3400
Change in poverty
-0.4346
4.8692
-1.3208
-11.9908
-31.4100
Growth effect
-4.9840
1.4220
-1.4700
-17.7180
-23.7950
Distribution effect
4.5494
3.4472
0.1492
5.7272
-7.6150
Overall price effect
54.7500
-9.0574
-3.2413
-9.8593
-5.2834
-23.6573
Gender
-4.5071
-1.2402
-2.1358
-3.6415
-18.7518
Education of household head
-2.5653
- 2.7900
-4.5846
-6.1859
7.4315
Experience
2.3506
0.8942
-2.3094
4.3264
-1.9533
Other member education
-3.0354
-0.8445
-2.6939
-0.7317
-3.7876
Asset: household status
0.1011
0.2532
2.7575
3.6566
-11.3646
Density
-0.1573
0.1516
0.1859
-1.2997
-0.6172
Access to secondary education
0.2261
0.0538
0.3456
0.5018
1.3611
Access to informal courses
-0.3796
-0.3336
0.0734
-0.0915
-0.0930
Access to credit
-0.9602
-0.6069
-0.5792
-0.4440
0.2368
Access to infrastructure
-1.1262
0.7863
-1.9201
-0.5716
3.8808
Access to communication
0.9959
0.4348
1.1480
-0.8022
0.0000
-0.4383
0.3194
0.0373
1.1552
-4.7895
Household occupation return
0.1043
-0.8473
-0.2655
1.3024
0.9739
Other member occupation return
1.2063
1.4087
1.7430
2.1657
2.2507
Endowment of occupational choice (oc)
-1.7489
-0.2421
-1.4402
-2.3128
-8.0142
Constant effect
14.1327
14.7461
11.3083
11.1295
5.0596
Non observables variance
0.5014
0.6781
0.09'-n
-0.1225
0.4355
Price + residual variance + occupational effect
5.1384
12.5023
1.5795
6.8787
-22.9518
Population eftect
-5.5730
-7.6331
-2.9003
-18 8695
-8.4582
Occupational-choice effect
Source: Susenas.
289
Annex III.4 Microeconomic simulation of changes in poverty

Poverty reduction is a complex dynamic process. Several inter-related factors can affect the process at the same time: for example, the
combined effects of changes in personal earnings, changes in occupational status, changes in the demographic composition of families,
and replacement of older cohorts by younger, better-educated and more productive cohorts. We need to identify the economic and
social causes behind these various phenomena, and the way they interact with each other.
How can we identify these factors and their related contributions to the change of poverty? We will use microeconomic simulations to
decompose changes of poverty in two different years and among different regions, into contributions of three sets of phenomena: (i)
changes in the socio-demographic structure of the population, the assets and personal or household characteristics in the population
(endowment or population effect); (ii) changes in the structure of earnings/expenditures,-the returns to those assets and characteristics
(price effect); and (iii) changes in occupational choice behavior, how people use those assets and characteristics in the labor market
(occupational-choice effect).
Why do we need to do micro-simulation decomposition in addition to the scalar decomposition of changes in poverty by population
subgroups? The scalar decomposition has several limitations. Although single inequality and poverty indices are useful summary
statistics, they do not tell what happened to the distribution. Scalar inequality measures, such as Gini coefficient or poverty indices, are
generally used to summarize distribution, but a single number does not tell the whole story. A decrease in Gini coefficient can be a story
of the poor getting better or the rich getting worse. We cannot infer whether poverty increases or decreases, and what happens to the
poor and the not-so-poor based on the Gini coefficient alone. Similarly a decline in the headcount index does not give us information
on what happens to the upper part of the distribution. This scalar decomposition also does not easily allow for controls. Therefore, we
cannot isolate impacts of key variables or distinguish contributions of changes in assets or changes in the market-returns to it. The
simulation of entire distributions addresses the shortcomings of scalar decompositions.
Using this simulation method, we want to first evaluate the effect of the change in the coefficients of the expenditure functions what we
have called price effect on the distribution of expenditure. The price effect is obtained by comparing the initial distribution and the
hypothetical distribution obtained by simulating the population observed at date t the remuneration structure observed at date t . To get
this simulated income, we replace in the 1999 sample the household expenditure by the value obtained using the function estimated for
2002, while keeping the observable and unobservable characteristics of the households constant. While the overall price effect is
obtained by modifying all the coefficients b at the same time, it is also interesting to evaluate the effect of a change in only a subset of
these coefficients, possibly a single coefficient. For instance, one may want to evaluate the effects of the change in the rate of return to
education or to experience. The analysis is only partial in the sense that we do not seek to understand how these three forces may have
interacted with each other and it also abstracts from general equilibrium impact. However, we believe this can yield insizghts into the
evolution of poverty and its relationship with the process of economic development in Indonesia.
We conduct two different kinds of simulation. The first is what we called absolute-approach simulation where we applied the 2002
coefficient to the 1999 data. The second, called the relative approach, is done by holding the mean of expenditure constant while we did
the partial simulations. The absolute approach will identify an effect as being pro-poor if it accompanied by absolute reduction of
poverty, while the relative approach considers an effect as being pro-poor if the poor benefit more from the changes than the rich, in
relative or absolute terms. The relative approach will identify the impact of the simulated changes on the distribution, whether it is an
equalizing or an inequality-increasing factor.
290
Annexes
Annex V. 1 Inpres development grants

Prior to the introduction of the Dana Alokasi Umum (General Allocation Fund, or DAU) in 2001, Indonesia s transfer system consisted
of the Subsidi Daerah Otonom (Autonomous Region Subsidy, or SDO) and Inpres. The SDO was to fully support routine expenditures
at the local level, while the Instruksi Presiden (Presidential Instruction, or Inpres) gave grants for local development activities, which
started as specific (earmarked) grants but developed into more general block grants over the period from the 1960s to the 1990s.
During this period, the Inpres grants changed considerably in their structure and function.
In an attempt to address regional disparities and national development objectives, the government launched an intergovernmental
financing scheme or development transfer known as Instruksi President (Presidential Instruction, or Inpres for short). These transfers
can be classified as specific or discretionary general purpose block grants from the national government to sub-national governments
to finance basic education, health services, reforestation or small economic activities. The Inpres grants were divided into the following
four categories: (i) Inpres Dati I are grants made to provincial level governments; (ii) Inpres Dati II are per capita grants provided for
infrastructure projects at the district/municipality level; (iii) Inpres Desa are grants for village development; and (iv) Inpres Desa
Tertinggal (IDT) are specific grants designed to assist village development in poor or left-behind areas.
The IDT began in 1994 with geographical targeting to provide small-scale credit to poor households living in more than 200 villages in
such left-behind areas across Indonesia. A total of US$200 million per year over a three year period (1994-96) was disbursed (Pangestu
and Azis, 1994, World Bank 1995). Each participating village receives a block grant of Rp 20 million (US$8,700)1 to Rp 60 million.
These grants are used by village working groups (Kelompok Masyarakat, or Pokmas) for infrastructure development, to address the
lack of quality infrastructure and its impact on poverty (Sumadiningrat).
Between 1995 and 2000, intergovernmental transfers made up a large portion of total regional government revenue of about 75 percent.
Lewis (2002) finds that SDO transfers were most important of around 38 percent of total regional government revenues, 31 percent for
provinces and 43 percent for districts/municipalities (Lewis, 2002). Inpres grants followed with 23 percent of total regional government
revenues, 6 percent of provincial and 28 percent of district/municipality revenues (Lewis, 2002).
Most analysts seem to have concluded that, in general, the Inpres system had a relatively positive impact on regional social and
economic development in Indonesia over three decades (Lewis, 2002). Furthermore, a significant amount of regional infrastructure was
financed by Inpres grants. But while Inpres grants have financed substantial improvements in local infrastructure and social services
they have not provided a lasting solution to the problem of financially weak or dependent local governments (Devas, 1989). Although
Inpres grants are not specifically designed as a pro-poor policy and targeted to the poor, the increased access to basic services through
the construction of schools and community health centers, rural electricity programs and expansion of roads and communication
networks, and improvements in supply of clean water, all benefit the poor.
Using exchange rate at Rp 2,300 per US$.
291
Nominal spending
(Rp billion,)
Real spending
(Rp billion 2004 prices,
adjusted with CPI index)
1,745.5
1,739.7
40,819.4
101
0.0
0.1
0.1
2.4
580,852
1,657.3
1,547.9
35,438.9
95
1,636.6
1,846.8
34,818.6
87
16,504.3
447.0
0.0
0.1
0.1
2.3
484,615
15,271.5
353.0
14,342.5
260.4
13,518.7
8,603.9
1.0
0.0
11,433.7
6,723.4
11,055.7
7,523.4
593.5
591.5
13,878.4
1.0
0.0
529.3
494.4
11,319.4
476.9
538.1
10,145.2
5,611.4
152.0
0.8
0.5
4,877.8
112.8
4,179.0
75.9
4,596.3
2,925.3
FY96
0.8
0.4
3,652.0
2,147.5
FY95
3,221.3
2,192.1
FY94
Central Spending1
Routine
0.8
Development
0.5
Province Level
Routine
1.0
Development
0.0
District/City Level
Routine
0.0
Development
0.1
INPRES
0.1
TOTAL
2.5
GDP Assumption 2000 index 406,796
Central Spending1
Routine
Development
Province Levels
Routine
Development
District/City level
Routine
Development
INPRES
TOTAL
Central Spending
Routine
Development
Province Level
Routine
Development
District/City Level
Routine
Development
INPRES
TOTAL
CPI index 1996=100
0.0
0.1
0.1
2.2
735,078
0.8
0.0
0.8
0.4
1,976.5
1,724.3
42,278.8
113
15,953.6
338.3
15,621.9
8,388.6
754.0
657.8
16,128.7
6,086.0
129.0
5,959.5
3,200.1
FY97
0.0
0.1
0.1
1.7
1,095,860
0.6
0.0
0.5
0.5
1,319.0
940.6
30,267.7
187
10,926.0
110.2
9,485.7
8,426.7
828.9
591.1
19,020.7
6,866.1
69.2
5,961.0
5,295.5
FY98
0.0
0.1
0.0
2.2
1,218,534
0.8
0.0
0.7
0.5
39,664.7
202
1,286.1
15,160.4
446.5
13,350.3
9,421.4
27,036.1
876.6
10,333.6
304.3
9,099.8
6,421.8
FY99
0.0
0.0
0.0
2.3
1,389,770
0.9
0.1
0.8
0.5
45,267.7
210
17,902.1
2,465.8
15,334.9
9,564.9
32,042.0
12,671.7
1,745.4
10,854.6
6,770.3
FY00
1.4
0.2
0.0
2.4
1,684,281
0.0
0.1
0.2
0.5
1.5
0.2
0.0
2.7
1,863,275
0.1
0.1
0.2
0.5
57,050.7
262
31,683.5
5,209.1
29,426.0
3,822.2
51,642.8
234
1,532.7
2,996.0
5,166.4
10,463.0
735.4
1,726.6
5,188.4
10,744.3
50,376.6
27,977.0
4,599.7
23,224.7
3,016.7
40,759.5
1,353.4
2,645.5
4,562.0
9,239.0
FY02
580.4
1,362.7
4,095.0
8,480.0
FY01
0.0
0.1
0.3
0.6
62,616.1
297
35,163.4
4,837.2
802.1
3,001.3
6,291.0
12,521.0
62,616.1
35,163.4
4,837.2
802.1
3,001.3
6,291.0
12,521.0
FY04
1.6
1.5
0.3
0.2
0.0
0.0
3.1
2.8
2,045,854 2,273,142
0.0
0.2
0.3
0.8
67,588.2
280
35,081.5
5,660.7
834.5
3,333.2
5,714.2
16,964.1
63,611.7
33,017.5
5,327.7
785.4
3,137.1
5,378.0
15,966.0
FY03
Source data: MoF, SIKD.

Notes:
1. Central spending includes: 1. Ministry of National Education (MONE) Spending: Total spending for the whole ministry, it includes for schools, sport, youth, art & culture programs and management costs, and Ministry of Religious Affairs (MORA) spending
for schools (Madrasah and IAIN+STAIN) only.
2. During pre-decentralization (up to year 2000), central government transfers fund to local governments through MOHA. SDO and SBPP-SDN were for routine spending while Inpres were for development spending.
3. The actual length of fiscal year 2000 is 9 months. All figures were adjusted to 12 months.
Education Spending
(% GDP 2000 index)
292
Supporting table for Figure 5.4 in the text
Annex V.2 Education spending in Indonesia 1994-2004 (Rp billion and % GDP)
Central Spending (MOH)

Routine
Development
Province Level
Routine1
Development
District/City Level
Routine
Development
INPRES
Total
Routine
Development
Province Level
Routine1
Development
District/City Level
Routine
Development
INPRES
Total
CPI 1996=100
Routine
Development
Province Level
Routine1
Development
District/City Level
Routine
Development
INPRES
Total
GDP Assumption 2000 index
757.7
638.8
532.6
97.8
219.1
338.7
2,246.0
2,371.7
1,999.5
1,667.1
306.0
685.8
1,060.2
7,030.2
95
0.2
0.1
0.1
0.0
0.0
0.0
0.1
0.5
484,615
443.0
72.1
196.9
412.0
1,902.5
1,917.4
2,167.8
1,520.1
247.2
675.7
1,413.7
6,528.2
87
0.1
0.2
0.1
0.0
0.0
0.0
0.1
0.6
406,796
FY95
558.8
631.8
FY94
0.0
0.0
0.1
0.6
580,852
0.1
0.0
0.2
0.1
759.9
1,658.8
8,056.3
101
1,837.0
342.5
2,666.8
2,450.1
258.4
564.1
2,739.7
624.7
116.5
906.9
833.2
FY96
0.1
0.0
0.1
0.3
426.7
1,945.0
8,544.5
187
1,386.9
105.7
1,809.6
4,815.5
268.2
1,222.5
5,370.6
871.7
66.5
1,137.4
3,026.8
FY98
0.0
0.0
0.0
0.0
0.1
0.1
0.5
0.6
735,078 1,095,860
0.1
0.0
0.2
0.1
830.3
2,242.4
8,340.3
113
1,936.2
274.5
2,937.1
2,362.1
316.8
855.6
3,182.3
738.8
104.7
1,120.7
901.3
FY97
0.1
0.1
0.1
0.2
0.0
0.0
0.0
0.0
0.0
0.0
0.7
0.6
1,218,534 1,389,770
0.1
0.0
0.2
0.3
0.2
0.1
0.0
0.5
1,684,281
0.1
0.0
0.0
0.1
0.2
0.1
0.0
0.6
1,863,275
0.1
0.0
0.0
0.1
12,030.1
262
11,719.7
234
11,875.8
202
11,298.4
210
4,529.0
1,524.0
3,990.8
1,566.6
525.4
2,766.1
397.0
1,046.3
2,905.9
1,639.5
1,046.2
2,294.7
1,881.4
2,841.0
4,281.3
1,490.6
719.5
1,937.2
436.6
2,996.1
6,108.9
10,625.0
9,251.8
8,096.4
7,999.1
4,000.0
1,346.0
3,150.4
1,236.7
464.0
2,443.0
FY02
270.6
826.0
2,294.0
FY01
1,448.0
924.0
1,624.6
1,332.0
2,011.4
3,031.1
FY00
1,176.7
568.0
1,320.7
297.7
2,042.6
4,164.8
FY99
FY03
0.2
0.1
0.0
0.8
2,045,854
0.1
0.1
0.0
0.3
17,011.0
280
4,869.5
3,068.4
1,443.7
1,553.1
482.3
5,594.0
16,013.5
4,583.9
2,888.5
1,359.1
1,462.00.
454.0
5,266.0
Source data: MoF and SIKD.

Notes:
1. Before decentralization, routine data by sector or institution at local government level is not available. Numbers from 1996/97 through to 2000 are estimated by assuming that 7% of total SDO is for routine health sector.
2. The actual length of fiscal year 2000 is 9 months. All figures were adjusted to 12 months.
Education Spending
(% GDP 2000 index)
Real spending
(Rp billion 2004 prices,
adjusted with CPI index)
Nominal spending
(Rp billion,)
Supporting table for Figure 5.9 in text
Annex V.3 Health spending in Indonesia 1994-2004 (Rp billion and % GDP)
0.2
0.1
0.0
0.7
2,273,142
0.1
0.1
0.0
0.2
16,703.2
297
4,976.0
3,132.0
1,248.2
1,752.0
619.0
4,976.0
16,703.2
4,976.0
3,132.0
1,248.2
1,752.0
619.0
4,976.0
FY04
Annexes
293
294
3,843.2
652.9
1,469.0
5,965.1
1,522.3
6.8
18.0
1,547.2
Roads Development Spending

Central
Province
District
Total
Electricity Development Spending

Central
Province
District
Total
1,847.7
9.5
19.5
1,876.7
3,549.9
768.9
1,668.5
5,987.4
940.9
128.8
45.9
1,115.7
847.6
68.2
17.6
933.4
9,913.1
7,186.1
975.5
1,751.6
FY95
1,523.0
5.7
19.1
1,547.9
3,728.6
862.9
1,921.6
6,513.0
1,256.7
140.1
52.7
1,449.4
862.0
96.0
18.9
976.9
10,487.2
7,370.3
1,104.7
2,012.2
FY96
1,779.6
4.1
26.5
1,810.1
4,863.9
959.0
1,987.0
7,809.9
1,259.2
141.4
52.6
1,453.1
879.8
99.3
34.3
1,013.4
12,086.5
8,782.4
1,203.8
2,100.3
FY97
3,118.5
2.4
21.7
3,142.6
4,460.1
707.8
1,615.5
6,783.4
1,517.5
231.0
52.1
1,800.5
2,057.3
47.3
21.2
2,125.9
13,852.3
11,153.4
988.4
1,710.5
FY98
1,517.9
7.8
27.7
1,553.3
3,559.4
990.7
2,000.0
6,550.1
1,589.8
239.8
83.2
1,912.9
1,064.4
147.3
51.8
1,263.5
11,279.8
7,731.5
1,385.6
2,162.7
FY99
925.4
6.5
50.7
982.7
2,637.3
1,115.4
3,043.3
6,795.9
1,590.0
236.8
175.0
2,001.8
1,230.6
123.5
88.3
1,442.3
11,222.8
6,383.2
1,482.2
3,357.3
FY00
530.0
5.2
99.4
634.6
1,933.7
1,641.9
5,427.4
9,003.1
2,132.9
216.2
424.0
2,773.1
2,135.9
189.5
312.8
2,638.2
15,048.9
6,732.5
2,052.8
6,263.6
FY01
1,391.5
35.4
239.5
1,666.5
3,692.2
2,543.9
6,534.8
12,770.8
1,550.9
260.1
646.3
2,457.2
1,834.4
379.8
358.3
2,572.4
19,466.9
8,468.9
3,219.2
7,778.8
FY02
Source: SIKD data for district and province spending, Ministry of Finance for Central Spending.
Notes:
1. Decentralized spending in East Timor is excluded from analysis for years 1994-1997 in order to ensure comparability across time.
2. Electricity sector spending at central level is taken as 40 percent of energy sub-sector spending.
3. Comparable decentralized spending by sectors after 2003 is not available due to changes in coding in SIKD data. Further work needs to be done to reclassify this data in a comparable
manner across time.
942.7
110.0
0.1
1,052.8
Irrigation Development Spending

Central
Province
District
Total
780.1
18.9
43.4
842.4
9,407.6
Total
Water and Sanitation Development Spending

Central
Province
District
Total
7,088.3
788.6
1,530.6
FY94
Central
Province
District
Total Spending on Water, Irrigation, Roads, Electricity
Supporting Table for Figure 5.14 and Figure 5.16 in text.
Annex V.4 Infrastructure development expenditures by sector and level of government (nominal Rp billion)
28,820.5
10,650.8
4,844
13,325.5
FY03
26,517.5
8,352.1
5,894
12,271.6
FY04
Annexes
Annex V.5 Only 18 percent of the urban poor have access to tap water
URBAN
TOTAL
(Source of drinking water by per capita expenditure quintiles)
Bottled Water
Tap Water
Pump
Protected Well
Unprotected Well
Protected Spring
Unprotected Spring
River
Rain Water
Other
Total
Those having access to tap water (%)
1 (Poor)
184,546
3,997,992
3,860,545
16,997,656
6,914,060
4,996,525
3,227,920
1,737,656
1,285,894
165,318
43,368,112
9.22
Per capita expenditure quintiles

2
3
4
325,857
499,092
906,583
5,244,701
6,988,516
9,596,061
4,854,166
6,010,679
7,638,176
17,929,366
16,716,186
14,804,233
5,784,827
5,133,612
3,935,833
4,053,708
3,481,889
2,795,909
2,066,331
1,610,966
1,217,104
1,525,658
1,403,568
1,119,693
1,387,382
1,304,578
1,139,223
195,041
217,862
215,179
43,367,037
43,366,948
43,367,994
12.09
16.11
22.13
5
2,955,297
14,180,639
8,790,111
11,017,387
2,272,535
1,788,992
553,632
633,822
990,106
184,087
43,366,608
32.70
Total
4,871,375
40,007,909
31,153,677
77,464,828
24,040,867
17,117,023
8,675,953
6,420,397
6,107,183
977,487
216,800,000
18.45
Urban 2004
Bottled Water
Tap Water
Pump
Protected Well
Unprotected Well
Protected Spring
Unprotected Spring
River
Rain Water
Other
Total
1 (Poor)
140,884
2,496,491
2,077,299
6,089,765
1,491,191
564,469
287,408
142,280
263,812
71,215
13,624,814
18.32
2
229,834
3,643,995
2,569,447
6,064,635
1,107,866
453,956
161,806
116,006
269,345
96,200
14,713,090
24.77
5
2,648,199
12,230,753
6,449,541
5,062,132
560,527
274,817
50,250
62,671
319,740
118,475
27,777,105
44.03
Total
4,157,438
31,318,851
19,411,916
28,902,634
4,829,843
1,936,576
731,116
510,281
1,471,404
446,732
93,716,791
33.42
5
307,098
1,949,886
2,340,570
5,955,255
1,712,008
1,514,175
503,382
571,151
670,366
65,612
15,589,503
12.51
Total
713,937
8,689,058
11,741,761
48,562,194
19,211,024
15,180,447
7,944,837
5,910,116
4,635,779
530,755
123,100,000
7.06
3
391,395
5,250,385
3,421,088
5,971,829
933,522
330,178
150,441
110,721
317,177
80,106
16,956,842
30.96
4
747,126
7,697,227
4,894,541
5,714,273
736,737
313,156
81,211
78,603
301,330
80,736
20,644,940
37.28
RURAL
Per capita expenditure quintiles
Bottled Water
Tap Water
Pump
Protected Well
Unprotected Well
Protected Spring
Unprotected Spring
River
Rain Water
Other
Total
1
43,662
1,501,501
1,783,246
10,907,891
5,422,869
4,432,056
2,940,512
1,595,376
1,022,082
94,103
29,743,298
5.05
2
96,023
1,600,706
2,284,719
11,864,731
4,676,961
3,599,752
1,904,525
1,409,652
1,118,037
98,841
28,653,947
5.59
3
107,697
1,738,131
2,589,591
10,744,357
4,200,090
3,151,711
1,460,525
1,292,847
987,401
137,756
26,410,106
6.58
4
159,457
1,898,834
2,743,635
9,089,960
3,199,096
2,482,753
1,135,893
1,041,090
837,893
134,443
22,723,054
8.36
295
296
Health Cards
Access to Healthcare
Subsidies for basic medical services, operational support for health
centers; medicine and imported medical equipment, family planning services, nutrition, and midwife services.
Budget allocation during the SSN program (1998/2000) was around

Rp 1 trillion. It then fell to around Rp 500 billion (2000-02). In
2005, the program was supplemented with fuel subsidy savings:
planned budget for 2005 is Rp 3.9 trillion.
As of 2003, the health card covered 14.3 percent of households,
while 22.2 percent of households in the poorest quintile were covered. The overall coverage of health cards has increased over time,
and this increase has occurred in all regions with Papua showing
the highest increase.
Health care programs consisted of several supply side components channeled through health service providers, and consequently
the pure benefit to the poor is difficult to estimate. The most direct
assistance to the poor was provided through the distribution of
health cards that can be used to obtain free medical services in
Puskesmas, Pustu, village midwife, and public hospital under the
referral scheme. Health card targeting was slightly pro-poor. 31
percent of health cards were distributed to households in the poorest quintile, while 53.7 percent of cards were in the lowest 2 quintiles
in 2004. Program targeting performance seems to be deteriorating
slightly from 1999 to 2004, when increased coverage predominantly benefited non-poor households.
Scholarships
Access to Education
Scholarships and block grants providing financing directly to elementary, junior and secondary school students, and block grants
to selected schools
Rp 2.7 trillion (in 2003), the program was expanded in 2005 to

Rp 6.3 trillion about Rp 5.1 trillion of which is allocated to block
grants to schools and only about Rp 272 billion in scholarships to
senior secondary school students.
Scholarship coverage among households with children attending
primary to senior high schools was around 9.5 percent in 2001,
decreased to 5.9 percent in 2002, and rose again to 7.6 percent in
2003 (with higher coverage in poorer regions such as NT/Maluku,
Papua and Kalimantan).Scholarship coverage for the poorest
quintile was 14.9 percent in 2001, 9.4 percent in 2002 and 12.1
percent in 2003.
Targeting is more pro-poor than other programs: 39 percent of

benefits accrue to poorest quintile while 65 percent of benefits accrue to the poorest 2 quintiles combined. To be able to reach the
target group, the program developed decentralized mechanism of
allocation and selection using a combination of geographical targeting in defining regional allocation and individual targeting using certain criteria in selecting beneficiaries usually carried out at
the community/school level.
Raskin (subsidized rice)
Food Security
Special Market Operation (OPK) program: sales of subsidized rice

to households. The Raskin (subsidized rice for the poor) program
is a government program intended to provide social protection to
poor families in meeting their food adequacy and reducing their
financial burden by providing rice in subsidized price. It is supposed to distribute 20 kg of rice per family per month at a price of
Rp 1,000 per kg at specified distribution points.
In 2002, GoI allocated almost Rp 4.23 trillion for the Raskin program, and has increased this Rp 4.83 trillion in 2003, Rp 4.83 trillion in 2004, and Rp 4.68 trillion in 2005
Target 8.6 million households in 2003 and 2004. However, Susenas

shows higher coverage with lower benefits accruing to each beneficiary: in 2003, the program covered 37 percent of households
(20 million households). On average, around 50-60 percent of
households in the poorest expenditure quintile bought Raskin while
around 20 percent of the highest quintile also enjoyed it. An average beneficiary households bought only 5 kg per month of Raskin,
far less than the 20 kg allowance per household
There was almost no targeting in the disbursement of Raskin: 29

percent of benefits accrued to the poorest quintile, while 52 percent
of benefits accrued to the bottom 2 quintiles.
Objective:
Description of Program
and Social Safety Net
Goal
Budget Allocation
Coverage
Targeting Performance
(Based on analysis conducted for the Indonesia Poverty Assessment by the SMERU Research Institute, 2005)
Annex VI.1 Parameters and performance of Indonesia s principal targeted social programs
In Maluku, 47.8 percent of cards were given to households

in the poorest quintile and 71.4 percent of cards were given
to lowest 2 quintiles. This is the best targeting performance
of health cards in Indonesia. Kalimantan also had successful targeting with 35.9 percent of health cards in lowest
quintile and 58.4 percent of cards in lowest 2 quintiles. In
Sulawesi and Papua there was almost no targeting with
only 21.7 percent and 20.6 percent of cards respectively
being distributed to the poorest quintiles. The program
guidelines did not specify eligibility criteria, hence there
were different criteria used in the selection of beneficiaries
by region.
There was evidence that the distribution of health cards
played important role in maintaining the use of health services and increased the utilization of public health services
(Pradhan). Another rapid assessment conducted in 1999
by the SMERU Team indicated that because of the onset of
the monetary crisis and particularly due to the distribution
of health cards, the number of visits to the Puskesmas had
increased.
There were several institutions involved in the implementation of SSN in health, and the same institutional set up
has been maintained during the PKPS-BBM period. Therefore, both programs adopt the same structure of program
administration and implementation as well as monitoring
and evaluation, with the exception in the establishment of
the independent monitoring unit, CIMU HNSDP, which existed only during the live of the SSN program. According
to the program s guideline, at the central level, the program is managed by program secretariat within the Ministry of Health (MoH), assisted by technical advisory team
and program advisory team consisted of bureaucrats at
MoH.
Targeting in NT/Maluku is better compared to other regions

during the 2001 to 2003 period. On the contrary, there is
not targeting in Kalimantan with the poor and rich having
the same likelihood of accessing scholarships. Other regions that show a decrease in targeting performance are
Sumatra and Papua. In Papua case, while Scholarship coverage increased in 2002-03 the expansion was not propoor. Given the low infrastructure condition on Papua and
Kalimantan, the targeting problem could be rooted in the
difficulty in reaching the poor that lived in remote areas,
while the students selected by the schools in accessible
locations might not be the poorest on the island.
Scholarship and grants programs had initially contributed
to preventing enrollment rates from declining sharply between the 1997/98 to 1998/99 academic years. The scholarship amount was not adequate to cover school expenses
for students in Java, though in outer islands the scholarship amount was sufficient. This was particularly true for
junior secondary school level scholarships that only covered about 55 percent of schooling costs. In addition, there
was also critique that the scholarship scheme did guarantee students to continue their study from primary to junior
high and to senior high school, the transition with higher
drop-out probability for the poor.
The program has some innovative design features in its
allocation and delivery systems: these include (i) establishment of various levels of committees with tasks to determine allocation based on pre-defined criteria and channeling information as well as increase public awareness of
the program, (ii) direct disbursement and channeling of
funds through post offices, and (iii) an independent monitoring unit (CIMU). Targeting is carried out in two stages,
geographically and then at the community level, in a decentralized fashion which seems to have worked.
The targeting was relatively better in NT/Maluku, one of the

poorest regions in Indonesia. The targeting in Kalimantan
and Papua regions were the worst, where the households
in the two lowest quintiles only received around 30 percent
of program benefits (compared to 52 percent in all of Indonesia). In Kalimantan, targeting performance deteriorated
over time; while the targeting performance in Papua has
improved improving over the course of 2001 03.
A beneficiary household received subsidy equivalent to

around Rp 8,393 to Rp 11,235 through Raskin in 2003.
Assuming that average household size was 4.9, this monthly
subsidy was equivalent to a per capita transfer of Rp 1,713
to Rp 2,293 or around 1.7 percent to 2.2 percent of the
official poverty line.
The Raskin program is implemented though a collaboration between various government agencies: The institutional arrangement seems to be too loose in the sense that
there is no single government agency or institution that is
mandated to lead the program, and that is accountable for
the overall program performance. The total subsidy received
by all Raskin beneficiaries ranges between Rp 2.0-2.7 trillion (which is equivalent to 42-56 percent of the Raskin
budget in APBN 2003). This estimate is calculated using
two market price assumption, which is the Bulog purchasing price (Rp 2,790 per kg rice) as the minimum price and
GoI purchase price from Bulog (Rp 3,340 per kg rice) as
the maximum price. This is a costly mechanism for extending benefits to the poor.
Targeting Performance
by Region
Impact and Outcomes
Institutional Arrangement and Effectiveness
Annexes
297
298
28.04
31.28
69.31
Unattended birth
Malnourishment (0-59 months)
Does not attend to pre-school ( 5 & 6 years old)
6 years
Inactivity-does not attend to school and does not work (19-25)
36.25
66.36
female
41.96
female
male
29.59
male
28.29
female
30.70
male
Unemployment (16-25) (4)
17.14
female
97.65
18.32
male
Does not attend university (19-25) (3)
5.54
Child labor (13-15 years)
25 year
30.63
Age does not correspond to grade level (13-15 years)
60.62
19.38
Does not attend primary nor secondary school (13-15 years)
Does not attend secondary school (16-18 )
3.85
Early child labor (7-12 years) (2)
From 16 to
10.58
Age does not correspond to grade level (7-12 years) (1)
15 years
3.73
Does not attend primary school (7-12 years old)
From 7 to
63.71
34.25
32.54
22.87
23.12
27.22
95.53
47.81
11.11
11.33
4.23
21.10
12.66
3.12
8.00
2.38
58.24
26.89
18.69
8.06
Mother did not receive pre-natal health care
From 0 to
52.04
24.33
14.91
11.96
16.29
19.50
77.42
27.19
4.64
4.21
3.28
14.33
5.22
2.77
6.71
1.35
49.33
21.81
7.29
4.97
(**)
(*)
13.16
poor
poor
Poor
Household risk indicator
Age
group
61.40
34.62
39.67
29.87
23.00
29.31
97.69
61.21
25.21
14.03
16.28
38.05
29.05
4.40
10.58
4.77
67.01
36.28
26.30
13.72
Poor
53.08
33.24
32.38
26.38
26.57
29.56
96.94
49.67
11.17
8.23
5.21
20.95
11.13
6.37
10.37
4.54
65.49
28.62
13.82
11.70
poor
Near-
22.00
29.09
13.19
17.16
14.94
23.41
69.69
28.92
5.34
7.15
4.64
15.85
8.28
3.19
7.98
1.51
52.08
21.74
7.87
9.64
poor
Non-
Female HH head
NearNon-
Male HH head
Urban households
63.20
21.76
45.69
25.59
20.86
19.62
98.89
78.07
21.51
17.94
13.95
42.07
28.30
8.71
13.38
6.02
80.10
32.99
48.18
24.08
Poor
63.65
20.28
42.22
23.77
19.27
18.24
98.58
69.07
17.85
13.66
10.04
32.27
21.04
5.27
9.01
3.15
76.56
30.10
38.78
16.71
poor
Near-
Male HH head
65.00
20.59
33.75
20.27
16.77
17.89
95.12
54.43
10.87
8.30
7.45
21.58
12.34
5.12
7.61
2.19
71.11
26.60
28.22
13.61
poor
Non-
58.78
17.25
41.23
27.10
23.21
17.85
99.01
83.87
21.47
25.14
17.12
47.47
34.84
10.45
14.11
7.20
79.28
35.52
45.91
24.52
Poor
52.10
17.81
37.31
28.14
18.82
19.60
98.66
78.06
20.87
18.39
12.35
35.98
26.55
5.21
9.26
3.91
76.50
24.29
33.45
17.18
poor
Near-
46.39
16.83
30.44
18.06
19.42
14.96
94.67
62.48
12.36
13.30
11.98
29.01
17.15
4.75
9.12
2.61
69.43
24.82
26.66
21.89
poor
Non-
Female HH head
Rural households
Risk incidence (as percent of relevant population)
Annex VI.2 Risk and vulnerability analysis tables
64 years
From 56 to
55 years
From 26 to
Age group
2.03
16.63
1.26
Permanently disabled
9.70
female
38.92
9.98
8.60
female
male
8.24
male
61.19
Reports health problems (sickness or accident)
Underemployment
Unemployed
male
32.00
Head of household with more than 5 dependents (8)

female
0.87
Economic inactivity
25.74
Reports health problems (sickness or accident)
12.47
16.83
male
female
99.63
99.72
male
female
19.21
30.11
male
female
6.10
8.89
female
57.68
male
female
male
Poor
Permanently disabled (7)
Underemployment ( percent of all employed) (6)
Non professional self-employed ( percent of self employed)
Informally employed ( percent of all employed
Unemployment
Economic inactivity (non-participation in labor force) (5)
0.75
41.09
11.84
10.64
7.41
9.65
59.19
15.43
20.65
0.61
24.54
13.28
9.37
99.70
99.19
29.03
19.81
7.34
5.38
57.48
1.61
Nearpoor
(*)
5.59
27.45
22.51
8.40
12.71
26.36
0.87
39.74
3.89
5.93
8.09
9.92
67.38
24.77
9.00
0.36
24.30
6.78
4.75
0.64
46.89
18.29
0.00
9.38
50.02
54.28
13.03
3.22
24.76
17.83
14.46
98.18 100
2.53
46.81
16.02
8.22
10.12
4.28
46.10
43.55
13.17
2.04
25.39
13.62
11.29
99.91
99.29
21.31
22.54
10.29
15.12
26.67
7.04
Nearpoor
0.64
48.98
8.49
4.51
9.59
12.81
49.65
42.64
4.88
1.61
26.20
8.38
7.63
98.73
96.26
23.18
17.58
9.24
14.82
24.11
4.72
Nonpoor
Female HH head
Poor
98.36 100
21.07
14.23
6.06
4.45
57.03
1.79
Nonpoor
(**)
Male HH head
Urban households
1.48
36.93
9.50
11.76
7.22
4.24
52.58
8.14
32.51
0.80
23.77
20.43
17.14
99.17
99.69
46.58
34.17
6.09
2.77
46.03
1.13
Poor
1.02
37.37
8.41
7.56
6.91
3.41
48.87
7.27
19.61
0.61
23.72
18.22
15.94
98.98
99.61
47.53
34.83
5.52
2.77
45.15
1.10
Nearpoor
Male HH head
0.77
39.01
7.67
8.28
5.77
3.83
50.96
8.20
8.69
0.49
25.21
14.77
12.48
99.49
99.64
47.38
33.48
5.48
2.84
47.86
1.07
Nonpoor
1.11
39.35
15.11
27.95
7.78
37.52
24.34
14.53
2.04
22.77
27.25
20.80
99.92
98.87
40.26
33.68
6.39
10.79
21.65
5.91
Poor
3.70
47.08
9.31
12.94
5.63
1.08
33.26
24.29
9.17
2.21
23.67
20.73
25.26
98.78
100
45.00
42.79
5.75
7.38
19.07
3.90
Nearpoor
Female HH head
Rural households
2.04
50.51
8.85
0.00
7.99
33.97
17.48
3.24
1.69
28.30
16.96
20.52
99.68
99.56
51.52
42.57
7.16
8.86
17.54
3.60
Nonpoor
Annexes
299
300
Notes:
Source:
3.58
52.63
8.13
7.84
male
5.55
7.16
52.10
21.88
10.80
17.02
female
female
female
female
male
Poor
1.74
51.86
5.20
5.94
8.95
7.01
56.01
20.68
14.16
21.88
Nearpoor
(*)
1.97
51.53
6.01
4.57
11.14
9.01
52.50
19.87
19.28
33.96
Nonpoor
(**)
Male HH head
5.69
53.46
22.96
16.87
3.14
6.61
3.14
34.74
26.03
Poor
Urban households
0.11
61.87
9.47
0.00
1.80
7.73
21.19
33.53
38.85
Nearpoor
2.35
58.31
5.40
0.00
15.00
8.74
35.57
31.24
57.22
12.32
Nonpoor
Female HH head
2.96
50.31
7.19
6.72
5.34
4.18
65.04
24.31
10.45
16.70
Poor
2.36
53.06
4.35
6.25
3.95
5.20
67.99
25.93
14.52
28.69
Nearpoor
2.71
52.13
4.37
5.23
4.46
6.08
71.15
27.14
28.76
47.07
Nonpoor
Male HH head
4.02
56.96
15.08
27.24
0.15
6.33
26.94
35.60
30.98
18.46
Poor
3.50
54.93
11.40
8.74
1.11
4.90
9.49
36.36
49.63
0.79
Nearpoor
2.66
60.44
6.96
7.89
7.08
6.99
37.91
42.14
77.04
5.45
Nonpoor
Female HH head
Rural households
1. Children that are behind in educational level according to their age.

2. Working children.
3. Includes diploma 1, diploma 2, diploma 3, diploma 4, undergraduate, master and PhD.
4. Includes persons without work but looking for work, persons without work who have established a new business/firm, persons who are discouraged from finding employment, and persons who have a job but which starts at a future point.
5. Calculated by subtracting all working and unemployed from the population group.
6. Working less than or equal to 35 hours/week, and willing to work more hours or seeking other employment.
7. Includes persons who are blind, deaf, mute, have speech impairments, physical impairment and mental impairment.
8. Dependents: children age below 19, elderly adult (>=65) and persons (19-64 years) who are not working.
Indonesia PA Team, using Susenas 2004 except for malnourished and permanent disabled using Susenas 2003.
* includes quintile 1 or 2 but not poor.
** includes quintile 3, 4 and 5.
Permanently disabled
Reports health problems (disability, sickness or accident)
Underemployment
male
Unemployed
male
Labor Force Participation
Elderly adult living alone or only with spouse
65 and
older
Age
group
Annexes
Appendix Table 6.2 Risk mitigation and coping instruments captured in Susenas
Instruments (percent observed in each group)
I.
2003
2004
Poor
Nearpoor
Nonpoor
Poor
Nearpoor
Nonpoor
Mitigation (ex ante a shock)

I.I.
Prevention (actions that lower likelihood of losses)
1.
Privately provided education
11.31
12.97
15.15
10.41
12.97
15.75
2.
Safe water
45.46
51.92
58.64
49.08
49.82
60.29
3.
Sanitation
20.01
32.41
57.51
19.13
32.51
60.83
I.II. Saving (inter-temporal transfer without compensation for risk differential)

4.
Investment in education
31.05
31.44
27.92
30.63
32.20
29.40
5.
Saving in financial institute
11.18
15.75
27.51
14.57
16.65
30.01
6.
Other financial assets
0.04
0.10
0.25
0.29
0.35
0.21
7.
JAMSOSTEK (provident fund for retirement)
1.07
2.71
7.18
1.04
2.87
7.73
I.III. Risk-pooling (inter-temporal transfer with compensation for risk differential))

8.
Private insurance policies
1.02
2.04
6.08
1.36
2.36
7.71
9.
JPKM (health coverage provided by local government)
0.94
1.18
0.47
1.52
0.75
0.70
10. Health Fund
0.46
0.29
0.73
0.47
0.38
0.75
11. Health Card
20.56
14.86
7.68
15.98
13.11
6.77
0.83
1.77
12.14
0.89
3.34
12.16
20.81
26.75
29.64
20.13
23.07
31.39
0.80
1.20
1.49
1.02
1.71
1.27
13.22
18.97
19.99
13.40
14.96
20.98
12. ASKES Health insurance (SI for civil servants)

13. Payments to Arisan (rotating credit scheme)
II.
Coping (ex post a shock)

II.I. Risk-pooling
14. Payment from insurance policy
15. Payment from Arisan (rotating credit scheme)
16. Use of public health services (at local clinics)
0.67
1.26
1.28
0.65
1.12
1.50
71.33
63.38
38.91
62.84
54.98
32.99
4.83
2.61
1.69
3.97
2.76
1.75
19. Money transfers (usually from family)
36.76
37.91
35.77
38.18
34.66
36.30
20. Food transfers (usually from family)
62.36
57.38
52.40
60.14
57.72
51.91
11.20
14.11
18.68
10.74
11.22
19.75
22. Payments from other financial asset
0.59
0.26
0.25
0.44
0.36
0.27
23. Pensions
0.97
1.51
4.74
0.48
1.59
4.68
24. Inherited assets
0.66
1.14
1.63
0.55
0.91
1.68
25. Repayment of money lent to others
3.73
5.16
5.62
3.20
4.50
5.77
17.82
18.67
15.87
16.41
17.35
16.32
27. Equity release payments
0.11
0.12
0.31
0.06
0.12
0.32
28. Pawning goods
1.52
1.35
1.35
1.47
1.65
1.24
29. Sale of HH assets
11.37
8.93
7.97
10.83
9.35
6.94
30. Cuts in spending in food/education
26.36
22.90
16.47
18.91
14.28
9.87
17. Benefits from social safety net program

18. Scholarships (other than safety net scholarships)
II.II. Saving
21. Withdrawal of financial savings
26. Borrowing
301
302
-0.139
-0.205
-0.301
-0.176
(-2.19)**
(-3.23)***
0.36
(-1.79)*
-0.064
-0.166
0.048
-0.100
-0.030
0.0046
6,343
Health losses only (H)
Force majeur and economic losses (FMER)
Force majeur and health losses (FMH)
Economic losses and health losses (ERH)
Force majeur, economic losses and health losses (FMERH)
R square
Observations (number of households)
Poor
-0.35
(-1.65)*
-0.63
(-2.38)**
(-2.09)**
(-3.61)***
-0.56
t-statistics
Source: Staff estimates using Susenas 2003 and 2004.

Notes:
1. *** significant at 1 percent ; ** significant at 5 percent; * significant at 10 percent.
2. Shock variables show significant differences between households that reported shocks compared with households that did not report shocks.
687
0.0325
-0.084
-0.135
(-3.65)***
-0.063
Economic losses only (ER)
-0.23
-0.070
Coefficient
-0.65
t-statistics
-0.022
All
Force majeur only (FM)
Coefficient
Dependent variable: Change in log monthly food consumption
Appendix Table 6.3 Model 1: Impact of reported shocks on household expenditure on food
1,238
0.0111
(dropped)
-0.025
(dropped)
-0.518
0.021
0.010
-0.044
-0.21
(-3.63)***
0.37
0.28
-0.58
t-statistics
Near-poor
Coefficient
4,418
0.0018
-0.026
-0.066
0.029
-0.045
-0.058
-0.047
-0.030
-0.18
-0.91
0.21
-0.69
-1.51
(-2.14)**
-0.76
t-statistics
Non-poor
Coefficient
(-3.42)***
0.36
-1.09
-0.170
0.047
-0.059
-0.017
1.13
-0.7
(2.77)***
0.000
0.005
0.000
0.006
0.000
Experience square of HH head
Schooling years of HH head
Schooling years square of HH head
Total schooling years of non HH head
Total schooling years square of non HH head
Formal agriculture 03
-0.178
-0.135
0.110
Informal service 03
Informal service 04
Employment of non HH head (control variables)
-0.170
0.131
Formal service 03
Formal service 04
-0.167
0.114
Informal industry 03
0.117
Formal industry 04
Informal industry 04
0.007
-0.129
Formal industry 03
-0.082
Employment of HH head (control variables)
-0.1
0.000
(-6.65)***
(5.45)***
(-6.62)***
(6.58)***
(-8.68)***
(3.75)***
(-5.3)***
(5.36)***
(-6.04)***
0.3
(-3.82)***
-1.39
-0.13
(2.75)***
0.043
Sex of HH head (if male = 1)
Experience of HH head
1.64
0.033
Home ownership (if owned = 1)
HH characteristics (control variables)
(-1.68)*
-0.048
-0.14
-1.47
(-3.33)***
-0.048
-0.056
t-statistics
All
Shocks
Coefficient
0.025
0.007
-0.031
0.087
-0.061
-0.011
-0.105
0.009
-0.050
0.056
-0.134
0.000
0.006
-0.001
0.004
0.000
0.006
0.073
0.024
-0.157
-0.170
-0.225
-0.262
-0.136
-0.140
-0.075
Coefficient
Poor
0.46
0.13
-0.57
1.42
-0.97
-0.15
-1.55
0.16
-0.9
0.96
(-2.84)***
-0.86
1.07
-0.83
0.31
-1.54
(1.71)*
(1.67)*
0.22
-0.65
-1.6
-0.47
(-2.98)***
(-2.02)**
(-3.77)***
-0.61
t-statistics
-0.179
0.107
-0.149
0.108
-0.210
0.113
-0.106
0.089
-0.047
0.000
-0.100
-0.001
0.014
0.000
-0.004
0.000
-0.001
0.116
0.036
(dropped)
-0.041
(dropped)
-0.523
0.018
-0.008
-0.103
Coefficient
(-3.82)***
(2.59)***
(-3.55)***
(2.21)**
(-4.44)***
(1.87)*
(-1.66)*
(1.85)*
-1.01
0.01
(-2.4)**
(-2.14)**
(2.86)***
-0.44
-0.38
-0.3
-0.3
(3.36)***
0.68
-0.36
(-3.82)***
0.33
-0.23
-1.41
t-statistics
Near-poor
-0.238
0.110
-0.150
0.124
-0.192
0.115
-0.189
0.109
-0.159
-0.011
-0.070
-0.001
0.009
0.000
-0.003
0.000
0.000
0.038
0.050
0.002
-0.030
0.020
-0.050
-0.034
-0.048
-0.053
Coefficient
(-6.19)***
(4.44)***
(-5.98)***
(5.33)***
(-8.43)***
(2.99)***
(-4.62)***
(4.17)***
(-6.12)***
-0.35
(-2.47)**
(-2.75)***
(3.72)***
0.35
-0.53
-0.28
-0.33
(2)**
(2.26)**
0.02
-0.43
0.15
-0.79
-0.93
(-2.25)**
-1.39
t-statistics
Non-poor
Appendix Table 6.3 Model 2: Impact of reported shocks on household expenditure on food including household and labor market
control variables
Annexes
303
304
-0.110
6,341
All
(5.58)***
(-7.02)***
(8.39)***
(-12.61)***
(4.07)***
(-5.14)***
(7.84)***
(-10.65)***
(6.64)***
(-10.13)***
(4.24)***
t-statistics
687
0.1147
0.046
-0.088
-0.039
-0.048
0.042
-0.054
0.142
-0.141
0.058
-0.084
-0.058
Coefficient
Poor
1.01
(-1.87)*
-0.78
-1.15
0.73
-0.88
(2.79)***
(-2.77)***
(2.61)***
(-3.51)***
-1.05
t-statistics

Notes:
1. *** significant at 1 percent ; ** significant at 5 percent; * significant at 10 percent
2. Shock variables show significant differences between households that reported shocks compared with households that did not report shocks
0.0866
0.085
R square
03
04
Informal service
Informal service
0.120
-0.163
0.099
-0.134
0.142
-0.184
0.081
-0.122
0.114
Observation
03
04
Formal service
Formal service
03
04
Informal industry
Informal industry
03
04
Formal industry
Formal industry
03
04
Informal agriculture
Informal agriculture
04
Formal agriculture
Coefficient
1,238
0.1378
0.060
-0.097
0.106
-0.188
0.065
-0.188
0.138
-0.169
0.105
-0.120
0.160
Coefficient
(1.86)*
(-2.8)***
(2.94)***
(-5.88)***
1.48
(-3.47)***
(3.75)***
(-4.89)***
(4.63)***
(-5.25)***
(3.52)***
t-statistics
Near-poor
Appendix Table 6.3 Model 2: Impact of reported shocks on household expenditure on food (continued)
4,416
0.0929
0.091
-0.120
0.126
-0.169
0.143
-0.146
0.142
-0.195
0.093
-0.121
0.146
(4.96)***
(-6.49)***
(7.79)***
(11.38)***
(4.4)***
(-4.5)***
(6.38)***
(-9.17)***
(5.16)***
(-7.39)***
(3.71)***
t-statistics
Non-poor
Coefficient
0.58
-0.070
-0.364
0.083
-0.142
majeur and health losses (FMH)
majeur, economic losses and health losses (FMERH)
ER*Sanitation

0.033
(dropped)
0.547
FM*Capital
-0.035
0.138
FM*Receivable
FM*Borrowing
FM*Pawning
0.531
0.092
FM*Food transfer
FM*Asset
1.29
0.96
1.06
1.27
-0.09
0.53
0.003
(dropped)
(dropped)
-0.039
(dropped)
(dropped)
(dropped)
(dropped)
0.263
0.03
-0.07
(2.34)**
(dropped)
1.070
0.027
FM*Insurance
0.34
(2.54)**
0.725
FM*Pension
0.148
(dropped)
(dropped)
(dropped)
(dropped)
0.139
(dropped)
(dropped)
-0.160
(dropped)
-0.94
-0.676
-0.43
(1.65)*
-0.035
-0.067
-0.052
-0.576
-0.203
0.292
0.139
(dropped)
1.515
(dropped)
-0.093
-0.054
-0.057
-0.139
(1.71)*
0.25
1.23
-0.79
-0.39
-0.1
(-2.03)**
-0.81
0.8
0.7
(2.06)**
-0.28
-0.35
-0.87
-0.73
t-statistics
Near-poor
Coefficient
0.421
1.31
-0.65
-0.53
0.57
-0.44
0.4
-0.48
0.07
0.5
(-3.36)***
-2.13
t-statistics
FM*Money transfer
0.639
(dropped)
Poor
FM*Scholarship
-0.65
(-1.78)*
-0.127
-0.134
FM*Health care provided by public hospitals directly
FM*Social Safety Net
-0.509
-0.592
-0.58
(-2.49)**
-0.059
-0.264
FM*Arisan
FM*Selling HH assets
(dropped)
-1.4
(1.69)*
-0.103
0.189
0.426
-0.155
0.138
-0.225
0.014
FM*Sanitation
-0.36
-0.61
0.093
-0.276
-0.735
Coefficient
FM*Saving in financial institute
Shocks * Instruments, force majeur only
-0.86
-0.008
nomic losses only (ER)

-0.12
0.23
(-3.26)***
0.017
t-statistics
-0.112
All
Shocks
Coefficient
-0.035
0.407
(dropped)
0.114
0.047
0.118
0.492
-0.007
0.725
0.395
-0.022
-0.139
-0.160
-0.210
-0.067
0.177
-0.159
-0.199
0.065
-0.567
0.030
-0.061
-0.061
0.037
Coefficient
-0.8
0.77
0.95
0.12
0.62
1.18
-0.08
(2.53)**
1.54
-0.23
-1.53
-0.76
-1.58
-0.56
1.47
(-1.89)*
-0.5
0.3 Force
-1.21
0.2 Force
-0.66
-1.31
0.45 Eco-
t-statistics
Non-poor
Appendix Table 6.3 Model 3: Impact of reported shocks on household expenditure on food and the effectiveness of mitigation
and coping
Annexes
305
306
1.5
0.45
0.9
0.048
0.015
0.003
0.067
0.054
0.062
0.060
0.214
0.002
-0.141
-0.070
ER*Social Safety Net
ER*Money transfer
ER*Scholarship
ER*Pension
ER*Insurance
ER*Food transfer
ER*Asset
ER*Receivable
ER*Borrowing
ER*Capital
ER*Pawning
-0.438
-0.328
-0.61
(-2.33)**
0.89
0.91
-0.133
-0.257
0.061
0.185
0.015
H*Money transfer
H*Scholarship
H*Pension
H*Insurance
transfer
H*Asset
0.07
0.057
(dropped)
-0.187
-0.117
0.219
-0.050
H*Social Safety Net
-0.81
(dropped)
-0.05
(-2.41)**
-0.007
-0.309
-0.156
(-2.15)**
-0.179
H*Selling HH assets
-0.252
-0.419
0.128
0.13
-0.85
-0.8
-1.58
-0.68
1.1
-1.4
-0.8
-1.64
0.55
(dropped)
0.697
-0.090
-0.669
0.267
-0.063
0.098
0.226
-0.029
0.160
-0.032
0.306
0.375
-0.438
-0.060
0.096
(-2.16)**
-0.36
-0.010
(dropped)
(dropped)
-0.031
0.89
(-2.12)**
0.455
-0.375
1.37
-0.61
(-2.17)**
0.75
-0.52
0.69
0.42
-0.18
0.74
-0.17
(1.99)**
(1.66)*
0.24
-0.7
-0.07
1.3
(-2.01)**
0.105
0.21
0.064
-0.615
(1.9)*
-0.65
-0.82
0.7
0.88
(-1.9)*
0.262
-0.160
-0.069
0.049
0.252
-0.264
-1.07
0.24
0.673
(3.11)***
1.48
0.67
1.29
-1.37
-0.089
0.023
t-statistics
Near-poor
Coefficient
(dropped)
0.382
0.057
0.099
H*Health care provided by public hospitals directly
0.34
(2.54)**
0.025
0.192
H*Saving in financial institute
0.47
-0.76
-0.77
0.05
1.43
0.03
0.42
1.44
-0.77
H*Arisan
0.031
H*Sanitation
0.71
-0.104
ER*Health care provided by public hospitals directly
0.47
0.045
(-1.87)*
-0.085
ER*Selling HH assets
-0.422
(-2.47)**
(-3.04)***
-0.307
-0.314
-1.07
0.51
-0.044
t-statistics
Poor
Coefficient
0.020
t-statistics
ER*Saving in financial institute
All
ER*Arisan
Coefficient
and coping (continued)
0.086
0.047
0.128
-0.208
0.374
-0.051
-0.223
-0.024
-0.132
0.176
0.107
-0.003
-0.012
-0.247
0.009
0.175
0.191
0.217
0.041
-0.020
-0.017
0.005
0.033
-0.183
-0.069
-0.014
-0.013
Coefficient
0.36
0.19
1.47 H*Food
-1.57
0.33
-0.62
(-2.5)**
-0.16
-1.15
(2.04)**
1.15
-0.04
-0.1
-1.19
0.18
1.12
(2.22)**
1.2
0.89
-0.2
-0.12
0.11
0.75
-1.04
-1.24
-0.27
-0.27
t-statistics
Non-poor
0.410
0.075
H*Capital
H*Pawning
-0.013
(2)***
0.136
1.286
2.116
0.238
0.703
FMH*Arisan
FMH*Selling HH assets
FMH*Health care provided by public hospitals directly
FMH*Social Safety Net
1.09
0.58
1.6
0.25
-2.146
FMH*Sanitation
-1.54
-0.03
FMH*Saving in financial institute
(dropped)
FMER*Capital
-0.066
FMER*Borrowing
FMER*Pawning
(-4.58)***
-1.919
FMER*Receivable
-0.44
0.16
(dropped)
0.092
1.36
FMER*Food transfer
0.219
FMER*Asset
(dropped)
0.186
FMER*Pension
(3.04)***
0.427
FMER*Money transfer
FMER*Scholarship
FMER*Insurance
-1.45
-0.177
0.56
(-1.73)*
-0.713
FMER*Health care provided by public hospitals
FMER*Social safety net
-0.62
-1.65
-0.111
-0.293
FMER*Arisan
FMER*Selling HH assets
0.02
(-1.7)*
0.002
-0.217
FMER*Sanitation
0.48
1.46
0.28
1.61
t-statistics
FMER*Saving in financial institute
0.585
0.019
H*Receivable
H*Borrowing
Coefficient
All
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
0.168
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
0.434
-0.300
(dropped)
-0.334
0.018
-0.161
-0.236
(dropped)
(dropped)
-0.069
(dropped)
Coefficient
Poor
0.61
1.23
-1.22
-0.93
0.04
-0.76
-0.83
-0.38
t-statistics
0.535
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
0.242
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
0.293
(dropped)
(dropped)
0.274
-0.117
0.47
0.57
0.58
-0.27
0.825
0.090
2.117
1.417
0.422
-2.071
(dropped)
(dropped)
-0.242
(dropped)
0.036
(dropped)
0.087
(dropped)
-0.040
0.398
-0.168
-0.914
0.028
(dropped)
0.000
0.014
-0.181
-0.271
0.433
0.076
1.23
0.21
1.61
(2.11)**
0.66
-1.49
-1.05
0.06
0.45
-0.11
(2.31)**
-1
(-2.16)**
0.1
0.08
-0.99
-1.23
1.3
0.77
1.46
t-statistics
Non-poor
Coefficient
(dropped)
(-1.83)*
(-1.77)*
1.28
0.41
t-statistics
-1.248
-0.906
(dropped)
0.312
(dropped)
0.055
(dropped)
Coefficient
Near-poor
Annexes
307
308
0.585
0.019
0.410
0.075
-0.365
(dropped)
(dropped)
0.548
(dropped)
0.888
(dropped)
(dropped)
(dropped)
(dropped)
-0.019
0.045
0.170
0.203
-0.671
-0.337
0.064
0.855
-0.295
-0.103
-0.324
0.010
-0.600
0.005
(dropped)
0.681
H*Receivable
H*Borrowing
H*Capital
H*Pawning

FMH*Money transfer
FMH*Scholarship
FMH*Pension
FMH*Insurance
FMH*Food transfer
FMH*Asset
FMH*Receivable
FMH*Borrowing
FMH*Capital
FMH*Pawning

ERH*Sanitation
ERH*Saving in financial institute
ERH*Arisan
ERH*Selling HH assets
ERH*Health care provided by public hospitals directly
ERH*Social Safety Net
ERH*Money transfer
ERH*Scholarship
ERH*Pension
ERH*Insurance
ERH*Food transfer
ERH*Asset
ERH*Receivable
ERH*Borrowing
ERH*Capital
ERH*Pawning
Coefficient
All
(1.85)*
-0.15
0.27
1.22
1.32
(-2.73)***
(-2.69)***
0.51
(3.22)***
-1.23
-0.79
-0.66
0.06
-1.07
0.04
1.3
0.66
-0.65
1.61
0.28
1.46
0.48
t-statistics
0.120
0.118
0.848
(dropped)
(dropped)
-0.322
-0.474
0.459
(dropped)
0.086
(dropped)
-0.410
(dropped)
-0.168
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
-0.069
(dropped)
(dropped)
Coefficient
Poor
-0.11
-0.53
0.18
-0.78
-0.72
0.73
0.12
0.12
0.78
-0.38
t-statistics
0.032
1.979
0.070
0.894
-1.245
-1.299
-1.390
(dropped)
(dropped)
-0.495
(dropped)
-0.411
(dropped)
-0.943
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
0.055
(dropped)
0.312
Coefficient
-1.51
-0.66
-0.8
0.04
(2.07)**
0.09
1.04
(-1.9)*
(-2.08)**
(-1.89)*
1.28
0.41
t-statistics
Near-poor
0.143
-0.264
0.092
-0.072
-0.951
-0.316
0.435
1.706
-0.342
-0.379
-1.155
0.252
-0.331
0.196
(dropped)
1.022
0.379
(dropped)
(dropped)
(dropped)
(dropped)
1.011
(dropped)
0.459
(dropped)
(dropped)
0.535
0.076
0.433
-0.271
Coefficient
(2.22)**
0.7
-1.01
0.43
-0.33
(-2.23)**
(-1.73)**
(1.76)*
(3.26)***
-1.34
(-2.03)**
(-1.89)*
0.76
-0.57
0.89
0.55
1.42
0.68
1.46
0.77
1.3
-1.23
t-statistics
Non-poor
0.410
0.075
H*Capital
H*Pawning
0.1139
6,341
R square
Observations (Number of Households)
1.19
-0.48
1.13
-0.28
-1.47
687
0.218
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
-0.353
0.146
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
-0.069
(dropped)
Coefficient
Poor
-0.55
0.26
-0.38
t-statistics
1,238
0.1948
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
0.312
(dropped)
0.055
(dropped)
Coefficient
1.28
0.41
t-statistics
Near-poor
4416
0.122
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
(dropped)
0.165
-0.119
-0.041
0.257
-0.720
(dropped)
(dropped)
1.261
0.533
-0.069
-0.271
0.433
0.076
0.535
Coefficient
0.24
-0.12
-0.07
0.32
-1.5
1.62
0.94
-0.14
-1.23
1.3
0.77
1.46
t-statistics
Non-poor

Notes:
1.
*** significant at 1 percent ; ** significant at 5 percent; * significant at 10 percent.
2.
Shock variables alone show significant differences between households that reported shocks compared with households that did not report shocks.
3.
Shock variables interacted with instruments, show significant differences between households who reported shocks and had the instrument and those who reported shocks but did not have the instruments.
4.
Household and employment control variables also included in the regressions, but not presented in this table there were no significant differences in the estimated coefficients on these control variables.
(dropped)
(dropped)
FMERH*Capital
1.502
FMERH*Pawning
(dropped)
FMERH*Receivable
FMERH*Borrowing
0.600
(dropped)
0.281
FMERH*Pension
FMERH*Insurance
-0.418
-0.039
FMERH*Food transfer
0.34
-0.160
FMERH*Money transfer
FMERH*Scholarship
FMERH*Asset
-0.07
-0.709
0.15
0.211
FMERH*Health care provided by public hospitals directly
FMERH*Social Safety Net
-1.63
(dropped)
-1.274
FMERH*Arisan
FMERH*Selling HH assets
-0.2
(2.42)**
-0.094
1.776
FMERH*Sanitation
0.48
1.46
0.28
1.61
t-statistics
FMERH*Saving in financial institute
0.585
0.019
H*Borrowing
H*Receivable
Coefficient
All
Annexes
309
Annex VI.3: Indonesia: Likely CCT program impacts on the MDGs in program areas by 2015
Indicators
2002
MDG 1: Reduce poverty and hunger by half:

- Extreme poverty (1 US/day) ( percent)
- Child malnutrition (percent of 0-5 yr. olds)
14.0
7.5
24.6
MDG 2: Universal basic education1

- Net enrollment rate primary
- Net enrollment rate junior secondary
- Net enrollment rate senior secondary
91.5
51.0
42.4
92.7
61.7
48.2
100
Significant increase in progression rates from primary to

secondary (e.g., Mexico, Colombia, and Brazil).
MDG 3: Gender equality

- Ratio of young literate females to males (percent
ages 15-24)
96.7
99.0
100
Substantial
Substantial. Cash transfers are given to women who make
a better use of funds.
MDG 4: Reduce by two-thirds under 5 mortality rate

- Infant mortality (1000 live births)
- Under 5 mortality (1000 live births)
- Immunization measles ( percent of children under
12 months)
60.0
91.0
58.0
32.0
43.0
76.0
20.4
30.9
100
Substantial. Vaccinations over 95 percent in program areas, health check ups and growth monitoring will prevent
death by diarrhea and other diseases.
31.7
310.0
64.2
<100.0
100
Substantial
Substantial. Prenatal care starting in first trimester, births
attended by skilled staff, vitamins, folic acid, iron, other.
MDG 5: Reduce maternal mortality by two-thirds

- Maternal mortality rate (100,000 births)
- Percent births attended by skilled staff
MDG 6: Combat HIV/AIDS, malaria and other diseases
Prevalence of HIV/AIDS, female (percent, ages 15-24)
MDG 7: Promote sustainable use of the environment
Reduce by half the percentage without:
- Water (percent coverage)
- Sanitation (percent coverage)
1
Figures from World Bank, 2004c.
Source: World Bank, 2004b.
310
Target
2015
Likely program impact
1995
7.0
Likely
Likely, if health package includes component on health
education.
0.1
71.0
47.0
78.0
55.0
Significant
Significant. As shown in Mexico, Colombia, others. Reductions in poverty gap and in incidence of extreme poverty.
84.5
73.5
Substantial
Substantial. If program includes water and sanitation component in program areas.
Annexes
Annex VI.4 Budgets for the PKPS-BBM programs for 2006

Size of PKPS-BBM programs (Rp billion)
2005
2006
3,342.1
3,874
1,574
2,300
6,272
1,006.6
5,136.9
3,342.1
3,874
11,075.5
10,273.9
544,8
UCT
4,650
13,950
Total
18,138.1
32,241.6
PKPS-BBM for Rural Infrastructure

PKPS-BBM for Healthcare
Basic Health
Askes Program
PKPS-BBM for Education
Scholarships
BOS
311
References
Abadi, Tulus. 2005. Biaya Sosial Akibat Merokok. www.antirokok.or.id Majalah
Tarbawi, 17 March 2005.
Abdallah, H. 2002. Cost-Effectiveness of Self-Assessment and Peer Review
in Improving Family Planning Provider-Client Communication in
Indonesia. Quality Assurance Project Case Study. Maryland: U.S.
Agency for International Development (USAID).
Abreu, Maria. 2005. Changing Patterns on Child Malnutrition in Indonesia.
Background Paper for the Making Services Work for the Poor in
Indonesia. World Bank Office Jakarta. Processed.
Aedo, Christi n and Sergio N nez. 2001. The Impact of Training Policies in
Latin America and the Caribbean: The Case of Programa Joven .
Washington, DC: Inter-American Development Bank.
Aedo, Cristi n, and Marcelo Pizarro Valdivia. 2004. Rentabilidad Econ mica
del Programa de Capacitaci n Laboral de J venes Chile Joven .
INACAP and Mideplan. Santiago de Chile. Processed.
Afiff, Saleh, and C. Peter Timmer. 1971. Rice Policy in Indonesia. Food
Research Institute Studies in Agricultural Economics, Trade, and
Development. Vol. X, no. 2, pp. 131-159.
Akhmadi, Usman and Suryadharma D. 2004. When Teachers Are Absent: Where
Do They Go and What Are the Impacts on Students? SMERU Research
Institute. Jakarta.
Alatas, Vivi. 2005. An Evaluation of Kecamatan Development Project. World
Bank Office Jakarta. Draft Mimeo.
Alatas, Vivi and Lisa Cameron. 2003. The Impact of Minimum Wages on
Employment in a Low Income Country: an Evaluation Using the
Difference-in-Difference Approach. Policy Research Working Paper
2985. Washington, DC: World Bank.
Alatas, Vivi, Menno Pradhan and Vijayendra Rao. 2006. Conceptions,
Connection and Compromise: How Community-Based Targeting
Works in Indonesia. World Bank Working Paper.
Alatas, Vivi and Francois Bourguignon. 2005. The Evolution of Income
Distribution during Indonesia s Fast Growth, 1980-1996.
Microeconomics of Income Distribution Dynamics in East Asia and
Latin America, World Bank and Oxford University Press.
Alatas, Vivi and Janelle Plummer. 2005. Urban Poverty in Indonesia.
Background Paper for the Making the New Indonesia Work for the
Poor. World Bank Office Jakarta.
Alatas, Vivi, Kurnya Roesad, and Hendratno Tuhiman. 2005. Understanding
Unemployment and Poverty Puzzle. Background Paper for the Making
the New Indonesia Work for the Poor. World Bank Office Jakarta.
312
Alatas, Vivi and Widya Sutiyo. 2005. Poverty and Regional Disparities.
Background Paper for Making the New Indonesia Work for the Poor.
World Bank Office Jakarta.
Alatas, Vivi and Hendratno Tuhiman. 2005a. What Account for Indonesias
Significant Decrease in Poverty since 1999? Background Paper for
Making the New Indonesia Work for the Poor. World Bank Office
Jakarta.
_______. 2005b. A Historical Validation Exercise of Poverty Changes.
Alisjahbana, Armida. 2004. Indonesia s Employment Protection Legislation:
Swimming against the Tide? Report prepared for Growth through
Investment in Agriculture and Trade Project (GIAT) and Faculty of
Economics, Padjadjaran University, Bandung.
Alisjahbana, Armida S., and Chris Manning. 2005. Trends and Constraints
Associated with Labor Faced by Non-Farm Enterprises, Prepared
for the Rural Investment Climate Assessment. World Bank Office
Jakarta. Processed.
Ananta, Aris, and Evi Nurvidya Arifin. eds. 2004. International Migration in
Southeast Asia. Institute of South East Asian Studies (ISEAS):
Singapore.
Andrew, Matthew. 2005. Performance-Based Budgeting Reform: Progress,
Problems and Priorities, in Shah, Anwar .ed. Fiscal Management.
World Bank.
Angelini, John and Kenichi Hirose. 2004. Extension of Social Security
Coverage for the Informal Economy in Indonesia: Surveys of the Urban
and Rural Informal Economy. International Labor Organization (ILO),
Working Paper No. 11.
Aran, Meltem and Melanie Juwono. 2006a. Benefit Incidence of Education
Expenditures in Decentralized Indonesia. Background Paper for the
Making the New Indonesia Work for the Poor. World Bank Office Jakarta.
_______.2006b. Benefit Incidence of Health Expenditures in Decentralized
Indonesia. Background Paper for the Making the New Indonesia Work
for the Poor. World Bank Office Jakarta.
Arifianto, Alex. 2004. Social Security Reform in Indonesia: An Analysis of the
National Social Security Bill (RUU Jamsosnas). SMERU Research
Institute Working Paper.
Arifianto, Alex and Ellen Tan et al. 2005. A Report on Health Financing
Mechanism (JPK-Gakin) Scheme in Kabupaten Purbalinggo, East
Sumba, and Tabanan. Prepared for the Making Services Work for
the Poor in Indonesia. World Bank Office Jakarta.
References
Asian Development Bank (ADB). 2000. Assisting Girl Street Children at Risk
of Sex Abuse.
Bourchier, David, and Vedi Hadiz. ed. 2003. Indonesia: Politics and Society.
London: Routledge.
_______.2004a. Agricultural and Rural Development Strategy (in

collaboration with International Food Policy Research Institute - IFPRI.
March 2004.
Breierova, Lucia and Esther Duflo. 2002. The Impact of Education on Fertility
and Child Mortality: Do Fathers Really Matter Less than Mothers?
The Massachusetts Institute of Technology (MIT), the Center for
Economic Policy Research (CEPR) and National Bureau of Economic
Research (NBER).
_______. 2004b. Sustainable Social Protection and Providing Assistance

for the Vulnerable and Poor in the Informal Sector. Volumes I and II,
INTEM Consulting Inc., Japan for ADB, DFID and Bappenas.
_______. 2006. Core Paper Asia 2015: Promoting Growth, Ending Poverty.
Presented at the Asia 2015 Conference .ADB, World Bank and DFID.
January 2006.
Asia Foundation. Indonesian Rapid Decentralization Appraisal (IRDA). Fifth
Report, November 2004.
_____. 2004. IRDA. Fourth Report
_____. 2003. IRDA. Third Report
_____. 2002. IRDA Second Report
_____. 2002. IRDA. First Report
Attanasio, O., M. Syed and M. Vera Hernandez. 2004. Early Evaluation of a
New Nutrition and Education Program in Colombia. The Institute for
Fiscal Studies. Briefing Note 44: 11, London.
Badan Pusat Statistik Indonesia (BPS) and ORC Macro.2003. Indonesia
Demographic and Health Survey 2002-03. BPS and ORC Macro,
Calverton, Maryland.
Baeza, Cristian and Truman Packard. 2005. Beyond Survival: Protecting
Households from the Impoverishing Impact of Health Shocks.
Regional Study, Office of the Chief Economist, Latin America and
Caribbean Regional Office. World Bank.
Baker, Richard W. et al . eds. 1999. Indonesia: The Challenge of Change.
Institute of Southeast Asian Studies (ISEAS). Singapore.
Beegle, K., E. Frankenberg and D. Thomas. 2000. Labor Market Transitions of
Men and Women during an Economic Crisis: Evidence from
Indonesia. Rand Labor and Population Working Paper 00-11.
Bergstr m, Staffan and Elizabeth Goodburn. 2001. The Role of Traditional
Birth Attendants within the Reduction of Maternal Mortality. Vincent
De Brouwere and Wim Van Lerberghe. eds. Safe Motherhood Strategy:
A Review of Evidence. Antwerp: ITG Press.
Bitran, Ricardo and Ursula Giedion. 2003. Waivers and Exemptions for Health
Services in Developing Countries Social Protection Unit. World Bank.
March 2003.
_______. Health, Nutrition and Population. World Bank. http://
web.worldbank.org
Block, Steven. 2002. Nutrition Knowledge Rice Prices and the Nutritional
Impact of Indonesia s Crisis of 1997/98. September 17, 2002.
Booth, Anne. 2002. Growth Collapses in Indonesia: A Comparison of the
1930s and the 1990s. Working Paper. SOAS, University of London.
Brook, Penelope and Suzanne Smith. eds. 2003. Contracting for Public
Services: Output-Based Aid and Its Applications. Washington, DC:
World Bank.
Cameron, L., and C. Worswick. 2003. The Labor market as a Smoothing Device:
Labor Supply Responses to Crop Loss. Review of Development
Economics 7(2): 327-341.
Castaneda, Tarsicio. 2005. How Can Conditional Cash Transfers (CCT) Help
the Poor in Indonesia? September 2005. Mimeo.
Castaneda, Tarsicio, Kathy Lindert, with B n dicte de la Bri re, Luisa Fernandez,
Celia Hubert, Osvaldo Larranaga, Monica Orozco, and Roxana Viquez.
2005. Designing and Implementing Household Targeting Systems:
Lessons from Latin American and the United States.
Chaudhuri, Shubham. 2003. Assessing Vulnerability to Poverty: Concepts,
Empirical Methods and Illustrative Examples. New York: Columbia
University.
Chaudhuri, Shubham, Jyotsna Jalan, and Asep Suryahadi. 2002. Assessing
Household Vulnerability to Poverty from Cross-sectional Data: A
Methodology and Estimates from Indonesia. Department of
Economics, Discussion Paper Series No. 0102-52. New York:
Colombia University.
Chaudhury, N., J. Hammer, M. Kremer, K. Muralidharan and H. F. Rogers. 2006.
Missing in Action: Teacher and Health Worker Absence in Developing
Countries. Journal of Economic Perspectives 20(1): 91-116.
Chetty, Raj, and Adam Looney. 2005. Income Risk and the Benefits of Social
Insurance: Evidence from Indonesia and the United States. Mimeo.
Center for Health Research. 2001. A Study of the Tobacco Industry. University
of Indonesia.
Coady, David, Margaret Grosh and John Hoddinott. 2004. Targeting of Transfer
in Developing Countries. World Bank.
Conning, Jonathan and Michael Kevane. 2002. Community-Based Targeting
Mechanisms for Social Safety Nets: A Critical Review.
de Ferranti, David, Guillermo E. Perry, Indermit S. Gill, Luis Serv n with William
Maloney, Nadeem Ilahi, Francesco Fereira and Martin Rama. 2000.
Securing Our Future in a Global Economy. Chapter 6: Helping
Workers Deal with the Risk of Unemployment. Washington, DC:
World Bank.
de Janvry, Alain. 2006. Conditional Cash Transfer Programs in the Bigger Picture
of Social Policy. Third International Conference on Conditional Cash
Transfers. Turkey. June 2006.
De Pee S., M.W. Bloom et al. 2002. Breastfeeding and Complementary Feeding
Practices in Indonesia, Nutrition & Health Surveillance System Annual
313
Report. Jakarta, Indonesia: Helen Keller Worldwide.

Department of Public Works. 2005. General Guidelines Program on
Compensation for Reduction of Oil-Based Fuel Subsidy Rural
Infrastructure Sector PKPS-BBM. Directorate General of Housing,
Planning and Urban Development of the Republic of Indonesia.
Deutsch, R., A. Morrison, C. Piras, H. Nopo. 2002. Working Within Confines:
Occupational Segregation by Gender for Three Latin American
Countries. Presented at the Seminar Women at Work: A Challenge
for Development. Santiago, Chile. March 17, 2001.
Devas, Nick. 1989. Financing Local Government in Indonesia. Monographs
in International Studies, Southeast Asia Series No. 84. Athens: Ohio
University.
Djaja, Sarimawar. 2000. Revealing Priority Problems in Effort to Reduce
Maternal Death in Indonesia. Badan Litbang Kesehatan dan
Kesejahteraan Sosial.
Duflo, Esther. 2001. Schooling and Labor Market Consequences of School
Construction in Indonesia: Evidence from an Unusual Policy
Experiment. The American Economic Review, Vol. 91 (4): 795-813.
Eckardt, Sebastian and Stefan Nachuk. 2003. Measuring Governance: Art or
Science? World Bank Jakarta Office. Draft Mimeo.
Ehrlich, Isaac, and Gary Becker. 1972. Market Insurance, Self-Insurance and
Self-Protection. Journal of Political Economy 80: 623-648.
Elbers, Chris, Jean O. Lanjouw, and Peter Lanjouw. 2002. Micro-Level
Estimation of Welfare. Research Working Paper 2911. Development
Research Group. Washington, DC: World Bank.
Friedman, Jed. 2004. The Varying Relationship between Poverty and Growth
across Indonesia s Disparate Regions.
Friedman, Jed and James Levinsohn. 2002. The Distributional Impacts of
Indonesias Financial Crisis on Household Welfare: A Rapid Response
Methodology. World Bank Economic Review. 2002.
Fuglie, Keith O. 2004. Productivity Growth in Indonesian Agriculture: 19612000. Bulletin of Indonesian Economic Studies, Vol. 40 (2): 209225.
Gadjah Mada University. 2004. Study on Governance and Decentralization
Survey, GDS+1. Center for Population and Policy Studies. Yogyakarta.
Gaduh, Arya and Laila Kuznezov. 2005. Case Study 6: Health Insurance reform
in Jembrana District, Bali Province. bPrepared for the Making
Services Work for the Poor in Indonesia Case Studies. World Bank
Office Jakarta, the Ash Institute for Democratic Governance and
Innovation, in collaboration with the Ford Foundation International
Innovation Liaison Group. www.innovations.harvard.edu Processed.
Galasso, Emmanuela and Martin Ravallion. 2001. Decentralized Targeting of
an Anti-Poverty Program Development Research Group. World Bank.
Gannon and Zhi Liu. 1997. Poverty and Transport World Bank TWU Series
paper, TWU-30, September 1997.
Gertler, Paul J. 2000. Final Report: The Impact of Progress on Health IFPRI:
Washington DC.
Emmerson, Donald K., 1999. Indonesia Beyond Suharto: Polity, Economy,

Society, Transition. M.E. Sharpe, Armonk.
Gertler, P., S. Martinez and M. Rubio. 2006. Investing Cash Transfers to Raise
Long-Term Living Standards. Mimeo.
Escobal, Javier and Carmen Ponce. 2002. The Benefits of Rural Roads: Enhancing
Income Opportunities for the Rural Poor. Working Paper No. 40,
GRADE. Lima.
Gibson, John. 2006. The Constraints Associated with Infrastructure faced by

Non-Farm Enterprises at the Kabupaten Level. Background paper
for the Rural Investment Climate Assessment. University of Canterbury,
New Zealand.
Estache, A and A. Goicoechea. 2005. How widespread were infrastructure reforms

during the 1990s. Research Working Paper 3595. Washington, DC:
World Bank.
Filmer, Deon, S. Lieberman et al. 2002. Draft. Indonesia and Education for All
(EFA). Development Research Group and Human Development Unit
of East Asia Region, World Bank, Washington, DC. Processed.
314
______. 2005. Can expanding access to basic healthcare improve children s

health status? Lessons from Indonesia s midwife in the village
programme. Population Studies 59 (1): 5-19.
Gill, Indermit, and Nadeem Ilahi. 2000. Economic Insecurity, Individual

Behavior, and Social Policy. Office of the Chief Economist, Latin
America and Caribbean Region. Washington, DC: World Bank.
Gill, Indermit, Truman Packard and Juan Yermo. 2004. Keeping the Promise
of Social Security in Latin America. Stanford University Press.
Frankenberg, E., Duncan Thomas and Kathleen Beegle. 1999. Economic

Shocks, Wealth and Welfare. Journal of Human Resources 38
(2): 280-321.
Government of Indonesia. 2003. General Guideline Social Assistance Program

for Informal Sector and Self-Employed Jobs. Ministry of Social
Affairs.
Frankenberg, E. and Duncan Thomas. 2001. Women s Health and

Pregnancy Outcomes: Do Services Make a Difference?
Demography 38 (2): 253-265.
______. 2001. Anti-Poverty Programs in Indonesia: Analysis, Prospects and

Policy Recommendations. Puguh B. Irawan, Erman A. Rahman,
Haning Romdiati and Uzair Suhaimi. BPS Catalogue: 1155.
Frankenberg, E., W. Suriastini et al. 2004. Can Expanding Access to Basic

Health Care Improve Children s Health Status? Lessons from
Indonesias Midwife in the Village Program . eScholarship Repository,
University of California.
______. 2002. Jamsostek: The Laws and Regulations of the Republic of

Indonesia on the Employees Social Security .
______. 2004. Indonesia Progress Report on the Millennium Development
Goals. National Development Planning Agency (Bappenas). Jakarta.
February 2004.
References
______. 2004. Indonesia Educational Statistics in Brief, 2003/2004. Ministry

of National Education. Jakarta.
______. 2004. The Tobacco Source Book: Data to Support a National Tobacco
Control Strategy (English translation). Ministry of Health Republic
of Indonesia. Jakarta.
______. 2005. National Land Policy Framework. Jakarta. June 2005.
______. 2005. Implementation Manual Health Services Program at
Puskesmas and In-patients Treatment at Third-Class Hospitals with
Government Guarantee. Ministry of Health Republic of Indonesia.
Jakarta. July 2005.
______. 2005. Implementation Guidelines on Operational Aid for Schools
Compensation for Fuel Subsidies Decrease Program in the Field of
Education (PKPS-BBM). Ministry of National Education and Ministry
of Religious Affairs of the Republic of Indonesia. Jakarta.
Graham, W. J., J. S. Bell, et al. 2001. Can Skilled Attendance at Delivery Reduce
Maternal Mortality in Developing Countries? Safe Motherhood
Strategies: A Review of the Evidence. Wim Van Lerberghe and V. D.
Brouwere (eds). Antwerp: ITG Press.
Guggenheim, Scott et al. 2004. Indonesia s Kecamatan Development
Program: A Large-Scale Use of Community Development to Reduce
Poverty. World Bank Office Jakarta.
Haggard, Stephan. 2000. The Political Economy of the Asian Financial Crisis.
Institute for International Economics. Washington, DC.
Hartono, Djoko. 2005. Analyzing and Explaining Human Development
Outcomes Regarding Transition from Primary to Junior Secondary
School. Background Paper for the Making the New Indonesia Work
for the Poor. World Bank Office Jakarta. Processed.
Heitzmann, Karin, R. Sudharshan Canagarajah, and Paul Siegel. 2002.
Guidelines for Assessing the Sources of Risk and Vulnerability.
Social Protection Discussion Paper, No. 0218.
Hill, Hal. 1996. The Indonesian Economy since 1966: Southeast Asia s
Emerging Giant. Cambridge University Press.
Hofman, Bert, Ella Rodrick-Jones, and Thee Kian Wie. 2004. Indonesia: Rapid
Growth; Weak Institutions. Paper prepared for the Shanghai
Conference on Scaling Up Poverty Reduction. May 28-29, 2004.
Holzmann, Robert and Steen Jogensen. 2000. Social Protection as Social
Risk Management: Conceptual Underpinnings for the Social
Protection Sector Strategy Paper. Journal of International
Development 11:1,005-27.
Hoogeveen, Johannes, Emil Tesliuc. Renos Vakis and Stefan Dercon. 2005.
A Guide to the Analysis of Risk, Vulnerability and Vulnerable Groups.
Social Protection Unit. Washington, DC: World Bank.
Hull, Terence. ed. 2002. The Marriage Revolution in Indonesia. Paper for
Conference of the Population Association of America, Atlanta. May
9-11, 2002.
_______. 2005. People, Population and Policy in Indonesia. Institute of
Southeast Asian Studies. Singapore: Equinox Publishing.
Hydroconseil. 2004. Inception Report Small Scale Water Providers in

Indonesia, supported by the Water and Sanitation Program. World
Bank Jakarta Office.
Inter-American Development Bank (IADB). 2005. Ex-post evaluation training
programs: Youth Labor Training Program (Pro-Joven) in Peru. Expost Project Report. IADB, April 2005.
International Foundation for Electoral Systems (IFES). Campaign Finance in
Indonesia. Jakarta. December 1999.
Indonesian Demographic and Health Survey (DHS). 2002. On Reported Deaths
over the period 1998 to 2002.
International Crisis Group (ICG). 2003. Indonesia Backgrounder: A Guide to
the 2004 Elections. Jakarta. December 2003.
International Monetary Fund (IMF). 2006. Draft Report on the Observance of
Standards and Codes (ROSC): Fiscal Transparency Module.
Ishihara, Yoichiro. 2005. Indonesia: Public Expenditure Review: Public
Investment, Fiscal Space and Expenditure Allocations. World Bank
Office Jakarta.
Ishihara, Yoichiro and Elif Yavuz. 2005. Indonesia: Binding Constraints to
Private Investment. World Bank Jakarta Office. Processed.
Jenkins, Robert and Anne Marie Goetz. 1999. Constraints on Civil Society s
Capacity to Curb Corruption: Lessons from the Indian Experience.
IDS Bulletin Vol. 30 No. 4 October 1999. Institute for Development
Studies (IDS).
JHPIEGO. 2004. Preventing Postpartum Hemorrhage: A Community-Based
Approach Proves Effective in Rural Indonesia. www.mnh.jhpiego.org.
Johnston, Don. 2005. Notes on the Financial System and Poverty. Prepared
for the Rural Investment Climate Assessment. World Bank Jakarta
Office. Processed.
Kamelus, Deno, Jessica Ludwig and Suhirman. 2004. Study on the Efficiency
and Effectiveness of the Planning and Budgeting Process in Selected
Districts in NTB/NTT. GTZ- PROMIS NT.
Kanbur, Ravi. ed. 2003. Q-Squared: Qualitative and Quantitative Methods of
Poverty Appraisal. Delhi: Permanent Black.
Kearney, M. 2000. Rise in teen prostitutes. The Strait Times. 14
November 2000.
Knowles, J, and James Marzolf. 2003. Health Financing for the Poor in
Indonesia World Bank Paper Prepared for Regional Study on ProPoor Health Financing. World Bank Office Jakarta.
Komisi Pemilihan Umum (Indonesian Election Commission). 2005. Jakarta
www.kpu.go.id
Kuznezov, Laila and Janes Ginting. 2005. Improving Budget Transparency in
Bandung City, West Java. Prepared for the Making Services Work
for the Poor in Indonesia Case Studies. World Bank Office Jakarta,
the Ash Institute for Democratic Governance and Innovation, in
collaboration with the Ford Foundation International Innovation Liaison
Group. www.innovations.harvard.edu
315
_______. 2005. The Community Block Grant Program in Blitar City, East
Java. Prepared for the Making Services Work for the Poor in Indonesia
Case Studies. World Bank Office Jakarta, the Ash Institute for
Democratic Governance and Innovation, in collaboration with the Ford
Foundation International Innovation Liaison Group.
www.innovations.harvard.edu
Lanjouw, Peter. 2001. Non-Farm Employment and Poverty in Rural El Salvador.
World Development 29 (3): 529-47.
Lanjouw, Peter, Menno Pradhan, Fadia Saadah, Haneen Sayed and Robert
Sparrow. 2001. Poverty, Education and Health in Indonesia: Who
Benefits from Public Spending? World Bank Policy Research Working
Paper Series No. 2739. Washington, D.C.
Lebo, Jerry and Dieter Schelling. 2001. Design and Appraisal of Rural Transport
Infrastructure. TWU Series Paper, TWU-45, April. World Bank.
Leipziger, Danny. 2003. Millennium Development Goals: The Infrastructure
Contribution. Presentation for Poverty Day 2003 Opening Session
and Plenary on Infrastructure and Poverty Reduction.
Leisher, Susannah and Stefan Nachuk. 2005. Creating Learning Communities
for Children in Polman District, West Sulawesi. Prepared for the
Making Services Work for the Poor in Indonesia Case Studies. World
Bank Office Jakarta, the Ash Institute for Democratic Governance and
Innovation Liaison Group. www.innovations.harvard.edu
______. 2005. Participatory Planning in Maros District, South Sulawesi.
Prepared for the Making Services Work for the Poor in Indonesia Case
Studies. World Bank Office Jakarta, the Ash Institute for Democratic
Governance and Innovation, in collaboration with the Ford Foundation
International Innovation Liaison Group. www.innovations.harvard.edu
Lewis, Blaine. 2002. Revenue Sharing and Grant Making in Indonesia: The
First Two Years of Fiscal Decentralization. Research Triangle Institute.
Lloyd, Grayson and Shannon Smith. eds. 2001. Indonesia Today: Challenges
of History. Institute of Southeast Asian Studies (ISEAS), Singapore.
Lieberman, S., and Puti Marzoeki. 2000. Indonesia Health Strategy in a PostCrisis: Decentralizing Indonesia. Washington, DC: World Bank.
Lindenthal, Roland. 2004a. The Challenge of Social Protection for All: Policy
Options for Indonesia. Joint UNSFIR-ILO Working Paper, No. 04/11.
______. 2004b. Social Policy in Indonesia: Concepts and Categories.
UNSFIR, No. 04/10.
Lokshin, Michael and Martin Ravallion. 2005. Lasting local impacts of an
economy-wide crisis. Policy Research Working Paper Series 3503.
Washington, DC: World Bank.
Lowry, Robert, 1996. The Armed Forces of Indonesia. Allen & Unwin: St.
Leonards.
LP3E FE UNPAD and GIAT. 2004. Indonesia s employment protection
legislation: Swimming against the tide. Faculty of Economics,
University of Padjajaran Bandung and Growth through Investment,
Agriculture and Trade (GIAT) Project.
316
Manning, Chris. 2005. Notes on Labor and Poverty. Prepared for the Rural
Investment Climate Assessment. World Bank Jakarta Office.
Processed.
Manning, Nick. 2000. Pay and Patronage in the Core Civil Service in
Indonesia. PRMPS. World Bank.
Martineau, Jean-Noel and Yves Guerard. 2005. Implementing an Indonesian
National Social Security System Phase I Report, FIRST Initiative.
Martini, Santi and Muji Sulistyowati. 2005. The Determinants of Smoking
Behavior among Teenagers in East Java Province, Indonesia.
Economics of Tobacco Control Paper No. 32. HNP Discussion Paper.
Marzolf, James. 2002. The Indonesia Private Health Sector: Opportunities for
Reform: an Analysis of Obstacles and Constraints to Growth.
Discussion Paper. World Bank Office Jakarta.
McCulloch, Neil, Peter Timmer and Julian Weisbrod. 2006. The Pathways
out of Poverty in Rural Indonesia- an empirical assessment. World
Bank Jakarta Office. Mimeo.
McGranahan, Gordon and David Satterthwaite. 2004. Improving access to
water and sanitation: rethinking the way forward in light of the
Millennium Development Goals , in Bigg, Tom (ed), Survival for a
Small Planet - The Sustainable Development Agenda. London:
Earthscan Publications, pp 268-282.
McIntyre, Andrew. 2003. Indonesia as a Poorly Performing State? Australian
National University. Processed.
McKinsey. 1997. Banking in Asia: The End of Entitlement. Counselor
(Education, Science and Training), Australian Embassy; Directorate
General of Customs and Excise.
McMahon, Walter, Nanik Suwaryani, and Elisabeth Appiah. 2001. Improving
Education Finance in Indonesia: Policy Review. Institute for Research
and Development. Ministry of National Education, UNICEF and
UNESCO.
Mietzner, Marcus. 2006. The Politics of Military Reform in Post-Suharto
Indonesia: Elite Conflict, Nationalism, and Institutional Resistance.
East West Center: Washington.
Mukherjee, N. 2005. Qualitative Study: Making Services Work for the Poor.
Consultations with the Poor at Eight Sites in Indonesia. World Bank
WSP, Washington, DC.
Myrdal, Gunnar. 1968. Asian Drama: An Inquiry into the Poverty of Nations.
Pantheon Books: New York.
National Democratic Institute (NDI). 2005. Towards a More Effective Indonesian
House of Representatives: Options for Positive Change by Legislators.
Washington, D.C.
Newhouse, D., and K. Beegle. 2005 The Effect of School Type on Academic
Achievement: Evidence from Indonesia. World Bank Policy Research
Working Paper No. 36043604, Washington, DC.
Organization for Economic Co-operation and Development (OECD). 2004.
Public Sector Modernization: Governing for Performance. OECD
Observer. October 2004.
References
Olken, Ben. 2004. Monitoring Corruption: Evidence from a Field Experiment

in Indonesia. NBER Working Paper. Washington, DC. November 2004.
Poverty Line. Policy Research Working Paper Series 3157. Washington,

DC: World Bank.
Packard, Truman. 2002. Pooling, Saving and Prevention: Mitigating Old Age
Poverty in Chile. Policy Research Working Paper No. 2849. Washington,
DC: World Bank.
Ravallion, Martin and Michael Lokshin. 2005. Who Cares About Relative
Deprivation? December 2005. World Bank Policy Research Working
Paper No. 3782.
Packard, Truman, Vivi Alatas and Lina Marliani. 2005. Indonesia: Risk and
Vulnerability Assessment. Background Paper for Making the New
Indonesia Work for the Poor.
Ravindra, Adikeshavalu. 2004. An Assessment of the Impact of Citizen

Report Cards on the Performance of Public Agencies. Washington,
DC: World Bank.
Pangestu, Mari and Iwan J. Azis. 1994. Survey of Recent Developments.

Bulletin of Indonesian Economic Studies 30, no. 2.
Rawlings, Laura and Gloria Rubio. 2003. Evaluating the Impact of Conditional
Cash Transfer Programs: Lessons from Latin America. World Bank.
Paqueo, Vicente and Robert Sparrow. 2005. Free Basic Education in Indonesia:
Policy Scenarios and Implications for School Enrolment. Working Paper.
Reardon, Tom, and C. Peter Timmer. 2005. Transformation of Markets for

Agricultural Output in Developing Countries Since 1950: How Has
Thinking Changed? Chapter 13 in R.E. Evenson, P. Pingali, and T.P.
Schultz. eds. Handbook of Agricultural Economics: Agricultural
Development: Farmers, Farm Production and Farm Markets, Volume
3A. Forthcoming.
Parakesit, Danang. 2006. Unpublished paper for the Indonesia Rural Investment
Climate Assessment. World Bank Jakarta Office.
Parker, E. and A. Roestam. 2002. The Bidan di Desa Program: A Literature and
Policy Review. JHPIEGO Corporation, CEDPA, JHU/CCP, and PATH.
Parker, S. and Emanuelle Skoufias. 2000. Final Report: The Impact of Progresa
on Work, Leisure, and Time Allocation. October. International Food
Policy Research Institute (IFPRI), Washington, D.C.
Papanek, Gustav F. 2006. Guaranteed Employment for the Poor or a Cash-forWork Program: How is it Different from Existing Employment
Programs? Can it Succeed? January. Mimeo.
Plummer, Janelle. 2006. Making Infrastructure Expenditure Work for the Poor
Pradhan, Menno, Fadia Saadah and Robert Sparrow. Did the Health Card
Program Ensure Access to Medical Care for the Poor during
Indonesia s Economic Crisis? World Bank Economic Review,
forthcoming.
Pritchett, Lant, Sudarno Sumarto and Asep Suryahadi. 2002. Targeted
Programs in an Economic Crisis: Empirical Findings from the
Experience of Indonesia. SMERU Research Institute. Working Paper.
________. 2001. Safety Nets and Safety Ropes: Comparing the Dynamic
Benefit Incidence of Subsidized Rice and Public Works Programs in
Indonesia. Paper presented at Vulnerability and Poverty Workshop,
Third Asian Development Forum, Bangkok. June 2001.
Purwoko, Bambang. 1999. Towards a Social Security Reform: The Indonesian
Case. Jamsostek.
Rama, Martin. 1996. Consequences of Doubling the Minimum Wage: the Case
of Indonesia. Policy Research Working Paper 1643. Washington, DC:
World Bank.
Ravallion, Martin. Inpres and Inequality: A Distributional Perspective on the
Centre s Regional Disbursement. Australian National University.
Ravallion, Martin and S. Chen. 2003. Measuring Pro-Poor Growth. World
Bank, Policy Research Working Paper 2666.
Reynolds, Catherine. 1998. Worshipping Cancer Sticks. Inside Indonesia No.

56, October-December 1998.
Ridao-Cano, Cristobal and Deon Filmer. 2004. Indonesia: Evaluating the
Performance of SGP and SIGP: A Review of the Existing Literature
and Beyond. Human Development Sector Unit, East Asia and Pacific
Region. Working Paper 2004-3. Washington, DC: World Bank.
Robinson, J. S., B. R. Burkhalter et al. 2001. Low-Cost on-the-Job Peer Training
of Nurses Improved Immunization Coverage in Indonesia. Bulletin
of the World Health Organization 79: 150-158.
Robinson, Marc and Stephen Sherlock. 2004. Assessing Capacity Building
Needs for Effective Legislative Oversight of the Budget Process.
Mimeo, May 2004.
Rodrik, Dani, Ricardo Hausmann, and Andr s Velasco. 2004. Growth
Diagnostics. Working Paper. October. Kennedy School of Government,
Harvard University. Processed.
Rohdewohld, Rainer. 1995. Public Administration in Indonesia. Melbourne:
Montech Pty. Ltd.
Sanghvi, Harshad et al. 2004. Preventing Postpartum Hemorrhage in
Homebirths in Indonesia through Community Education and
Distribution of Misoprostol. JHPIEGO Maternal and Neonatal Health
Program. Maryland: JHPIEGO.
Sarimawar, Djaja. 2000. Revealing Priority Problems in Effort to Reduce
Maternal Death in Indonesia. Badan Litbang Kesehatan Departemen
Kesehatan dan Kesejahteraan Sosial. Jakarta.
Schwarz, Adam. 1994. A Nation in Waiting: Indonesia in 1990s. Westview Press.
Schady, Norbert R. 2006. Conditional Cash Transfer Programs: Reviewing the
Evidence. Third International Conference on Conditional Cash Transfers,
Turkey. June 2006,
Scheil-Adlung, Xenia. 2004. Indonesia: Advancing Social Health Protection
for the Poor. Joint UNSFIR-ILO Working Paper.
Ravallion, Martin and Michael Lokshin. 2003. On the Utility Consistency of
317
Shoemaker, J. 2005. International Family Planning Perspectives . Vol. 31

(3). Baltimore: JHPIEGO.
Simatupang, Pantjar. 2005. Agricultural Marketing and Competition.
Prepared for the Rural Investment Climate Assessment. World Bank
Jakarta Office. Processed.
Singh, Janmejay and Parmesh Shah. 2003. Making Services Work for Poor
People-The Role of Participatory Public Expenditure Management.
Social Development Notes. World Bank. March 2003.
Skoufias, Emmanuel. 2005. PROGRESA and Its Impacts on the Welfare of
Rural Households in Mexico. IFPRI Research Report No. 139.
International Food Policy Research Institute (IFPRI). Washington, D.C.
SMERU Research Institute. 2001. Wage and Employment Effect of Minimum
Wage Policy in the Indonesian Urban Labor Market. Jakarta.
_______. 2002. An Impact Evaluation of Systematic Land Titling under the
Land Administration Project (LAP). June 2002.
_______. 2004. Mapping Assistance Programs to Strengthen Micro
business. SMERU News No. 10. April-June 2004.
______. 2005a. Progress Report and Preliminary Assessment of District
Government Capacity. Part of Study on Capacity.
_______. 2005b. Development of Local Government through Assisting
Participatory Poverty Assessment (PPA). August.
_______. 2006a. PKPS-BBM Education Program Rapid Assessment:
Preliminary Findings. March 2006. Jakarta.
______. 2006b. PKPS-BBM for Rural Infrastructure and Clean Water.
Background Paper for the Making the New Indonesia Work for the Poor.
_______. 2006c. SSN and PKPS BBM for Education Sector School Grants
and Scholarships. Background Paper for the Making the New
Indonesia Work for the Poor. January 2006.
______. 2006d. SSN and PKPS BBM for Health Sector. Background Paper
for the Making the New Indonesia Work for the Poor.
Songco, Jocelyn A. 2002. Do Rural Infrastructure Investments Benefit the
Poor? Evaluating Linkages: A Global View, a Focus on Vietnam.
Policy Research Working Paper 2796. Washington, DC: World Bank.
Sood, S., U. Chandra et al. 2004. Measuring the Effects of the SIAGA
Behavior Change Campaign in Indonesia with Population-Based
Survey Results.
Strauss, John, Kathleen Beegle, Agus Dwiyanto, Yulia Herawati, Daan
Pattinasarany, Ewan Satriawan, Bondan Sikoki, Sukamdi, Firman
Witoelar. 2004. Indonesian Living Standards: Before and After the
Financial Crisis. Rand Corporation, USA and Institute of Southeast
Asian Studies.
Strauss, John, Kathleen Beegle, Bondan Sikoki, Agus Dwiyanto, Yulia Herawati
and Firman Witoelar. 2004. The Third Wave of the Indonesia Family
Life Survey (IFLS): Overview and Field Report. WR-144/1-NIA/NICHD.
Suharto, E. 2002. Human Development and the Urban Informal Sector in
Bandung, Indonesia: the Poverty Issue. In New Zealand Journal of
Asian Studies, 4, 2. December 2002.
318
Sulastomo. Social Security Reform in Indonesia. Mimeo.

Sumarto, Sudarno, Asep Suryahadi, and Lant Pritchett. 2001. Safety Nets and
Safety Ropes: Comparing the Dynamic Benefit Incidence of Subsidized
Rice and Public Works Programs in Indonesia Paper for presentation
at the Vulnerability and Poverty Workshop, Third Asian Development
Forum. Bangkok. June 12-14.
Sumarto, Sudarno, Asep Suryahadi and Wenefrida Widyanti. 2004.
Assessing the Impact of Indonesian Social Safety Net Programs
on Household Welfare and Poverty Dynamics. SMERU Research
Institute Working Paper.
Supratikto, G., G. Wirth, E. Achadi, S. Cohen and C. Ronsmans. 2002. A DistrictBased Audit of the Causes and Circumstances of Maternal Deaths in
South Kalimantan. Indonesia Bulletin of the World Health Organization
80 (3): 228-34.
Suryadarma, Daniel, Rima Prama Artha, Asep Suryahadi, and Sudarno Sumarto.
2005. A Reassessment of Inequality and its Role in Poverty Reduction
in Indonesia. SMERU Research Institute Working Paper.
Suryadarma, Daniel, Wenefrida Widyanti, and Asep Suryahadi. 2003. Access
to Income: An overview of the Broader Dimensions of Inequality in
Indonesia. SMERU Research Institute.
Tan, Ellen. 2005. Determinants of Maternal Mortality in Indonesia. Background
Paper for the Making the New Indonesia Work for the Poor. World
Bank Office Jakarta.
Tan, Ellen, C. Clarita Kusharto and Sri Budiyati. 2005. Rewarding Educational
Performance in Tanah Datar, Sumatra. Prepared for the Making
Services Work for the Poor in Indonesia Case Studies. World Bank
Office Jakarta, the Ash Institute for Democratic Governance and
Innovation Liaison Group. www.innovations.harvard.edu
_______. 2005. Vouchers for Midwife Services in Kabupaten Pemalang,
Central Java. Prepared for the Making Services Work for the Poor in
Indonesia Case Studies. World Bank Office Jakarta, the Ash Institute
for Democratic Governance and Innovation, in collaboration with the
Ford Foundation International Innovation Liaison Group.
_______. 2005. Water Supply and Health in Lumajang District, East Java
Province. Prepared for the Making Services Work for the Poor in
Indonesia Case Studies. World Bank Office Jakarta, the Ash Institute
for Democratic Governance and Innovation, in collaboration with the
Ford Foundation International Innovation Liaison Group.
Temple, Jonathan. 2001. Growing into Trouble: Indonesia after 1966. Working
Paper Department of Economics, University of Bristol. Processed.
The Economist. Digging for Dirt. London. 16 May 2006.

Thorbecke, Erik. 1995. The Political Economy of Development: Indonesia and
the Philippines. The Frank H. Golay Memorial Lecture. Cornell
Southeast Asia Program. Ithaca, NY.
References
Timmer, C. Peter. 1975. The Political Economy of Rice in Asia: Indonesia,

Food Research Institute Studies, 14 (3): 197-231.
_______. 1992. Agriculture and Economic Development Revisited.
Agriculture Systems (40): 21-58.
_______. 1997. How Well do the Poor Connect to the Growth Process?
Harvard Institute for International Development for the USAID/CAER
project. Processed.
_______.2002. Agriculture and Economic Growth. In Bruce Gardner and
Gordon Rausser, eds., The Handbook of Agricultural Economics, Vol.
II, 1,487-1,546. Amsterdam: North-Holland.
_______.2003. Food Security and Rice Price Policy in Indonesia: The
Economics and Politics of the Food Price Dilemma. In Mew, T. W.,
Brar, D. S., Peng, S., Dawe, D., and Hardy, B. eds. Rice Science:
Innovations and Impact for Livelihood. Proceedings of the International
Rice Research Conference, 16-19 September, 2002, Beijing, China.
International Rice Research Institute, Chinese Academy of Engineering,
and Chinese Academy of Agricultural Sciences, pp. 777-788.
_______.2004. The Road to Pro-Poor Growth: The Indonesian Experience
in Regional Perspective. Bulletin of Indonesian Economic Studies.
Vol. 40, No. 2 (August): 173-203.
_______. 2005. Operationalizing Pro-Poor Growth: A Country Case Study
of Indonesia. Poverty Reduction and Economic Management.
Washington, DC: World Bank. Processed.
Torrens, Anthony. 2005. KDP Economic Impact Analysis Study. World Bank
Office Jakarta. Mimeo.
West, Andrew. 2003. At the Margins: Street Children in Asia and Pacific Region.
Asian Development Bank (ADB) Working Paper. Poverty and Social
Development Unit.
Wetterberg, Anna. 2005. Crisis, social ties, and household welfare: testing
social capital theory with evidence from Indonesia. Social
Development Sector (EASSD). Report Number 34223. Washington,
DC: World Bank.
Willoughby, Christopher. 2004. How important is infrastructure for achieving
pro-poor growth? DAC Network on Poverty Reduction Global Picture
for Infrastructure and Pro-Poor Growth Room Document One. Paris.
March 29-30, 2004
Wodon, Quentin. 1997. Food Energy Intake and Cost of Basic Needs:
Measuring Poverty in Bangladesh. Journal of Development Studies
No. 34: 66-101.
Woodhouse, Andrea. 2002. Village Corruption in Indonesia: Fighting
Corruption in the World Bank s Kecamatan Development Program.
World Bank. 1993. The East Asian Miracle: Economic Growth and Public Policy.
New York: University Press.
_______. 1994. Indonesias Health Work Force: Issues and Options Report
No: 12834-IND Population and Human Resources Division,
Washington, D.C.
_______. 1997. Training and the Labor Market in Indonesia: Policies for
Productivity Gains and Employment Growth. Sector Report No. 16990.
Townsend, R. 1994. Risk and Insurance in Village India. Econometrica 62(3):

539-591.
_______. 2000. Managing Social Risks in Argentina. Human Development

Department, Latin America and Caribbean Regional Office.
Washington, D.C.
Transparency International. 2004. Transparency International: Global Corruption

Report 2004. London: Pluto Press. www.transparency .org
_______. 2001a. Social Protection Strategy: from Safety Net to Spring Board.
Social Protection Sector Board. Washington, D.C.
UNDP and Government of Malaysia. 2005. Malaysia: Achieving the Millennium

Development Goals, Successes and Challenges.
_______. 2001b. Indonesia: The Imperative for Reform. World Bank

Office Jakarta.
Usui, Norio and Armida Alisjahbana. 2004. Local Development Planning and
Budgeting in Decentralized Indonesia: Key Issues.
_______. 2002a. Indonesia Social Safety Net Adjustment Loan.

Implementation Completion Report. No. 24320. Washington, D.C.
Usui, Norio and Catur Sugiyanto. 2003. Development Planning, Budgeting

and Service Delivery: A Case of Lombok Tengah.
_______. 2002b. Accountability to the Poor: Experiences in Civic Engagement

in Public Expenditure Management.
von Luebke, Christian. Political Economy of Local Business Regulations:

Findings on Local Taxation and Licensing Practices from Four District
Cases in Central Java and West Sumatra. Prepared for the Rural
Investment Climate Assessment Case Studies. World Bank Office
Jakarta. Forthcoming.
_______. 2002c. A Sourcebook for Poverty Reduction Strategies (2 Volumes).

Edited by Jeni Klugman. Vol. 1: Core Techniques and Cross Cutting
Issues. Washington, DC: World Bank.
Wallace, W., Wolfgang Fengler, and Bastian Zaini. 2006. Increasing subnational government resources: Magnitude and implications. World
Bank Office Jakarta. Mimeo.
Warr, Peter G. 1984. Exchange Rate Protection in Indonesia. Bulletin of
Indonesian Economic Studies. Vol. XX (2): 53-89.
Wagstaff, A., and Menno Pradhan. 2003. Evaluating the Impacts of Health
Insurance: Looking Beyond the Negative.
_______. 2003a. Achievements Indonesia Maternal and Neonatal Health

Program. Maternal and Neonatal Health Technical Review. World Bank
Office Jakarta.
_______. 2003b. Health Financing in Indonesia. World Bank Office Jakarta.
_______. 2003c. Indonesia Education Sector Review. Human Development
Network. World Bank Office Jakarta.
_______. 2003d. Decentralizing Indonesia: A Regional Public Expenditure
Review. World Bank Office Jakarta.
319
_______. 2004a. Concept Paper on Operationalizing Pro-Poor Growth.

A Research Project Sponsored by AFD, DFID, GTZ, KfW and PREM.
May 11. Washington, D.C.
_______. 2004b. Indonesia: Averting an Infrastructure Crisis. Washington,
D.C: World Bank.
_______. 2004c. Shocks and Social Protection: Lessons from the Coffee
Crisis in Central America. Human Development Department, Latin
America and Caribbean Regional Office. Washington, D.C.
_______. 2004d. World Development Report 2004: Making Services Work
for the Poor. A co-publication of the World Bank and Oxford University
Press.
_______. 2004. Initiatives for Local Governance Reform Report. World Bank
Office Jakarta. July, 2004.
_______. 2005a. Indonesia - Tobacco Control Policy Options (draft). Health,
Nutrition and Population Unit, Human Development Network and East
Asia and Pacific Region. Washington, DC: World Bank.
_______. 2005b. Information and Communication Technologies for Rural
Development: Issues and Options. Washington, DC: World Bank.
_______. 2005c. Pro-poor growth in the 1990s: Lessons and insight from
14 countries. A Research Project Sponsored by DFID, AFD, GTZ,
Kfw. June. Washington, DC: World Bank.
_______. 2005d. Teacher Employment and Deployment Study. Human
Development Network. World Bank Office Jakarta.
_______. 2005e. Civil Service Diagnostics and Road Map for Reform in Cluster
2 Region (West Sumatra), World Bank Office, Jakarta.
_______. 2005. Public Expenditure Review. Forthcoming. World Bank Office.
Jakarta.
_______. 2005. East Asia Update. Washington, DC: World Bank.
_______. 2005. Strategy Note for Government of Indonesia: Expansion of
a National Poverty Reduction Program. World Bank Jakarta Office.
May 2005.
_______. 2006a. Impediments to Effective Rural Infrastructure: Roads in
Manggarai District, NTT Province. Rural Investment Climate Case
Study 5. World Bank Office Jakarta.
_______. 2006b. Indonesia: An Assessment of Public Spending in
Infrastructure Sectors. World Bank Office Jakarta. Mimeo.
_______. 2006c. Indonesia Enabling Water Utilities to serve the Urban Poor.
Infrastructure Network. A co-publication of the World Bank East Asia
Infrastructure Department and Indonesia Country Program.
_______. 2006d. Infrastructure and the Climate for Rural Investment in Indonesia:
A Question of Resource Allocation. Prepared for the Indonesia Rural
Investment Climate Assessment. World Bank Office Jakarta.
320
_______. 2006e. Investing for Growth and Recovery. CGI Brief Paper. World
_______. 2006f. Making Services Work for the Poor in Indonesia, Case
Study: Vouchers for Midwife Services in Pemalang District, Central
Java. Prepared for the Making Services Work for the Poor in
Indonesia. World Bank Office Jakarta.
_______. 2006g. Making Services Work for the Poor in Indonesia, Case
Study: Water Supply and Health in Lumajang District, East Java.?
Prepared for the Making Services Work for the Poor in Indonesia.
________.2006h. Revitalizing the Rural Economy: An Assessment of the
Rural Investment Climate in Indonesia, Consultative Draft. Jakarta.
Forthcoming.
_______. 2006i. The Impact of Formal and Informal Labor Regulation on
Business in Serang District. Rural Investment Climate Case Study
2. Jakarta.
_______. 2006j. World Development Report 2006: Equity and Development.
Washington, DC: World Bank.
_______. 2006k. Making Services Work for the Poor in Indonesia: Focusing
on Achieving Results on the Ground. Jakarta. Forthcoming.
_______. 2006l. The Nuts and Bolts of Brazil s Bolsa Familia Program:
Implementing Conditional Cash Transfers in a Decentralized Context.
Draft for Discussion, June 2006.
_______. 2006m. Indonesia: Sustaining Economic Growth, Rural
Livelihoods, and Environmental Benefits: Strategic Options for
Forestry Assistance in Indonesia. Jakarta.
_______. 2006. Evaluation Fiscal Equalization in Indonesia. Policy Working
Paper 3911. Washington, DC: World Bank. May 2006.
_______. 2006. Opportunities and Constraints for Civil Service Reform in
Indonesia. World Bank and Partnership for Governance Reform.
_______. 2006. Education Public Expenditure Review. Forthcoming. World
_______. Poverty Library. http://povlibrary.worldbank.org/files/
4978_chap18.pdf
World Health Organization (WHO). 2003a. Indonesia Reproductive
Health Profile.
_______. 2003b. Country Profile: Indonesia. www.who.org
_______. 2004. Indonesia Public Health Expenditure Survey 2004.

Making The New Indonesia - WB Poverty

Загружено:

Сведения о документе

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Making The New Indonesia - WB Poverty

Загружено:

Авторское право:

Доступные форматы

Work for the Poor

MAKING THE NEW INDONESIA WORK FOR THE POOR

THE WORLD BANK OFFICE JAKARTA

Photographs by Jacqueline Koch Photography

Inside cover photo: Copyright

Jez O Hare Photography

MAKING THE NEW INDONESIA WORK FOR THE POOR

MAKING THE NEW INDONESIA WORK FOR THE POOR

Asian Development Bank

Development Planning Agency at Sub-National Level (Badan Perencanaan Pembangunan Daerah)

Community Achievement Fund

Special Allocation Fund (Dana Alokasi Khusus)

Food and Agriculture Organization

Governance and Decentralization Survey

Human Development Index

Indonesian Agency for Agricultural Research and Development

National Institute of Administration (Lembaga Administrasi Negara)

MAKING THE NEW INDONESIA WORK FOR THE POOR

Reconstruction of Aceh Land Administration Project

Primary School (Sekolah Dasar)

Senior High School (Sekolah Menengah Atas)

Pension fund for civil servants

Unconditional Cash Transfer

Conservation Information Kiosk (Warung Informasi Konservasi)

Sandalwood Education Foundation (Yayasan Pendidikan Cendana)

Editorial and data note

MAKING THE NEW INDONESIA WORK FOR THE POOR

Chapter 2 A History of Growth and Poverty Reduction

A Troubled History and Chronic Poverty

A Period of Growth and Poverty Reduction

The Financial Crisis

Post-Crisis: Stabilization, Democracy and Decentralization

Conclusion: Learning from History to Support Poverty Reduction

Chapter 3 Understanding Poverty in Indonesia

A Profile of the Poor

The Determinants of Poverty

The Determinants of Recent Poverty Changes: a Dynamic Analysis

Conclusion: Poverty Diagnostics Provide Pointers for Poverty Reduction Efforts

Portraits from the Regions

Chapter 4 Making Growth Work for the Poor

What Are the Pathways out of Poverty?

Linking Pathways out of Poverty to Policies for Pro-Poor Growth

Maintaining a Stable Macroeconomy

Investing in the Capabilities of the Poor

Connecting the Poor to Growth

Conclusion: Making Growth more Pro-Poor

Spotlight on Gender: Remittances from female migrant workers: A lifeline to communities

Aggregate Spending Levels and Fiscal Space

MAKING THE NEW INDONESIA WORK FOR THE POOR

Human Development Sectors

Focus on Reducing Maternal Mortality

Conclusion: Achieving Synergies through the Right Development Spending Mix

The Evolution of Social Protection Institutions in Indonesia: from Universal Subsidies to

Conclusion: Towards a Third-Generation System

Spotlight on Innovation: Poverty maps: A powerful tool in targeting the poor

Policy, Planning and Budgeting Systems

Monitoring and Assessment of Poverty and Poverty Reduction Interventions

Conclusion: Towards Government Services Focused on the Poor

Focus on Annual Plan and Budget Preparation for2006

Poverty Reduction Policy Matrix

Indonesias elections and the consolidation of democracy

Poverty definitions and measures