essays on decentralisation, public services and well-being in indonesia
DESCRIPTION
Essays on decentralisation, public services and well-being in IndonesiaTRANSCRIPT
-
Essays on decentralisation, public services andwell-being in Indonesia
A thesis submitted to the University of Manchester for the degree of
Doctor of Philosophy in the Faculty of Humanities
2013
Sujarwoto
School of Social Sciences
-
Contents
Abstract vii
Declaration and Copyright viii
Acknowledgements ix
Abbreviations and Glossary x
1 Introduction 11.1 Decentralisation in Indonesia: from centralistic government to radical
decentralisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.2 Assessing decentralisation literature on developing countries . . . . . . 12
1.2.1 Positive associations . . . . . . . . . . . . . . . . . . . . . . . . 171.2.2 Negative associations and inconclusive findings . . . . . . . . . . 221.2.3 Why does decentralisation work and not work? . . . . . . . . . 25
1.3 Current decentralisation studies: three research gaps . . . . . . . . . . 321.3.1 The limitations of cross-country analysis and a few studies based
on local government reform . . . . . . . . . . . . . . . . . . . . 341.3.2 Insufficient analysis of the consequences of decentralisation for
subjective well-being . . . . . . . . . . . . . . . . . . . . . . . . 351.3.3 A limited analysis of the contextual effects of decentralisation on
well-being . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 371.4 Research contribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . 381.5 Research aims and questions . . . . . . . . . . . . . . . . . . . . . . . . 391.6 Data, measures and methods . . . . . . . . . . . . . . . . . . . . . . . . 41
1.6.1 Survey data and official statistics . . . . . . . . . . . . . . . . . 421.6.2 Decentralisation and well-being measures . . . . . . . . . . . . . 461.6.3 Multilevel analyses and instrumental variables method . . . . . 52
1.7 Description of chapters . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
2 Political decentralisation and public service performance in Indonesia:Multilevel multiple indicators multiple causes (MIMIC) analysis 572.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 582.2 Political decentralisation and local public service performance . . . . . 62
i
-
2.3 The political accountability system and decentralisation in Indonesia . 672.4 Data, determinants and method . . . . . . . . . . . . . . . . . . . . . . 70
2.4.1 The Governance Decentralisation Survey 2006 and official statistics 702.4.2 Local government performance measure . . . . . . . . . . . . . . 722.4.3 Local government and household determinants . . . . . . . . . . 742.4.4 Multilevel analyses . . . . . . . . . . . . . . . . . . . . . . . . . 76
2.5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 782.6 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 862.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
3 Decentralisation and citizen happiness: Multilevel analysis of self-rated happiness in Indonesia 933.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 933.2 Decentralisation and happiness . . . . . . . . . . . . . . . . . . . . . . 973.3 Democracy and political transition in the contemporary Indonesia . . . 1003.4 Data, determinants and method . . . . . . . . . . . . . . . . . . . . . . 103
3.4.1 The Indonesian Family Life Survey (IFLS) 2007 and official statis-tics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
3.4.2 Happiness and decentralisation measures . . . . . . . . . . . . . 1053.4.3 Local government determinants . . . . . . . . . . . . . . . . . . 1073.4.4 Individual determinants . . . . . . . . . . . . . . . . . . . . . . 1103.4.5 Multilevel analyses . . . . . . . . . . . . . . . . . . . . . . . . . 113
3.5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1143.6 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
3.6.1 Local government economic growth, community social capital andhappiness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
3.6.2 Individual determinants of happiness . . . . . . . . . . . . . . . 1253.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
4 Decentralisation, social capital and child health in Indonesia: Instru-mental variables estimation method 1294.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1304.2 Social capital and health outcomes . . . . . . . . . . . . . . . . . . . . 1324.3 Community development and health in Indonesia . . . . . . . . . . . . 1374.4 Data, measures and method . . . . . . . . . . . . . . . . . . . . . . . . 140
4.4.1 The Indonesian Family Life Survey (IFLS) 2007 . . . . . . . . . 1404.4.2 Measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1414.4.3 Instrumental variables estimation method . . . . . . . . . . . . 1484.4.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1504.4.5 Mothers social capital and child health . . . . . . . . . . . . . . 1534.4.6 Mothers social capital and child health: two-way causality? . . 155
4.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1574.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
ii
-
5 Public spending and healthcare demand in Indonesia: Multilevel finitemixture analysis 1635.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1645.2 Decentralisation, public health spending and health care demand in de-
veloping countries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1685.3 Indonesian healthcare reform . . . . . . . . . . . . . . . . . . . . . . . . 1745.4 Data and method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
5.4.1 The Indonesian Socio-Economy Survey (Susenas) 2009 and offi-cial statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
5.4.2 Healthcare demand and fiscal decentralisation measure . . . . . 1815.4.3 Local government determinants . . . . . . . . . . . . . . . . . . 1835.4.4 Individual socio-demographic determinants . . . . . . . . . . . . 1845.4.5 Multilevel finite mixture analyses . . . . . . . . . . . . . . . . . 186
5.5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1905.6 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1955.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
6 Conclusion 2026.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2026.2 Key findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2026.3 Main contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2076.4 Policy implications for current decentralisation and development in In-
donesia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2096.5 Limitations and suggestions for further research . . . . . . . . . . . . . 213
6.5.1 Add qualitative research to enrich understanding of the processof decentralisation . . . . . . . . . . . . . . . . . . . . . . . . . 213
6.5.2 Extend research to include longitudinal data . . . . . . . . . . . 2146.5.3 Extend research to other well-being measures . . . . . . . . . . 2156.5.4 Extend research to other decentralisation and local government
capacity measures . . . . . . . . . . . . . . . . . . . . . . . . . . 2166.5.5 Extend research to longer period of data . . . . . . . . . . . . . 2186.5.6 Extend research to cover a greater number of local government
authorities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2196.5.7 Extend research to include spatial multilevel analyses . . . . . . 220
A Appendix 1: Multilevel regression results and correlations 261A.1 Chapter two . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
A.1.1 Multilevel regression results with factor analysis of perceived localpublic service performance . . . . . . . . . . . . . . . . . . . . . 261
A.2 Chapter three . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262A.2.1 Multilevel regression results of self rated happiness with xtmixed 262A.2.2 Bivariate correlation of self rated happiness determinants . . . . 263
A.3 Chapter five . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266
iii
-
A.3.1 Bivariate correlation of healthcare demand determinants . . . . 266A.3.2 The distribution of two component multilevel finite mixture neg-
ative binomial for unconditional being ill sample . . . . . . . . . 267A.3.3 The distribution of two component multilevel finite mixture neg-
ative binomial for conditional being ill sample . . . . . . . . . . 267
B Appendix 2: Stata and MPlus code 268B.1 Stata code for chapter two . . . . . . . . . . . . . . . . . . . . . . . . . 268B.2 MPlus code for chapter two . . . . . . . . . . . . . . . . . . . . . . . . 281B.3 Stata code for chapter three . . . . . . . . . . . . . . . . . . . . . . . . 284B.4 Stata code for chapter four . . . . . . . . . . . . . . . . . . . . . . . . . 298B.5 Stata code for chapter five . . . . . . . . . . . . . . . . . . . . . . . . . 328B.6 MPlus code for chapter five . . . . . . . . . . . . . . . . . . . . . . . . 340
iv
-
List of Tables
Table 1.1: Decentralisation in Indonesia and other developing countries . . . 9Table 1.2: Summary of decentralisation studies in developing countries . . . 13Table 1.3: Research gaps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33Table 1.4: Data sources and samples used in the study . . . . . . . . . . . . 42Table 1.5: Decentralisation and well-being measures . . . . . . . . . . . . . . 49
Table 2.1: Description of perceived local public service performance determi-nants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
Table 2.2: Summary statistics of analytic sample . . . . . . . . . . . . . . . 74Table 2.3: Centroids correlation of public service index . . . . . . . . . . . . 79Table 2.4: Results of multilevel regression of local public service performance 84
Table 3.1: Summary statistics of analytic sample . . . . . . . . . . . . . . . 107Table 3.2: Centroid correlations of self-rated happiness . . . . . . . . . . . . 115Table 3.3: Results of multilevel regression of self-rated happiness . . . . . . . 117Table 3.4: Results of multilevel regression of self-rated happiness for cross
level interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
Table 4.1: Distribution of mothers social capital . . . . . . . . . . . . . . . . 139Table 4.2: Summary statistics of analytic sample . . . . . . . . . . . . . . . 142Table 4.3: Bivariate correlation of selected determinants . . . . . . . . . . . 153Table 4.4: Results of second stage regression . . . . . . . . . . . . . . . . . . 154Table 4.5: Results of first stage regression . . . . . . . . . . . . . . . . . . . 156
Table 5.1: Distribution of length of hospital stay . . . . . . . . . . . . . . . . 182Table 5.2: Summary statistics of analytic sample . . . . . . . . . . . . . . . 182Table 5.3: Centroids correlation of average length of hospital stay . . . . . . 191Table 5.4: Model selection results . . . . . . . . . . . . . . . . . . . . . . . . 192Table 5.5: Results of multilevel finite mixture negative binomial . . . . . . . 193
v
-
List of Figures
Figure 1.1: Map of Indonesia 2009 . . . . . . . . . . . . . . . . . . . . . . . 3Figure 1.2: Trend of government spending in Indonesia 1994-2007 . . . . . . 5Figure 1.3: Distribution of intergovernmental fiscal transfer in Indonesia be-
tween 2001 and 2009 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6Figure 1.4: Percentage of GDP allocated to public spending, 2008 . . . . . . 7Figure 1.5: Local direct elections in Indonesia 2005-2009 . . . . . . . . . . . 8Figure 1.6: A framework of the linkages between decentralisation, its condi-
tions and outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28Figure 1.7: Schematic research framework for this study . . . . . . . . . . . 53
Figure 2.1: Distribution of perceived local public service performance across120 local governments . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
Figure 2.2: Distribution of highest and lowest perceived service performance 79Figure 2.3: Bivariate correlation of selected local government determinants . 82
Figure 3.1: Distribution of average self-rated happiness across 262 local gov-ernments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
Figure 3.2: Five highest and lowest ranks of aggregate happiness . . . . . . . 115
Figure 4.1: Distribution of social groups across local governments . . . . . . 138Figure 4.2: Distribution of height for age across local governments . . . . . . 151Figure 4.3: Distribution of weight for age across local governments . . . . . . 152Figure 4.4: Distribution of height and weight for age by mothers social capital 152
Figure 5.1: Indonesia health spending from 1995 to 2008 . . . . . . . . . . . 176Figure 5.2: Transfer of health spending since decentralisation . . . . . . . . . 177Figure 5.3: Distribution of average length of hospital stay in 471 local govern-
ments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
vi
-
The University of ManchesterSujarwoto
PhD in Social ChangeEssays on decentralisation, public services and well-being in Indonesia
7th February 2013Abstract
Decentralisation has been viewed as one means for promoting citizen well-being bybringing public goods and services closer to local needs. Yet empirical evidence acrossdeveloping countries shows mixed results. This study aims to examine the associationbetween decentralisation, public services and well-being in decentralised Indonesia. Itargues that decentralisation enhances citizen well-being through improved local gov-ernment capacity better able to deliver public goods and services. With lack of localgovernment capacity and accountability, decentralisation decreases well-being.
This study contributes to existing research on decentralisation and well-being inthree ways. First, while most studies discuss decentralisation and well-being in a cross-country context, this examines the relationship in a cross-local government context, andspecifically within a developing country. Second, while most decentralisation studiesfocus on objective measures of well-being, this study uses both a subjective measure (i.e.happiness and citizen satisfaction with public services) and an objective measure (i.e.child health and healthcare demand). Third, while most studies use either aggregate orindividual analyses to examine the effect of decentralisation on well-being, this studyuses multilevel analysis to examine the effect of local government determinants onindividual well-being.
This study uses unique datasets which combine individual and household level dataand local government data. Individual and household level data is taken from the Gov-ernance and Decentralisation Survey (GDS) 2006, the Indonesian Family Life Survey(IFLS) 2007, and the Indonesia socio-economic survey (Susenas) 2009. Local govern-ment level data comes from the national village census (Podes) 2006-2008, the nationalelection database 2004-2007, the national health database 2009, local development bud-get and expenditure information 2004-2008, and the consumer prices indices 2004-2009.
The main findings show that well-being among Indonesians varies across local gov-ernments, and that disparities in both public services and well-being appear betweenmore developed and less developed regions. Well-being is not only associated with indi-vidual and household determinants, but also with local government determinants. Theresults are consistent, namely, that variation in well-being is associated with the ca-pacity of local governments to deliver public goods and services. Citizens report beinghappier and more satisfied when local governments are able to provide better publicgoods and services for them (i.e. able to spend more of their budget on providing publicservices). In contrast, well-being decreases in the face of local corruption and of weakcapacity to govern. These findings suggest that improving local government capacity toprovide effective policies and good public services is vital to improve citizen well-beingin decentralised Indonesia.
vii
-
Declaration
No portion of the work referred to in the thesis has been submitted in support of anapplication for another degree or qualification of this or any other university or otherinstitute of learning.
Copyright
1. The author of this thesis (including any appendices and/or schedules to this thesis)owns certain copyright or related rights in it (the Copyright) and s/he has givenThe University of Manchester certain rights to use such Copyright, including foradministrative purposes.
2. Copies of this thesis, either in full or in extracts and whether in hard or electroniccopy, may be made only in accordance with the Copyright, Designs and PatentsAct 1988 (as amended) and regulations issued under it or, where appropriate, inaccordance with licensing agreements which the University has from time to time.This page must form part of any such copies made.
3. The ownership of certain Copyright, patents, designs, trade marks and other intel-lectual property (the Intellectual Property) and any reproductions of copyrightworks in the thesis, for example graphs and tables (Reproductions), which maybe described in this thesis, may not be owned by the author and may be ownedby third parties. Such Intellectual Property and Reproductions cannot and mustnot be made available for use without the prior written permission of the owner(s)of the relevant Intellectual Property and/or Reproductions.
4. Further information on the conditions under which disclosure, publication andcommercialisation of this thesis, the Copyright and any Intellectual Propertyand/or Reproductions described in it may take place is available in the UniversityIP Policy (see www.campus.manchester.ac.uk/medialibrary/policies/intellectual-property.pdf), in any relevant Thesis restriction declarations deposited in the Uni-versity Library, The University Librarys regulations (see www.manchester.ac.uk/library/aboutus/regulations) and in The Universitys policy on presentation ofTheses.
viii
-
Acknowledgements
Over the last four years, I have never been short of other people to guide me in writingmy thesis, to spur my development as a scientist, or to help me relax after (or during)a day of hard work. For their help in starting, continuing, and finishing this thesis, Ifeel very much obliged to express my special gratitude to some of them.
First and foremost of the long list of those to whom I must express my gratitude aremy supervisors, Gindo Tampubolon and Ronnie Ramlogan for their excellent supervi-sion and persistent support throughout my study of the doctoral programme. Gindohas given me supervision in many respects. He has taught me how good quantitativeresearch is done and has given me enormous assistance in dealing with survey datausing statistical packages such as STATA and MPLUS. Overall, I am thankful for allhis contributions of time, ideas and techniques. Thank you for your advice and sug-gestions, and for reading my papers, drafts, and chapters over and over again. Yoursupervision has certainly made me a better researcher.
The members of Institute for Social Change at the University of Manchester havecontributed to the completion of this work. I am especially grateful to David Cutts,Edward Fieldhouse and Jennifer Birchall, to name just a few, for their assistance,comments and discussions on the work at various stages. To my colleagues Adi, Devi,Citra and Asri, what moments of great joy, enormous stress, and huge ambition havewe shared the last couple of years. It is hard to imagine how I would have managedto complete my thesis without your suggestions and companionship. Having goodcolleagues is very helpful in writing a PhD thesis, as is having nice friends.
I must also express my gratitude to Indonesian government, for the funding of mydoctoral programme. It would have been impossible to have come to a smooth comple-tion of the research project without its grant assistance. A very special debt is owedto Mbak Ria in University of Brawijaya for her work in financial administration andlogistic assistance. I also thank the Harvard Kennedy School Indonesia for giving mean opportunity to spend two months at the Harvard Kennedy School of Government asa visiting scholar. Strolling around Boston and Cambridge was not only a life changingexperience in itself, my stay at Harvard has also truly propelled my career as a socialscientist.
Finally, I would like to thank my family members for all their love: for my parentswho choose me with care and love; for my wife, Tri Yumarni, and my son, Danish, fortheir understanding, support and encouragement. A special acknowledgment goes tomy late father-in-law, Muji Pranoto, who had lived with my core family for years andhad spent much time taking care of my son but very sorrowfully passed away beforethe completion of the doctoral thesis.
ix
-
Abbreviations and Glossary
AIC Akaike Information Criterion.arisan or binda a form of rotating savings and credit association in Indone-
sian culture.Askes (Asuransi kesehatan) health insurance for Indonesian civil
servant.Askeskin (Asuransi kesehatan masyarakat miskin) health insurance
for poor people in Indonesia.BIC Bayesian Information Criterion.BPS (Badan Pusat Statistik) The Indonesian governments Cen-
tral Bureau of Statistics.Bupati/walikota mayor or head of local government in Indonesia.CFI Comparative Fix Index.CPI Consumer Price Index.DPRD (Dewan Perwakilan Rakyat Daerah) local government par-
liament body in Indonesia.GDP Gross Domestic Product.GDS Governance Decentralisation Survey.GLLAMM generalised linear latent and mixed models.Golkar (Golongan Karya) the ruling political party during
Suhartos autocratic regime.Gotong royong Generalised reciprocity both in rural and urban areas.ICC Intra Class Correlation.IDR Indonesia Rupiah.IFLS Indonesian Family Life Survey.IRT Item Response Theory.Jamkesmas (Jaminan kesehatan masyarakat) Community health protec-
tion scheme in Indonesia.Kabupaten/kota district or local government in Indonesias term.Karangtaruna youth group organisation in Indonesia culture.Kelompok pengajian muslim religious group in Indonesia culture.Kelompok kebaktian christian/catholic religious group in Indonesia culture.Kelompok wanita women groups in Indonesia.Kerja bakti a form of voluntary labour in Indonesia culture.Koperasi a form of business organisation which is organised based on
common interest within communities.
1
x
-
MIMIC Multiple Indicators Multiple Causes.MoH The Ministry of Health.Musyawarah a form of discussions and sharing of ideas or opinions among
community members to solve problems within community.Ndihma Ekonomike the name of Albanias poverty-alleviation program in the
early 1990s.OECD Organisation for Economic Co-operation and Development.Orde Baru New Order regime in Indonesia from 1965 to 1998.Panchayats the oldest system of local government in the Indian subcon-
tinent. The word panchayat literally means assembly (ayat)of five (panch) wise and respected elders chosen and acceptedby the local community.
PDI (Partai Demokrasi Indonesia Perjuangan) Indonesia Demo-cratic Party.
PPP (Partai Persatuan Pembangunan) Indonesia United Devel-opment Party.
PKK (Pendidikan Kesejahteraan Keluarga) a group of women inIndonesia, guided by the idea that improving family welfareby providing village women with improved basic skills buildsthe foundation for a better society.
Persatuan kematian funeral group in Indonesia culture.Pilkadal (Pemilihan kepala daerah langsung) direct local government
election in Indonesia.Podes (Potensi Desa) Village Potential Census.Propinsi Province in Indonesia government structure.PT Askes Indonsia state owned enterprise engaged in health insurance.Puskesmas (Pusat kesehatan masyarakat) Community health centre.RMSEA Root-Mean Square Error of Approximation.rukun warga neighbourhood association in Indonesia culture.SIKD (Sistem Informasi Keuangan Daerah) Indonesian develop-
ment budget and expenditure information system.Susenas Indonesian Socio Economic Survey.UNDP United Nation Development Programme.USD United State Dollars.WHO-CDC World Health Organisation, Centers for Disease Control and
Prevention.WLSV weighted least-square with mean and variance adjustment.
1xi
-
Chapter 1
Introduction
In her book Going local: decentralisation, democratisation, and the promise of good
governance (2007), Merilee S. Grindle writes that in the early twenty-first century the
quality of local governments presents new relevance to the well-being of people across
developing countries. Although having long played some role in the management of local
affairs, they have now been given many new responsibilities, provided with increased
resources, and allowed greater autonomy to decide local policies and services aimed at
promoting citizen well-being. Decentralisation of fiscal, administrative, and political
responsibilities has transformed local government into the central actor in delivering
local public services and enhancing citizen well-being. Bardhan and Mookherjee (2006)
explain that this phenomenon is geographically widespread, from Latin America, Africa,
and Asia to Eastern Europe. The earliest changes were initiated in the 1970s, picked
up momentum in the 1980s, and accelerated after the 1990s (World Bank, 2008a).
Focusing on this increasing role of local government, this study aims to examine
the consequences of decentralisation reform on public services and citizen well-being,
and to find answers to the following: When local governments are charged with new
responsibilities and provided with new resources, to what extent can they promote
public services and citizen well-being? Why are some local governments more effective
than others in promoting well-being? What are the implications of decentralisation
1
-
reform for local public service performance and citizen well-being? To answer these
questions, I refer to the decentralisation experience of Indonesia.
Throughout this study, decentralisation refers to the formal and informal mecha-
nisms and rules that allocate political authority and resources downward to the various
levels of government (Grindle, 2007). My main focus is the decentralisation of political
authority and resources to local governments below state or provincial level. Local
government thus refers to district level government. This distinction is meaningful in
countries such as Indonesia which are of a large enough size that provincial governments
form an intermediate tier between national and local governments1. This distinguishes
this enquiry from prior studies which provide a comparative perspective of how ad-
ministrative decentralisation affects states or provinces within federated systems across
developed and developing countries (see for instance Rodden et al., 2003). In such con-
texts, my main interest is the devolution of responsibilities and resources from central
to local government2.
1.1 Decentralisation in Indonesia: from centralistic
government to radical decentralisation
Made up of over 13,000 islands covering 8.8 million square kilometres (a distance roughly
equal to that between London and Baghdad), Indonesia is home to over 228 million
people. The largest islands are Sumatra, Java (the most populous), Bali, Kalimantan
1As in China and India, a typical province in Indonesia is larger (in terms of size of population)than most countries in the world, and so decentralisation in the sense of devolution of power to theprovincial state government may still keep power over people somewhat centralised.
2In countries with a long history of centralised control such as Russia, China or India, by decen-tralisation public administrators often mean the dispersion of some responsibilities to regional branchoffices at the local level. For the purpose of this study, I distinguish decentralisation in the sense ofdevolution of political decision-making power from such mere administrative delegation of functions ofcentral government to their local branches.
2
-
(Indonesias part of Borneo), Sulawesi (Celebes), the Nusa Tenggara islands, and Papua,
the western part of New Guinea. Its neighbour to the north is Malaysia and to the east
is Papua New Guinea.
Figure 1.1 : Map of Indonesia 2009
Sumatra
Java
Kalimantan SulawesiPapua
Nusa tenggara
Jakarta
BaliSource: author calculated based on Indonesia shape files 2007
Indonesia is a unitary state with a presidential system. As such, provincial and
local governments are the creation of central government. Hence, local governments
generally act based on authority delegated to them by legislation or by the directives of
central government. From 1965 until 1998, the country was under a centralised govern-
ment. During this period, local governments were controlled by the highly centralistic
government of the New Order. They had little right and no authority to decide local
laws and preferences, as all policy were decided by central government agencies. Any
notion of local development was non-existent, since all development initiatives came
from central government in Jakarta. This centralistic political order resulted in not
only a lack of democracy but also a high dependency of local governments on central
government3.
3See Ricklefs (2001) for elaboration of the history and historiography of the New Order regime inIndonesia. See Turner et al. (2003) for an elaboration on political system and decentralisation inIndonesia before and after 1999.
3
-
A number of attempts were made by Indonesian government since its independence
to enhance local government autonomy, but these efforts were often side-tracked by po-
litical considerations4. However, the economic and political crisis in Indonesia in 1998
triggered radical decentralisation reform in 19995. This reform has changed Indonesias
political system from one of highly centralised to highly decentralised government, a
radical reform which has devolved all central government functions to local government
level, except for national defence, international relations, justice, police, monetary pol-
icy, development planning, religion, and finance, which are retained at the centre. Local
governments are now obliged to perform a set of key functions. These include the pro-
vision of health, education, environmental and infrastructure services, and are designed
to perform any functions not explicitly reserved for either the centre or the provinces
(World Bank, 2003).
This radical decentralisation also brings substantial resources to local governments.
The bill on intergovernmental fiscal transfer set out a revised framework for the coun-
try (Kaiser and Hofman, 2003; World Bank, 2003). This replaces the old system of
earmarked grants and centralised control over local finances with one in which a gen-
eral allocation fund or dana alokasi umum and shared taxes provide the bulk of local
revenue6. Figure 1.2 describes government spending trends in Indonesia between 2001
4See Devas (1989), Kaiser and Hofman (2003) and Turner (2003) for further discussion of theseissues.
5The Asian financial crisis that began to affect Indonesia in mid-1997 became an economic andpolitical crisis. Its effects were severe: by November 1997, rapid currency depreciation had seen publicdebt reach USD$60 billion, imposing a severe strain on the governments budget. In 1998, actual GDPcontracted by 13.7%. The economy reached its lowest point in mid-1999 and actual GDP growth for theyear was 0.3%. Inflation reached 77% in 1998 but slowed to 2% in 1999. The rupiah, which had been inthe range of IDR2,600 to USD1 at the start of August 1997 fell to IDR11,000 to USD1 by January 1998,with spot rates around IDR15,000 for brief periods during the first half of 1998. It returned to aroundIDR8,000 to USD1 at the end of 1998 and has generally traded in the IDR8,000-10,000 to USD1 rangeever since, with fluctuations that are relatively predictable and gradual (World Bank, 2003). Thiseconomic crisis de-legitimised the regime. Widespread social unrest erupted, and after the regime fellin May 1998, a number of fundamental changes in the Indonesian political and administration systemtook place. One of the most fundamental was the radical decentralisation formulated in 1999.
6The general allocation fund or dana alokasi umum is a grant through which equalisation paymentsare made, and is designed to ensure that local governments have adequate and similar capacities. By
4
-
and 2007. Decentralisaton almost doubled the regional share of government spending,
from about 17% in 1994 to over 30% of total government expenditure after 2001. By
2007, this share had grown to over 40% (Figure 1.2).
Figure 1.2 : Trend of government spending in Indonesia 1994-2007
Con
stan
t 200
0 pr
ices
, Rp.
Tril
lion
Decentralisation bills
Transfer to local governments
Central government spending
0
50
100
150
0
50
100
150
1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007
Subsidies
Source: author calculated based on World Bank (2008c)
The distribution of intergovernmental transfer since decentralisation has also sig-
nificantly increased. Most local governments have received more than double since the
beginning of decentralisation. Some local governments (Papua and Aceh) received an
even larger transfer in 2009, about triple that received in 2001 (Figure 1.3).
law, at least 26% of net domestic revenue has to be transferred to local governments through this fund(Ministry of Finance, 2009). The general allocation fund is the most important tool of the transfersystem, financing more than 80% of local government expenditure (World Bank, 2008). The schemehas substantially contributed to closing fiscal gaps in Indonesias poorer regions.
5
-
Figure 1.3 : Distribution of intergovernmental fiscal transfer in Indonesia between
2001 and 2009
Source: author calculated based on Indonesian local government development budget and expenditure information (SIKD) 20012009 data
> 200%150200%100150%< 100%
Decentralisation also encourages local governments to increase their public spending.
In 1998, the national public spending average was about 23% of GDP, increasing to
about 35% of GDP in 2008. This increase is mainly due to the significant growth
in public spending at local government level. Figure 1.4 presents the share of public
spending in terms of GDP for a selected range of local governments in 2008 (World
Bank, 2008c). For example, the local governments of Kepulauan Aru and Paniai (in
Papua province) allocated about 90% of GDP for public spending in 2008, a proportion
higher than that of the United Kingdom, France, Denmark and Sweden. However, there
is also evidence that some local governments are lagging behind in their spending. For
example, the local government of Kupang in Nusa Tenggara Timur province has spent
below 20% of its GDP for public spending, a proportion comparable to Bangladesh and
Afghanistan over the same period.
6
-
Figure 1.4 : Percentage of GDP allocated to public spending, 2008
Kab. Kepulauan Aru
Kab. Kutai Barat
Kab. Kayong Utara
Kab. Gianyar
Kab. Paniai
Kab. Maros
Kab. Buton Utara
Kota Kupang
Iraq
FranceDenmark
Indonesia
Sweden
BangladeshAfghanistan
Kab. Yapen Waropen
Kab. Wakatobi Kab. Simelue
United Kingdom
0
20
40
60
80
100Pu
blic
spen
ding
as
% o
f GDP
Local governments Countries Source: author calculated based on SIKD 2008 and World Development Indicators 2008
Decentralisation not only brings change to the fiscal transfer system but also to the
local political system, shifting the balance from a largely top-down form of government
to one of local political accountability (Erb and Sulistiyanto, 2009). Between 1999 and
2004, local parliaments were directly elected. However, mayors or bupati/walikota were
indirectly elected. Under this system, local parliaments elect mayors, and can dismiss
them through rejection of their annual accountability speech. Further powers rested
in the local parliaments or Dewan Perwakilan Rakyat Daerah include approval of the
annual local budget and of local laws and regulations.
Concerns around indirect political accountability triggered a second wave of local
government electoral reform in terms of direct elections or pilkada langsung (Erb and
Sulistiyanto, 2009). Starting in 2005, mayors were elected along direct (and more pres-
idential) rather than parliamentary lines. This reform made the mayor more directly
accountable to the people. Legislation regarding local government elections also stip-
ulated that the mayor should (1) manage their jurisdiction along guidelines laid down
7
-
by local parliament, (2) implement local laws, including budget law, (3) present annual
accountability reports to the local parliament and central government, and (4) provide
information to citizens on the governments performance.
Based on the local government election bill or pilkada amendment, central govern-
ment decided to conduct the first batch of direct local government elections in June
2005 in those local governments where the local parliament speakers had come to the
end of their tenure. By 2009 all local governments had directly elected mayors (Figure
1.5).
Figure 1.5: Local direct elections in Indonesia 2005-2009
Direct election
Indirect election
0
100
200
300
400
500
0
100
200
300
400
500
2005 2006 2007 2008 2009
Nu
mb
er
of
loc
al
go
ve
rnm
en
ts
Source: author calculated based on Indonesian local government election data (The Ministry of Home Affairs, 2010)
Radical decentralisation reform in Indonesia has unique characteristics compared
to that which has taken place in Africa, Asia and South America countries. Bardhan
and Mookherjee (2006) for example identify three forms of decentralisation in several
countries across these continents (including Indonesia); they base their classification
8
-
on motive for decentralisation, and the nature of each countrys political and economic
decentralisation. Table 1.1 presents these categories and the countries within each.
Table 1.1: Decentralisation in Indonesia and other developing countries
Countries Motives Nature of political Nature of economicdecentralisation decentralisation
1 2 3 4 5 6 7 8Indonesia (2001) Yes Yes Yes Yes Yes Yes Yes YesMalaysia No No No Yes No No No NoPhilippines Yes No No Yes Yes Yes No YesChina (1978) No No No Yes No Yes No NoIndia (1994) No No No No Yes No No NoPakistan (2000) No No Yes Yes No No No NoBrazil (1988) Yes Yes No No Yes Yes No YesUganda (1987-2004) No No No No Yes No No NoSouth Africa (2000) Yes Yes No Yes Yes Yes Yes NoNote: 1: transition to democracy, 2: change in political party power, 3:radical approach, 4: uniform, 5: direct election, 6: financial autonomy, 7: sizeof fiscal transfer >30%, 8: responsibility of most public functions
1
In 2001, Indonesia embarked on a programme of comprehensive decentralisation that
was implemented simultaneously throughout the entire country within a short period
of time. The reform was introduced as part of the transition to democracy following
the collapse of the autocratic regime and subsequent weakening of central government
power in the face of strengthening regional interests. The previous centralised regime
had been associated with high levels of inequality in terms of interregional allocation of
funds, poor service delivery, and high levels of corruption. External pressures (such as
international donor agenda and ideological considerations) played a relatively small role
in the transition (Kaiser and Hofman, 2003; Kaiser et al., 2006; Kruse et al., 2012; Erb
and Sulistiyanto, 2009; World Bank, 2003). This motivation behind decentralisation is
noticeably different from that of other countries, such as China, Malaysia, India, Pak-
istan and Uganda. China and Malaysia for example are characterised by a long tradition
of centralised political power monopolised by a single ruler or one party. Decentral-
9
-
isation in these countries represents more a combination of more extensive economic
decentralisation with the near absence of local democracy (Bardhan and Mookherjee,
2006). Pakistan represents a different extreme, where state bureaucrats have become
subordinate to elected district officials, at least at the de jure level. China represents
the other extreme, with unelected local government officials subordinate to those at
more central level.
Decentralisation in Indonesia was both political and economic in nature. Local
governments were provided with political authority independent of the centre, regular
direct local mayoral and parliamentary elections were conducted, and responsibility
over most local social services (including primary education, health, infrastructure, and
welfare) was devolved from the centre. This mode of decentralisation and the degree of
public function transfer differ from those which occurred in Malaysia and China (where
there is no regular direct election for mayors and parliaments) and Malaysia and India
(where local governments have no jurisdiction over the administration of basic services,
such as primary education and health). The common feature of decentralisation in
these countries is that local governments remain essentially subordinate to the dominant
power at national level, are elected (if at all) in a partyless election, have no independent
(or at least limited) constitutional authority or protection, and can be dismissed or
neutralised by higher levels of government. As a consequent, local governments typically
have limited authority to formulate and to implement policies or programmes related
to most public functions (Bardhan and Mookherjee, 2006). In Malaysia and China for
example local governments have little authority over the personnel they hire, especially
with respect to pay and service conditions (and salaries represent a significant part of
local government costs).
Also in contrast to many other decentralised developing countries, the size of fiscal
decentralisation in Indonesia is specific in terms of devolution of fiscal resources and
10
-
generation of revenue power. Local governments here have received substantial fund
transfers - more than thirty per cent of central government budget (World Bank, 2008),
substantially more than other Asian countries such as Philippines and Malaysia (where
it is less than twenty per cent) and certain countries in South America. Furthermore,
intergovernmental transfers in Indonesia are largely formula-bound, greatly reducing
the scope for interference at upper levels of government (Kaiser et al., 2006). Local
governments thus are able to exercise relatively strong power and have the freedom
to generate local, own-source revenue from taxes. In most other developing countries,
spending mandates have not been comfortably covered by devolved funds, giving rise
to considerable fiscal strain. In Malaysia for example, local governments have no right
to generate taxes.
It is widely agreed that radical decentralisation in Indonesia has provided local
governments with increased authority to devise their own policies and programmes (see
for example Kaiser and Hofman (2003); Kaiser et al. (2006); Kruse et al. (2012);
Erb and Sulistiyanto (2009); World Bank (2003)). This has widely increased their role
in managing public services for the benefit of local citizens. Local governments also
have better access to sources of revenue. However, although research has discussed the
consequences of decentralisation in Indonesia in terms of various development outcomes,
relatively few studies have examined the relationship between this decentralisation and
the well-being of its citizens. The next section reviews the literature on decentralisation
in developing countries and outlines the research gaps which this study attempts to
address.
11
-
1.2 Assessing decentralisation literature on devel-
oping countries
This section outlines some of the attempts that have been made to empirically evaluate
the impact of decentralisation on the delivery of public services and on well-being in
developing countries. Even though decentralisation is continuing to be implemented in
many of these countries, quantitative evidence on its impact is rather scarce. There
are a number of scattered studies that I will try to arrange in terms of the nature of
empirical methodology followed. Overall, the conclusions are mixed and the relationship
they attempt to establish is at times elusive. Table 1.2 provides a summary of some
of the prominent studies in this area from the last two decades which show positive,
negative and inconclusive relationships between decentralisation and public services and
various measures of well-being across developing countries.
12
-
Tab
le1.
2:Sum
mar
yof
dec
entr
alis
atio
nst
udie
sin
dev
elop
ing
countr
ies
Unit
ofan
alys
is(l
ocat
ion)
met
hod
imp
orta
ntob
serv
atio
ns/
consi
der
atio
ns
stre
ngt
hs
and
wea
knes
sof
the
study
1.Positiveassociation
One
munic
ipal
ity,
Bra
zil
Qual
itat
ive
study
Dec
entr
alis
atio
nim
pro
ved
par
tici
pat
ory
budge
ting
and
pla
nnin
gin
public
serv
ices
(San
tos,
1998
)R
ich
anal
ysis
onth
ehis
tory
ofpar
tici
pat
ory
budge
t-in
gan
dpla
nnin
gin
public
serv
ices
.H
owev
er,
the
resu
lts
isve
rylim
ited
for
gener
alis
atio
n.
Munic
ipal
itie
s,B
oliv
iaP
rinci
pal
com
po-
nen
tan
alys
isD
ecen
tral
isat
ion
impro
ved
consi
sten
cyof
public
ser-
vice
sw
ith
loca
lpre
fere
nce
san
dqu
alit
yof
acce
ss(F
aguet
,20
01)
Use
vari
ous
mea
sure
ofpublic
serv
ices
per
form
ance
,but
the
study
only
info
rms
asso
ciat
ion
rath
erth
anca
usa
leff
ect.
Hou
sehol
ds,
Alb
ania
Tob
itre
gres
sion
wit
hfixe
deff
ect
Mod
est
gain
sin
targ
etin
geffi
cien
cyan
dco
st-
effec
tive
nes
sfo
llow
ing
dec
entr
alis
atio
n(A
lder
man
,19
98)
Use
man
yye
ars
ofdat
aan
dco
ntro
lled
for
unob
serv
edhet
erog
enei
ties
inlo
cal
gove
rnm
ent
leve
l.H
owev
er,
the
study
do
not
info
rmp
oten
tial
ofel
ite
captu
rein
the
pro
gram
me.
234
villag
es,
India
Fix
edeff
ect
wit
hIV
regr
essi
onA
nin
crea
sein
the
dem
ogra
phic
wei
ght
ofth
ela
ndle
sshou
sehol
ds
ina
villag
eunder
dem
ocra
tic
dec
entr
ali-
sati
onhas
ap
osit
ive
effec
ton
allo
cati
onof
public
re-
sourc
esto
road
const
ruct
ion
and
aneg
ativ
eeff
ect
onth
atto
irri
gati
onfa
ciliti
es(F
oste
ran
dR
osen
zwei
g,20
01)
Use
pan
eldat
aan
din
form
causa
leff
ect,
but
this
study
do
not
info
rmdat
aca
ptu
ring
elit
eca
ptu
reon
the
pro
gram
me.
83vi
llag
es,
India
Tob
itre
gres
sion
Dec
entr
alis
edm
anag
emen
tth
rough
Pan
chayat
ad-
vance
dp
over
tyal
levi
atio
ngo
als
inW
est
Ben
gal
(Bar
dhan
and
Moo
kher
jee,
2003
)
Use
spec
ific
case
ofdec
entr
alis
atio
nin
Pan
chayat
syst
emac
ross
India
nvi
llag
es.
How
ever
,th
est
udy
do
not
hav
edat
aab
out
the
dis
trib
uti
onof
pro
gram
ben
efits
acro
ssso
cio-
econ
omic
grou
ps
and
geog
raphy
.C
omm
unit
ies,
Ban
glad
esh
Tob
itre
gres
sion
Dec
entr
alis
edfo
odfo
red
uca
tion
pro
gram
inB
angl
ades
hw
asm
ildly
pro
-poo
r(G
alas
soan
dR
aval
-lion
,20
01)
Bas
edon
smal
lsca
lear
eas
(com
munit
ies)
toev
aluat
eth
eim
pac
tof
dec
entr
alis
atio
nin
par
ticu
lar
pro
gram
.T
he
study
how
ever
do
not
captu
reco
nsi
stdat
aw
hic
hin
form
pot
enti
alel
ite
captu
rew
ithin
the
pro
gram
.Sch
ools
,A
rgen
tina
IVre
gres
sion
Dec
entr
alis
atio
nof
educa
tion
led
toim
pro
vem
ent
insc
hoo
lac
hie
vem
ent
scor
es(E
skel
and
and
Film
er,
2002
)
Usi
ng
stro
ng
stat
isti
cal
met
hod
olog
yto
esta
blish
causa
leff
ect
wit
hin
spec
ific
pro
gram
me.
How
ever
,th
isst
udy
only
use
smal
lnu
mb
erof
schoo
ls.
Sch
ools
,N
icar
agua
Fix
edeff
ect
regr
es-
sion
Dec
entr
alis
edm
anag
emen
tof
schoo
lsle
dto
impro
ve-
men
tin
achie
vem
ent
ofb
ette
rp
erfo
rman
ce(K
ing
and
Ozl
er,
1998
)
This
study
eval
uat
eth
eeff
ect
ofdec
entr
alis
atio
nw
ithin
spec
ific
pro
gram
.B
ut,
iton
lyuse
smal
lnu
m-
ber
ofsc
hoo
lsin
the
anal
ysis
.T
wo
count
ries
,In
dia
and
Sou
thK
orea
Qual
itat
ive
study
Ove
r-ce
ntra
lise
dto
pdow
nm
anag
emen
tac
com
pa-
nie
dw
eak
mon
itor
ing
cont
ribute
dto
corr
upti
onan
dp
oor
del
iver
yp
erfo
rman
cefo
rca
nal
irri
gati
onin
In-
dia
(Wad
e,19
97)
This
study
pro
vides
rich
info
rmat
ion
onth
ehis
tory
and
pro
cess
ofdec
entr
alis
atio
nin
thes
eco
unt
ries
.H
owev
er,
the
lim
ited
case
sm
ean
that
this
study
can-
not
be
gener
alis
edto
other
count
ries
.16
munic
ipal
itie
s,C
olom
bia
Qual
itat
ive
study
Com
pet
itio
nfo
rp
olit
ical
office
incr
ease
sre
spon
sibil-
ity
and
faci
lita
tes
lead
ersh
ipin
nov
atio
n(F
iszb
ein,
1997
)
This
study
pro
vides
rich
info
rmat
ion
abou
tth
ero
leof
pol
itic
alco
mp
etit
ion
and
lead
ersh
ipw
ithin
dec
en-
tral
isat
ion
inC
olom
bia
.H
owev
er,
itis
lim
ited
only
for
smal
lnu
mb
erof
munic
ipal
itie
s.1,
808
firm
s,In
don
esia
IVR
egre
ssio
nA
dm
inis
trat
ive
dec
entr
alis
atio
nle
dto
low
erco
rrup-
tion
asfirm
sre
loca
ted
toar
eas
wit
hlo
wer
bri
bes
(Hen
der
son
and
Kunco
ro,
2004
)
Use
larg
enu
mb
erof
firm
san
dso
phis
tica
ted
stat
is-
tica
lm
ethod
olog
y.H
owev
er,
iton
lyuse
dcr
oss
sec-
tion
aldat
a.20
7m
unic
ipal
itie
s,In
don
esia
IVR
egre
ssio
nP
osit
ive
effec
tof
dec
entr
alis
edhea
lth
spen
din
gfo
rhea
lth
care
use
ofp
oor
peo
ple
(Kru
seet
al.,
2012
)U
selo
cal
gove
rnm
ents
asunit
ofan
alys
isan
dso
-phis
tica
ted
met
hod
olog
y,but
ituse
dcr
oss
sect
ional
dat
a.
1
13
-
Unit
ofan
alysi
s(l
oca
tion
)m
ethod
imp
orta
nt
obse
rvat
ions/
consi
der
atio
ns
stre
ngt
hs
and
wea
knes
sof
the
study
Hea
lth
centr
esan
dhos
pit
al,
Zam
bia
Tim
ese
ries
Dec
entr
alis
atio
nin
hea
lth
sect
orw
ith
loca
lco
ntr
olof
reso
urc
esco
uld
be
anal
tern
ativ
eto
the
trad
itio
nal
vert
ical
dis
ease
pro
gram
mes
appro
ach
for
pri
orit
yin
-te
rven
tion
s(B
las
and
Lim
bam
bal
a,20
01)
This
study
exam
ines
dec
entr
alis
atio
nw
ithin
spec
ific
pro
gram
me
and
use
tim
ese
ries
ofdat
a.H
owev
er,
iton
lypro
vid
esin
-dep
than
alysi
son
the
impac
tof
the
pro
gram
me.
Munic
ipal
ity,
Bra
zil
Qual
itat
ive
study
Dec
entr
alis
atio
nre
form
thro
ugh
par
tici
pat
ory
bud-
geti
ng
pro
cess
can
incr
ease
wel
fare
oflo
cal
citi
zens
(Bai
occ
hi,
2001
)
This
study
pro
vid
esri
chin
form
atio
nab
out
par
tic-
ipat
ory
budge
ting
pro
cess
inso
me
munic
ipal
itie
s.H
owev
er,
the
resu
lts
are
lim
ited
tob
ege
ner
alis
edfo
rot
her
pla
ces.
26pro
vin
ces,
Chin
apan
eldat
aan
alysi
sM
ore
dec
entr
alis
edpro
vin
ces
per
form
bet
ter
wit
hre
-sp
ect
tohea
lth
outc
omes
ifa
funct
ionin
gtr
ansf
ersy
stem
esta
blish
edb
etw
een
the
pro
vin
cean
dco
unty
leve
lsan
dco
unty
gove
rnm
ents
ow
nfisc
alca
pac
ity
stre
ngt
hen
ed(U
chim
ura
and
Ju
ttin
g,20
07)
Use
pan
eldat
aan
dco
ntr
olunob
serv
edhet
ero-
genei
ties
,but
itis
lim
ited
onin
dic
ator
tom
easu
refisc
aldec
entr
alis
atio
n.
20co
untr
ies
Pan
eldat
aan
alysi
sA
sign
ifica
nt
pos
itiv
eeff
ect
ofex
pen
dit
ure
dec
entr
al-
isat
ion
onp
erca
pit
ain
fras
truct
ure
del
iver
ies
(Es-
tach
ean
dSin
ha,
1995
)
Use
pan
eldat
aan
alysi
s,but
only
use
lim
ited
num
ber
ofco
untr
ies.
42co
untr
ies
OL
SD
ecen
tral
ised
road
mai
nte
nan
ceim
pro
ved
road
con-
dit
ions
(Wor
ldB
ank,
1994
).T
his
study
exam
ines
the
effec
tof
dec
entr
alis
atio
non
spec
ific
pro
gram
me.
Ital
souse
sre
lati
vely
larg
enum
ber
ofco
untr
ies.
How
ever
,it
does
not
esta
blish
causa
leff
ect
onth
edec
entr
alis
atio
npro
gram
me.
80co
untr
ies
IVre
gres
sion
Dec
entr
alis
atio
nco
ntr
ibute
dto
impro
vedel
iver
yof
public
goods
and
serv
ices
(Huth
eran
dShah
,19
95)
Use
larg
enum
ber
ofco
untr
ies
and
use
robust
stat
is-
tica
lm
ethod
toes
tablish
causa
leff
ect.
How
ever
,th
isst
udy
islim
ited
wit
hit
cros
sse
ctio
nal
des
ign.
4dev
elop
ing
countr
ies
Qual
itat
ive
study
Dec
entr
alis
atio
nle
dto
enhan
ced
tran
spar
ency
and
reduce
din
ciden
ceof
corr
upti
on(C
rook
and
Man
or,
2000
)
This
study
pro
vid
esri
chin
form
atio
nab
out
pot
en-
tial
sas
wel
las
chal
lenge
sof
dec
entr
alis
atio
nas
am
ean
for
impro
vin
gtr
ansp
aren
cyan
dco
rrupti
onin
thes
eco
untr
ies.
How
ever
,it
lim
its
for
gener
alis
atio
nto
other
conte
xts
.78
countr
ies
IVR
egre
ssio
nT
axdec
entr
alis
atio
nw
asp
osit
ivel
yas
soci
ated
wit
him
pro
ved
qual
ity
ofgo
vern
ance
(De
Mel
loan
dB
aren
stei
n,
2001
)
This
study
use
sla
rge
num
ber
ofco
untr
ies
and
robust
stat
isti
cal
met
hodol
ogy,
but
itlim
its
onth
em
easu
reof
qual
ity
ofgo
vern
ance
take
into
acco
unt.
59co
untr
ies
IVR
egre
ssio
nA
sign
ifica
nt
neg
ativ
eas
soci
atio
nb
etw
een
exp
endi-
ture
dec
entr
alis
atio
nan
dco
rrupti
onm
easu
res
(Fis
-m
anan
dG
atti
,20
02)
This
study
use
sla
rge
num
ber
ofco
untr
ies
usi
ng
ro-
bust
stat
isti
cal
met
hodol
ogy.
How
ever
,it
lim
its
onth
em
easu
reof
corr
upti
onuse
d.
30co
untr
ies
OL
SD
ecen
tral
isat
ion
supp
orts
grea
ter
acco
unta
bilit
yin
the
public
sect
oran
dre
duce
dco
rrupti
on(G
urg
ur
and
Shah
,20
02)
This
pro
vid
essp
ecifi
can
alysi
son
pol
itic
aldec
entr
al-
isat
ion
and
the
role
ofac
counta
bilit
y.H
owev
er,
itis
only
bas
edon
cros
sse
ctio
nal
dat
aan
dlim
ited
num
-b
erof
countr
ies
and
stat
isti
cal
met
hodol
ogy.
43dev
elop
ing
countr
ies
IVR
egre
ssio
nP
ublic
spen
din
gm
atte
rsfo
rin
crea
sing
hea
lth
out-
com
esan
dpri
mar
yed
uca
tion
atta
inm
ent
inco
un-
trie
sw
ith
good
gove
rnan
ce(R
ajk
um
aran
dSw
aroop
,20
07)
This
study
exam
ines
the
spec
ific
link
bet
wee
npublic
spen
din
gan
dhea
lth
outc
omes
and
educa
tion
per
for-
man
ce.
Itis
bas
edon
pan
eldat
aan
duse
dro
bust
stat
isti
cal
met
hod.
66co
untr
ies
OL
SM
ore
spen
din
gor
reve
nue
dec
entr
alis
atio
nra
ises
wel
l-b
eing
(Bjo
rnsk
ovet
al.,
2008
).T
his
study
use
sla
rge
num
ber
ofco
untr
ies
and
exam
-in
esp
ecifi
cm
easu
reof
wel
l-b
eing.
How
ever
,it
only
info
rms
the
asso
ciat
ion
and
bas
edon
cros
sse
ctio
nal
des
ign.
29E
uro
pea
nco
untr
ies
OL
SP
osit
ive
rela
tion
bet
wee
ndec
entr
alis
atio
nan
din
di-
vid
ual
hap
pin
ess
(Dia
z-Ser
rano
and
Rodri
guez
-Pos
e,20
12)
This
study
exam
ines
the
effec
tof
dec
entr
alis
atio
nan
dhap
pin
ess
inE
uro
pea
nco
untr
ies.
How
ever
,it
lim
its
onit
scr
oss
sect
ional
des
ign.
1
14
-
Un
itof
an
aly
sis
(loca
tion
)m
eth
od
imp
ort
ant
ob
serv
ati
on
s/co
nsi
der
ati
on
sst
ren
gth
san
dw
eakn
ess
of
the
stu
dy
2.Negativeassociation
Pro
vin
ces,
Arg
enti
na
OL
SP
oore
rp
rovin
ces
are
less
succ
essf
ul
as
dec
en-
trali
sati
on
gen
erate
dsu
bst
anti
al
ineq
uali
tyin
pu
bli
csp
end
ing
(Gala
sso
an
dR
avall
ion
,1998)
Th
isst
ud
yex
am
ines
spec
ific
effec
tof
dec
en-
trali
sati
on
and
ineq
uali
tyin
pu
bli
csp
endin
g.
How
ever
,it
lim
its
on
pro
vin
cial
level
an
dst
a-
tist
ical
met
hod
olo
gy.
Cou
nti
es,
Ch
ina
Des
crip
tive
an
aly
sis
Dec
entr
ali
sati
on
resu
lted
inlo
wer
level
of
pu
b-
lic
serv
ices
inp
oore
rre
gio
ns
(Wes
tan
dW
on
g,
1995)
Th
isst
udy
exam
ine
the
effec
tof
dec
entr
al-
isati
on
on
pub
lic
serv
ices
inco
unti
esle
vel
.T
he
wea
kn
ess
of
this
stu
dy
isth
at
itis
base
don
lim
ited
nu
mb
erof
cou
nti
esan
dst
ati
stic
al
met
hod
olo
gy.
Mu
nic
ipali
ties
,In
don
esia
Qu
ali
tati
ve
stu
dy
Incr
ease
dd
isp
ari
ties
insp
end
ing
bet
wee
ncu
-ra
tive
an
dp
reven
tive
hea
lth
serv
ices
(Soer
ojo
an
dW
ilso
n,
2001)
Th
isst
ud
yp
rovid
esri
chan
dd
eep
an
aly
sis
on
the
ass
oci
ati
on
bet
wee
nd
ecen
trali
sati
on
an
dcu
rati
ve/
pre
ven
tive
hea
lth
serv
ices
.H
ow
ever
,it
ison
lyb
ase
don
small
nu
mb
erof
mu
nic
ipal-
itie
s.S
ixm
un
icip
ali
ties
,M
exic
oQ
ualita
tive
stu
dy
Pu
bli
cse
rvic
esca
nsu
ffer
as
are
sult
of
dec
en-
trali
sati
on
(Gri
nd
le,
2007)
Th
isst
ud
yp
rovid
esri
chu
nd
erst
an
din
gab
ou
tth
ep
roce
ssan
dd
eter
min
ant
of
dec
entr
ali
sa-
tion
inM
exic
ou
sin
gm
ixed
met
hod
olo
gy.
Th
em
ain
wea
kn
ess
of
this
stu
dy
isth
at
itis
on
lyb
ase
don
six
mu
nic
ipali
ties
.M
un
icip
ali
ties
,S
ou
thA
fric
aQ
uali
tati
ve
stu
dy
No
inh
eren
tre
aso
nw
hy
dec
entr
ali
sed
gover
n-
men
tssh
ou
ldb
eany
more
dem
ocr
ati
cth
an
centr
ali
sed
gover
nm
ents
(Hel
ler,
2001)
Th
isst
ud
yp
rovid
esri
chan
aly
sis
on
the
cur-
rent
dec
entr
ali
sati
on
refo
rmin
Sou
thA
fric
a,
bu
tit
isli
mit
edon
lyon
sever
al
case
s.28
pro
vin
ces,
Ch
ina
OL
SF
isca
ld
ecen
trali
sati
on
contr
ibu
ted
tolo
wer
pro
vin
cial
gro
wth
(Zh
an
gan
dZ
ou
,1998)
Th
isst
ud
yex
am
ines
the
effec
tof
fisc
al
dec
en-
trali
sati
on
on
pro
vin
cialgro
wth
inC
hin
a.
Th
em
ain
wea
kn
ess
of
this
stu
dy
isth
at
itis
base
don
cross
sect
ion
ald
ata
an
dth
em
eth
od
do
not
contr
ol
for
un
ob
serv
edh
eter
ogen
eiti
esb
ias.
Cou
ntr
ies,
Bra
zil
an
dA
rgen
tin
aT
ime
seri
esD
ecen
trali
sati
on
cou
ldle
ad
toin
crea
sed
fisc
al
defi
cits
an
dim
per
ilm
acr
oec
on
om
icst
ab
ilit
y(D
illi
nger
an
dW
ebb
,1999)
Th
isst
ud
yex
am
ines
the
effec
tof
dec
entr
al-
isati
on
on
fisc
al
cap
aci
tyan
dm
acr
oec
on
om
icst
ab
ilit
y,b
ut
itis
on
lyb
ase
don
two
cou
ntr
ies.
Mu
nic
ipali
ties
,U
gan
da
Qu
ali
tati
ve
stu
dy
Dec
entr
ali
sati
on
lead
toso
cial,
poli
tica
l,and
econ
om
icco
nfl
icts
(Gre
en,
2008)
Th
isst
ud
yp
rovid
esri
chan
aly
sis
on
dec
entr
al-
isati
on
an
dth
ein
cid
ence
of
vari
ou
sco
nfl
ict
inU
gan
da.
How
ever
,it
isd
on
ot
exp
lain
qu
ite
dep
thab
ou
tth
eh
isto
ryof
con
flic
tsb
efore
de-
centr
ali
sati
on
inth
eco
untr
y.C
ou
ntr
ies,
Lati
nA
mer
ica
Qu
alita
tive
stu
dy
Inst
ead
of
incr
easi
ng
the
rob
ust
nes
sof
lo-
calta
xati
on
,lo
calgover
nm
ents
have
incr
ease
dth
eir
dem
an
ds
on
centr
algover
nm
ent
for
more
reven
ue
shari
ng
(Wib
bel
s,2005)
Th
isst
ud
yp
rovid
esri
chan
aly
sis
on
the
pro
-ce
ssof
dec
entr
ali
sati
on
an
dit
sim
pact
of
loca
lgover
nm
ent
beh
avio
ur
inL
ati
nA
mer
ica
cou
n-
trie
s.A
sa
qu
ali
tati
ve
stu
dy
the
main
wea
k-
nes
sof
this
an
aly
sis
isn
ot
pro
vid
ed
epth
un
-d
erst
an
din
gon
the
chara
cter
isti
cof
soci
ety
on
each
case
stu
dy.
Cou
ntr
ies,
85
cou
ntr
ies
OL
SD
ecen
trali
sed
cou
ntr
ies
have
hig
her
per
ceiv
edco
rru
pti
on
and
poore
rse
rvic
ed
eliv
ery
per
-fo
rman
cein
pu
bli
ch
ealt
hse
rvic
es(T
reis
man
,2000)
Th
isst
ud
yex
am
ines
the
rela
tion
ship
bet
wee
nd
ecen
trali
sati
on
an
dco
rru
pti
on
usi
ng
larg
enu
mb
erof
cou
ntr
ies.
How
ever
,th
em
eth
od
do
not
esta
bli
shca
usa
leff
ect
on
both
vari
ab
les.
1
15
-
Unit
ofan
alysi
s(l
oca
tion
)m
ethod
imp
orta
nt
obse
rvat
ions/
consi
der
atio
ns
stre
ngt
hs
and
wea
knes
sof
the
study
Cou
ntr
ies,
Eas
tern
and
Cen
tral
Euro
pe
Des
crip
tive
anal
ysi
sP
ublic
serv
ices
can
suff
eras
are
sult
ofdec
en-
tral
isat
ion
(Lit
vack
etal
.,19
98)
This
study
only
pro
vid
esdes
crip
tive
anal
ysi
son
the
rela
-ti
onsh
ipb
etw
een
dec
entr
alis
atio
nan
dpublic
serv
ice
per
for-
man
ceac
ross
Eas
tern
and
Cen
tral
Euro
pe.
Cou
ntr
ies,
3dev
elop
ing
countr
ies
Qual
itat
ive
study
Dec
entr
alis
atio
nin
Col
ombia
,W
est
Ben
gal
and
Bra
zilhas
achie
ved
litt
lein
impro
vin
gse
r-vic
edel
iver
y(C
rook
and
Sve
rris
son,
1999
)
Ric
hst
ory
ofdec
entr
alis
atio
nin
thre
ediff
eren
tco
untr
ies
isex
amin
ed,
but
iton
lyca
ptu
res
the
pro
cess
and
his
tory
ofdec
entr
alis
atio
nin
few
pla
ces.
Cou
ntr
ies,
30co
untr
ies
OL
SC
oor
din
atio
nfa
ilure
sin
inte
rgov
ernm
enta
lre
-la
tion
sw
ere
like
lyto
resu
ltin
adefi
cit
bia
sin
dec
entr
alis
edp
olic
y-m
akin
g(D
eM
ello
,20
00)
The
effec
tof
dec
entr
alis
atio
non
inte
rgov
ernm
enta
ltr
ansf
eris
exam
ined
,w
ithou
tro
bust
stat
isti
cal
appro
ach.
Cou
ntr
ies,
46co
untr
ies
Pan
eldat
aan
alysi
sD
ecen
tral
isat
ion
was
asso
ciat
edw
ith
slow
erec
onom
icgr
owth
(Dav
oodi
and
Zou
,19
98)
The
effec
tof
dec
entr
alis
atio
nan
dec
onom
icgr
owth
isex
am-
ined
usi
ng
pan
eldat
a,but
itunab
leto
captu
reunob
serv
edhet
erog
enei
ties
acro
ssco
untr
yth
atm
ayre
sult
inw
eak
af-
fect
ofdec
entr
alis
atio
n.
Cou
ntr
ies,
6co
untr
ies
OL
SA
neg
ativ
eim
pac
tof
dec
entr
alis
atio
non
eco-
nom
icgr
owth
for
Mex
ico
and
the
Unit
edSta
tes
but
no
impac
tfo
rG
erm
any,
India
,It
aly
and
Spai
n(R
odri
guez
-Pos
ean
dB
wir
e,20
03)
The
impac
tof
dec
entr
alis
atio
nan
dec
onom
icgr
owth
isex
-am
ined
by
com
par
ing
dev
elop
edan
ddev
elop
ing
countr
ies,
but
the
model
isunab
leto
contr
olbia
sre
sult
edfr
omun-
obse
rved
het
erog
enei
ties
wit
hin
countr
y.3.
Incon
clu
sive
fin
din
gs
Sch
ool
s,C
hile
OL
ST
hou
ghed
uca
tion
dec
entr
alis
atio
nim
pro
ved
the
effici
ency
ofpro
vis
ion,
adec
line
was
exp
e-ri
ence
din
cogn
itiv
ete
sts
scor
es(W
inkle
ran
dR
ounds,
1996
)
This
study
use
dO
LS
toex
amin
eth
eeff
ect
ofdec
entr
al-
isat
ion
oned
uca
tion
per
form
ance
.T
he
lim
itat
ion
ofth
em
ethod
use
dis
that
itis
unab
leto
contr
olfo
runob
serv
edhet
erog
enei
ties
bia
sat
school
san
dco
mm
unit
yle
vel
that
may
affec
ted
uca
tion
per
form
ance
.L
oca
lgo
vern
men
ts,
Uga
nda
OL
SD
idnot
find
any
pos
itiv
eim
pac
tsof
dec
entr
al-
isat
ion
oneffi
cien
cyan
deq
uit
yof
loca
lpublic
serv
ices
(Azf
aret
al.,
2000
)
This
study
use
sO
LS
toex
amin
eth
eim
pac
tof
dec
entr
alis
a-ti
onan
dlo
cal
public
serv
ice
per
form
ance
.In
this
anal
ysi
s,it
use
dp
erce
ived
per
form
ance
rath
erth
anob
ject
ive
per
for-
man
ceof
public
serv
ices
.B
ias
esti
mat
esm
ayocc
ur
onth
ees
tim
atio
n.
Munic
ipal
itie
s,In
don
esia
OL
SD
ecen
tral
ised
hea
lth
spen
din
gis
not
sign
ifi-
cantl
yas
soci
ated
wit
him
pro
vehea
lth
care
use
and
hea
lth
outc
omes
(Wor
ldB
ank,
2008
b)
This
study
use
sla
rge
num
ber
ofdis
tric
tsin
Indon
esia
.H
owev
er,
the
stat
isti
cal
met
hodol
ogy
isunab
leto
captu
reth
eva
riat
ion
ofhea
lthca
redem
and
acro
ssdis
tric
tsan
din
-div
idual
s.L
owan
dm
iddle
inco
me
coun-
trie
s,14
0co
untr
ies
OL
SW
hile
dec
entr
alis
atio
nim
pro
ved
the
imm
u-
niz
atio
nco
vera
gein
low
inco
me
countr
ies,
op-
pos
ite
resu
lts
wer
eob
tain
edfo
rm
iddle
inco
me
countr
ies
(Khal
eghia
n,
2003
)
This
study
exam
ines
the
rela
tion
ship
bet
wee
ndec
entr
ali-
sati
onan
dim
munis
atio
nco
vera
geusi
ng
larg
enum
ber
ofco
untr
ies.
Its
met
hodol
ogy
how
ever
unab
leto
contr
olunob
-se
rved
het
erog
enei
ties
bia
sin
the
countr
yle
vel
whic
hm
ayaff
ect
onb
oth
vari
able
s.D
evel
opin
gco
untr
ies,
19co
un-
trie
sD
escr
ipti
vean
alysi
sT
he
impac
tof
dec
entr
alis
atio
non
pov
erty
re-
duct
ion
ism
ixed
(Ju
ttin
get
al.,
2004
)T
his
study
pro
vid
esri
chan
alysi
son
the
effec
tof
dec
entr
al-
isat
ion
onp
over
tyre
duct
ion
inso
me
countr
ies.
Itpro
vid
essi
mple
fram
ewor
kto
under
stan
dch
annel
sfo
reff
ecti
veor
ineff
ecti
vedec
entr
alis
atio
nfo
rp
over
tyre
duct
ion.
Cou
ntr
ies,
20le
ssdev
elop
edan
d17
dev
elop
edco
untr
ies
Pan
eldat
aan
alysi
sN
ore
lati
onsh
ipb
etw
een
fisc
aldec
entr
alis
atio
nan
dec
onom
icgr
owth
insa
mple
ofle
ssdev
el-
oped
countr
ies
(Phillips
and
Wol
ler,
1997
)
This
study
exam
ines
the
rela
tion
ship
bet
wee
nfisc
aldec
en-
tral
isat
ion
and
econ
omic
grow
thusi
ng
pan
eldat
aan
alysi
s.H
owev
er,
iton
lyuse
ssm
all
num
ber
ofco
untr
ies.
Cou
ntr
ies,
OE
CD
and
4le
ssde-
velo
ped
countr
ies
Cro