ED/EFA/MRT/2015/PI/11

Background paper prepared for the

Education for All Global Monitoring Report 2015

Education for All 2000-2015: achievements and

challenges

Decentralization and the Quality of Education

Anila Channa

2015

This paper was commissioned by the Education for All Global Monitoring Report as background
information to assist in drafting the 2015 report. It has not been edited by the team. The views
and opinions expressed in this paper are those of the author(s) and should not be attributed to
the EFA Global Monitoring Report or to UNESCO. The papers can be cited with the following
reference: Paper commissioned for the EFA Global Monitoring Report 2015, Education for All
2000-2015: achievements and challenges For further information, please contact
efareport@unesco.org

Background paper prepared for

Education for All Global Monitoring Report 2015

Decentralization and the Quality of Education

Anila Channa
April 2014
Abstract
Although decentralization has increasingly been advocated as a way of enhancing educational
quality, its potential in this area is still subject to some debate. This paper traces the popularity of
education decentralization over the past few decades to highlight that the post 2000 era has been
characterized by a deepening of reforms implemented earlier, an enhanced focus on school
decentralization interventions, and a notable increase in schemes in Africa. The article then
examines the empirical evidence on the relationship between decentralization and educational
quality to show that although the scholarship is limited in size and quality, a handful of rigorous
studies suggest that decentralization has the potential to enhance quality. Detailed case studies on
Mexico, Indonesia and Kenya supplement these two sets of analyses to not only showcase how
decentralization policies have evolved over time in these countries, but to also illustrate why
different decentralization approaches can result in dramatically different quality outcomes.
Table of Contents
1. Introduction ........................................................................................................................................................ 2
2. Decentralization and quality of education ......................................................................................................... 3
3. Evolution in education decentralization policies: impetus and popularity ........................................................ 4
3.1. Efforts in the 1980s and 1990s .................................................................................................................... 4
3.2. Education decentralization in the 2000s ..................................................................................................... 5
4. Evidence on decentralization and the quality of schooling ................................................................................ 8
4.1. Studies using randomized controlled trials ................................................................................................. 9
4.2. Studies using quasi-experimental techniques or panel data ...................................................................... 9
4.3. Studies using cross-sectional data analysis ............................................................................................... 11
4.4. Concluding remarks ................................................................................................................................... 11
5. Education decentralization case studies........................................................................................................... 13
5.1. Mexico ....................................................................................................................................................... 14
5.2. Indonesia ................................................................................................................................................... 17
5.3. Kenya ......................................................................................................................................................... 20
6. Prerequisites for success .................................................................................................................................. 24
7. Concluding remarks .......................................................................................................................................... 26
8. References ........................................................................................................................................................ 27
9. Appendix ........................................................................................................................................................... 31

Index of Tables
Table 1: Key Decentralization Reforms in Selected Developing Countries ............................................................. 4
Table 2: A Basic Introduction to Selected Identification Strategies........................................................................ 8
Table 3: Case Studies Factsheet............................................................................................................................ 23
Table 4: Summary of Selected Evidence on Decentralization and Quality ........................................................... 31

1 of 34

1. Introduction
Decentralization is probably the single most advocated reform for improving the provision of basic
services such as education in developing countries. Proponents argue that by taking decision-making
closer to the people, decentralization in service delivery can increase relevance in decisionmaking and enhance accountability. Both of these, the advocates posit, can in turn translate into
tangible improvements in the quality of education.
The aims of this paper are twofold: (1) to investigate key trends in the evolution of education
decentralization policies since 2000 and (2), to examine the empirical relationship between
decentralization and educational quality. The paper addresses these aims by first reviewing the
decentralization experiences of several countries more generally, and then by undertaking detailed
case studies of Mexico, Indonesia and Kenya to illustrate how different approaches towards
education decentralization can result in differing quality outcomes. Over the past two decades,
developing nations have made rapid progress in increasing participation in schooling. Less progress,
however, has been made in ensuring that the education dispensed is of adequate quality. Enhancing
our understanding of decentralizations potential in this arena can thus have broader policy
implications for how nations address the endemic low quality challenge in the future.
At the onset, it is important to establish that although education decentralization initiatives are now
ubiquitous, they are not uniform in content. Rather, there are countless configurations of
decentralization schemes across the globe, as reforms differ substantially based on what education
decision-making responsibility has been devolved and what level it has been devolved to. For the
sake of simplicity, this paper distinguishes between two forms in particular decentralization of
education to (1) local governments and (2) schools. More formally, the first form is defined as the
transfer of authority for decision-making, finance, and management to quasi-autonomous units of
local governments (Litvack and Seddon 1999: pp. 3) specifically in the arena of education. The
second form, referred to interchangeably as school decentralization or school-based management
(SBM) in the text is defined as ..a form of decentralization that identifies the individual school as the
primary unit of improvement and relies on the redistribution of decision-making authority as the
primary means through which improvement might be stimulated (Malen et al. 1990: pp. 290)1.
This paper proceeds in the following manner. Section two briefly discusses the link between
decentralization and educational quality. Section three traces the popularity of the scheme over
time, highlighting that the 2000s have witnessed a deepening of reforms implemented earlier, an
enhanced focus on school decentralization interventions, and a notable increase in schemes in
Africa. Section four summarizes the findings of the relevant empirical literature. It argues that while
the evidence base is still limited in size and quality, a handful of rigorous evaluations lend support to
SBMs potential in enhancing quality. Section five presents three case studies in order to showcase
different experiences with the intervention. It highlights that realizing decentralizations potential is
often predicated on design and implementation, both of which in turn are often shaped by economic
and political conditions. Section six considers prerequisites; and the final section concludes.
1

See Rondinelli (1981) for a further categorization of decentralization to local governments into the forms of
deconcentration, delegation and devolution. See Leithwood and Menzies (1998) for a further categorization of school
decentralization into the forms of administrative control, professional control, community control and balanced control.

2 of 34

2. Decentralization and quality of education

The production of education is a complex process. Besides inputs such as desks, textbooks and
teachers, the quality of delivery often also depends on understanding what local learning challenges
exist and on providing the right incentives to address these challenges.
By borrowing arguments from the first generation literature on decentralization (see Tiebout 1956;
Oates 1972), proponents posit that decentralization of education has the ability to improve the
quality of education. This claim is usually based on two arguments:
1. First, advocates contend that locating decisions regarding education closer to those
responsible for delivering it can enhance the relevance of decision making through greater
knowledge of local needs and preferences. The standardized delivery by central
governments, on the other hand, is assumed to be unable to address these heterogeneous
demands. In practice, through this route, local governments or schools could for instance
improve learning by directing greater resources to customized areas of need, or by tailoring
class plans to focus on topics that local students appear to be struggling with the most.
2. Second, supporters assert that decentralization can increase accountability in the education
system by locating decision-makers closer to parents and the community. These
stakeholders can then voice their concerns, as well as monitor education delivery more
directly. This argument is more commonly expounded using a framework from the 2004
World Bank Development Report: Making Services Work for Poor People, which indicates
that decentralization creates a shorter route of accountability for service delivery by
providing a direct link between citizens and providers. Via this mechanism, parents can thus
affect learning by, for example, monitoring teachers to ensure that they are actually
teaching, or by voicing concerns over poor performance to schools or local officials who
have the authority to make decisions regarding its improvement.
Of course, following decentralization there is no guarantee that the quality of education will actually
improve through either of these mechanisms. In contrast to the optimism in the earlier literature,
the second generation literature on decentralization is in fact significantly less decisive of its
benefits. This body of work proposes a world of self-serving officials and politicians whose incentives
are crucial in determining the service delivery outcomes of the intervention (see Ahmad and Brosio
2009). Critics in this vein thus highlight the possibilities of elite capture, the chances of inequity and
the lack of capacity of local governments and schools to deliver quality education (see Winkler 1989;
Fiske 1996). They further emphasize that inconsistent curricular and quality standards, resistance
from teachers unions, and the reluctance and inability of parents to make technical decisions on
education matters may in fact reduce, not increase, the effectiveness of education service delivery.
The lack of a clear theoretical prediction on how decentralization affects educational quality forces
us to rely on the empirical evidence instead. As this paper will demonstrate in later sections, this
empirical scholarship indicates that the reform does have the potential to enhance learning.
However, whether or not it realizes this potential is usually predicated on how economic conditions
and political forces shape the design and implementation of reforms.
3 of 34

3. Evolution in education decentralization policies: impetus and popularity

Decentralization has proven to be a popular reform in the developing world. According to World
Bank (2008), most client countries have decentralized responsibilities to at least one level of lower
government. Just as widespread has been the adoption of school-based management practices
countries as diverse as Argentina, Thailand, Israel and Senegal have all experimented with the
initiative in one form or another (Barrera-Osorio et al. 2009). Table 1 highlights reforms for a sample
of selected countries to illustrate decentralizations prevalence, both before and after the year 2000.
3.1. Efforts in the 1980s and 1990s
Amongst the earliest decentralizing nations, the
impetus for transferring decision-making authority
to local governments was usually a broader political
or economic reform. In much of Latin America, for
instance, decentralization formed an integral part of
wider political democratization movements (Litvack
et al. 1998). In the former Soviet Union and Eastern
Europe, decentralization of education authority
accompanied the important shift from a command
to a market economy (De Grauwe 2005). In other
countries such as South Africa and Sri Lanka,
decentralization emerged as a tool for addressing
ethnic diversity and conflict (World Bank 2008).

Table 1: Key Decentralization Reforms in

Selected Developing Countries
Decentralization to local
governments
Pre 2000
Argentina
Bangladesh
Bolivia
Brazil
Burkina Faso
Chile
China
Colombia
Ethiopia
India
Russia
South Africa
Sri Lanka
Tanzania
Uganda
Post 2000
Albania
Cambodia
Indonesia
Nepal
Pakistan
Rwanda
Sierra Leone
Vietnam

Decentralization to
schools
Argentina
Brazil
El Salvador
Guatemala
Honduras
Hong Kong
Israel
Mexico (AGE)
Nicaragua
Thailand

Many school-based management interventions

Benin
Gambia
accompanied the education reforms that were
Indonesia
decentralizing authority to lower levels of
Kenya
Madagascar
governments. Many others still were motivated by a
Mexico (PEC)
myriad of other factors. One of the earliest SBM
Niger
efforts from the developing world, for instance,
Qatar
Senegal
came from El Salvador in 1991 in the form of the
Source: Authors own compilation from literature
celebrated EDUCO community school model. The
innovative intervention was specifically driven by a desire to expand schooling after the end of a civil
war (Jimenez and Sawada 1999). In Asia, on the other hand, Hong Kong and Thailand were two of
the earliest adopters of the school decentralization reform interestingly, both stated enhancing
school effectiveness as a key goal of the reforms adoption (Bruns et al. 2011).
Regardless of the explicit goals proffered by governments for decentralizing education, sceptics
observe that the programme has seldom been the result of grassroots pressure to improve the
quality of education (Smoke 1993; De Grauwe 2005; King and Guerra 2005). On the contrary, these
critics posit that education decentralization has instead usually been the consequence of a push
from multilateral agencies to decentralize. This push is certainly evident in the World Banks
strategy, which has supported the reform since the 1990s both through its advocacy and by
dedicating a growing proportion of its financing to decentralization projects. Estimates indicate that

4 of 34

between 1990 and 2006, the Banks commitment on projects with a decentralization component
stood at approximately USD 32 billion, spread over almost 90 countries (World Bank 2008).
Other scholars note that the earliest decentralizing nations were probably more concerned with
dismantling large, costly bureaucracies than in specifically improving learning outcomes (Caldwell
2005). Consistent with this proposition is the fact that evidence on the efficacy of education
decentralization at that point in time was still overwhelmingly pessimistic. In a review of 56 studies
published from the 1990s, for example, Shah et al. (2004) noted that decentralization to lower levels
of governments in some cases improved and in many others worsened service delivery. Similarly, in
a comprehensive survey of 83 empirical studies on SBM, Leithwood and Menzies (1998) argued that
there was no evidence of any effect of the reform on student outcomes at all.
3.2. Education decentralization in the 2000s
Since the mid-1990s, there has been significant progress in the literature on education and
decentralization. The most crucial development perhaps has been the growing consensus that inputs
such as desks, textbooks and blackboards are not enough to enhance learning (see Hanushek 1995;
Glewwe and Kremer 2005). This consensus has understandably resulted in an enhanced focus by
governments and donors alike on governance reforms such as decentralization instead. A
concomitant development has been the increased availability of data on student attainment, which
has permitted more rigorous evaluations of the effect of decentralization on quality. A handful of
studies now suggest that the scheme can be beneficial to educational quality after all (see next
section). This has further invigorated enthusiasm for the reform amongst the stakeholders involved.
Besides advances in these two sets of literatures, the persistent, widespread issues of poor
educational quality in developing countries on one hand, and the strong push by donors towards
achieving the Education for All (EFA) goals on the other, have contributed to making decentralization
even more popular today. This popularity, however, makes an inventory of all the education
decentralization programmes that have been implemented since 2000 not only close to impossible,
but also of limited analytical value in understanding policy evolution. Instead, this paper presents
three interrelated trends that have been particularly prominent in recent years together, these
three trends - which are described in some detail below - characterize the kind of decentralization
policy changes most commonly seen in the post 2000 period.
3.2.1. Deepening of decentralization efforts
First, although there were a few notable fresh interventions such as those seen in Indonesia and
Pakistan, many more decentralization policies in this period involved the deepening of existing
reforms. One type of deepening initiative commonly seen was that of strengthening legal
frameworks for decentralization. Both in Brazil and in the Indian states of Karnataka and Andra
Pradesh, for instance, new laws were promulgated expounding on local fiscal responsibilities.
Another area for improvement that proved popular in the post 2000 era was that of fiscal design
countries such as Russia and Uganda, for example, moved to formula-based allocations in order to
improve transparency in their fiscal systems. Finally, electoral accountability reforms were also
adopted in a number of countries. One prominent example of this is Burkina Faso, which had
decentralized its government system in the 1990s but held its first local elections for lower levels of
government in 2006 as part of a deepening effort.
5 of 34

One reason for the prevalence of this particular trend was the fact that most countries had already
decentralized in some form or other by the late 1990s for these countries, deepening was simply a
form of organic growth in public sector reform. The much more common reason, however, was the
multiple challenges most countries faced in their early experiments with the scheme. A whole strand
of literature in fact became dedicated to exploring these early instances of decentralization gone
wrong (see Prudhomme 1995; Tanzi 1995). Scholars in this vein highlighted, among other
challenges, misaligned incentives, weak or stalled implementation, unfunded mandates and serious
issues in capacity and training. Unsurprisingly, throughout the 2000s, countries suffering from such
challenges turned to the multilateral agencies that had advocated decentralization in the first place
for both technical advice and funding support on how to strengthen their reforms.
Stimulated by these challenges, as well as the second generation literature on decentralization that
had by this time shifted from a normative stance to a more political economy approach, donors
altered their approach towards reform adoption - particularly so in the late 2000s. Increasingly, the
World Bank, for instance, advocated undertaking a political economy analysis in order to understand
whether decentralizations benefits could be realized in a particular context (for e.g. see Eaton et al.
2010). The politics of the reform gained centrality in this new approach, and policymakers not only
became more conscious of the incentives of different actors in the decentralization process, but also
began to acknowledge their own role as political actors in the implementation of these schemes.
Examples of projects funded by the World Bank in this post 2000 era for strengthening previous
education decentralization reforms, to name a few, included: capacity building initiatives in Bolivia
and Ethiopia, elementary decentralization support efforts in El Salvador, and interventions targeting
marginalized communities in Brazil. In fact, even the newly decentralized countries of Indonesia and
Pakistan pursued deepening efforts through much of the 2000s with the assistance of the Bank.
3.2.2. Focus on school decentralization schemes
School decentralization reforms in particular gained popularity in the past 14 years. In most
instances, these schemes involved the empowerment of school councils comprising of school
teachers, parents and community members. Many, albeit not all, reforms also provided grants that
schools could use based on school improvement priorities set by these councils.
In many countries, implementing SBM was a kind of deepening effort, a logical extension of broader
education decentralization interventions that had already been implemented. Philippines, for
example, implemented a Big Bang reform in the 1990s and supplemented it with school autonomy
programmes in the early 2000s. Similarly Mexico, one of the case studies described in detail later in
this paper, underwent decentralization to local governments in 1992, and then went on to
implement its biggest SBM programme in 2001. Of course, not all countries adopting school
decentralization schemes decentralized education to democratic local governments first. Qatar is an
important example. The emirate, which in contrast to the Philippines and Mexico is a constitutional
monarchy, launched independent schools that were funded by the government but managed locally
in 2003 as part of its experiment with school decentralization.
The popularity of this form of decentralization is actually not that surprising - the most optimistic
evidence available today is in fact on SBM (see next section). Nonetheless, it is important to note
6 of 34

that many of the evaluations of the intervention were sponsored by the Bank itself, who by this time
had become a key supporter of SBM. Trends in the Banks education project funding are telling of
this support. Between 2000 and 2006, to illustrate, 10% of the World Banks education portfolio
supported SBM initiatives (Barrera-Osorio et al. 2009). In the fiscal year 2012, on the other hand,
almost 50% of the 17 operations approved by the World Banks Education Sector Board contained
school decentralization elements.
3.2.3. Increased decentralization activity in Africa
Finally, recent years have also seen African countries as diverse as Kenya, Morocco, Chad and Benin
all decentralizing elements of their education systems.
Political reforms and increased stability in the region contributed significantly to these increased
decentralization efforts. In Congo and Sierra Leone, for instance, the end of civil wars in the early
2000s prompted renewed interests in decentralization to local bodies. In Rwanda, stability after the
end of genocide in the 1990s allowed the country to also embark on decentralization to lower tiers
of government in the 2000s. By this time, many nations in the region had embraced democracy and
multi-party political systems, both of which resulted in significant shifts in political power in
individual nations. In 2000, for example, the opposition party in Senegal won elections after 19 years
of rule by the same party. Similarly, in Guinea, the first civilian election since independence was held
in 2010. This general trend of political reform was more often than not accompanied by public sector
reform in general, and decentralization programmes in particular.
Besides political reform, a significant increase in funding for social sectors in recent years also
contributed to the decentralization activity in Africa (Gerschberg and Winkler 2003). The Highly
Indebted Poor Countries debt initiative (HIPC), for instance, was accelerated in the 2000s forgiving
the external debt of many African countries in exchange for their commitments to invest in basic
education. Further, slow progress in achieving the Millennium Development and the Education for
All goals resulted in a large injection of donor funds through the Global Partnership in Education
initiative in several countries in the region. This increase in funds also encouraged the launch of
decentralization programmes as most donors believed that the reform would provide the right
institutional environment in which to expand access and quality (Gerschberg and Winkler 2003).
In line with first trend, the multiple programmes implemented in Africa often involved deepening of
existing reforms such as those seen in Ethiopia and in Tanzania - both countries launched
programmes to strengthen capacity for their previously implemented schemes. Similarly, in line with
the second trend, countries such as Benin and Niger experimented with SBM during this period.
Other countries that had education decentralization projects funded by the World Bank in the post
2000 era in the region included: Cameroon, Chad, Ghana, Malawi, Mali, and Nigeria. In addition,
three of the four prominent randomized controlled trials in the SBM arena in the 2000s were
conducted in Kenya, The Gambia and Mozambique, further highlighting the increased
decentralization activity in the region.

7 of 34

4. Evidence on decentralization and the quality of schooling

The previous section highlighted just how widespread decentralization is today. In spite of this, the
empirical evidence linking decentralization and quality of schooling is as yet limited in size and
quality. Varying reform content, limited data availability, and the difficulty of disentangling
decentralizations effects from those of other reforms that tend to accompany it have severely
restricted our ability to draw conclusions on what decentralization can and cannot do.
That said, the evidence base in this arena is fortunately not only growing, but is also increasingly
employing more sophisticated methods to isolate the causal effects of education decentralization.
Establishing causality is the hallmark of good impact evaluation. And in the field of economics,
ascertaining whether relationships can be considered causal generally relies on employing what is
referred to as a credible identification strategy an empirical strategy that allows for the
measurement of a counterfactual; or in a laypersons terms, a strategy that permits the
measurement of what would have happened had the participants not received the decentralization
intervention (see Gertler et al. 2007 for a more detailed discussion on establishing causality in school
decentralization interventions).
This section summarizes key studies that
attempt to quantitatively establish a causal link
between education decentralization and the
quality of schooling, as measured by
achievement scores or failure, repetition and
dropout rates2. Along Channa and Faguet (2012),
the evidence presented is organized by the
established hierarchy of identification strategies
in economics as widely taught in graduate
programmes today (see Table 2 for a brief
introduction
to
selected
identification
strategies). Thus, at the top of the hierarchy are
randomized controlled trials (RCT), which are
considered the gold standard in establishing
causal links, while at the bottom are mostly
cross-sectional studies that are unable to control
for biases common in this type of analysis. In the
middle category fall various studies that try to
address biases through other methods, such as
those that use difference-in-differences
techniques or employ the use of panel data with
fixed effects. The evidence described in this
section is summarized in the Appendix.

Table 2: A Basic Introduction to Selected

Identification Strategies
An identification strategy can be defined as the manner in
which a researcher uses observational data to approximate a real
experiment (Angrist and Pischke 2009: pp. 7).
Randomized
Considered the gold standard for establishing
controlled trial causality in econometrics, this method
or experiment
randomly assigns participants into treatment
and control groups and then compares the
difference in their outcomes to estimate
causal effects. Because participants are
randomly assigned, the control group is
similar in all respects to the treatment group
except for the actual treatment applied.
Difference-inThis method is considered a quasidifferences
experimental technique. It relies on
comparing the change in the treated group to
the change in a plausible comparison group,
before and after an intervention. The key
benefit of this method is that is allows us to
control for differences between the
treatment and comparison groups that are
constant over time.
Fixed effects This method also relies on using data over
model
on multiple periods of time and is technically
panel data
close to the difference-in-differences
technique. In models using two way fixed
effects, the key benefit of this method is that
it allows us to control for differences that are
(1) constant over time but vary across
entities, as well as those that are (2) constant
across entities but vary over time.
Cross-sectional This method relies on analysis using data
data regression from a single period. Unless complemented
analysis
by other techniques, this method is unable to
address estimation biases.

Source: Ravallion (1999); Angrist and Pischke (2009)

For a more comprehensive review of the literature on the relationship between decentralization and service delivery in
developing countries, see Channa and Faguet (2012). For a review of empirical evidence from OECD and other developed
nations, refer to Ahmad and Brosio (2009). For good reviews of the school decentralization literature, see Bruns et al.
(2011) and Barrera-Osorio et al. (2009).

8 of 34

4.1. Studies using randomized controlled trials

Understandably, there is no experimental evidence on education decentralization to lower levels of
government. There are, nevertheless, four contributions that use randomized trials to investigate
SBMs potential in enhancing attainment. These four papers yield somewhat positive findings.
The most optimistic evidence is due to Duflo et al. (2007), who examine a randomized controlled
trial from Kenya. The trial tested a number of interventions on 210 primary schools, one of which
involved an SBM component that empowered school councils to hire and monitor contract teachers.
Duflo et al. compare the SBM groups to their counterparts in the control group to show that
students in the treatment cell scored 0.18 and 0.24 standard deviations higher in Mathematics and
Language than their non-treated counterparts (see also Kenya case study).
Just as positive is a trial from Indonesia, which was implemented in approximately 500 public
schools (Pradhan et al. 2011). In particular, the RCT tested the impact of providing a block grant of
approximately USD 800 to treated schools, in addition to a combination of three initiatives including
training, election of SC members, and collaboration with the village council. Pradhan et al. (2011)
find that the training intervention had no effect on learning, but that both the election and linkage
interventions were successful. Student test scores increased by 0.17 standard deviations for linkage,
and 0.22 standard deviations for linkage and elections (see also Indonesia case study).
Blimpo and Evans (2011) study of an experimental trial in The Gambia is marginally less optimistic,
yet suggests that SBM can have positive effects on absenteeism. The Gambia experiment comprised
of a grant and school training programme in 273 primary schools. The authors find that four years
into the intervention, there was no discernible impact on student test scores. They do though report
that the trial resulted in a 21% and 23% reduction in student and teacher absenteeism, respectively.
To explain the lack of impact on quality, they posit that a minimum of 45% of adult literacy may be
required before the intervention can yield positive effects on learning outcomes.
Finally, Glewwe and Maiga (2011) present the least optimistic experimental results in this category.
They examine a randomized trial in Madagascar, which involved management reforms at three
levels district, sub-district and school. In a sample of 30 districts, sub-districts and schools were
randomly sorted into treatment and control groups. Glewwe and Maiga document some school
improvements in the first six months, but by the end of two years find no impact on aggregated test
scores. They conclude not against the reform per se, arguing instead that results may be driven by
the short time since implementation of the intervention.
4.2. Studies using quasi-experimental techniques or panel data
Evidence using quasi-experimental from Argentina suggests that decentralization of education
decision-making authority to lower levels of government can enhance student achievement. Like
many other countries, Argentina undertook devolution of personnel and budgeting decision to
provinces as part of a broader structural reform. Galiani et al. (2008) perform a difference-indifferences estimation, comparing changes in student scores in secondary schools that had always
been under provincial control to changes in schools that were under federal control until the 1991
reform. They show a positive association between decentralization and Mathematics and Spanish
scores - after five years, they report a 4.9% and 6.9% increase compared to the mean, respectively.
9 of 34

Four country studies that also rely on the difference-in-differences method use it to investigate the
effect of school decentralization instead, and are similarly favourable. Gertler et al. (2011) consider
the AGE intervention from Mexico, which is an SBM reform that provides training and small grants
to parent associations in disadvantaged schools to invest in infrastructure and materials. Gertler et
al. exploit the phased implementation of the AGE programme, comparing schools that adopted AGE
earlier to those that adopted it later. Their analysis suggests that participation in the AGE
programme is associated with a reduction in the proportion of students failing and repeating a grade
of 5.4% and 4%, respectively. Skoufias and Shapiro (2006) are equally optimistic about school
decentralization in the country. They combine their difference-in-differences method with a
matching technique to examine the PEC programme. PEC is another SBM reform in Mexico that
provides annual grants to disadvantaged schools to improve educational quality. They find that
participation in PEC is significantly associated with a 0.24, 0.24 and 0.31 reduction in dropout, failure
and repetition rates, respectively. Murnane et al. (2006) and Bando (2011) concur, arguing that
participation in PEC is indeed related to improvements in quality (see also Mexico case study).
From Brazil, Paes de Barros and Mendonca (1998) study the three SBM changes of financial
autonomy of schools, head teacher election and establishment of school councils using data from
1981 to 1993. They find that financial autonomy is associated with a significant drop in repetition
rates and also report positive trends in the mean level lags in grades attended and share of children
with lag. Equally supportive of school decentralization is an investigation from Asia by Khattri et al.
(2010). The authors evaluate the impact of school-based management reforms implemented in 2003
in 23 districts in the Philippines, which provided funding as well as training to principals and parents.
Khattri et al. compare the test score performance of students in schools that implemented the
intervention in the first phase of implementation to those that implemented the intervention in a
later phase. They report that as compared to the control group, the treatment group showed a 1.45
percentage point improvement in overall attainment.
In contrast to the above papers in this category, the results of two cross-country studies on SBM are
mixed, albeit not entirely pessimistic. Gunnarsson et al. (2009) consider the effects of two
components of school decentralization reforms - school autonomy and participation - in 10 Latin
American countries. The authors employ an instrumental variable technique in which they use
principal attributes and legal structure of a country as instruments for school autonomy and
participation. They find a negative and significant association between test scores and school
autonomy, but a positive one with parental participation. The second cross-country study is due to
Hanushek et al. (2011), who use data from four waves of PISA test scores to establish the
relationship between student achievement and autonomy in curricular, personnel and budgeting
areas. Their dataset contains data on 1 million students from 42 countries, of which 25 are classified
as high income nations. Using a two way fixed effects model, Hanushek et al. find the relationship
between the parameters of interest to be negative, albeit heterogeneous across countries based on
income levels. A disaggregated analysis suggests that school autonomy is related to positive
outcomes in developed and high-performing nations, but to negative ones in developing and lowperforming nations mostly due to a lack of institutional capacity.

10 of 34

4.3. Studies using cross-sectional data analysis

The studies employing primarily cross-sectional data analysis are not unanimous where the benefits
of decentralization to lower levels of government are concerned. Freinkman and Plenakanov (2009),
for instance, find that the statistical relationship between test scores of students from Russia and
regional fiscal decentralization is consistently positive. Treisman (2002), on the other hand, finds
that the presence of constitutional autonomy and electoral accountability at the local level are both
associated with a higher level of youth illiteracy in data from up to 166 nations. Meanwhile, Di
Gropello (2002) shows conflicting results on the impact of municipality level devolution on quality in
Chile. She finds that both devolved wage incentives and training expenditure at the municipal level
are associated with higher scores, while greater financial autonomy is not.
Numerous country-level contributions from the Latin American region also report contradictory
findings on the impact of school decentralization on quality. Jimenez and Sawada (1999) examine
EDUCO from El Salvador, perhaps the most celebrated case of SBM, in which community schools
were established to enhance access in rural areas. They find no significant difference in test scores,
although do observe that student absenteeism is lower in EDUCO schools. In a follow-study, they
contend that EDUCO students also have higher continuation rates. Sawada and Ragatz (2005) use a
different econometric technique on the same dataset, still finding no impact on student
achievement but noting significantly lower teacher absenteeism.
In an examination of SBM in Nicaragua, King and Ozler (2000) argue that it is not school autonomy
on paper (de jure), but rather autonomy in practice (de facto) that that improves student
performance. Parker (2005) provides mixed support for the same reform. Using more nationally
representative data, Parker shows that third graders in autonomous schools scored significantly
higher than their counterparts in centralized schools in Mathematics, but for sixth graders this effect
was negative. In neither case were results significant for Spanish. In a similar vein, Di Gropello and
Marshall (2005) assess the impact of participating in a PROHECO community school in Honduras.
The authors report marginally lower dropout and repetition rates and higher Science achievement,
but no change in Mathematics and Language scores. In Argentina, on the other hand, Eskeland and
Filmer (2007) find that autonomy is indeed significantly associated with student test scores in
Mathematics, but not in Language.
4.4. Concluding remarks
Answers to policy questions should place a greater weight on what higher quality evidence has to
say. So, what does all the above evidence tell us about the link between the quality of education and
decentralization of education decision-making authority to (1) lower levels of governments and (2)
to schools? The evidence on the former link is limited in size and quality and is inconclusive at best,
with positive associations reported in Argentina and Russia but not in Chile or in a cross-country
analysis. Evidence on the latter is fortunately significantly larger. Moreover, on balance the stronger
evidence indicates that school-based management can improve test scores - as it did in Kenya,
Indonesia and the Philippines - as well as reduce repetition, failure or dropout rates - as it did in
Mexico and Brazil. Further, evidence from The Gambia indicates that decentralization can reduce
both student and teacher absenteeism.

11 of 34

That said, mixed findings from cross-country analyses of the Latin American region and of 42
countries participating in PISA suggest that the effect of SBM may be heterogeneous across
countries. Thus, the detailed case studies that follow in the next section consider why this might be
the case. They do so by exploring the different contexts of, motives for, and nature of
decentralization reforms in Mexico, Indonesia and Kenya.
In addition, randomized trials from Madagascar and The Gambia respectively demonstrate that
interventions may take time to be beneficial or may be predicated on prerequisites such as
community literacy. These prerequisites to realizing the benefits of decentralization, as well as
others suggested in the literature, are briefly discussed in the final section of this paper.

12 of 34

5. Education decentralization case studies

The three case studies in this section describe the distinct approaches to education decentralization
adopted by Mexico, Indonesia and Kenya. The three countries are similar in that each one of them
has achieved a satisfactory level of primary enrolment, yet has continued to struggle with poor levels
of educational quality. They are dissimilar in almost all other respects - they represent different
geographies, differing levels of income, and varying models of education decentralization.
The Mexico study describes the case of a middle income country whose decentralization trajectory
has been significantly influenced by prevailing economic conditions, the vested interests of a
powerful teachers union as well as a change in political leadership in the 2000s. Since the 1980s the
government has been gradually decentralizing more authority to local governments, although
Mexicos education system continues to be criticized for being too centralized. Evidence on the
nations school decentralization initiatives, however, is optimistic multiple studies indicate that
SBM interventions have been successful in improving both quality and crucially, equity in education.
In sharp contrast to this, evidence on the efficacy of the Indonesian reform is both thin and
contradictory. Indonesia, a lower middle income country, implemented a Big Bang
decentralization scheme in 2000 with the assistance of the World Bank. However, design and
implementation issues are commonly acknowledged to not only have limited decentralizations
potential in enhancing learning, but to also have contributed to increasing inequity. The Indonesia
case study thus illustrates the importance of both these elements in the success of decentralization
schemes, while also emphasizing some unintended consequences of the reform on equity.
Unlike Mexico and Indonesias experiences with decentralization, Kenyas experiment with
decentralization is in flux with reform implementation still underway. Historically, the low income
country has lacked a consistent decentralization strategy. Consequently, its education system today
is an incoherent mix of centralized and decentralized elements, which has resulted in poor
accountability on one hand and excessive redundancy on the other. Counterintuitively though, the
strongest evidence in favour of school decentralization, as we saw in the previous section, also hails
from Kenya this evidence highlights the potential of decentralization if the country can get its
policy right. Importantly, the continued fluidity in this case exemplifies the status of several
countries that have demonstrated a renewed interest in decentralization in the post 2000 period.
Each case study that follows briefly describes the context prior to decentralization, the motivation
for the reform itself, its actual content and evolution in design and finally the evidence available on
how the scheme has affected the quality of education. Together, the case studies highlight that
design and implementation have the ability to alter decentralizations effects on quality. They also
demonstrate that design and implementation policies in turn are usually shaped by economic
conditions, as well as the politics of key stakeholders such as donors, political parties and teachers
unions. Table 3 follows this text with a summary of key comparative statistics for each country.

13 of 34

5.1. Mexico
At the primary level, net enrolment in Mexico had reached close to universal as early as the mid1980s. However, compared to other OECD countries, quality in the education system remained poor
even after taking the countrys lower level of development into account (Hagerstrom 2006).
Poor quality in fact plagued the education systems of most Latin American nations at the time. The
region had one of the highest repetition rates in the world, and tests showed that students learned
less than in other regions (Murillo 1999). The first PISA test in 2000 confirmed this lack of learning almost all Latin American nations underperformed on the assessment. Out of the 32 countries that
participated, Mexicos 15 year olds statistically outperformed their counterparts in mathematics in
just three other countries (PISA website).
5.1.1. Early decentralization efforts
The debt crisis of 1982 and the economic crisis of 1994 both affected Mexico acutely. Mexico thus
embarked on a series of adjustment programmes to reduce government expenditures and raise
revenues throughout the 1980s and 1990s (Murillo 1999).
As part of these adjustment programmes, as well as a broader shift towards a more open economy,
Mexico began to pass autonomy to states in the early 1980s (Ornelas 2000). Educational institutions
at this time were notorious for their rigidity, inefficiency and unresponsiveness (Murnane et al.
2006). Thus, in his inaugural speech in 1982, President De la Madrid, announced the transfer of basic
education to the states, citing the goals of improving quality and equity. Although there appeared to
be no grass roots movement for the reform at the time, development agencies such as the IADB and
the World Bank welcomed these interventions as they fit in with their own priorities (Ornelas 2000).
The education system itself underwent two key reforms in this early decentralization period. The
first came about in 1978, and involved the deconcentration of the federal governments Public
Education Secretariat (known as SEP) to the state level. Through this deconcentration, states formed
education delegations that collected basic data and performed regional planning roles under the
auspices of the SEP (Ornelas 2000). The second, and significantly more substantial, reform was
enacted in 1992 through the National Agreement for Basic Education Modernization. The agreement
was signed by the federal government, the governors of all 31 states, and the national teachers
union. Importantly, it legally transferred the education secretariats to state control.
A critical aspect that shaped the content of the Mexican reform was the role played by the national
teachers union or SNTE. The SNTE was a powerful political actor in the education arena that had
been closely aligned to the ruling political party for many years (Murillo 1999). The fear of losing the
power of collective bargaining prompted the SNTE to oppose efforts to decentralize through much
of the 1980s. While decentralization efforts in other Latin American nations continued to be stymied
due to a similar resistance by teachers unions, Mexicos government instead negotiated significant
concessions to gain union support. Thus, through the 1992 agreement, SNTE consented to the
reform in exchange for wage increases and new schemes to support career progression (Ornelas
2000). As part of the agreement, wage negotiations were allowed to remain centralized, much to the
delight of the SNTE (Hagerstrom 2006).

14 of 34

Sceptics criticized the decentralization scheme, arguing that decision-making in fact became even
more centralized as a result (Gerschberg 1999). Besides authority over personnel, the Centre
continued to maintain control over the curriculum, testing, and budgeting. The lack of meaningful
devolution was evident in the financial arena as well. Financing remained predominantly federal at
over 80% well after the reforms implementation (ibid.). Moreover, financial transfers to states were
determined by annual negotiations rather than transparent criteria-based formulae. On the
expenditure side, devolution was likewise limited - states effectively gained control of only 10% of
expenses as the other 90% comprised of teachers salaries, which were determined centrally.
The 1992 reform also called for the establishment of councils of social participation of education,
comprising of teachers, parents and community members (Ornelas 2000). These councils, however,
remained dysfunctional for a long time. To mobilize them and also as part of a broader equalization
programme for schools, in 1996 Mexico launched the Apoyo a la Gestion Escolar (Support to School
Management) or AGE initiative. AGE gave cash grants of USD 500 to USD 700 to school councils in
highly disadvantaged communities to spend on avenues they considered appropriate (BarreraOsorio et al. 2009). Grants were generally meant to be used on infrastructure projects and parents
had to commit to greater involvement in schools. Although the authority given to parents and
schools was limited, AGE did represent the first time such authority was devolved.
Over the next decade, some states made efforts to further decentralize to the municipal level, while
the Centre itself also bundled modest increases in state autonomy into other education reforms. In
spite of this, the countrys political system remained one of the most centralized in the world
(Grindle 2007).
5.1.2. Decentralization in the 2000s
The year 2000 saw the dominant political party, which had ruled for over 70 years, lose power to the
opposition. Historically, quality improvement initiatives had been implemented in a uniform manner
across all schools. With this change in political leadership, however, the education bureaucracy
became open to new education policies and procedures (Murnane et al. 2006).
The key SBM type reform adopted as a result was the Programa Escuela de Calidad (Quality Schools
Programme) or PEC intervention, which was implemented in 2001. The PEC intervention empowered
school management and parents to jointly develop a five year quality improvement plan for a
school, and then provided annual grants of up to USD 15,000 to implement these plans (Skoufias and
Shapiro 2006). The involvement of parental associations in implementing and monitoring of the
grant was mandatory. Additionally, in the first four years, 80% of the grant had to be spent on
infrastructure and learning materials, while in the fifth year most of the grant went to teacher
training and development (ibid.).
Although participation in PEC was voluntary and any primary school could participate, the
programme did target disadvantaged urban schools. In 2001, some 2000 schools participated in the
programme (Murnane et al. 2006). By 2004, the number of participating schools had increased to
20,000, which is over 10% of Mexicos public school (Skoufias and Shapiro 2006).

15 of 34

5.1.3. Effects on quality of schooling

The literature examining the effect of decentralization to local governments on the quality of
education is limited, most likely due to the restricted content of the reform implemented. On the
other hand, as the largest SBM reform in Mexican schools, the PEC intervention has been widely
evaluated. Authors not only conclude that it limited corrupt practices and improved school
infrastructure and security (Skoufias and Shapiro 2006; Bruns et al. 2011), but three papers also use
quantitative analysis to argue for a causal relationship between participation in the PEC programme
and improved educational quality indicators.
In the first, Skoufias and Shapiro (2006) use panel data from 2000 to 2003 to examine the
relationship between participation in PEC and educational quality. They adopt an econometric
technique known as difference-in-differences with matching to find statistically significant decreases
in dropout rates of 0.24 points, failure rates of 0.24 points and repetition rates of 0.31 points. In the
second evaluation, Murnane et al. (2006) use a similar econometric methodology but add data for
an additional year of PEC participation to corroborate the previous studys positive findings. They
report that each year of participation in the PEC programme resulted in a decline in the dropout rate
of 0.11 percentage points. Moreover, they also demonstrate that PEC participation facilitated the
largest reduction in dropout rates in states with middle values on the human development index.
Three years of participation reduced dropout rates in these schools by an average of 0.33
percentage points. On the other hand, states with low values on the human development index had
negative values in the initial years that became positive in the later years but was never statistically
significant. The authors posit that this difference may be due to the varying levels of capacity in
education departments of different states. Interestingly, Murnane et al. find no significant
relationship between PEC and student failure rates.
Bando (2010) provides the third evaluation, which not only takes into account dropout outcomes
data up to 2006 but also incorporates data on standardized test scores. Her findings on the former
indicator supports the results of the previous two evaluations PEC participation is related to a
reduction in dropout rates, the magnitude of which increases as schools get more experience with
the programme. Conversely, on average PEC participation is not significantly related to student
achievement. However, schools that participate in PEC for five years show increases in Math and
Spanish scores of 0.09 and 0.07 standard deviations, respectively. Both results, she argues, highlight
that it takes time and experience with SBM reforms before benefits can be realized.
Although it was considered by many as a weak form of SBM-type reform (see Bruns et al. 2009), a
rigorous evaluation of the AGE programme also yielded positive findings. Exploiting a phased roll-out
of the intervention, Gertler et al. (2011) find that AGE reduced the proportion of students failing by
5.4% and of repeated a grade by 4%. They find that impacts are larger in the first three years of
primary school. Through qualitative research, the paper also suggests that the channel for
improvement is the increased participation of parents in decision-making. In addition, the authors
investigate whether the effect of AGE differs by socioeconomic status of communities. They report
that although AGE affects indicators of schooling quality positively in poor communities, it has no
effect in the poorest of communities. They postulate that his may be due to the parents in the
poorest communities lacking the ability and stature to voice and assert their preferences (pp. 11).

16 of 34

5.1.4. Concluding Remarks

The 2012 PISA results for Mexico are encouraging. Both performance in mathematics and equity in
educational opportunities improved dramatically compared to previous years (PISA website).
Decentralization has undoubtedly contributed to this progress, together with other education
initiatives on nationalized testing, teacher training and accreditation, and early education. That said,
55% of students still do not meet the benchmark for mathematics, indicating that there is room for
quality improvement in the education system yet. Much of the literature on the decentralization
scheme from the country indicates that greater autonomy devolution to states and schools over
financing and human resource management may have to ability to yield additional benefits. On the
equity front, targeted SBM programmes for disadvantaged schools have contributed to improving
opportunities for students with a low socioeconomic status. The main challenge in the future will be
to improve overall educational quality without negatively affecting this above average equity.
5.2. Indonesia
By the middle of the 1980s Indonesia had achieved universal primary enrolment, leading many to
proclaim it as a model of excellence for the rest of Asia. Quality of schooling in the country, however,
remained poor with most students leaving the education system without the necessary literacy and
numeracy skills (Behrman et al. 2002). According to the 1999 TIMSS study, the performance of
Indonesias eighth grade students stood below international standards, with an average score in
mathematics of 403 against the benchmark of 500 (TIMSS website).
Besides quality, equity was also a subject of great concern in the pre-decentralization era remote
parts of the populous island nation suffered from a lack of resources such as desks and textbooks
(UNESCO 2006), while disparities in enrolment at the secondary level between the children of the
richest and the poorest 20% of households were a significant 37 percentage points (Kristiansen and
Patrikno 2006). Moreover, one fifth of the countrys districts had junior high participation rates that
stood at lower than 60% (ibid.).
5.2.1. Decentralizing Indonesia
In the aftermath of the Asian Financial Crisis in the late 1990s, the rupiah was devalued, inflation
increased and there was a significant contraction in Indonesias economy. This mounted pressure on
government expenditure for the public education system (Behrman et al. 2002). The crisis also
posed the additional challenges of maintaining enrolment and ensuring progression to higher levels.
In 1999, Indonesia promulgated two ground-breaking laws on decentralization as part of a broader
democratization reform. The nations Big Bang decentralization followed almost three decades of
highly centralized, autocratic rule under Suharto and encompassed sweeping electoral, governance
and constitutional reforms. Importantly, the scheme also passed responsibility of service delivery,
including education, down to over 30 provinces comprising of more than 400 districts. Beginning in
2005, accountability in service delivery was further enhanced by electing, instead of appointing, the
heads of these devolved local governments (Skoufias et al. 2011).
Although many authors contend that the evolving political landscape and growing dissatisfaction
within provinces over power-sharing were the main driving forces behind the decentralization (see
for e.g. Simatupang 2009), international agencies also played a prominent role in its implementation.
17 of 34

Not only was education decentralization on the list of recommendations made by the World Bank to
deal with the issues of low quality and increasing financial pressure of schooling (Behrman et al.
2002), but the broader reform also served as a key condition in the post-crisis IMF rescue package
offered to Indonesia (Kristiansen and Pratiko 2006).
5.2.2. Education decentralization reforms
Although the Indonesian government had already demonstrated its commitment to education by
making nine years of schooling compulsory in the 1990s, the education system had remained
extremely centralized. This of course changed dramatically with the implementation of the
decentralization scheme in 2001 when districts became responsible for establishing new schools and
for setting local education policies (World Bank 2012). As is common in most decentralization
programmes, the Centre continued to maintain control over setting and maintaining national
competency standards, curriculums and education calendars, as well as over implementing
evaluations (King and Guerra 2005).
Unlike the previous case study of Mexico in which the central government did not devolve any
staffing decisions to lower levels of governments, Indonesias reform also made provisions for
sharing the human resource management responsibilities for teachers. District governments were
given the responsibility of employing all teachers in the public schooling sector (with the exception
of those for madrasah schools), as well as the authority to hire and dismiss contract teachers.
Because teachers were civil servants, salary levels and ranks continued to be set centrally.
Nonetheless, district governments could transfer teachers, recommend promotions and provide
supplementary benefits and incentives (UNESCO 2006).
Through regulations and additional directives in 2003 and 2005, Indonesia incorporated important
school-based management principles into its education system (World Bank 2012). Schools were
given the authority to manage operations such planning and budgeting. Teachers themselves were
meant to maintain control over pedagogy, learning plans and the selection of textbooks as long as
they met the minimum standards set centrally (UNESCO 2006). To further establish the autonomy of
the school, block grants were issued by the government that could be used based on a schools own
priorities. Moreover, schools were mandated to form school committees (SC) of parents and
prominent community members. Whereas earlier SCs had only raised funds, committees were now
given the enhanced task of providing input on all school matters, participating in the management of
funds given to schools and collecting additional money in support of education.
Following decentralization, changes were also made to financing. Districts were, for example,
allowed to generate their own taxes and also borrow within limits (UNESCO 2006). This local
financing, nonetheless, remained limited and to date represents only 10% of total local revenues
(Skoufias et al. 2011). Local governments thus continue to rely heavily on central transfers or block
grants, although a move towards formula-based grants to provinces and districts has been made.
5.2.3. Effects on quality of schooling
The evidence base on the impact of Indonesias ambitious reform is thin and contradictory. Most
studies in recent years have focused on improving the fiscal design of the initiative, rather than on
the rigorous evaluation of the outcomes of decentralized service delivery. Simatupang (2009) is an
18 of 34

important exception. Using data from 1994 to 2006, Simatupang reports that more than half of
Indonesias municipalities experienced changes in education outcomes in the post reform period.
She uses a quasi-experimental methodology to find an overall statistically significant reduction in
drop-out rates at the primary and high school post decentralization.
Along a different vein, Kristiansen and Pratikno (2006) study four districts to provide moderate
support for the interventions perceived impact on quality of education. Based on the results of over
500 household surveys, they report that 81% of parents believed the quality of their childrens
schooling had improved after the reform. SC members, school management and district officials
interviewed for a comprehensive World Bank study in 2012 concurred they noted that block grants
had had a positive effect on transition rates to junior high schools, enrolment rates of poor students
and had overall also improved student academic performance.
On the other hand, there are those who show that the impact of decentralization reforms has been
limited to date. Skoufias et al. (2011) for instance find that although electoral reforms have resulted
in higher revenue generation from own sources, more district-level budget surpluses and greater
expenditure on education, there was no significant impact on service delivery outcomes in the two
years following implementation. Where SBM is concerned, the World Bank study mentioned above
yields similarly pessimistic findings. It concludes that the status of SBM implementation in schools
was not statistically correlated with student test scores. Both investigations suggest that a short time
since implementation may be the key reason why benefits of decentralization are not discernible.
5.2.4. Decentralization challenges
Besides a short time since reform, there are several other challenges that may be preventing
decentralizations benefits from being realized in Indonesia. One key issue in the country has been
uneven implementation. Take school councils for instance. Many schools did not have an active
school committee for several years after the reform was implemented (Pradhan et al. 2011). Where
they did exist, members had often been selected instead of being elected, meetings between the
principal and full committee members were rare, and parents tended to be overly deferential
towards school management on decisions (World Bank 2012). Yet, a randomized controlled trial on
the subject suggests that providing block grants together with election of SC members and/or
linkages with a village council can result in significant improvements in achievement (see Pradhan et
al. 2011). This finding indicates that had implementation been stronger across the board, more
improvements in quality would have been likely.
A related implementation concern has been a lack of clarity in the roles and responsibilities of the
different layers in the education bureaucracy. Devolution had originally been propagated as a means
of clarifying the ambiguity and redundancy prevalent in the system previously, yet there remained
considerable overlaps between roles in the sector even after the intervention (Kristiansen and
Pratikno 2006). Even years after reform implementation, several reports indicate that both
government officials and SC members were unclear about their responsibilities (Brodjonergora 2004;
World Bank 2012).
A final area where Indonesia has struggled has been on fiscal design and transparency. Toi (2010)
highlights that the disparity in revenue sharing between local governments actually widened as a
19 of 34

result of decentralization, thus increasing inequity in education. Although transfers to local

governments are now formula-based, imperfections in the formula imply that the equalizing
property of these grants is weak (Hofman and Guerra 2005). A lack of self-sufficiency in revenues
and the inadequacy of the transfer grant from the Centre have further resulted in a notable neglect
of development expenditures at the local level (Brodjonegora 2004). Moreover, according to
Kristiansen and Pratikno (2006), even five years after decentralization no expenditure data had been
made available to local oversight bodies or the civil society, thus resulting in a total lack of
accountability on this front. Due to the lack of transparency, local government officials often also
remained unclear about what level of funding was available to them (King and Guerra 2005).
5.2.5. Concluding Remarks
Despite its promises, decentralization has not been a panacea for Indonesias quality challenges.
Quality of education remains low - the country ranked second last on the PISA 2012 mathematics
achievement test (PISA website). Besides inadequate facilities, scholars highlight that there are
serious issues related to poor teacher attendance and corruption in the bureaucracy (UNESCO 2006;
Kristiansen and Pratikno 2006). More capacity building and training, as well as enhanced
transparency and horizontal accountability may thus be useful in improving decentralizations ability
to address these concerns. On the equity front, even after decentralization inequalities based on
socioeconomic status and geography have persisted (Kristiansen and Pratikno 2006). Although the
Centre has supported the Periphery through special funds allocated to national programmes on
education, they have not been sufficient in addressing the equity challenge. The above suggests that
a lot more remains to be done if Indonesia wants to enhance student learning.
5.3. Kenya
On a continent where coups and revolutions were prevalent, Kenya had witnessed noteworthy
economic and political stability up to the 1990s (Smoke 1993). This stability, together with the
abolishment of school fees in primary schools in 2003 resulted in large increases in enrolment (Bruns
et al. 2011). Gross enrolment ratios at the primary level crossed the 100 mark in 2003, and by 2005
primary net enrolment had risen to approximately 75 (UIS).
The higher enrolment did not imply higher quality according to one of the few assessments
available in the country, in 2006 only three out of 10 children in third grade could read a story in
English or do simple division calculations from the second grade curriculum (Bold et al. 2012).
5.3.1. Early decentralization efforts
At the time of independence in 1963, Kenya inherited a system of local government from the British.
These local governments had functioned primarily during the interim years before independence
(1960 -1963), had performed reasonably well in the provision of basic services, and had been well
supported by a decent revenue base and central government grants (Menon et al. 2008).
The original constitution of the nation thus built on this existing system and provided for strong
provinces that were charged with some responsibilities surrounding delivery of basic services such as
primary education (Rocaboy et al. 2013). However, shortly afterwards, the countrys first President
initiated changes to the Constitution to limit the authority of provinces. Many note that he did this in
order to prevent other ethnic groups from building local power bases that could challenge his rule
20 of 34

(Kremer et al. 2003). Ostensibly though, the government argued that centralization in service
delivery was necessary to not only promote tribal unity in an ethnically fractionalized nation, but
also to address increasing financial pressures and poor service delivery performance (ibid.).
Over time, in spite of both local and international pressure to strengthen them, the local
governments became more and more marginalized (Menon et al. 2008). In 1966, for example, a
commission charged by the President recommended ambitious reforms to empower local
governments. But rather than implementing these reforms, in 1969 the parliament implemented the
Transfer of Functions Act, which abolished most of the financing provided to the Periphery, and took
away their control over primary education (Smoke 1993).
Recommendations made by the IMF to stop the weakening of local governments in the 70s and 80s
continued to be largely ignored by the ruling regimes, and government responsibilities were
increasingly recentralized. The share of spending by local governments to GDP thus fell from 3.26
percent in 1970 to a mere 1.22 in 2000 (Rocaboy et al. 2013). The local governments with their
limited decision-making authority nonetheless continued to exist side by side a deconcentrated
central government system that had been responsible for implementing directives made by the
Centre since independence (Menon et al. 2008).
The centralization of the broader system notwithstanding, education in Kenya had always
incorporated at least some decentralized community control of schools through the institution of
Harambee. Harambee literally translates to Lets pull together and is a movement with precolonial
roots that encouraged local communities to work together to raise funds for schools and other
public goods (Kremer et al. 2003). To promote enrolment in the early years after independence in
particular, the government had made an open commitment indicating that if local communities built
schools and kept them functioning for a short period through Harambee, the Centre would allocate
teachers to these schools. According to Kremer et al. (2003), however, the way this local institution
was used created distortions in the system. They elaborate that although this open commitment
allowed a rapid expansion in education, the disconnect between local authority and central financing
resulted in more than an optimal number of schools and less than an optimal level of financing for
non-teacher inputs. Both of which probably reduced, rather than enhanced, educational quality.
5.3.2. Decentralization in the 2000s
By the 2000s, the system had evolved into a mix of centralized and decentralized elements. Ad hoc
amendments and approvals given had resulted in an inconsistency in the services that had been
devolved to local bodies in different geographies, and redundancy between the local government
and deconcentrated system persisted in many key functions (Smoke 1993; Rocaboy et al. 2013).
Moreover, financial allocations to local bodies were dictated by political decisions rather than a
transparent formula, further adding to the opacity of decision-making. Critics highlighted the related
issues of a poor institutional environment, low capacity building, corruption and weak citizen
participation (Smoke 1993; Menon et al. 2008; Bold et al. 2012).
To address some of these concerns, the Kenya Local Government Reform Programme (KLGRP) was
launched in 1995 with the assistance of donors. Its goal was to streamline the delivery of services
and increase the financial resources available to local governments. By the early 2000s, the initiative
21 of 34

was successful in giving at least some financial independence back to local bodies (Rocaboy et al.
2013). However, there remained restrictions about how local governments could dispose of these
increased funds. A concomitant directive also delegated authority to electoral districts or
constituencies to develop local projects for service delivery under an elected member of parliament.
The move added yet another alternative type of local bureaucracy to compete with local
governments and the deconcentrated administrative system (ibid.).
In 2003, a new government promised a renewed commitment towards devolution. Because the legal
basis for local governments was still grounded in the original 1963 Local Government Act, an
attempt was made by the government to establish a new decentralization framework into the
Constitution in 2005 (Menon et al. 2008). The provision failed to win a popular vote referendum. In
August 2010, yet another Constitution was proposed and approved through a referendum for
gradual implementation this Constitution restores decision-making autonomy to local bodies. A
new local government law was also drafted to clarify the role of local bodies (Rocaboy et al. 2013).
Given these recent changes, the fate of decentralization in the country still remains in flux.
5.3.3. Effects on quality of schooling
Given the status of decentralization in Kenya, very little can be said about the relationship between
the intervention and the quality of education. Understandably, much of the literature on the topic
from the country focuses on design issues and challenges, not on service delivery outcomes.
The one exception is due to Duflo et al. (2007). The authors evaluate a randomized controlled trial of
an additional teacher reform with an SBM component that was implemented with funding
assistance from the World Bank and a regional NGO. The intervention, piloted between 2005 and
2007, provided funds to 140 schools to hire an extra teacher for primary school. PTAs in half the
schools were then randomly selected to receive training on monitoring the contract teacher, as well
as on soliciting performance information. The authors report that students in all the schools that
received an extra teacher performed better than those that did not. Moreover, the effect on test
scores was larger in schools that had received training on monitoring teacher performance, thus
indicating support for SBM-type initiatives. The authors also note a reduction in overall teacher
absence as a result of the intervention.
5.3.4. Concluding Remarks
Some thirty odd constitutional amendments later, Kenyas government is finally decentralizing. Until
recently, local bodies had limited constitutional authority, and the multiple institutional
arrangements that existed at the local level both diluted accountability, and created redundancy in
education service delivery. Yet, an experimental trial does suggest that SBM type reforms have the
potential to improve educational quality in the country. To convert this potential into reality, Kenya
not only needs a strong legal framework for decentralization, but also has to rationalize service
delivery, reduce distortions in the system and substantially enhance the authority of local
governments and schools. Of course many of these initiatives are already underway with the
adoption of the new Constitution and a renewed interest in education decentralization. How
successful these efforts will be in the area of education delivery, however, as yet remains to be seen.

22 of 34

Table 3: Case Studies Factsheet

Mexico

Indonesia

Kenya

Population
116 million (2013 estimate)
GDP per capita (PPP)
USD 15,600 (2012 estimate)
World Bank classification
Middle income
Key Education Statistics (year recorded in brackets)

254 million (2014 estimate)

USD 5,200 (2013 estimate)
Lower middle income

45 million (2014 estimate)

USD 1,800 (2013 estimate)
Low income

Primary GER
Secondary GER
Primary NER
Secondary GER
Survival to end of primary
Achievement statistics

104 (2011)
84 (2011)
96 (2011)
67 (2011)
95 (2010)
PISA 2012 mathematics ranked 53 out of
64 nations tested

109 (2011)
81 (2011)
94 (2011)
75 (2011)
88 (2010)
PISA 2012 mathematics ranked 63 out of 64
nations tested

112 (2009)
60 (2009)
82 (2009)
50 (2009)
78 (2004)
No international comparisons available

1821 from Spanish rule

Federal republic

1945 from Dutch rule

Federal in first few years after independence.
Then unitary republic
Big bang in 2000, with additions through 2000s

Basic Information

Governance and Decentralization

Independence date
Government structure
Decentralization reform date
Education decentralization content

SBM content

Impact on quality

1963 from British rule

Federal in first year after independence.
Then unitary republic
1992, with additions through next two
Original local government act in 1963.
decades
Reforms in flux since then
Weak to Moderate
Moderate
Weak
Local governments responsible for education Local governments responsible for education
Local governments have in the past few
delivery but fiscal and human resource
delivery but fiscal devolution limited
decades had little authority over education
devolution limited
delivery. System highly centralized
Moderate
Moderate
Moderate in specific aspects
AGE and PEC interventions both give grants
Local councils receive BOS funds and are
Harambee system gave communities control
as well as some authority over how to use
allowed to use based on school priorities
over establishing schools. RCT tested a
funds
stronger form of SBM
Evidence base on SBM strong. Indicates that Evidence base thin and contradictory.
Evidence base thin overall.
both AGE and PEC participation enhances
RCT on SBM type intervention of monitoring
schooling quality.
contract teachers suggests improvements in
quality.
Source: Authors own compilation. Statistics are from CIA Factbook, UNESCO Institute of Statistics and PISA websites

23 of 34

6. Prerequisites for success

A large policy literature suggests that decentralization reforms that are incomplete, badly designed,
poorly implemented or lacking in strong political sponsorship are seldom successful (see Litvack and
Seddon 1999; WDR 2004; Caldwell 2005). Conversely, schemes that encourage accountability, are
designed with the local context in mind, build capacity as well as foster key stakeholder buy-in are
the ones that are able to actually enhance service delivery. Empirically, the importance of some of
these design and implementation elements was highlighted in the case studies presented earlier.
In recent years, the empirical evidence has provided support for many of the other prerequisites
suggested in the policy literature. Studies on school decentralization from El Salvador (Jimenez and
Sawada 1999) and Honduras (Di Gropello and Marshall 2005), for example, suggest that community
participation in schools can indeed increase accountability. Research from Bolivia (Faguet 2004)
indicates that community oversight is particularly important for decentralization schemes to be
effective. Yet in many countries such as Pakistan (Cheema 2007) and Indonesia (Pradhan et al. 2011),
many participatory mechanisms remained inactive several years after implementation. Moreover, in
contexts where there is widespread social inequality, chances of elite capture usually increase, thus
limiting reform success (see Prudhomme 1994). Further, inherent characteristics of populations
shall as ethnic diversity have been shown to restrict the efficacy of community participation in
countries like Kenya (Miguel and Gugerty 2005) and Ghana (Akramov and Asante 2009). In order to
ensure that community participation and oversight do in fact translate into greater accountability, it
may be important to take steps to prevent projects from falling into these common pitfalls.
Another prerequisite that has seen growing support from the empirical body of work is capacity
building and training. Research from Kenya (Duflo et al. 2009) and Central America (Di Gropello
2006) indicates that successful SBM applications spent considerable time in training their school
councils. Evidence from Gambia (Blimpo and Evans 2011), Peru (Loayza et al. 2011) and Mexico
(Murnane et al. 2006) also demonstrates that factors such as literacy of the community, the
capability of local government officials, and the level of human development in a region can all
influence whether or not communities can benefit from education decentralization efforts. As a
consequence, governments have been increasing efforts on this front and rightly so.
While it is important that adequate decision-making authority over say financial and personnel
matters be devolved to the appropriate level, decentralization does not imply that the central
government should abandon its role altogether. Rather, much of the empirical evidence points to
the importance of a continued role for the Centre. Besides continued efforts towards capacity
building as mentioned above, another area that probably requires the Centres involvement is
equalization transfers that help to address inequity. Compensatory programmes, such as those
implemented in Mexcio, can go a long way in reducing gaps in opportunities between students of
rich and poor backgrounds. Additionally, there is also evidence that some level of vertical
accountability may actually enhance outcomes. Both Chaudhury et al. (2005) and Olken (2009)
support this line of reasoning, arguing that higher government monitoring is more effective than
local community scrutiny in reducing teacher absenteeism and misuse of infrastructure project
funds, respectively.

24 of 34

Of course, even if the above prerequisites are present, the benefits of decentralization may still
require a certain amount of time and experience before they become evident. Putnam (1993)
argues that decentralization should be evaluated not over years, but over decades. In the years
following decentralization, governments learn by doing, and in this period adjustments in design,
receptivity and participation are common.
Borman et al. (2003) offer evidence in this regard specifically for school decentralization schemes.
They examine over 800 SBM designs in the US to posit that it takes 5 years for changes of the reform
to be institutionalized, and 8 years for SBM to actually affect student learning or outcomes. In a
study of the Chicago SBM reform, Hess (1999) illustrates this very argument, showing that after an
initial fall in scores, student attainment recovered by the fifth year. Evidence from Mexico (Bando
2010) lends further support to this prerequisite, demonstrating that more experience with certain
reforms may improve the ability of participates to benefit from them.

25 of 34

7. Concluding remarks
As this paper has shown, decentralization in the education arena is a popular reform by now
almost all developing nations have experimented with the reform in one form or another. Yet the
size and quality of the evidence linking decentralization to improvements in learning is limited. A
handful of more rigorous evaluations - such as those from Argentina, Brazil, Mexico, Indonesia, and
the Philippines do nonetheless demonstrate that decentralization has the potential to address
quality concerns in developing countries.
Certainly, a lot more empirical research and particularly research with a stronger empirical design is
required before firmer conclusions than the above can be drawn on the relationship between the
two. As the size of these rigorous evaluations grows, research needs to focus on two areas in
particular: first, what specific forms of decentralization work well, and which work less well and
second, what prerequisites allow decentralization to achieve its potential in enhancing learning.
Where the former is concerned, the review indicated that effects may be heterogeneous across
countries. The case studies in turn illustrated how at least some of this heterogeneity may be the
result of reform design and implementation. As of now, however, the scholarship is too small to
draw firm conclusions on specific elements of these two factors. In the latter area of prerequisites,
the empirical evidence is fortunately larger. Thus far it has shown support for community
participation, capacity building and training, a continued role for the Centre, and time and
experience. There are, however, instances when even the presence of these conditions has not
resulted in better quality this suggests that there may be additional factors that are relevant.
Based on the education literature, one important factor that comes immediately to mind is that of
the traditional inputs physical resources such as desks, textbooks and flipcharts. Although
consensus is growing that inputs alone are not enough to enhance learning, more and more scholars
also agree that the effects of inputs on learning may be heterogeneous. In developing countries,
which often lack basic inputs, there may still be a basic level of resources required before
governance reforms can be successful.
In addition, it is important that we do not overlook one of the most critical inputs of all teachers.
Empirical evidence from The Gambia, El Salvador and Brazil has shown that devolving even a little
authority over personnel matters can facilitate reduced teacher absenteeism. By contrast, research
from countries such as Mexico has demonstrated that teachers unions can have a significant impact
on the content and implementation of reforms. Together, these factors suggest that a key agenda
item post 2015 needs to be initiatives that not only better incorporate teachers into reforms, but
also those that teach local government officials and community members how teacher performance
can best be managed for better results. Even developed nations are now increasingly focusing on
training, motivating and professionalizing teachers in order to ensure that those at the frontline are
up to the task. At this critical juncture for EFA, it is important that developing countries follow suit.
Finally, many countries that have decentralized education have also struggled with widening gaps
between different geographies, as well as those between students from different social
backgrounds. Needless to say, as countries deepen their decentralization efforts, initiatives that can
narrow such gaps and thus enhance equity must sit firmly at the top of the post 2015 agenda.
26 of 34

8. References
Ahmad, E. and G. Brosio. 2009. Does decentralization enhance service delivery and poverty reduction?
Cheltenham, UK: Edward Elgar.
Akramov, K., and F. Asante. 2008. Decentralization and local public services in Ghana: Do geography and ethnic
diversity matter? IFPRI GSSP Background paper 16.
Angrist, J. and S. Pischke. 2008. Mostly harmless econometrics: An empiricist's companion. Princeton: Princeton
University Press.
Barrera-Osorio, F., H. Patrinos and T. Fasih. 2009. Decentralized Decision-Making in Schools. Washington DC:
World Bank.
Bando, R. 2010. The Effect of School Based Management on Parent Behavior and the Quality of Education in
Mexicos. PhD Thesis, University of California, Berkeley.
Behrman, J., A. Deolalikar, and L. Soon. 2002. Promoting effective schooling through education decentralization
in Bangladesh, Indonesia, and Philippines (No. 23). Manila: Asian Development Bank.
Blimpo, M. and D. Evans. 2011. School-based management and educational outcomes: lessons from a
randomized field experiment. Unpublished manscript.
Bold, T., M. Kimenyi, G. Mwabu, and P. Welcome. 2012. Interventions & Institutions Experimental Evidence on
Scaling up Education Reforms in Kenya. Unpublished manuscript
Borman, G., G. Hewes, L. Overman and S. Brown. 2003. Comprehensive school reform and achievement: A
meta-analysis. Review of Educational Research 73(2):125.
Brodjonegoro, Bambang (2004). Three Years of Fiscal Decentralization in Indonesia: Its Impacts on Regional
Economic Development and Fiscal Sustainablitiy. presented at the Symposium on Fiscal Decentralization.
Bruns, B., D. Filmer and H. Patrinos. 2011. Making schools work: New evidence on accountability reforms.
Washington DC: The World Bank
Caldwell, B. 2005. School-based management (Vol. 3). Paris: UNESCO Institute for Educational Planning.
Channa, A. and JP Faguet. 2012. Decentralization of health and education in developing countries: a qualityadjusted review of the empirical literature. Economic organisation and public policy discussion papers 38
Chaudhury, N., J. Hammer, M. Kremer, K. Muralidharan and H. Rogers. 2005. Teachers and health care
providers absenteeism: a multi-country study. Washington DC: World Bank Development Research Group.
Cheema, A. 2007. Governance Impediments to Pro-Poor change in Pakistan. Asian Development Bank Paper.
CIA Factbook. https://www.cia.gov/library/publications/the-world-factbook/ Accessed March 18, 2014.
De Grauwe, A. 2005. School-based management (SBM): does it improve quality?" Background paper for EFA
Global Monitoring Report 2005.
Di Gropello, E. 2002. An assessment of the impact of decentralization on the quality of education in Chile.
Washington DC: World Bank
Di Gropello, E. 2006. A comparative analysis of school-based management in Central America. Washington DC:
World Bank Publications.
Di Gropello, E. and J. Marshall. 2005. Teacher effort and schooling outcomes in rural Honduras. In Incentives
to Improve Teaching, ed. E. Vegas. Washington DC: World Bank.
Duflo, E., P. Dupas, and M. Kremer. 2008. Peer effects, pupil-teacher ratios, and teacher incentives: Evidence
from a randomized evaluation in Kenya. National Bureau of Economic Research Paper No. 14475.
Eaton, K., K. Kaiser and P. Smoke. 2010. The Political Economy of Decentralization Reforms: Implications for aid
effectiveness. Washington DC: World Bank.
Eskeland, G and D. Filmer. 2007. Autonomy, Participation and Learning: Findings from Argentine Schools, and
Implications for Decentralization" Education Economics Vol. 15, No. 1, March pp. 103-127.
Faguet, JP. 2004. Does Decentralization Increase Responsiveness to Local Needs? Evidence from Bolivia.
Journal of Public Economics, 88(4): 867-894.
Fiske, E. 1996. Decentralization of education: Politics and consensus. Washington DC: World Bank

27 of 34

Freinkman, L. and A. Plekhanov. 2009. Fiscal decentralization in rentier regions: Evidence from Russia. World
Development 37(2):503-512.
Galiani, S., P. Gertler and E. Schargrodsky. 2008. School decentralization: Helping the good get better, but
leaving the poor behind. Journal of Public Economics 92(10-11):2106-2120.
Gershberg, A. 1999. Education Decentralization Processes in Mexico and Nicaragua: legislative versus
ministry-led reform strategies. Comparative Education, 35(1), 63-80.
Gershberg, A. and D. Winkler. 2003. Education decentralization in Africa: A review of recent policy and practice.
Washington DC: World Bank.
Gertler, P., M. Rubio-Codina, and H. Patrinos. 2012. Empowering Parents to Improve Education: Evidence
from Rural Mexico. Journal of Development Studies. 99(1), 68-79
Gertler, P., H. Patrinos and M. Rubio-Codina. 2007. "Methodological Issues in the Evaluation of School-Based
Management Reforms." Unpublished manuscript, World Bank
Glewwe, P. and M. Kremer. 2006. Schools, teachers, and education outcomes in developing countries.
Handbook of the Economics of Education, 2, 945-1017.
Glewwe, P. and E. Maga. 2011. The Impacts of School Management Reforms in Madagascar: Do the Impacts
Vary by Teacher Type? Available at http://www.povertyactionlab.org/publication/impacts-schoolmanagement-reforms-madagascar-do-impacts-vary-teacher-type
Grindle, M. 2007. Going local: decentralization, democratization, and the promise of good governance.
Princeton University Press.
Gunnarsson, V., P. Orazem, M. Snchez and A. Verdisco. 2009. Does Local School Control Raise Student
Outcomes? Evidence on the Roles of School Autonomy and Parental Participation. Economic Development
and Cultural Change 58(1):25-52.
Hanushek, E. 1995. Interpreting recent research on schooling in developing countries. The World Bank
Research Observer, 10(2), 227-246.
Hanushek, E., S. Link and L. Woessman. 2011. Does School Autonomy Make Sense Everywhere? Panel
Estimates from PISA. National Bureau of Economic Research Paper No. w17591.
Hagerstrom, M. 2006. Basic Education in Decentralized service delivery for the poor: Background paper.
Washington DC: World Bank
Hess, G. 1999. Understanding achievement (and other) changes under Chicago school reform. Educational
Evaluation and Policy Analysis 21(1):67.
Hofman, B. and S. Guerra. 2005. Fiscal Disparities in East Asia: How Large and Do They Matter?. In East Asia
Decentralizes: Making Local Government Work. Washington DC: World Bank.
Jimenez, E. and Y. Sawada. 1999. "Do community-managed schools work? An evaluation of El Salvador's
EDUCO program." The World Bank Economic Review 13(3):415.
Khattri, N., C. Ling, and S. Jha. 2010. The effects of school-based management in the Philippines: An initial
assessment using administrative data. World Bank Policy Research Working Paper 5248.
King, E. and S. Guerra. 2005. Education Reforms in East Asia: Policy, Process and Impact. In East Asia
Decentralizes: Making Local Government Work. Washington DC: World Bank.
King, E. and B. Ozler. 2000. Whats decentralization got to do with learning? Endogenous school quality and
student performance in Nicaragua. Development Research Group, World Bank.
Kremer, M., S. Moulin and R. Namunyu. 2003. Decentralization: A cautionary tale. Poverty Action Lab Paper 10.
Kristiansen, S. and Pratikno. 2006. Decentralising education in Indonesia. International Journal of Educational
Development, 26(5), 513-531.
Leithwood, K. and T. Menzies. 1998. Forms and effects of school-based management: A review. Educational
policy, 12(3), 325-346.
Litvack, J., J. Ahmad and R. Bird. 1998. Rethinking decentralization in developing countries. Washington DC:
World Bank.
Litvack, J. and J. Seddon. 1999. Decentralization briefing notes. Washington DC: The World Bank.

28 of 34

Loayza, N., J. Rigolini, V. Jamele and O. Calvo-Gonzalez. 2011. More than you can handle: decentralization and
spending ability of Peruvian municipalities. World Bank Policy Research Working Paper No 5763.
Washington DC: World Bank.
Malen, B., R. Ogawa, and J. Kranz. 1990. What do we know about school-based management? A case study of
the literature-A call for research. Choice and control in American education 2:289-342.
Menon, B. , J. Mutero and S. Macharia. 2008. Decentralization and Local Governments in Kenya. International
Studies Programme Working Paper 08-32. Georgia State University
Miguel, E. and M. K. Gugerty. 2005. Ethnic diversity, social sanctions, and public goods in Kenya. Journal of
Public Economics 89(11-12):2325-2368.
Murillo, M. 1999. Recovering political dynamics: Teachers' unions and the decentralization of education in
Argentina and Mexico. Journal of Inter-American Studies and World Affairs, 41(1), 31-57.
Murnane, R., J. Willet, and S. Cardenas. 2006. Did participation of schools in Programa Escuelas de Calidad
(PEC) influence student outcomes? Working Paper, Harvard University Graduate School of Education
Oates, W. E. 1972. Fiscal federalism. New York: Harcourt Brace Jovanovich.
Olken, B. A. 2009. Corruption perceptions vs. corruption reality. Journal of Public Economics 93(7-8):950-964.
Ornelas, C. 2000. The politics of the educational decentralization in Mexico. Journal of Educational
Administration, 38(5), 426-442.
Paes de Barros, R. and R. Mendonca. 1998. The impact of three institutional innovations in Brazilian
education. Organization matters: agency problems in health and education in Latin America: 75.
Parker, C. 2005. Teacher Incentives and Student Achievement in Nicaraguan. In Incentives to Improve
Teaching, edited by E. Vegas. Washington DC: World Bank.
PISA Website. http://www.oecd.org/pisa/ Accessed March 05, 2014.
Pradhan, M., D. Suryadarma, A. Beatty, M. Wong, A. Alishjabana and A.Gaduh. 2011. Improving educational
quality through enhancing community participation: results from a randomized field experiment in
Indonesia. World Bank Policy Research Working Paper 5795. Washington DC: World Bank
Prudhomme, R. 1995. On the Dangers of Decentralization. World Bank Research Observer 10:210
Putnam, R. D. 1993. Making Democracy Work: Civic Traditions in Modern Italy. Princeton University Press.
Rocaboy, Y., F. Valliancourt and R. Hugounenq. 2013. Public Finances of Local Government in Kenya in The
Political Economy of Decentralization in Sub-Saharan Africa: A New Implementation Model in Burkina Faso,
Ghana, Kenya and Senegal, eds. Dafflon and Madies. Washington DC: World Bank.
Ravallion, M. 1999. The mystery of the vanishing benefits: Ms speedy analyst's introduction to evaluation.
World Bank Publications Vol. 2153.
Rondinelli, D. A., G. S. Cheema and J. Nellis. 1983. Decentralization in Developing Countries: A Review of
Recent Experience. World Bank Staff Working Paper No.581. Washington, DC: World Bank.
Sawada, Y. and A. Ragatz. 2005. Decentralization of education, teacher behavior, and outcomes. In Incentives
to Improve Teaching, ed. E. Vegas. Washington DC: World Bank.
Shah, A., T. Thompson, and H.Zou. 2004. The impact of decentralization on service delivery, corruption, fiscal
management and growth in developing and emerging market economies: A synthesis of empirical
evidence. CESifo DICE Report, 1/2004: 10-14.
Simatupang, R. 2009. Evaluation of decentralization outcomes in Indonesia: Analysis of health and education
sectors. PhD Dissertation. Georgia State University.
Skoufias, E. and J. Shapiro. 2006. Evaluating the impact of Mexico's quality schools program: the pitfalls of
using nonexperimental data. World Bank Policy Research Working Paper.
Skoufias, E., A. Narayan, K. Kaiser, and B. Dasgupta. 2011. Electoral Accountability, Fiscal Decentralization and
Service Delivery in Indonesia. Policy and Research Series Paper 5614. Washington DC: World Bank.
Smoke, P. 1993. Local government fiscal reform in developing countries: Lessons from Kenya. World
Development, 21(6), 901-923.
Tanzi, V. 1995. Fiscal federalism and efficiency: a review of some efficiency and macroeconomic aspects.
Tiebout, C. 1956. A pure theory of local expenditures. The Journal of Political Economy 64 (5):416-424.

29 of 34

TIMSS website. http://timssandpirls.bc.edu/ Accessed March 01, 2014.

Toi, A. 2010. An empirical study of the effects of decentralization in Indonesian junior secondary education.
Educational Research for Policy and Practice, 9(2), 107-125.
Treisman, D. 2002. Decentralization and the Quality of Government. Unpublished paper, Department of
Political Science, UCLA.
UNESCO. 2006. Decentralization of Education in Indonesia. Unpublished manuscript, UNESCO
UNESCO Institute of Statistics (UIS) Website. http://www.uis.unesco.org/Pages/default.aspx Accessed March
15, 2014.
Winkler, D. 1989. Decentralization in Education. Washington, DC: World Bank
World Development Report 2004. Making Services Work for Poor People. Washington DC: World Bank
World Bank. 2008. Decentralization in Client Countries: An Evaluation of the World Bank Support, 1990-2007.
Washington DC: World Bank Publications
World Bank. 2012. Implementation of School-based Management in Indonesia. Jakarta: Indonesia

30 of 34

9. Appendix
Table 4: Summary of Selected Evidence on Decentralization and Quality
Country

Authors

Date

Type

Programme

Method of Analysis

Sample

Results

3.1. Studies using Randomized Controlled Trials

Kenya

Duflo et al.
(2007)

2005

SBM

Randomized trial which gave

school councils money,
autonomy and training to hire
extra teachers and monitor their
performance

Comparison of treatment
and control groups.

21k students from 210

schools

SBM associated with

Increase in Mathematics scores of
0.24 standard deviations
Increase in Language scores of 0.18
standard deviations

Indonesia

Pradhan et
al. (2011)

2007

SBM

Randomized trial which gave

school councils a grant, in
combination with training, help
with school council elections
and/or linkages to village council.

Comparison of treatment
and control groups.

520 schools

No impact of training alone

Village linkage associated with
increase in test scores of 0.17 standard
deviations
Village linkage and elections
associated with increase in test scores
of 0.22 standard deviations

The Gambia

Blimpo and
Evans (2011)

SBM

Randomized trial which gave

school councils a grant in
combination with training

Comparison of treatment
and control groups.

237 primary schools

No significant association with test

scores
20% reduction in student
absenteeism
23% reduction in teacher
absenteeism

Madagascar

Glewwe and
Maiga (2011)

SBM

Randomized trial in which

materials, training and greater
accountability is given to three
levels of districts, sub-districts
and schools

Comparison of treatment
and control groups.

20k students from 30

districts over period
2006 to 2007

No significant association with test

scores

Difference in differences

Almost all secondary

schools over period
1994 to 1999

D associated with higher Math and

Spanish scores
Increase of 4.9% and 6.9% in scores,
respectively

2005

3.2. Studies using Quasi-experimental Techniques or Panel Estimations

Argentina

Galiani et al.
(2008)

1991

LG

Decentralized financing, staff

management and budgeting to
LG

31 of 34

Table 4: Summary of Selected Evidence on Decentralization and Quality

Country
Mexico

Authors
Gertler et al.
(2011)

Date
1996

Type
SBM

Programme
Small grants to parent councils
and parental training targeted at
disadvantaged areas AGE

Method of Analysis
Difference in differences

Sample
30,000 students from
6,000 schools over
1997 - 2001

Mexico

Skoufias and
Shapiro
(2006)

2001

SBM

Annual grants of up to USD 15k

given to schools/SMCs to
improve education quality - PEC

Difference-in-differences
with matching

75000 schools over

period 2001 2003

Mexico

Murnane et
al. (2006)

2001

SBM

Annual grants of up to USD 15k

given to schools/SMCs to
improve education quality PEC

Difference in differences

Data from 2001 - 2004

Mexico

Bando
(2010)

2001

SBM

Annual grants of up to USD 15k

given to schools/SMCs to
improve education quality PEC

Fixed effects on panel

data

Data from 2001 to

2006

Brazil

Paes de
Barros and
Mendonca
(1998)

1982

SBM

SBM with three key innovations:

Financial autonomy of schools
Ability to elect principals
Presence of school councils

Difference in differences state-level

18 states over period

1981 - 1993

Philippines

Khattri et al.
(2010)

2003

SBM

Training and direct funding for

school improvement

Difference in differences
with matching

5k schools from 23
districts over 2003 to
2005

32 of 34

Results
SBM associated with
Reduction in failure rates by 5.4%
Reduction in repetition rates by 4%
No impact on drop-out rates
SBM associated with
0.24 lower dropout rates
0.24 lower failure rates
0.31 lower repetition rates
Each year of participation resulted in
decline in dropout rate of 0.11
percentage points
No impact on failure rates
Five years of participation is
associated with increases in Math
scores of 0.09 standard deviations
Five years of participation is
associated with increases in Spanish
scores of 0.07 standard deviations
Lower repetition rates associated
with financial autonomy
Lower mean grade level lag
associated with financial autonomy
and school councils
Lower proportion of students with lag
associated with school council
presence.
SBM associated with
1.45 percentage points overall
improvement
1.82 percentage points improvement
in Science
1.32 percentage points improvement
in English
1.88 percentage points improvement
in Mathematics

Table 4: Summary of Selected Evidence on Decentralization and Quality

Country
Crosscountry

Authors
Gunnarsson
et al. (2009)

Date
Various

Type
SBM

Programme
Various

Method of Analysis
Instrumental variable

Sample
17k students from 10
Latam countries - 1997
survey data

Results
Autonomy associated with lower test
scores
Participation associated with higher
test scores

Crosscountry

Hanushek et
al. (2011)

Various

SBM

Various

Regression with country

fixed effects

1mn students from 42

countries - 4 waves of
PISA from 2000 to
2009

Overall negative association between

autonomy and scores

3.3. Studies using Cross-sectional Data Regression Analysis

Russia

Freinkman
and
Plekhanov
(2009)

Phased
beginning
in 1994

LG

Increased fiscal powers with rulebased transfers, responsibility for

public services

Regression analysis using a

between effects model

Secondary school
results from 73 out of
83 regions, tested in
2004 and 2005

D associated with higher test scores

Crosscountry

Treisman
(2002)

Various

LG

Various

Regression analysis

D associated with higher youth

illiteracy

Chile

Di Gropello
(2002)

Early
1980s

LG

Some increase in devolved funds

to LG, greater responsibility for
public services. In 1990s, greater
pedagogical devolution to schools

Regression analysis

Up to 166 countries
with cross-sectional
data collected from
mid-90s
50 municipalities (out
of 355) - Student tests
conducted in 1996

El Salvador

Jimenez and
Sawada
(1999)

SBM

Community schools where SMCs

can hire/ fire teachers, manage
school funds and maintain
infrastructure

Regression analysis with

Heckman correction
model

605 3rd grade students

from 162
municipalities - data
from 1996

No association with Math or English

test scores
Students in EDUCO schools have
lower absenteeism

1991

33 of 34

Municipal financial autonomy not

significant
Municipal training spend and wage
incentives positively associated with
test scores
School involvement in financial
decision-making positively associated
with test scores
School pedagogical and curricular
autonomy positively associated with
test scores

Table 4: Summary of Selected Evidence on Decentralization and Quality

Country
El Salvador

Authors
Sawada and
Ragatz
(2005)

Date
1991

Type
SBM

Programme
Community schools where SMCs
can hire/ fire teachers, manage
school funds and maintain
infrastructure

Method of Analysis
Matching

Sample
605 3rd grade students
from 162
municipalities - data
from 1996

Results
No association with scores
Lower teacher absenteeism in EDUCO
schools

Nicaragua

King and
Ozler (2000)

1991

SBM

Autonomous schools with SMCs

that can hire/ fire teachers,
manage school funds and
maintain infrastructure

Matching

3000 students from

primary and secondary
schools over period
1995 1997

No impact of de jure Autonomy

Positive association of de facto
Autonomy with Math and Spanish, no
association with Language

Nicaragua

Parker
(2005)

1991

SBM

Autonomous schools with SMCs

that can hire/ fire teachers,
manage school funds and
maintain infrastructure

Matching

1000 3rd and 6th grade

students - tested in
2002

SBM associated with

Higher third grade Math scores
Lower sixth grade Math scores
No association with Spanish scores

Honduras

Di Gropello
and Marshall
(2005)

1999

SBM

Community schools where SMCs

can hire/ fire teachers, manage
school funds and maintain
infrastructure

Regression analysis with

Heckman correction
model

200 rural schools

tested in 2002 and
2003

SBM associated with

Higher science scores but with no
change in Math or Language test
scores
Marginally lower dropout rates

Argentina

Eskeland and
Filmer
(2007)

1978

SBM

Schools choose textbooks and

teaching methods.

Regression analysis

24000 6th and 7th

grade students from
urban schools

SBM associated with

Higher Math but no change in
Language scores
Effect is stronger for poorer
households

34 of 34