Education Reforms in East Asia: Policy, Process, and Impactsiteresources.worldbank.org/INTEAPDECEN/Resources/Chapter-9.pdf · Education Reforms in East Asia: Policy, ... and the political

In most developing countries, responsibility forproviding primary and secondary education hasresided with the central government. However, agrowing number of countries throughout theworld, including those in East Asia, are transfer-ring this responsibility away from the center, typi-cally as part of a broader reform to decentralizegovernment functions. This transfer has taken var-ious forms, including devolving fiscal responsibil-ity and management to lower levels of govern-ment, making public schools autonomous,requiring the participation of communities inoperating schools, expanding community financ-ing, allowing families to choose their schools, andstimulating private provision of education. Theimpetus for decentralization has often been politi-cal or financial rather than educational, yet sup-porters of decentralization would argue that it canaddress difficult problems confronting educationsystems, especially those relating to performanceand accountability. Education systems areextremely demanding of the managerial, technical,and financial capacity of governments, so thepotential returns to making such systems moreefficient and effective are great.

The promise of decentralization lies in givingmore voice and power to local leaders and school

179

9

Education Reformsin East Asia:

Policy, Process, and Impact

Elizabeth M. King and Susana Cordeiro Guerra

personnel, who presumably know more about localeducational problems than national officials, andwho have an incentive to lobby for more resourcesand to innovate. Indeed, as the broader decentraliza-tion literature suggests, the benefits of decentraliza-tion lie in reinforcing accountability among thoseresponsible for delivering services—between the cen-tral government and local governments, betweengovernments and school personnel, and betweenschool personnel and the communities they serve(Ahmad et al. 1998).1 In countries as large and diverseas China and Indonesia, generating local solutions toeducational problems and mobilizing local energiesand resources can yield dividends for all.

Despite its promises, however, decentralizationis not a policy panacea. As this chapter shows,choosing an appropriate design for transformingan education system is difficult. What’s more, thereform process is never smooth. It is likely to bepunctuated by bursts of progress and frequent set-backs, which may lead to rising frustration andgrowing mistrust among stakeholders who seethemselves as losers under the reform process.

This chapter reviews the experiences of EastAsian countries in decentralizing their educationsystems, with the goal of understanding thechallenges of designing reforms, distilling lessons

east_179-208.qxd 5/17/05 11:43 AM Page 179

180 East Asia Decentralizes

on implementation, and examining the impact oneducational development. Decentralization lawstypically stipulate dramatic reallocations ofauthority and responsibility among levels of gov-ernment and also transfers of resources. However,the experience in East Asia—and, indeed, innearly all countries that have decentralized—sug-gests a lack of congruence between design andimplementation, or between de jure and de factodecentralization.

Several factors have given rise to this incongru-ence: incomplete design and implementation lags,which may be due to weak technical and adminis-trative capacity, and lack of broad political supportfor reform. For example, central agencies are notshy about transferring responsibilities for financingand delivering education services to local govern-ments but are not as eager to share correspondingauthority and resources, and so find ways ofreasserting control. Local governments that aresupposed to yield some decision-making authorityto schools may also hold back from doing so.Indeed, two common challenges are to align func-tions, powers, and resources among levels of gov-ernments, and to define an appropriate role for thecentral authority within a decentralized system.Achieving a better alignment of functions, powers,and resources is primarily a matter of improvingdesign in some countries, and of improving imple-mentation in others.

The next section examines the rationale fordecentralization in East Asian countries. Theprincipal motives rarely relate to expanding orimproving public services, so the allocation offunctions and resources often does not provide acoordinated framework for managing servicesmore effectively. The third section reviews thenature and design of education reforms in thesecountries, as well as their implementation. This sec-tion focuses on the overall legislative framework:how decentralization has changed governance andmanagement; which responsibilities and functionscountries have devolved; whether resources areadequate to act on these; whether the structure ofthe system is aligned with the changes; and whatfunctions the central agency has retained. EastAsian countries reveal common design features butalso important differences, emerging partly fromdifferences in motivation for reform, initialconditions, and the political milieu.

The fourth section reviews evidence on theimpact of decentralization and the factors that haveinfluenced its effectiveness. Because educationaldevelopment is rarely the rationale for decentral-ization, there is no guarantee that the reform will,in fact, improve education outcomes. With theexception of China, East Asia’s experience withdecentralizing education is fairly recent and re-search on its impact nascent, so the review focuseson shifts in education expenditures and on inequal-ity, and then relies on lessons from around theworld to evaluate the impact of decentralization onlearning. The final section summarizes key findingsand lessons about decentralization given experi-ences in the East Asian countries.

The Impetus for DecentralizingEducation

Educational achievement in parts of East Asia ismuch admired. Japan, Korea, Singapore, Taiwan,and Hong Kong have achieved high enrollmentrates and high-quality education, with their stu-dents consistently topping international tests (seetable 9.1) (Martin et al. 2004a and 2004b). OtherEast Asian countries have not done as well, but they,too, have achieved high enrollment rates (see figure9.1). These countries face other educational chal-lenges: The emerging economies of China, Indone-sia, Malaysia, the Philippines, and Thailand mustbetter manage their education systems to reducedisparities between wealthier and poorer regions,and to improve the overall quality of education. Thepoorer countries—Laos, Cambodia, and Papua NewGuinea—must expand the number of children whoenter school, cut the number who drop out at theprimary level, ensure that the system producesenough talent to support economic growth anddevelopment, and address difficult problems infinancing and managing their education systems.

These challenges, however, have not been the pri-mary rationale and main driving force behind effortsto decentralize the education systems in these coun-tries. Rather, political factors and fiscal concerns havebeen the impetus.2 Key design aspects of reform—including central-local transfers, local tax authority,and civil service rules—may therefore ignore legal,financial, and administrative issues that are critical forachieving national education goals, and may establishstructures and incentives that imperil those goals.

east_179-208.qxd 5/17/05 11:43 AM Page 180

In China, decentralization of education can betraced to the decollectivization and economic liber-alization reforms of the 1970s, which laid thegroundwork for transferring responsibility forsocial services to local governments. Fiscal con-straints on the central government were also seen asa primary motive for that transfer (Hawkins 2000;Bray 1999; Cheng 1997). In Indonesia, politicalfactors—a national call for democracy, the end ofthe Soeharto regime, the failures of the highly cen-tralistic government, intensified by the financialcrisis of 1997—drove the decision to decentralize

all but a few sectors in 1999 (World Bank 2003a). Inthe Philippines, the 1987 Constitution mandateddecentralization, and the 1991 Local GovernmentCode provided legal guidelines for transferringresponsibility for providing services to subnationalgovernments. Except for the transfer of construc-tion and maintenance of school buildings to localgovernments, however, the Philippines has notformally decentralized governance of elementaryeducation. Political considerations underlie thisexception. One often-cited reason is that publicschoolteachers have traditionally counted votes

Education Reforms in East Asia: Policy, Process, and Impact 181

TABLE 9.1 Student Performance on Mathematics and Science Tests(ranking among 38 countries)

Mathematics score and rank Science score and rank

Country 1999 2003 1999 2003

Singapore 604 (1) 605 (1) 568 (2) 578 (1)Korea 587 (2) 589 (2) 549 (5) 558 (3)Taiwan 585 (3) — 569 (1) —Hong Kong 582 (4) 586 (3) 530 (15) 556 (4)Japan 579 (5) 570 (5) 550 (4) 552 (6)Malaysia 519 (16) 508 (13) 492 (22) 510 (21)Thailand 467 (27) — 482 (24) —Indonesia 403 (34) 411(35) 435 (32) 420 (27)Philippines 348 (36) 378 (42) 345 (36) 377 (43)

Sources: Trends in International Mathematics and Science Study 1999 and Martin et al. 2004a and 2004b.Note: — � not available. Scores reported are for eighth grade. Ranking is among 38 countries (1999) and46 countries (2003).

FIGURE 9.1 Net Enrollment Rates in East Asia, 2000

0

10

20

40

60

80

100

30

50

per

cent

70

90

Cambo

dia

Lao

PDR

Mon

golia

China

Philip

pines

Thail

and

Mala

ysia

Kore

a, Re

p. o

f

Vietn

am

Note: Countries are listed according to their gross domestic product per capita for 2002—from Cambodia toRepublic of Korea. Data on Thailand’s and China’s net enrollment rates at the secondary level are not available.

Sources: UNESCO 2002/2003; World Bank 2003b.

Indo

nesia

primary level secondary level

east_179-208.qxd 5/17/05 11:43 AM Page 181

during elections, so decentralization would makethem vulnerable to local politics, possibly compro-mising election results (Loehr and Manasan 1999).3

In Thailand, decentralization is said to resultfrom the groundswell of support for greater democ-racy, shared powers and resources between the cen-tral government and local levels, and greateraccountability, culminating in the 1997 Constitu-tion (Mutebi 2003; Weist 2001). The motivation fordecentralization in Cambodia was also predomi-nantly political: building democratic governance ina country ruled by centralized power for most of itsmodern history. Because the regime that emergedfrom a long civil war was marked by rigid organiza-tion, inefficiency, leakage of funds, budget allocationdifficulties, and little community participation, civilsociety and the development community pushed todeconcentrate government functions to improveservice delivery, especially for the poor (RoyalKingdom of Cambodia 2001).

The Design and Practice of Education Decentralization

The design of decentralized education across EastAsian countries reflects common features. One isthat devolved education systems rest on multilay-ered governance and management structures, withthe result that forging a coherent national policyrequires a much larger effort. Central and interme-diate (provincial, state, municipality, and district)levels of government generally continue to governpost-basic education, but the lowest level ofgovernment, and even schools themselves, governbasic education.

China’s policy stipulates multiple layers of edu-cational supervision involving the National Educa-tional Supervision Agency as well as correspondingagencies in local governments (Hawkins 2000;Wang 2004). The provincial level takes responsibil-ity for developing specific local policies and regula-tions in line with national education objectives.The local government—the township level in ruralareas (the lowest level of the bureaucracy withouteducation offices), and the district level in urbansettings (with education offices)—has responsibil-ity for ensuring that all children receive nine yearsof compulsory education. Earlier implementationrevealed inadequate capacity of township govern-ments to manage schools, so local responsibility for

financing and managing basic education in ruralareas was transferred from township to the countylevel in 2001.4 In 2002, the People’s Congress passedthe Private Education Promotion Law, whichdefined the legal status as well as the rights andresponsibilities of the private sector, further open-ing the door for diversified provision and multiplesources of funding for education (Wang 2004).

In Indonesia, Laws 22 and 25 of 1999 transferredgovernance and management of primary and jun-ior secondary education to district governments,and the upper secondary level to provincial govern-ments, while the central government retains con-trol of the tertiary level. The Education Law 20 of2003 takes decentralization a step further, movingcontrol of basic levels of education from districts toschools (World Bank 2004a). In Cambodia, recentlaws have transferred functions and powers—including the provision of public services—to com-munes, and the country plans to boost accountabil-ity further by increasing the“operational autonomy”of schools and postsecondary institutions (RoyalKingdom of Cambodia 2001).5

A second common feature of decentralizededucation across East Asia is that, at the deepestlevel, the vehicles for governance and manage-ment are typically community councils andschool committees involving local officials, civicleaders, and parents. In Thailand, each school issupposed to have a board composed of represen-tatives of parents, teachers, community organiza-tions, alumni, and students. Parents’ organizationswith jurisdiction over schools are to establish aquality assurance system, and communities areurged to “participate in educational provision bycontributing their experience, knowledge, expert-ise and local wisdom for educational benefits”(Kingdom of Thailand 1999). In Indonesia, eachschool is supposed to have a School Committee—declared an independent body by the 2003 Educa-tion Law—to provide advice, direction, and sup-port for managing schools (Government ofIndonesia 2003). In China, school principals arecharged with greater responsibility than in thepast but also enjoy more autonomy. They areexpected to generate additional resources for theschool and ensure teaching quality, because theycan choose teachers without much interventionfrom the district or county, as well as determineincentives for teachers (Wang 2004).


east_179-208.qxd 5/17/05 11:43 AM Page 182

Such deep decentralization is common outsidethe region, too, a means not only of mobilizinglocal resources but also of fostering greateraccountability and better performance. In Brazil,reform in several states has entailed establishingschool councils, allowing the direct transfer ofresources to schools, and giving communities thepower to elect the local principal. El Salvador’sCommunity-Managed Schools Program—betterknown as EDUCO (Educación con Participación dela Comunidad)—transferred management of eachpreprimary and primary school to an elected Com-munity Education Association composed mostly ofparents and other community members (Jimenezand Sawada 1999). These councils are legallyresponsible for running the schools, raising funds,and hiring and firing teachers, with the goal ofimproving accountability, attendance, and achieve-ment. Nicaragua’s school autonomy reform gaveschool councils—composed of principals, parents,teachers, and students—the authority to hire andfire teachers, veto power over principals’ decisions,and discretion over the sanctions of the Ministry ofEducation and the obligations of teachers and stu-dents (King and Ozler 1998). In Australia, undersite-based management reform, school councilsdevelop a school charter, which is a contractbetween the school and the government, while par-ents play a supervisory role through the council(Pascoe and Pascoe 1998). Overall, internationalexperience suggests that deep decentralization withempowered, accountable schools presents the bestopportunity for improving schools.

One important difference between experience inLatin America and East Asia, however, is that schoolcouncils in Latin America elect their members fromthe community, whereas this is not yet the practicein East Asian countries. Elected council memberscan truly represent the interests of the communityand provide built-in accountability. In East Asia,the duties and legal powers of school committees—in general and relative to school principals, whotend to sit on the committees—are often unclear.

The Locus of Decision Making

Countries do not devolve responsibility and powerto lower levels of government and schools whole-sale. While a central government may transfer gov-ernance and overall management of basic educa-

tion to lower levels, it may retain control of peda-gogical matters, personnel management, andfinancing and resource allocation, or it may decen-tralize those functions to school councils. Howeach country assigns these specific decision areas is,in many respects, a more accurate measure of itsdegree of decentralization.

In 1998, the World Education Indicators survey,conducted under the auspices of the Organisationof Economic Co-operation and Development(OECD), collected information on the locus ofdecision making in lower secondary education in asmall sample of developing countries, including afew East Asian countries (OECD 1998). The surveyexamined 38 decisions pertaining to instructionalcontent, personnel management, and resources andfinancing. To update the resulting data for thischapter, we undertook a similar, though moremodest, information-gathering effort in East Asiancountries. (See the chapter annex for details on themethod we used and a list of functions comparingthe two sets of data.)

Both databases reveal de facto rather than dejure decentralization in East Asia.6 However, theseassessments of decision-making authority are sub-jective for at least two reasons: First, practice canvary widely within each country, so country-levelinformation is impressionistic rather than a“weighted average” of practice across areas. Second,periodic assessments are likely to reflect variationin implementation of legislation, a change in legis-lation, or both. These factors affect these two datasets. We present the results briefly, nonetheless, as arough indication of the change in the degree ofdecentralization of each country relative to othercountries.

In 1998, the proportion of decisions related tosecondary education made by the central govern-ment varied widely in East Asia—from one-fifth inChina to three-fifths in Indonesia, indicating thatChina’s education system was then much morehighly decentralized than Indonesia’s (see table 9.2).At that time, China’s provincial and local govern-ments were making one-third of such decisionsand schools about one-half, while Indonesia’sprovinces made less than one-tenth of such deci-sions and schools about one-third. By 2003, theallocation of decision making in Indonesia andChina had shifted considerably. China seems tohave retracted powers from schools but widened


east_179-208.qxd 5/17/05 11:43 AM Page 183

provincial and local power (among counties andtownships)—not uncommon in the country’sdecentralization history. In Indonesia, the 1999decentralization reform assigned powers andresponsibilities to district governments, quadru-pling the proportion of education-related decisionsby these governments.

The numbers from 1998 and 2003 for thePhilippines and Thailand are puzzling: they implythat these countries have recentralized rather thandecentralized during this period. The 1998 datasuggest that the central government was makingonly about one-third of decisions in secondaryeducation in the Philippines, and about one-half ofthe decisions in Thailand. These countries seem tohave pushed back decentralization in 2003, withthree-fifths and three-fourths of decision makinglying with the central government and the role ofschools greatly reduced. Historical background,however, suggests that decentralization to thedegree suggested by the 1998 numbers did notexist, and that the 2003 assessment more accuratelyreflects reality. In Thailand, following the 1997Constitution and the Decentralization Act of 1999,a Decentralization Master Plan approved in 2000stipulated details for transferring responsibilities.However, implementation has been slow. In thePhilippines, the legal framework for decentraliza-tion has not transferred overall management ofsecondary education to local government, althoughstrong local governments nevertheless use theirsubstantial autonomy under the Local GovernmentCode to supplement their administrative authority.

In sum, despite the transfer of governance oflower secondary education to lower levels of gov-

ernment in these countries, decision making onspecific functions actually occurs at different levelsof government and in schools.The next section askswhether these countries reveal a pattern in allocat-ing specific decision areas, and whether such allo-cation is likely to improve the way the educationsystems operate and thus promote better outcomes.

Who Makes Which Decisions?

With different levels of government involved in mul-tiple areas of decision making, the goal is to ensuredelineation and alignment of responsibilities, coor-dination, and information sharing. Although thesechallenges may have existed before decentralization,pressure to address them has intensified. Decentral-ization is not likely to improve the education systemif local governments have the authority to hire andfire teachers but not to influence their promotion,compensation, and development, or if schools havethe authority to choose teaching methods but nottextbooks.

Patterns emerge in the locus and mode of 22 of38 decisions related to secondary education in theEast Asian countries (see table 9.3). Setting curricu-lum content, instruction time, and teachers’salaries, and allocating resources to schools, remainthe domain of the national or state and provincialgovernments. In contrast, all five countries leavethe choice of teaching methods and support activi-ties for students (such as remedial classes) entirelyto schools; four of five countries also assign deci-sion making on teachers’ careers to local entities.Cambodia is the most centralistic with respect tothe 22 decisions:


TABLE 9.2 Percentage of Decisions Related to Lower Secondary Educationat Each Level of Government, 1998 and 2003

Central State/provincial/government local government School

Country 1998 2003 1998 2003 1998 2003

Cambodia — 75 — 11 — 14China 21 3 33 77 46 20Indonesia 63 36 7 28 30 35Philippines 37 62 24 20 39 18Thailand 55 75 0 6 45 20

Sources: OECD 1998; World Bank survey for this study, 2003.Note: — � not available. Percentages may not add up to 100 because of rounding.

east_179-208.qxd 5/17/05 11:43 AM Page 184

Instructional matters. All these countries arereluctant to delegate standard setting and decisionson core curricula to local governments and schools,reflecting the widely held belief that the education

system helps promote a national identity as well asshared values and culture. The management andquality assurance functions of local educationcouncils and school committees remain fairly


TABLE 9.3 Locus and Mode of Key Decisions in Lower Secondary Education, 1998 and 2003

Cambodia China Indonesia Philippines Thailand

1998 2003 1998 2003 1998 2003 1998 2003 1998 2003

Instructional mattersInstruction time — � � � � � � � � �

Designing programs of study — � � � � � � � � �

Defining course content — � � � � � � � � �

Choosing textbooks — � � � � ◊ ◊ � ◊ ◊Teaching methods — ♦ ♦ � � ◊ ◊ � ♦ �

Mode of grouping students — ◊ ♦ ♦ ♦ ◊ ♦ ♦ ♦ ♦

Support activities for students — � ♦ ♦ ♦ ♦ ♦ ♦ ♦ ♦

Creation/closure of schools — � � � � � � � � �

Creation/abolition of grades — � � � ♦ ♦ � � � �

Setting qualifying exams — � � � � � � � ◊ ♦

Credentialing — � � ♦ � ◊ � ♦ ◊ —Methods for assessing

students’ regular work — � ◊ � ♦ � ♦ � ♦ ◊

PersonnelmanagementHiring teachers — � ♦ � � � � � � �

Hiring principals — � � � � � � � � �

Fixing teacher salaries — � ◊ � � � � � � �

Fixing principal salaries — � ◊ � � � � � � �

Career of teachers — � ♦ ◊ ♦ � ◊ � ♦ �

Career of principals — � � � ♦ � � � � �

ResourcesAllocation to school

for teaching staff — � � � � � � � �

Allocation to school for nonsalary current expenditure — � ♦ � � � � � � �

Allocation to school for capital expenditure — � � � � � � � � �

Use in school for capital expenditure — � � � � ◊ ◊ � � �

Sources: OECD 1998; World Bank survey for this study, 2003.Note: — � missing data. Symbols indicate locus and mode of decision according to this legend: Decisionmade in full autonomy: �, Central government; �, Intermediate government; �, Local government; ♦, School. Decision made in consultation or within framework: �, Central government; �, Intermediategovernment; �, Local government; ◊, School.

east_179-208.qxd 5/17/05 11:43 AM Page 185

limited, bounded by a national framework. China’scentral government continues to keep a close watchon curriculum, selection of textbooks, school-leav-ing qualifications, and teacher education, and alsoretains control over core subject areas such asmoral-political education (Bray 1999; Shen 1994;Hawkins 2000). A national curriculum frame-work—developed primarily by the centralgovernment with some consultation with localgovernments and adopted in 1992—specifiescompulsory courses. Local autonomy in educationcontent appears to be limited to art, music, andsports. Continuing to take control over the nationalcore curriculum, in 1999 the central governmentdeveloped new curriculum standards for 18 subjectareas for the nine-year compulsory education level.These standards emphasized the need for the cur-riculum to respond to rapid changes in technologyand China’s economy (Wang 2004). The new corecurriculum also allows for local and schoolcurricula, however.

According to Indonesia’s Education Law 20, thecentral government still determines the curricu-lum framework and structure for basic and sec-ondary education (Article 38). The central govern-ment is also establishing minimum servicestandards for education. However, district govern-ments, given constrained financial and technicalresources, may have trouble meeting these stan-dards. Likewise, in Thailand, the Commission ofBasic Education—a pillar of the central educationagency—is responsible for proposing standardsand the core curricula for basic education in linewith the National Scheme of Education, Religion,Art, and Culture. Parents’ groups will provideinternal oversight of each school, while a centralagency will develop criteria and methods forassessing student performance and school quality.In the Philippines, the central government alsoretains responsibility for policy, curriculum, per-sonnel, and operations.

Decisions on creating or closing a school aremade at the national level in the Philippines andThailand but at the local level in Cambodia, China,and Indonesia. These countries usually decentralizedecisions on textbooks and teaching and assessmentmethods to provincial and local governments. Allcountries except Cambodia allow schools to makeautonomous decisions regarding grouping studentsand providing extracurricular activities.

Countries outside East Asia show a similar reluc-tance to surrender control over the substance andquality of education to subnational governmentsand schools. For example, in Chile, the central min-istry maintains curriculum-setting, regulatory, andquality assurance functions (Delannoy 2000). InAustralia, the Curriculum Standard Frameworkdefines eight key learning areas, incorporating bothcontent and process standards (Pascoe and Pascoe1998). Likewise, the British school reform estab-lished a national curriculum with learning objec-tives for core subjects each year and at each keystage (Rodríguez and Hovde 2002). In Spain, whichis less centralized, the decentralization law estab-lished that the Ministry of Education defines65 percent of the instructional material taught inall schools, while autonomous communities maydefine 35–45 percent of domains that reflectregional interests (Hanson 2000). But in otherOECD countries, schools choose teaching methods,textbooks, and techniques for assessing studentsday-to-day, although usually within a frameworkestablished by a higher level of government (OECD1998). This is similar to the approach among somestates in Brazil, such as Paraná, Pernambuco, andRio de Janeiro, where most schools are responsiblefor elaborating pedagogical proposals and inte-grating them into the core curriculum process(Machado 2002).7

Teacher management. Different levels of govern-ment make decisions regarding teachers, oftenleading to confusion and inefficiencies. These deci-sions range from teacher training to recruitment,deployment, performance evaluation, humanresources databases, payroll, and redeployment.Some countries decentralize some functions, suchas hiring and firing of teachers, while keeping oth-ers effectively centralized, such as setting compen-sation levels.

Indonesia illustrates a mixed—and confus-ing—policy regarding teacher management. The2003 Education Law stipulates that the central anddistrict governments share responsibility for “get-ting educators and education personnel to ensurethe implementation of good quality educationprograms” (Article 41), and that these govern-ments will “supervise and develop educationalpersonnel in education units” (Article 44). How-ever, many aspects of teacher management remaincentralized, including managing the personnel


east_179-208.qxd 5/17/05 11:43 AM Page 186

database, registering personnel actions, and trans-mitting this information to the payroll system.While districts manage personnel and payroll, therecording of such actions—necessary to triggerthe payroll—is still centralized, and, according tocivil service law, the central government retainsmuch authority over teacher wages, positionallowances, family and rice allowances, and evenhonoraria. In focus-group discussions, teachersreported that while they support decentralization,they prefer central management of their employ-ment (World Bank 2004a). According to teachers,given that processes such as promotion stillrequire the center’s approval, decentralization hasslowed action on personnel matters because it hasadded a bureaucratic layer. Teachers also claimthat management processes are neither moretransparent nor better monitored, even thoughthey occur at the district level. Without authorityor significant influence over teacher-related mat-ters, local governments and schools lack the singlemost important tool to influence the quality ofeducation.

In countries outside East Asia, the approachto managing teachers is also mixed and reveals awillingness to experiment. In Chile and Mexico,control over contracts is centralized, and a nationalsalary scale standardizes teachers’ pay. Other coun-tries have encouraged greater local participation. InEl Salvador, community education associations arelegally responsible for hiring and firing teachers. Inthe United Kingdom, while the national level sets aminimum pay scale and qualifications for educa-tors, public schools are responsible for hiring andpaying their own teachers. In Brazil, communitiesacross an increasing number of municipalities relyon direct elections to select school directors (Namode Mello 2005).

Financing and resource allocation. This decisionarea is the most decentralized, as countries havesought to mobilize local funds for schools—but notwithout second thoughts. Initial enthusiasm forgranting revenue-raising authority to local govern-ments has been dampened by inequalities, followedby attempts to rein in the tendency of local govern-ments to impose many new taxes. Nonetheless,declines in subsidies from the central governmentand emerging fiscal gaps have forced communitiesand schools to seek supplementary funding, oftenby raising user charges.

Fiscal decentralization was a key feature of China’sreform, with the central government reducing itssubsidies to local schools, and local governmentsintensifying their efforts to find alternative fundingfor basic education through taxes, community con-tributions, and income from enterprises (Hawkins2000; Tsang 2002). As the central government cutschool subsidies, the share of nongovernmentalsources rose from 19 percent in 1993 to 24 percent in2000 (Hawkins 2000). Reform documents suggestedsix sources of funding: urban and rural surchargeslevied by local governments, contributions fromindustry and social organizations, donations by indi-viduals and community organizations, tuition fees,income from school-run enterprises, and centralauthorities. In 1994, however, the central govern-ment reversed itself and removed certain tax author-ity from local governments, and has continued tofund teachers’ salaries and certain capital expendi-tures, citing growing disparities across regions(Tsang 2002; West and Wong 1995). The practice ofcharging fees is prohibited by the central governmentbut encouraged by local governments, which usesome of these additional resources to fund a compen-satory mechanism. The local government defines thefee scale and collects a certain percentage from feerevenues. For example, district governments receive25–50 percent of fee revenues collected by schools.The revenues remitted to local governments are thenused to help other schools repair their school build-ings and improve their facilities (Wang 2004).

Financing for education in Indonesia is alsomeant to be a “shared responsibility” of the centralgovernment, district governments, and communi-ties (Article 46), but legislation has sent mixed mes-sages about how autonomous local governmentsactually are in raising funds (Government ofIndonesia 2003; World Bank 2004a). On the onehand, laws have expanded the revenue-raising abil-ity of district governments and allowed them todetermine their own financial management,accounting, and procurement systems within broadguidelines. On the other hand, three design featuresof the reform limit local autonomy. First, a suppos-edly temporary hold-harmless component assignspart of the block grant to districts to cover thesalaries of existing teachers. Second, the reform ear-marks at least 20 percent of the national budget anda similar percentage of the regional budget (net ofsalaries) for education (Article 49 of the Education


east_179-208.qxd 5/17/05 11:43 AM Page 187

Law). Third, the funding mechanism for educationis still too diverse and fragmented. Given thesefeatures, some regional education officials haveexpressed frustration at not knowing the total levelof resources actually available to them (World Bank2004a). Without information or transparency, theyfind it difficult to plan ahead, to develop coherentand effective educational programs, and to monitorand assess the flow of funds through the system.

In the Philippines, education financing is morecentralized because public education is not for-mally decentralized, but local governments spendtheir own resources for education nonetheless.8

The sources of local government financing are theInternal Revenue Allotment (IRA), which the cen-tral government sends to each municipality andcity, and the Special Education Fund (SEF), whichis a 1 percent tax on assessed values of real proper-ties owned by a municipality or city. One-half ofthe SEF is spent at the municipality or city level andthe other half is remitted to provinces for educationprojects. The provincial Local School Board deter-mines the allocation of this fund among munici-palities.9 Because the Local Government Codedevolved construction and maintenance of elemen-tary and secondary school buildings to municipali-ties and cities, the SEF also sometimes finances suchconstruction and repair, as well as equipment, edu-cational research, books and periodicals, and sportsdevelopment. Many local governments have alsoshown initiative in using the fund to establish newsecondary schools and hire more teachers, or to topoff the salaries of the centrally hired public teachers(Azfar et al. 2001; Manasan 2002).10 User chargeshave also boosted local funding: the share of schoolfees in education spending by households rose to17 percent in 1997 (Manasan 2002).11 Fearing thatfees might reverse gains in enrollments, in 2001 thecentral Department of Education prohibited ele-mentary schools from collecting user charges.

If the central government adopts a strong com-pensatory policy in distributing funds acrossregions, then local financing and modest usercharges can boost performance by allowing parentsand the community to exert greater control overschool operations. In the Philippines, schools thatrely more heavily on local sources—includingcontributions from the local school board, munici-pal government, and parent-teacher associations—are more efficient. A 1 percent increase in the share

of financing from local sources can lead to a0.14 percent decline in total costs (Jimenez andSawada 1996). In Indonesia, local governmentspending and parental contributions boost schoolefficiency: cost per student falls as the local share offunding rises, though at a diminishing rate (Jameset al. 1996). In sum, if used with an eye towardequity concerns, local funding can improve effi-ciency without worsening inequality.

The role of the central education agency. Underdecentralization, the central education authorityneeds to redefine its role and reform its structuresand processes so it can fulfill its new core functions.Lower levels of governments simply have no incen-tive to carry out some policies and programsbecause they cannot fully capture the returns, theyare unable to perform them well because of a lackof economies of scale, or they do not have enoughresources. These policies and programs includesetting goals and standards for service provision,experimenting when needed, rewarding innovationfrom other parts of the system, disseminatinginformation widely and regularly, establishing andenforcing a transparent regulatory framework, andensuring more equitable education spending.

East Asian countries recognize these roles. The1985 decision by the Central Committee of China’sCommunist Party on reforming the educationsystem retains a guiding and monitoring role for thecentral government on major policies, principles,and the general plan. In 1993, the State Councilissued the Program for China’s Educational Reformand Development to address weaknesses in the edu-cation system and to emphasize the link betweenthe country’s economic development and education.This policy guideline provided for local govern-ments to assume more responsibility for managingand financing basic education, and encouraged thegradual establishment of community-sponsoredschools, while retaining the central government’srole as the arbiter of rules and regulations (Hawkins2000). This mandate of local responsibility and dif-ferentiated levels of management was reinforced bythe State Education Commission in 1995 throughthe Education Law of the People’s Republic of China(Wang 2004).

Thailand’s National Education Act of 1999assigned administration and management functionsrelated to academic matters, budget, and personnelto “educational service areas, educational insti-


east_179-208.qxd 5/17/05 11:43 AM Page 188

tutions, and local administration organizations”(Section 9) (Kingdom of Thailand 1999). However,this legislation maintains a large role for the centralauthority in designating standards and definingprocedures while supporting local governancethrough boards and committees.

International experience illustrates the role ofthe central government in reducing educationinequalities within a decentralized setting. In bothSpain and Chile, the central government usedrevenue-sharing and transfer mechanisms toimplement this goal. Besides a block grant budgettransfer, Spain created the Inter-Territorial Com-pensation Fund (FCI). As a result, in 1996 Andaluciareceived 38 percent of state redistributed incomeand 39 percent of the FCI, while Madrid receivedless than 1 percent of state redistributed income(Hanson 2000). In Chile, the P-900 programdesigned pedagogic support initiatives for ruralstudents and the least-advantaged 10 percent of theprimary school population. Chile used severalmeasures to improve equity in the 1990s: a capita-tion grant to rural schools; scholarships for indige-nous, low-income, and distinguished students;school feeding programs; and an expansion of pre-school education (Delannoy 2000). In 1998, theMexican government adopted a formula-drivensystem for allocating transfers to states. Underthe new formula, states receive at least the sameamount as the previous year, as well as budgetaryincrements based on the number of needed schoolsand teachers (Lopez-Acevedo et al. 2003). In Brazilin the mid-1990s, the Lei de Diretrizes e Bases daEducação Nacional assigned the federal governmentthe role of narrowing inequalities in access andfinance, and launched the Fund for the Mainte-nance and Development of Basic Education andTeacher Appreciation (FUNDEF) to equalizefinancing for basic education. This fund guaranteesa minimum per pupil expenditure in primaryschools throughout the country and partiallyequalizes per pupil funding within states.

The Impact of Decentralization:Educational Dividends

East Asia’s experience with decentralization hasbeen relatively brief, so it is too early to assess thereal impact of decentralization reforms on manymeasures of educational development. This section

presents two measures of such impact: the level ofspending on education, and geographical dispari-ties in enrollment and literacy rates. The sectionfocuses on three of the five countries—China, thePhilippines, and Indonesia, which have had slightlylonger experience with decentralization. The evi-dence is suggestive rather than conclusive, becauseit reflects not only the effects of decentralizationbut also those of other reforms and developments.Even if we could capture the direct impact ofdecentralization, the results would reflect its design,procedural and implementation capacity, andpolitical maturity more than its inherent flaws andbenefits. This section also presents the impact ofdecentralization on student performance, butfocuses on lessons from international experiencebecause of lack of data on East Asia.

Greater but More Unequal Education Spending

Has total spending on education grown as a resultof decentralization? Are funding levels appropriate?Answers to these questions need to considerwhether a country has created an appropriate bal-ance between assigned expenditure responsibilitiesat various levels and allocated revenues (“verticalequalization”). In other words, local governmentsshould control resources commensurate with theirassigned responsibilities, and transfers from thenational government should supplement what theylack. Has the spending share of local governmentsexpanded relative to the share of the central gov-ernment? Has the type of government spending foreducation changed? Have central governmentstended to spend more on, say, capital investmentsthan recurrent items such as salaries and opera-tional expenditures?

China. This country’s share of education expen-ditures in total fiscal spending more than doubled—from 6.7 to 18 percent—from 1978 to 2001. Thecentral government devoted 16.3 percent of itsbudget to education in 2001 (People’s Republic ofChina 2002). From 1986 to 1992, the budget alloca-tion and out-of-budget funds grew annually by3.5 percent and 19.7 percent in real terms, respec-tively, while per-student budgeted spending rose by9.6 percent at the primary level and 5.1 percent atthe secondary level. Yet because China’s economygrew rapidly, the share of education spending ingross domestic product (GDP) fell from 2.9 percent


east_179-208.qxd 5/17/05 11:43 AM Page 189

ernments provide a financial subsidy to poor areas,the subsidy is small and an ad hoc instrumentrather than a regular part of financing for compul-sory education. The result is that teachers are notpaid on time, many schools are in poor physicalcondition, and the goals of the Universal Compul-sory Education program have been delayed. Despitethe program’s efforts to raise the minimum provi-sion of education in poor regions, they provide lesseducation in terms of quantity and quality and passmore costs along to families. And, contrary to thelaw, some county governments are borrowing fromthe private sector to finance their schools (Tsang2002). In addition, although tuition fees in compul-sory education are forbidden by law, fees in publicschools are often collected in the form of a “jointconstruction fee” or as voluntary donations (Wang2004).

Philippines. In this country, total public spend-ing on education as a percent of GDP rose at anaverage annual rate of 6 percent from 1987 to 2001(see figure 9.2). During this 15-year period, spend-ing declined only during times of fiscal adjustment(1990–94) and financial crisis (1998–99), mainlybecause of dips in central government spending.GDP grew by 4.1 percent in real terms, so this largershare of education meant a substantial rise in realspending for education. As a share of total spendingby local governments, education spending rosefrom 3.8 percent in 1990–91 to 7.8 percent in1998–2000, with the Special Education Fund (SEF)fueling a growing part of this local spending, rising


TABLE 9.4 China: Education as a Share ofTotal Fiscal Expenditures

1997 1998 1999

Hezheng County 17.3 14.3 13.3Jinshishan County 19.0 18.4 18.3Linxia Prefecture 16.0 17.2 16.1Gansu Province 14.9 15.5 16.0CHINA 16.7 16.0

Source: World Bank 2000.

TABLE 9.5 China: Per Student Educational Expenditure, 1989, 1997, and 2000

Primary level Lower secondary level

1989 1997 2000 1989 1997 2000

Highest-spending provinces 393 2,351 2,756 788 3,425 2,788

Lowest-spending provinces 75 255 261 174 491 420

Ratio of highest-to-lowest-spendingprovinces 5.2 9.2 10.6 4.5 7.0 6.6

Mean 166 593 492 353 1,096 680

Source: Tsang 2002.Note: Data for 1989 and 1997 pertain to total education expenditures by provinces; 2000 data pertainonly to recurrent expenditures. In 1997, for which both total and recurrent expenditure data areavailable, the ratios of recurrent spending by the highest- and lowest-spending provinces are similar(9.4 and 7.1, respectively).

in 1991 to 2.5 percent in 1997, and to 2.2 percent in1999 (Tsang 2002). Even as a share of total govern-ment spending, education expenditures fell some-what from 16.7 percent in 1997 to 16.0 in 1999,such as other fiscal expenditures rose faster (seetable 9.4). Trends varied across counties, however.In some, as Hezheng and Jinshishan, education’sshare of total spending declined. In Gansu province,which began the period with a lower share for edu-cation, that share rose.

Interprovincial disparities in per-student spend-ing have also widened. The highest-spendingprovinces spend many times more for primary andlower secondary education than the lowest-spend-ing provinces, and these gaps have grown, especiallyfor primary education (see table 9.5). Observersconclude that the lack of a clear equalizationscheme is a fundamental weakness of the system’sfinancing (Hawkins 2000; Tsang 2002; West andWong 1997). While the central and provincial gov-

east_179-208.qxd 5/17/05 11:43 AM Page 190

from 57 percent in 1992 to 79 percent in 1999. Theresult has been a shift in the shares of central andlocal governments in education spending. In 1991,the local government share was only 2.5 percent,rising to 7.4 percent in 2001. Meanwhile, the sharesof the central government in both capital expendi-tures and operating expenditures for educationhave declined.12

Data on average SEF spending per pupil indicatewide variation across income classes of municipali-ties and cities (see table 9.6). The poorest munici-palities spend only 13 percent of the SEF per pupilresources of the richest municipalities, and only 3percent of the SEF resources of the richest cities.These numbers plainly show that transferringresponsibility for funding basic education to localgovernments leads to wide regional disparities inschool inputs.

Indonesia. It is still too early to tell whetherdecentralization will raise overall public spendingon education in Indonesia, but early signs havebeen positive. Before decentralization, Indonesiawas spending the smallest share of GDP on educa-tion among East Asian countries: only 1.4 percent.This share rose to more than 4 percent in 2002—asignificant expansion in resources for the sector.In 2001, district governments accounted for abouttwo-thirds of total education spending, whereasthe share of provincial governments was only4 percent.

On average, district governments in Indonesiado have more resources at their disposal than in thepast, and the allocation formula stipulates thatpoorer districts should receive larger transfers.However, the central government expects districtsto mobilize more of their own resources to supple-ment the transfers. Herein lies the risk that inequal-ities among districts will grow, as in China and thePhilippines.13 Decentralization laws have giventaxing authority to district governments if the cen-tral government authorizes the taxes and districtsabide by principles in Law 34 of 2000.14 The reality,however, is large inequalities in local revenue bases.Many district governments have limited capacityto raise taxes from land, buildings, and naturalresources, which constitute only about 5 percentof their revenues. Provincial governments have alarger own-resource base but must share these rev-enues with district governments. For example, theper capita GDP (excluding oil and gas) of the rich-est province, Jakarta, is almost nine times largerthan that of the poorest province, East NusaTenggara (Akita and Alisjahbana 2002). Partly as aresult, per student allocations for recurrent andcapital expenditures vary widely, with districts atthe lower end of the range surely not meeting anykind of education standard (see table 9.7).

Summary. The trends in education spending inChina, the Philippines, and Indonesia show that total


TABLE 9.6 Philippines: Median Valuesof SEF Resources per Pupil

FIGURE 9.2 Central and Local GovernmentEducation Spending in the Philippines

Source: Manasan 2002.

0

0.5

1.5

2.5

% G

DP

3.5

4.5

1.0

2.0

3.0

4.0

1987

1988

1989

1990

199119

9219

9319

9419

9519

9619

9719

9819

9920

0020

01

total government spending

central government spending

local government spending

Income 40th classification Median percentile

City class 1 590 450City class 2 382 270City class 3 341 190City class 4a 140City class 5a 120Class 1 132 101Class 2 65.7 51Class 3 50.2 40Class 4 46.7 37Class 5 32.5 26Class 6 17.6 11

Source: Manasan and Atkins 2004.Note: The classes pertain to income classifica-tions: 1 to 5 for cities and 1 to 6 for othermunicipalities.a. These numbers represent few observations.

east_179-208.qxd 5/17/05 11:43 AM Page 191

resources for education have grown under decentral-ization; yet whether these increases are larger thanthey would have been without reform is difficult tosay. It is also clear that the share of education spend-ing by local governments has grown, partly becausethe central government has devolved resources andresponsibilities for spending those resources to localgovernments, and partly because local governmentsare expected to generate their own resources to meetthose expenditures. However, local governments arefar from equal in their ability to mobilize their ownresources, and thus the gap in education expendi-tures per student between wealthier and poorer areascan only widen. Central governments clearly need toestablish a mechanism for equalizing educationresources across municipalities and cities.

The block grant system—which gives local deci-sion makers latitude to act on local goals with gen-erally unrestricted funds—does not guarantee thatofficials will spend enough resources on education.On the one hand, local decision makers may chooseto finance budget items that promise a quicker andmore stimulating effect on the local economy. Onthe other hand, they may respond to the desire oflocal voters for more and better schools, or to thefact that schools provide local employment and canbe a source of prestige for the community and itsleaders. Greater local funding is expected to createpublic pressure to spend resources more wisely andthus make the sector more efficient, given anappropriate system of accountability.

Reducing Educational Disparities within Countries

As mentioned, developing countries in East Asia

have generally made important progress in basic


education. Enrollment rates at the primary level are

high, and rates are rising even for post-basic educa-

tion. These numbers hide large inequalities within

each country, however. These inequalities predate

decentralization reforms, but those reforms could

exacerbate them. Large inequalities in the distribu-

tion of resources among geographical regions can

produce large disparities in education outcomes.

Transferring fiscal responsibilities to local areas and

relying on local resources and expertise is likely to

widen educational gaps between areas with a strong

revenue base and those that are less prosperous

while weakening the central government’s ability to

close these gaps.China. So how large are within-country dif-

ferences in education outcomes, and have theyincreased or decreased since decentralization?China’s overall enrollment rates in basic educationare high, but provinces differ widely in literacyrates and in enrollment rates at the secondarylevel. Literacy rates (for those aged 10 and above)rose significantly from 1982 to 1999—in a fewprovinces by as much as 25 percentage points—and inequality fell (see figure 9.3). The coefficientof variation for literacy rates declined from 0.19 in1982 to 0.13 in 1999. With the sole exception ofTibet, where literacy was only 35 percent in 1999,literacy rates across China exceeded 70 percent.Despite this progress toward equalization,undoubtedly the result of the national policyof universal basic education, provinces such asQinghai, Gansu, Guizhou, and Yunnan lag bynearly 20 percentage points behind the most liter-ate provinces of Liaoning, Jilin, and Tianjin. Tibet’sliteracy rate also improved, but its gain was one ofthe smallest.

TABLE 9.7 Indonesia: Per Capita Education Spending, 2001–2(in rupiah)

2001 (actual) 2002 (planned)

Total 134,000 175,058(1,586/463,753) (1,193/540,479)

Recurrent 126,118 159,460(998/450,789) (1,013/539,287)

Capital 16,185 21,692(177/205,044) (402/415,463)

Source: Sistem Informasi Keuangan Daerah (SIKD), Ministry of Finance.Note: Numbers in parentheses indicate minimum and maximum values.

east_179-208.qxd 5/17/05 11:43 AM Page 192

Education of minority ethnic groups inpoorer, interior rural regions has been a concern.Enrollment rates are 15 and 10 percentage pointslower for minority girls and boys than for Han girlsand boys, respectively (Hannum 2002). Secondaryenrollment rates are also unequal among provinces.In 2000, Shanghai and Beijing had enrollment ratesclose to 90 percent, while Tibet, Guizhou, Guangxi,and Hainan had enrollment rates one-third lower.But while the coefficient of variation of enrollmentrates is smaller (at 0.10) than that for literacy rates,the tendency is for provinces that had higherliteracy rates in 1982 to have higher secondaryenrollment rates in 2000, suggesting that laggingprovinces will continue to fall behind.

Philippines. Similarly, in the Philippines, literacyrates and enrollment rates vary widely acrossprovinces. Literacy rates (for the population age10 years and above) increased substantially over thefive-year period from 1989 to 1994—by as much asnearly 20 percentage points in Western Mindanao,and by almost 15 percentage points in threeother regions (see figure 9.4). With these gains, theliteracy gap narrowed among the regions, but by1994 literacy rates still ranged from 61 percent inthe Autonomous Region of Muslim Mindanao(ARMM) to 92 percent in the national capitalregion (metropolitan Manila).15 These trends donot indicate that decentralization has helpedreduce education gaps.

Enrollment rates at the secondary level have alsoshown significant gains since decentralization, withincreases fairly equal across regions.16 The notableexceptions are the regions in Mindanao: NorthernMindanao is the only region in which secondaryenrollment rates declined, and the increases in theother Mindanao regions are smaller than in anyother region. In 2002, enrollment rates varied fromjust 32 percent in ARMM to 94 percent in theIlocos Region.

The experience of ARMM is noteworthy becauseit is the region with the fullest autonomy, includingin managing its education system.17 Legislationpassed in 2001 contains detailed provisions that theregion’s schools will adopt the basic core courses,minimum curriculum, and textbooks required bythe national government, but will have the preroga-tive and responsibility to add other courses andinstructional materials that reflect Islamic values.With respect to the two indicators considered above,


FIGURE 9.3 Literacy and Enrollment Ratesin China, by Province

Source: People’s Republic of China 2001.

Note: Provinces are arranged in descending order according tovalues in the more recent year.

TibetGuizhouGuangxi

HainanAnhui

YunnanNingxia

HeilongjiangGuangdong

HenanJiangxiGansuFujian

ShanxiHunan

ShandongHebei

SichuanChongqing

LiaoningHubei

XinjiangJilin

QinghaiJiangsu

ShaanxiZhejang

Inner MongoliaTianjinBeijing

Shanghai

Literacy rates, 1982 and 1999

percent

percent

20 30 40 50 60 70 80 90 100

TibetQinghai

GansuGuizhouYunnanNingxia

AnhuiShandong

FujianShaanxiJiangsu

SichuanInner Mongolia

HenanZhejang

HubeiChongqing

HainanJiangxi

GuangxiHebei

HunanHeilongjiang

XinjiangGuangdong

ShanxiShanghai

TianjinLiaoning

JilinBeijing

Secondary enrollment, 2000

50 60 70 80 90 100

19821999

east_179-208.qxd 5/17/05 11:43 AM Page 193

however, ARMM ranked dead last, with changesnot large enough to allow it to catch up with otherregions. Although the enrollment gap in primaryeducation between ARMM and the rest of the coun-try has narrowed since the 1990s, it remains sub-stantial: nearly 20 percentage points separateschool-age children in the poorest quintile inARMM and those in other regions (World Bank2003b). ARMM children begin school later and areonly about half as likely to continue through theelementary grades, and the transition to high school

is particularly difficult. As a result, fewer than 2 of10 children who enter grade one complete highschool. In sum, while at least two other factors mightexplain the region’s poor education indicators—high poverty and protracted armed conflict—it isreasonable to conclude that greater autonomy hasnot produced better outcomes.

Indonesia. Indonesia’s enrollment rates, especiallyat the primary level, compare favorably with thoseof East Asian countries with higher per capitaincome. With decentralization relatively nascent inIndonesia, these overall gains cannot be attributed tothe reform. But differences among regions revealthe equity challenge for decentralization. In 2002, theenrollment rate of youths aged 13 to 15 (roughlythe junior secondary level) ranged from 68 percentin South Sulawesi to 94 percent in Yogjakarta (seefigure 9.5).18 Increases since 1998 have been modest.


FIGURE 9.5 Enrollment Rates in Indonesia,by Province

Enrollment rates of13- to 15-year-olds, 1998 and 2002

percent

60 70 9080 100

19982002

Sources: Susenas 1998 and 2002.

Note: Figure reflects provincial divisions in 1998; the provincesof Aceh, Irian Jaya, Maluku, and North Maluku have beenexcluded because their sample sizes in 2002 were much smallerand limited than in 1998. Provinces are arranged in descendingorder according to values in the more recent year.

Yogyakarta

Jakarta

North Sumatera

East Kalimantan

West Sumatera

Riau

Bali

Lampung

Central Java

Jambi

East Java

Bengkulu

Central Kalimantan

SE Sulawesi

West Java

West Kalimantan

North Sulawesi

South Sumatera

N.T. Barat

South Kalimantan

N.T. Timur

Central Sulawesi

South Sulawesi

FIGURE 9.4 Literacy and Enrollment Rates inthe Philippines, by Region

Literacy rates, 1989 and 1994

percent

percent

50 60 70 9080 100

Secondary enrollment rates,1991 and 2002

20 40 806010 30 50 9070 100

ARMM

West Mindanao

Central Mindanao

Cordillera Admin

South Mindanao

Eastern Visayas

Central Visayas

Western Visayas

Bicol Region

North Mindanao

Ilocos Region

Cagayan Valley

Central Luzon

Southern Tagalog

Metro Manila

ARMM

North Mindanao

West Mindanao

South Mindanao

Central Mindanao

Eastern Visayas

Bicol Region

Central Visayas

Cagayan Valley

Western Visayas

Central Luzon

Cordillera Admin

Southern Tagalog

Metro Manila

Ilocos Region

Source: Department of Education Statistical Yearbook,several years; National Statistical Coordination Board.

Note: Provinces are arranged in descending order according tovalues in the more recent year.

19912002

19891994

east_179-208.qxd 5/17/05 11:43 AM Page 194

Disparities within provinces are even larger thanvariation between provinces. Only about one-fourth to one-third of inequality in enrollmentrates in primary and secondary education is due todifferences among provinces—the rest reflects dif-ferences among districts within a province. Thismeans that equalization among districts withineach province is a greater challenge than equaliza-tion among provinces.

Overall, these changes in the levels and inter-provincial distribution of education expenditures,and in basic indicators, in China, the Philippines,and Indonesia point to both positive and negativeeffects of reform. However, the analysis is still at anaggregate level. Ultimately, the test of decentraliza-tion’s effectiveness is whether schools are better andstudents are learning better. Decentralization lawsencourage greater local and community participa-tion in providing and financing education, but thisfeature exposes inequalities between prosperousand poor areas, and the inability of poor areas tomobilize adequate resources. Central transfermechanisms need to equalize resources acrossareas. Indeed, the push for greater local mobiliza-tion of resources in decentralizing countries riskswidening disparities between regions with a strongrevenue base and those that are less prosperous. Butimplementing an equalization scheme is a consid-erable political challenge. On the one hand is theissue of how much inequality in educational out-comes the political system can tolerate; on the otherhand is the question of to what extent the systemcan redistribute from richer areas to poorer areas,and from urban areas to rural areas.

Improving the Quality of Education

Good education is not only about quantitativetargets, such as boosting the supply of classrooms,textbooks, and teachers, but also about incentivesthat lead to better instruction and learning. Thelevel and use of public spending for education isonly part of the educational story in East Asia.19

East Asian countries do not yet have a record of theimpact of decentralization on student outcomes,especially on learning. There are several reasons forthis: Decentralization is a wide-ranging reform,encompassing and influencing school functions indifferent ways, so isolating its impact from otherchanges in the education system and the economy

at large is difficult. Student-level data on nationalexam results are usually not available even toresearchers. Summary test results do not allowstudy of whether differences in test scores resultfrom changes in students’ economic conditions orfrom aspects of the decentralization reform. Lastly,most East Asian countries have relatively briefexperience with decentralization. However, evalua-tions of the experiences of countries outside EastAsia provide lessons on the potential impact ofdecentralization on student performance.

In the United States, two examples are illustra-tive. The 1995 Chicago School Reform AmendatoryAct modified a 1988 autonomy reform by estab-lishing stronger central support functions andrequiring external accountability mechanisms.20

Although attributing causality to either the 1988 orthe 1995 reform is difficult because of their com-plementary nature, student scores in elementaryreading and math have improved consistently since1995. The percentage of students scoring at orabove the mean in elementary reading tests rosefrom 26.5 percent in 1995 to 36.1 percent in 1999.21

In 1995, Memphis introduced a similar set ofreforms that differed in one important feature: cityschools received a menu of eight different restruc-turing models from which to choose. Before 1995,the schools that later became autonomous hadlower student achievement; two years later, theirscores were substantially higher than those in acontrol group (Ross et al. 1998).

In Chile, two phases of reform appear to haveproduced significantly different results. One evalu-ation concluded that the first phase had either anegative effect on student performance or no effect.A confounding factor was that education expendi-tures declined during the same period. A later eval-uation concluded that the reform did not improvethe quality of public schools, and that test scores forthe majority of students declined. Another evalua-tion found that test scores were higher in privateschools, but concluded that this was largely becausethose schools chose better students. Yet anotherstudy found that teacher autonomy exerts greaterpositive effects on student performance whendecision-making authority is also decentralized(Winkler and Gershberg 2000; Prawda 1992;McEwan and Carnoy 1999; Hsieh and Urquiola2001; Vegas 2002). Evaluations of the second phaseshow more positive results, with an 18 percent rise


east_179-208.qxd 5/17/05 11:43 AM Page 195

in language and math test scores on standardizedtests. But again, isolating the effects of decentraliza-tion from other influences, such as the substantialrise in education expenditures throughout thedecade, is difficult.

An evaluation of Nicaragua’s 1991 school auton-omy reform shows that schools took some time toexercise the new functions and powers given toschool councils. Controlling for de jure and defacto autonomy, the results indicate that schoolautonomy—especially in decisions related to staffingand monitoring of teacher activities—improves stu-dent performance (King and Ozler 1998). Moreover,math and language scores were significantly higherin schools where teachers felt more empowered andinfluential in decision making.

Brazil’s reform remains to be evaluated at thestudent level, but researchers have already usedstate-level measures of educational performance toassess progress. While school councils and the directtransfer of resources are not significantly related tobetter student performance, the election of schooldirectors is positively associated with higher testscores.22 Rather than testing the reform as a whole,the researchers decided to break it down into threecomponents and analyze the impact of eachinnovation on educational performance.23 Amongreform-minded Brazilian states, the most promi-nent is Minas Gerais, whose reform included schoolautonomy and dramatically changed schools’ inter-nal structure of accountability (Guedes et al.1997).24 As in Nicaragua, however, qualitative stud-ies have shown that while de jure autonomy rarelyexerts any influence in most schools, de facto auton-omy appears to boost teacher motivation, and thusthe potential for improving student learning andparticipation in the classroom (Cordeiro Guerra2003).

An evaluation of El Salvador’s EDUCO reformfound that parents in EDUCO schools participatemore actively in school affairs, feel they have moreinfluence over decision making in the school, andhave a more direct relationship with teachers thanparents in traditional public schools. This greaterlocal participation has had a positive effect on edu-cation outcomes. Controlling for school andstudent characteristics, a study found that studentsin EDUCO schools do not perform worse onachievement tests despite the fact that they comefrom poorer families, and that student absences

owing to teacher absences are significantly lower inEDUCO schools (Jimenez and Sawada 1999).

In sum, international experience yields mixedevidence on the impact of decentralization on stu-dent performance. U.S. experience provides posi-tive evidence, but the experience in Latin Americayields ambiguous results. Reform-oriented schoolsin the cities of Chicago and Memphis have substan-tially higher test scores than schools in the controlgroup. In Latin America, whereas decentralizationappears to have improved student performance inNicaragua and Brazil, and to have reduced teacherand student absenteeism in El Salvador, evaluationsof Chile’s long-running reform are inconclusiveconcerning the impact of greater local participationand school choice on student performance.

Lessons about Decentralizing Education

Experiences in the five East Asian countries arebeginning to provide lessons for implementingdecentralization—the factors that have affectedtheir experiences, the sources of resistance or sup-port they have encountered, and the risks and chal-lenges that have emerged. Actual practice oftendeviates from formal rules on decentralization, andit is important to understand why. As mentioned,China has had the longest experience with decen-tralization and thus offers valuable lessons. WhileIndonesia and Thailand previously tried to decen-tralize some functions, their formal decentraliza-tion reforms are recent. Cambodia’s reform is moreaccurately characterized as deconcentration, whilereform in the Philippines is largely a side effect ofa broader decentralization that has formallybypassed education. Indonesia has chosen the rapidBig Bang approach, while Cambodia and Thailandare proceeding at a much more measured pace.

Lessons from these experiences include:

Strive to clarify the assignment of functions, sim-plify new processes and structures, and provide mech-anisms to coordinate and foster a shared understand-ing of reform at different levels of government, as wellas to adjudicate disagreements.

Decentralizing education systems requires har-monizing a complex set of functions at each leveland type of education, and is a difficult reform todesign and implement. Central governments tend todevolve management of education to different levels


east_179-208.qxd 5/17/05 11:43 AM Page 196

of government, but devolution does not happenwholesale. The central government may retain indi-vidual decision areas that may or may not cut acrosslevels of education, or may devolve them even moredeeply. This complexity leads to confusion, redun-dant bureaucracies, and weak implementation.Mixed signals from legislation and policies make itdifficult for subnational governments and schools tofulfill their functions efficiently and effectively.

Despite two decades of implementing decentral-ization in China, for example, the division ofresponsibilities between county and townships isstill unclear. The 1994 Guidelines for the Reformand Development of Education state that bothcounty and township governments are responsiblefor delivering compulsory education, although theformer manages education revenues and the lattersafeguards the right to compulsory education forschool-age children and adolescents. The legisla-tion is unclear on how these responsibilities differ,and also seems to conflict with the Budget Law,which clearly states that each level of governmentshould budget separately for its own jurisdiction(Hawkins 2000).

In Thailand, implementation difficulties stemfrom vague compromises on the overall decentral-ization plan and time frame. Local governments arepoorly informed about their roles and responsibili-ties as well as the decentralization plan and timeframe. Legislation is itself unclear about the decen-tralization process. For example, the 1999 NationalEducation Act transfers authority from the centralgovernment’s provincial and district offices to 175school districts or local education areas (LEAs),each with its own committee and office. This aspectof the reform, if implemented, would requirestaffing cuts and redeployment totaling half of alleducation administrators in various provinces(Mutebi 2003).25 In response, implementers of theNational Decentralization Act argue that they havethe authority to transfer power to local govern-ments only after they fulfill a set of readiness crite-ria. The latter piece of legislation would seem toretrieve some of the autonomy delegated by theNational Education Act. Parties have reached acompromise to merge both acts, which entailsadministrative deconcentration of central power toLEAs in the short to medium term, and gradualdecentralization of responsibilities from LEAs tolocal governments in the long term (World Bank

2003d).26 Ironically, even this compromise suffersfrom vagueness on a time frame.

In Cambodia, the initial framework for decon-centrating and decentralizing education is clearabout delegating authority to provincial and districtauthorities, but is much less clear about the roles ofthe school cluster system and commune councils.The Cluster School Policy, created in 1996, encour-ages decentralized management of resources, but isambiguous about what functions school clustersneed to perform. Similarly, commune councils,though endowed with new financing mechanisms,lack clearly defined roles and responsibilities. Thislack of clarity weakens the institutional structuresthat are closest to the community, ultimately weak-ening accountability as well.

The central government plays important butdifferent roles in a decentralized context, and needs totransform both its structure and skills.

Though stripped of some traditional functionsunder decentralization, the central authority needsto step into its new roles quickly to facilitatesuccessful reform. These roles include settingstandards and performance measures for usethroughout the education system; ensuring thatdecentralized units can meet the standards throughsystems development, training, and funding; andbrokering links between local governments to cap-ture economies of scale. The central governmentalso needs to perform overall system planning andforecasting and prioritize investment, including ofteacher supply and demand; design and implementan equalization scheme, fostering support fromwealthier regions; and stimulate experiments andspread lessons learned.

The central government clearly has a criticalrole to play in designing and implementing equal-ization schemes. Decentralization laws encouragegreater local and community participation in pro-viding and financing education, but this featureexposes inequalities between prosperous and poorareas and the inability of poor areas to mobilizeadequate resources for education, and risks widen-ing those disparities. Appropriate transfer mecha-nisms can equalize resources across regions. How-ever, equalization efforts are not only about infusingmore money into local systems but also—andmore importantly—about changing incentives.Implementing an equalization scheme can be a


east_179-208.qxd 5/17/05 11:43 AM Page 197

considerable political challenge, as efforts to redis-tribute from richer areas to poorer areas, and fromurban areas to rural areas, may meet resistance.

Furthermore, the relative importance of theroles of the central government is likely to changeas decentralization matures, requiring the centralgovernment to boost its involvement in certainfunctions at certain times. For example, it makesmore sense for the central government to helpbuild capacity in countries like Cambodia than inthose like China, which have a larger national sup-ply of experts. Ultimately, redefining the centralgovernment’s roles means distinguishing betweenfunctions requiring critical involvement through-out the process, those requiring periodic involve-ment at different stages, and those requiring one-time initiatives with frequent follow-ups.

Decentralization puts the complex architectureneeded to operate an education system—personnel,finances, procurement, student assessment, and infor-mation management systems—under pressure. Thissystem needs to be reorganized to reflect the newintergovernmental relationships and decentralizedfunctions, and capacity needs to be strengthened.

In decentralized education systems, replacinginappropriate structures and building the capacityto work within new arrangements are key chal-lenges. A principal bottleneck is a lack of adequatetechnical and managerial experience among respon-sible parties. A lack of viable and coordinated man-agement systems linking central agencies to localgovernments and schools exacerbates problemsresulting from weak local capacity.

Countries may be tempted to slow the pace ofdecentralization because of fears that district andprovincial governments do not have enoughcapacity to fulfill their newly assigned responsibil-ities effectively. Indeed, the traditional approachhas been to build local capacity before transfer-ring responsibility and authority because of con-cerns about irresponsible spending, local corrup-tion, regional inequities, and service collapse.However, although capacity building is important,local governments may have more capacity thanmost central policy makers assume. As decentral-ization proceeds, local talents and capacities arelikely to emerge and to improve with practice, as isbecoming clear in Cambodia, Indonesia, and thePhilippines. Furthermore, in China, the capacity

of local governments to manage compulsory edu-cation seems to have improved substantially, inareas such as training personnel, collecting andusing information, expanding the use of technol-ogy, and incorporating research findings and con-sulting experts in the decision-making process(World Bank 2004c; Kerr 1999; Tsang 2002).

For decentralization to exert a positive impact onstudent performance, information and evaluationsystems—as accountability mechanisms—must notonly be in place but must also function iterativelythrough participation.

Information problems become much moreacute in a decentralized context. Dramatic shifts inresponsibilities and powers often lead to the break-down of information and evaluation systems,which typically depend on the central governmentto extract information from lower levels of govern-ment and schools. In Spain, decentralizing admin-istrative functions to regions greatly underminedthe country’s capacity to collect and disseminatenational statistics. Newly autonomous regionsbegan to produce their own statistics using incom-patible methods, and although the Spanish govern-ment took a strong stance on cooperation, it stillmet resistance. The Education Law in 1985 tried toovercome the lack of coordination across regionsthrough a Conference of Counselors which includedthe Minister of Education and autonomous com-munity counselors.

Information on performance at all levels is keyto accountability. Countries can use sample testing,national surveys, and the census to assess theimpact of programs, allocate resources, and iden-tify geographic areas requiring special attention(Asian Development Bank 1999). Local govern-ments, communities, and schools also need infor-mation and diagnostic tools to evaluate perfor-mance in specific subject areas, define learningchallenges in different communities, and comparedifferent pedagogical approaches and teachertraining mechanisms. School-level data can alsocommunicate results to parents and the larger com-munity. At the deepest level, information is instru-mental for greater accountability and control. Citi-zens rely mostly on frequent reports, magazines,and participatory workshops to gain informationon student performance. However, linking teacherpromotion to predetermined inputs and outputs


east_179-208.qxd 5/17/05 11:43 AM Page 198

could create perverse incentives for transmittinginformation. Some analysts, including the Educa-tion Commission in the Philippines, argue that thegovernment should use incentives to make teachersand school heads directly accountable for studentoutcomes (see box 9.1 for an account of the infor-mation system of the Brazilian state of Paraná).

One area that deserves monitoring and evalua-tion is the flow of funds through the system. InCambodia, the Department of Finance designedan improved system for monitoring financial per-formance in 2001. Supported by training and tech-nical assistance, new budget management forms forschools, districts, provinces, and central depart-ments, along with provincial and program reports,will feed into a computerized system. A fund track-ing system will monitor inputs and outputs andrelate them to strategic outcomes, and will includeincentives for transmitting information. Schoolinspection reports will now focus more on schoolperformance indicators, such as development plan-ning, financial management, community partner-ships, teaching and learning processes, the learningand school environment, and promotion rates.

Accurate and timely information on enrollments,teachers, and school inputs is essential, especially forassessing the needs of remote areas and underservedpopulations. Widely conflicting statistics on thesevariables are still too common to support robustplanning and policy making. Most problematic ofall are student testing systems: the challenge is tomake national tests comparable over time to allow

policy evaluation, and to ensure that they reflectexisting or desired curriculum content.27

To advance education outcomes, school stakehold-ers must have greater voice and exercise some controlover school operations.

Decentralization is not just about increasinglocal management and technical skills, but alsoabout strengthening the voice of the community inthe delivery of public services. In many centralisticsystems, local communities are not used to govern-ing themselves, electing politicians who representtheir interests, and using their right to vote to maketheir will known. Breaking out of this mold imme-diately is difficult. Despite commune councils inChina, lack of popular representation persists at thelocal level. In Cambodia, the central governmentstill appoints provincial governors and districtheads. And in Indonesia, although the majority ofschools now have school committees, they hardlymeet and do not yet fulfill their designated func-tions (World Bank 2004a).

Beyond strengthening voice, experiences incountries outside East Asia suggest that giving par-ents, teachers, and other stakeholders decision-making authority in key areas such as curriculum,training, and pedagogical approaches leads to bet-ter student performance. This occurs throughgreater commitment from teachers, more focus onlearning, stronger school leadership, and a sense ofresponsibility for results (King and Ozler 1998;Winkler and Gershberg 2000).


BOX 9.1 School Report Cards in Paraná: A New Incentive System

In the state of Paraná in Brazil, a new system ofincentives at the school level entails producingschool report cards. These report cards—knownas Boletim da Escola—include three main setsof information. These are results from theStatewide Student Learning Testing (coveringPortuguese, math, and science), school censusdata (on student promotion, retention, dropoutrates, enrollment, teacher-pupil ratios, andteacher profiles), and surveys of school life (fromstudents, parents, and school managers). Thereport cards therefore focus on individual schoolperformance while allowing for cross-school

comparisons in the state and region. This wealthof information enhances competition acrossschools, thus strengthening incentives and over-all accountability at the state level. If appliedconsistently across time, this system will alsohelp schools assess the impact of their policieson student performance. One lesson from theseexperiences is that such systems are not sustain-able if they lack local ownership, and if schoolsdo not have the technical capacity to maintainand use them.

Source: Vasconcelos-Saliba 2004.

east_179-208.qxd 5/17/05 11:43 AM Page 199

Top-down decentralization has largely precludedthe involvement of teachers and teachers’ unions indesigning education reform.28 Such limited agencyreflects not only the impetus for decentralizationbut also political and unions’ inability to mobilize.Unions have been less politically active (China) andhave had fewer opportunities to play a significantrole (Cambodia and Philippines) than in LatinAmerican countries, where teachers’ unions havewielded more political power and played criticalroles in shaping education reform.29 The LatinAmerican experience shows that these responsescan be diverse, including resistance (Mexico),negotiation (Bolivia and Chile in the 1990s),cooperation (Brazil), and inaction (Chile inthe 1980s). Decentralization inevitably drawsmany stakeholders—namely, local authorities andassociations—into the decision-making process ingovernment and schools. Fearing a loss of negotiat-ing power on critical issues such as salaries and thehiring and firing of personnel, unions have stronglyopposed efforts to decentralize (Gaynor 1998).These experiences foreshadow the extent to whichunions’ political influence might affect both thepace and depth of decentralization in East Asia.

A review of 83 studies of school-based manage-ment in North America and among members of theOECD revealed greater teacher commitment, morecollaboration and information sharing, and achange in classroom instruction. In India, studentdropout rates declined and teacher attendanceimproved—from 33 percent to 78–86 percent—after village education committees began participat-ing in schools and teacher monitoring (Leithwoodand Menzies 1998; Pandey 2000). In the Philippines,about 2,000 schools adopted school-based manage-ment on a trial basis in connection with a WorldBank–supported education project. Preliminaryresults show that greater involvement among teach-ers in planning and managing schools has greatlyimproved their motivation and enthusiasm, whileprincipals’ efforts to fully involve teachers in identi-fying problems and needs, and to improve commu-nication, have helped yield solutions.

Countries vary in the extent to which they havedecentralized functions directly affecting teach-ers, such as recruitment, deployment, promotion,and salary scales. Nonetheless, underlying thesearrangements are incentives that influence thebehavior of teachers and ultimately affect the qual-

ity of teaching and student learning. Research onteacher incentives has shown that teachers are notonly sensitive to incentives but also responsive.Designing targeted incentives that translate intoimproved classroom performance and studentlearning is extremely difficult because teacher effortis both difficult and costly to measure (Murnaneet al. 1991; Hanushek 1986; Waterreus 2003).Several countries in Latin America, in parallel withdecentralization, have implemented policy reformsaffecting teacher incentives, such as Mexico’sCarrera Magisterial, Chile’s Sistema Nacional deEvaluación del Desempeño, and Brazil’s FUNDEF.

Annex: Research and Data Collection Methods

The 2003 World Bank study on which this chapteris based followed a method similar to that used inthe 1998 World Education Indicators survey byOECD-INES (Indicators of Education Systems).Yet the Bank study also differs in significant ways.What follows is a description and comparison ofour methodology and research approach, includingthe conceptual framework of the questionnaire,data collection procedure, and calculation of indi-cators. This annex also outlines the methods weused to make the two studies comparable.

Conceptual Framework of the Questionnaire

Decentralization focuses on the distribution ofpower between levels of government. The OECD-INES survey examined two dimensions of decen-tralization: the locus of decision making—that is,the level of government with authority—and themode of decision making, or the degrees to whichlevels of government are autonomous or shareauthority. The World Bank study examined the for-mer: the locus of decision making.

While the OECD-INES survey distinguished sixlevels of government (central, state, provincial/regional, subregional/intermunicipal, local, andschool), the World Bank study focused on threemain levels of government:

National/central: The central government consistsof all bodies at the national level that make deci-sions or participate in different aspects ofdecision making, including both administrative(government bureaucracy) and legislative bodies.


east_179-208.qxd 5/17/05 11:43 AM Page 200

State/provincial/local: The state is the first ter-ritorial unit below the nation in federal countries,or in countries with similar types of governmen-tal structures. The province or region is the firstterritorial unit below the national level in coun-tries that do not have a federal or similar type ofgovernment structure, and the second territorialunit below the national level in countries witha federal or similar types of governmentalstructures. The municipality or community isthe smallest territorial unit with governmentalauthority; the local authority may be the educa-tion department within a general-purpose localgovernment, or a special-purpose governmentwhose sole area of authority is education.

School: This level applies to individual schoolsand includes school administrators and teachersor a school board or committee established exclu-sively for that individual school. The decision-making body or bodies for this school may be anexternal school board, which includes residentsfrom the larger community; an internal schoolboard, which could include headmasters, teach-ers, other school staff, parents, and students; andboth an external and an internal school board.The study considered parents and teachers as anelement of the school level.

In practice, however, the decision-making processis not that simple. In determining at what level deci-sions are made, numerous unclear situations arise. Insome cases, a higher level of government may haveformal or legal responsibility for decision making,but in practice that level delegates its authority to alower level of government. In describing the actualdecision-making process, we identified the lowerlevel of government as the decision maker. Similarly,a higher level of government may provide a lowerlevel of government with choices in a particular area,such as the selection of textbooks, even though thehigher level establishes the framework for the deci-sion. In that case, too, we designated the lower levelof government as the actual decision maker. Finally,one level of government may have responsibility foran individual decision, but inaction results in a deci-sion by a lower level. If a decision is left to the discre-tion of a lower level through lack of determination athigher levels, then we chose the level that actuallymakes the decision.

Although the OECD-INES survey includedfewer indicators (35) than the World Bank study

(53), both studies organized those indicators intofour broad categories. The main items within thosecategories included:

Organization of instruction: Decisions regardingwhich school students will attend, schoolcareers, instruction time, choice of textbooks,grouping of pupils, assistance to pupils, teachingmethods, and methods for assessing pupils’regular work.

Personnel management: Decisions regardingthe hiring and firing of the principal and teachingand nonteaching staff, their duties and conditionsof service, their salary scales, and their careers.

Planning and structures: Decisions on creatingand abolishing schools and grade levels, selectingand designing programs of study and subjectstaught at a particular school, defining coursecontent, setting qualifying exams for certificatesor diplomas, and credentializing students.

Resource allocation and use: Decisions on allo-cating resources to a school for teaching staff, non-teaching staff, capital expenditures, and operatingexpenditures, and on using resources for staff,capital expenditures, and operating expenditures.

Data Collection Procedure

The OECD-INES approach to collecting data wassimilar to that of the World Bank, although the twostudies differed fundamentally in the composition ofexpert panels formed to assess decision makingat different levels. In the OECD-INES approach,researchers created a panel for each level of educa-tion, composed of one member from each of thethree decision-making levels. These groups com-pleted the questionnaire and arrived at a consensuson all questions. The researchers also composeda second panel for each level of education, againcomposed of one member from each of the threedecision-making levels, and repeated the process.The INES Network C representative or nationalcoordinator for the World Education Indicators sur-vey then compared the results of the two surveys.Where the responses differed, the INES Network Crepresentative used source documents and consultedthe national coordinator to reconcile disagreements.

We conducted the World Bank study in twophases, relying on intermittent consultationbetween two different panels of experts. The firstpanel was composed of World Bank education


east_179-208.qxd 5/17/05 11:43 AM Page 201


TABLE 9A.1 Cross-Study Comparison of Functions

World Bank Study (2003) OECD-INES (1998)

Pedagogical Organization of instruction

Admissions criteria to enter school Decision on what school a child should attendMode of grouping pupils/class size Mode of grouping pupilsNo. of periods of instruction/classroom hours Number of periods of instructionSelection of textbooks/teaching material Choice of textbooksDesign of programsDesign of program subjectsChoice of subject mattersDefinition of curriculumTeaching methods Teaching methodsProvision for extra help Assistance to pupilsExtracurricular activitiesEvaluation methods Methods of assessing pupils’ regular workStudent promotion Decisions affecting pupils, streamingSetting of equivalenciesSetting of goals/targets for the school

Administrative Planning and structureOpening/closure of school Creation and closure of a schoolSchool calendarCreation/abolition of grade(s) Creation and abolition of gradesDistribution of textbooks/teaching materialSetting of qualifying examinations Setting of qualifying examinationsCollection of student data (enrollment/exams)Community outreachAwarding of credentials Awarding of credentialsAccreditation of new schools Design of programs for a specific school type

Definition of course contentPersonnel management Personnel managementHiring and firing of staff Hiring and firing of staffHead of school PrincipalsTeaching staff TeachersGeneral staff Nonteaching posts

Terms of service and duties Duties and conditions of service of staffHead of school PrincipalsTeaching staff Teachers

Nonteaching postsFixing of salary levels/benefits/incentives Fixing of salary scales for staffHead of school PrincipalsTeaching staff Teachers

Nonteaching posts

experts with knowledge and work experiencewithin each of the countries. This group completedthe questionnaire and arrived at a consensus on allquestions. A second expert panel included WorldBank education experts located in the field inthe respective countries. This group reviewed theresults from the first round of surveys. Wherethe responses differed, this panel reconciled differ-ences with the first panel.

Calculating the Indicators

The OECD-INES study gave equal importance to

the indicators within each of the four domains.

Each domain contributes 25 percent to the results.

Because each domain includes a different number

of items, each item is weighted by the inverse of the

number of items in its domain. The World Bank

study followed the same approach.

east_179-208.qxd 5/17/05 11:43 AM Page 202


World Bank Study (2003) OECD-INES (1998)

Evaluation methods/supervisionHead of schoolTeaching staff

Promotion Influence over staff careers Head of school PrincipalsTeaching staff Teachers

Nonteaching postsTrainingHead of schoolTeaching staff

Certifying staffHead of schoolTeaching staff

Allocation and use of resources Allocation and use of resourcesAllocation of resources to the school Allocation of resources to the school Salaries and benefits Teaching staff; nonteaching staffAdministrative/operating costs Operating expendituresCapital expenditure Capital expenditures

(infrastructure/maintenance)Scholarship/awardsSchool projects/activities

Budget/financesUser charges/monthly fees/voluntary feesFundraising activitiesAllocation from state/subsidies

Procurement and disbursement Use of resources in the school Salaries and benefits StaffAdministrative/operating costs Operating expendituresCapital expenditures (infrastructure) Capital expendituresScholarship/awardsSchool projects/activities

Relationship with unions

Sources: OECD 1998; World Bank study 2003.

If a single item produced multiple answers,although the instructions for the questionnaire didnot specifically allow this, each answer received halfof the original weight of that item.

The following table compares the indicators usedin the two studies:

Endnotes

1. Accountability means more than “answerability” or“enforceability.” It implies a precise set of relationshipsbetween principals and agents encompassing five mainfeatures: delegation, finance, performance, informationabout performance, and enforceability (World Bank 2004c).

2. This, however, is not different from what had occurred inother countries that had decentralized earlier, such as Spain

and Chile (political reasons) and the United Kingdom andAustralia (concerns about state finances). After Franco diedin 1975, a new Spanish Constitution promoting a transitionto democracy devolved central functions to regional gov-ernments (Hanson 2000). Through an opposite politicaltransition in Chile—from a democratic to a militarygovernment—newly empowered neoliberal economistsand social planners also pushed for more decentralization.In contrast, Australia’s fiscal situation provided the princi-pal motives for public service reform. The VictorianCommission of Audit assessed public expenditures in keysectors as too high and pushed for incremental change(Pascoe and Pascoe 1998).

3. One impediment to further devolution of education func-tions is increased politicization of elections at the locallevel. In the Philippines, teachers serve on the Board ofElection Inspectors and thus play an important role incounting ballots (Manasan 2002).

east_179-208.qxd 5/17/05 11:43 AM Page 203

4. This was undertaken with the 2001 Decision on the Reformand Development of Basic Education by the State Council(Wang 2004).

5. Cambodia’s policy goals for education reform are con-tained in the Education Strategic Plan (ESP) for 2001–5and implemented through the Education Sector SupportProgram and the Priority Action Program rather thanthrough a single piece of legislation. The ESP reiterates thecentral government’s vision: “The Ministry’s vision of aninclusive education system also includes broad-based par-ticipation al all levels of Government and civil society intaking responsibility for planning and implementation ofeducation services . . . An associated goal would be toincrementally delegate greater decision-making and spend-ing authority to districts, possibly communes, and schools.In this way, all national stakeholders would also have toassume responsibility for frank and open evaluation of howthe system is performing and in taking steps to put thingsright” (Royal Kingdom of Cambodia 2001, p. 1).

6. The OECD-INES method clearly distinguishes between dejure and de facto decision-making power: “The descrip-tions of ‘at what level’ and ‘how’ educational decisions aremade reflect the actual decision-making process. In somecases, a higher level of government may have formal orlegal responsibility for decision-making, but in practice,that level of government delegates its decision-makingauthority to a lower level of government. In describing theactual decision-making process, the lower level of govern-ment is identified as the decision-maker. Similarly, a higherlevel of government may provide a lower level of govern-ment with choices in a particular area of decision-making(e.g., the selection of textbooks for particular course). Inthat case too, the lower level of government is the actualdecision-maker, but within a framework established by ahigher level of government. Finally, there are cases in whichone level of government may have the responsibility for anindividual decision, but inaction by the higher level resultsin a decision being made by a lower level within the educa-tional system” (OECD 1998, p. 407).

7. Although references are frequently made to state-leveldecentralization in Brazil, the process has progressed atsuch a rapid pace with FUNDEF’s implementation that, asmunicipalities and mayorships are given responsibility forthe first four years of basic education, discussing decentral-ization at the state level does not fully reflect the nationalcontext.

8. Local government expenditures on education rose nearly14-fold, from P= 0.8 billion in 1991 to P= 11.6 billion, in 2001(Manasan 2002).

9. The Local School Board is cochaired by the local chief exec-utive and the division superintendent. Other membersinclude the chair of the education committee of the locallegislature, the local treasurer, a representative of the fed-eration of local youth councils, the president of the localfederation of teacher-parent associations, a representativeof the local teachers’ organization, and a representative ofthe nonacademic personnel of the local public schools(Manasan 2002).

10. Because of a shortage of teachers at the local level, manylocal governments hire and pay supplementary teachersdespite the fact that this function is one of the primaryresponsibilities of the central Department of Education.

11. This high percentage may be due to the unclear distinctionbetween fees and voluntary contributions.

12. The share of central government spending in capital expen-ditures fell from 10.5 percent in 1990–91 to 4.6 percentin 1996–2000, and the central share of maintenance andoperating expenditures fell from 18.7 percent in 1990–91 to10.4 percent in 1996–2000 (Manasan 2002).

13. With the transfer of authority and management to regionalgovernments in Indonesia, “It is hoped that the local gov-ernments are capable of obtaining more funds in theirregions and in managing them more efficiently and effec-tively. It is expected that the local governments are able touse or exploit the sources of funds in the regions, such asthe private sector and other education stakeholders, infunding education” (Government of Indonesia 2001).

14. Until recently, the national government retained 10 percentof total tax revenues for its own use and provided 9 percentto local tax offices to assist with collections, 16 percent toprovincial governments, and 65 percent to local govern-ments (Lewis 2002).

15. The coefficient of variation in literacy rates declined from0.11 in 1989 to 0.08 in 1994.

16. There was no change in the coefficient of variation ofenrollment rates across the regions.

17. The creation of ARMM was a direct result of the 1996peace treaty between the national government and theMoro National Liberation Front. Republic Act No. 9054—the law creating ARMM—contains detailed provisions thatpertain to the governance, regulation, and funding ofhuman development sectors in the region. However, the1991 Local Government Code had already mandated devo-lution of many functions and responsibilities related tohuman development—notably health and social welfareservices—to provinces, cities, municipalities, and barangays(World Bank 2003b).

18. These comparisons exclude a few provinces where thesamples tended to be in urban areas. According to theSUSENAS Work Manual (Government of Indonesia 2002),“Because of the unfavorable security situation, in the fol-lowing provinces/regencies SUSENAS 2002 is only con-ducted in Banda Aceh (Aceh), Ambon (Maluku), Ternate(North Maluku), Sorong (West Irian Jaya), Timica (CentralIrian Jaya), and Jayapura (East Irian Jaya).” Because thesamples were not representative, enrollment rates in theseprovinces appear to be surprisingly higher than in otherprovinces.

19. Education systems suffer from ineffective and substandardschools, persistent shortages of good textbooks, and unpre-pared and absent teachers. The nondelivery of publiclysupplied textbooks and chairs at the beginning of eachschool year in several of these countries deprives millionsof children of the chance to do better in school. In thePhilippines, this problem was estimated at 30–60 percentof total contracts with the education central agency(Chua 1999).

The proportion of teachers who are absent from theclassroom is too high, according to a recent survey of teach-ers in several countries. For example, in Indonesia, 17 per-cent of teachers in primary schools were not on the schoolpremises during school hours. And too many teachers donot know their subject matter better than their students, asindicated by the performance of teachers on tests con-ducted for a recent study of the quality of primary schoolsin Vietnam (World Bank 2004b). These problems are alsofound in a much wider set of countries for which policyand program evaluations do not point conclusively to gains


east_179-208.qxd 5/17/05 11:43 AM Page 204

from higher per pupil spending or from investments inspecific school inputs (Glewwe 2002).

20. The newly established Academic Accountability Council,along with the Office of Accountability, was jointly respon-sible for overseeing a system of review and analysis of schoolperformance, while local school councils lost some of theirindependence.

21. See the Chicago Public Schools website: www.cps.k12.il.us/.22. Paes de Barros and Silva Pinto de Mendonça (1998) con-

ducted a broad evaluation of decentralization acrossmost Brazilian states. Their research base includes allgeographical units in the country except the Federal Dis-trict, northern states, and the state of Alagoas. The studyexamined data between 1981 and 1993 and included 220observations.

23. The measures of educational performance used includedthe repetition rate (school census); the proportion ofchildren outside school and two measures of grade-level lag(National Household Survey Sample, or Pesquisa Nacionalpor Amostra de Domicílios); and student achievement(National Basic Education Evaluation System, or Sistema deAvaliação do Ensino Básico).

24. The 10-year-old Basic Education Quality ImprovementProject (Próqualidade) aimed to strengthen school auton-omy by providing managers in central and regional officeswith tools designed for information-based decision mak-ing; and help school directors assume their new roles asboth leaders and managers in a more decentralized stateeducation system. The project also aimed to increase teach-ers’ access to training opportunities; deliver packages ofinstructional materials to public schools; and upgradefacility management to ensure equitable access of pupils toclassroom time as well as rational use of school space.

25. The LEAs were created according to population distribu-tion and density, number of institutions, geographic char-acteristics, sociocultural considerations, and the extent towhich planned areas overlapped with existing districtboundaries.

26. This act was supported by the National DecentralizationCommittee and part of the Decentralization Action Planpassed by Parliament in early 2002.

27. In China, the monitoring and evaluation system is quitecomprehensive. Throughout compulsory education, stu-dents must take exams and tests following each semesterand school year to graduate. In primary schools, studentsmust pass tests in Chinese and mathematics, while tests inthe remaining subjects are usually used as checkup. In sec-ondary schools, exam subjects reflect the general subjectstaught to a given graduating class, while exams in theremaining subjects are again used for checkup (China Edu-cation and Research Network 2003). Junior secondaryschool graduates seeking promotion to senior secondaryschools must pass locally organized entrance exams priorto full admission.

In Indonesia, final exams are set at both the nationaland school levels. While the national portion is based onmultiple choice, the school portion is based on laboratorywork, written portfolios, or demonstration of a given set ofskills. Students are tested on core knowledge in three mainsubject areas: mathematics, Indonesian, and English. At thecentral level, test results play a strong role in resource alloca-tion. At the school level, they provide comparable informa-tion on the relative strengths and weaknesses of individualschools. At the client level, parents and the community can

use exam results to hold providers accountable. However,the new exam system does not include primary school, anddoes not provide information about the performance ofchildren below the ninth grade. Confronted with this,provinces and districts can create their own complementarytesting system using guidance available from the NationalEvaluation Center.

28. Some have argued, however, that even when reformsinvolve quality issues, unions still oppose them, given thatsuch reforms require more effort and political sacrifices(Corrales 1999). Similar debates stem from the joint func-tion of unions as both professional organizations aiming topromote efficacy and public knowledge and agents of col-lective bargaining (World Bank 2004c).

29. In Mexico, given strong leadership by the teachers’ union,the government quickly realized that it could neither con-front nor ignore it, and so openly included it in the processof negotiation on greater decentralization. Unions in Chileare also politically powerful and “have systematically stoodin the way of true curriculum reform and teacher account-ability, and the political class has been unwilling to con-front them” (Edwards 2003). Less contentious than inMexico and Chile, reform in Bolivia has shown progress innegotiating with teachers. Nonetheless, teachers’ unionsare also extremely powerful and reform has stalled at sev-eral points. In contrast, unions in Brazil did not stallreform; union representatives were active stakeholdersin reform negotiations, and decentralization progressed tothe school level. And, finally, in El Salvador, teachers’unions took a strong stance against the EDUCO model,but government dissuaded them with evidence of innova-tion in teaching practices (Marques and Bannon 2003;Stavenhagen 1999).

Bibliography

Ahmad, Junaid, Richard Bird, and Jennie Litvack. 1998.“Rethinking Decentralization in Developing Countries.”PREM Sector Studies Series. Washington, DC: World Bank.

Akita, Takahiro, and Armida S. Alisjahbana. 2002.“RegionalIncome Inequality in Indonesia and the Initial Impact of theEconomic Crisis.” Bulletin for Indonesian Economic Studies38 (2): 201–22.

Asian Development Bank. 1999. Philippine Education for the 21stCentury. 1998 Philippines Education Sector Study. Manila:Asian Development Bank.

Azfar, Omar, Satu Kahkonen, and Patrick Meagher. 2001.“Conditions for Effective Decentralization Governance: ASynthesis of Research Findings.” College Park, MD: Univer-sity of Maryland, IRIS Center.

Bray, Mark. 1999. “The Private Costs of Public Schooling:Household and Community Financing in Primary Educa-tion in Cambodia.” Paris and Buenos Aires: InternationalInstitute for Educational Planning.

Cheng, Kai-ming. 1997. “The Meaning of Decentralization:Looking at the Case of China.” In International Handbook ofEducation and Development: Preparing Schools, Students, andNations for the Twenty-First Century, eds. William Cum-mings and Noel F. McGinn. New York: Pergamon Press.

China Education and Research Network. 2003. www.edu.cn/HomePage/english/index.shtml.

Chicago Public School System. n.d. “Chicago Public Schools:Every Child, Every School.” www.cps.k12.il.us/.


east_179-208.qxd 5/17/05 11:43 AM Page 205

Chua, Yvonne T. 1999. Robbed: An Investigation on Corruption inPhilippine Education. Manila: Philippine Center for Inves-tigative Journalism.

Cordeiro Guerra, Susana. 2003. “A Proposal for TeacherEmpowerment in Minas Gerais.” Cambridge, MA: HarvardUniversity.

Corrales, Javier. 1999. “The Politics of Education Reform: Bol-stering the Supply and Demand—Overcoming InstitutionalBlocks.” Education Reform and Management Series 2 (1).Washington, DC: World Bank.

Delannoy, Françoise. 2000. “Education Reforms in Chile,1980–98: A Lesson in Pragmatism.” Education Reform andManagement Series 1 (1). Washington, DC: World Bank.

Edwards, Sebastian. 2003. “How Chile Can Make the Most of ItsU.S. Trade Deal.” Wall Street Journal, January 3.

Gaynor, Cathy. 1998. Decentralization of Education: TeacherManagement. Directions in Development Series. Washing-ton, DC: World Bank.

Glewwe, Paul. 2002.“Schools and Skills in Developing Coun-tries: Education Policies and Socioeconomic Outcomes.”Journal of Economic Literature 40 (2): 436–82.

Government of Indonesia. 1998 and 2002.“Survei SosialEkonomi Nasional” [SUSENAS]. Jakarta: Bureau of CentralStatistics.

———. 2000. Statistik Indonesia [Statistical Yearbook of Indone-sia]. Jakarta: Badan Pusat Statistik.

———. 2001. “Decentralization of Education.” National Com-mittee for Education, Subcommittee III, draft report. Trans-lated into English. Jakarta: Government of Indonesia.

———. 2003. Act of the Republic of Indonesia Number 20, Year2003, on National Education System. Jakarta: NationalGazette.

Guedes, Andrea, Thereza Lobo, Robert Walker, and Ana LúciaAmaral. 1997. Gestión Descentralizada de la Educación en elEstado de Minas Gerais, Brasil. Grupo de DesarrolloHumano América Latina y La Región del Caribe, informeno. 11. Washington, DC: World Bank.

Hannum, Emily. 2002.“Educational Stratification by Ethnicityin China.” Demography 39 (1): 91–117.

Hanson, Mark E. 2000. Democratization and Educational Decen-tralization in Spain. Education Reform and ManagementSeries. Washington, DC: World Bank.

Hanushek, Erik A. 1986. “Production and Efficiency in PublicSchools.” Journal of Economic Literature 24 (3): 1141–77.

Hawkins, John N. 2000.“Centralization, Decentralization, Recen-tralization: Educational Reform in China.” Journal of Educa-tional Administration 38 (5): 442–55.

Hsieh, Chang-Tai, and Miguel Urquiola. 2001. “When SchoolsCompete, How Do They Compete? An Assessment of Chile’sNationwide School Voucher Program.” elsa.berkeley. edu/~chsieh/c10.pdf.

James, Estelle, Elizabeth King, and Ace Suryadi. 1996. “Finance,Management, and Costs of Public and Private Schoolsin Indonesia.” Economics of Education Review 15 (4): 387–98.

Jimenez, Emmanuel, and Vicente Paqueo. 1996. “Public for Pri-vate: The Relationship between Public and Private SchoolEnrollment in the Philippines.” Economics of EducationReview 15 (4): 389–99.

Jimenez, Emmanuel, and Yasuyuki Sawada. 1999. “DoCommunity-Managed Schools Work? An Evaluation of ElSalvador’s EDUCO Program.” World Bank Economic Review13 (3): 415–42.

Kerr, Graham. 1999.“Local Technical and Managerial Capacity.”In Decentralization Briefing Notes, eds. Jennie Litvack andJennica Seddon. Washington, DC: World Bank Institute.

King, Elizabeth, and Berk Ozler. 1998. “What’s DecentralizationGot to Do with Learning? The Case of Nicaragua’s SchoolAutonomy Reform.” Washington, DC: World Bank, Devel-opment Research Group.

Kingdom of Thailand. 1999. National Education Act, B.E. 2542.Bangkok: Office of the National Education Commission.

Leithwood, Kenneth A., and Teresa Menzies. 1998. “Forms andEffects of School-Based Management: A Review.” Educa-tional Policy 12 (3): 325–46.

Lewis, Blane D. 2002. “Revenue-Sharing and Grant-Making inIndonesia: The First Two Years of Fiscal Decentralization.” InIntergovernmental Transfers in Asia, ed. Paul Smoke. Manila:Asian Development Bank.

Loehr, W., and Rosario Manasan. 1999. “Fiscal Decentralizationand Economic Efficiency: Measurement and Evaluation.”IMCC Consulting Assistance for Economic Reform II.Cambridge, MA: Harvard Institute for International Devel-opment and U.S. Agency for International Development.

Lopez-Acevedo, Gladys, Vicente Paqueo, and Suhas Parandekar.2003. “On the Use of Transparent Formulae to AllocateFederal Education Transfers.” Washington, DC: World Bank,Latin America and the Caribbean Region.

Machado, Maria Aglaê de Medeiros. 2002. “School Autonomy inBrazil: General Overview and Selected States.” Brasília:World Bank.

Manasan, Rosario. 2002. “Education Financing and ServiceDelivery: Selected Issues.” Background Paper for PhilippinesPublic Expenditure: Procurement & Financial ManagementReview 3 (annex VI).

Manasan, Rosario, and John M. Atkins. 2004. “NormativeFinancing in Basic Education.” Manila: Philippine Instituteof Development Studies.

Marques, José, and Ian Bannon. 2003. “Central America: Educa-tion Reform in a Post-Conflict Setting—Opportunities andChallenges.” Conflict Prevention and Reconstruction Work-ing Paper. Washington, DC: World Bank, Social Develop-ment Department.

Martin, Michael O., Ina V. S. Mullis, Eugenio J. Gonzalez, StevenJ. Chrostowski. 2004a. TIMSS 2003 International Mathemat-ics Report: Findings from IEA’s Trends in Mathematics andScience Study at the Fourth and Eighth Grades. Boston: Inter-national Association for the Evaluation of EducationAchievement (IEA).

———. 2004b. TIMSS 2003 International Science Report: Find-ings from IEA’s Trends in Mathematics and Science Study atthe Fourth and Eighth Grades. Boston: International Associa-tion for the Evaluation of Education Achievement (IEA).

McEwan, Patrick J., and Martin Carnoy. 1999. The Effectivenessand Efficiency of Private Schools in Chile’s Voucher System.Stanford, CA: Stanford University Press.

McMahon, W., with N. Suwaryani, Boediono, and E. Appiah.2001. Improving Education Finance in Indonesia. Jakarta:Ministry of National Education, Institute for Researchand Development, Policy Research Center; UNICEF; andUNESCO.

Murnane, Richard, Judith Singer, John Willett, James Kemple,and Randall Olsen. 1991. Who Will Teach? Policies ThatMatter. Cambridge, MA: Harvard University Press.

Mutebi, Alex M. 2003. “Thailand’s Decentralization Experiment:Evolution: Dimensions and Challenges.” Singapore: NationalUniversity of Singapore, Lee Kuan Yew School of PublicPolicy.

Namo de Mello, Guiomar. Forthcoming. “Introdução: BreveHistórico.” Background paper for Lemann Foundation.http://www.fundacaolemann.org.br.


east_179-208.qxd 5/17/05 11:43 AM Page 206

National Statistical Coordination Board. 2003. Philippine Statis-tical Yearbook. Manila: National Economic and DevelopmentAuthority.

OECD (Organisation for Economic Co-operation and Develop-ment). 1998. Education at a Glance: OECD Indicators. Paris:OECD, Centre for Educational Research and Innovation.

Paes de Barros, Ricardo, and Rosane Silva Pinto de Mendonça.1998. “The Impact of Three Institutional Innovations inBrazilian Education.” In Organization Matters: Agency Prob-lems in Health and Education in Latin America, ed. WilliamD. Savedoff. Baltimore, MD: Johns Hopkins University Press.

Pandey, Raghaw. 2000. “Going to Scale with Education Reform:India’s District Primary Education Program, 1995–99.” Edu-cation Reform and Management Publication Series 1 (4).Washington, DC: World Bank.

Pascoe, Susan, and Robert Pascoe. 1998. “Education Reform inAustralia, 1992–97.” Education Reform and ManagementPublication Series 1 (2). Washington, DC: World Bank.

People’s Republic of China. 2001. Essential Statistics of Educationin China, Vol.1. Beijing: Ministry of Education, Departmentof Development and Planning.

———. 2002. China Statistical Yearbook, Vol. 21. Beijing:National Bureau of Statistics, China Statistics Press.

Prawda, Juan. 1992. “Educational Decentralization in LatinAmerica: Lessons Learned.” Washington, DC: World Bank,Latin America and the Caribbean Region, Human ResourcesDivision.

Rodríguez, Alberto, and Kate Hovde. 2002. “The Challenge ofSchool Autonomy: Supporting Principals.” Washington, DC:World Bank, Latin America and the Caribbean Region,Department of Human Development.

Ross, Steven M., William L. Sanders, S. Paul Wright, and SamStringfield. 1998. “The Memphis Restructuring Initiative:Achievement Results for Years 1 and 2 on the Tennessee Value-Added Assessment System (TVAAS).” Memphis, TN: Univer-sity of Memphis, Center for Research in Educational Policy.

Royal Kingdom of Cambodia. 2001. Education Sector Plan, sectionI (1). Phnom Penh: Ministry of Education, Youth, and Sport.

Shen, A. 1994.“Teacher Education and National Development inChina.” Journal of Education 176 (2): 57–71.

Stavenhagen, Rodolfo. 1999. “Latin America’s Challenge for the21st Century.” Cambridge, MA: Harvard University, DavidRockefeller Center for Latin American Studies.

Trends in International Mathematics and Science Study. 1999.http://isc.bc.edu/timss1999.html.

Tsang, Mun C. 2002.“Intergovernmental Grants and the Financ-ing of Compulsory Education in China.” New York: Colum-bia University, Teachers College.

Turner, Mark. 2002. “Whatever Happened to Deconcentration?Recent Initiatives in Cambodia.” Public Administration andDevelopment 22 (4): 353–64.

UNESCO (U.N. Education, Scientific, and Cultural Organiza-tion). 2002/2003. “Education Statistics.” Paris: UNESCO,Institute for Statistics.

Vasconcelos-Saliba, Alcyone. 2004. “Governance and Account-ability Tools for Accountability: School Report Cards—The Case of Paraná (Brazil).” Presentation at the Interna-tional Conference on Governance Accountability in SocialSector Decentralization. Washington, DC: World Bank.

Vegas, Emiliana. 2002. “School Choice, Student Performance,and Teacher and School Characteristics: The Chilean Case.”Washington, DC: World Bank.

Wang, Yidan. 2004. “Governance of Basic Education: ServiceProvision and Quality Assurance in China.” Washington,DC: World Bank.

Waterreus Ib. 2003. “Lessons in Teacher Pay: Studies on Incen-tives and the Labor Market for Teachers.” Dissertation.Amsterdam: University of Amsterdam.

Weist, Dana. 2001. “Thailand’s Decentralization: Progress andProspects.” Paper presented at the KPI Annual Congress IIIIon Decentralization and Local Government in Thailand,November.

West, Loraine A., and Christine P. W. Wong. 1995. “Fiscal Decen-tralization and Growing Regional Disparities in RuralChina: Some Evidence in the Provision of Social Services.”Oxford Review of Economic Policy 11 (4): 70–84.

———. 1997. “Equalization Issues.” In Financing Local Govern-ment in the People’s Republic of China, ed. Christine P. W.Wong. Hong Kong: Oxford University Press, pp. 283–311.

Winkler, Donald, and Alec Gershberg. 2000.“EducationDecentralization in Latin America: The Effects on theQuality of Schooling.” In Decentralization and Accountabil-ity of the Public Sector, eds. Shahid Javed Burki et al. Pro-ceedings of the 1999 Annual World Bank Conference onDevelopment in Latin America. Washington, DC: WorldBank.

World Bank. 2000. “China National Development and Subna-tional Finance: A Review of Provincial Expenditures.”Washington, DC: World Bank, East Asia and Pacific Region,Poverty Reduction and Economic Management Unit.

———. 2003a.“Decentralizing Indonesia.” Regional PublicExpenditure Review overview report. Washington, DC:World Bank, East Asia and Pacific Region, Poverty Reduc-tion and Economic Management Unit.

———. 2003b. “From Cyclical Recovery to Long-Run Growth:Regional Overview.” Washington, DC: World Bank, East Asiaand Pacific Region, Poverty Reduction and EconomicManagement Unit, East Asia Update.

———. 2003c.“Human Development for Peace and Prosperityin the Autonomous Region in Muslim Mindanao.” Washing-ton, DC: World Bank, East Asia and Pacific Region, HumanDevelopment Sector Unit.

———. 2003d. Thailand Country Development Partnership:Governance and Public Sector Reform. Phase I monitoringworkshop.

———. 2003e.“Philippines: Decentralization and ServiceDelivery—From Promise to Performance. Cross-CuttingThemes in Local Government Functioning.” Washington,DC: World Bank, East Asia and Pacific Region, PovertyReduction and Economic Management Unit.

———. 2003f. “Vietnam: Delivering on Its Promise.” Develop-ment Report. Washington, DC: World Bank, East Asia andPacific Region, Poverty Reduction and Economic Manage-ment Unit.

———. 2004a. “Education in Indonesia: Managing the Transi-tion to Decentralization.” Indonesia Education Sector Review.Prepared in partnership with the Government of Indonesia,Ministry of National Education; and the Indonesia EducationDonor Partnership. Jakarta: World Bank.

——— . 2004b. “The Quality of Education at the End ofPrimary School: Achievement in Grade 5—Baseline Data forthe Nation and Each Province,” Vol. 2. Washington, DC, andHanoi: World Bank.

———. 2004c. World Development Report: Making ServicesWork for Poor People. Washington, DC: World Bank; andNew York: Oxford University Press.

Zhang, Tiedao, Zhao Minxia, Zhao Xueqin, Zhang Xi, and WangYan. 2003.“Universalizing Nine-Year Compulsory Educationfor Poverty Reduction in Rural China.” Case study on scalingup poverty reduction for Shanghai Poverty Conference.


east_179-208.qxd 5/17/05 11:43 AM Page 207

east_179-208.qxd 5/17/05 11:43 AM Page 208

Documents

Education Reforms in East Asia: Policy, Process, and Impactsiteresources.worldbank.org/INTEAPDECEN/Resources/Chapter-9.pdf · Education Reforms in East Asia: Policy, ... and the political