Donald R. Winkler andAlec Ian Gershberg*

Nº 17

Education Decentralizationin Latin America:

The Effects on the Qualityof Schooling

APRIL, 2000.-

* Donald R. Winkler is Lead Specialist for Human Development in theLAC Region of the World Bank.

* Alec Ian Gershberg is Assistant Professor at the Milano School ofManagement and Urban Policy, the New School University, andFaculty Research Fellow at the National Bureau of EconomicResearch.

The expressed opinions in this document are responsability of theauthors and don’t compromise Preal.

2

As we have seen, over the past decadedecentralization of government has become commonthroughout Latin America. The education sector isno exception, and there has been a rapid increase inthe number of countries implementing significantdecentralization reforms (see Figure 1). At the sametime, there has been a worldwide trend to give schoolsgreater decision-making autonomy, in the interest ofimproving school performance and accountability.School systems as diverse as those in Victoria, Aus-tralia; Memphis, Tennessee; and Minas Gerais,Brazil, have given authority to school heads, and thenthrough a variety of mechanisms held themresponsible for school performance.

Figure 1

The two types of education decentralization—tolower levels of government and to individualschools—have very different origins and aims. Thedecentralization of education to lower levels ofgovernment has almost without exception beenundertaken in the context of a more generaldecentralization of government, the causes of whichvary widely. The decentralization of education to in-dividual schools, on the other hand, has typically beenmotivated by concerns about poor school performan-ce. Both types of education decentralization are wellrepresented in Latin America, and this chapterreviews the evidence to date on their various impactson schooling.

The literature on education decentralization isgrowing rapidly, but it is still primarily descriptivein nature. Attempts to assess the impacts ofdecentralization have suffered from weak baseline

data and poor research designs, mainly resulting frominadequate data. Weak evaluations are not limited toLatin America or developing countries. For example,Summers and Johnson (1991) reviewed more than600 evaluations of school-based management in theUnited States and found only two with an adequateresearch design.

Several recent studies and evaluations of primaryand secondary education, both in Latin America andin other regions, provide the basis for this chapter, ofwhich three merit mention. The World Bank recentlycompleted several studies on educationdecentralization worldwide (Fiske 1996; Gaynor1998); the Inter-American Development Banksponsored research on the effects of differentorganizational arrangements in education in Brazil,Chile, and Venezuela (Savedoff 1998); and the Cen-tro Estudios para America Latina (CEPAL) workedwith researchers in five countries (Bolivia, Brazil,Colombia, Mexico, and Nicaragua) to assesseducation decentralization strategies (di Gropello1998). In addition, this chapter draws on severalcountry-specific evaluations from Latin America andselected evaluations from outside the Region.

RATIO NALE FOR EDUCATIONDECENTRALIZ ATION

The economic rationale for decentralizingeducation is to improve technical and social efficiency(Winkler 1994). Decentralized decision-making, itis argued, will give local voter-consumers greatervoice in the service mix that they receive and, hence,raise their welfare. Presumably, the more local thedecision the greater the voter-consumer voice willbe—that is, greater at the school level than the muni-cipal level, and greater in single-purpose (forexample, school district) than general-purposegovernments. If the finance and supply of educationis determined locally, the improvement in socialwelfare will be still greater, for the median voter-consumer will tax himself or herself only up to thatpoint where the marginal tax costs and marginaleducational benefits are equal.

However, these arguments presume a world inwhich democracy works well, and in which allexternalities are captured locally. If there is the riskthat local elites capture local decision-making, so-

Countries Implementing Education DecentralizationReforms

0

2

4

6

8

10

12

14

1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998

Num

ber

of C

ount

ries

ES, BRZ,ARG, COL

GUA, BOL

HON, BOL, CHI

MEX, PAR, NIC, PAN1998

1997

1995

1994VEN

1993COL, NIC

1992MEX

1991

BRZ1988

CHI1981

ARG1976

3

cial welfare may not improve; this risk may be higherin societies with little experience in participativedemocracy at the local level. If the externalitiesalleged to result from education, especially basiceducation, are distributed beyond the confines of thelocality, there is a strong argument for a highpercentage of financing coming from centralizedsources. Ensuring equality of educationalopportunity, as measured at a minimum by equalityin educational spending, is a further argument for ahigh degree of centralized financing in countrieswhere income inequality is high.

Improved technical efficiency is the other rationalefor education decentralization. Here the argument hasseveral elements. First, to the extent that prices andproduction processes vary across localities, there areobvious efficiencies resulting from letting localdecision-makers allocate budgets across inputs.Second, in situations where the capacity of central-government ministries to monitor and supervise lo-cal schools has been weak, devolving theseresponsibilities to local voter-consumers mayincrease the accountability of the school for its per-formance. The interest of local voter-consumers maybe higher, if they are also contributing resources—financial or non-financial—to the school.

A final argument for decentralization is thathaving many suppliers rather than just one supplieris likely to lead to a wider variety of experiences andinnovations. If there are adequate means forcommunicating and exchanging information on theseexperiences, a decentralized system may lead to morerapid innovation and change than a centralized one.There is some evidence for this argument in the caseof Brazil (Xavier, Sobrinho, and Marra).

THE EDUCATIO NAL CONTEXT OFDECENTRALIZ ATION

The problem of access to basic schooling has beensolved for most children in Latin America. Now, thereis a growing consensus that it is the quality ofeducation that must be improved, especially in thepublic schools and especially for poor children(Summit of the Americas II 1998). Low quality isreflected in high rates of repetition and dropout andlow performance on standardized tests of scholasticachievement. The Latin American and Caribbean

countries that have participated in international testsof science and mathematics have scored slightlyabove African countries and well below East Asiancountries (see Figure 2).

Figure 2

Source: International Association for the Evaluation of Educational Achievement(IEA). (1996). Mathematics Achievement in the Middle School Year: IEA’s ThirdInternational Mathematics and Science Study (TIMSS). Center for the Study ofTesting, Evaluation, and Educational Policy, Boston College. November.

Figure 3

Average Math AchievementTest Scores of Eight Graders,

Selected Countries

643607

527

454

385354

100

200

300

400

500

600

700

Mean Score

Singapore

Korea

Ireland

Portugal

Colombia

S. Africa

0

150 200 250 300 350 400

Argentina

Bolivia

Brazil

Chile

Colombia

Cuba

Honduras

Mexico

Paraguay

Dom. Rep.

Venezuela

Third-Grade Language Achievement Scores(Median, 25%, 75%)

Source: UNESCO data in PREAL/CINDE 1999.

In addition, the evidence coming from a UnitedNations Educational, Scientif ic and CulturalOrganization (UNESCO) test of educationalachievement administered in 11 LAC countries showsthat, excluding Cuba, the performance of mostcountries in LAC does not differ greatly, suggestingthat most LAC countries would fare poorly oninternational achievement tests (see Figure 3) (La-boratorio Latinoamericano de Evaluación de la Ca-lidad de la Educación 1999). The low quality of basiceducation constrains the quality of higher levels ofeducation and puts LAC at risk in its capacity to com-pete economically with the rest of the world. Inaddition, while children from all income groups nowhave access to basic schooling, there remain largeinequalities in educational opportunity as measuredby quality of schooling. Compared with children fromeconomically advantaged homes, children from poorhouseholds are likely to receive lower schoolinginvestments from both the home and the school.

While the rationale for decentralization is at leastas much political as it is educational, the proponentsof decentralization expect one impact to be improvedquality. Other possible effects are changes inefficiency and equity. Due to the importance of raisingquality and the limited information available onefficiency and equity, this paper focuses on the impactof decentralization on educational quality in LAC.

TYPOLOGY

Decentralization takes many forms. It varies bythe level of government to which decisions are de-volved, the kinds of decisions moved to other levelsof government, and the orientation of thedecentralization—emphasis on governance changesversus. emphasis on pedagogic changes.

Level of Decentralization

The level to which educational decisions aredecentralized ranges from regional and localgovernment to the community and the school. Inmany federal countries—Brazil, Canada, Germany,India—the states or provinces that make up thefederation have had a constitutional responsibility foreducation. In other countries—Argentina, Mexico,

4

Venezuela—education responsibilities havehistorically been situated in the central government,but they have been largely devolved to states orprovinces over the past decade.

Local governments quite often have educationalresponsibilities, especially for primary and secondaryschooling. In the United States, most stategovernments have devolved educational managementto single-purpose local governments, or schooldistricts. In other countries—Brazil, Chile, Colom-bia—municipalities have been given increasededucational responsibilities over the past decade.

Finally, some countries have given school councilsand schools significant autonomy in managing (butrarely financing) education. The Netherlands isperhaps the best example of a country that hasempowered parents to create their own schools withf inancing and other support from the centralgovernment. Recently, in cities like Chicago andMemphis in the United States, it is the school districtthat has given the school significant managementautonomy.

Decision-Making Powers

Some educational functions are decentralizedeven within centralized systems, and others arecentralized even within decentralized systems. AnOECD survey of its members, for example, showsthat, even in centralized systems, schools make mostof the decisions about the organization of instruction.These decisions include choice of teaching methods,textbooks, criteria for grouping students withinschools, and day-to-day methods of studentassessment. On the other hand, in most Europeancountries, most personnel-management decisions aremade at a central level.

The OECD methodology for measuring the degreeof education decentralization divides educationalfunctions into four groups: the organization ofinstruction, personnel management, planning andstructures, and resources. For the purposes of thispaper, we adapted these definitions to be consistentwith Latin American experience and availableinformation. The content of each group is given inTable 1.

5

Table 1Types of Decisions That May Be Decentralized

Organization of Instruction Select school attended by student. Set instruction time. Choose textbooks.Define curriculum content. Determine teaching methods.

Personnel Management Hire and fire school director. Recruit and hire teachers. Set or augmentteacher pay scale. Assign teaching responsibilities. Determine provisionof in-service training.

Planning and Structures Create or close a school. Selection of programs offered in a school.Definition of course content. Set examinations to monitor schoolperformance.

Resources Develop school improvement plan. Allocate personnel budget.Allocate non-personnel budget. Allocate resources for in-serviceteacher training.

Structure and Content

Just as the composition of educational functions thatare decentralized varies across countries, so too doesthe goal and orientation of the decentralization reforms.In some reforms, local control is the goal, either forpolitical reasons or to strengthen accountability by theschools to its clients. The focus of these reforms is onstructure—that is, transferring decision-makingpowers and responsibilities to lower levels ofgovernment or to school councils. Implicit in thesereforms is the expectation that local control andaccountability will improve efficiency, both in the usesof resources and in the match between client demandand the supply of school services.

In other reforms, the goal is improved learning, andthe transfer of decision-making powers is simply avehicle for attaining that goal. These reforms put moreemphasis on the content of education reform than onthe structure itself. Parental participation is valued bythese reforms because it is viewed as contributing tothe success of education and not because it improvesaccountability. Matching client demand with what theschools offer is important only to the extent that clientdemand is consistent with raising quality.

While it is tempting to contrast structural reformswith reforms that emphasize content, this typology isin fact a continuum, with most decentralization reformsencompassing elements of each.

TYPOLOG Y APPLIE D TO RECENT L ATINAMERICAN EXPERIENCE

Education decentralization has taken many formsin Latin America and the rest of the world. It alwaysincludes the transfer of authority and responsibilityfrom higher to lower levels of government, but it variesconsiderably in terms of which decision-makingpowers are decentralized and who receives those newpowers. Figure 4 illustrates the wide variety in LatinAmerican and OECD countries in the location ofimportant educational decisions. In addition, sinceeducation decentralization is often part of a broadereducation reform effort, there is considerable variationin practice in terms of accompanying schoolimprovement measures.

In the discussion that follows, the typology will beapplied to the experiences of Argentina, Brazil (with afocus on Minas Gerais State), Chile, El Salvador,Mexico, and Nicaragua.

Figure 4

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Neth. N.Z. Spain Chile Parag. Arg. U.S. France

Central

Regional

Local

School

Level of Decision-Making in Education Sector

Source: OECD 1998

6

Level of Decentralization

The level of education decentralization varieswidely within Latin America. In Argentina, primaryand secondary education and the normal schools weretransferred from the central government to the provin-cial governments (in 1976 and 1991, respectively), andtoday most decision-making authority remainsconcentrated in the provincial education ministries. Inthis respect—the concentration of decision-makingauthority at the regional level—Argentina presents aunique model in Latin America, although Mexicoappears to be quickly evolving in a similar fashion.

Brazil has a long tradition of decentralizededucation, with most authority concentrated at the stategovernment level. The state’s pre-eminent role insecondary education was confirmed by the 1988constitution, and municipalities were given the pre-eminent role in financing and delivering primary andpreschool education. In addition, during the 1990s,some states (for example, Minas Gerais) havetransferred significant decision-making authority to thelevel of the school.

Chile’s education decentralization effort is long andcomplicated. It began in 1981 with the transfer ofdecision-making authority to the municipalities, on theone hand, and to nonprofit schools, on the other. Itcontinued in the 1990s with the central government’sexercising stronger pedagogic leadership and workingdirectly with the schools to bring about school-levelimprovements.

El Salvador’s decentralization effort was not uni-versal but, instead, targeted rural areas where centralgovernment schools failed to function during the civilwar. Hence, while for traditional public schoolseducational decision-making remained concentrated atthe level of the central government, the new ruralschools, called EDUCO (the Spanish acronym forEducation with the Participation of the Community)were given significant decision-making authority andautonomy. The success in implementing the EDUCOmodel has led to current efforts to decentralizetraditional schools as well.

Mexico’s education decentralization is acombination of the Argentine and Salvadoran models.The 1993 the Ley General de la Educación transferred

most educational decision-making authority forprimary and secondary schools to the stategovernments, but the central government’s importantrole in financing education through negotiated transfersto the states resulted in de facto continuedcentralization. Real decentralization to the statesoccurred only in 1998 when education transfers becameautomatic. In addition, the central governmentcontinues to directly operate a system of rural schools,called CONAFE (the Spanish acronym for NationalBoard for Educational Improvement), to ensurelearning opportunities for remote rural, and especiallyindigenous, children. While not nearly as autonomousas El Salvador’s EDUCO schools, the CONAFEschools give parents a considerably more importantrole than is found in the traditional public schools.

Finally, Nicaragua’s education decentralization hasevolved from an emphasis in the early 1990s onmuncipalization, to a clear policy in the late 1990s totransfer most important educational management andfinance decisions to the level of the school.

Several other countries in the region have alsoadopted education decentralization policies during the1990s. Colombia decentralized primary and secondaryeducation to departments (regional governments) andmunicipalities, and Bolivia is slowly implementing asimilar policy. Guatemala and Honduras have followedthe model of El Salvador’s EDUCO schools. In theregion, only Costa Rica, Ecuador, Panama, and Uru-guay have chosen to retain centralized educationalsystems.

Decision-Making Powers

What does it mean that education has beendecentralized to a particular level? As noted earlier forOECD countries, several educational decisions, suchas choosing textbooks, selecting teaching methods, andresponsibility for implementing school improvementplans, tend to be situated at the school level irrespectiveof the level of decentralization. Others, like setting thecore curriculum or administering and reporting resultson achievement examinations, tend to be located atthe national level irrespective of the level ofdecentralization. Table 2 illustrates the focus of keyeducational decisions in several countries in LatinAmerica.

7

Table 2The Locus of Key Educational Decisions and Responsibilities

Group Decisions Arg Min Ger Chile El Sal Mex NicOrganization Level of decentralization R S L S R S

Choose textbooks S S S S N SDetermine teaching methods S S S S S S

Personnel Hire/fire school director R S L S R SRecruit/hire teachers R R L S R SSet or augment teacher pay R R L N N S

Planning Set performance exams N R N N N NImplement school improvement plan S S S S

Resources Determine expenditures R R N,L N R N,SAllocate personnel budget R R L N R SAllocate non-personnel budget R S L S R S

N = national, R = regional, L = local, S = school.

Decentralization is mainly characterized by thelocus of decisions on personnel and budgets. Thegreatest consistency is found around teacher and schooldirector recruitment and hiring decisions, and thebudgeting of non-personnel expenditures. Thus, inArgentina and Mexico these decisions are situated atthe regional (provincial) level, in Chile at the local(municipal) level, and in El Salvador and Nicaragua atthe school level. Teacher pay decisions are sometimesretained at higher levels of government (as in MinasGerais, El Salvador, and Mexico), and in most casesare heavily influenced by national policy that setsminimum pay conditions (for example, Chile) ornational decisions about education finance (forexample, Minas Gerais).

Of course, simple descriptions of decentralizationfail to capture important nuances. A case in point isthe school improvement plan. Almost every countryin LAC now requires that schools or local jurisdictionsdevelop improvement plans, but as a recent assessmentof the Chilean experience illustrates, such plans areoften carried out as a bureaucratic exercise and fail tomeet minimum standards of quality and communityparticipation. When schools do develop plans, theyoften lack the authority to implement them, as in Co-lombia. And even when they have the authority toimplement, they may have no source of financing.

Another case in point is the allocation of thepersonnel budget. The multiple constraints of nationalor regional pay scales, collective bargaining agreementson working conditions, including class size, andnational curriculum requirements may translate intolittle real discretion at the decentralized level.

Structure and Content

Have decentralization reforms in LAC been mainlystructural in nature—focused on increasing local con-trol and raising accountability—or have they been moreconcerned with content and viewed as a vehicle to raisequality? The answer, of course, is not a simple one.

The education decentralization experiences of Ar-gentina, Chile in the 1980s, El Salvador, and Mexicocan be viewed as mainly structural in nature, but forvery different reasons. In Argentina, primary andsecondary education were devolved to provincialgovernments for mainly fiscal reasons. Hence, the goalof the reform was simply to move expenditureresponsibilities to the provincial governments. Therewas little concern as to whether this would lower orraise quality.

In Chile, the Pinochet government simultaneously

8

introduced a modif ied voucher scheme andmunicipalized public education to increasecompetition between schools for students and therebyraise the accountability of schools to parents. In ElSalvador, the EDUCO model has put the emphasison the creation of school councils to receive andmanage government funds for the purpose ofproviding schooling. While the main objective ofEDUCO has been educational—to improve accessin rural areas—its primary focus has not beeninterventions to alter the content and raise the qualityof schooling. In Mexico, education decentralizationhas been an integral part of a broader decentralizationof powers to state governments in keeping with thepolitical liberalization of the country. Finally,Nicaragua’s policy of school autonomy as the prin-cipal focus has been giving voice to parents and civilsociety on educational issues and, in this way,increasing operational efficiency (Arcia and Belli1998).

In contrast to the cases described above, MinasGerais and Chile (since 1990) have focused onchanging the content of education and raising itsquality through decentralization. Minas Geraisgranted a significant degree of autonomy to the publicschools financed by the state government to definetheir goals, develop a school pedagogical project, andmanage financial resources with the overall goal ofimproving education. Chile since 1990 has attemptedto balance the structural reforms of the 1980s withcontent reforms to raise educational quality,especially for the poor. While the recent reforms havebeen top-down in their design and the goals theypursue, they have attempted to deepen thedecentralization process and move pedagogicdecision-making to the level of the school. Forexample, beginning in 1992, teachers have beenencouraged to work together to develop schoolimprovement projects, which the education ministryfunds on a competitive basis. The Teachers’ Statutewas revised in 1995 to allow school directors tomanage funds directly and to provide school-basedfinancial incentives for performance. Further,beginning in 1997, a competition to fund the besteducation improvement projects proposed bysecondary schools both provides financial incentivesfor performance and gives school directors fullmanagement responsibility for implementing theprojects.

EVAL UATION OF DECENTRALIZ ATION

While the reasons for the decentralization ofeducation in Latin America are often political or fis-cal in nature, from an educational perspective thereis the expectation that decentralization will improveschooling outcomes. Schooling outcomes can bedefined in a variety of ways, but at a minimum involvemeasures of the level and distribution of learning andyears of schooling attained by schoolchildren.

For three reasons it is difficult to use thesemeasures to evaluate education decentralization.First, time series of these measures are seldomavailable. Second, these school outcomes usuallychange slowly in response to any kind of educationalintervention, including decentralization. Third, it isvery difficult to control for external shocks—rangingfrom natural disasters and fiscal crises to teacherstrikes and changes in national educationleadership—that may also influence schooloutcomes.

Given the difficulty of isolating the effects ofdecentralization on learning and educationalattainment, our approach is look at howdecentralization changes factors known to be relatedto learning. First, we ask what is the received wisdomon what are the characteristics of effective or high-performing schools. Second, we ask how thesecharacteristics are reflected in the schoolenvi ronment. And, third, we ask how doesdecentralization directly or indirectly affect any ofthese factors.

High-Performing Schools

There is a growing qualitative and quantitativeresearch literature on the characteristics of high-performing or effective schools (Mohrman andWohlstetter 1994; Creemers 1994; Darling-Hammond 1997) that mirrors the much largerliterature on successful organizations (Barzelay 1992;Lawler 1992). This literature concludes that high-performing schools are characterized by strongleadership, highly qualified and committed staff, afocus on learning, and responsibility for results.Another set of literature reviews the evidence on theprocess by which schools improve, and it yieldsconclusions that are consistent with the effective-

9

schools research. For example, in an evaluation ofschool improvements on three continents, Dalin (withothers 1994) concludes that essential ingredients insuccessful reforms are a sustained commitment toquality improvement, local empowerment to adaptprograms to local conditions, strong emphasis onschool and classroom practice, and strong support

linkage between education authorities and the school“via information, assistance, pressure and rewards”(see Annex Box 4.A.1). In the discussion that follows,we group the variables associated with high-performing schools into four characteristics:leadership, excellent teachers, learning focus, andaccountability.

Table 3Characteristics that Can Be Stimulated through Decentralization

Characteristics of Effective Schools Decentralization Variables that Can Contribute to SpecificCharacteristics of Effective Schools

Leadership School directors are selected by the community using transparentcriteria. School improvement plans are developedlocally. Resources are transferred to schools for the implementationof school plans.

Skilled and committed teachers Schools are given the authority to make curriculum and pedagogicchanges. Teachers have significant responsibility for developingschool improvement plans. Directors are given the authority toprovide a substantive evaluation of teachers’ performance. Schoolsare given the authority (and resources) to make their own decisionsas to the type of training to be provided to teachers.

Focus on learning results The school improvement plan emphasizes goals of improvinglearning (and associated results, such as reducing dropout andrepetition). Information on learning at the level of the school istransparent.

Responsibility for results Directors have fixed-term appointments which may not be renewedif improved learning goals are not met.

Strong leaders have the capacity to effectivelydevelop and communicate a schoolwide andcommunitywide commitment to a common missionand vision for the school, and to manage theimplementation of the school’s improvement plan.The common mission and vision fosters teamworkinside and outside the school, and, most importantly,the process of developing them makes teachers andparents the “owners” of efforts to improve learning.Leadership is especially important in a serviceindustry like education, where the contribution ofindividual teachers is difficult to measure, and thusdifficult to directly reward. In the absence of strongindividual incentives, leaders must motivate teachersto improve. These characteristics can be stimulated

through decentralization. Table 3 summarizes ourfindings.

Decentralization cannot, of course, convert schooldirectors who are used to passively following minis-terial orders into dynamic leaders overnight, but itcan and often does provide a transparent, competitiveselection process for school directors that selects inpart for leaders. A good example of this is the MinasGerais decentralization, which (1) established aprocedure for certifying qualified candidates to com-pete for school director positions, (2) requiredcandidates to present their proposals for schoolimprovements as part of the competition, and (3)empowered school councils to make the final

10

selection of the school director.

Excellent teachers commit to the high goals andstandards of the school, have the strong teaching skillsrequired to meet those goals, continually work toimprove teaching and student learning, and do theirwork in a supportive work environment. Teachercommitment is essential to developing the teamworkrequired for schools to continually diagnose their ownproblems and devise their own solutions. Teamworkis also essential to permit the sharing of teachingexperience required to continually improve teachingpractices. Effective evaluations of teaching perfor-mance is critical to giving teachers information onwhat they need to improve and how to improve it.The time required to participate in the managementof the school and the improvement of teaching isunlikely to be forthcoming in a work environmentwhere teachers are not given time for these activitieswithin their normal work schedule. In many LACcountries, where double and triple shifts are common,it may be logistically challenging to find the spaceand time for teacher participation.

Decentralization can contribute to excellentteaching in a variety of ways. When decisions onsignificant pedagogic matters are transferred toschools, teachers are empowered and motivated towork collectively to improve the services deliveredto students. When school directors are given theauthority to carry out meaningful evaluations ofteaching staff, teachers can focus their training onwhat they need to improve. When resources fortraining and training decisions are given to the school,teachers and directors can purchase the training theyneed (demand-driven) rather than the supply-driventraining provided by the education ministry.

Excellent teaching focuses on student learning.A school system that is focused on learning providesa pedagogy, a curriculum, and resources appropriateto student needs. In most cases, it is the local schooland its teachers who are best placed to diagnose andfind pedagogic solutions to individual student andcollective school learning problems. Different kindsof students—rural, indigenous, poor, urban youth,and so forth—are also likely to have different learningneeds with implications for the distribution offinancial resources to schools by higher levels ofgovernment. Rural children may require smaller class

sizes, reasonable commuting distances, or bustransportation. Indigenous children may require morecostly bilingual instruction. Poor children mayrequire school lunches and subsidized textbooks.

Decentralization can facilitate and reinforce afocus on student learning by providing theinformation required to assess learning problems,devolving appropriate pedagogic decision-making tothe school, and allocating additional resources toschools with special needs. The visible product ofthis process is a solid school improvement plan,constructed with the active participation of teachersand the community, and with real possibilities ofbeing implemented. Good information on studentlearning, and on the value-added of the school, isessential to the diagnosis of learning problems thatis an essential part of the school improvement plan.Good information is also essential to monitoringprogress toward attaining learning goals. Thedevolution of appropriate pedagogic decisions iscritical to the local design of solutions to locallearning problems. Finally, financing is important,both because it is a means of implementing schoolimprovement plans and because it permits theadoption of pedagogy that meets special needs. Inparticular, in the absence of additional resources,children from educationally disadvantaged homes areunlikely to meet the educational goals required forthem to escape their parents’ poverty.

Establishing responsibility for results providesthe incentives necessary for sustained educationalimprovement. A school system with responsibilityfor results requires a set of measurable learning goals,up-to-date information on school performance towardmeeting those goals, rewards for meeting goals andsanctions for not meeting them, and active monitoringof progress. The actor held accountable is typicallythe school director or the staff of the school. The ac-tor holding the school accountable may be theeducation ministry, a school council, or both. In LatinAmerica, the failure by ministries to hold schoolsaccountable is often cited as the rationale for thecreation of elected school councils, which have lo-cal knowledge of the school but often lacksophistication to systematically evaluate performan-ce.

There can be no accountability at the local or

11

school level in the absence of devolution of authorityto make pedagogic and resource-allocation decisionsat the local level. Decentralization can contribute toaccountability at the local level by devolvingdecision-making; establishing performance contractsbetween schools and financing bodies (includingcentral-government ministries and parent-led electedschool councils) that specify learning goals; creatinginformation systems, including standardized tests ofstudents’ knowledge, to permit contract enforcement;and creation of performance-related rewards andsanctions, including dismissal of school directors. Forexample, the decentralization reform in the Chicago,Illinois school system replaced tenure for schooldirectors with four-year contracts and required eachdirector to sign an annual performance contract withthe system specifying measurable goals for the year.Schools that consistently fail to meet goals may seetheir director dismissed and teaching staff reassigned(see Table 4.A.2).

THE CONSEQUENCES OF SCHOOLDECENTRALIZ ATION

In this section, we attempt to evaluate each of theeducation decentralization cases discussed in thispaper in terms of its potential to raise learning,especially among children from poor households. Insome cases, such as Argentina, decentralization wasjust one component of a larger education reform. Inother cases, such as Chile, education reform andchanges in decentralized responsibilities have evolvedover more than a decade. Given the complexities ofevaluating reforms, we do not attempt to separate outthe “decentralization” component for evaluation, nordo we try to evaluate the initial reform. Rather, we tryto make an assessment of the reform as it looks today.

The criteria for this evaluation are thecharacteristics of decentralization that the researchliterature and professional opinion attribute to high-performing schools. Below we give a summaryassessment for each country reviewed in this paper;more complete information on each country’seducation decentralization is given in the Annex.

Table 4Assessment of Education Decentralization

Characteristics of Decentralization Variables

Effective Schools Related to Effective Schools Arg Min Ger Chile El Sal Mex Nic

Leadership Community selects director ✔ N/A ✔

School improvement plans ✔ ✔ ✔

Transfer funds to school ✔ ✔ ✔ ✔

Skilled and committed teachers School curriculum authority ✔ ✔ ✔

Teachers develop improvement plans ✔ ✔ ✔

Directors evaluate teachers ✔ ✔ N/A ✔ ✔

Schools decide training ✔ ✔ ✔

Focus on Learning Learning goals specified ✔

Transparent information ✔ ✔ ✔ ✔

Responsibility for Results Fixed-term appointments for directors ✔ N/A ✔

Competition for students ✔ N/A ✔

Parents have effective voice ✔ ✔ ✔

12

Leadership

The decentralization experiences reviewed here varygreatly in terms of the extent to which they have createdthe conditions that may give rise to strong localleadership. Neither Argentina nor Mexico has givenschool directors any signif icant authority andresponsibility. Chile has recently granted more authorityto directors of municipal schools, and of course thedirectors of the private subsidized schools have longhad a high degree of authority. The EDUCO schools ofEl Salvador are mostly small and often without schooldirectors, and school autonomy is only slowly beinggranted to the traditional public schools. Minas Geraisand Nicaragua are the two examples where schooldirectors have significant authority, and in the MinasGerais, in particular, the open selection processimplicitly values the leadership qualities of candidates.

Teacher Excellence

Strengthening the teaching capacity of teachershas been a high educational priority for mostcountries in Latin America in recent years. Argenti-na has embarked on a major upgrading of its normalschools. Minas Gerais has emphasized the use ofdistance education to upgrade teacher skills. Chilehas provided competitive grants to universities toimprove their teacher training programs and has sentlarge numbers of teachers abroad to strengthen theirteaching skills. Mexico has introduced the CarreraMagisterial to strengthen teacher evaluation and per-formance incentives.

However, few of the region’s efforts to upgradeteaching capacity have been accompanied by in-depthevaluation of teachers, additional compensated timeto participate in school activities and prepare lessons,and incentives for teachers to work and learn inteams—all factors that appear to contribute to schoolimprovement (Dalin et al. 1994). Among thecountries reviewed here, Chile has the policies bestaligned with changing teacher behavior and training.Teamwork among a school’s teachers in Chile isencouraged through (1) competitive funding ofteacher-designed and implemented school-improvement plans, (2) bonuses (equal on averageto one month’s salary) to the 25 percent highest-performing schools as assessed using school perfor-mance indices, and (3) provision of staff time to

participate in professional development circles, withfinancial support from the education ministry.

Focus on Learning

The emphasis on improving quality and raisingstudent achievement is clear in the Argentineeducation reform, the Minas Gerais decentralizationreform, the evolving Chilean reform of the 1990s,and some of the policies and programs carried out inMexico. It is less clear in El Salvador, where theemphasis has been more one of raising access, andNicaragua, where the focus has been more on parentalparticipation than on scholastic achievement.However, even in those countries where nationaleducation reforms and policies are focused on studentlearning, the conditions are not always present foreffectively creating a school-based focus on learning.

Argentina has adopted an ambitious reform totrain teachers, provide sophisticated feedback on in-dividual student performance (at the secondary level),and provide additional financing for children withspecial needs. However, schools, teachers, and localcommunities have almost no authority to diagnosetheir own needs and design their own interventions.Minas Gerais, in contrast, encourages schools todiagnose, monitor, and evaluate; schools are expectedto produce school improvement plans, and the stategovernment provides funding for these plans and fee-dback on student achievement. However, the focusof all this effort is not necessarily specific learninggoals, and teachers and community members are notalways active participants in the process.

As in Argentina, the Mexican education reform hasbeen guided and driven at the national level. Whiledecentralization efforts have not been focused onimproving learning, other components of the reform,including changes in teacher evaluation and pay, andproviding additional resources for poor and indigenousrural children, are focused on learning. However, excludingthe CONAFE schools, teachers and parents are not yetactively engaged in bringing about learning improvementsat the level of the school (Gershberg 1998a).

Chile’s reform efforts since 1990 have beenfocused on student learning, especially for poorchildren. Teachers have been actively involved indiagnosing their own needs and developing their own

14

teacher training decisions. Another key finding of theevaluation is that the degree of decision-makingactually exercised by autonomous schools variesgreatly, and there is a positive and statisticallysignificant relationship between the degree ofdecision-making exercised and student achievement.Furthermore, the strongest positive relationship tolearning was found for variables measuring decision-making on teacher staffing and monitoring of teacheractivities. Nicaragua also illustrates the potential roleof the central government within the context ofdecentralization: A recent qualitative assessment ofNicaragua’s school autonomy discovered thateducators strongly welcome the active interventionof the central government in promoting a pedagogyof active learning (Fuller and Rivarola 1998).

The Minas Gerais reform has not beensystematically evaluated, but the results of theBrazilian national education test put Minas Gerais ator near the top of student achievement in every gradeand subject matter (INEP 1997). The reformsundertaken by the state of Minas Gerais in Brazil havebeen replicated in part by several other states. In par-ticular, several states have now adopted (1) the esta-blishment of school councils, (2) the direct transferof resources to schools, and (3) the local election ofschool directors. Using state-level pooled time-series,cross-sectional data, Paes de Barros and Silva Pintode Mendonça (1998) have analyzed the relationshipbetween these reforms and a number of schoolingoutcomes—gross enrollment rates, repetition rates,age-grade lags, and student achievement as measuredby the Brazilian national educational test, SAEB. Theyfound statistically significant but mixed results. Theestablishment of school councils and the directtransfer of resources are associated with increasedattendance and reduced age-grade lags, but have nostatistically significant relationship to studentachievement. The local election of the school direc-tor, on the other hand, is positively associated withstudent achievement gains, but not with the othermeasures of schooling outcomes.

As noted earlier, Chile has passed through tworeform phases. The first, begun in 1981, emphasizedchanging the structure or organization of educationthrough municipalization and the introduction ofcompetition and choice. A simple comparison ofstudent achievement scores across the 1980s shows

a decline in learning, but during this period real per-student education expenditures also declined, makingit difficult to isolate the reform effect. However, a1998 study by McEwan and Carnoy assembledschool-level panel data to examine how the degreeof competition and choice across municipalities andover time affects public school quality, as measuredby changes in student achievement test scores. Theyconclude that this aspect of Chilean education reformhas had no effect on public school quality. Thisfinding confirms the qualitative evaluations made byother scholars that municipalization did not lead toany substantive changes in behavior and achievementin the public schools (Espinola 1997).

The second phase of the Chilean reform began in1990 and, as noted earlier, simultaneously deepeneddecentralization and set clear goals of raising qualityand equity. In contrast to the 1980s, studentachievement on Chile’s standardized exam, theSIMCE, increased significantly, both in language andmathematics (Cox and Lemaitre 1999). Nationally,the number of correct answers increased by about 18percent. However, here, too, it is difficult to separatethe effects of decentralization reforms, such asintroduction of school improvement projects, fromother reforms (for example, in teacher training), andfrom significantly increased spending over the decade.

The findings for El Salvador, Nicaragua, and Chileare complemented by two careful evaluations carriedout in two large U.S. cities having large populationsof poor and minority students—Chicago andMemphis. As discussed in Box 1, Chicago introducedlargely structural reforms in 1988, and followed upwith a much stronger content-based reform in 1995.A consortium of academic institutions led by theUniversity of Chicago has carefully monitored andevaluated the Chicago reform from Day One. Themost recent evaluation report concludes that year-to-year gains in student learning have risensignificantly (for example, a 19 percent gain inachievement for fifth graders between 1992 and 1996)since the beginning of the reform, despite the factthat the socioeconomic level of students has beengradually decreasing (Bryk, Thum, Easton, andLuppescu 1998). Earlier evaluations demonstrated,also, that school reform efforts resulting fromautonomy are as likely to be initiated in poorer as inricher neighborhoods.

15

Box 1Chicago: An Initial Emphasis on Governance

Chicago has adopted two education reforms. The first, initiated in 1988, focused on governance, while thesecond, adopted beginning 1995, decentralized some powers and put the focus on improving learning. The 1988reform created elected, parent-led school councils with the power to hire and fire the school director. The councilworks with the director to prepare and monitor a school development plan. Tenure for directors was replaced byfour-year contracts. Directors were given increased powers to hire teachers, increased discretion in allocating thebudget, and increased control over curriculum decisions.

By 1995 there was the widespread perception that educational improvements were not occurring rapidly enoughin Chicago. As a result, the mayor took control and named a central district school board and a corporate-stylemanagement team. The board was given the right to impose sanctions on poorly performing schools, includingdisbanding the school council and evaluating and dismissing principals (in conjunction with the councils). One ofits first actions was to put 109 of the 557 public schools in Chicago on probation because of poor academic perfor-mance. The 1995 reform also established a central body responsible for the review and evaluation of the performan-ce of each school, with recommendations for actions to improve performance. Finally, it increased the budgetaryautonomy of each school, including giving each director the freedom to outsource a wide variety of school services.

In contrast to Chicago, the Memphis reform hasbeen heavily content-based from the beginning (seeBox 2). The evaluation of the Memphis school reformconf irmed the Chicago results of sustainedimprovements over time. Prior to implementation ofthe reform, the experimental schools (thosesubsequently undertaking school-based reforms) hadsmaller student gains in learning than a group of con-trol schools. After one year of implementation, the

gains of the experimental and control schools werethe same, and after two years of implementation,student achievement gains in the experimentalschools were significantly higher than in the con-trol schools (Ross, Sanders, Wright, andStringfield 1998). Finally, an evaluation of theMemphis decentralization conf irmed thatleadership by school directors and teacher buy-into reforms are critical to their implementation.

Box 2Memphis: Decentralization Focused on Improving Learning

The schools of Memphis, Tennessee, serve a largely poor and educationally disadvantaged population. Frustratedwith the persistently poor academic performance by students, the city decided in 1995 to grant limited autonomy toindividual schools with the objective of stimulating school-level educational reforms. Each school formed an advisoryschool council comprising the director, teachers, parents, and community members. The principal function of eachcouncil is a technical one—diagnosing needs, agreeing on reforms, and monitoring progress in student learning—and while it is legally advisory in nature, its opinions are taken seriously.

Each school in the Memphis district was required to adopt a school-based reform from a menu of eight differentschool restructuring models. While the pedagogic orientation of the models differ, they share several characteristics:increased school autonomy (especially, on pedagogic matters); a common vision of school goals reflected in theschool development plan; performance contracts with specific, quantifiable targets between the school director andthe central administration; extensive teacher development activities at the school level; teamwork within the school;and constant monitoring of progress, including the use of standardized examinations.

The central Memphis education office continued to play a strong role in setting high standards (for example, allstudents in grades 3 through 8 must pass set exams in mathematics and science in order to be promoted); mandatingminimum standards and core curriculums; facilitating teacher development by offering a broad menu of trainingoptions and opportunities; providing additional financing to cover the costs of implementing school developmentplans (with larger amounts for schools serving the poor); and establishing monitoring and evaluation systems toprovide constant feedback to individual schools on their performance.

16

Taken together, the El Salvador, Nicaragua, Chi-le, Chicago, and Memphis evaluations provide strongevidence that educational decentralization canimprove learning. What is notable is that those casesdemonstrating the largest positive gains haveemphasized school autonomy with pedagogic reform,especially true in Chicago since 1995, in Memphis,and since 1990 in Chile.

Summary

Education decentralization is a worldwidephenomenon, and Latin America is no exception.While there are economic and education argumentsfor decentralization, the particular forms ofdecentralization in most Latin American countrieshave been driven more by politics. Given themagnitude of education decentralization efforts in theRegion over the past decade and the forms they havetaken, it is timely to assess their effects.

The evaluation of decentralization reforms isdiff icult due to (1) lack of baseline data, (2)incomplete implementation of many reform elements,and (3) lags between implementation and the changesin such factors as behavior and resource allocation,which affect learning. The difficulty in evaluatingreforms argues for caution in interpreting results. Thelack of much rigorous evaluation of Latin Americanexperiences has led us to rely to some extent on goodevaluations of decentralization efforts outside theregion for our overall conclusions.

The fact that few evaluations exist of the impactof decentralization on learning outcomes has also ledus to an alternative approach to infer impacts bylooking at the extent to which characteristics ofdecentralization reforms are consistent with thecharacteristics associated with high-performingschools. The fact that two well-evaluated andsuccessful U.S. school reforms—in Chicago andMemphis—have shared the decentralizationcharacteristics professional educators associate withpublic schools lends credence to this approach.Interestingly, many of the recommendations madeby educators for creating effective schools areconsistent with the prescriptions economists mightmake.

Designing decentralization reforms to improve

learning is complicated by the nature of education.For example, it is difficult for any actor external tothe school to monitor and hold the school’s perfor-mance accountable. After all, the outputs of the schoolare several, and almost all are difficult to measure.Experience has shown it is especially difficult tomeasure the value-added of the school in producingscholastic achievement (Ladd 1996). In addition,when teachers work in isolation they have thecapacity to shirk their duties, with little risk ofnegative consequences. Finally, strong labor unionsand regulatory protection (often embodied in teacherstatutes in Latin America) make it difficult to penalizepoor-performing teachers even when they can beidentified.

To economists, these agency problems argue fora number of solutions. First, intense efforts shouldbe made to provide good information on the perfor-mance of schools and teachers, taking into accountthe complexity of the educational production process.This may require establishing an independent agencyto carry out external audits of schools that go beyondmerely identifying outputs, and provide diagnosesof problems and propose solutions as well. Second,school directors should be given a large degree ofauthority; they have considerably better capacity tomonitor school and teacher behavior than do localpolitical agencies, including school councils. Third,teaching should be organized in a way that minimizesshirking and provides peer rewards and sanctions forperformance. This requires that teachers shareexperiences and work together as much as possible.Fourth, given the high risk of shirking, teachers mustthemselves become the proponents and owners ofefforts to improve teaching, including deciding ontheir own training. Externally imposed (that is, top-down) solutions to educational problems are likelyto fail in the absence of an effective communicationscampaign to enlist the support of teachers.

Of the Latin American reforms reviewed here,two—those in Chile and Minas Gerais, Brazil—entaila large number of the elements that arguably giverise to the characteristics of effective schools. Neitherreform has yet been subjected to rigorous evaluation,although the available evidence for Chile is positive.Two other Latin American reforms—more limitedin scope than Chile and Minas Gerais—have beenevaluated in terms of impact, with somewhat

17

contradictory results. El Salvador’s EDUCOprogram has not yet demonstrated positive effectson learning, while Nicaragua’s charter schoolprogram has. Nicaragua’s reform granted substantialauthority to school directors, which Brazilianresearch has found to be associated with learninggains.

In sum, there is growing evidence that at leastsome of the characteristics of educationdecentralization reforms that focus on schoolautonomy, as opposed to municipal or regionalautonomy, contribute to higher-performing schools.Decentralization to subregional governments mayalso yield some educational benefits by allowinggreater innovation and greater flexibility to adaptresource allocation to local prices, but they have notyet been proven.

Bibliography

Aedo, Cristián (1998). “Difference in Schoolsand Student Performance in Chile.” I n William D.Savedoff, ed., Organization Matters: AgencyProblems in Health and Education. Washington,D.C.: Inter-American Development Bank.

Arcia, Gustavo, and Humberto Belli (1998).“Rebuilding the Social Contract: School Autonomyin Nicaragua.” LCHSD Paper Series No. 40.Department of Human Development, Latin Americaand the Caribbean Region. Washington, D.C.: WorldBank.

Barzelay, Michael (1992). Breaking ThroughBureaucracy: A New Way for Managing inGovernment. Berkeley, CA: University of CaliforniaPress.

Bryk, Anthony; David Kerbow; and SharonRollow (1997). “Chicago School Reform.” In D.Ravitch and J. Viteritti, eds., New Schools for a NewCentury: The Redesign of Urban Education. NewHaven: Yale University Press, 164–200.

Bryk, A. S.; Y. M. Thum; J. Q. Easton; and S.Luppescu (1997). “Assessing School ProductivityUsing Student Achievement: The Chicago PublicElementary Schools,” mimeo. Consortium on

Chicago School Research.

Bryk, A. S.; Y. M. Thum; J. Q. Easton; and S.Luppescu (1998). “Examining Productivity: Ten-YearTrends in the Chicago Public Schools.” Chicago:Consortium on Chicago School Research, March.

Catholic University (Pontífica Universidad Cató-lica de Chile) (1998). Informe Final: Evaluación dela Línea de Proyectos de Mejoramiento Educativo(PME). Convenio Ministerio de Educación—Facul-tad de Educación. Santiago, Chile, August.

Cox, Cristián, and María José Lemaitre (1999).“Mar ket and State Principles of Reform in ChileanEducation: Policies and Results.” In Danny Leiptzigerand Guillermy Perry, eds., Chile: Recent PolicyLessons and Emerging Challenges. WDIDevelopment Studies Series. Washington, D.C.: WorldBank.

Creemers, Bert P. M. (1994). The EffectiveClassroom. New York: Cassell.

Creemers, Bert P. M.; Tony Peters; and DaveReynolds, eds. (1989). School Effectiveness andSchool Improvement: Proceedings of the SecondInternational Congress. Rotterdam: InternationalCongress for School Effectiveness 1989.

Dalin, Per, with Tekle Ayono, Anbesu Biazen,Birhanu Dibaba, Mumtaz Jahan, Matthew B. Miles,and Carlos Rojas (1994). How Schools Improve: AnInternational Report. New York: Cassell.

Darling-Hammond, Linda (1997). The Right toLearn: A Blueprint for Creating Schools that Work.San Francisco: Jossey-Bass Publishers.

di Gropello, Emanuela (1998). “Descentralizaciónde la Educación y la Salud: Un Analisis Comparativode la Experiencia Latino Americano.” Chile: CEPAL,Document No. LC/L1132.

Dillinger, William (1995). Better Urban Services:Finding the Right Incentives. Washington, D.C.: WorldBank.

Espinola, Viola (1997). “Descentralización del sis-tema Educativo en Chile: Impacto en la gestion de

18

las escuelas.” Informe No. 10., May. Department ofHuman Development, Latin America and theCaribbean Region. Washington, D.C.: World Bank.

Fiske, Edward B. (1996). “Decentralization ofEducation—Politics and Consensus.” Directions inDevelopment Series. Washington, D.C.: World Bank.

Fuller, Bruce, and Magdalena Rivarola (1998).“Nicaragua’s Experiment to Decentralize Schools:Views of Parents, Teachers, and Directors.”Development Economics Research Group, WorkingPaper Series on Impact Evaluation of EducationReforms, No. 5, February. Washington, D.C.: WorldBank.

Gaynor, Cathy (1998). “Decentralization ofEducation—Teacher Management.” Directions inDevelopment Series. Washington, D.C.: World Bank.

Gershberg, Alec Ian (1995). “FiscalDecentralization and Intergovernmental Relations:An Analysis of Federal Versus State EducationFinance in Mexico.” Review of Urban and RegionalDevelopment Studies 7 (2): 199–142.

Gershberg, Alec Ian (1998a). “Decentralizationand Re-centralization: Lessons from the SocialSectors in Mexico and Nicaragua.” Office of the ChiefEconomist, Working Paper No. 379. Washington,D.C.: Inter-American Development Bank.

Gershberg, Alec Ian (1998b). “Decentralization,Re-centralization, and Performance Accountability:Building an Operationally Useful Framework forAnalysis.” Development Policy Review 16 (4) :405–31.

Gershberg, Alec Ian (1999a). “Decentralization,Citizen Participation, and the Role of the State: TheAutonomous Schools Program in Nicaragua.” LatinAmerican Perspectives, 26(4):8-38.

Gershberg, Alec Ian (1999b). “Fostering EffectiveParental Participation in Education: Lessons from aComparison of Reform Processes in Nicaragua andMexico. World Development 27 (4).

Guedes, A., A. L. Amaral, R. Walker, and T. Lobo(1995). “Brazil—Decentralized Management of

Education in Minas Gerais.” Education Sector, July.Washington, D.C.: World Bank.

Hanson, E. Mark (1997). “EducationalDecentralization: Issues and Challenges.” OccasionalPaper No. 9. Chile: Partnership for EducationalRevitalization in the Americas.

Hanushek, Eric A. (1994). Making Schools Work:Improving Performance and Controlling Costs. Was-hington, D.C.: Brookings Institution.

Haslam, M. Bruce (1997). “How to Rebuild aLocal Professional Development Infrastructure.”Arlington, VA: New American Schools. GettingBetter by Design Series No. 4.

Henevald, Ward, and Helen Craig (1996). SchoolsCount: World Bank Project Designs and the Qualityof Primary Education in Sub-Saharan Africa. Was-hington, D.C.: World Bank.

IDB (Inter-American Development Bank) (1994).“Special Report: Fiscal Decentralization.” InEconomic and Social Progress in Latin America:1994 Report. Washington, D.C.

INEP (Instituto Nacional de Estudios e Pesqui-sas Educacionais) (1997). Sistema Nacional deAvaliação da Educação Básica (SAEB): PrimeirosResultados. Brasilia: INEP.

Javed Burki, Shahid, and Guillermo E. Perry(1998). Beyond the Washington Consensus:Institutions Matter. Washington, D.C.: World Bank.

Jimenez, Emmanuel, and Yasuyuki Sawada(1998). “Do Community-Managed Schools Work?An Evaluation of El Salvador’s EDUCO Program.”Development Research Group, Working Paper Se-ries on Impact Evaluation of Education Reforms, No.8, February. Washington, D.C.: World Bank.

Kellor, Eileen M., and Allan Odden (1999).“Cincinnati: A Case Study Design of a School-BasedPerformance Award Program.” Paper presented to theAmerican Education Finance Association, Seattle,WA, March 18–21.

King, Elizabeth; Laura Rawlings; Berk Özler;

19

Patricia Callejas; Nora Gordon; and Nora Mayorga(1996). “Nicaragua’s School Autonomy Reform: AFirst Look.” Working Paper Series on ImpactEvaluation of Education Reforms, No. 1, October 15.Washington, D.C.: World Bank.

King, Elizabeth, and Berk Özler (1998). “What’sDecentralization Got To Do with Learning? The Caseof Nicaragua’s School Autonomy Reform.”Development Economics Research Group, WorkingPaper Series on Impact Evaluation of EducationReforms, No. 9, June. Washington, D.C.: WorldBank.

Laboratorio Latinoamericano de Evaluación dela Calidad de la Educación (1999). “Primer EstudioInternacional Comparativo sobre Lenguaje, Matemá-tica y Factores Asociados en Tercero y Cuarto Gra-do.” Santiago, Chile: UNESCO/OREALC.

Ladd, Helen, ed. (1996). Holding SchoolsAccountable: Performance-Based Reform inEducation. Washington, D.C.: Brookings Institution.

Lawler, Edward E. (1992). The UltimateAdvantage: Creating the High InvolvementOrganization. San Francisco: Jossey-Bass.

McEwan, Patrick, and Martin Carnoy (1998).“Public Investments or Private Schools? AReconstruction of Educational Improvements in Chi-le.” Processed. Washington, D.C.: World Bank.

Mohrman, Susan Anlbers, and PriscillaWohlstetter, eds. (1994). School-Based Management:Organizing for High Performance. San Francisco:Jossey-Bass.

Navarro, Juan Carlos, and Rafael de la Cruz(1998). “Federal, State and Nonprofit Schools inVenezuela.” I n William D. Savedoff, ed.,Organization Matters: Agency Problems in Healthand Education. Washington, D.C.: Inter-AmericanDevelopment Bank.

OECD (1998). Education at a Glance – OECDIndicators. Paris: Organization for EconomicCooperation and Development.

Paes de Barros, Ricardo, and Rosane Silva Pintode Mendonça (1998). “The Impact of Three

Institutional Innovations in Brazilian Education.” InWilliam D. Savedoff, ed., Organization Matters:Agency Problems in Health and Education. Washing-ton, D.C.: Inter-American Development Bank.

PREAL (Partnership for EducationalRevitalization in the Americas) (1998). The Futureat Stake: Report of the Task Force on Education,Equity, and Economic Competitiveness in LatinAmerica and the Caribbean.” Chile.

PREAL/CINDE (Programa de Promoción de laReforma Educativa en América Latina y el Caribe/Corporación de Investigaciones para el Desarrollo)(1999). “Laboratorio Latinoamericano difunde estu-dio comparativo de lenguaje y matemáticas.” PREALInforma No. 2 (February).

Ross, Steven M.; Marty Alberg; and John Nunnery(1998). “Selection and Evaluation of LocallyDeveloped Versus Externally Developed SchoolwideProjects.” Mimeo. Memphis, TN: Center for Researchin Educational Policy, The University of Memphis,September.

Ross, Steven M.; William L. Sanders; S. PaulWright; and Samuel Stringfield (1998). “TheMemphis Restructuring Initiative: AchievementResults for Years 1 and 2 on the Tennessee Value-Added Assessment System (TVAAS).” Memphis,TN: Center for Research in Educational Policy, TheUniversity of Memphis, June.

Savedoff, William, ed. (1999). OrganizationMatters: Agency Problems in Health and Education.Washington, DC: Inter-American DevelopmentBank.

Schmelkes, Sylvia (1996). The Quality of PrimaryEducation: A Case Study of Puebla, Mexico. Paris:UNESCO, International Institute for EducationalPlanning.

Shipps, Dorothy; Joe Kahne; and Mark Smylie(1997). “Been Down So Long, It Looks Like Up toMe: Decentralization in the Chicago Public Schools,1988–1997.” Paper presented to the annual meetingof The University Council for EducationalAdministration, Orlando, FL, October 31–November2.

20

Slavin, Robert E. (1996). “Salas de Clase Efecti-vas, Escuelas Efectivas.” Occasional Paper No. 3.Chile: Partnership for Educational Revitalization inthe Americas.

Summers, Anita A., and Amy W. Johnson (1991).“A Review of the Evidence on the Effects of SchoolBased Management Plans.” Paper for the Panel onthe Economics of Educational Reform, March 29.Washington, D.C.: The Brookings Institution.

Summit of the Americas II (1998). Plan of Action.Santiago, Chile, April 1–3.

Winkler, Donald R. (1994). “The Design andAdministration of Intergovernmental Transfers: Fis-cal Decentralization in Latin America.” DiscussionPaper, No. 235, July. Washington, D.C.: World Bank.

Wong, Kenneth K.; Robert Dreeben; and LaurenceE. Lynn Jr. (1997). “Integrated Governance as aReform Strategy in the Chicago Public Schools.”Office of Educational Research and Improvement,Department of Education, Laboratory for StudentSuccess Publication Series, Report No. L97–10.

World Bank (1997). “El Salvador’s Community-Managed Schools (EDUCO) Program: A FirstReport.” Development Research Group, WorkingPaper Series on Impact Evaluation of EducationReforms, No. 4, July. Washington, D.C.

World Bank (1994). “El Salvador—CommunityEducation Strategy: Decentralized SchoolManagement.” Education Sector, Final Sector Report,No. 13502, December. Washington, D.C.

Xavier, Antonio Carlos da Ressurreição, JoséAmaral Sobrinho e Fátima Marra (Orgs.) (1994).Gestão escolar: desafios e tendências. INSTITUTODE PESQUISA ECONOMICA APLICADA. IPEA.Livros. Série IPEA, 145. Brasília: 375p.

Alexandra BlackhurstC:\CDRIVE\valdivia\Winkler.doc02/07/00 2:40 PM

Ve·A·CeVe·A·CeVe·A·CeVe·A·CeVe·A·CePhone: (56-2)371 3350 - Fax: (56-2)363 9598

E-mail: info@veace.clSantiago - Chile