Does Local Financing Make Primary Schools More Efficient?

Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized

Policy,Planning, arid Rmarwh

WORKING'

PAPERS

Eduation and Employment

Population andHumanResources Department TheWorldBank August1988 WPS 69

Does Local Financing Make Primary Schools More Efficient? The Philippine Case EmmanuelJimenez, Vicente Paqueo, and Ma. Lourdesde Vera

Philippine schools that rely more heavily on local sources of income are irore cost-effective than those that are more dependent on central funding.

the findings of wodk in progress and to distnbutcs PPR Wo*kingPapers to diannte ThePdicy. PiantindRgsRchCanhplex aicounge the exchange of ideas among Bank staff and alUothes intesed in development issues. These papers caty de nam of theauthsn, refletaoldy their views, and should beued and cited accordingly.The fundings.interprtations, and cmtclusins are the World Bank, its Boardof Directos, itsmanagruecnt, orany ofitsmenbercautries. authdseown. They should not beattributothe

Po,Pnlenrtk, *nd Rewch

Educnion and Enpboyment

In the Ehighly centralized system of the Philippines, local funding provides the only source of flexibility to meet specific and urgent needs.

forced to consider the effect of their behavior on the people who live and work in the local community.

The government in Manila, which pays all teacher salaries, finds it easier politicaUy in times of fiscal belt-tightening to cut recurrent costs. Although local funds are relatively small percentage of the education budget, they make an important contribution to covering maintenance and operating costs. For example, the quality of both textbooks and school buildings appears to irncreasewith the level of local funding.

The policy implications of these findings for the Philippines are important. They strongly suggest that decentralization will increase efficiency. Without an increase in local funding, the quality of primary education will suffer. Other developing nations, facing similar situations, might also consider more community funding for school systems.

The total cost of education per student also appears to lower in schools with greater local financing, regardless of the perceived quality of the school. Administrators and teachers have greater incentive to be cost-effective when

This paper is a product of the Education and Employment Division, Population and Human Resources Department Copies are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Teresa Hawkins, room S6-224, extension 33678.

The PPRWorkingPaperSeriesdisseminawes the findngs of workunderway in theBank'sPolicy,Planing, and Research Complex.An objectiveof the seriesis to get thesefmdingsout quickly,evenif presentationsare less than fullypolished. The findings,intepretations,and conclusionsin thesepapersdo not necessarilyrepresentofficialpolicyof the Bank. Copyright0 1988by the Intnational Bankfor ReconstructionandDevelopmentlTeWorldBank

Does Local Financing Make Primary Schools More Efficent, The Philippine Case by Emmanuel Jimenz, Vicente Paqueo, and Ma. Lourdes de Vera Table of Contents Introduction ...................................................... 2 Conceptualand EmpiricalSeting. ........................ 3........................ 3 Local Funding and School Efficiency. .................................................. 6 Data and Some SinPle Comparisons........................ .8 DescTiptiveStadstics..................................................... ... 9 Empirical Results ...................................................... 12 Are Schoolswith Proportionately More Local Funding More Efficient?.. 12 Why are Schoolswith Relatively More Local Fmancing More Efficient?.. 15 Condusions .. ...... 16 References......................................................................... 19

INTRODUCTION

In many developingcountries,the central governmentprovides the bulk of financingfor public education. Given tighteningbudgetary restraintson the central government'spurse, alternativesources of financinghave to be considered. One option is to rely, to a greater extent, on support from individualusers or through contributionsfrom the local community (World Bank 1986). This policy would presumablyenhance what has been called 'externalefficiency"in the educationeconomics literatiwre,because it would raise additional resources for education.

The additionof locallymobilized resourceswould also affect allocationwithin the school system. In a highly centralizedsystem,money from local sources that are under the control of the school often provide its only source of funding specificneeds, particularlythose non-salary items affectedby acts in national appropriations. Furthermore,schools that are financiallyaccountableto the communitiesthey serve may be more responsiveto their clients -- i.e., the studentsand their parents. In order to survive,schoolswhose existencedepends on local fundingwould provide the type and quality of schooling the communitydesires. In addition, administratorsand teacherswould have a greater incentiveto be cost-effective,since they must considerthe effect of their behavioron the financialcontributionsof people who live and work In the immediate vicinity of the school. Although these argumentsare theoreticallycompelling,there has been relativelylittle empirical researchon these issues in developing,or

-2-

even, as far as we can tell, developedcountries. Are public schools that rely on local resourcesfor a greater share of their financingmore cost effective? In policy terms, would financialdecentralizationlead to more erficientschools? This paper attemptsto answer these questionsusing cost, financial sources and student achievementdata from Philippineprimary schools. The first section,which follows immediatelybelow, describesthe Philippine setting and how a school'sdegree of financialdecentralizationcan affect efficiency. The degree of decentralizationis measuredby the proportionof a school'stotal expendituresfinancedby local contributions,such as contributionsfrom the local school board, municipal government, parent-teachersassociations(PTA) and other sources. The s--,

; section

describesthe sample and basic characteristicsof the differe..

pea of

schools. The third sectionpresents the results of estimatedcost functions which determinewhether this proportionis correlatedwith school expenditures,holding constantfor student size and school quality.A brief concludingsection then follows.

CONCEPTUALAND EMPIRICALSETTING

The Philippinepublic elementaryschool system,which accountsfor 95 percent of enrollmentat that level, is highly centralizedwith several layers of bureaucracy. Administratively,schools are organized into districts,which are in turn grouped into divisions and then regions. Schools are headed by principalswho are directlysupervisedby district supervisors. And the divisionsuperintendentin turn overseesthe district supervisors. Almost all of the importantadministrativeissues are decided -3 -

at the regionaland the Departmentof Education,Culture and Sports (DECS) central office -- e.g. appointments,determinationof salary scales, and promotions. Administrativematters,however, are initiatedat the sub-regionallevel, such as request for teachersby principals,processing of teacher applicationsand drawingup of recomme"ndations for promotionsof supervisors,principals,teachersand support staff. The bulk of financialsupport of the public elementaryeducation, which is provided free and accountsfor 70 percent of the DECS budget, comes from natio.al appropriations. The systems of budgeting,resourceallocation and procurementare highly centralized. The responsibilities for budget control and executionare limited to the upper administrativelevels and stop at the division office. All administrativelevels, however, go through the motion of budget planning. To illustratethe extent of centralization, it is interestingto note that it is at the regionalor divisionoffice where actual procurementof the maintenanceand operatingexpense items of the schools are undertaken. Schools receive those items in kind. Consequently,they have very little flexibility. The degree of centralizationis similar for personnelservices and capital outlay. The availabilityof local funds, though small comparedto the financialsupport from the nationalgovernment,provides the only source of flexibilityand opportunityat lower administrativelevels for matching resourcesand urgent school needs. The main source of local funds is the Special Education Fund (SEF) created under RepublicAct No. 5447. The SEF is financed from its share of the additionaltax on real property collected by the municipalityor city and a certain portion of taxes on a certain type of cigarettesand duties on importedtobacco. The SEF share of the revenuesfrom the additionalreal property tax collected by the municipalityis distributedas follows:50 percent to -4 -

be retainedby the municipality,20 percent goes to the provincialtreasurer and the rest is remittedto the national treasurerto finance expenditures exclusivelyfor stabilizingthe SEP in the municipalities,cities and provinces. In the case of cities, 60 percent are retainedand the rest are transmittedto the nationaltreasurer. Local school boards for every city, municipalityand provincewere created to administerthe SEF under RA 5447 and decide on its utilization. They are constitutedto representthe school administration,the local governmentand the parents. The municipal (city) board is headed by the districtsupervisor (city superintendent)and includesas members the meyor, a representativeof the municipal (city),and the president or the duly-electedrepresentativeof the League of Parent-TeachersAssociations. It appears then that RA 5447 provides the mechanism whereby schools can get additionalfunds and the ability to respondwith some flexibilityand-timelinessto emergingspecificlocal needs and problems. It also creates a greater sense of accountabilityamong school administratorsand sensitivityto parentalconcerns. Another source of funds for schools is the local government appropriationfund, which consistsof all the revenuesof the municipalor city governmentfrom taxes and other income generatingsources available for its own use. In this regard the superintendentmust submit requestsfor funding to the mayor. Grants from this fund require the mayor to monitor its receipt and use. Finally,there are funds from the Parents-Teachers Association (PTA) and other private contributions. These include donations and revenues from income generatingactivitiessponsoredby the PTA. Parents have the lead role in the PTA and the only ones qualifiedto become its officers. Parents,however, generallyelicit suggestonsfrom the teachers and principals. It is interestingto note that the PTA treasurer - 5 -

and th- principalmaintain a joint bank account in their names rather than the school's. The motivationof this arrangementis to avoid bureaucratic red tape. The magnitudes from all of these sources together are far from trivial. According tc.the gov r ent's 1983 Householdand Matching Survey (HSMS),two-thirdsof public pri sry schools had some sort of extra-budgetaryfunding from local sources.Some 19 percent of schools relied on local contributionsfor 5-25 percent of their expenditure. Another 7 percent of them had local funding amountingto more than 25 percent of their budget. The criticalquestion is whether this source of funding can affect school efficiency.

Local funding and school efficiency: One reason why the ability of local school authoritiesto selffinance can improve school efficiencyis that it can lead to a more appropriate input-mix-- i.e., one consistentwith the relativeeconomiccosts of labor and non-labor inputs. Since teachers are paid directlyby the central educationdepartmentand are assignedto schools,schoolswhich have no other funding sources are consignedto live with the input-mixesmandatedby the central government. In recent years, the pressureson the public budget have forced the central authoritiesto change this mix. For politicalreasons, it is simply easier to make real cuts in the non-salariedrather than the salariedbudget. (This "recurrentcost' problem,while endemic to all sectors, is particularlyacute in education,6 sector in which a larger share of the budget pays for personnelcosts --

see Heller, 1979.) Locally

generated funds, while relativelysmall comparedto all school expenditures, can make an importantcontributionto maintenanceand non-teacheroperational expenditures. Across all public schools (whetherthey receive local - 6 -

contributionsor not), the average amount of local contributionsis enough to cover 5% of all education expenditures-- but over 40% of average

non-

personnel expenditures. These figures imply that local financingcan improve efficiencyby a more appropriatebalance between personnel and nonpersonnel recurrentcosts. A second reason why decentralizedfunding can improvemanagerial efficiencyis through the implicitincentivestructure. As stated in an earlier document: In a centralizedsystem, administrators, studentsand parentsplay only a marginal role in determining--indirectly through t,heirchoice of schools -- how school resourcesare to be allocated. Typically,school administratorsare accountablenot to parents and studentsbut to central authorities,such as ministriesof education. Since the costs of monitoring,inspectingand enforcing detailed guidelinesfor individualschools are likely to be high, these ministriesset norms, such as for the distributions of budgetary allocationsbetween teachers'salaries and other inputs. If norms do not match the school's needs or the community'spreferences,as is often the case, school administratorshave neither the financialpower nor the incentiveto change them. As a result, the use of school resources is inefficient.(World Bank, 1986, p. 11) In the Philippines,the local contributionto the financingof centralizedpublic schools is done through formal as well as informalassociations. About 601 of all public schools receive some funds from PTA or other sources. However,the average amount receivedfrom these sources is relativelysmall (about 5,000 pesos per annum, per school or 101 of average expenditureson maintenanceand operatingexpenses). Although a smaller percentageof schools receivesaid from local governmentsor school boards, the average contributionsfrom these sources is large -- an average of about 25,000 pesos per annum, the equivalentof about one-half of average expenditureson maintenanceand operatingexpenses. The goal of this paper is to determinewhether these reasons, which are plausible in the Philippinecase, can lead to increasedefficiency - 7 -

in primary

schools. We do so by estimatingcost functionsfor schools.

This methodologyenables us to hold constant for other factorswhich could significantlyaffect school costa, such as enrollmentand input prices.

ln

addition,we assume a multi-productframeworkto enable us to hold constant for the other dimensionof school output -- quality,as measured by average achievementscores. We use a rich data set which containsboth cost and school quality data from a nationwidesample of Philippineprimary schools. of the cost functionsare These data, as well as the exact specificationi describedmore fully in the next sections.

DATA AND SOMESIMPLE COMPARISONS

The sample schools in our analysiswere obtained from the ongoing Household School Matching Survey Project (HSMS),conductedby the EDPITAF team of the then Ministry of Education. The data for these schoolswere obtained from a stratifiednatLonalsample of barangays (villages).These data are part of a broader eftort to collect integratedbaseline information for policy analysisand the !mpact evaluationof the Program for Decentralized Educationand Development(PRODED). Aside from school level data, the HSMS data base includesinfomnationat the level the of community,household and individuals. The HSMSdata were collected December 1983, from 260 barangays

nationwide

between May 1982 and

and 4990 households.

These were chosen on

the basis of a two-stage stratifiedrandom samplingscheme. The identificationof the barangay samples involvedthe selectionof 20 barangays in each region acrose urban and rural and affectedand unaffected strata. ('Affected"samles are those belongingto school districts found -8-

by PRODED to be below the nationalcut-off in these indicators:achievement scores, school part cipationrate and survival rate). rhe villageswere randomlydrawn with probabilityproportionalto size, defined as the number of householdsin the barangay. Equal allocationwas used, i.e., five barangayswere drawn across each stratum in every region. From the sample barangayswero drawn the sample households. The samplingframe was a listing of all householdsin the sample communities with at least one member less than 25 years old. From the listing of members in each sample householdwas made a group of elementaryschool childre and their schoolswere identified. Data on those schools w(-collectv4 including (among others) informationon finance,costs, inputs and other schcol characteristis. A random sample of 20 grade four pupils were given school achievementtests. Their average score, which can be used as a summary indicatorof school quality,was computed for each of the 586 sample elementaryschools.

DescriptiveStatistics Table 1 below shows the mean characteristicsof the three categoriesof schools,by degree of financialindependencefrom national appropriations,as measuredby the proportionof school expendituresfrom local sources. Schools that are financiallyless dependenton DECS appear to have an advantageover the more dependentschools in terms of performance in achievement that

scores.

are more urbanized WJth respect

schools relative

is completely financial

The former

also

and character

'ed

to achievement

as their

in communities

wages.

performance,

This is true

-9-

to be situated

by higher

test

in the same order

independence.

tend

degree for all

the rank order of of the subject

areas

Table lt

School Characteristics

School Characteristic Achievementscore: Overall Math English Pilipino

by Degree of Local Financing

Proportionof ExpendituresFrom Local Contribution GT 25 2 5-25 2 LT 5 2

42.26 39.68 40.93 46.19

47.60 45.00 45.29 52.52

48.25 44.10 46.13 54.51

721.57 42.56 13.39

1214.81 46.76 15.07

821.34 46.48 17.10

0.54 0.51

0.36 0.44

0.28 0.52

Labor inputs: N of teachers X of teacher with MS Teaching experience Training Supervisorvisits Principalexperience

24.62 0.47 15.56 0.54 4.53 5.32

38.74 0.34 16.30 0.52 6.67 6.85

27.76 0.38 13.52 0.64 5.65 10.80

Non-labor inputs: Desk per classroom Studentsper class Textbookper student Floor area Quality of floor N of teach. manuals workbooks

14.46 36.15 3.46 1334.60 58.84 69.83 38.62

18.62 36.56 3.66 1547.17 62.96 92.72 32.94

16.60 35.19 4.67 1125.64 64.97 95.88 26.92

Other (process)inputs: PTA meetings

5.27

5.02

5.40

469.37 41.80

294.18 21.27

226.37 34.91

34.96 459.48 675.42

16.47 287.12 318.67

29.89 255.07 265.46

66 0.19

25 0.07

Student inputst School size Avg. mental ability (MAT) Wage of Unskld adult (backgroundmeasure) Rural school Affected Barangay

Cost per student: Personnel Maint. and oper. Maint. and oper. less textbooks Total less text Total Number of schools Proportionof schools

254* 0.74

*143 of these schools (or 562) have F of less than 1%.

-

10

-

except in math where tne average score of schoolswith the local share of funding at 5-251 rank*d

slightlyhigher than thoso schoolswith a share

greater than 251, though the average score in the moet DECS dependent schools (less than 51 share) remainedthe lowest. The same pattern of .-elationshipwith degree of financialdecentralization was observed for their mental abilitytest (MAT) scores. Average recurrentcost rer student,on the other hand, increases with degree of dependenceon nationalappropriations. The unit recurrent cost

is 675 pesos for schools with less than 5 percent of its expenditures

accounted share pattern.

for by local

is 25 percent In contrast,

between maintenance

sources

or more. there

compared to 2;5 pesos Personnel

cost

for schools

per student

follows

whose local a similar

does not seem to be a monotonic relationship

and operating expenditureper student and degroe of

relianceon local funds. In terms of the quality of labor used, it is interestingthat while schoolswith relativelymore local funding tend to have less experiencedteachers, they have in contrastmore experiencedprincipals. On the other hand, the relationshipis not clear for tx-aining, supervisor visits, and percentageof teacherswith griduate level courses. Regardingnon-labor inputs,textbookper student and quality of school buildings (as measured by the percentageof the floor area in good condition)appear to be increasingwith degree of local finance. But, there is no clear monotonic relationshipbetween local finance and desks per classroom,class sizo, floor area per student,teachingmanual per teachers and workbooks per student.

-

11

-

Finally,the simple cross tabulationsdo not indicate that frequencyof PTA meetings, a measure of parental involvement,is correlated with the extent to which schools are funded by local sources.

EMPIRICALRESULTS

The empiricalportion of this paper seeks to address two questions:whether schools that rely less on central funding sources are more efficient;and, if so, why.

Are schoolswith proportionatelvmore local fundingmore efficient? To answer this question,we compare the cost of schools that have more local funds relativeto their budget with those that have less. However, a simple comparisonof average expenditureis not enough. Other factors,such as school size and the relativecost of attractinglabor to the school,must be held constant. To do so, we use cost function analysis (see Fox 1980). A school's short-runtotal cost is a functionof outputs, input prices and its fixed capital stock. Economictheory suggests that the choice of functlinalform of this functionis arbitrary (i.e., any positive, homogeneous,non-decreasing,concave functionof factor prices is a cost function). In this exploratorystudy, we estimatethe simple double-log(or Cobb-Douglas)form.1

'To test the robustnessof our results,we also estimatedother forms, such as the linear and quadratic. Since our resultswere insensitiveto these specifications,we do not present the results. Of course, all of the above specificationsimpose prior parametricrestrictionson the production process. For example,the Cobb-Douglasassumes a unit elasticity of substitutionbetween any two factors. In subsequentversions of the paper, we may use more general specifications(see Jimenez 1986). - 12 -

We use two basic cost functionmodels. In the first, we assume the traditionalone-output (enrolment)model. For the ith school, short-run cost (C) is determinedby: student enrolment (S), a proxy variable for labor cost confrontedby the school (W - the daily wage rate of unskilled adult workers in the communitywhere the school is located),and, as our innovativecomponentto measure x-efficiencywithin schools,the proportion of current school expenditurefrom local sources (F). Thus, for the ith school, the cost functionis as follows:

(1)

ln Ci - a + b ln Si + c ln Wj + d Fi + U$

where ln signifiesnatural logarithm;a, b, c and d are parametersto be estimatedand u is a randomlydistributederror term. We note that the proxy variable to control for differencesin factor prices is an imperfect one. For example, it could also be an indicatorof the average socioeconomic status of the populationfrom which the school is drawing its students. We neverthelessinclude it in this exploratorypaper. In subsequent studies,we hope to get a better measure, such as averagewages of teachers by region. The estimated regressionequation,using OLS, is presented as equation (1) of Table 2. The principalvariablewith which we are concerned is the correlationof the financialdependencyvariablewith school expenditure. Its coefficient (d) is negativeand significant. This implies that per student cost is lower among schools that are more dependenton local finance. This finding,ho%.'ever, does not yet show that they are more efficient. The issue may be raised that lower per student cost may have been achieved at the expense of school quality.Consequently,we estimate a

-

13 -

Table 2: Cost Functions

Total Current School Expenditure (1)

Personnel Maintenanceand Expenditure OperatingExpenditure

(2)

(3)

(4)

PROPLOC (F)

-1.61

-1.66

-2.400

( 6.27)

( 6.27)

( 7.35)

-1.454 ( 2.97)

Ln STUDENT SIZE (S)

.76 (15.84)

.740 (12.87)

.745 (10.36)

.355 ( 3.30)

.196

-.366 (1.10)

.697 (1.40)

-.299 (1.96)

-.504 (2.20)

-.0004 (-.00)

-.329 (1.72)

Ln SCHOOL QUALITY

(Q)

( .71)

RURAL (R)

( .34)

AFFECTED (A)

( .42)

-.046 -.044

Ln WAGE (W)

.127 1.11)

( .90)

.162 (1.20)

-.346 (1.72)

CONSTANT

7.19 (19.40)

6.67 (6.01)

8.592 (6.77)

5.566 (2.92)

.576 226

.573 226

.418 338

.137 338

R2 N

.105

second basic cost functionmodel in which the simple regression specificationwas expandedto include school quality and other control variables. Specifically,the followingequationwas estimated:

(1)

ln Ci - a + b ln Si + c ln Wi + d Fi + g Ri + k Ai + m ln Qi + ui

where Q - school quality;R - 1 if rural, 0 otherwise;A - 1 if the school is in an "affected'or disadvantagedarea, as defined by PRODED,

-

14 -

O otherwiselg,

k, and m are parametersto be estimated. School quality is

measured as the averageachievementscore of grade 4 pupils who were tested by HSMS in mathematics,English,andPilipino.These are the groups of 20 grade 4 students that HSMS randomlyselected from each of the sample schools and were given curriculumbased achievementtests. The dummy variables R and A were includedto further control for local differences. The estimates for this expandedspecification,which are presented in column 2, Table 2, confirm our earlier finding of a significantnegative correlationbetween school expenditureand dependenceon local finance.It is intert..ing that the coefficientof F and its t statistichardly changed, suggestingthat the correlation between cost and degree of local finance is robust.

Why are schoolswith relativelymore local financingmore efficient? The regressionestimatesalso reveal that, while student size clearly determinesschool expenditure,the coefficientof school quality is not statisticallysignificant.We conjecturethat most public schools are not getting the most out of their budget in terms of student learningand/or the number of studentsbeing taught. If current expenditureper student is lower for schools that are more dependent on local finance,holding school quality constant,where are the cost similar

savings

coming from? To examine this

to equation

and maintenance

(2) above was estimated

and operation

expenditure

separately

an expenditure

function

for personnel

cost

(MOE).

In both the MOE and the personnel of Table 2) the coefficient

issue,

of F is negative

cost

funtions

and significant.

(columns It

3 and 4 is

interestingthat the absolutevalue the coefficientis larger for MOE than for personnel expenditure.This implies that on a percentagebasis an

-

15 -

increase in the relative Importanceof local finance in school budgets tends to be associatedwith greater cost savings in personnel expenditurethan in MOE. If indeed schools in the Philippinesare inside the production possibility frontier,why is expendituresystematicallycorrelatedwith number of student but not with school quality?We conjecturethat the explanationlies probably in the prevailingstructureof incentives. Budgetaryallocationis clearly tied to the number 'f students in school and not to their learningachievement. Furthermore,performanceindicator for student size is troretransparentthan learning.Consequently,it may be expectedthat a budget maximizingbureacrata la Niskanen i1971)would tend to accommodateenrolmentsfirst and keep expenditureper student in the neighborhoodof the standardallocationformula normallyused. Furthermmore, in the absence of a similarmotive force for a cost-effectivepursuit of school quality,there may be a tendencyto treat as 'residualvariables' student learning achievementand the cost of improvingit. In the context of this environmentit is, therefore,not unreasonableto expect these variables to be largely determinedby the idiosyncraciesof individual teachers and school officials. Hence, school quality and cost tend to be uncorrelated.

CONCLUSIONS

The main purpose source

of this'

paper was to test the hypothesisthat the

of funding matters in determiningcost of schools. Our conclusion,

based on this preliminaryevidence, is that schoolswhich rely more heavily on local sources are more efficient -- i.e., have lower cost, holding con- 16 -

stant for enrolment and quality. Decentralizedschools appear to save on personnel costs. Since teachersare assignedcentrally,most of these savings are likely due to fewer non-teachingpersonnelor in lower salaries. As far as we know (and we would be gratefulif readers can point out other referenceswhich we have not uncovered)this evidence is the first empiricalindicationin the literaturethat local financecan lead to improvedcost performance,given size and quality. Because this finding is so new, other studies need to be undertakento test its robustness. However, if found consistent,the policy implicationsare very importantfor the Phi ippines. Since the early 1980's the Philippinegovernmenthas been addressing financing-related problems at the primary level. In 1981 it launched the Program for DecentralizedEducationalDevelopment (PRODED),which seeks, among others to improve policies in financingand administration.Previously,the regionaleducationbudgets had been appropriatedusing a standard formula based on enrollmentexpansionbut totally ignoringthe performance of the regions in terms of qualitativecriteriaas student achievement, participationrate and cohort survival rates. Recently,a new financing policy has been introducedwhich is supposedto take both quantitative and qualitativefactors into account. Furthermore,the national government has decided to finance the salariesof all public school teacher,a move that could weaken local accountability. The financingproblem is further complicatedby the new constitutional mandate to establish free public secondaryeducation. Unless new sourc;es of funds are tapped and/or internalefficiencyin education improved,the quality cf public elementaryeducationcould suffer. A possible restructuringof the DECS budget could occur favoringmore allocations to secondary educationat the expense of primary education.Consequently, - 17 -

there is a need for the governmentto explore efficientand equitablesupplementaryfinancing arrangements. One possiblearea is to tap local communities and householdswhose potentialcontributionshave been limitedby the current financingarrangements,an observationwhich is common in many developingcountries (World Bank 1986).

-

18-

HF-

ICzS

Fox, W.F. "Reviewingeconomicsof size in education,"Journal of Educational Finance, 5:273-96, 1980. Heyneman, S., Jamison,D. & Montenegro,X. Textbooks in the Philippines: Evaluationof the PedagogicalImpact of a NationwideInvestment. EducationalEvaluationand Policy Analysis, 6, 2, 139-150, 1983. Jimenez,Emmanuel "The Structureof EducationalCosts: MultiproductCost Functions for Primary and SecondarySchools in Latin America," Economicsof EducationReview, 5(1): 25-39, 1986. Nlskanen,William A. Bureaucracyand RepresentativeGovernment,Chicago: Aldine-Atherton,1971. World Bank FinancinzEducation in DevelopingCountries:A review of issues and options,Washington,DC 1986.

_

19

-

PPR WorkingPaper Series Title

Author

Black Morket Premia,ExchangeRate UnificatiQnand InflationIn Sub-SaharanAfrica

Brian Pinto

Grants in Intergovernmental DevelopingCountries

Larry S-hroeder

WPS39

Fiscal Policy in Low-IncomeAfrica

WPS40

WPS37

WPS38

WPS41

WPS42

WPS43

Date

Contact

July 1988

R. Blade-Charest 33754

StephenA. O'Connell

July 1988

R. Blade-Charest 33754

FinancialDeregulationand the Globalizationof CapitalMarkets

Eugene L. Ve.sluysen

August 1988

R. Blade-Charest 33754

Urban Proper:y Tax Reform In DevelopingCountries

WilliamDillinger

GovernmentPay and EmploymentPolicies and GovernmentPerformancein DevelopingEconomies

David L. Lindauer

August 1988

R. Blade-Charest 33754

Tax Administration Division

07/13/88

Tax AdministrationIn Developing Countries: Strategiesand Tools of Implementation

WPS44- Assignedto WDR WPS60 WPS61

WPS62

WPS63

WPS64

StudentPerformanceand SchoolsCosts in Privateand Public High Schools in the Philippines

EmmanuelJlmenez VicentePaqueo Ma. Lourdesde Vera

Statusof Universitiesin the Arab Countriesof the Middle East and North Africa

George 1. Za'rour

Does Japan ImportLess Than It Should? A Review of the Econometric Literature

Kenji Takeuchi

July 1988

J. Epps 33710

Cocoa and Coffee PricingPolicies in Cote d'lvoire

TakamasaAkiyama

August 1988

D. Gustafson 33714

PPR WorkingPaper Series Title WPS65

WPS66

WPS67

WPS68

WPS69

WPS70

WPS71

WPS72

WPS73

WPS74

Interactionof InfantMortalityand Fertillltand the Effectivenessof Health and Family PlanningPrograms Slowingthe Stork: Better Health for Women throughFamily Planning

Price and Tax Policy for SemiSubsistenceAgriculturein Ethiopia

A Comparisonof Lamps for Domestic L;ghting in DevelopingCountries Does Local FinancingMake Public Primary SchoolsMore Efficient? The PhilippineCase

VocationalEducationand the Economic Environment: Conflictor Convergence?

School Effectson StudentAchievement in Nigeriaand Swaziland

The RelativeEfficiencyof Public Schools in DevelopingCountries

Taxationand Output Growth: Evidence from AfricanCountries Fiscal Stabilizationand ExchangeRate Instability:A TheoreticalApproach and Some Policy ConclusionsUsing Mexican Data

Author

Date

Contact

Howard Barnum

AnthonyR. Measham Roger W. Rochat

Robert D. Weaver Saad All Shire

August 1988

D. Gustafson 33714

August 1988

T. Hawkins 33678

MarlaineLockheed Andre Komenan

August 1988

T. Hawkins 33678

EmmanuelJimenez MarlaineLockheed VicentePaqueo

August 1988

T. Hawkins 33678

Robert van der Plas

EmmanualJimenez VicentePaqueo Ma. Lourdesde Vera

Arvil V. Adams AntoineSchwartz

JonathanSkinner

Andrew Feltenstein StephenMorris