Education Outcomes in the Philippines

7/30/2019 Education Outcomes in the Philippines

1/69

ADB EconomicsWorking Paper Series

Education Outcomes in the Philippines

Dalisay S. Maligalig, Rhona B. Caoli-Rodriguez,

Arturo Martinez, Jr., and Sining CuevasNo. 199 | May 2010


2/69


3/69

ADB Economics Working Paper Series No. 199

Education Outcomes in the Philippines

Dalisay S. Maligalig, Rhona B. Caoli-Rodriguez,

Arturo Martinez, Jr., and Sining Cuevas

May 2010(Revised: 17 January 2011)

Dalisay Maligalig is Principal Statistician; and Rhona Caoli-Rodriguez, Arturo Martinez, and Sining Cuevas are

Consultants at the Development Indicators and Policy Research Division, Economics and Research Department,

Asian Development Bank. This study was carried out under Regional Technical Assistance (RETA) 6364:

Measurement and Policy Analysis or Poverty Reduction. The authors beneted greatly rom the insightul

comments o Anil Deolalikar, Socorro Abejo, Jesus Lorenzo Mateo, and Joel Mangahas. They also thank thePhilippine National Statistics Ofce and the Department o Educations Research and Statistics Division or

providing the datasets used in this study. Any remaining errors are the authors.


4/69

Asian Development Bank

6 ADB Avenue, Mandaluyong City

1550 Metro Manila, Philippines

www.adb.org/economics

2010 by Asian Development BankMay 2010

ISSN 1655-5252

Publication Stock No. WPS102229

The views expressed in this paper

are those of the author(s) and do not

necessarily reect the views or policies

of the Asian Development Bank.

The ADB Economics Working Paper Series is a forum for stimulating discussion and

eliciting feedback on ongoing and recently completed research and policy studies

undertaken by the Asian Development Bank (ADB) staff, consultants, or resource

persons. The series deals with key economic and development problems, particularly

those facing the Asia and Pacic region; as well as conceptual, analytical, or

methodological issues relating to project/program economic analysis, and statistical data

and measurement. The series aims to enhance the knowledge on Asias development

and policy challenges; strengthen analytical rigor and quality of ADBs country partnership

strategies, and its subregional and country operations; and improve the quality and

availability of statistical data and development indicators for monitoring development

effectiveness.

The ADB Economics Working Paper Series is a quick-disseminating, informal publication

whose titles could subsequently be revised for publication as articles in professional

journals or chapters in books. The series is maintained by the Economics and Research

Department.


5/69

Contents

Abstract v

I. Introduction 1

II. Conceptual Framework 3

A. Data Sources 5

B. Statistical Models 13

III. Results 18

A. Individual Education Outcomes 18

B. School Outcomes 21

C. Quality of Education Outcomes 23

IV. Policy Implications 26

A. Deployment of Teachers and Effective Class Size 26

B. Decentralization 30

C. On Making Access to Primary Education Equitable 32

D. On Working Children 36

E. Other DepEd Programs to Keep Children in School 38

F. On Gender Disparity 39

G. Age of Ofcial Entry to Primary School 40

V. Conclusions and Recommendations 41

Appendix 1: Education for All Targets and Accomplishments, Primary Education 47

Appendix 2: Indicators from Basic Education Information System 48

Appendix 3: Preliminary AnalysisAPIS 50Appendix 4: Reasons for Not Attending School 52

References 58


6/69


7/69

Abstract

This paper identies key determinants of individual, school, and quality of

education outcomes and examines related policies, strategies, and project

interventions to recommend reforms or possible reorientation. Two sets of data

were used: (i) data on school resources and outputs from the administrative

reporting systems of the Department of Education; and (ii) the 2002, 2004,

and 2007 Annual Poverty Indicator Surveys. Analysis of individual, school,

and quality of education outcomes showed that although school resources

such as pupilteacher ratio is a key determinant for both individual and school

outcomes, and that per capita miscellaneous operating and other expenses aresignicant factors in determining quality of education outcome, socioeconomic

characteristics are stronger determinants. Children of families in the lower-income

deciles and with less educated household heads are vulnerable and less likely

to attend school. Girls have better odds of attending school than boys. Working

children, especially males, are less likely to attend secondary school. On the

basis of these results, recommendations in the areas of policy and programs

are discussed to help address further deterioration, reverse the declining trend,

and/or sustain gains so far in improving basic education system performance

outcomes.


8/69


9/69

I. Introduction

Filipino parents value education as one of the most important legacies they can impart

to their children. They believe that having a better education opens opportunities that

would ensure a good future and eventually lift them out of poverty. Thus, they are willing

to make enormous sacrices to send their children to school (Dolan 1991, De Dios 1995,

LaRocque 2004).However, with a poor familys severely limited resources, education

tends to be less prioritized over more basic needs such as food and shelter. Hence, the

chances of the family to move out of poverty are unlikely. It is therefore, important that

the poor be given equitable access to education.

The 1987 Philippine Constitution declares that education, particularly basic education, is

a right of every Filipino. On this basis, government education policies and programs have

been primarily geared toward providingaccess to education for all. The Philippines is

committed to the World Declaration on Education for All (EFA) and the second goal of the

Millennium Development Goals (MDG) to achieve universal primary education by 2015.

EFAs framework of action has six specic goals in the areas of: (i) early childhood care

and education (ECCE); (ii) universal primary/basic education; (iii) life skills and lifelong

learning; (iv) adult literacy; (v) gender equality; and (vi) quality. In line with this framework

of action, the Philippine EFA 2015 National Action Plan (UNESCO 2010) adopted in 2006

was formulated as the countrys master plan for basic education.

In 2000, the Philippines reported that it has achieved substantial improvement in terms

of access to basic education, but still faces challenges in the areas of early childhood

care and development, internal efciency, and learning outcomes (NCEFA 1999).

Through the governments efforts to achieve the 2015 MDG targets, recent studies such

as the Philippines Midterm Progress Report on the MDGs (NEDA and United Nations

Country Team 2007, Table 1) assess that the probability of achieving universal primary

education (MDG 2) in the country is low (based on net enrollment rate, cohort survival

rate, and completion rate). Similarly, the 2009 EFA Global Monitoring Report (UNESCO

2008) identied the Philippines to be among the countries with decreased net enrollmentrate from 1999 to 2006, and with the greatest number of out-of-school children (more

than 500,000). The Philippiness current performance in education based on the trends

identied by the EFA and MDG indicators as shown in Appendix Table 1 is not also

promising. It is quite likely that the EFA and MDG targets will not be met by 2015.


10/69

Overall, the Philippines has suffered a setback in most education outcome indicators.

Although signs of recovery have been registered by some indicators, national targets for

key EFA indicators such as intake and enrollment rates will still likely be missed in 2015.

How can the decline in the performance of EFA indicators of education outcomes beaverted and improvements in those that registered recovery be sustained? This paper

aims to address this question by identifying key determinants of selected major education

outcomes, and on this basis, examine concomitant or related policies, strategies, and

project interventions for purposes of recommending reforms or possible reorientation.

Previous studies have suggested that poverty incidence (socioeconomic status),

government expenditure on education (as a percentage of gross domestic product

[GDP]) and pupilteacher ratio (PTR) are key determinants of school attendance or net

enrollment rate. Except for a few studies covering a specic area in the country, most

related studies in the Philippines examine the relationships of education outcomes and

inputs using exploratory correlations and regressions of inputs and factors that mayaffect education outcomes. These studies do not have an explicit theoretical model to

guide the analysis, and hence could be considered to have been done on a piecemeal

basis, without being able to explore the relationships of all the major factors in one

comprehensive analysis. For example, Maligalig and Albert (2008) concluded that there

is evidence that government expenditure on education and poverty incidence are directly

related to net enrollment ratio, but failed to ascertain the degree of the relationships as

well as the efcacy of other factors that may affect school enrollment.

There are many other methods that could be employed in identifying key determinants

of education outcomes, such as the education production function, which has been

used by many studies cited throughout this paper. Another method is the randomizedevaluations that have already been done in other countries like Kenya, Nicaragua,

and United States; or the natural experiments study conducted in Indonesia by Duo

(2001); or the qualitative methods that are being conducted as part of the Trends

in International Mathematics and Science Study. The education production function

approach usually refers to a mathematical equation between outcomes and inputs and

a statistical method for estimating those relationships. The success of this approach

is contingent upon available data and the application of suitable statistical methods in

estimating the production function. Both randomized evaluation and natural experiments

render controlled comparisons. However, both require extensive planning prior to the

implementation of the study.

For the purposes of this study, as randomized evaluations and natural experiment were

not possible, key determinants of education outcomes were identied by estimating an

education production function based on the combination of data from the Department of

Education (DepEd) administrative reporting systems, and the Annual Poverty Indicator

Survey (APIS) conducted by the National Statistics Ofce (NSO) in between the Family

2 | ADB Economics Working Paper Series No. 199


11/69

Income and Expenditure Survey (FIES). Section II of this paper identies the conceptual

framework that was used; Section III presents the results; while Section IV discusses the

policy implications. The last section presents the conclusions and recommendations of

the study.

II. Conceptual Framework

Many studies on the determinants of education outcomes are based on an education

production function that denes a mathematical relationship between inputs and education

outcome1Ysuch as

Y Y I F R e= ( ) +, , (1)

where Yis a function of Iand F,which are individual characteristics and family

socioeconomic factors, respectively, Ris school resources, and e represents unmeasured

factors inuencing schooling quality. Depending on the availability of data, this

mathematical relationship is estimated using suitable statistical models, of which the

best is identied through evaluation of the models goodness of t and adherence to

assumptions.

The output of an education production function is usually some achievement that can

be measured through indicators. Among these are intake and enrollment rates, cohort

survival rate, dropout rate, and repetition rate, which are all EFA indicators. Another

key education outcome indicator is the learning achievement rate or learning outcomes

usually measured through national standardized tests.

The education production function described in equation (1) requires both measures of

individual and family socioeconomic characteristics as well as school resources. Previous

studies in the Philippines as well as in other countries indicate that there are individual

and household characteristics that inuence childrens participation and performance in

basic education (Bacolod and Tobias 2005, DeGraff and Bilsborrow 2003, UIS 2005).

These studies suggest that family background and socioeconomic factors are as

important as school resources in determining whether a child will attend school, survive,

and complete an education level, and achieve an acceptable level of learning outcome.

In fact, Hanushek (1986) concluded that socioeconomic factors are stronger determinantscompared to school resources.

Individual characteristics such as age, sex, and parents educational attainment are

important factors in achieving better education outcomes. For example, based on the

1 In economic theory, this should be output, which is the result o the production unction, while outcome would be the utility othe output. However, in this study, output and outcome are used interchangeably.

Education Outcomes in the Philippines | 3


12/69

2004 APIS, Maligalig and Albert (2008) concluded that, assuming all other factors stay

the same (ceteris paribus), boys are 1.39 times more likely not to attend school than

girls. Similarly, in examining Indonesias 1987 National Socioeconomic Survey, Deolalikar

(1993) found that males have signicantly lower returns to schooling than females at

the secondary and tertiary levels. The returns to university education are 25% higher forfemales than males. Deolalikar also cited some evidence that older household heads and

better-schooled female household heads provide relatively more schooling opportunities

for their female relatives. Furthermore, community characteristics such as proportion of

villages in the district of residence having access to all-weather roads, access by water,

lower secondary school, etc. have relatively few signicant effects on school enrollment.

School resources, on the other hand, are typically the basic inputs in education, the

most fundamental being the classrooms and teachers. Other important inputs are the

curriculum, textbooks and other instructional materials, water and sanitation facilities such

as toilets, libraries, and science laboratories. Bacolod and Tobias (2005) nd that the

presence of electricity is an important school input positively affecting learning outcome inCebu. As measure of school quality, school resources are expressed as PTR and pupil

classroom ratio, among others.

Previous studies have mixed observations on the effects of school resources on

education outcomes. Case and Deaton (1999) found that prior to the democratic elections

in South Africa in 1999 and conditional on age, lower test scores, and lower probabilities

of being enrolled in education, schools with high PTRs discourage educational attainment.

In their study of time series data from 58 countries, Lee and Barro (2001) found strong

relationships between measures of school resources and measures of outcomes such

as subject test scores, dropout rate, and repetition rate. On the other hand, Hanushek

and Kimko (2000) concluded, based on data from 39 countries, that traditional measuresof school resources such as PTR and per capita education expenditures do not have

strong effects on test performance. Also, Hoxby (2000) on her study of 649 elementary

schools in the United States concluded that reduction in class size has no effect on

students achievement. Hanushek (2003) compiled 376 production functions from 89

individual publications on education outcomes across the United States and concluded

that the evidence on the PTR as an important determinant of education outcomes is

not conclusive. These studies, however, differ on the statistical methods and data used.

The suitability of the econometric methods was not considered nor was data quality

examined. As Case and Deaton (1999) have pointed out, many of these studies were

concerned with the estimation of detailed educational production functions that try to sort

out effects of different resources on education such as PTR, textbook-to-student ratio,pupilclassroom ratio, school buildings, presence of library, per capita expenditure on

education, among others.



13/69

A. Data Sources

Education production functions will be modeled using two major sources: (i) the 2002, 2004,

and 2007 APIS conducted by the NSO; and (ii) administrative data obtained from the Basic

Education Information System (BEIS) and the National Educational Testing and ResearchCenter (NETRC) of DepEd as well as from its budget appropriations.

The rst source of data consists of three rounds of APIS that used almost the same

questionnaire. These surveys are of national coverage with regions as domains,

barangays or enumeration areas as primary sampling units, and housing dwellings as the

ultimate sampling units. Households in the selected housing dwellings are enumerated on

the households income and expenditures and the socioeconomic characteristics of each

member of the household. A responsible adult in the household was asked about each

members age, sex, educational attainment, school attendance, reason for not attending

school, as well as household income and expenditures, among others. More than 50,000

households were surveyed covering the 85 provinces in the Philippines.

The APIS is undertaken during the intervening years of the FIES. Beginning 2004, the

2003 master sample design was used for all household surveys of national coverage

including APIS. The basis of the sampling frame for the 2003 master sample is the 2000

Census of Population and Housing as well as results of past national surveys, such as

the 2000 FIES, the 2001 Labor Force Survey, and the 1997 Family Planning Survey.

Administrative data from DepEds reporting systems stored at the division level could

either be from a province or an independent city. For purposes of consistency with APIS,

the province was set as the unit of analysis. Data were on the most recent ve years

(20022007).

The APIS gathers information on the demographic, economic, and social characteristics

of households, which include health and education data on each family member. Data on

education include school attendance, highest educational attainment, and reasons for not

attending school. Among the cited reasons for absence from school are cost of education,

distance between home and school, availability of transportation, existence of illness or

disability, and whether the member is working or looking for work (Appendix 4).

BEIS was established in 2002 to improve the monitoring and evaluation of basic

education performance. Prior to BEIS, the basic education data system was laden with

an almost 3-year backlog. The BEIS signicantly reduced data backlog with its quicker

consolidation and validation process. It includes data on school inputs (number of

teachers, classrooms, other school facilities) and outcome indicators crucial in assessing

basic education performance in terms of access, internal efciency, and quality. For

school resources, the BEIS uses a color coding system that indicates the status of

divisions and even schools with respect to these resources.



14/69

The BEIS uses three modules. Module I is the Quick Count Module, which gets total

data from the schools (e.g., total enrollment, total number of teachers etc.) by the end of

December every year. The information is used for planning and budgeting for the next

school year. Module II is the School Statistics Module, which collects school data in detail

(e.g., enrollment by grade/year, age proles of enrollees, etc.). This module is designedto collect information from both public and private schools. Module III is the Performance

Indicators Module, which processes the data and presents the outcome indicators.

Figure 1 describes the BEIS data collection process. Annual data collection starts upon

the issuance of a DepEd order to collect public school proles. The order is disseminated

down to the schools where base data on enrollment, dropouts, repeaters, number of

classrooms, teachers, etc. are manually recorded using annual data gathering forms

(government school prole forms for elementary and secondary levels) under Module

II. These forms are submitted to the division ofces where they are encoded and

consolidated in MS Excel les. The division ofces are also responsible for validating

the accuracy of information with the schools before they are submitted to the regionalofces for further consolidation. The regional ofces then submit the data to the central

ofces Research and Statistics Division, which maintains and updates the BEIS annually,

processes the data, and presents the outcome indicators under Module III. The data

remains in MS Excel les that because of their bulk cannot be uploaded on the DepEds

website. Researchers and other users can only access from the internet a one-page fact

sheet on basic education statistics showing the national aggregates of major indicators

for the last 5 years. The researchers may obtain more information from the BEIS through

a written request addressed to the Research and Statistics Division, which provides the

information in soft copy. The BEIS is also internally accessible among DepEds various

ofces and units through its local area network.

Figure 1: DepEd-BEIS Data Source and Collection

National Level: consolidation in BEIS; interpretation, evaluation, and reporting

Regional Level: consolidation of divisional data into regional data

Division Level: consolidation of school data; validation of data with the schools;

computation of gross and net intake rate; computation of gross and net

enrollment rates

School Level: collection of data on enrollment, existing resources, resource gaps,

drop-outs, repeaters; computation of pupil-teacher ratio, pupilclassroom ratio,drop out rate, repetition rate, cohort survival rate



15/69

The DepEd intends to continuously improve BEIS. Under the BESRA, a proposal for

Enhanced BEIS is being explored. This involves developing an automated database

system where even data down the schools (School Information System) can be accessed

from the web. Moreover, DepEd is currently in the process of adopting an ICT-based data

collection scheme that will put in place effective quantitative and qualitative data collectionas well as student tracking systems.

Gross and net intake rates, gross and net enrollment rates, dropout rate, repetition rate,

and cohort survival rate are the key outcome indicators estimated and compiled by BEIS.

These indicators gauge the level of the childrens access to formal basic education and

the school effectiveness in keeping the children.

Indicators such as repetition rate, dropout rate, cohort survival rate, PTR, etc. are

computed based on actual intake and year-to-year enrollment. As such they can be

estimated at the school level and aggregated upward to district, division, regional,

and national levels. Intake and enrollment rates, however, can only be computedat the division level based on the consolidated actual enrollment data, because the

disaggregation of population estimate from the NSO are available down to the division

level only.

The gross intake rate is the total number of enrollees in Grade 1, regardless of age,

expressed as a percentage of the population in the ofcial primary education entry age,

which is currently 6 years old. On the other hand, net intake rate accounts for Grade 1

enrollees expressed as a percentage of the 6-year-old population. The gross enrollment

rate is dened as the total number of children, regardless of age, enrolled in a particular

education level, measured as a proportion of the age group corresponding to that

level. Meanwhile net enrollment rate (NER) accounts for the participation of childrenwho fall within a dened ofcial school-age group.2 While the gross enrollment rate

reects total participation and, to some extent, the capacity of the education system, the

net enrollment rate is indicative of both the quantity and quality of education system

performance and effectiveness with respect to the target age group.

2 Gross enrollment rate can be more than 100% as they include underaged and overaged children but unlike net enrollment

rate it does not reect the quality o participation o the ocial school-age group. In a desirable situation, NER should be or

approaching 100%. It should be noted that values exceeding 100% are recorded in areas/divisions such as Pasig City and Cebu

City and other highly urbanized areas. One possible reason or such condition is that children rom neighboring divisions

(usually rom the province where the city is or rom the peripheral provinces) also attend schools in these cities/divisions,

thus, the enrollment exceeds the school-age population in the host division. But it does not mean that the division has 100%

participation. For additional discussion on NER, reer to Box 1.



16/69

Box 1: Investigating the Accuracy o the Philippiness Net Enrollment Rate

One of the key education indicators is the net enrollment rate (NER), which is chiey used to

measure developments in primary education. In fact, both the EFA and MDG programs utilizethis to evaluate the progress in their respective Goal 2 objectives. On the basis of the NER

current trends (Box Figure 1), it is projected that the Philippines will not likely attain universal

primary education by 2015.

The NER is the ratio of the enrollment for the age group corresponding to the ofcial school

age in the elementary/secondary level to the population of the same age group in a given year.

The ofcial school-age population for the primary level in the Philippines is 611 years; thus, in

order to estimate for the NER, the total enrolled students aged 6-11 must be divided by the total

population of the same age group. In theory, NER should range from 0 to 100%. However, in

practice, as shown in Box Figure 2 where the box plots of NERs of provinces and independent

cities are shown, there are many data points with more than 100% NERs.

This situation merits a closer look at how the data are compiled. There are three possible

sources of errors: (i) the population projections in the 611 age group in provinces and cities

are not accurate; (ii) the total enrollment of ages 611 is not properly captured; or (iii) there are

many cross-provincial enrollees for some provinces and these are not captured at all in the

DepEd administrative reporting system (BEIS).a

Box Table 1 shows the comparison between APIS and DepEd data. The gures for total

population in the 611 age group that DepEd used to compute NER grew at a steady 2.34%

annually from 2002 to 2006 and dropped by 0.14% in 2007. The constant growth rate for 2002

to 2006 is equal to the national annual average population growth rate that the NSO computed

on the basis of the 1995 and 2000 Census of Population and Housing. To derive the 611

population in 2007, DepEd then adjusted the growth rate used and applied the average annualgrowth rate from 2000 to 2007b on the 2000 Census 611 population. With a lower growth

a This can only be validated by a special survey that captures the school location and residence o the children o respondent

households. There is no strong evidence, however, to suggest that there is a signicant number o cross-provincial enrollees.b 2000 and 2007 are census years.

continued.

92

90

88

86

84

82

80

78

90.3

88.7

87.1

84.4

83.2

84.8

2002 2004 2005 20072003 2006

Box Figure 1: Net Enrollment Rate

Trend, 20022007 (percent)

250

200

150

100

50

2002 2004 2005 20072003 2006

Box Figure 2: Net Enrollment Rate

Distribution, 20022007 (percent)



17/69

Box 1. continued.

rate basis of 2.04%, the 2007 population consequently exhibited a declining trend since theadjustment was not back-tracked. Usually, when new census gures become available, the

population projections are also updated. This is not yet the case in the current NER.

Therefore, the use of 2007 Census of Population and Housing estimates without back tracking

the series may have caused an articial increase in the 2007 NER.

Box Table 1: Total Population and Enrollment o Children Aged 611, 20022007

Year Population, Aged 611

(millions)

Total Enrollment,

Aged 611 (millions)

NER

(%)

Growth (DepEd)

(%)

APIS DepEd APIS DepEd APIS DepEd Popu-

lation

Enrollment

2002 11.76 12.00 10.37 10.83 88.2 90.3

2003 12.28 10.90 88.7 2.34 0.59

2004 12.59 12.57 11.11 10.95 88.2 87.1 2.34 0.45

2005 12.86 10.86 84.4 2.34 -0.80

2006 13.16 10.95 83.2 2.34 0.86

2007 13.04 13.14 11.59 11.15 88.9 84.8 -0.14 1.81

... means not available or not applicable.

Note: Annual population growth is 2.34% or 19952000 based on the 2000 census; and 2.04% or 20002007 based

on the 2007 census.

Another point investigated is the use of national population growth estimates instead of age-

specic population growth rates. The 2.34% growth rate applied by DepEd to the 20022006

population is the 19902000 average annual growth rate of the Philippines. Similarly, the

2.04% growth used for the 2007 estimate is the also the rate at the national level for the years

20002007. However, if the national average annual population growth rate projections for

20012005 is to be computed, it is only about 2.1%. And if the estimation is to be agespecic,

the average annual population growth rate for the 611 age group is only about 1.04%. c These

two gures are lower than the 2.34% that DepEd employed to project total population of ages

611. Box Figure 3 shows the various NER trends based on (i) the 2.34% population growth

rate used by DepEd for 20022006; (ii) the 2.04% rate if the population adjustment will be back

tracked; and (iii) the 1.04% rate, if the age-specic 611 growth rate is to be applied. Thus, the

type of population estimator used by DepEd has contributed to the rate of decline in NER from

2002 to 2006.

c Estimated based on the 2000 Census o Population and Housing population projections by age group that NSO publishes in

its website, and by assuming that the population counts are evenly distributed across ages in an age group.

continued.



18/69

To validate the total enrollment as compiled by BEIS, similar estimates from the Annual

Poverty Indicator Survey were derived. The APIS is a survey of national coverage that the

NSO conducts in the intervening years of the Family Income and Expenditure Survey. All

family members are asked about his/her age, whether he/she is attending school and if not,

the reason for not doing so, among others. Hence, APIS could also provide estimates of the

population in the primary age group as well as the population in the same age group who

are in school. The total enrollment estimates from APIS are within acceptable error margin

(one standard error) compared to the DepEds total enrollment and hence, there is no strong

evidence that DepEds total enrollment data is not accurate.

It should be noted, however, that based on APIS data, a substantial number of 6-year-olds are

not yet in primary school even though by DepEds guidelines, the ofcial age of entry to primary

school is at 6 years old. About 830,900 6-year-old children were not in primary school in 2007;

37.5% have not started school yet; while 62.5% were still in preschool. This is equivalent to

about 6.4% of the total population in the 611 age group. On the other hand, examination of the

composition of enrolled 7-year old students showed that, although by DepEd guidelines, they

should be in the Grade 2 level, most of them are still in Grade 1. In 2002, half of the 7-year olds

who are enrolled are in Grade 1. And although this proportion steeply declined in 2004, it rose

again in 2007 resulting to a nearly equal number of 7-year-old students in Grade 1 and Grade 2.

This is an unexpected occurrence since it is anticipated that because DepEd has implemented

its guidelines on the ofcial age of entry to primary school in 1995, the number of enrolled 7

year-olds in Grade 1 should have been declining since then. These ndings suggest that though

the ofcial school age starts at 6 years, there is still a signicant percentage of families sending

their children to primary school at a later year, thus contributing to the articial decline of the

NER.

Box Figure 4 shows the APIS and DepEd estimates of NER, which is another form of validation

that was used. While DepEds NER is steadily declining, the equivalent APIS indicator remained

steady between 2002 and 2004, and showed a slight increase by 2007.

Box 1: continued.

92

90

88

86

84

82

80

782002 2004

NER at 2.34% population growth NER at 2.04% population growth

NER at 1.04% population growth

2005 20072003 2006

Percent

Box Figure 3: Comparative NERs Based on Alternative Population Growths

continued.



19/69

Box Figure 4: NER Trends, 20022007 (percent)92

90

88

86

84

82

80

78

DepEd

APIS 6-11

2002 2004 2005 20072003 2006

90.3

88.288.7 88.2

87.1

84.4

83.2

84.8

88.9

The four indicators discussed aboveNER, gross enrollment rate, net intake rate,

gross intake rateare compiled in BEIS at the division level using data from schools

as numerator and as denominator, the population projections for the corresponding age

groups from the NSO. A closer examination (see Box 1) of the net enrollment rate, which

is the main indicator for universal primary or universal basic education goals of both EFA

and MDG, reveals that there are aws in the estimation process. For example, the fast

decline of NER as reected in the BEIS data series seems to be caused by the higher

population projections from NSO.

Once the children are in school, the next order of business is how to keep them engagedso that they are able to acquire the identied skills and levels of competencies dened

in the curriculum. How well the schools can keep the children from leaving before

completing a particular education level gauges the schools internal efciency. Indicators

of internal efciency include cohort survival rate, dropout rate, and repetition rate. The

cohort survival rate in a certain education level is the percentage of a cohort of pupils/

students enrolled in the rst year of that level who reach the last grade/year of that

particular education level. It indicates the holding power of the school. A desirable pattern

is that it should approach 100% and that its movement should have a negative relation

with the dropout rate.

Distortions in cohort survival rate are mainly the result of high dropout and repetition

rates. Dropout rate accounts for those pupils/students who leave school during the year

and those who complete the previous grade level but do not enroll in the next grade/

year level the following school year. It is expressed as a percentage of the total number

of pupils/students enrolled during the previous school year. Repetition rate serves to

measure the occurrence of pupils/students repeating a grade. It is technically dened as

Box 1: continued.



20/69

the percentage of a cohort of pupils enrolled in a grade at a given schoolyear who study

in the same grade the following schoolyear.

The National Achievement Test (NAT) is the primary indicator of school effectiveness

based on pupil/student scores in subjects like language, science, and math. The NAT isadministered by DepEd through its National Educational Testing and Research Center,

whose functions include analysis and interpretation of data for policy formulation and

recommendation. Making a time-series comparison of NAT results from 2002 to 2007 is

problematic since the tests are administered at different grade or year levels annually.

The NAT was rst administered in 2002 to Grade 4 and 1st year high school students. It

included a diagnostic component conducted at the start of schoolyear to determine the

academic weaknesses or learning gaps of the pupil/students based on the curriculum-

prescribed learning competencies at a particular level. The results of this diagnostic test

are compared with the achievement tests administered to the same group of pupils at

the end of the schoolyear to determine learning progress. In the following schoolyears,

however, the NAT was administered in different grades and years.

Two indicators of school resources that will be used in the models are the miscellaneous

operating and other expenses budget (MOOE) and the personnel salary (PS) budget.

The budgeting division, working closely with Ofce of Planning Services, computes for

the MOOE based on a formula (per capita student cost and school-based). They use

the quick count data from BEIS to estimate the next schoolyears enrollment and the

MOOE. However, they also request the regional ofces to submit MOOE proposals that

they only use for validation purposes. The budget for PS is computed based on current

staff complement and increases only for new hires and promotions. Data on PS and

MOOE used in this study were taken from various Congress-approved Government

Appropriations Acts based on the National Expenditure Program proposed by thegovernment. Using the DepEd budget, however, does not present the complete basic

education nancing because it does not account for the contributions of private schools,

which comprise 8% of total elementary school enrollment and 21% of secondary school

enrollment.

These data also do not include the contributions of the private sector and local

government units. DepEd has forged partnerships with private and business sectors

in projects such as Adopt-a-School and is implementing other private sector initiatives

that have resulted in valuable contributions that are also quantiable but are not being

captured in the BEIS or by any DepEd unit. Local government units also contribute

signicantly to basic resources needed by the schools. Among these local sources is theSpecial Education Funds (SEF) coming from the 1% real property tax earned by local

governments and earmarked forbasic education as provided for in the Local Government

Code. The SEF is used for construction and rehabilitation of classrooms as well as for

funding salaries of locally hired teachers.



21/69

The available administrative data do not include individual and household characteristics

of the pupils/students (e.g., socioeconomic status and ethnic or linguistic variation).

Moreover, accuracy is often an issue with administrative data, especially since the

collector and processor of information are also its main users. As a result, over-reporting

or under-reporting to inuence decisions on funding and other incentives can happen(UIS 2008).

A more rigorous study that is also the approach taken by this research is to combine

education administrative data with census or household surveys. Although often

conducted less regularly, household surveys provide more information on the

characteristics of individuals and households that often inuence decisions related to

education services made available by the government. Corresponding to the two major

data sources described above, two datasets were constructed: (i) the household/individual

data that combines APIS and the provincial-level PTR; and (ii) provincial-level data that

consists of data from BEIS, NETRC, and the Financial Management System but which

also includes provincial-level indicators from APIS such as the proportion of females,median educational attainment of the household head, and median household per capita

income.

B. Statistical Models

On the basis of the available data described above, a modeling framework was

developed (see Figure 2). In this framework, the decision to attend school is considered

as an investment that promises future returns. First, it is hypothesized that the decision

whether to attend school or not is mainly inuenced by personal circumstances. The

process of deciding whether to attend school or not usually starts at the household

level and is depicted by the dotted arrows pointing directly from household, personalresources, to the decision of attending school. Once the household decides to send

the child to school, there are different possible education outcomes that are measured,

such as dropout rate, survival rate, repetition rate, and NAT score, among others. These

education outcomes are directly inuenced by education inputs, but household and

personal resources are also contributing factors.



22/69

Figure 2: Model Framework

Household, Personal

ResourcesEducation Inputs

(School Resources)

(Individual

Outcome)

Decision to

attend school

School Outcomes

Repetition

Rate

Dropout

Rate

Survival

Rate

NAT

Score

Individual outcome (decision to attend school) is modeled using a combination of the

household/individual data from APIS and the provincial PTR from BEIS. All schooloutcomes including the quality of education outcome are modeled using the combined

administrative data and provincial estimates of key individual and household variables

from APIS.

In the case of the APIS dataset, for each year (2002, 2004, and 2007), a probability

sample is drawn and hence, the set of households and individuals in the data set were

selected randomly. Because of this, a random effects model is explored, such that

subject specic parameters i{ } are treated as draws from an unknown population(and thus may be considered random). Moreover, the outcome that will be modeled for

this data set is school attendance, a binary variable that can be modeled suitably by a

logistic regression using random effects likelihood estimation. Unlike the administrativedataset, individuals, which are the unit of analysis, are only measured once; therefore,

if individuals are considered the subject in the model, a longitudinal analysis approach

is not possible. However, since the regions are the domains of the APIS and housing

dwellings are drawn from clusters or primary sampling units from strata dened within

regions (but are not similar across regions), the random effects that can be accounted for

clustering of responses are within the domains (region) and across years, such that

lnP y

P y

tdi td

tdi td

td

=( )=( )

= +

1

0

xtdi . (1)

where ytdi is the education outcome of the ith individual in region dand yeart, x

tdiis the

corresponding vector of explanatory variables, and td is the domain-specic nested intime parameter representing heterogeneity across time and regions. The results of the

random effects model are also compared with that of the more commonly used ordinary

logistic model.



23/69

Three types of explanatory variables are considered in the models: (i) individual

characteristics such as sex and age; (ii) household characteristics such as household

per capita expenditure, and age and educational attainment of the household head;

and (iii) PTR at the provincial level representing school resources. The factor other than

household characteristics that could affect the parents decision to send their childrento school is their perception on the capacity of the school. A measure of this perception

that is available is PTR because in general, parents believe that their children would get

better education if the classrooms are not crowded. Other indicators of school resources

were considered but dropped from the model because they were not used by parents or

individuals in their decision to attend school or not. These are the proxy for the average

teachers salary and the per capita MOOE. Moreover, these two indicators cover only the

public school system and there are no corresponding data from the private schools.

For school education outcomes such as the NAT overall rating, NAT average test scores

in Science, Math, English, and Filipino; dropout rate; cohort survival rate; and repetition

rates were considered. Since the BEIS dataset is the major data source for modelingthese education outcomes, the unit of analysis was the province, since this is the lowest

disaggregation level at which the full set of data across the most recent 5 years is

available. Also, for most of the provinces, data have been recorded for the most recent

5 years. Thus, longitudinal analysis3 was conducted instead of cross sectional analysis.

Longitudinal analysis is more complex than regression or time series analysis but it has

the ability to study dynamic relationships and to model differences among subjects. It

can be shown that the educational outcomes signicantly vary across provinces. Hence,

provincial-specic parameters will be included in the model such that

E yit i it ( ) = + x (2)

where i

is the ith province-specic parameters, yit is the educational outcome at year

t and province I, while xit

is the vector of explanatory variables. These variables are

further described herein. There are two distinct approaches for modeling the quantities

that represent heterogeneity among the subjects (in this case, provinces) i{ } : (i) xed-

effects model in which i{ } are treated as xed yet unknown parameters that need to

be estimated and (ii) random effects model in which i{ } are treated as draws from anunknown population and thus are random variables such that

E yit i i it ( ) = + x (3)

Considering that measures from all provinces that are the subjects or units of analysisare included in the datasets, and that provincial-level measures were derived from data

3 Longitudinal analysis is a combination of various features of regression (cross-section and time series analysis). It is

very much like regression analysis because it examines a cross-section of subjects (unit of analysis). On the other

hand, it is similar to time series because subjects are observed over time. In this paper, instead of using the 5-year

BEIS data, modeling is restricted for the years when APIS were conducted since some APIS variables were merged

in the BEIS data.



24/69

of all schools in the province, the possibility of a provincial measure to vary because of

a random draw (sample) can be eliminated and hence, xed effects model is deemed

appropriate.

Since the education production function is not complete without socioeconomiccharacteristics that are not found in BEIS or any other government administrative

reporting system, some provincial-level indicators from the APIS such as the proportion

of females, median education attainment of the household head, and median household

income were combined with the dataset. As a consequence, only 2002, 2004, and 2007

data were included in the nal data set.

There are many situations in educational and behavioral research in which multiple

dependent variables are of interest. Usually, separate analyses are conducted for each

of these variables even though they are likely to be correlated and have similar although

not identical set of predictor variables. In this research, a good example would be the

average NAT scores for English, Science, and Math that are also available for most ofthe provinces. These subject NAT scores are highly correlated and hence, to accurately

capture this situation, an alternative modeling approach, the seemingly unrelated

regression (SUR) was used. SUR is a technique for analyzing a system of multiple

equations with cross-equation parameter restrictions and correlated error terms.

The SUR technique estimates separate error variances for each equation; hence separate

R2s can be computed. Numerous parameter restrictions employed in SUR, however,

may lead to negative R2.A potential advantage of its application in panel data analysisis to allow for same parameter estimates of the xed effects using different correlated

dependent variables. Further, it moves away from the potential problem that unbalanced

data may cause under xed or random effects framework.

Since separate data series for primary and secondary schools are provided in the

administrative dataset, separate models for primary and secondary age groups were

derived and examined. To apply these models in the APIS dataset, the primary and

secondary age groups have to be designated. The issue of the ofcial age of entry to

primary education arose in the process. Per DepEds policy, the ofcial entry age to

formal primary education is 6 years old. However, preliminary analysis of APIS revealed

that a substantial numbers of6-year-olds were not yet in school (21.5% for 2002, 17.5%

for 2004, and 15.2% in 2007) and a signicant proportion is still in preschool (27.2% for

2002, 26% for 2004, and 25.3% for 2007) (Table 1).



25/69

Table 1: Age-Specic Enrollment Rates, APIS 2002, 2004, 2007 (percent)

Age 2002 2004 2007

Enrolled Pre-

school

Primary Secondary Enrolled Pre-

school

Primary Secondary Enrolled Pre-

school

Primary Secondar

6 78.55 27.18 51.37 82.5 25.96 56.54 84.8 25.33 59.487 93.91 2.97 90.94 94.02 3.46 90.56 94.19 3.07 91.12

8 96.78 0.89 95.89 96.87 0.69 96.18 96.2 0.5 95.7

9 97.86 0.33 97.53 97.37 0.18 97.19 97.32 0.26 97.06

10 97.79 0.15 97.53 0.11 96.79 0.18 96.61 96.83 0.04 96.79

11 97.84 0.01* 93.6 4.23 96.76 91.92 4.73 96.26 0.06* 91.3 4.

12 94.87 0.01* 56.65 38.21 94.16 56.23 37.88 94.44 0.1* 52.76 41.5

13 92.41 22.37 70.04 90.62 23.32 67.21 90.36 0.05* 21.74 68.5

14 88.66 10.46 78.1 86.56 11.09 75.33 86.76 10.29 76.4

15 84.62 4.39 79.33 82.85 4.76 76.67 82.2 0.04* 4.91 74.0

16 74.32 2.3 57.87 70.72 2.28 53.45 66.97 2.06 43.4

17 60.12 0.03* 0.76 23.73 56.6 1.01 23.07 54.38 1.16 20.8

Zero values.

* Nonzero values; suspected to be encoding errors.

Source: Authors computations using APIS 2002, 2004, and 2007.

In fact, both the DepEd administrative and APIS data across years (2002 to 2007)

showed that less than half of 6-year-old children are not yet in primary school. BEIS

reported that 63.36% of Grade 1 enrollees are older than 6 years. Of these overaged

Grade 1 pupils, 63.44% are 7 years old. Parents appear to postpone enrollment at 6

years old and tend to send their children to school when they get older. And since this

study does not aim to determine when the child is sent to school but the decision whether

the child is sent to school or not, the age groups that will be used for primary and

secondary school were 712 and 1316 years old, respectively.

In addition to data availability and results of previous studies, endogeneity issues are

also considered in determining the explanatory variables that will be included in the

models. Explanatory variablessuch as total enrollment, number of teachers, budget

for personnel salary and wages, and budget for miscellaneous operating and other

expenseswhich also vary according to the school size and consequently, the size of

the province are taken out of the list and instead, corresponding variables that are not

robust to school size are considered, such as PTR, average teacher salary, and per pupil

MOOE. The median per capita household income, median educational attainment of the

household head, and proportion of females for the appropriate school age group that

were estimated from APIS at the provincial level represent the household and individual

characteristics.



26/69

Preliminary analysis of APIS data for 1316-year-olds as presented in Table 2 shows

that a sizeable number of 1316-year-olds are already working and may not be able to

attend school. Hence, a binary variable corresponding to working or not could be a good

explanatory variable for the secondary school age group individual outcome model. But

having work can be viewed as an outcome of a childs time allocation process (Khanamand Ross 2005), and in this case, a possible endogeneity problem may exist. Moreover,

it is difcult to identify the true effect of work on school attendance since the factors

that encourage children to work tend to be the same conditions that discourage school

attendance. These issues, however, do not apply in the case of the APIS dataset in which

each family member was asked for his/her reason for not attending school. One of the

major reasons cited is already working.

Table 2: Working 1316-Year-Olds by Age and Sex

Year Age Total Population (thousands) Already Working (percent)

Male Female Total Male Female Total

2002

13 910.52 893.16 1,803.69 11.51 6.07 8.8114 864.14 814.48 1,678.62 17.05 7.96 12.64

15 948.41 848.66 1,797.07 21.57 8.62 15.45

16 821.95 758.80 1,580.75 27.28 12.57 20.22

All 3,545.01 3,315.10 6,860.12 19.21 8.67 14.12

2004

13 1,011.76 980.78 1,992.54 11.09 6.10 8.64

14 974.99 903.81 1,878.80 17.43 7.02 12.42

15 960.09 1,006.47 1,966.56 22.68 7.98 15.16

16 957.82 944.84 1,902.66 29.68 10.85 20.33

All 3,904.66 3,835.89 7,740.55 20.09 7.98 14.09

2007

13 1,142.57 1,082.80 2,225.37 9.68 5.11 7.45

14 1,078.04 1,062.66 2,140.70 13.91 7.52 10.74

15 1,082.29 1,182.89 2,265.18 20.55 9.84 14.96

16 1,055.42 1,119.36 2,174.78 27.63 14.85 21.05

All 4,358.32 4,447.71 8,806.03 17.77 9.39 13.54

Note: Values may not add up to totals due to rounding of.

Source: Authors computations using APIS data.

III. Results

A. Individual Education Outcomes

Table 3 presents the best models for log odds of attending school for the 712 and 1316

age group. For the primary age group, age, sex, per capita expenditure of the household,

highest educational attainment of the household head, and PTR are the signicant

explanatory variables.



27/69

Table 3: Random Eects Models or Log Odds o Attending School

Explanatory Variablesa Random Eects Logistic

Age: 712 Age: 1316 Age: 712 Age: 1316

Age = 8 0.69** 0.69**

Age = 9 1. 00 ** 1.00**Age = 10 0.93** 0.93**

Age = 11 0.79** 0.79**

Age = 12 0.21** 0.21**

Age = 14 (0.36)** (0.36)**

Age = 15 (0.68)** (0.68)**

Age = 16 (1.48)** (1.48)**

Sex (1 = male) (0.43)** (0.30)** (0.43)** (0.3)**

log(per capita household expenditure) 1.03** 0.86** 1.04** 0.86**

(1 = i household head is male) 0.02 0.07** 0.02 0.08*

Age o household head 0.00 0.01** 0.00 0.01**

(1 = i household head is working) (0.05) 0.23** (0.05) 0.24**

Highest educational attainment o household head 0.13** 0.11** 0.13** 0.11**

Pupilteacher ratio (0.02)** (0.01)** (0.01)** (0.01)**

(1 = i child is working) (2.29)** (2.28)**

Variance (random intercept due to year diferences) 0.05 0.05Variance (random intercept due to regional

diferences)

0.13 0.17

Log likelihood o model (13376.87) (18530.94) (13333.15) (18469.04)

Pseudo R2 based rom simple logistic model 0.14 0.28

Rescaled R2 0.02 0.11

Number o observations 91243 57011 91243 57011

AIC 26783.75 37089.87 26726.29 36996.08

BIC 26925.07 37215.18 27008.93 37255.66

** means statistically signicant at 5% (p-value is at most at 0.05); * means signicant at 10% (p-value is at most 0.10).

0.0 means magnitude is less than hal o a unit.a Similar models were estimated incorporating sex-slope interaction with pupilteacher ratio. The results are presented in

Appendix 3. The variable is signicant or the primary school model but not or the secondary school model.

Note: P-value is the probability o observing an extreme or more extreme value or the test statistic under the null hypothesis

that the parameter coecient or the variable under consideration is zero. Smaller p-values suggest statistical signicance.

The models use random intercepts to incorporate random variations due to diferences in years and regions where theobservations come rom. Random efects are characterized by their variance components.

Statistical signicance o random efects is not directly estimated. Note that some multilevel-structural estimation

methods such as this do not allow the use o weights. But a preliminary analysis on the ordinary logistic regression results

reveals that there is no substantive diference between weighted and unweighted models. Results provided above are all

unweighted.

The Rescaled R2 provides a measure o the improvement on the amount o variation captured by including xed efects in

the model (i.e., the null log likelihood is estimated rom a pure random intercept-model).

Source: Authors computations using BEIS and APIS data.

Assuming all other variables stay in the same level (ceteris paribus), the following

conclusions can be derived from the model:

(i) As the child gets older up to 9 years old, the more she/he would be likely inschool. However, the odds taper off after 9 years old. In fact, when the child

reaches 12 years old, for the elementary age group model, the odds of attending

school decreased dramatically. In particular, the odds of attending school at

age 12 is approximately half than that of age 9. Figure 3 provides a graphical

representation of age-specic enrollment rates.



28/69

(ii) Girls are1

0 4342021exp .( )or 1.54 times more likely to attend school than boys.

(iii) A 1% increase in per capita household expenditure will translate to about 1.03%increase in the odds for attending school.

(iv) The more educated the household head, the better the odds of the child to be in

school. In fact, the odds of attending school increase by 13% for every year of

increase in the educational attainment of the household head.

(v) A unit increase in PTR will reduce the odds of attending school by 2%.

In the case of the model for secondary school age children, all the explanatory variables

were signicant. However, in terms of magnitude of the coefcients, the explanatory

variable with the strongest inuence is if the child is working or not. If the child is working,the odds of him/her not attending school is 9.87 times greater than when he/she is not

working, all other variables being equal. Other results on ceteris paribus assumption are

as follows:

(i) Older children are less likely to be attending school. From age 13 to 16, the odds

of attending school uniformly decrease. The steep decline is noticeable especially

between age 15 and 16.

(ii) Girls are 1.35 times more likely to attend school than boys.

(iii) A 1% increase in per capita household expenditure translates to about 0.86%increase in the odds for attending school.

(iv) The more educated the household head, the better the odds of the child to be

in schoolaround an 11% increase for every year of increase in the educational

attainment of the household head.

(v) The child in a household with a head who is working is 1.26 times likely to be

attending school than a child whose household head is not working.

(vi) A unit increase in PTR will reduce the odds of attending school by 0.8%.

To probe further the odds of attending school at a different age, we can examine Figure 3

in which the proportion of school attendance by age group for the 2002, 2004, and 2007

APIS is presented. This gure illustrates the shift in signs for age when modeling odds

of attending school. Until the age of 9 or 10, there seems to be an upward trend of age-

specic enrollment rates, thereafter, age-specic enrollment rate declines.



29/69

Figure 3: Age-Specic Enrollment Rates (percent)

100

90

80

70

607 8

2002 2004 2007

9 10

Age

11 12 13 14 15 16

Source: Authors computations using APIS data.

B. School Outcomes

On the basis of variability of education outcomes across observations from the panel data

considered, dummy variables for time period (year) and provinces were introduced to

explain heterogeneity across years and the variation across provinces, respectively.

Tables 4 and 6 present the estimates of the coefcients of the models, the p-values of the

corresponding tests of signicance, and other model diagnostics for school efciency andquality of education outcomes, respectively.

Except for survival rate in secondary schools, the models above have good R2 values,4

which for this type of statistical model is a good measure of t. Note, however, that there

are two modelsprimary dropout rate and survival ratethat do not have signicant

explanatory variables but have signicant provincial effects, though not reected in

the table. This implies that the variations of primary dropout rate and survival rate

are largely determined by the variations of the dependent variables across provinces.

These variations represent those explanatory variables that were omitted in the models.

For example, the quality of school management varies across provinces, as well as

the nancial support of local government units. These explanatory variables were notrepresented in the models because there were no readily available and comprehensive

measures to represent them.

4 R2 measures the proportion o variation o the dependent variable (in this case, education outcome) that is explained by themodel. R2 ranges rom 0 to 1. I it nears 1 it implies that the model has adequately explained the variations in the dependent

variable.



30/69

Table 4: Fixed Eects Models or Dropout Rate and Survival Rate

Explanatory Variables Education Outcomes

log(dropout rate) log(survival rate)

Primary Secondary Primary Secondary

log(per pupil MOOE) (0.07) (0.10) 0.04* (0.11)

Pupilteacher ratio 0.03 (0.01) (0.02)** (0.00)

log(teachers salary)a 0.03 (0.12) (0.01) 0.33**

Median household head

educational attainment

(0.00) (0.06)** (0.01) 0.01

Median provincial household per

capita income

(0.00) 0.00 0.00 0.00**

Proportion o emales (0.62) (0.42) 0.00 (0.27)

2004 (0.02) 0.00 (0.00) 0.02

2007 (0.01) (0.00) 0.01 (0.00)


Test or heteroskedasticity 0.11 0.00 0.00 0.01

Adjusted R2 0.82 0.58 0.70 0.18

** means statistically signicant at 5% (p-value is at most 0.05); * means signicant at 10% (p-value is at most 0.10).

0.0 means magnitude is less than hal a unit.a Similar statistical models where the proxy variable or teachers salary was normalized as a proportion o provincial per capita

income were also estimated. Still at the 0.05 level, the variable is not statistically signicant.

Note: Unit o analysis is province or the years 2002, 2004, and 2007.

P-value is the probability o observing an extreme or more extreme value or the test statistic under the null hypothesis


For models that do not satisy constant variance assumption, robust standard errors are used and the corresponding

p-values are reported.

The results above are based on the traditional view o xed efects models where the panel efects (in this case, provincial

efects) are treated as parameters to be estimated. Estimation o xed efects model using dummy variable regression

usually leads to high R2.


On the basis of the estimated xed effects computed from the models presented in

Table 4, the top and bottom provinces were identied and listed in Table 5. The xed

effects represent the characteristics that are unique to the provinces and hence, it may

be benecial to have a closer look at the best performers to identify why they were above

the rest; and also, to examine those that need improvement the most to identify the

characteristics that could be enhanced.

Table 5: Key Perormers in Selected Primary School Efciency Indicators

Best Perormers Needs Improvement

Dropout Rate Cohort Survival Rate Dropout Rate Cohort Survival Rate

Bataan 2nd District Bohol Basilan

Batangas 3rd District Iloilo Lanao del Sur

Davao del Sur 4th District Northern Samar Negros Occidental

Misamis Oriental Bulacan Quirino Sarangani

Mt. Province Rizal Sultan Kudarat Sulu

Note: In coming up with the list, provinces are ranked according to the computed xed efects.

Source: Authors computations using BEIS data.



31/69

As indicated above, the secondary cohort survival rate R2

0 1797=( ). has the lowestmodel t. This implies that even with the provincial effects that were used to represent

omitted variables that vary by province, there are still explanatory variables (not varying

by province) that are lacking in the secondary cohort survival rate model. A strong

possibility is that secondary-age children chose not to stay in school and work instead asshown in the model for individual outcomes (decision to attend school).

For secondary schools dropout rate, the signicant explanatory variable is median

household head educational attainment. An increase of 1 year in the median educational

attainment of the household head would result into a 5.9 percentage point reduction of

the dropout rate. Similarly, an increase of Pesos (P) 1,000 in the median provincial per

capita household income will increase the cohort survival rate by 2.3%. School resources,

represented by per pupil MOOE and PTR in the model, did not render signicant

coefcients. There are two possible explanations for this. One, the school resources vary

widely across school districts within a province, but these variations cannot be reected

in the provincial average that is used in the model, hence the relationship between

outcomes and school resources are not well estimated. Two, it is simply socioeconomic

characteristics that are more important in inuencing school education outcomes.

C. Quality o Education Outcomes

Contrary to their minimal inuence on school outcomes, per pupil MOOE and PTR have

a signicant impact on the quality of education outcomes based on the result of modeling

NAT scores.

For the secondary repetition rate, the per pupil MOOE is signicant but its sign is

counterintuitive. This is perhaps because per pupil MOOE only covers the public schools

that comprise only 79% of all secondary schools enrollment,and hence can only reect

the public schools situation.

Per pupil MOOE and PTR are both signicant determinants of primary NAT score. Ceteris

paribus, a 1% increase in per pupil MOOE translates to a 4.7% increase in the NAT

score, while a unit increase in the PTR results to a decrease of the NAT score by 1.18.

Note that the only budget school heads have a certain level of control over is MOOE.The

school MOOE is released to division ofces that can disburse it directly to the schools in

the form of cash advance. The schools can exercise exibility by realigning across the

MOOE items (e.g., participation in seminars/meetings and supplies) according to theiractual needs.Hence, in the model, per pupil MOOE can be viewed as the proxy indicator

for decentralization. On the other hand, the PS budget represented in the model by the

average teachers salary (the ratio of the budget for PS and the number of teachers) can

be taken as the proxy indicator for the status quo (no decentralization). That per pupil

MOOE is a signicant determinant for the primary NAT score while the average teachers

salary is not provides support to the potential of the continuing decentralization process. If



32/69

school heads are given the authority to determine and manage funds such as the MOOE

in accordance with their school development targets, then it can signicantly affect quality

of education outcome such as the NAT score.

In addition to the MOOE and PTR, the median provincial per capita income is also asignicant determinant of primary NAT score outcome. Assuming all variables stay at the

same level, an increase of P1,000 in the median income translates to an 18.3% increase

in the NAT score. On the other hand, the median household head educational attainment

is the signicant determinant of secondary school enrollment. A year increase in the

educational attainment results to an additional 1.14 to the NAT score.

Table 6: Quality of Education Production Functions

Education Inputs Education Outcomes

log(repetition rate) NAT Score

Primary Secondary Primary Secondary

log(per pupil MOOE) 0.06 0.40** 4.70** 2.73*

Pupilteacher ratio 0.01 0.01 (1.18)** (0.19)

log(teachers salary)a 0.02 (0.35) (1.43) (0.06)

Median household head educational

attainment

0.00 (0.04) 0.74 1.15**

Median provincial per capita income 0.00* 0.00 0.00** 0.00

Proportion o emales (0.71) (0.89) 1.47 0.42

2004 0.02 (0.13)** 2.30** 0.58

2007 0.02 0.00 1.01** 0.34


Test or heteroskedasticity 0.00 0.00 0.10 0.45

Adjusted R2 0.84 0.55 0.56 0.72


0.0 means magnitude is less than hal a unit.a Similar statistical models where the proxy variable or teachers salary was normalized as a proportion o provincial per capita

income were also estimated. Still at the 0.05 level, the variable is not statistically signicant.

Note: P-value is the probability o observing as extreme or more extreme value or the test statistic under the null hypothesis that

the parameter coecient or the variable under consideration is zero. Smaller p-values suggest statistical signicance.

The results above are based on the traditional view o xed efects models where the panel efects (in this case, provincial

efects) are treated as parameters to be estimated. Estimation o xed efects model using dummy variable regression

usually leads to high R2.


A large part of the variations of the quality of education outcomes is explained by the

provincial effects and therefore, could be useful to identify which of the provinces are

the best-performing and least performing. On the basis of consistency of belonging to

the top 10 (or bottom 10) highest provincial average NAT scores between 2003 to 2007,

the best performing provinces for primary schools are Bataan, Biliran, Cavite, Eastern

Samar, Ilocos Norte, Leyte, Romblon, Surigao del Norte, and Surigao del Sur. The least

performers are Basilan, Lanao del Sur, Maguindanao, Sulu, and Tawi-tawi. For secondary

schools, the best performing provinces are Agusan del Sur, Biliran, Eastern and Western

Samar, Northern Samar, Southern Leyte, and Surigao del Norte; the least performing



33/69

are Basilan, Cotabato City, Maguindanao, Sarangani, Sulu, Tawi-tawi, and Zamboanga

Sibugay. Notably, all are in Mindanao and most of them in the Autonomous Region of

Muslim Mindanao, the region with the largest number of out of school children in the

primary school age group (83,520 or 14.1% of children in that age group) and secondary

age group (78,888 or 21.5%).

Since the NAT scores for English, Science, and Math are highly correlated, SUR modeling

was applied,5 where almost similar observations as discussed above can be observed

(Table 7). Note that a unit increase in PTR tends to have a negative impact on primary

NAT scores on key subjects (English, Science, Math) while educational attainment of

household head seems to yield a positive impact on average secondary NAT scores.

Table 7: Seemingly Unrelated Regression (SUR) Models or NAT Scores on English,

Science, and Math

Education Inputs NAT Score

Primary Secondary

log(per pupil MOOE) 0.31 1.09

Pupilteacher ratio (0.30)** (0.04)

log(teachers salary) (3.87) 0.58

Median household head

educational attainment

0.25 0.63**

Median provincial per capita

income

0.00 (0.00)

Proportion o emales (11.38)** (2.77)

R2 (%) 0.87, 3.26, 2.49,

6.75,2.93, 4.42, 2.88,

0.88, 4.98, 1.80, 5.80,3.40, 4.49, 1.12, 2.10

(2.72), (2.37), (0.12), (3.71),

(3.34), (0.60), (2.75), (2.55),

(4.26), (2.74), (4.33), (2.10),2.18, (00.32), 1.52


0.0 means magnitude is less than hal o a unit.

Note: The system has 15 equations where the dependent variables are the scores on national achievement tests in language,

science, and mathematics rom 2003 to 2007. Each equation has a diferent intercept to allow or varying degrees o

diculty in each test.

P-value is the probability o observing an extreme or more extreme value or the test statistic under the null hypothesis


Sources: Authors computations using BEIS and APIS data.

5 Additional discussion is provided in the Statistical Models section.



34/69

IV. Policy Implications

Modeling the individual, school, and quality of education outcomes provided concrete

evidence on their key determinants. The PTR affects the individual outcomes for both age

groups and also has a direct effect on the NAT score at the primary level. Meanwhile,the per pupil MOOE is signicant in determining the NAT score at the primary level.

Socioeconomic characteristics (whether children were working, household income,

educational attainment of household head) proved to be the stronger determinants for all

types of education outcomes. Provincial effects are signicant for both school and quality

of education outcomes. This section discusses how these results affect policy.

A. Deployment o Teachers and Eective Class Size

The result of this study on the effect of PTR on the odds of attending school and pupil/

student learning outcome reinforces the theory that quality schools attract families and

encourage them to access available education services (Bray 2002, UNICEF-UNESCO

2006). On the other hand, parents commonly equate overcrowding with low-quality

education and are thus discouraged to send their children to overcrowded schools. Bray

(2002) also noted that teachers morale tends to erode as the class size grows. It is

therefore vital for the education system to recognize this relation and examine current the

teacher hiring and deployment system.

The average PTR at the national level is 33.64 for primary schools and 39.36 for

secondary schools, both of which are considerably lower than 50, which is the target

of the Philippine EFA plan. However, provincial-level PTR varies widely from a very low

11.5853.05 with a standard error of 6.88 for primary schools, and 10.6684.54 with astandard error of 7.98 for secondary schools (see Appendix Tables 5.1 and 5.2). These

ranges could be much wider if statistics are summarized at the district school level. These

summary statistics suggest that there is overcrowding in some areas like Maguindanao,

Rizal, and Lanao del Sur that may adversely affect individuals decisions to attend

school and their learning outcome (Figures 4 and 5). Overcrowding in schools tends to

put off families as it is recognized that for big classes, the teaching-learning quality is

compromised.



35/69

Figure 4: Distribution o PupilTeacher Ratios, Primary Education (pupils per teacher)

70

60

50

40

30

20

10

0

Maguindanao

Rizal

MaguindanaoRizal

Maguindanao

RizalRizal

Maguindanao

Rizal

Maguindano

Rizal

2002 2004 2005 20072003 2006

Note: The rectangular box in the graph represents the 25th (lower hinge) and the 75th

(upperhinge) percentile o the data or each year. The line that cuts through the

rectangle shows the median point. The dots show the outliers in the set, as wellas the minimum and maximum values.


Figure 5: Distribution o PupilTeacher Ratios, Secondary Education (pupils per teacher)

80

70

60

50

40

30

20

10

0

2002 2004 2005 20072003 2006

Lanao del SurRizalBoholTawitawiSultan Kudarat

Lanao del Sur

RizalBohol Rizal

Lanao del SurMaguindanaoRizal

Bulacan

Lanao del Sur

RizalLaguna

Lanao del Sur

Rizal

Note: The rectangular box in the graph represents the 25th (lower hinge) and the 75th

(upperhinge) percentile o the data or each year. The line that cuts through the

rectangle shows the median point. The dots show the outliers in the set, as well

as the minimum and maximum values.


The wide variation of PTRs across provinces suggests that the deployment of teachers

may not be equitable. One of the major impediments to rational distribution of teaching

assignments is Republic Act (RA) No. 4670 or the Magna Carta for Teachers of 1966,

which provides that teachers cannot be reassigned without their consent. The teachers



36/69

are thus protected from being transferred from one post to another based on whimsical

decisions from or abuse of power by school principals/heads and other higher ofcials.

However, when there is a real and urgent need for transfer arising from a shortage of

teachers in schools in other areas, RA 4670 can also be invoked. As early as 1999,

studies like the Philippine Education Sector Study (ADB and World Bank 1999, 60)concluded that the Magna Carta constrains the ability of local education authorities to

deploy teaching staff to meet local requirements and to redeploy teachers in response

to demographic shifts and to address teacher performance issues or for exposure and

training purposes.

Recognizing this limitation, the Medium-Term Philippine Development Plan 20042010

included, among its priority legislative agenda, the amendment of this law with the vision

to balance teachers rights and privileges with responsibility and accountability. This

includes the promotion of the general welfare of teachers such as provision of additional

compensation, sufcient hardship allowance, and salary increment as warranted by

special assignments.

At present, the Magna Carta provides for special hardship allowance for teachers in

areas where they are exposed to hardship such as difculty in commuting to the place

of work or other hazards peculiar to the place of employment (Section 19). It is also

provided that determining the areas considered to be difcult shall be the responsibility of

the DepEd Secretary. The hardship allowance shall be no less than 25% of the teachers

monthly salary. The allocation of the hardship allowance is determined and proposed by

division ofces and are provided in the Government Appropriations Acts under the lump

sum allowances of regional ofces. In cases where the allocation is insufcient, savings

from the DepEd eld ofces are tapped. The Department of Budget and Management

provided the updated Guidelines on the Grant of Special Hardship Allowance (NationalBudget Circular Number No. 514, 5 December 2007).6 However, these additional

allowances and any incentive such as additional hazard pay (from budget savings) do not

seem attractive enough for effective deployment of teachers.

On the other hand, most pending initiatives in the legislature, such as the Senate7, to

amend the Magna Carta are focused on strengthening the rights and benets of teachers,

and do not sufciently address the issue on demand-based equitable deployment.

Technical deliberations on these bills are progressing slowly while the government,

despite the provision in the Medium-Term Philippine Development Plan, does not seem

to be taking a stronger stand on the amendment owing to its potentially political nature.

6 The guidelines cover classroom teachers and heads/administrators assigned to hardship posts, multigrade teachers, mobileteachers, and nonormal education or alternative learning system (ALS) coordinators. Hardship posts are public schools or

community learning centers (in the case o ALS) located in areas characterized by transport inaccessibility and diculty o

situation (e.g., places declared calamitous, hazardous due to armed conict and extremely dangerous locations).7 For example, Senate Bill Nos. 72, 156, 166. In 2008, a technical working group in the Senate was convened to review the Magna

Carta, study the diferent bills seeking to amend it, and consider the other proposed legislations related to the welare and

benets o teachers. The technical working group, which invites representatives rom relevant government agencies, aims to

produce a consolidated bill that would address all the issues.



37/69

Any amendment to the Magna Carta should equally and sufciently address both the

deploym

Documents

Education Outcomes in the Philippines