DECEMBER 2021

DISCUSSION PAPER SERIES NO. 2021-27

If You Pay Peanuts, You Get Monkeys?

Education Spending and Schooling Quality
in the Philippines

Michael R.M. Abrigo

The PIDS Discussion Paper Series constitutes studies that are preliminary and subject to further revisions. They are being circulated in a limited number of copies only for
purposes of soliciting comments and suggestions for further refinements. The studies under the Series are unedited and unreviewed. The views and opinions expressed are
those of the author(s) and do not necessarily reflect those of the Institute. Not for quotation without permission from the author(s) and the Institute.

CONTACT US:
RESEARCH INFORMATION DEPARTMENT
Philippine Institute for Development Studies

18th Floor, Three Cyberpod Centris - North Tower publications@mail.pids.gov.ph

https://www.pids.gov.ph
EDSA corner Quezon Avenue, Quezon City, Philippines (+632) 8877-4000
If You Pay Peanuts, You Get Monkeys? Education Spending
and Schooling Quality in the Philippines

Michael R.M. Abrigo

PHILIPPINE INSTITUTE FOR DEVELOPMENT STUDIES

December 2021
Abstract

Cross-country comparison of education financing is often limited to public sector spending,

which only provides a partial view of the global education financing landscape. Using recent
National Transfer Accounts estimates of public and private education consumption for 74
economies around the world, we uncovered important trends in education financing in the
Philippines and benchmarked its performance against its peers. Based on a synthetic measure
of basic education consumption, we showed that education spending per person in the country
has grown robustly over the past 25 years, even surpassing the growth in per capita income.
Despite this feat, the Philippines’ education spending levels trail behind its regional and
aspirational peers, which contributes to its poor performance in international standardized
student assessments. While such is the case, there may still be opportunities to improve
schooling quality by identifying and scaling cost-effective education interventions that better
translate resource inputs to desired education outcomes.

Keywords: Education financing, National Transfer Account, Philippines

i
Table of Contents
1. Introduction .......................................................................................................... 1
2. How much are Filipinos spending on education? ................................................ 2
3. What drives education spending? ........................................................................ 9
4. Does spending level matter? ............................................................................. 10
5. Policy implications ............................................................................................. 13
References ............................................................................................................... 15
Annex A. Stochastic frontier estimation .................................................................... 18

List of Tables
Table 1. Education expenditure (PhP billion, 2012=100): Philippines, 2005-2019 ..... 3
Table 2. Public expenditure per student: Philippines and selected countries ............. 4

List of Figures
Figure 1. Education consumption per capita by age: Philippines, 1990-2015 ............ 5
Figure 2. Basic education consumption as share of labor income ............................. 7
Figure 3. Cumulative public and private basic education consumption ...................... 8
Figure 4. Government share in cumulative basic education consumption.................. 9
Figure 5. Correlates of cumulative basic education consumption ............................ 11
Figure 6. Education quality and cumulative basic education consumption ............... 12
Figure 7. Technical efficiency estimates................................................................... 14

ii
 If You Pay Peanuts, You Get Monkeys? Education
Spending and Schooling Quality in the Philippines
Michael R.M. Abrigo 1

1. Introduction

In 2015, the National Economic and Development Authority (NEDA) commissioned a national
survey to learn about Filipinos’ aspirations, values, and principles (NEDA, 2016). In this
Ambisyon Nation 2040 survey, practically all respondents who either have or intended to have
children expressed that the education of their (future) children is important to attain their
ambitions or plans in life. Ninety percent of these respondents were confident that their (future)
children will finish their studies. However, 86% of them also said that insufficient livelihood
or family income may be a possible challenge that could hinder them from achieving their
education dream.

Across the world, the public sector plays an important role in providing education services.
This may be warranted for at least two reasons. First, education is known to have many positive
benefits that spillover from individuals and households to the larger society, including
increasing incomes (e.g. Acemoglu and Angrist, 2000), lowering crime rates (e.g. Lochner and
Moretti, 2004), and breaking intergenerational poverty transmissions (e.g. Duarte, et. al.,
2017). 2 If left on their own, myopic households who generally do not internalize these positive
externalities to society and to future generations in their decision processes will typically
underinvest in education. Second, government concerns for equity may also drive public
spending on education. In particular, paternalistic governments may set minimum social
protection guarantees for primary goods, including education (c.f. Rawls, 1971; Gutmann,
1980), to which everyone is entitled and even compelled to consume.

That being said, both households and governments spend substantial amounts for education.
Recent basic education reforms in the Philippines, particularly through the passage of Republic
Act 10533 or the Enhanced Basic Education Act of 2013, together with expanding school-aged
population and household incomes, has contributed to greater aggregate education spending in
the country. In 2019, Filipino households spent a total of PhP753 billion on education
(Philippine Statistics Authority [PSA], 2021), while the national government spent PhP655
billion 3 (Department of Budget and Management [DBM], 2021). These values are more than
double the aggregate spending levels ten years prior.

Beyond similar within-country analyses, little is known of how the Philippines fare in
comparison with its regional and aspirational peers with regard education spending. The dearth
in comparable international education financing data has effectively hampered cross-country

1
Fellow II, Philippine Institute for Development Studies. The author is grateful for the excellent research
assistance by Ms. Zyra Eunice Diego. The paper has benefited from thoughtful discussions by seminar
participants at the 2021 PIDS Research Workshop series. All remaining errors are by the author.
2
See McMahon (2004) for a review.
3
This value excludes other education-related government spending, such as those for the conditional cash
transfer Pantawid Pamilyang Pilipino Program (4Ps) implemented by the Department of Social Welfare and
Development (DSWD). In 2019, DBM reports that the 4Ps spent PhP138.8 billion on program benefits and
administration.

1
appraisals and benchmarking, which would have been beneficial for policymaking. Several
countries have estimated National Education Accounts 4 (NEA) to complement their country’s
System of National Accounts (SNA). However, global education spending data are largely
confined to those based on public administrative records, which only provide a partial view of
the whole education sector financing. But even with public sector records, only as much as 60
percent of countries report data on total public education expenditures, and largely with
considerable time lag among reporting countries (UNESCO, 2016b).

In this study, we draw from recent National Transfer Account (NTA) estimates for the
Philippines and countries around the world to uncover and describe trends in global education
financing with particular focus on the Philippines as a case study. The NTA is a system of
accounts that measure how different generations produce, consume, and reallocate resources
in a manner that is consistent with the United Nations (UN) SNA (UN, 2013; Lee and Mason,
2011). Earlier studies (e.g. Abrigo, et. al., 2018; Mason and Lee, 2011; Tung, 2011) have used
NTA to characterize human capital investments, i.e., including both education and health, and
found wide variation in spending patterns across economies.

We focus on three critical questions on education financing. First, how much are countries,
particularly the Philippines, spending on education? Second, what drive the differences in
spending levels? Finally, does education spending levels matter? Providing evidence-based
answers to these questions may help guide future policy directions, and open new research on
education financing using alternative data sources, such as the NTA.

2. How much are Filipinos spending on education?

Over the last 15 years, total education expenditures in the Philippines increased by about 6.4%
annually, with a large part of the growth happening more recently. In 2019, the total current
education bill, i.e., excluding capital outlay, for all school levels reached almost PhP1.2 trillion
(in 2012 prices) from only PhP0.8 trillion in 2015 and PhP0.5 trillion in 2005 (Table 1). During
this period, households bear majority of the expense, reaching as high as 59.3% in 2005,
although increasing government expenditures contributed to the decline in the household share,
settling at 54.5% in 2019.

Education spending has been increasing as a share of the country’s gross domestic product
(GDP). By 2019, education spending in the country represent 7.5% of GDP, a considerable
increase compared with the recorded 5.8% share in 2005. Over the same period, public sector
spending on education increased from 2.1- to 3.1-percent of GDP 5 – an important feat, but still
behind the benchmark 4- to 6-percent by the Education 2030 Incheon Declaration.

The Philippine Constitution guarantees the highest budgetary priority to education. 6 In 2019,
actual education spending by the national government reached 18.1% of its total expenditures.
Programmed and proposed education spending for 2020 and 2021, on the other hand, are both

4
National Education Account is an accounting framework consistent with the UN system of National Accounts
that measures the financial flows in the education sector. In 2001, the Philippine National Statistical
Coordination Board, now part of the Philippine Statistics Authority, compiled a similar set of accounts, named
the National Education Expenditure Accounts, covering 1991 to 1998, but the series has since been discontinued
(United Nations Educational, Scientific and Cultural Organization [UNESCO], 2016a)
5
Including expenditures for capital outlay raises the shares to 2.5- to 3.4-percent in 2005 and 2019, respectively.
6
Article XIV, Section 5(5)

2
about 17% of the national budget. These figures are around the 15- to 20-percent thresholds
suggested by the Education 2030 Incheon Declaration for this metric.

Among government units, basic education expenditures by the Department of Education

(DepEd) comprise about 75- to 80-percent of the total current public education spending
between 2005 and 2019. Public higher education, represented by state universities and colleges
(SUCs) and the Commission on Higher Education (CHED), on the other hand, comprise about
11- to 17-percent. Public technical and vocational education through the Technical Education
and Skills Development Authority (TESDA) captures about 1- to 2-percent. Finally, local
governments, which fund different education and manpower training services, together with
museums, education institutions in other government offices, and other education-related
institutions, represent about 5- to 10-percent.

Table 1. Education expenditure (PhP billion, 2012=100): Philippines, 2005-2019

2005 2010 2015 2017 2019
Households 289.0 360.0 453.5 518.8 636.1
Government 198.5 254.1 326.7 352.8 531.1
National government 181.5 239.3 312.5 337.4 511.2
DepEd 150.5 203.7 261.9 267.7 410.5
TESDA 2.8 3.1 4.9 6.4 10.3
SUCs 24.4 26.8 33.9 44.2 48.3
CHED 1.1 1.7 5.8 17.1 28.2
Others 2.8 3.9 5.9 2.1 13.8
Local government 17.0 14.8 14.2 15.4 20.0
All financing agents 487.5 614.1 780.1 871.6 1,167.3
Source: PSA (2021), DBM (various years), DOF-BLGF (various years). Note: Values only include
current education expenditures, and exclude expenditures for capital outlay. DepEd – Department of
Education; TESDA – Technical Education and Skills Development Authority; SUCs – State
Universities and Colleges; CHED – Commission on Higher Education.

Compared with other countries in the Asia and the Pacific region, the Philippine’s public
spending per student in the primary and secondary education levels lag behind its regional and
aspirational peers (Table 2). While per student public spending appears to be strongly
correlated with per capita income, the Philippines spends only about 60- and 72-percent of
Indonesia’s per student public spending for primary and secondary levels, respectively, despite
the Philippines’ per capita income being 84% of Indonesia for the years presented. Singapore,
which leads the set of countries presented in Table 2, spends PPP$16,704 and PPP$20,632 per
primary- and secondary-level student, respectively. The Philippines, on the other hand, spends
PPP$813 and PPP$777 per student in the same respective education levels.

An important issue with relying solely on public expenditures to gauge education financing is
that it may not be representative of the total education costs faced by society in general, and by
students and their families in particular. It may be the case that government shares in total
education spending vary quite significantly across countries, which may affect the ordering of
country rankings in per student spending. Estimates based on NEAs presented in UNESCO
(2016c), for instance, show households’ share of aggregate education expenditures ranging
between 24- and 57-percent in their four-country case study. However, estimation of NEAs has
not gained much traction across governments for a number of reasons, including the technical

3
complexity of the estimation process and the limited supply of international and national NEA
expertise (UNESCO, 2016a).

Table 2. Public expenditure per student: Philippines and selected countries

Public expenditure per student
Per capita GDP
Country Year (2017 PPP$)
(2017 PPP$)
Primary Secondary
Singapore 2017 95,350 16,704 20,632
Brunei Darussalam 2016 60,867 5,401 14,392
Japan 2016 40,031 8,719 9,628
South Korea 2016 39,815 11,087 11,219
Malaysia 2017 26,662 4,302 6,024
Thailand 2013 15,767 3,676 2,838
Indonesia 2015 10,150 1,348 1,068
Philippines 2018 8,516 813 777
Lao PDR 2014 6,193 564 776
Myanmar 2018 5,029 393 518
Timor-Leste 2014 3,265 272 259
Source: World Bank [WB] (2021a). Note: Philippine expenditure per student estimates based on
author’s calculation using enrollment data (PSA, 2021), public education expenditure by level (DBM,
2020), and GDP in constant 2017 international (purchasing power parity, PPP) dollars (WB, 2021a).
Per student public spending for the rest of the countries are calculated based on WB (2021a) data.

An alternative to NEA is NTA. Similar to NEA, NTA is an accounting framework consistent

with the UN-SNA. Unlike NEA that is designed specifically as a satellite education account to
the SNA (UNESCO, 2016a), NTA is an elaboration of the SNA to capture economic flows
across age groups (UN, 2013). There are important conceptual differences between the two
frameworks. First, NTA records the value of consumption, i.e., expenditures less taxes net of
subsidies on products, while NEA includes taxes as part of expenditures. Second, the main
economic agents in NTA are the generations or age groups in an economy, while these are the
private and public sectors in the NEA. In NTA, the public and private sectors are only viewed
as mediators through which economic resources flow. Third, NEA has more elaborate
treatment of economic flows on education, while in NTA this is confined to age-specific public
and private consumption and the related resources financing the consumption. Fourth, although
it is possible to disaggregate NTA education consumption by levels of education, like in NEA,
this is seldom reported, and only aggregates by age are usually available. 7

A key advantage of NTA over NEA, however, is that NTA is more widely estimated, although
not particularly to compare cross-country education spending. In 2021, NTA has been
estimated for more than 90 countries, representing more than 80% of the world’s population.
In addition, modeling exercises has allowed the estimation of NTAs for the rest of the countries
around the world (e.g. Mason, et. al., 2017).

7
Estimates of age-specific NTA education consumption by education level are usually available to NTA country
team compilers as part of the NTA estimation process, but these are not part of the standard template reported
in the global database.

4
Figure 1 presents estimates of per capita NTA public and private education consumption in the
Philippines for age groups 5 to 24 years between 1990 and 2015. 8 These values capture age-
specific differences in school attendance and schooling costs, as well as the many different
factors that affect these variables. Public education consumption includes government
expenses on direct provision of education services in public institutions, excluding capital
outlays. Private education consumption, on the other hand, includes private-sector procured
education supplies, tuition and other fee payments, transportation costs related to education,
etc., regardless whether a student is enrolled in a public or private education institution. These
accounts also include non-schooling, but education-related consumption, i.e., general public
education, such as maintenance of museums and cultural centers, cost of books not specific to
a schooling level, and the like.

There are some notable patterns in per capita education consumption over the 25-year period
covered in Figure 1. First, per capita education consumption up to around age 16 hovers at
around PhP10,000 (in 2012 prices), although there is a large drop in 2005. Second, per capita
education consumption starts to increase rapidly after age 16 and peaks around age 20,
representing the relative costs of higher education compared with basic education, then tapers
off as students leave formal education. The peak appears to be shifting to the right across years.
Third, public consumption dominates in younger ages, i.e., in primary and secondary school-
ages, while private consumption dominates in later ages, i.e., in tertiary school-ages and
beyond. Fourth, public education consumption per capita has become flatter, potentially
showing increased school attendance rates in higher grade levels in more recent years.

Figure 1. Education consumption per capita by age: Philippines, 1990-2015

1990 1995 2000
5 10 15 20 25

5 10 15 20 25

5 10 15 20 25
Education consumption per capita
in constant '000 PhP (2012=100)
0

5 10 15 20 25 5 10 15 20 25 5 10 15 20 25

2005 2010 2015

5 10 15 20 25

5 10 15 20 25

5 10 15 20 25
0

5 10 15 20 25 5 10 15 20 25 5 10 15 20 25
Age
Private Public

Source: Abrigo, et. al. (2020)

8
For a discussion of the quinquennial Philippine NTA between 1990 and 2015, see Abrigo, et. al. (2020). For a
discussion of the NTA estimation methodology, please see the UN NTA Manual (UN, 2013).

5
Focusing on basic education, Figure 2 presents total education consumption up to age 16 as a
share of labor income. 9 While this measure may be rather unconventional 10 it has a particularly
attractive interpretation. The values presented in Figure 2 may be seen as the tax rate on labor
income that is needed to finance basic education if this is to be wholly financed by working.

Basic education consumption as a share of labor income varies widely among the 73 economies
represented in Figure 2. It ranges from less than one percent in 2010 Guinea Bissau to 15.7%
in 2010 Mexico. The Philippines spent 3.7% of labor income for basic education in 2015, only
a little ahead of Vietnam’s 3.5% in 2012. Other countries in the Asia and the Pacific region
spends significantly larger shares: Singapore (4.7%), Japan (6.0%), Thailand (8.3%), South
Korea (9.2%), Malaysia (10.1%), and Indonesia (10.5%).

The division between public and private spending also differ substantially across countries. In
Asia and the Pacific region, Vietnam and Cambodia stand out with less than 25% of their basic
education consumption financed by government. Other countries in the region rely more
heavily on government. In Thailand and Singapore, in particular, more than 80% of basic
education consumption are financed through government. Meanwhile, financing basic
education in the Philippines is roughly equally shared between the public and private sectors. 11

The substantial share of the private sector in education financing is quite surprising given the
dominant role of governments in providing basic education services as reflected in school
enrollment rates. In East Asia and the Pacific, only about a tenth of primary school pupils and
a fifth of secondary school students are enrolled in private schools (UNESCO, 2021). This may
suggest that complementary spending by the private sector, primarily by households, are
significant even with a large public education sector, which is reflected as private consumption
in the NTA. While education instruction may be financed by government, other important
education inputs, such as school supplies, may be financed privately.

The metric summarized in Figure 2 is based on the aggregate basic education consumption of
countries in a particular snapshot, and depends not only on spending per capita but also on the
distribution of school-age children. That is, the estimates are not really comparable since an
economy with more school-age population will appear to spend more as a share of labor income
compared with a comparable economy with the same education spending and labor income per
capita by age but with less school-age population.

Figure 3 presents an alternative indicator based on a synthetic cohort. The figure shows the
expected cumulative per capita public and private education consumption of an individual who
would have lived up to age 16 if faced with the particular schedule of per capita consumption
for the year of the NTA estimate. A higher (lower) value would indicate greater (lesser)
spending on basic education consumption over the entirety of the basic education cycle. The
values are converted into 2017 international dollars in purchasing power parity (PPP$) to
further facilitate comparison. 12

9
The upper age limit of 16 is chosen to roughly match the education cycle in the Philippines in 2015. This has
since been increased to 18, similar to the usual completion age for basic education in many other countries.
10
Usual cross-country measures include education spending per capita, as share of GDP, and as share of total
government spending.
11
Based on 2015 Philippine NTA estimates. More recent data suggest an increasing share of government.
12
A related measure is the education component in WB’s human capital index that measures the expected
number of years of schooling a child would complete by age 18 given the cross-sectional age pattern of school
enrollment adjusted for education quality, as measured by harmonized test scores (WB, 2021b).

6
Figure 2. Basic education consumption as share of labor income

IDN, 2012
MYS, 2009
KOR, 2016

THA, 2013

JPN, 2004

SGP, 2013

PHL, 2015

VNM, 2012

Private
Public

0 5 10 15
Education consumption as share of total labor income (%)

Source: Author’s calculations based on recent NTA data from www.ntaccounts.org. Number in labels
represent coverage year of NTA estimate. Labelled Asia and the Pacific countries are as follows: IDN
– Indonesia; JPN – Japan; MYS – Malaysia; KOR – South Korea; PHL – Philippines; SGP – Singapore;
THA – Thailand; and VNM – Vietnam.

7
Figure 3. Cumulative public and private basic education consumption

More private spending →

100
(in thousand 2017 PPP$, log scale)

SGP
30
Private education spending

VNM MYS
10

IDN

THA
3

KHM
PHL 2015
1

LAO

PHL 1990
0.1

More public spending →

0.01

0.01 0.1 1 3 10 30 100

Public education spending
(in thousand 2017 PPP$, log scale)

Source: Author’s calculations based on recent NTA data from www.ntaccounts.org. Note: Values are
the simple sum of per capita education consumption up to age 16. Number in labels represent coverage
year of NTA estimate. Labelled countries are as follows: IDN – Indonesia; KHM – Cambodia; MYS –
Malaysia; PHL – Philippines; SGP – Singapore; THA – Thailand; and VNM – Vietnam.

Over a span of 25 years, the Philippines’ cumulative basic education consumption almost
trebled, although much of the growth happened more recently. Between 1990 and 2010,
cumulative basic education consumption increased from PPP$3,390 to PPP$6,430, or about
3.3% per year. In the next five years, the measure would grow by 6.3% per annum to PPP$8,720
by 2015. It is noteworthy that these rates are significantly larger than the annual growth in per
capita GDP, which registered at 1.7% between 1990 and 2010 and 4.3% between 2010 and
2015. The Philippines’ private and public sectors spend about equal shares to finance basic
education over the course of this period.

Similar to insights from Figure 2, Cambodia and Vietnam still both rely more on the private
sector, unlike other countries in the region, using the synthetic cohort measure of basic
education consumption in Figure 3. But in addition, and more importantly, Figure 3 also shows
that public spending appears to not crowd-out private spending on basic education, with private
spending increasing by about 0.4% for every percent increase in public spending. 13 Further,
the figure also shows a higher tendency towards greater public share in basic education
financing as an economy develops.

This tendency may be better visualized in Figure 4, which shows a scatter plot of the share of
the public sector in per capita cumulative basic education consumption against per capita GDP.

13
This pertains to education financing only, which covers private spending in public school systems, among
others. Indeed, evidence from the Philippines suggest that expansion of the public education system crowds out
private school enrollment, at least at the secondary level (e.g. Jimenez and Sawada, 2001).

8
Figure 4. Government share in cumulative basic education consumption
100
80
Government share (%)
60

PHL 2015
40

PHL 1990
20
0

1 3 10 30 100
Per capita GDP (in thousand 2017 PPP$, log scale)

Source: Author’s calculations based on recent NTA data from www.ntaccounts.org, and per capita GDP
data from WB (2021a). Note: The government shares in cumulative basic education consumption are
based on the simple sum of per capita public and private education consumption up to age 16. PHL –
Philippines.

The government share in cumulative basic education consumption varies considerably among
lower-income countries, spanning almost the whole 0 to 100 percent. However, the span of the
distribution narrows and the average share increases with higher per capita GDP. Similar
observations of such “financing transition” have been observed, for example, in health
spending (e.g. Fan and Savedoff, 2014), in human capital investments (e.g. Abrigo, et. al.,
2018), and in consumption in general (Peacock and Wiseman, 1961).

3. What drives education spending?

Unless there is widespread child labor or capital markets allow loans to children, basic
education consumption is primarily financed through government tax-transfer systems or
through inter vivos transfers by older generations, especially family members. Motivations for
such inter-generational transfers have been well studied. 14 Earlier theoretical work were based
on parents deriving satisfaction from their children’s consumption (e.g. Becker, 1974; Barro,
1974), which were later expanded to allow other possible motivations for inter-generational
transfers, including for parents benefitting from investments on their children (e.g. Becker and
Murphy, 1988; Becker, 1991), among others.

Two important empirical studies in the Philippines exemplify these lines of thought. Estudillo,
et. al. (2001a, 2001b) documented sex-differentiated giving norms in agrarian settings, with

14
See Arrondel and Masson (2006), and Laferre and Wolff (2006) for a comprehensive discussion.

9
male children usually inheriting land while female children being compensated by greater
education investments. Yamauchi and Tiongco (2013), on the other hand, showed that greater
propensity of parents getting future financial support from their female children may be a
possible alternative driver of differential education investments on children.

Although it may not be possible to disentangle these competing motivations using estimates of
representative populations based on NTA, we show in Figure 5 that (log) cumulative basic
education consumption increases with (log) per capita income (Pearson’s correlation
coefficient, ρ=0.96), and decreases with (log) fertility rate (ρ=0.88). These observations are in
line with predictions of Becker’s (1960) classic child quality-quantity tradeoff theory, and
supported by household-level empirical evidences, such as in the Philippines (e.g. Orbeta,
2009; Yamauchi and Tiongco, 2013). It is interesting to note that public and private cumulative
basic education consumption separately have lower correlation coefficients when compared
against per capita income and fertility rate.

Between 1990 and 2015, the Philippines appear to be spending at about the cumulative basic
education consumption of the average country for the specified per capita income. However,
for its fertility rate the Philippines is spending significantly below the average country,
especially in more recent years. In 2015, for instance, the Philippines spends at the rate of only
about half of the expected cumulative basic education consumption of PPP$18,350 (standard
error = PPP$1,860) for a country with a total fertility rate of about 2.8 births per woman.

4. Does spending level matter?

Greater education spending does not automatically lead to better schooling outcomes. A series
of papers by Hanushek (1981, 1986, 1989), for instance, found no strong relationship between
expenditures and student achievements, although more recent studies suggest some positive
associations not only with schooling outcomes, but also in some later-life welfare measures
(e.g. Card and Kreuger, 1992; Gupta, et. al., 2002; Baldacci, et. al., 2003; Huang, et. al., 2021).
Other factors have been found to be also important in explaining schooling success, including
family background (e.g. Plug and Vijverberg, 2003) and rent-seeking in government (e.g.
Suryadarma, 2012), among others.

Figure 6 presents scatter plots of cumulative basic education consumption up to age 15 and
corresponding country-average scores in the 2018 Program for International Assessment
(PISA) by the Organization for Economic Co-operation (OECD, 2019). PISA is an
international standardized test that evaluates the performance of 15-year old students in
science, mathematics, and reading. The presented cumulative basic education consumption are
modelled estimates using per capita GDP, total fertility rate, and government share in final
consumption expenditure as predictors. This allows us to use all 77 economies in the 2018
PISA, instead of the 33 that have overlap with NTA country estimates. The results are
qualitatively similar using either the 33 or the 77 data points.

10
Figure 5. Correlates of cumulative basic education consumption
A. Per capita income
300
Cumulative education expenditure, age 3 to 16
(in thousand 2017 PPP$, log scale)
100
30
10

PHL 2015
3

PHL 1990
1

1 3 10 30 100
Per capita GDP (in thousand 2017 PPP$, log scale)

B. Fertility rate
300
Cumulative education expenditure, age 3 to 16
(in thousand 2017 PPP$, log scale)
100
30

PHL 2015
10
3

PHL 1990
1

1 3 5 7
Total fertility rate (births per woman)

Source: Author’s calculations based on recent NTA data from www.ntaccounts.org, and per capita GDP
and total fertility rate data from WB (2021a). Note: Cumulative basic education consumption values
are based on the simple sum of per capita public and private education consumption up to age 16. PHL
– Philippines.

11
Figure 6. Education quality and cumulative basic education consumption
A. Science 600
PISA 2018 average score
400 500

Philippines
300

0 100 200 300

Cumulative education expenditure, age 3 to 15
(in thousand 2017 PPP$)

B. Mathematics
600
PISA 2018 average score
400 500

Philippines
300

0 100 200 300

Cumulative education expenditure, age 3 to 15
(in thousand 2017 PPP$)

C. Reading
600
PISA 2018 average score
400 500

Philippines
300

0 100 200 300

Cumulative education expenditure, age 3 to 15
(in thousand 2017 PPP$)

Source: Author’s calculations based on recent NTA data from www.ntaccounts.org, and OECD (2019).
Note: Cumulative basic education consumption values are based on the simple sum of per capita public
and private education consumption up to age 15. Presented values are modelled estimates using per
capita GDP, total fertility rate and government share in consumption as predictors.

12
Similar to findings by Schleicher (2019), who used a different approach, Figure 6 shows that
average schooling quality increases with cumulative basic education consumption. The rate of
change is much larger at lower levels of spending, and gradually flattens with higher
expenditures. 15 This observation is true for the science, mathematics, and reading scores. While
this association may not be interpreted as causal, it is suggestive that greater resources may be
needed to raise schooling quality, especially in resource-poor settings.

The figure also shows that better schooling quality may be achieved using the same level of
resources. For example, Belarus, and Bosnia and Herzegovina both spend about PPP$50
thousand in cumulative education consumption up to age 15, but their average PISA scores
differ by about 70 points. This suggests that some education systems may be better at
converting education inputs into outputs.

We formalize this idea in Figure 7 that shows the estimated technical efficiency (TE) scores
among different education systems in the 2018 PISA. The TE scores are estimated using a
multi-output, in this case PISA scores by subject, stochastic frontier model (Aigner, et. al.,
1977; Meeusen and Van den Broeck, 1977) with cumulative education consumption up to age
15 as the only input. 16 The TE scores compares the distance of a country from a theoretical
frontier based on a specified education production function. A TE score of 100% indicates that
a country is at the frontier, while lower TE scores imply some inefficiency in converting inputs
to outputs relative to the production possibility frontier. We provide two sets of estimates,
wherein we exclude China, represented by Beijing, Shanghai, Jiangsu and Zhejiang, from the
estimation sample in one set to assess robustness against outliers.

TE scores vary widely among the 77 PISA countries. With China in the estimation sample, TE
scores range between 56.6% and 93.9%. Excluding China, on the other hand, leads to upward
revisions in TE scores across countries, with TE scores now ranging between 75.0% and
99.4%. The Philippines scores behind other East and South-east Asia countries included in the
2018 PISA, although its TE score in the no-China set is at a more acceptable value of 86.5%.
If the Philippines have fully maximized the use of its inputs it should have scored closer to 400
points in each of the test subjects, instead of the 340 to 360 observed range.

5. Policy implications

The analyses we have presented highlights important trends in Philippine education financing.
First, using a synthetic measure of basic education consumption, we have uncovered that
education spending per person in the country has grown robustly over the past 25 years, even
growing faster than recorded GDP per person over the same period. Second, the private sector
contributes about as much as government to finance basic education. Third, while the
Philippines is spending at the average level for its per capita GDP, it is spending substantially
less for its level of fertility. Fourth, despite the robust growth in spending per person, its basic
education spending level still lags behind its regional and aspirational peers, which contributes
to its poor performance in international standardized student assessments. Finally, while the
Philippines lag behind in education spending per capita there may be opportunities to improve
schooling quality by optimizing the translation of inputs to outputs.

15
Using the 33 countries with NTA data overlap, the fitted line is positively sloped until about PPP$80 thousand,
then flattens thereafter. The output elasticities are around 0.12 for each of the test subjects.
16
See methodological annex for details.

13
Figure 7. Technical efficiency estimates

CHN

JPN

SGP

KOR

IDN
MYS

THA

PHL

0 20 40 60 80 100
Technical efficiency (%)

With China Excluding China

Source: Author’s calculations based on recent NTA data from www.ntaccounts.org, and OECD (2019).
Note: Cumulative basic education consumption values are based on the simple sum of per capita public
and private education consumption up to age 15. Presented values are modelled estimates using per
capita GDP, total fertility rate and government share in consumption as predictors. The technical
efficiency scores are estimated using a multi-output stochastic frontier model with total cumulative
education consumption up to age 15 as the only input.

Although increasing resources available for education may be desirable to improve schooling
quality, our cross-country comparison shows that this may be difficult to achieve as per capita
education spending is intimately linked with an economy’s particular economic development
and fertility levels. Raising education spending per capita may therefore entail more than
rallying resources for the education sector, but also ensuring that robust economic opportunities

14
are available to improve average household incomes, as well as assisting households to achieve
their desired fertility levels.

Poor schooling quality need not be the necessary and only outcome of subpar education
spending levels. A more important and arguably more urgent challenge for government is to
identify and to scale cost-effective education interventions that better translate resource inputs
to desired education outcomes.

References

Abrigo, M.R.M., R.H. Racelis, J.M. Ian Salas, A.N. Herrin, D.A.P. Ortiz, and Z.C. Tam (2020).
Are we missing out on the demographic dividend? Trends and prospects. Research Paper
Series No. 2020-02. Quezon City, Philippines: Philippine Institute for Development
Studies.

Abrigo, M.R.M., S.-H. Lee, and D. Park (2018). Human capital spending, inequality, and
growth in middle-income Asia. Emerging Markets Finance and Trade, 54, 1285-1303.

Acemoglu, D., and J. Angrist (2000). How large are human-capital externalities? Evidence
from compulsory schooling laws. NBER Macroeconomics Annual, 15, 9-59.

Baldacci, E., M.T. Guin-Siu, and L. De Mello (2003). More on the effectiveness of public
spending on health care and education: A covariance structure model. Journal of
International Development, 15(6), 709-725.

Card, D. and A.B. Kreuger (1992). Does school quality matter? Returns to education and the
characteristics of public schools in the United States. Journal of Political Economy,
100(1), 1-40.

Department of Budget and Management [Philippines] (2021). Budget of expenditures and

sources of financing. Manila, Philippines: DBM.

Duarte, R., S. Ferrando-Latorre, and J.A. Molina (2017). How to escape poverty through
education? Intergenerational evidence in Spain. Applied Economics Letters, 25(9), 624-
627.

Fan, V.Y., and W.D. Savedoff (2014). The health financing transition: A conceptual framework
and empirical evidence. Social Science and Medicine, 105, 112-121.

Gupta, S., M. Verhoeven, and E.R. Tiongson (2002). The effectiveness of government
spending on education and health care in developing and transition economies. European
Journal of Political Economy, 18(4), 717-737.

Gutmann, A. (1980). Children, paternalism, and education: A liberal argument. Philosophy and
Public Affairs, 9(4), 338-358.

Hanushek, E.A. (1981). Throwing moneys at schools. Journal of Policy Analysis and
Management, 1(1), 19-41.

Hanushek, E.A. (1986). The economics of schooling: Production and efficiency in public
schools. Journal of Economic Literature, 24(3), 1141-1177.

15
Hanushek, E.A. (1989). The impact of differential expenditures on school performance.
Educational Researcher, 18(4), 45-51+62.

Huang, X., S. Huang, and A. Shui (2021). Government spending and international income
mobility: Evidence from China. Journal of Economic Behavior and Organization, 191,
387-414.

Jimenez, E., and Y. Sawada (2001). Public for private: The relationship between public and
private school enrollment in the Philippines. Economics of Education Review, 20(4), 389-
399.

Lee, R. and A. Mason (Eds.) (2011). Population aging and the generational economy: A global
perspective. Cheltenham, UK and Northampton, MA, and Ottawa, Canada: Edward Elgar
Publishing, and International Development Research Centre.

Lochner, L., and E. Morretti (2004). The effect of education on crime: Evidence from prison
inmates, arrests, and self-reports. American Economic Review, 94(1), 155-189.

Mason, A., and R. Lee (2011). Population aging and the generational economy: Key findings.
In R. Lee and A. Mason, eds. Population aging and the generational economy: A global
perspective. Cheltenham, UK and Northampton, MA, and Ottawa, Canada: Edward Elgar
Publishing, and International Development Research Centre.

Mason, A., R. Lee, M.R.M. Abrigo, and S.H. Lee (2017). Support ratios and demographic
dividends: Estimates for the world. Technical Paper No. 2017/1. New York: United
Nations, Department of Economic Affairs, Population Division.

McMahon, W.W. (2004). The social and external benefits of education. In G. Johnes and J.
Johnes, eds. International Handbook on the Economics of Education. Cheltenham, UK,
and Northampton, MA: Edward Elgar Publishing.

National Economic and Development Authority [Philippines] (2016). Ambisyon Natin 2040: A
long-term vision for the Philippines. Mandaluyong, Philippines: NEDA.

Philippine Statistics Authority (2021). Household final consumption expenditure (2000-2020).

Online database. Accessed on 30 June 2021 from https://psa.gov.ph/national-accounts/
base-2018/data-series

Plug, E., and W. Vijverberg (2003). Schooling, family background, and adoption: Is it nature
or is it nurture? Journal of Political Economy, 111(3), 611-641.

Rawls, J. (1971). A theory of justice. Cambridge, MA: Belknap Press.

Schleicher, A. (2019). PISA 2018: Insights and interpretations. Paris, France: OECD.

Suryadaram, D. (2012). How corruption diminishes the effectiveness of public spending on

education in Indonesia. Bulletin of Indonesian Economic Studies, 48(1), 85-100.

Tung, A.C. (2011). Consumption over the lifecycle: An international comparison. In R. Lee
and A. Mason, eds. Population aging and the generational economy: A global
perspective. Cheltenham, UK and Northampton, MA, and Ottawa, Canada: Edward Elgar
Publishing, and International Development Research Centre.

16
United Nations Educational, Scientific and Cultural Organization (2016a). Methodology of the
National Expenditure Accounts. Montreal, Canada: UNESCO Institute for Statistics.

United Nations Educational, Scientific and Cultural Organization (2016b). Global education
monitoring report 2016. Paris, France: UNESCO.

United Nations Educational, Scientific and Cultural Organization (2016c). Who pays for what
in education? The real costs revealed through National Education Accounts. Montreal,
Canada: UNESCO Institute for Statistics.

United Nations Educational, Scientific and Cultural Organization (2021). Data for the
Sustainable Development Goals. Online database. Montreal, Canada: UNESCO Institute
for Statistics.

World Bank (2021a). World Development Indicators. Online database. Accessed on 24

September 2021 from https://databank.worldbank.org/source/world-development-
indicators

World Bank (2021b). The human capital index 2020 update: Human capital in the time of
COVID-19. Washington, D.C.: The World Bank.

17
Annex A. Stochastic frontier estimation

Stochastic frontier models (Aigner, et. al., 1977; Meeusen and Van den Broeck, 1977) are used
to estimate and analyze technical (in)efficiency based on firm production theory. A firm, in
this case countries, are assumed to produce a set of outputs, in this case PISA test scores, using
a common technology given some production inputs, in this case cumulative basic education
consumption. A firm is perfectly efficient when it produces the maximum possible output for
the input it uses, i.e. when it is at the production possibility frontier.

We estimate a multi-output, single input stochastic frontier model using the following
linearized production function

ln(𝑦𝑦𝑖𝑖 ) = ln(𝑐𝑐𝑖𝑖 ) ⋅ 𝛽𝛽 + 𝛾𝛾 + (𝑣𝑣𝑖𝑖 − 𝑒𝑒𝑖𝑖 )

where 𝑦𝑦𝑖𝑖 is the column vector of PISA scores by test subject for country 𝑖𝑖 = 1,2, … ,77, and 𝑐𝑐𝑖𝑖
is its cumulative basic education consumption up to age 15, a scalar. The conformable vector
𝛽𝛽 are production elasticities that relate a unit of input 𝑐𝑐𝑖𝑖 to output 𝑦𝑦𝑖𝑖 . The vector 𝛾𝛾 is a subject-
specific shifter that captures differences in production technology across test subjects. The
model residual is composed of two parts. The country- and test-specific vector 𝑣𝑣𝑖𝑖 is assumed
to follow a normal distribution with zero mean and variance 𝜎𝜎𝑣𝑣2 , and captures idiosyncratic
shocks on production. In the general case, the country-specific vector 𝑒𝑒𝑖𝑖 is assumed to follow
some non-negative random distribution, which when exponentiated ranges between zero and
one and provides an estimate of production efficiency, with a value of one indicating
production at the frontier. In our estimation, we assume 𝑒𝑒𝑖𝑖 to be distributed as half-normal
𝑁𝑁 + (𝜇𝜇𝑣𝑣 , 𝜎𝜎𝑣𝑣2 ), and common across test subjects for a given country.

There is significant under-coverage in countries with both PISA scores and NTA estimates. In
order to expand our sample, we modelled cumulative basic education consumption up to age
15 using per capita GDP, total fertility rate, and government share in final consumption
expenditure as predictors. This allows us to use all 77 economies in the 2018 PISA, instead of
the 33 that have overlap with NTA country estimates. The results are qualitatively similar using
either the 33 or the 77 data points. Estimates are available from the author upon request.

We estimated two sets of models based on the above specification. We include all countries in
our baseline model. In a second model, we remove data for China to assess the robustness of
our estimates to outliers. As shown in Figure 7, the statistical ordering of technical efficiency
scores across countries although magnitudes differ between specifications. Table A1 provides
a summary of the parameters of our estimated stochastic frontier models.

We also did an alternative stochastic frontier model using corrected ordinary least squares
(COLS), where we did not assume any parametric distribution for 𝑒𝑒𝑖𝑖 . Instead, we estimated a
country fixed-effects model, with the estimated fixed-effects 𝑢𝑢𝑖𝑖∗ split into two components:
𝑒𝑒𝑖𝑖 ≥ 0 capturing technical efficiency, and 𝑢𝑢𝑖𝑖 capturing the contribution of other country-
specific test-invariant characteristics

ln(𝑦𝑦𝑖𝑖 ) = ln(𝑐𝑐𝑖𝑖 ) ⋅ 𝛽𝛽 + 𝛾𝛾 + (𝑣𝑣𝑖𝑖 + 𝑢𝑢𝑖𝑖∗ )

𝑢𝑢𝑖𝑖∗ = 𝑢𝑢𝑖𝑖 − 𝑒𝑒𝑖𝑖
𝑒𝑒𝑖𝑖 = max(𝑢𝑢𝑖𝑖∗ ) − 𝑢𝑢𝑖𝑖∗ ≥ 0

18
The results are qualitatively the same as the models presented in Table A1 and in Figure 7,
although magnitudes differ. In COLS, the most efficient firm is assigned a technical efficiency
of 100% following the applied correction above.

Table A1. Stochastic frontier model: PISA test scores

Model 1 Model 2
ln(ci) 0.116 *** 0.115 ***
(0.013) (0.011)
Test-subject
Math (= 1) -0.061 ** -0.062 **
(0.027) (0.027)
Reading (= 1) -0.045 * -0.041
(0.027) (0.027)
Test-subject x ln(ci)
Math (= 1) x ln(ci) 0.014 ** 0.015 **
(0.006) (0.006)
Reading (= 1) x ln(ci) 0.008 0.007
(0.006) (0.006)
Constant 6.001 *** 5.719 ***
(1.247) (0.044)

𝜎𝜎𝑣𝑣2 0.001 0.001

(0.000) (0.000)
𝜎𝜎𝑒𝑒2 0.006 0.010
(0.001) (0.005)
𝜇𝜇𝑒𝑒 0.375 0.044
(1.246) (0.059)

With China Yes No

N 222 219
BIC -696 -713
Source: Author’s calculations based on recent NTA data from www.ntaccounts.org, and OECD (2019).
Note: Cumulative basic education consumption values, ci, are based on the simple sum of per capita
public and private education consumption up to age 15 modelled using per capita GDP, total fertility
rate and government share in consumption as predictors. N – Number of observations; BIC –
Bayesian information criterion

19