Buku Institusi

Institution, Human Capital and Development
in Indonesia
Edited by:
Gumilang Aryo Sahadewo
Evi Noor Afifah
Eny Sulistyaningrum
Devanto S. Pratomo
Alin Halimatussadiah
Budy P. Resosudarmo
IRSA Book Series On Regional Development No. 20

Institution, Human Capital and Development in Indonesia
Penyunting :
Gumilang Aryo Sahadewo,
Evi Noor Afifah,
Eny Sulistyaningrum,
Devanto S. Pratomo,
Alin Halimatussadiah, and
Budy P. Resosudarmo
Desain sampul :
Arief A. Yusuf
Tata letak :
Deni Lesmana
Proofreader :
Lilies Achmadi
Diterbitkan oleh :
Perkumpulan Ilmu Regional Indonesia (IRSA)
Sekretariat IRSA, Jl. Cimandiri No. 6 Bandung 40115
West Java-Indonesia
Email : info@irsa.or.id
ISBN : 978-623-94282-4-2
Cetakan pertama, Juni 2022
Hak Cipta dilindungi Undang-Undang

TABLE OF CONTENTS
CONTRIBUTORS.................................................................................................. v
FOREWORD........................................................................................................... x
Chapter 1 Institutional and Human Capital Aspects of Development in
Indonesia
Gumilang Aryo Sahadewo, Evi Noor Afifah, Eny Sulistyaningrum,
Devanto S. Pratomo, Alin Halimatussadiah, and Budy P.
Resosudarmo................................................................................ 1
PART I. INSTITUTION
Chapter 2 Productivity Impact of Political Dynamics in A Decentralised
Indonesia: The Case of the 2016 Centralisation Policy of Batam
Free Trade Zone*
Bangkit A. Wiryawan..........................................................................15
Chapter 3 Rent Seeking Undercover: Corruption and Public Spending
Composition in Indonesia
Jesita Wida Ajani, I Gede Sthitaprajna Virananda, and Ghifari
Ramadhan Firman................................................................................39
Chapter 4 Political Competition and Deforestation: Evidence from
Indonesia
Harnum Yulia Sari, Raka Rizky Fadilla, and Zihaul Abdi.................65
PART II. HUMAN CAPITAL
Chapter 5 The Role of Socioeconomic Factors in Voting Preferences in
Indonesia’s Local Electionsa
Dharra Mahardhika and Ekki Syamsulhakim......................................89
Chapter 6 Heritability of Cognitive Ability and Its Relation to Economic
Achievement in Adolescence
Siti Munawaroh and Ni Made Sukartini........................................111
iii
Chapter 7 The Effect of Educational Attainment on Female Labour Force
Participation in Indonesia
Rizka Isnaini Husna and Eny Sulistyaningrum...............................131
PART III. DEVELOPMENT
Chapter 8 Forestry Agrarian Reform in Balumpewa Village, Central
Sulawesi Province
Kasmiati, Mohamad Shohibuddin, Ganies Oktaviana, and Ahmad
Jaetuloh...............................................................................................151
Chapter 9 Elements of Maritime Logistics Transaction Costs in Islands
Province of North Maluku, Eastern Part of Indonesia
Chairullah Amin, Heti Mulyati, Eva Anggraini, Tridoyo Kusumastanto,
and D.S. Priyarsono...........................................................................177
Chapter 10 Institutional Access and Poverty of Capture Fisheries Households
in Indonesia
Silviana Taniu, Dyah Wulan Sari, and Dias Satria..........................201
iv Institution, Human Capital and Development in Indonesia

CONTRIBUTORS
Ahmad JAETULOH
Sajogyo Institute, Bogor, Indonesia
E-mail: Jaetulohahmad@gmail.com
Alin HALIMATUSSADIAH
Department of Economics, Faculty of Economics and Business
University of Indonesia, Depok, Indonesia
E-mail: alin.halimah@gmail.com
Bangkit A. WIRYAWAN
Department of Politics and Government
Diponegoro University, Semarang, Indonesia
E-mail: bwiryawan@lecturer.undip.ac.id
Budy P. RESOSUDARMO
Arndt-Corden Department of Economics
Australian National University, Canberra, Australia
E-mail: budy.resosudarmo@anu.edu.au
Chairullah AMIN
Khairun University, Ternate, Indonesia
E-mail: chairulamin79@gmail.com
D.S. PRIYARSONO
Department of Economics, Faculty of Economics and Management
Bogor Agricultural University, Bogor, Indonesia
E-mail: priyarsono@apps.ipb.ac.id
v
Devanto S. PRATOMO
Brawijaya University, Malang, Indonesia
E-mail: devanto@ub.ac.id
Dharra MAHARDIKA
Department of Economics, Faculty of Economics and Business,
Padjadjaran University, Bandung, Indonesia
E-mail: dharra.mahardhika@gmail.com
Diah Wulan SARI

Universitas Airlangga, Surabaya, Indonesia
E-mail: dyah-wulansari@feb.unair.ac.id
Dias SATRIA
Brawijaya University, Malang, Indonesia
E-mail: dias.satria@ub.ac.id
Ekki SYAMSULHAKIM
Padjadjaran University, Bandung, Indonesia
E-mail: ekki.syamsulhakim@fe.unpad.ac.id
Eny SULISTYANINGRUM
Universitas Gadjah Mada, Yogyakarta, Indonesia
E-mail: eny@ugm.ac.id
Eva ANGGRAINI
Department of Resource and Environmental Economics
E-mail: evaanggraini@apps.ipb.ac.id
vi Institution, Human Capital and Development in Indonesia

Evi Noor AFIFAH
E-mail: evinoor.afifah@ugm.ac.id
Ganies OKTAVIANA
Sajogyo Institute, Bogor, Indonesia
E-mail: ganies@sajogyo-institute.org
Ghifari Ramadhan FIRMAN

Harris School of Public Policy
University of Chicago, Chicago, USA
E-mail: ghifari.r@mail.ugm.ac.id
Gumilang Aryo SAHADEWO

E-mail: gasahadewo@ugm.ac.id
Harnum Yulia SARI

E-mail: harnumyuliasari@gmail.com
Heti MULYATI
Department of Management, Faculty of Economics and Management
IPB University, Bogor, Indonesia
E-mail: heti@apps.ipb.ac.id
I Gede Sthitaprajna VIRANANDA

E-mail: i.gede69@ui.ac.id
vii
Jesita Wida AJANI
Universita Gadjah Mada, Yogyakarta, Indonesia
E-mail: jesita.wida.ajani@mail.ugm.ac.id
KASMIATI
Faculty of Agriculture and Forestry
Universitas Sulawesi Barat, Majene, Indonesia
Email: kasmiatis35@gmail.com
Mohamad SHOHIBUDDIN
Faculty of Human Ecology
E-mail: m-shohib@apps.ipb.ac.id
Ni Made SUKARTINI
Universitas Airlangga, Surabaya, Indonesia.
E-mail: ni-made-s@feb.unair.ac.id
Raka Rizky FADILLA

E-mail: raka.rizky@ui.ac.id
Rizka Isnaini HUSNA

E-mail: rizka.husna@gmail.com
Silviana TANIU
Department of Development Studies, Faculty of Economics
University of Gorontalo, Gorontalo, Indonesia
E-mail: silvianayunus@gmail.com
viii Institution, Human Capital and Development in Indonesia

Siti MUNAWAROH
Universitas Airlangga, Surabaya, Indonesia.
E-mail: siti.munawaroh-2017@feb.unair.ac.id
Tridoyo KUSUMASTANTO
Department of Resource and Environmental Economics
E-mail: prof.kusumastanto@gmail.com
Zihaul ABDI
E-mail: abdizihaul@gmail.com
ix
FOREWORD
Indonesian government aspires to escape from the middle-income trap

before 2045, marking 100 years of the country’s independence. However,
looking at her economic growth performance of the last two decades (of
slightly above 5% per year), this target is likely hard to achieve. To guarantee
the economic growth necessary to escape the middle-income trap, we need a
sustained high economic growth—a high growth that can be maintained for
a lengthy period of time.
We cannot achieve this growth by relying on exploitation of natural
resources but through improving technical knowhow. As the literature on
deep determinant of growth suggests, countries that have been successful
in improving technical knowhow usually those having institutions with
excellent qualities. This idea of institutions as the root cause of economic
growth is not new but often forgotten.
This book titled “Institution, Human Capital and Development in
Indonesia” contains recent research on the regional aspects of the interlinkage
between institutions, human capital in the context of Indonesian economic
development. Each chapter of the book was selected from the annual
conference of Indonesian Regional Science Association (IRSA) held in
Yogyakarta in 2021.
Publication of this book is possible because of excellent materials from
the conference papers selected, followed by a rigorous editorial process by a
team of respected Indonesian regional scientists together with all the chapter
authors. Without their hard work, the publication of this book will not be
possible.
x Institution, Human Capital and Development in Indonesia

Hence, this book is important for two reasons. First, its topic is a
reminder of the importance of institutions in economic development in
Indonesia. Second, it is also a reminder that producing high quality of book
publication of international standards is possible with a collective effort of
like-minded individuals.
Regards,
Arief A. Yusuf
President, Indonesian Regional Science Association
Bandung, June 2022
xi
xii Institution, Human Capital and Development in Indonesia
Chapter 1
Institutional and Human Capital Aspects

of Development in Indonesia
Gumilang Aryo Sahadewo, Evi Noor Afifah, Eny Sulistyaningrum,
Devanto S. Pratomo, Alin Halimatussadiah, and Budy P. Resosudarmo
INTRODUCTION
Initial policies to supress the outbreaks of the COVID-19 pandemic were
mainly strict restriction of human movements and social distances (Baldwin
and Evenett, 2020). These policies, although have been compensated with
fiscal expansion policies, negatively affected the economies around the
world. The global economy was contracted in 2020. Vaccines, fortunately,
were invented approximately a year after the start of the pandemic and
were able to be widely distributed since then. This situation allowed many
governments to relax restriction on human movements and social distances.
The economy, hence, has been able to bounce back since 2021.
Such a situation also happened in Indonesia when the economy
experienced a contraction in 2020 (Lewis and Witoelar 2021; Resosudarmo
et al. 2021). Nevertheless, a year later, as vaccines have been relatively
widely distributed and government’s expansion fiscal policy have been fully
implemented, the economy bounced back (Halimatussadiah et al., 2022).
Figure 1 shows growths of gross domestic product (GDP) per capita, as a
welfare measure of the country, during several years before the COVID-19
till 2021. It can be seen that several years before the pandemic, the country’s
welfare has been growing with a rate of almost 5 percent. This performance
has been relative better than many developing countries. Furthermore,
1
Figure 1 also shows that welfare in all regions in Indonesia has been growing
as fast as the levels before the pandemic.
Figure 1. Growth GDP per Capita (%)
20
%
15
10 Sumatra
Java-Bali
Kalimantan
5
Sulawesi
Eastern Indonesia
0 Indonesia
2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021
-5
-10
Source: BPS via CEIC.

During the first year of COVID-19, the consequence of the contraction
of the economy, in average at the national level, people welfare was decline
by approximately 3 percent. Welfare contractions in Eastern Indonesia and
Sulawesi were higher than the national average. The spread of the virus and
this contraction would certainly have some short- and long-term impact,
particularly on human development (Resosudarmo and Irhamni 2021).
In 2021, as the economy bounced back, welfare of the country has been
bounced back again almost as high as the condition before the pandemic.
Confidence, in general, has been built among Indonesians in such that
they aim to recovery better. Not only, that the country expects to be able to
accelerate the growth of its welfare, but also to transform its economy into a
green economy (LPEM-FEB-UI and BAPPENAS 2020; Kurmala 2022).
Since early 2022, the President, for several times, has been instructing all
government agencies to develop strategies to achieve these goals of high rate
of welfare improvement and transformation to a green economy. Several
experts, nevertheless, argue that, to achieve these goals, Indonesia needs
to improve the qualities of institutions and human capitals (OECD, 2021;
Mulyani and Kuncoro, 2021).
This book is intended to discuss several issues related to institution,
human capital and development in Indonesia. Particular attention, whenever
2 Institution, Human Capital and Development in Indonesia

possible, will be on regional perspectives of these issues. This chapter will
outlay several general issues of institution and human capital issues in
Indonesia. Last part of this chapter will provide summaries of other chapters
presented in this book.
INSTITUTION IN INDONESIA
The importance of institution quality of economic growth has been argued
in the literature for at least since 1990s. Examples would be the works by
Barro (1996), by Perotti (1996), and by Alesina et al. (1996). Since mid 2000s,
the most prominent scholars arguing this relationship are Acemoglu and
Robinson through many of their seminal works (Acemoglu and Robinson,
2008 and 2012; Acemoglu, et al., 2019). They show both theoretically and
empirically that better quality of institution would produce higher economic
growth in a country. This section, hence, will provide slight picture on the
quality of Institution in Indonesia.
In general, it can be confirmed that the quality of institutions in Indonesia
has been improving. However, it can also be argued that these institutions
have not yet having quality as good as those in developed countries. One
indicator of institutional quality that is publicly available is the World
Governance Indicators (WGI) produced by the World Bank. Figure 2
provides information on for several countries published by the World Bank.
WGI is an index consisting of several measures of institutional quality
in a country. These measures are (1) voice and accountability, (2) political
stability and absence of violence/terrorism, (3) government effectiveness,
(4) regulatory quality, (5) rule of law, and (6) control of corruption. Figure 2
shows that Indonesia’s WGIs from 2000 till 2020 have been improving. It was
-0.9 in 2000 and became 0.1 in in 2020, which is a major progress.
Institutional and Human Capital Aspects of 3

Development in Indonesia
Figure 2. World Governance Indicator
(0,50) 0,00 0,50 1,00 1,50 2,00
Australia
Brazil
Germany
India
Indonesia
Mexico
Netherlands
Philippines
United Kingdom
United States
2000 2010 2020
Source : http://info.worldbank.org/governance/wgi/
Note that the World Governance Indicator consists of six dimensions of
governance, which are (1) voice and accountability, (2) political stability and
absence of violence/terrorism, (3) government effectiveness, (4) regulatory
quality, (5) rule of law and (6) control of corruption. The worse situation would
be -2.5 and the best would be 2.5.
Comparing to several developed countries, however, the quality of
institution in Indonesia is still relatively low. WGIs for Australia, Germany,
Netherlands, and UK are all above 1. Though institutional quality in India, as
another large democratic country in Asia, has been indicated to be declined
in the last 10 years or so, still it is better than Indonesia. The key question on
how Indonesia could improve the performance as its institutions is, hence,
a valid issue.
To better understand in what aspects of institutional quality that
Indonesia is lagging, we could compare Indonesia and Germany in each
aspect of governance indicators. Figure 3 shows these comparisons.

Figure 3. Gaps between Indonesia and Germany in Dimensions of World
Governance Indicator
Control of Corruption
Rule of Law
Regulatory Quality
Government Effectiveness
Political Stability and Absence of
Violence/Terrorism
Voice and Accountability
0 0,5 1 1,5 2 2,5 3 3,5 4
2000 2010 2020
Source: http://info.worldbank.org/governance/wgi/
As indicated in Figure 3, while gaps of all governance dimensions
between Indonesia and Germany have been steadily narrowing from
2000 to 2020, Indonesia has been progressing in all issues of governance—
particularly in terms of its institutional quality that has been improving.
Large progresses have been taking place with regard to the issues of political
stability, absence of violence/terrorism, and government effectiveness. The
two key issues—which Indonesia particularly needs to pay attention—are
control of corruption and rule of law. This fact indicates that corruption is
still rampant in Indonesia and enforcement of laws is relatively weak.
Since early 2000s, Indonesia has conducted activities to reduce
corruption in the country. In 2002, Indonesia set up a special anti-corruption
committee, known as Komisi Pemberantasan Korupsi (KPK). This committee
has so far managed to bring many important corruption cases to the court
that punished those involved in such crime. However, cases of corruption
have been argued not to significantly decline. One argument is that law
enforcements have not been that strong so far. Indonesia’s efforts so far
might not be sufficient to resolve the issues of corruption and enforcement
of laws. Better measures to combat corruption and to better enforce the laws
are needed. Should it be successful, Indonesia could expect having a higher
economic growth rate accordingly.

HUMAN CAPITAL IN INDONESIA
Investment in human resources through education is considered as one
of the most effective ways for countries to increase their national welfare
and reduce poverty in the long term. A good quality of human capital will
result in high productivity, which will in turn affect growth and economy
positively. The superior human resources can be generated from educational
institutions. Undoubtedly, promoting educational attainment can increase
the average standard of living and contribute to the reduction of absolute
poverty by enabling individuals to generate income and access better paid
jobs.
Indonesia government has made this issue of improving human capital
as one of the national priorities through ensuring an increase in national
budget allocation annually. It is necessary to the government through
budgetary policies in the education sector to provide funding in increasing
the number of people who are able to complete high school and tertiary
education.
Although the universal education programme has been increased
into 12 years since 2015, Indonesia is still far from achieving the objective
of universalisation of twelve-year compulsory basic education. In fact, as
shown by National Labour Force Survey (SAKERNAS), the Indonesian
workforce is still dominated by those with lower level of education.
Figure 4. Average School Attainment in 2018: Percentage of Population 25
Years and Older Graduated from Junior High School
120
100
80
60
40
20
0
Source: World Development Indicator

In general, almost 60% of the total Indonesian labour force have an
education equivalent to junior high school or below, while only 43% of the
labour force have an upper secondary and tertiary education in 2018 (Figure

4). However, the youthful age groups (20-29 years old) showed a significant
positive trend of improving education, providing a dominant share of upper
secondary and tertiary education.
While the Gross Enrolment Rate for lower secondary education is
almost universal in Indonesia (Suharti 2013), the transition from junior
secondary school to senior secondary school needs greater consideration.
Based on the 2019 SUSENAS data, almost 30% of 16-18 years children still
excluded from the school, because of the poverty reasons (need to work
to support family income). Among the enrolled students, 61% of children
from the richest households reach grade 12, but only 23% of children from
the poorest households reach that level. The other reasons include early
marriage, disability, and remoteness (World Bank, 2020). Poverty remains
an important determinant in school enrolment, and poor households often
have to drop out their children after lower secondary education in order to
add the family income.
In terms of education, regional disparities also remain wide, which is
reflected by the lack of upper secondary schools in some remote regions,
particularly outside Java. Although there are upper secondary schools (both
academic and vocational streams) in each district (Kabupaten) in Indonesia,
it is now likely to require building schools at sub-district level (Kecamatan).
World Bank (Al-Samarrai and Cerdan-Infantes 2013) noted that more than
30% of students should access four or more kilometres to attend the school
at upper secondary level. Otherwise, relevant programmes that should be
taken into consideration include educational expenses other than tuition
fees, such as subsidizing transportation, which can encourage poor students
in rural or remote areas to attend school.
Regarding the quality of education, given the low cognitive score of
Indonesian students, represented by low PISA and TIIMS scores, global
evidence showed that incorporating socio-emotional skills and higher-order
cognitive skills into the curriculum is also the best learned when the students
actively work with more observation and practice (World Bank 2020). World
Bank Survey in 2008 also showed that two thirds of employers complained
about finding qualifies graduates for professional and management position
(World Bank 2020). Recent policy changes for secondary school shows a
right direction where the national exam will be terminated and changed into
assessment of minimum competency, combining literacy, numeracy, and
character education.

Re-skilling and up-skilling opportunities have immensely positive
impacts on Indonesian labours, given the difficulties faced by young and
unskilled workers in the labour market with the objective to substitute
and in some cases to supplement formal education with the context of
lifelong learning. In some occupations and industries, having a technical
qualification or certification and keeping skills up to date can be critical for
improving competitiveness. Therefore, providing jobs reskilling—which
enables individuals to learn new skills—is needed. Midcareer reskilling will
become important more important as the skills mix needed for a successful
career change. Recently, countering the impact of the COVID-19 pandemic
on employment, the Indonesian government has successfully provided
subsidised training to employers through Kartu Prakerja which is not only
as a safety net but also upgrading the skills of affected workers.
INSTITUTION, HUMAN CAPITAL, AND

DEVELOPMENT ISSUES RAISED IN THIS BOOK
This book is organised into three parts: Part 1, Part 2, and Part 3, which
cover current issues on institution, human capital, and development in
Indonesia, respectively. Part 1 consists of three chapters—i.e., Chapter
2, 3, and 4—that discuss issues of centralisation, corruption and political
party polarisation. In particular, Chapter 2 written by Bangkit Wiryawan
discusses productivity impact of the 2016 centralisation policy of the Batam
Free Trade Zone. According to Wiryawan, the policy is mainly directed to
improve the business climate in the zone that was deemed underperformed
due to prolonged central-local conflict. This chapter uses survey data from
the Statistics of Indonesia covering the years of 2015 and 2017. Difference-in-
difference estimator is employed to measure the impact with firms located
in neighbouring provinces being set as the control group. The main finding
of this chapter is that the policy has led to a productivity loss at around 32%,
with similar large drop in terms of growth. It is further found that locational
choice, investment origin, as well as export orientation determine their
resiliency against this policy change.
Meanwhile, Chapter 3 raises an issue on corruption and public
spending composition in Indonesia. The authors—Jesita Wida Ajani, I
Gede Sthitaprajna Virananda, and Ghifari Ramadhan Firman—argue that
government spending presents opportunities for public officials to commit
corruption and extract rents. In Indonesia, public officials in decentralised
local governments might have the incentive to allocate more of the local

government budget towards spending categories that are more accessible
for corruption. Therefore, the authors assumed that corruption levels are
associated with spending composition. They evaluate this hypothesis using
a novel dataset of corruption cases from the Indonesian Supreme Court’s
legal records. They show that corruption cases per capita at the provincial
level has a positive correlation with the province’s budget allocation for
social protection programmes and spending on health, i.e., the provincial
governments’ social and health programmes might be prone to rent-
seeking behaviour of bureaucrats. These results suggest that adverse hidden
intentions still play a leading role in formulating annual budgeting decisions
across provinces in Indonesia.
In Chapter 4 the authors—Harnum Yulia Sari, Raka Rizky Fadilla,
and Zihaul Abdi—investigate the relationship between local political
competition and deforestation in Indonesia. They further establish a new
database of local parliamentary seats configuration and deforestation that
covers 241 districts/cities in Indonesia ranging from 2009 to 2018. Political
competition is measured using the Political Party Fractionalisation Index
and Political Party Polarisation Index. Their analysis shows a negative
relationship between deforestation and fractionalisation, and, on the other
hand, a positive relationship between deforestation with polarisation. They
further argue that there is no significant association between political party
ideological fractionalisation and/or polarisation of political parties and
deforestation. This finding further confirms that the political competition at
the local level matters for deforestation but the political party ideology in
Indonesia differs little on matters of policy, especially on deforestation.
Meanwhile, the second part of this book—i.e., Part 2—is related to human
capital issues and their socio-economic consequences. This part consists of
3 chapters—Chapter 5, 6, and 7. The first chapter in this part, Chapter 5,
observes the role of socio-economic factors in voting preferences during
Indonesia’s local head elections. Dharra Mahardika and Ekki Syamsulhakim
argue that religion and gender of candidates can be two key factors in
receiving votes. This chapter aims to investigate further and measure the
influence of voters’ socioeconomic characteristics on considering candidate’s
religion and gender as an important attribute in making their decisions
during local head elections in Indonesia. They use the 2007 and 2014 IFLS
data comprising about 30,000 pooled individuals, and Moslem voters tend
to consider the religion and gender of the candidates, compared to the non-
Moslem for their heads of district or their mayors of city.

Chapter 6, which is written by Siti Munawaroh and Ni Made Sukartini,
puts forward issues on heritability of cognitive ability and its relationship to
economic achievement in adolescence. This chapter uses microdata from the
Indonesia Family Life Survey (IFLS) of 2000 and 2014. Based on a sample of
603 individuals analysed in this chapter, it reveals that, first, better cognitive
abilities in children are inherited from their parents, especially mothers.
Intelligent daughters and sons correlate with educated mothers. Second,
socioeconomic status, such as educational expenses and location of residence
and birth order are also associated with children’s cognitive abilities. Lastly,
cognitive scores in their youth have a strong positive correlation with
educational outcomes and wages during adolescence.
The last chapter in Part 2—i.e., Chapter 7—discusses the effect of
education on female labour force participation (FLFP) in Indonesia. In this
chapter, the authors –i.e., Rizka Isnaini Husna and Eny Sulistyaningrum—
aim to examine the correlation between educational attainment, represented
by years of schooling, and FLFP is tested using a probit model over the
Indonesian Family Life Survey (IFLS) data for 2000, 2007, and 2014. This
chapter confirms the U-shape relationship between educational attainment
and FLFP. The relationship shows that FLFP increases after female education
attainment reaching the turning point of 7 years of schooling. Evidence
suggests that policies which increase women’s access to higher education
and support working mothers to go back to school could encourage higher
FLFP.
Furthermore, Part 3 that consists of Chapter 8, 9, and 10 presents several
issues of regional development in Indonesia. Chapter 8, written by Kasmiati,
Mohamad Shohibuddin, Ganies Oktaviana, and Ahmad Jaetuloh, presents a
discussion on the implementation of forestry agrarian reform in Balumpewa
Village, Sigi District, Central Sulawesi Province. This chapter discuss the
difficulties in implementing forestry agrarian reform in Balumpewa Village,
even though the local government of Sigi district has set up a special task
force to conduct the reform. Among the 168,000 hectares set up in 2017 to be
reformed, only 3,000 hectares by 2020 have been up approved to be distributed
to local communities. Given the various difficulties in conducting this reform,
this chapter argues that implementation of agrarian reform requires a special
law particularly for this issue. Conventional regulations could discourage
effective participations of local communities in this forestry agrarian reform.

In Chapter 9 Chairullah Amin, Heti Mulyati, Eva Anggraini, Tridoyo
Kusumastanto, and D.S. Priyarsono, present a discussion on maritime logistics
transaction cost. Using the case of the islands province of North Maluku
and adopting the prospective analysis method, this chapter measures factors
affecting high transaction costs in Eastern parts of Indonesia’s maritime
logistics system. Among the factors analysed are port infrastructure, port
services availability, accessibilities, loading uncertainty, island economies of
scale, weather, regulations, organisational governance, employee behaviour,
maritime logistics costs, and traffic volume throughput. The results suggest
that port infrastructure, cargo uncertainty, weather, human behaviour,
economic scale of island, and regulation have the most decisive influence
on maritime logistics’ transaction costs. For example, an increase in the
island capacity-based economic activity could reduce maritime logistics
transaction costs, which will then positively impact an economic growth of
the archipelagic region of Indonesia.
Furthermore, Chapter 10 written by Silviana Taniu, Diah Wulan Sari,
and Dias Satria discusses institutional access and poverty among captured
fisheries households in Indonesia. Poverty in this chapter is measured
using (1) monetary indicator only and (2) multidimensional indicators.
The household survey on agricultural and captured fisheries household
in 2013 by the Statistics Indonesia is utilised to measure those two poverty
indicators. In general, the results show institutional access—such as access to
credit, utilisation of cooperatives, membership of fishermen’s groups (KUB),
access to production facilities, and market access—negatively correlates
with being both monetarily and multidimensional poor. Meanwhile, among
several fishery grant programmes, only the fishing gear grant is significantly
correlated with monetary poverty status.
REFERENCE
Acemoglu, D. and J. Robinson. 2008. The Role of Institutions in Growth and
Development. Commission on Growth and Development Working
Paper No. 10. World Bank, Washington, DC. https://openknowledge.
worldbank.org/handle/10986/28045 License: CC BY 3.0 IGO.
Acemoglu, D. and J. Robinson. 2012. Why Nations Fail: Origins of Power,
Poverty and Prosperity. New York: Crown Publishers.
Acemoglu, D., S. Naidu, P. Restrepo, and J. Robinson. 2019. Democracy Does
Cause Growth. Journal of Political Economy, 127(1), pp. 47-100.

Alesina, A., S. Özler, N. Roubini, and P. Swagel. 1996. Political Instability
and Economic Growth. Journal of Economic Growth, 1(2): 189–211. http://
www.jstor.org/stable/40215915
Al-Samarrai, S. and P. Cerdan-Infantes. 2013. “Where did all the money
go? Financing basic education in Indonesia”. D. Suryadarma and G.W.
Jones (eds.), Education in Indonesia, Singapore: ISEAS Publishing, 109-
138. https://doi.org/10.1355/9789814459877-007
Baldwin, R. and S.J. Evenett. 2020. COVID-19 and Trade Polcy: Why Turning
Inward Won’t Work. London: CEPR Press.
Barro, R.J. 1996. Democracy and Growth. Journal of Economic Growth, 1(1):
1-27.
Halimatussadiah, A., R. Edwards, F.R. Moeis, and R.F. Maulia. 2022.
Agriculture, Development and Sustainability in the Covid-19 Era.
Bulletin of Indonesian Economic Studies, 58:1, 1-30
Indrawati, S.M. and A. Kuncoro. 2021. Improving Competitiveness through
Vocational and Higher Education: Indonesia’s Vision for Human Capital
Development in 2019–2024. Bulletin of Indonesian Economic Studies, 57:1,
29-59.
Kurmala, A. 2022. Green Economy for Job Creation. Antara, 12/02/2022.
https://en.antaranews.com/news/214909/green-economy-for-job-
creation.
Lewis, B. and F. Witoelar (Eds.). 2021. Economic Dimensions of COVID-19 in
Indonesia: Responding to the Crisis. Singapore: ISEAS.
LPEM-FEB-UI and BAPPENAS. 2020. Thinking Ahead: Indonesia’s Agenda
on Sustainable Recovery from COVID-19 Pandemic. Jakarta: LPEM-
FEB-UI.
Organisation for Economic Co-operation and Development (OECD). 2021.
OECD Economic Surveys: Indonesia. Economic Snapshot, Paris: OECD.
Perotti, R. 1996. Growth, Income Distribution, and Democracy: What the
Data Say. Journal of Economic Growth, 1(2): 149–187. http://www.jstor.
org/stable/40215914
Resosudarmo, B.P. and M. Irhamni. 2021. “Consequences of the COVID-19
Pandemic on Human Capital Development”. B. Lewis and F. Witoelar
(Eds.). Economic Dimensions of COVID-19 in Indonesia: Responding to the

Crisis. Singapore: ISEAS, 170-189.
Resosudarmo, B.P., T. Mulyaningsih, D.S. Priyarsono, D. Pratomo and
A. Yusuf (Eds.). 2021. Regional Perspectives of COVID-19 in Indonesia.
Bandung: IRSA Press.
Suharti. 2013. “Trends in Education in Indonesia”. D. Suryadarma and G.W.
Jones (eds.), Education in Indonesia, Singapore: ISEAS Publishing, 15-52.
https://doi.org/10.1355/9789814459877-007
World Bank. 2020. The Promise of Education in Indonesia. Jakarta: The World
Bank and Australian Government.

Chapter 2
Productivity Impact of Political

Dynamics in A Decentralised Indonesia:
The Case of the 2016 Centralisation
Policy of Batam Free Trade Zone*
Bangkit A. Wiryawan
INTRODUCTION
During his presidential visit to Singapore in July 2015, President Joko Widodo
attempted to lure more investment to the neighbouring island of Batam, the
capital of Riau Island province. The island hosts Indonesia’s first successful
Free Trade Zone (FTZ). It showed stellar performance in the first half of the
1990s with the majority of investments coming from Singapore (Damuri et
al. 2015; Negara and Hutchinson 2020; Wulandari 2012) . In the meeting with
Singaporean business leaders during the visit, he laid out a plan to restructure
the FTZ governance in order to better improve its business climate. The
purpose of this effort is two folds. First, he tried to resolve the long-lasting
conflict between the central and local governments in the zone which was
detrimental towards industrialisation attempt in the region (Jati 2016).
Second, considering the regional context the measure was seemingly done to
anticipate a major upcoming change, i.e., the implementation of the ASEAN
Economic Community (AEC) at the end of that year. AEC could potentially
lessen the policy incentive provided in the zone. The utilisation of placed-
based policies such as the FTZ to boost industrial output has been a growing
practice in many parts of the world (Farole & Akinci 2011). Its instrumentation
in the Asian Tiger economies, and particularly in promoting export-oriented
* This manuscript is part of author’s dissertation research and the findings have been presented at
the JASID Tokai Conference for Young Researcher, Nagoya 21 August 2021.
15
activities, has managed to sustain unprecedented high growth spanning for
more than two decades. The remarkable success of China’s industrialisation
from the 1980s to the 2010s was also largely contributed by their location-
based policy. China has been successfully harnessed foreign investment by
establishing Special Economic Zones in Shenzhen, Zhuhai, Shantou, and
Xiamen in the early days of the open-door policy. The success has led to
the larger experimentation of this approach, leading to the open coastal and
open cities policy between 1988 and 1992 (Ge 1999).
This chapter aims to estimate the impact of vertical institutional change
in the management of Batam FTZ on firms’ labour productivity. One of the
latest changes, following the president’s aforementioned visit to Singapore,
was started in 2015 where the central government attempted to revitalise
the zone through centralising its administration. This move was finally
formalised in February 2016 with the Presidential Decree (Keppres) no. 8.
DEVELOPMENT OF BATAM FTZ

The island of Batam has not been originally developed as a free trade zone.
Despite its strategic location in the busy Malacca Strait and shortly across the
rapidly developing Singapore (see Figure 1), before 1970 its main economic
activities used to be largely fishery and agriculture (BP Batam, 4 February
2022). The island was lacking sufficient infrastructure and was sparsely
populated that prevented the local industry to develop. Back then, instead
of focusing their attention on the island, the government had prioritised the
relocation of the province’s capital from Tanjung Pinang island, just next to
Batam, to the main Sumatra island in the city of Pekanbaru in 1960 (BP Batam,
2021). The abundance of natural resources was one of the main reasons for
this administrative change.

Figure 1. Map of Batam FTZ (Insert: Map of Southeast Asia)
Source: databasin.org
The initiative to make Batam a free trade zone began to emerge in October
1971 with Presidential Decree No. 74. It was then followed by another decree
in 1973 that assigned Batam Island Industrial Area Development Authority
(BIDA) as the agency responsible for the zone’s development (BP Batam
2021). This placed the zone under Jakarta’s direct influence with BIDA serves
as de facto government of the island. This is in contrast to most other areas in
Indonesia where regional and local governments are the ones responsible for
the development in their respected jurisdiction, even when they were merely
an extended arm of the central government. President Suharto’s Minister of
Research and Technology, BJ Habibie, was appointed as the head of BIDA,
in the hope to propagate not just industry but also promoting industrial
upgrading (Hutchinson 2017). However, for nearly two decades Batam still
encountered difficulties in attracting investments, which soon had changed
with the establishment of a growth triangle plan better known as SIJORI.
In 1989, the Indonesian government was one part of the tripartite SIJORI
Growth Triangle plan. SIJORI stands for Singapore, Johor, and Riau, where
the latter two regions are the state/province of Malaysia and Indonesia,
respectively. The triangle comprised a development plan with Singapore
at the center and both Johor and Riau as its periphery areas. The plan was
announced by then Deputy Prime Minister of Singapore Goh Cok Tong and
was responded with much enthusiasm by all parties involved.
Following the announcement of the SIJORI plan, massive investment
flocked into Batam FTZ. Between 1991 and 1997 total foreign investment was
Productivity Impact of Political Dynamics in A Decentralised Indonesia: 17

The Case of the 2016 Centralisation Policy of Batam Free Trade Zone
USD535.5 million, averaging at nearly USD70 million per year (BKPM 2020).
This was also followed by increasing domestic investment at a similar rate
of growth, suggesting an increasing number of joint-venture projects and
economic spillovers to the island. As a result, GDP growth reached an all-
time high at 18% on average between those years.
Nonetheless, the unanticipated Asian Financial Crisis in 1997 had caused a
massive problem for Indonesia’s industrialisation as it was one of the worst-
hit countries. It took around four years to recover its GDP loss. In the Batam
FTZ, however, the picture was a little bit contrasting. Despite GDP growth
was down sharply to 3% in 1998, FDI increased more than twofolds from
USD 196.47 million in 1997 to USD 514.11 million in 1998. In the two years
that follow, Batam FTZ enjoyed close to USD 200 million of investment on
a yearly average. This resilience was also observed by Hutchinson (2017) as
he studied the electronic industries in the zone. However, following massive
institutional change in 2000, the high investment period has not returned to
the island, at least not until 2007 when the government revitalised the FTZ
approach.
Decentralisation Reform and Revitalisation of Batam FTZ

The 1997 Asian crisis was followed by nationwide unrest that led to Indonesia’s
economic and political reform by May the next year. This ended the heavily
centralised regime of President Suharto. The new administration in Jakarta,
led by President BJ Habibie, subsequently responded to the demand for a
stronger local government and hastily applied the decentralisation bill in
1999. Batam FTZ has not been immune to this change. Almost overnight, a
newly autonomous local government of the city of Batam was established.
Prior to this, there has been already a local government that was established in
1983, but with limited autonomy, and its developmental role was practically
overtaken by the FTZ authority1.
Following the decentralisation zeitgeist, Batam’s Riau Province was split
separating the archipelagic part of the region from the Sumatra mainland
and creating the new Riau Islands Province. It consists of seven districts,
with Batam serves as the main industrial center as well as its most populous
areas. The emergence of autonomous local governments both at the province
as well as the city level complicates the business climate in the FTZ. There
1 The establishment of the City Government of Batam in 1983 was part of the deconcentration
programme under Law no. 5 in 1974 that allowed for larger local government autonomy. However,
the programme was merely focusing on administrative matters instead of providing real autonomy,
as local policies were still dictated by the central government in Jakarta.

were overlapping responsibilities between the city government and the FTZ
authority (BIDA) such as in facilitating investment and development of
infrastructure. This dual authority problem has been analysed by Damuri et
al. (2015), Aritenang (2017), as well as Zaenuddin et al. (2017), among others.
At the same time with the bottom-up decentralisation change, as part
of the attempt to restore macroeconomic stability following the crisis, and
as part of IMF’s Structural Adjustment Programme, the government of
Indonesia was required to abandon its industrial policies. In relation to the
FTZ, the first regulation issued was the suspension of value-added tax and
luxury goods tax exemptions on the island in March 1998. It was not until
mid-2000 that the government reverted to the old regulation, which was
later renewed in December 2003.
The political dynamics that follow the bottom-up (decentralisation) and
the top down (the IMF Structural Adjustment Program) institutional change
in Batam had not immediately brought down business climate in the zone,
as suggested by Broadfoot (2003). Annual growth was maintained at nearly
7% until 2003 (see Figure 2). However, investment flow was affected sharply
as it was slashed at just a third sequentially from 2001 to 2003, with central-
local friction serves as one of the underlying causes (Damuri et al. 2015). The
government tried to prevent this spiraling down further.
Figure 2. Investment and Growth Dynamics in the Batam FTZ
Source : BKPM, 2021

Note : (1) Black line marks bi-yearly average FDI inflow (log current USD mil.)
pegged at the left axis; (2) Yellow bar denotes bi-yearly average GDP
growth pegged at the right axis; and, (3) BOD is the Board of Director of the
zone.

The road to revitalising the FTZ strategy in Batam took place in 2006,
with leaders of Indonesia and Singapore initiating a joint ministerial meeting
on economic cooperation (Wong and Ng 2009). One of the focuses was to
develop Batam and its surrounding areas. This was then followed by the
establishment of a national team in Indonesia, led by the Coordinating
Minister for Economic Affairs, to study several policy options. The bilateral
agreement with Singapore was an obvious demonstration of Jakarta’s
international influence that was not possessed by the local government.
Figure 3. Batam FTZ Structure under Different Periods of Government
Source: peraturan.bpk.go.id, 2021
In the year after, the central government in Jakarta issued the long-
awaited FTZ Law (44/2007) which was accompanied by a presidential
decree (Perpres) effective in 2008. Under the new regulations, both Governor
of Riau Island and the Mayor of Batam served as the chair and vice-chair of
the board, respectively (see Figure 3). The membership of the board under
the regulation has also consisted of local officials. The dynamics had swung
the zone’s political pendulum from a centralised to a decentralised one.
The revitalisation plan dramatically increased FDI flow by 45% whilst
also sustaining GDP growth at around 7% between 2008 and 2013 as shown
in Figure 2. It was only interrupted in 2009 due to the global financial crisis.
These numbers suggest that the decentralised policy works, accommodating
local government’s role in the FTZ administration helped in improving
climate change which eventually results in higher investment and growth.
However, as we can see in Figure 2, nearing the implementation of AEC in
2015 growth begins to falter.

RESEARCH GAP AND PROBLEMS
President Jokowi’s visit to Singapore in mid-2015 was met with enthusiasm
and concern from the country’s business chamber (sbf.org.sg, 28 July 2015).
One of the often-expressed concerns, as implied in his speech in front of
150 Singaporean businessmen, was related to the dual-authority situation
in Batam. The FTZ revitalisation strategy in 2007 which decentralized the
zone management, was not considered competitive enough to attract
larger investment.Despite increasing total FDI inflow to the zone, the scale
of investment is getting smaller and smaller. During its glory days in the
1990s foreign investment in Batam averaged above USD5 million per project.
However, by 2014 under the new FTZ policy (Perpres No. 9 in 2008) average
FDI per project was only USD2.1 – 2.3 million (BKPM, 2021). This could be
a contributing factor that caused the declining growth trend between 2010
and 2015 as shown in Figure 2. To overcome this situation the government
decided to go in one direction, by centralising the FTZ administration. The
FTZ authority, now known as BP Batam (Free Zone Authority of Batam),
is placed under the central government’s Coordinating Ministry for the
Economic Affairs. The zone is then headed by one of its deputy ministers,
Mr. Lukita Diansyah Tuwo. This change, however, did not manage to
improve growth on the island. In 2016 and 2017, GDP growth plummeted
to less than 4%, below national average. FDI per project coming to the zone
was also down to below USD1 million per project (BKPM 2021), lowest in its
history. Various literatures have studied the competitiveness of Batam FTZ
(Wahyuni et al. 2010), its impact on FDI and industrialisation in the region,
or on the dual authority issue that occured due to the policy (Damuri et al.
2015; Zaenuddin et al. 2017). This study is among the first to focus itself in
the recent institutional change in Batam with the 2016 policy. Centralising
the zone under Jakarta’s command, in a largely decentralised state structure,
is a large-scale experiment worth of investigating.
DATA AND ESTIMATION STRATEGY

The research design follows quasi-experimental method that is very popular
in policy evaluation studies (Besstremyannaya 2013; Card, 1992; Card and
Krueger 1994; Duflo 2001) including in the case of Batam FTZ development
(Aritenang and Chandramidi 2020). Hence, the study is aimed to calculate
the difference in the expected firms’ output and productivity in the treatment
group against firms’ output and productivity in the selected control group to
measure the impact of the policy change. The treatment group, comprising

firms in Batam FTZ, is measured using firms in the rest of Riau Islands
Province as proxy. Province level data are readily available and much easier
to set up for the purpose of this study. Furthermore, using the province data
as a proxy is justified, as Negara and Hutchinson (2020) described that the
city of Batam contributed 84% of the number of firms to the province based
on BPS survey.
Data Description
The main data source was taken from Statistics of Indonesia’s Large and
Medium Industrial survey that is conducted yearly. However, there are some
gap years in accessing the data, leaving with only odd years data selected.
In choosing the data to be used, first the identification of the recent policy
change in Batam’s FTZ and its parent province Riau Islands is conducted.
From there the data in between its implementation were selected, since
the targeted policy intervention was the 2016 FTZ centralisation policy. It
was then followed by selecting the industrial data of 2015 and 2017 for the
estimation.

Table 1. Descriptive Statistics
Variables All Control Treat
Obs. Mean s.d. Obs. Mean s.d. Obs. Mean s.d.
Before Policy Intervention
Dependent var.
Output (log current IDR mil.) 1394 10.73 2.12 1126 10.56 2.17 268 11.06 1.83
Productivity (log current 1394 5.98 1.51 1126 5.99 1.56 268 5.95 1.28
IDR mil.)
Independent var.
Labor cost (log current IDR 1394 3.22 0.50 1126 3.14 0.48 268 3.51 0.49
mil.)
Mac. purchase (log current 1394 1.58 2.90 1126 0.84 2.13 268 4.36 3.86
IDR mil.)
Material cost (log current 1394 10.01 3.27 1126 10.42 2.68 268 8.29 4.68
IDR mil.)
Foreign ownership (%) 1394 19.59 38.45 1126 9.87 28.20 268 60.47 47.83
After Policy Intervention

Dependent var.
Output (log current IDR mil.) 1392 10.88 2.08 1125 10.88 2.16 267 10.89 1.71
Productivity (log current 1392 6.15 1.46 1125 6.22 1.54 267 5.88 1.02
IDR mil.)
Independent var.
Labor cost (log current IDR 1392 3.42 0.52 1125 3.34 0.49 267 3.75 0.48
mil.)
Mac. purchase (log current 1392 2.39 3.06 1125 2.03 2.87 267 3.92 3.33
IDR mil.)
Material cost (log current 1392 10.26 3.09 1125 10.62 2.66 267 8.75 4.15
IDR mil.)
Foreign ownership (%) 1392 19.32 38.43 1125 9.18 27.45 267 62.03 47.65
Source : Author, calculated based on the SI (Industrial Survey) data of 2015 and 2017
As indicated in the statistics of variables used (Table 1), the total

number of observations for the two periods (2015 and 2017) is at 2786. As
an additional robustness measure, the observation was controlled to include
only firms that reported their production activity for at least ten months.
This eliminates firms that reported their figures for more than 9 months or
less.
In terms of output, the result shows that the treatment group comprising
firms located in the Batam FTZ has a higher mean score (11.06) compared to
the control group (10.56) before the policy intervention. After 2016 the gap

is narrowing with productivity growth in the control group outweighing
the treatment one. In contrast, the score per worker productivity is lower,
albeit only slightly, which was 5.95 in the treatment group against 5.99 in the
control group. However, after 2016 both experienced productivity growth,
but the score in the control group is noticeably larger (6.22) compared to the
treatment gorup, i.e., those located in Batam FTZ (5.88).
Estimation Strategy
The treated region (treat) is a binary variable mark with 1. The period before
the policy came into effect (post) is marked as 0, and 1 is given for the period
following the policy implementation which is 2016. The initial equation set-
up in predicting the impact of policy on productivity is written as follows:
… (1)
resembles output or per worker productivity change in firm i at the
year t. All the beta coefficients ( ) are unknown parameters. Hence, the
parameter of interest is , which resembles the DID coefficient. Menahwile,
is the error term, expected to be uncorrelated with the other variables.
When the expected condition is met, then it can be confirmed that the parallel
trend assumption is fulfilled (Albouy 2004).
Using the OLS, the model specification is formulated as follows:
… (2)
X is a vector of endogenous variables that includes (1) unit labor cost,
(2) machinery purchase, (3) raw material cost, and (4) foreign ownership
(% share); whereas and are region-specific and individual firm fixed
effects, respectively, which control unobservable heterogeneity that might
occur due to culture, language, and supply-chain network. Lastly, is an
error term not correlated to the main independent variable.
Parallel Trend Test

To ensure the validity of the DID model as laid above it is necessary to study
the parallel trend assumption in the pre-estimation stage. A failure to confirm
a parallel trend could lead to a biased result. Therefore, it is necessary to
gather the pre-treatment periods of firm productivity data in 2013, 2015,
and 2017. Figure 4 shows almost virtually similar trend of growth between
Batam FTZ and the control group in the period before the policy change,
thus confirming the parallel trend assumption. After the policy took place in

2016, productivity in the treatment group is now lower than in the control
group. The counterfactual line predicts Batam’s growth projection should
the policy was never in place.
Figure 4. Parallel Trend Graph for Output and Productivity
A. Output B. Productivity
Note : Batam FTZ is set as the treatment group while the neighboring provinces (North
Sumatera, Riau, South Sumatera, and Lampung) are set as the control group.
Vertical thick line denotes beginning of policy change; the centralization of FTZ
in February 2016. Dashed black line marks the counterfactual line should the
policy was not applied.
Source : Author’s own calculation based on the SI (Industrial Survey) data of 2013, 2015,
2017.
The parallel trend graph was confirmed by testing it using lead (post)
and lag (pre) data. Using the firm output as the dependent variable does
not allow it to pass the lead-trend test, but it does manage to pass the time-
trend one (see Table 2, Column 1). The time-trend result does not suffice the
DID assumption that treatment coefficient is zero in the pre-period (Pischke
2005). Our main variable of interest, productivity per worker, managed to
pass both the lead-trend as well as the time-trend tests (see Table 2, Column
3). The F-test score probability indicates whether the variables pass or do not
pass the test.

Table 2. Parallel-trend Test Result.
Parallel Trend Test Output Productivity
(1) (2)
Lead-trend
F-stat. 9.46 1.127
Prob. F 0.002 0.259
Result Not-passed Passed
Time-trend
F-stat. 0.65 0.72
Prob. F 0.419 0.395
Result Passed Passed
Source : Author’s own calculation based on the SI (Industrial Survey) data
of 2013, 2015, 2017.
RESULTS AND DISCUSSION

This part is divided into five sub-sections. The first one presents the
unconditional or also known as the baseline result. This serves as the
benchmark as more control variables were added into the model in Sub-
Section 2 that shows the robustness of the full estimation result by comparing
it to a different control group using neighboring provinces. Meanwhile,
Sub-Section 3 presents an extension analysis in order to look at the different
policy impact based on firm characteristics. The last sub-section provides a
discussion based on the result findings.
Baseline Result
Table 3 presents the baseline result, without controlling for endogenous
variables. Columns (1) and (2) describe the result using output as a dependent
variable. First, it can be seen that the DID coefficients are significantly
negative, either calculated using the standard OLS or with Fixed-Effects
OLS. Second, using the main dependent variable, productivity, in Columns
(3) and (4), the coefficients are consistently negative and significant for both
the output and productivity, suggesting that the result could be robust.

Table 3. Baseline Estimation Result
Variables Output Productivity
OLS FE-OLS OLS FE-OLS
(1) (2) (3) (4)
Treat 0.833*** 3.563*** 0.280*** 2.390***
(0.132) (0.373) (0.092) (0.319)
Post 0.222*** 0.233*** 0.227*** 0.226***

(0.084) (0.031) (0.059) (0.030)
Post*Treat -0.392** -0.423*** -0.303*** -0.309***

(0.175) (0.090) (0.116) (0.085)
Constant 10.225*** 6.739*** 5.671*** 3.611***

(0.068) (0.773) (0.047) (0.739)
Province FE Yes No Yes No

Firm FE No Yes No Yes
Provinces 5 5 5 5
Observations 2786 2786 2786 2786
Deg. of freedom 6 1454 6 1454
Adj. R-sq 0.132 0.935 0.166 0.882
F-statistics 70.217*** 13.245*** 91.878*** 6.844***
Note : - Productivity is per worker output, reported in log current IDR million.
- Variable post denotes the year after 2016 where central-local friction regarding
Batam FTZ governance has been on the rise.
- Post*Treat is our difference-in-difference estimation.
- Standard errors are in parentheses with * 0.1 ** 0.05 *** 0.01 denote significance
level.
Source : Author’s own calculation based on the SI (Industrial Survey) data of 2015, 2017
Main Result
Let us continue the baseline result above by adding endogenous control
variables necessary to explain productivity. Table 4 shows the main estimation
results. Column (1) shows that the negative coefficient for firm output
stays (-0.368) after controlling for unit labor cost, machinery purchase, raw
materials purchase, and foreign ownership share. In terms of productivity,
the same trend can be observed as shown in Column (5). The coefficient
has smaller magnitude (-0.279) compared to previous specification, but the
negative trend remains.

Column (2) shows the comparison between Batam FTZ with its
closest neighbouring province, North Sumatra. This resulted in a similarly
negative coefficient with the previous column. The negative effect in terms
of productivity is confirmed in Column (6). In the next specification, the
control to the next neighboring province, Riau, is changed into Column (2).
Here, it can seen that the coefficient is significantly smaller, the same with
productivity shown in Column 7. Riau province is the former administrative
province for Batam FTZ before it was splitted into the new Riau Island
province in 2004. Next, Batam FTZ was also compared to South Sumatera
(Column 4) where the result is negative, albeit not significant, the same as
the productivity (Column 8).

Table 4. Main Estimation Result
Output Productivity
Variables
(1) (2) (3) (4) (5) (6) (7) (8)
Post 0.064** 0.084** 0.009 -0.104 0.074*** 0.066** 0.091 -0.064
(0.027) (0.033) (0.066) (0.096) (0.028) (0.033) (0.073) (0.096)
Treat 2.436*** 2.328*** -1.463* -1.448* 1.379* 1.337* -2.003*** -1.03

(0.822) (0.858) (0.778) (0.846) (0.794) (0.804) (0.748) (0.791)
Post*Treat -0.368*** -0.375*** -0.271*** -0.175 -0.279*** -0.267*** -0.242** -0.106

(0.084) (0.085) (0.099) (0.120) (0.077) (0.077) (0.100) (0.117)
Unit labor 0.307*** 0.340*** 0.250** 0.338*** 0.324*** 0.362*** 0.211* 0.285**
cost (0.059) (0.080) (0.122) (0.129) (0.058) (0.075) (0.110) (0.117)
Mach. 0.066*** 0.090*** 0.064*** 0.077*** 0.051*** 0.058*** 0.057*** 0.061***

Purchase (0.010) (0.014) (0.015) (0.019) (0.009) (0.012) (0.014) (0.018)
Raw material 0.168*** 0.144*** 0.097*** 0.098*** 0.159*** 0.137*** 0.102*** 0.107***
(0.021) (0.023) (0.023) (0.023) (0.018) (0.019) (0.021) (0.021)
Foreign own. 0.000 0.000 0.000 0.002 0.000 0.000 0.000 0.003**
(%) (0.001) (0.002) (0.002) (0.003) (0.001) (0.002) (0.001) (0.001)
Constant 7.962*** 7.812*** 9.651*** 8.645*** 3.277*** 3.172*** 4.633*** 3.723***

(0.278) (0.329) (0.501) (0.524) (0.254) (0.294) (0.436) (0.445)
Control All N. Riau S. All N. Riau S.

province Sumatra Sumatra Sumatra Sumatra
Firm FE Yes Yes Yes Yes Yes Yes Yes Yes
Observations 2786 1878 874 737 2786 1878 874 737
R-sq. 0.286 0.289 0.189 0.201 0.270 0.263 0.195 0.208
F-statistics 33.010 23.282 8.057 7.128 36.942 23.830 9.982 8.134
level.
Source : Author’s own calculation based on the SI (Industrial Survey) data of 2015, 2017.
Across all specifications shown in Table 4, the time-trend indicator (post)
indicates a positive coefficient except for the comparison with South Sumatra,
suggesting that for both groups the output and the productivity growths
are positive. Furthermore, it is interesting to see the difference between

those positive trends. A larger trend in the control group resulting in the
negative DID coefficient. The control variables (labour cost, machinery, and
raw materials) show expected positive signs. With the exception of foreign
ownership which shows virtually negligible magnitude, the variables are
statistically significant. Omitted variable bias is still present as explained
by the full model in Column (1) and (5) showing only around 27-29% of
the changes in output and productivity. Since we have observed similar
treatment effect for both dependent variables
Extension Analysis
For all the estimations presented in this sub-section, the same control variables
used in Table 4 (Column 3) in the previous section are applied. However, this
part is focused on firm productivity, which is the main dependent variable. To
begin with, the extension analysis is started by dividing firms based on their
size. The size definition follows national standards where large firm status is
assigned for a firm with 100 or more workers. Meanwhile, the medium-size
firm is given to a firm with 20 to 99 workers. Table 5 presents the estimation
result based on firm size with additional legal ownership status.

Table 5. Extension Analysis: Heterogeneity of Firm Size and Location
Productivity
Medium Large Non-IP IP IP + IP + FDI IP + FDI
firm firm DDI + Ex.
(1) (2) (3) (4) (5) (6) (7)
Post 0.075* 0.022 0.007 0.016 0.083 -0.458 -1.492***
(0.040) (0.039) (0.030) (0.149) (0.172) (0.402) (0.378)
Treat 1.02 -0.7 1.763*** -2.071** 0.57 -2.570*** 0.6

(0.735) (0.652) (0.675) (1.047) (1.476) (0.898) (1.032)
Post*Treat -0.229** -0.346*** -0.513*** 0.051 -0.012 0.469 1.275***

(0.101) (0.130) (0.124) (0.183) (0.296) (0.403) (0.334)
Constant 4.110*** 2.644*** 3.901*** 3.700*** 3.680* 3.374*** 2.439

(0.354) (0.366) (0.316) (0.716) (1.879) (0.955) (1.622)
Control Yes Yes Yes Yes Yes Yes Yes

variables
Firm FE Yes Yes Yes Yes Yes Yes Yes
Observations 1470 1316 2029 757 472 285 171
Deg. of 817 787 1234 572 402 188 128
freedom
R-sq. 0.91 0.93 0.95 0.92 0.93 0.92 0.93
F-statistics 8.460*** 8.372*** 11.802*** 3.540*** 2.317*** 6.072*** 4.364***
level.
- IP refers to firm residing inside Industrial Park, while FDI refers to Foreign Direct
Investment firm (with 10% or more share owned by foreign entity) while DDI is
Domestic Direct investment. Ex. = Exporting firm.
Source : Author’s own calculation based on the SI (Industrial Survey) data of 2015 and
2017.
The result for medium firms is presented in Columns (1) where the DID
coefficient is showing a negative trend (-0.229), and the effect is significant.
Meanwhile, Column (2) shows the result for large firms. Here the DID
coefficient becomes significantly larger while it still retains its significances.
Firm locations (Columns 3 and 4) seem to provide different results. Those
located outside of Industrial Park (IP) show a largely negative (-0.513) and

significant effect whereas those located inside the IP do not show negative
trend although the result is not significant. This observation is relevant as
back in 2009 the Indonesian government issued Governmental Regulation
(PP No. 24 in 2009) that requires all industrial activities to be carried out
inside industrial parks.
Inside the IP, firm origin shows different signs with domestic firms
(DDI/Domestic Direct Investment) being negative and the foreign one (FDI/
Foreign Direct Investment) being positive. Both, however, are not statistically
significant. Lastly, combining firm location and origin of investment with
orientation (exporting firm) shows us a contrasting result with the others. It
yields positive treatment effect (1.275) under 0.01 significance level (Column
7). The magnitude is exceptionally large with small number of observation
(171), which is worth for further studying to confirm this result.
Discussion
Through multiple exercises in the previous section, we have observed that
there is a convincingly negative productivity impact of the 2016 centralisation
policy of Batam FTZ. From the main result, the -0.279 coefficient magnitude
corresponds to around 32% drop in productivity. The same negative result
(45% drop) is also observed for firm growth. This section will be focused on
(1) the issue of central-local relation in the currently decentralised Indonesia,
and (2) possible determinants of industrial resilience against policy change.
The central government, through its affiliated agency, has been the
dominant agent in the development and industrialisation of Batam FTZ
during the new order era (1967-1998). The centralisation strategy worked
well given the centralised nature of the institution at that time. However,
under the currently contrasting political landscape since decentralisation
reform in 2001, applying a centralised approach is expected to bring a
counter productive consequence as shown in Table 3. Hence, the finding is
similar to the previous studies (Aritenang 2017; Aritenang and Chandramidi
2020; Negara and Hutchinson 2020) that failed to find positive impact of
political and policy change in the Batam FTZ. Poor institutions that come
with the policy change have also been pointed out to contribute to the
industrial decline in the case of electronic sector in the island (van Grunsven
and Hutchinson 2017).
Despite its ambitious attempt to strengthen national coordination, the 2016
policy was still not able to resolve the dual authority issue, if not rather

exacerbating the dualism. This leads to the increasingly negative perception
toward the FTZ administration in the period following the policy (Sindonews
2016; Tribunnews 21 April 2017). Previously, one of the largest civil acts that
demands to end the duality of zone administration happened in 2007 (MKRI
21 March 2013) although it was not successful. It is not too far fetched to
think that local actors and politicians who have strong linkage to the local
government of Batam support or work behind these actions. This contributes
in lowering down competitiveness in the zone as suggested by Negara and
Hutchinson (2020) and translates to the declining productivity shown in this
study.
The often mentioned problem related to the dualism of authority is
land ownership, where most strategic location is owned by BP Batam due
to its strong role during the Suharto administration. The new established
autonomous government of Batam in 1999, and later the establishment of
Riau Island provincial government in 2004, had experienced a number of
difficulties to find space for development in the island. Another issue is
regarding the management of strategic resources. For example, in providing
services regarding the issuance of land permit and license, the disfunction
and overlapping roles between the two institutions have led to legal
uncertainty to some 22,000 land titles (Negara and Hutchinson 2020). The
result of heterogeneity analysis shows us that the negative effect persists
across firm size, whether they are medium or large.
From a policy point of view, this is something that is rather difficult
to control. However, it is easier for the government to intervene for firms’
choice of location, as was done with the 2009 Industrial Park policy. The
result shown in Table 4 confirms that firms located in Industrial Park (IP) are
less exposed to negative impact of policy change. One frequent complaint
in Batam, as discussed by Hutchinson (2017) among others, is regarding the
double bureaucracy and overlapping role between BP Batam and the local
government. Dealing with a complex bureaucratic matter has been the job
of IP (Industrial Park) management in many well-managed parks in Batam.
Therefore, firms inside IP can concentrate more on the production side rather
than wasting time dealing with regulations and permits.
Furthermore, in addition to location issue, firms’ characteristics such
as its origin and orientation could also affect their performance. Foreign
firms which engage in exporting activity were not negatively affected by
the policy change. Instead, they managed to retain their productivity level

relative to the control group. Thus, in times of policy discontent such as the
one experienced by Batam, maintaining these kinds of firms is essential to
prevent deeper productivity drop.
CONCLUSION
After it was perceived to fail to perform well under the local government
leadership, president Jokowi attempted to recentralise Batam FTZ
administration in 2016 by placing it under the Coordinating Minister of
Economic Affairs. However, this does not resolve the still existing dual
authorities issue and rather increasing the tension between the central and
local government.
As mentioned previously, this research aims to investigate the impact
of that policy change on firms’ productivity. The difference-in-difference
estimation strategy is employed to exploit the impact, using Batam’s
neighbouring provinces as the control group. Prior to such analysis, the data
passing the parallel trend test for growth (passing the time trend test) and
productivity (passing both lead and time trend tests) were set. Main results
suggest that the policy has caused the output growth to decline by 45% and
32% drop in terms of productivity growth. The impact was serious that led
the government to take drastic action to revert the policy decision by 2019.
There are two important things that stand out from this study. First,
firms’ choice of location can determine their severity against policy change.
Those located in the Industrial Park are not necessarily affected by the
negative impact of the policy. Second, foreign and export oriented firms are
more resilient to the change. The result of the analysis has explained that
those types of firms are able to record positive growth relative to the control
group in the period after the policy implementation. Regardless of future
policy options decided by the government, maintaining the operations of
these firms is important to prevent deeper productivity loss.

REFERENCES
Albouy, D. 2004. Program evaluation and the difference in difference
estimator. Economics 131: 1–4.
Aritenang, A.F., 2017. Special economic zone at the crossroads: The case of
Batam. Jurnal Ilmu Sosial dan Ilmu Politik 21: 132–146.
Aritenang, A.F., and A.N. Chandramidi. 2020. The Impact of Special
Economic Zones and Government Intervention on Firm Productivity:
The Case of Batam, Indonesia. null 56, 225–249. https://doi.org/10.108
0/00074918.2019.1643005
Besstremyannaya, G. 2013. The impact of Japanese hospital financing reform
on hospital efficiency: A difference-in-difference approach. The Japanese
Economic Review 64: 337–362.
BKPM. (n.d.). National Single Window for Investment. Retrieved on 4
February 2022, from https://nswi.bkpm.go.id/
BP Batam. (n.d.). History of Batam. Retrieved on 1 February 2022, from
https://bpbatam.go.id/en/about-batam/
BPK. Database Peraturan. (n.d.). Database peraturan. Retrieved on 2 February
2022, from https://peraturan.bpk.go.id/
Broadfoot, R. 2003. “Batam Risk Report.” Political and Economic Risk
Consultancy, Hong Kong.
Card, D. 1992. Do Minimum Wages Reduce Employment? A Case
Study of California, 1987–89. ILR Review 46: 38–54. https://doi.
org/10.1177/001979399204600104
Card, D. and A. Krueger. 1994. Minimum Wages and Employment: A Case
Study of the Fast-Food Industry in New Jersey and Pennsylvania.
American Economic Review 84: 772–93.
Damuri, Y.R., D. Christian, and R. Atje. 2015. “Kawasan Ekonomi Khusus
dan Strategis di Indonesia: Tinjauan atas Peluang dan Permasalahan.”
CSIS Report.
https://www.csis.or.id
Data Basin, (n.d.). Maps. Retrieved 3 April 2021, from https://databasin.
org/maps
Duflo, E. 2001. Schooling and Labor Market Consequences of School

Construction in Indonesia: Evidence from an Unusual Policy Experiment.
The American Economic Review 91, 795–813.
Farole, T., and G. Akinci. 2011. Special economic zones: progress, emerging
challenges, and future directions. World Bank Publications.
Ge, W. 1999. Special economic zones and the economic transition in China.
World Scientific.
Hutchinson, F.E. 2017. Rowing Against the Tide? Batam’s Economic Fortunes
in Today’s Indonesia. Southeast Asia, no. 8.
Jati, G.P. 2016. Jokowi Perintahkan Dualisme FTZ Batam Dihentikan. [online]
ekonomi.
MKRI. 2013. Sebabkan Dualisme Pemerintahan di Batam, UU Kawasan
Perdagangan Bebas Diuji. https://www.mkri.id/index.php?page=web.
Berita&id=8249
Negara, S.D. and F.Hutchinson. 2020. Batam: Life after the FTZ? Bulletin of
Indonesian Economic Studies 56, 87–125.
Pischke, J.-S. 2005. Empirical methods in applied economics: Lecture notes.
Downloaded 24 July 2015.
Romadi, A. 2016. Tak Hanya demo, Warga Batam Juga Gelar Aksi Jahit Mulut.
Sindonews.com. https://daerah.sindonews.com/berita/1151942/194/
tak-hanya-demo-warga-batam-juga-gelar-aksi-jahit-mulut
Singapore Business Federation. 2015. Bilateral Trade and investment relations
between Singapore and Indonesia are enhanced with the signing of a strategic
cooperation agreement between SBF and the Indonesian Chamber of Commerce
& Industry (Kadin Indonesia).
https://www.sbf.org.sg/sbf-kadin-indonesia-strategic-cooperation-
agreement-2015
Romadi, A. 2016. Tak Hanya demo, Warga Batam Juga Gelar Aksi Jahit
Mulut. Sindonews.com. November 1. https://daerah.sindonews.com/
berita/1151942/194/tak-hanya-demo-warga-batam-juga-gelar-aksi-
jahit-mulut
van Grunsven, L. and F.E. Hutchinson. 2017. The evolution of the electronics
industry on Batam Island (Riau Islands Province, Indonesia): an
evolutionary trajectory contributing to regional resilience? GeoJournal
82: 475–492.

Wahyuni, S., I.K. Djamil, S.A. Astuti E.S., and T. Mudita. 2010. The study of
regional competitiveness in Batam, Bintan and Karimun. International
Journal of Sustainable Strategic Management 2: 299–316.
Wong, P.K., and K.K. Ng. 2009. Batam, Bintan and Karimun–Past History
and Current Development Towards Being A SEZ. Singapore: Asia
Competitiveness Institute, Lee Kuan Yew School of Public Policy,
National University of Singapore.
Wulandari, S. 2012. Batam Free Trade Zone. Hong Kong: Asia Monitor
Resource Center.
Zaenuddin, M., Saleh W. Kumorotomo, and A.H. Hadna. 2017. Dualisme
Kelembagaan antara Pemerintah Kota dan Badan Pengusahaan Batam
Serta Dampaknya terhadap Kinerja Perekonomian di Kota Batam.
Journal of Applied Business Administration 1: 219–231.

Chapter 3
Rent Seeking Undercover: Corruption

and Public Spending Composition in
Indonesia
Jesita Wida Ajani, I Gede Sthitaprajna Virananda, and Ghifari Ramadhan
Firman
INTRODUCTION
Public officials committing corruption might skew spending towards
categories of activities that provide the most lucrative opportunities to extract
rents (Tanzi and Davoodi 1997; Gupta, deMello, and Sharan 2001; Hessami
2010). Studies by Mauro (1998), Delavallade (2006) and Baraldi (2008), for
example, found that corruption decreases public spending particularly on
education. Corruption might also be committed by marking up spending
predominantly in sectors such as public construction and public health. These
deemed most corruption-prone sectors are driven by the aforementioned
behaviour of the bureaucrats, increasing their shares in the government’s
fiscal spending, which may result in a decrease in other sectors allocation
and in turn in the welfare loss on the part of the whole society.
This chapter aims to investigate whether the rent-seeking activities
are prominent in the local public spending composition. To this end, it
is focused on the sub-national level data because there are variances and
heterogeneities that much more happened at the local level rather than at
the aggregate national level. To accommodate our research intentions and
design, we use a novel dataset of corruption cases in Indonesia from “Cegah
Korupsi” set up by the Faculty of Economics and Business, Universitas
Gadjah Mada. The data comprises individual level corruption history in
39
which we aggregate it into provincial level. We also combined it with the
data from the Central Bureau of Statistics (BPS) of Indonesia to gather other
complementary covariates.
We ran regressions to find the relationship between corruption cases
and public spending allocation. We divided the public spending into
several major categories: public service, health, social protection programs,
economic programs, education and housing. Our main independent variable
is corruption cases per capita, which is the number of corruption cases per
province divided by the number of population. To control heterogeneity
across provinces, we employ fixed effect regressions.
Based on the data provided by Cegah Korupsi, Papua and South
Kalimantan rank the first, as they occupy the highest level of corruption cases
per capita by far (Figure 1). Meanwhile, Banten, West Jawa, Central Jawa and
East Jawa, all have the lowest corruption cases per capita, clearly affected
by their substantially large population sizes as denominators. Finally, our
fixed-effects regression finds three spending categories that are significantly
associated with corruption per capita, namely social protection, health, and
housing. The sign is positive for the first two while negative for housing.
This result could imply that social protection and health present lucrative
rent-seeking opportunities, affirming the urgency to reform institutions in
these spending areas.
Figure 1. Mean of Corruption per Capita and Population from 2007-2016

PREVIOUS LITERATURE ON CORRUPTION
The composition of public spending is determined by various factors, both
economic and political. Budget allocation on certain expenditure items is
likely driven by the corresponding needs of the constituents, at least to some
extent. Such needs might be reflected by aggregate economic indicators,
such as the overall level of output or income. For instance, Fan et al. (2008)
found empirically that as developing countries become richer, they tend
to spend more on education, health and social security. More specifically,
spending composition should be driven by the different demands of different
groups in society. For instance, the proportion of young people seems to
affect the demand for education spending (Marlow and Shiers 1999), while
aging populations command more spending on health and social security
(Feldstein 1997). Literature from political economy and public choice have
also extensively delved into how the characteristics of interest groups affect
their ability to lobby the government for favourable policies (Becker 1983),
including certain public expenditures.
On the other hand, public officials who propose budget plans and
approve them should also affect public spending allocation. Deacon (1978)
portrayed the government as a benevolent social planner which determines
spending allocation to maximise a weighted social welfare function, which
is defined in terms of public services consumed by individuals, subject to
a budget constraint. Similarly, the political preference function expresses
a policymaker’s utility as a function of welfare from different groups or
entities in society (Bullock 1994). More recent literature tends to consider
public officials as more self-serving, for instance by not only maximizing
social welfare but also electoral support, responding to voters’ preference
(Tridimas 2001). Selfish interests of public officials, such as elite capture
and rent-seeking, are believed to drive public spending composition. For
example, Besley et al. (2012) found that heads of administrative units in
India allocate more money to villages where they originate.
Corruption, which encompasses rent-seeking activities such as bribery,
has been proposed as a factor influencing public spending composition.
The idea behind this is that different spending items provide different
opportunities to extract rents. Shleifer and Vishny (1993) believe that corrupt
public officials favour investment projects which generate the highest bribes
and not necessarily the most efficient. Mauro’s (1996) model adapted from
Barro (1990) implies that if corruption were treated as a tax on income
Rent Seeking Undercover : Corruption and Public Spending 41

with different rates across sectors, then the composition of government
spending would not be independent of corruption. Indeed, there are certain
expenditure sectors that stand out as being more prone to rent-seeking, such
as natural resource extraction (Sachs and Warner 1995; Ades and Di Tella
1999; Leite and Weidmann 2002). Based on the aforementioned theories,
corruption would skew public spending allocation to these sectors.
On the empirical front, multiple studies have investigated how corruption
might affect public spending composition at both the cross-country and
within country levels. Mauro (1998) conducted cross-section estimation with
corruption indices from the International Country Risk Guide as proxy for
corruption in over 100 countries. He addressed potential endogeneity using
multiple instrumental variables, including ethnolinguistic fractionalisation,
colonial history, and trade openness, finding robust results that corruption
decreases allocation to education expenditures. Erhlich and Lui (1999)
similarly found that corruption decreases human capital investments. These
results affirm the hypothesis that spending on education is less prone to
rent-seeking because its value is not difficult to exact and more visible to
the public, e.g., textbook prices and teacher salaries compared to military
equipment or infrastructure projects costs. By contrast, Gupta et al. (2001)
found that higher corruption is associated with higher military spending.
Focusing on developing countries, Delavallade (2006) used the three-stage
least squares method to show that higher public corruption leads to less
education, health and social protection while spending on public services
and order, fuel and energy, culture and defence increases. She proxied
corruption using an indicator produced by the World Bank, which is also
perception-based.
More recently, some works have obtained administrative data which
could serve as a better proxy for corruption. One example is United States
corruption convictions data, with breakdown by state, from the Public
Integrity Section (PIN) of an annual report to Congress by the Department
of Justice. An early use of this data was Leeson and Sobel (2008), which
concluded that public corruption is substantially higher in states frequently hit
by natural disasters. On the question of public spending composition, Cordis
(2014) employs a more detailed database of public employees’ convictions for
official misconduct or misuse of office gathered by the Transactional Records
Access Clearinghouse (TRAC), an organisation associated with Syracuse
University. Using political instrumental variables related to the age of state
constitution and eligibility to vote, she discovers that corruption decreases

government spending’s share for education (especially higher education),
corrections and public welfare, as well as the “other and unallocable budget
items” category. On the other hand, the portion of government expenditure
for health and hospitals, housing and community development, and natural
resources increased with corruption.
While administrative data should be more reliable than perceptions of
corruption, the quality of such data may vary across sources. For instance,
Cordis and Milyo (2015) argue that results from the PIN data may be
misleading due to having no detail on the level of public officials convicted
(federal, state or local) and including many conviction cases of postal service
employees stealing and destroying mail. They went as far as to argue that
Leeson and Sobel’s (2008) conclusion was driven by these data issues, showing
that the results do not replicate with TRAC data. Finally, case statistics data
are still subject to the hiddenness and sensitivity gaps, where actual cases of
corruption might not be prosecuted and their information withheld due to
political considerations (Guo 2015). Furthermore, the number of cases that
arrive in court is driven by both corruption and anti-corruption activities,
with failure to distinguish them possibly misleading.
Evidence from Indonesia

Several pieces of evidence show how corruption can affect economic
outcomes. For instance, Olken (2006) found that then subsidised rice
programme known as Raskin (Beras untuk Rumah Tangga Miskin) suffered
severely from leakages that the welfare loss was much bigger than the gain.
The beneficiaries, on average, received 18% less rice from the programme
and the number is exacerbated in a more ethnically heterogeneous and sparse
setting. However, this relationship is not just a one-way causality. A research
from Kuncoro (2006) states that the political and business uncertainty induce
bribery behaviour among companies.
Nonetheless, it is not just the uncertain part; the whole political process
itself is also correlated with corruption prevalence. The study by Lewis
and Hendrawan (2019) shows that the majority of coalitions across districts
in Indonesia have caused a shift in spending towards health allocation.
Although it seems good at first, the authors then found out that it became a
rent-prone sector for the bureaucrat. This is also confirmed by the research
from Henderson and Kuncoro (2011) where the composition of parties in the
local government affected the prevalence of corruption1.
1 Few research, nevertheless, argue that the relationship between corruption and economic outcomes

A question then arises: what can be done? Combating corruption is
surely a hugely hard task since the efforts will include both political and
economic aspects, as well as the entrenched nature of corruption (Hamilton-
Hart 2001). Therefore, the government needs to start from improving the
governance of its annual budget, as suggested by Suryadarma (2012), for
the educational spending. In this case, it can be expected to minimizing gaps
for any embezzlement activities. Second, a top-down monitoring approach
is proven effective in tackling corruption behaviour, especially on the local
level (Olken 2007). Third, involving the civil society organisations at the local
level to monitor corruption can also serve as an alternative (Setiyono and
Mcleod 2010), considering the decentralisation has increased the prevalence
of local corruption (Rinaldi et al. 2007).
DATA
The main data we used in this chapter comes from corruption database
named Cegah Korupsi-5 (5th wave) produced by Faculty of Economics and
Business Universitas Gadjah Mada. The data originated from a web scraping
process based on the supreme court website. It comprises individual level
corruption history, such as personal data, the charges, verdict in every level
of judgeship, loss caused by corruption and the name of judges involved.
The data supposedly covers all the corruption cases that have already taken
place from 2000 to 2019, since our data were collected from the verdict
documents at district, provincial and supreme court level.
has not been easily detected. Resosudarmo and Subiman (2006), for example, could not find the
correlation between firm bribery behaviour and the use of illegal logging among wood-based
industries in Indonesia.

Figure 2. Distribution of the Recorded Corruption Cases from 2001-2019
Nonetheless, there are several problems with the data. First, not all
cases are recorded on the website of the supreme court. Figure 2 shows how
the recorded cases are normally distributed. We suspect that for early years,
the number is relatively small because there were not many corruption cases
exposed since the corruption eradication commission was formed only
recently. For recent years, most cases that are still ongoing. These cases do
not have any verdict documents, hence not recorded in the dataset. Notably
there are some heterogeneities across courts in Indonesia.
As shown in Table 1, the differences are not that high at glance. A
simple OLS regression is conducted—as presented in Table 2—to figure
out whether there is an association between province and corruption cases.
Here, Java island was set as the base variable, of which the results show that
on average the number of corruption cases is heterogenous across provinces
and statistically different with Java. This might be caused by endogenous
trait of the corruption itself where it is more prominent in region with
more economic activity, or maybe because of the limitation of corruption
eradication commission in exposing cases on some regions. Unfortunately,
no further evidence was found to answer this question.

Table 1. Distribution of the Recorded Corruption Cases from 2001-2019
Province Frequency %
Aceh 119 2.91
Bali 48 1.18
Banten 46 1.13
Bengkulu 83 2.03
DKI Jakarta 312 7.64
Daerah Istimewa Yogyakarta 85 2.08
Gorontalo 62 1.52
Jambi 80 1.96
Jawa Barat 336 8.23
Jawa Tengah 275 6.73
Jawa Timur 371 9.08
Kalimantan Barat 101 2.47
Kalimantan Selatan 316 7.74
Kalimantan Tengah 80 1.96
Kalimantan Timur 117 2.86
Kalimantan Utara 17 0.42
Kepulauan Bangka Belitung 53 1.30
Kepulauan Riau 33 0.81
Lampung 74 1.81
Maluku 54 1.32
Maluku Utara 43 1.05
NTB 92 2.25
NTT 174 4.26
Papua 91 2.23
Papua Barat 31 0.76
Riau 43 1.05
Sulawesi Barat 41 1.00
Sulawesi Selatan 205 5.02
Sulawesi Tengah 101 2.47
Sulawesi Tenggara 55 1.35
Sulawesi Utara 110 2.69
Sumatera Barat 167 4.09
Sumatera Selatan 101 2.47
Sumatera Utara 168 4.11
Total 4084 100.00

Table 2. Regression of Number of Cases on Province
VARIABLES Number of Corruption Cases

Sumatera -9.342***
(1.181)
Kalimantan -6.529***
(1.442)
Bali and Nusa Tenggara -8.397***
(1.661)
Sulawesi -8.273***
(1.435)
Maluku and Papua -11.388***
(1.435)
Observations 464
R-squared 0.174
Note : Standard errors are in parentheses. *, ** and *** indicates 10%, 5% and 1%
statistical significance, respectively. Regression is done by using provincial
dummy variables, categorical year variable as control and Java as base.
Second, there are many duplicated observations. Third, many variables
suffer from prevalent missing values, thus not all of them can be analysed
directly. Fourth, the corruption cases tend to be small in number for the
early and latter year of the period. For example, in the early 2000s and years
around 2019 there were a small number of corruption cases. Based on the
practices by supreme justice, the cause of this problem is that not all cases
are uploaded on the supreme court website, and if they are, it happens quite
randomly. There is no tendency of certain characteristics that cause the data
being not uploaded. Fourth, the data do not have an explicit date of when
the corruption occurred, since the embezzlements did not happen in just a
one-time moment. Indeed, it may happen in a period of time, nonetheless
that data cannot provide that. Fifth, what type of corruption they have been
convicted with is not stated in detail.
Our approach in tackling these problems, first, we conducted a thorough
data cleaning process to drop any duplicated observations. Second, we used
2010-2016 data only since most observations are concentrated at these time.
This would raise a concern about selection problem cause by truncated. We
firmly believed that the exclusion of other years would not yield biased
results since the error of other years is assumed random, as aforementioned
before. Overall, we got around 4000 observations that were good enough to
be included. We also used prosecution date as our time period benchmark

because this date is an available time identifier that is close enough with the
actual corruption incident.
Since the analysis focused on provinces in Indonesia, we aggregated
our data at provincial level. Regarding the data corruption were taken from
“Cegah Korupsi”, we can infer that Papua has the highest corruption per
capita case (Figure 3). This might be due to a relatively small population
and high number of corruption cases compared to other regions. The second
province with the highest corruption per capita is South Kalimantan, followed
by West Sulawesi. The cases of corruption in West Sulawesi, however, is
a unique case where 22 legislative members were accused of local budget
misuse in 2009, causing an outlier in the regional corruption data. Banten
is the province with the lowest number of corruption per capita, plausibly
caused by a higher number of population causing the corruption per capita
index to fall.
Figure 4 displays the results of the descriptive statistics of government
spending allocation in all provinces in Indonesia. The allocated spending
comprises economic, health, public service, social protection, education,
housing, and security spending. The chart suggests that the provincial
government spending for public service purposes is the highest compared
to other types of spending, this is consistent among all provinces. The rank
of the highest to lowest government spending allocation is similar for all
provinces. However, different magnitudes of spending allocation between
provinces are expected. Social protection, on the other hand, stood as the
lowest spending allocation purpose. This may be caused by the fact that
most social protection allocation and spending are managed solely by the
central government, leaving the provincial government to only be in charge
of minor social protection programmes.

Figure 3. Corruption per Capita per Province from 2007-2016

Figure 4. Budget Allocation per Province from 2007-2016

It is further investigated whether prevalence of corruption is affecting
public spending by merging the corruption data with other provincial
characteristics covariates collected from the Central Bureau of Statistics
(BPS). Nevertheless, this chapter limited its analysis on the aggregate
provincial budget only, without diving into much more detail, e.g., transfer
from central government. In general, public service has the highest allocation
which is around 33%, on average. Education appears as the second highest
with an average of 25% of budget allocation. While social protection has the
lowest average around 1.4%. The control variables are gross rate of secondary
school participation, GDP per capita and the number of population. Table 3
provides the descriptive statistics of all variables we utilised in this chapter.
(1) (2) (3) (4) (5)
VARIABLES N Mean Standard Min Max
Deviation
Dependent Variables
Share of Economic Spending 301 9.642 2.072 4.662 15.22
Housing Spending per Capita 302 16.16 4.738 5.777 31.11
Health Spending per Capita 299 9.334 1.732 4.536 13.88
Social Protection Spending per Capita 295 1.409 0.334 0.604 2.928
Public Service Spending per Capita 297 32.95 4.788 19.38 49.05
Education Spending per Capita 303 25.63 6.314 7.452 44.07
Main Independent Variable
Corruption per Capita 303 0.211 0.211 0.0088 1.899
Control Variables
Ln Population 303 15.29 0.971 13.77 17.69
Ln GDRP 303 17.27 1.242 14.67 19.98
Gross Rate of Secondary School Participation 303 86.35 7.175 65.09 107.5
METHODOLOGY
In general, our regression model is constructed as follows:
where variable is the amount of per capita provincial

budget allocation for sectors, i.e., health, public service, housing, education,
economic programmes, and social programmes. We can be sceptical on this
model: there is a simultaneity relationship between all of these outcomes.

However, as Cordis (2014) and Haupt & Oberhofer (2000) point out, using
regression with the same explanatory variables across regression will yield
same results as seemingly unrelated regression (SUR) with the same variables
specification. Our variable interest is corruption cases per capita, which is
number of corruption cases divided by population size during year t, and
the effect is shown by . Subscript i indicates the province cross section.
is a set of relevant control variables, i.e., population of the provinces, gross
rate of secondary school participation and GDP per capita; whereas and
are provincial and year fixed effects, respectively.
The data from the year of 2010 to 2016 were then added, allowing them
to appear in panel format. Nevertheless, apparently there is unobserved
heterogeneity across provinces that might affect both public spending and
the number of corruption cases. Presumably these unobserved factors are
unique in every province and time invariant, i.e., social-cultural aspects.
Hence, a fixed effect needs to be included to control the unobserved
heterogeneity. If is statistically significant, it can be concluded that the
number of corruption cases can drive the budget allocation of the provincial
government. Whether the sign is negative or positive, it is still a sign that
corruption does matter in public spending discourse. Hence, our hypothesis
is that any significance shows a rent-seeking prone sector.
To check whether our results are robust, we also regress equation (1)
using lag three years of corruption per capita. This is to help ensure that our
results do not suffer reverse causality bias (Vadila and Resosudarmo 2020).

Main Result
Tables 4A, 2B and 2C display the results of regressing each spending
category’s share on corruption per capita. We exclude Papua province from
our regressions due to its corruption per capita values being extremely
high, primarily driven by its particularly small population. Our regression
results are considerably different from simple correlations, implying that
the controls do matter in determining public spending composition. Our
preferred results are those with province and year fixed effects (columns 2
and 4 in Tables 4A, 4B and 4C). We find that three spending categories have
a statistically significant positive association with corruption per capita,
namely health, public service, and social protection. Whereas the prosecution

cases per capita, is negatively associated with the share of local government
expenditure in housing in a given year.
Table 4A. Main Regression Results
Economic Education
VARIABLES
(1) (2) (3) (4)
Corruption per Capita 0.713** 0.123 4.047* 0.132
(0.329) (0.242) (2.029) (1.114)
Ln Population 0.119 -0.665 7.941* 3.240
(0.787) (0.834) (4.137) (4.450)
Ln GDRP per Capita -1.305*** -1.221* 0.513 1.659
(0.475) (0.674) (2.191) (2.851)
Gross Rate of Secondary -0.0324*** -0.0286 0.0126 -0.0165
Education (0.00917) (0.0180) (0.0554) (0.0534)
Constant 13.08 23.93* -98.77 -30.12
(11.42) (13.34) (59.81) (70.33)
Province fixed effects Yes Yes Yes Yes
Year fixed effects No Yes No Yes
Observations 301 301 303 303

R-squared 0.146 0.338 0.071 0.525
Number of prov_id 32 32 32 32
Table 4B. Main Regression Results

Health Housing
VARIABLES
(1) (2) (3) (4)
Corruption per Capita 0.0946 0.524** -3.809*** -1.784***
(0.224) (0.245) (1.085) (0.649)
Ln Population 5.745*** 0.535 -4.375 -5.729*
(0.931) (1.277) (2.594) (3.178)
Ln GDRP per Capita 3.560*** -0.297 -1.243 -3.909*
(0.561) (0.794) (1.514) (2.114)
Gross Rate of Secondary -0.00639 -0.0223 -0.0171 0.00257
Education (0.0141) (0.0208) (0.0315) (0.0371)
Constant -85.16*** 1.677 87.82** 111.3**
(13.91) (20.94) (38.78) (51.82)

R-squared 0.459 0.627 0.094 0.405

Table 4C. Main Regression Results
Public Service Social Protection
VARIABLES
(1) (2) (3) (4)
Corruption per Capita 2.088* 2.267* 0.326*** 0.227**
(1.116) (1.115) (0.104) (0.106)
Ln Population -4.727 -3.212 0.643*** 0.306
(3.516) (5.102) (0.208) (0.242)
Ln GDRP per Capita 0.824 -0.442 0.197 0.117
(1.897) (3.455) (0.138) (0.170)
Gross Rate of Secondary -0.00830 0.106 -0.00583* -0.00875
Education (0.0566) (0.0711) (0.00315) (0.00659)
Constant 103.9* 72.13 -8.393** -2.935
(52.46) (81.22) (3.108) (4.070)

R-squared 0.032 0.239 0.127 0.295
The positive coefficient of social protection suggests that corruption

is likely to induce local governments to spend more on social protection,
implying that such spending presents opportunities for rent-seeking. While
major social protection programmes, such as the conditional cash transfer
(Program Keluarga Harapan) and food aid (Bantuan Pangan Non Tunai) are
run by the central government, local governments could also distribute their
own social protection programmes. Some high-profile corruption cases,
including the one committed by the Minister of Social Affairs himself in 2020,
demonstrate that social protection programmes are prone to corruption.
For instance, social protection in the form of in-kind aid usually involves
procurement, where public officials could mark up the budget and collude
with suppliers to extract rents.
There are also anecdotes of local officials taking cuts along the chain of
aid transfer from the disbursing agency down to the recipients, who end up
receiving less than they are supposed to. While an increasing share of social
protection spending seems benign, it is actually a problem if much of the
spending is corrupted. Hence, the finding affirms the importance of proofing
the institutions and mechanisms of social protection against rent-seeking
activities. One popular proposal has been to prioritise direct transfers instead
of in-kind and channel them directly to recipients through bank accounts,

skipping multiple levels of public officials.
Another spending category that was found to be positively associated
with corruption is health. Interestingly, there was not any significant relation
from our simple correlation analysis. Health expenditures encompass
investment in hospitals and health equipment, salaries for health officials,
and other public health programs. We posit that this positive relation is
driven by the first type of spending, which would be particularly prone to
corruption similar to construction and procurement programs in general. A
more detailed breakdown of local health expenditures, which is unfortunately
still scant, would be needed to investigate this hypothesis.
Nevertheless, Cordis (2014) also finds in the United States that health
and hospitals are among the categories positively related with corruption
cases. Transparency International also estimates that at least 5% of health
services spending worldwide may be lost to corruption. It is worth noting
that healthcare as another area in health expenditure might also be prone
to rent seeking. This is due to the sheer size of the healthcare programme
in Indonesia, popularly known as BPJS (Badan Penyelenggara Jaminan
Sosial Kesehatan or Social Health Insurance Administration Body), and its
relatively recent establishment, which might mean that the system might
still have many loopholes for corruption.
Furthermore, anecdotal evidence suggests that the limited amount
of funds available for the massive payouts has led to competition among
healthcare providers for BPJS disbursements, opening opportunities for
bribery. However, this particular example of health spending corruption
would not be included in our sample, since BPJS was only instituted
around 2014. Even worse, the share of local government spending on
housing seems to be negatively related to corruption per capita. This result
is particularly surprising since we expect housing projects, where the cost
structure is oftentimes less obvious and more easily marked up, to be prone
to corruption. There could also be collusion opportunities when contractors
are involved. Looking at our sample, several provinces with a relatively high
share of housing spending are Kalimantan Timur, Sumatera Selatan, and
Riau. Aside from Kalimantan Timur in 2010-2011, these provinces indeed
have low levels of corruption per capita prosecuted.
Many provinces with low housing spending share—such as, Bali, Jawa
Tengah, and Daerah Istimewa Yogyakarta—also have low corruption per
capita. In fact, a simple correlation analysis shows only a negligible negative

coefficient that is far from statistical significance. With such inconclusive
evidence, we consider the possible explanation that corruption opportunities
might simply be more lucrative in other spending categories. Unfortunately,
no rigorous evidence has been found to further support this argument.
Thus, opening a further research development on this issue. Moreover,
many infrastructure projects that are considered to be prone to corruption
are likely categorised in non-housing spending, for example hospitals in the
health category (positive and statistically significant) and infrastructure in
the economic category (positive but not statistically significant).
Regarding the spending categories which are not statistically
significant—i.e., public service, economic, and education—, the first
category (public service) is particularly interesting because the coefficient
of corruption per capita is positive and statistically significant when control
variables are excluded or Papua is included in our sample. This might be due
to public service that takes the highest share of local government spending in
almost all provinces, resulting in relatively less variation between provinces,
until several socioeconomic variables are controlled.
Furthermore, the inclusion of Papua distorts the result as the province
happens to have both outlier values of corruption per capita and the highest
share of public service spending nationally. Nevertheless, we hesitate to
make conclusions about public service since it is one of the more general
spending categories and detailed breakdown of the spending is unavailable.
Similarly, the economic spending category is also a relatively general
category covering a diverse array of programmes, with significant positive
coefficient for corruption when control variables are excluded.
Finally, the share of the education spending category is positive but not
statistically significant. While we expected education to have the least relation
with corruption, we find it interesting that the coefficient is not negative and
statistically significant, as commonly found in previous studies (Mauro 2006;
Cordis 2014). Nevertheless, the statistically insignificant result affirms that
education is not a particularly attractive field for corruption. Furthermore,
education spending in Indonesia is restricted by a constitutional mandate of
minimum 20% of budget, both at national and regional level, hence allowing
less room to reallocate spending related to education.
Robustness Check
To check whether the results are robust, particularly not suffering from a

possible reverse causality bias, Equation (1) is regressed using lag three
years of corruption per capita following Vadila and Resosudarmo (2020).
Tables 5A dan 5B show the results of these regressions. We found some of
the signs change but no statistically significant association of corruption per
capita on budget allocation. These results seem promising since it indicates
that changing the year lead is not yielding statistically significant and similar
results from our main regression, indicating lower possibility of reverse
causality bias.
If our hypothesized mechanism holds that corruption in a given fiscal
year affects spending by altering the incentives of policymakers/politicians
during the budgeting process in that same year or the year before. In this
case, it should not be future corruption that affects current spending. If
future corruption affects current spending then there could be a separate
ongoing trend, which happens to correlate with future corruption, that is
affecting some budget spending but not current corruption.
Table 5A. Robustness Check
(1) (2) (3)
VARIABLES
Economic Education Health
Corruption per Capita -0.108 0.0141 0.383
(Three years Lag) (0.294) (1.015) (0.449)
Ln Population -1.126 7.941* 3.929
(1.598) (4.137) (3.010)
Ln GDRP per Capita -1.739 0.513 2.747
(1.531) (2.191) (2.288)
Gross Rate of Secondary -0.0170 0.0126 -0.0122
Education (0.0230) (0.0554) (0.0224)
Constant 30.88 -98.77 -56.37
(26.44) (59.81) (50.22)
Province fixed effects Yes Yes Yes
Year fixed effects Yes Yes Yes
Observations 205 207 203

R-squared 0.138 0.667 0.407
Number of prov_id 32 32 32

Table 5B. Robustness Check
(1) (2) (3)
VARIABLES Housing Public Service Social
Potection
Corruption per Capita 1.054 -0.374 0.0794
(Three years Lagged) (0.701) (0.650) (0.164)
Ln Population -4.375 -2.588 0.652
(2.594) (6.536) (0.450)
Ln GDRP per Capita -1.243 1.713 0.570
(1.514) (6.380) (0.471)
Gross Rate of Secondary -0.0171 0.0831 -0.00443
Education (0.0315) (0.0777) (0.00763)
Constant 85.44 61.16 -9.386
(75.61) (107.5) (7.699)
Province fixed effects Yes Yes Yes
Year fixed effects Yes Yes Yes
Observations 206 201 199

R-squared 0.488 0.225 0.273
Number of prov_id 32 32 32
Heterogeneity Analysis
Table 6 illustrates heterogeneities across province in Indonesia. For provinces
in Java, housing has a negative association and positive relationship for
public service and social protection on corruption cases per capita. For Bali-
NT and Maluku-Papua, the negative and positive association was shown
by housing and public service spending, respectively. Meanwhile, different
cases take place for Kalimantan where education has positive relationship
with corruption per capita, and Sulawesi in the case of health. Interestingly,
no evidence on the relationship between corruption case and the economic
allocation of regional budget allocation can be found.

Table 6. Heterogeneity Results of Corruption per Capita
(5) (6) (7) (8) (9)
Spending
Variable Sumatera Java Bali-NT & Kalimantan Sulawesi
Mal-Pap
Economic -1.247 0.316 -0.597 1.188 0.114
(0.982) (1.015) (1.014) (0.559) (0.365)
Education -0.797 4.222 0.547 4.081* -0.639

(2.079) (2.542) (3.121) (1.462) (1.619)
Health 0.589 1.290 -0.270 1.217 0.935***

(0.864) (1.042) (0.862) (0.552) (0.138)
Housing 2.070 -4.509** -5.233* -0.635 -2.020**

(3.721) (1.293) (2.465) (0.964) (0.628)
Public Service 3.306 5.588** 9.471* -4.658 1.389

(2.341) (1.657) (4.185) (2.596) (0.914)
Social Protection 0.361 0.627*** 0.262 0.213** 0.124

(0.318) (0.116) (0.135) (0.0605) (0.102)
Note : Mal = Maluku and Pap = Papua. All models are with control variables and
province and year fixed effects.
CONCLUSION AND POLICY IMPLICATIONS

Using a novel dataset of corruption prosecutions from Indonesia’s Supreme
Court database, we investigate the relation between corruption and public
spending composition in Indonesia’s local governments. We are motivated by
the notion that if some public spending categories present more rent-seeking
opportunities, then corruption might steer public spending towards these
categories. We map provinces by their level of corruption cases per capita
and check simple correlations with shares of certain spending categories. We
then exploit variation across provinces by running fixed effect regressions
of each major spending category on corruption per capita and some control
variables. Our results show that corruption per capita is positively associated
with the share of local spending towards social protection and health. Indeed,
several high-profile corruption cases related to social protection and public
health anecdotally show that these two fields are prone to corruption.
Meanwhile, other spending categories such as public spending,

economic, and education do not exhibit statistically significant relations
with corruption, which we suspect could be due to the general nature of
these categories, the relatively low opportunities for corruption, or other
institutional factors, such as the constitutional mandate restricting education
spending. Moreover, there is also an evidence for heterogeneity across regions
in Indonesia where housing is negatively correlated with public spending in
Java and Sulawesi, and positive association for public service in Java, Bali-
NT & Maluku-Papua, social protection and education in Kalimantan, and
health in Sulawesi.
Our results affirm the urgency to reform institutions related to social
protection and health spending in particular, in order to protect them
against rent seekers. As an example, there is a popular proposal to channel
social protection transfers through bank channels directly to recipients.
We would also like to note several limitations mainly related to our data.
First, our dataset does not adequately include specific information on the
corruption committed by the convict, such as date of corruption and the
form of corruption. Furthermore, we cannot ascertain whether some
corruption cases are not yet inputted to the database, though we assume that
any omission should be random. Second, we caution possible endogeneity
due to likely reverse causality from a particular spending category’s share
towards corruption level. We encourage future research to address these
limitations by completing the corruption database and employing methods
such as instrumental variable regression to establish evidence of possible
causal relation.
REFERENCES
Ades, A. and R. Di Tella. 1994. “Competition and Corruption. Institute of
Economics and Statistics.” Discussion Papers 169, University of Oxford.
Baraldi, A., 2008. Effects of electoral rules, political competition and corruption
on the size and composition of government consumption spending: An
Italian regional analysis. The BE Journal of Economic Analysis & Policy, 8:
1–37.
Barro, R. 1990. Government spending in a simple model of endogenous
growth. Journal of Political Economy, 98 (5): S103–S125.
Becker, G., 1983. A theory of competition among pressure groups for political

influence. Quarterly Journal of Economics, 98 (3): 371–400.
Besley, T., R. Pande, and V. Rao. 2012. Just rewards? Local politics and
public resource allocation in South India. World Bank Economic Review,
26 (2):191–216.
Bullock, D.S. 1994. In search of rational government: What political preference
function studies measure and assume. American Journal of Agricultural
Economics, 76 (3): 347–361.
Cordis, A. 2014. Corruption and the composition of public spending in the
United States. Public Finance Review, 42 (6): 745-773.
Cordis, A., and J. Milyo. 2015. “Don’t Blame the Weather: Federal Natural
Disaster Aid and Public Corruption.” Mercatus Working Paper,
Mercatus Centre at George Mason University, Arlington, VA, February
2015.
Deacon, R.T. 1978. A demand model for the local public sector. Review of
Economics and Statistics, 60 (2): 180–202.
Delavallade, C. 2006. Corruption and distribution of public spending in
developing countries. Journal of Economics and Finance, 30 : 222–39.
Ehrlich, I., and F. Lui. 1999. Bureaucratic corruption and endogenous
economic growth. Journal of Political Economy, 107 : 270–93.
Fan, S., B. Yu, B., and A. Saurkar. 2008. Public spending in developing
countries: Trends, determination, and impact. In S. Fan (ed.), Public
Expenditures, Growth, and Poverty. Baltimore, MD: Johns Hopkins
University Press, 20–55.
Feldstein, M. and Samwick, A. 1997. The Economics of Prefunding Social
Security and Medicare Benefits. NBER Macroeconomics Annual, 12 : 115–
148.
Guo, Y. 2008. Corruption in transitional China: An empirical analysis. The
China Quarterly, 194 : 349-364.
Gupta, S., L. deMello, L., and R. Sharan. 2001. Corruption and military
spending. European Journal of Political Economy, 17 : 749–77.
Hamilton-Hart, N. 2001. Anti-corruption strategies in Indonesia. Bulletin of
Indonesian Economic Studies, 37 (1): 65-82.
Henderson, J.V. and A. Kuncoro. 2011. Corruption and local democratization
in Indonesia: The role of Islamic parties. Journal of Development

Economics, 94(2): 164-180.
Hessami, Z. 2010. Corruption and the Composition of Government Expendi-
ture—Evidence from the OECD Countries. Working paper. Konstanz,
Germany: University of Konstanz.
Khattak, N., and J. Khan. 2012. The contribution of education to economic
growth: evidence from Pakistan. International Journal of Business and
Social Science, 3(4): 145-151.
Kuncoro, A. 2006. Corruption and business uncertainty in Indonesia. ASEAN
Economic Bulletin, 23(1): 11-30.
Leeson, P., and R. Sobel. 2008. Weathering Corruption. Journal of Law and
Economics, 51: 667–81.
Leite, C. and J. Weidmann. 2002. Does mother nature corrupt? Natural
resources, corruption, and economic growth. In G. Abed and S.
Gupta (eds.). Governance, Corruption and Economic Performance.
Washington, DC: International Monetary Fund, 159–196
Lewis, B. D. and A. Hendrawan. 2019. The impact of majority coalitions
on local government spending, service delivery, and corruption in
Indonesia. European Journal of Political Economy, 58 (C): 178-191.
Marlow, M. L., and A.F. Shiers. 1999. Do law enforcement expenditures
crowd-out public education expenditures?’ Applied Economics, 31(2):
255-266.
Mauro, P. 1996. The effects of corruption on investment, growth, and
government expenditure. International Monetary Fund Working Paper
96/98.
Mauro, P. 1998. Corruption and the composition of government expenditure.
Journal of Public Economics, 69: 263–279.
Olken, B. A. 2006. Corruption and the costs of redistribution. Journal of
Public
Economics, 90(4-5): 853-870.
Olken, B. A. 2007. Monitoring corruption: Evidence from a field experiment
in Indonesia. Journal of Political Economy, 115(2): 200-249.
Resosudarmo, B.P. and N.L.Subiman. 2006. The link between firm
characteristics, bribery, and illegal logging in Indonesian wood-based
industries. ASEAN Economic Bulletin, 23 (3): 360-372.

Rinaldi, T., M. Purnomo, and D. Damayanti. 2007. Fighting corruption in
decentralized Indonesia: Case studies on handling local government
corruption. Jakarta: The World Bank.
Sachs, J., and A. Warner. 1995. “Natural Resource Abundance and Economic
Growth.” NBER Working Paper 5398.
Setiyono, B. and R. Mc.Leod. 2010. Civil society organisations’ contributions
to the anti-corruption movement in Indonesia. Bulletin of Indonesian
Economic Studies, 46 (3): 347-370.
Schleifer, A. and R. Vishny. 1993. Corruption. Quarterly Journal of Economics,
CIX: 599–617.
Suryadarma, D. 2012. How corruption diminishes the effectiveness of public
spending on education in Indonesia. Bulletin of Indonesian Economic
Studies, 48 (1): 85-100.
Tanzi, V. and H.Davoodi. 1997. “Corruption, Public Investment, and
Growth.” Working paper. Washington, DC: International Monetary
Fund.
Tridimas, G., 2001. The economics and politics of the structure of public
expenditure. Public Choice, 106 (3–4): 299–316.
Vadila, Y. and B.P. Resosudarmo. 2020. Tariff reform and income inequality
in Indonesia. Regional Science Policy and Practice, 12 (3): 455-475.

Chapter 4
Political Competition and Deforestation:

Harnum Yulia Sari, Raka Rizky Fadilla, and Zihaul Abdi
INTRODUCTION
It has been argued that the root cause of environmental degradation in
Indonesia is a combination of weak institutions and poor governance.
Dearth of transparency, lack of accountability, and limited participation
of stakeholders in formulating and implementing forest land management
policies have contributed to the continued deforestation (Bisson et al. 2003;
Cisneros et al. 2020; Smith et al. 2003). As a result, over the last two decades,
Indonesia has lost 9.75 million hectares of its humid primary forest, making
up 36% of its total tree cover loss in the same period (Global Forest Watch
2021).
Decentralisation has been blamed as one of the major institutional factors
driving local capture that trigger deforestation in Indonesia (Adrison 2014).
Decentralisation in Indonesia gave local government discretionary power
over the supply of logging permits, which has led to excessive logging and
bribe-taking (Smith et al. 2003). Furthermore, the direct mayoral elections
did not yield an increase in political accountability and resulted in even
more deforestation (Burgess et al. 2012), possibly by changing the structure
of political rent extraction, with fewer candidates running for office and
larger governing coalitions exerting their influence.
However, theoretically, Indonesia can still benefit from the
decentralisation system through the check and balances mechanism between
the executive and legislative branches at the local level. This ideal version
of decentralisation requires a strong local parliament—characterised by
65
high competition among political parties—that is expected to hold the local
leaders accountable for their policy choice, leading to better governance and
ultimately reducing deforestation.
A body of research has found that political competition plays a key
role in influencing socio-economic performance. A high degree of political
competition encourages efficiency in public goods provision and dissuades
rent-seeking behaviours. Political competition and the checks and balances
imposed in a well-functioning democracy restrict the ability of governments
to engage in rent-seeking behaviour (Olson 2000; North 1990; North and
Weingast 1989). Political competition has built-in mechanisms for increasing
the probability of detection and punishment due to the competition among
parliament members of the ruling party and opposition parties. Studies
dealing with environmental policies also showed that political competition
raises the stringency of environmental policies (Fredriksson et al. 2005). That
being said, it is important to investigate to what extent does the competition
among political parties at the local level influences the environmental
outcomes, specifically around forest management.
In Indonesia, a higher degree of political competition appears to
improve the economic outcome (Rezki 2020) and local services delivery
(Skoufias et al. 2014) through pro-business policies and better expenditure
choices, respectively. While evidence highlights the link between local
political incentives and deforestation in Indonesia (see Cisneros et al. 2021),
there is no empirical evidence to date linking local political competition and
deforestation.
Our research, therefore, attempts to explore the association between
political competition and the rate of deforestation. We hypothesise that a
higher degree of political competition within the parliament, hence more
accountable government, is associated with a decreasing rate of deforestation.
This chapter offers a novel measurement of political competition by
simultaneously employing fractionalisation and polarisation index. The
Fractionalisation index measures the degree of political party diversity in a
parliament. While this measurement has been widely used in the previous
literature to illustrate the degree of political party fragmentation (Wang
2014), we argue that such an index is insufficient to measure the likelihood
of conflicting ideas in the policymaking process. We, therefore, complement
the measurement with the polarisation index.
Using subnational-level data of Indonesia from 2009 to 2018, we

empirically test whether the political party fractionalisation and political
party polarisation at the local parliament are associated with deforestation
rates, which we define as the annual primary forest loss. We find evidence
that political party fractionalisation and political party polarisation are
indeed relevant determinants of the deforestation rate, but the association is
opposite with each other. On the one hand, this research shows a negative
association between political party fractionalisation with deforestation rate,
i.e., the deforestation rate is declining along with a higher degree of political
party fragmentation. On the other hand, we also find that the increasingly
polarised local parliament is positively associated with the deforestation
rate.
This chapter proceeds as follows. The following section will provide the
institutional context of electoral politics, political parties, and their association
with deforestation in Indonesia. The third section presents the data and
empirical strategy of this research. Section four provides the analysis of the
regression and the subsequent discussion. Section five concludes.
GOVERNMENT AND DEFORESTATION IN INDONESIA

A Brief History of Indonesia’s Election and Political Party
In Indonesia, the election of local parliament (DPRD) members is part of the
general election (pemilihan umum) that takes place once every five years.
Since its inception in 1955, general elections have been held twelve times.
However, the election for DPRD members only started in 1971. There were
nine political parties and one mass organisation (GOLKAR) competing in
that first general election of the New Order era. President Suharto initiated
a restructuring of the party system in 1973. Except for Golkar, all political
parties and groups were forced to fuse under two political parties: the United
Development Party (PPP) and the Indonesian Democratic Party (PDI). Only
these two parties that were allowed to compete with Golkar in the general
election for more than two decades (Liddle 1978).
This hegemonic system of political parties lasted until the reformation
in 1998. The party system in Indonesia entered a new phase in the 1999
general election. The process of transition in late 1998 led to full-blown
democratisation of the political regime, paving the way for the founding
elections of the second Indonesian democracy in June 1999. Out of more
than 100 political parties founded in the following months, 48 passed the
Political Competition and Deforestation : 67

requirements to participate in the first free and fair elections Indonesia had
seen since 1955 (Sulistyo 2002).
The Role of Local Parliament in the Policymaking Process

After the big bang decentralisation in 2001, most public service provision
responsibilities were deconcentrated to the district level. District
governments have a legislature and executive branches like their national-
level counterparts. Each district has a local house of representatives (DPRD)
and an elected head of government (regent/bupati or mayor/wali kota),
who present the drafts of the development budget and other plans to the
DPRD for approval. The local parliaments have three functions: legislation,
budgeting, and supervising (Blomkamp, E. et al. 2017).
Prior to the fall of the authoritarian regime in 1998, DPRDs merely
rubber-stamped executive drafts with little capacity to analyse executive-
initiated policy. The member of the parliament could not conduct their
independent analyses or seek citizen input in the process (USAID 2008).
However, free, and fair elections in 1999 and 2004, together with new
decentralisation reforms, have allowed DPRDs to become more responsive.
Their authorities and powers have significantly increased, and DPRDs have
claimed their legitimate role in local governance as the legislative branch of
government. DPRDs are actively involved in drafting local regulations and
play critical roles in the local budgeting process. Local governments consult
with DPRDs on key policy decisions, and DPRDs are actively taking part in
the development planning of the economy and society of their regions.
Despite improvements in recent decades, transparency, and citizen’s
participation in the policymaking process at the local level remains a challenge.
As USAID (2008) noted, although public consultation and civic engagement
in the legal drafting process are institutionalised by local policies, citizens’
capacity to advocate, lobby and contribute during public consultations are
still very limited. Consequently, the DPRD-citizen relationship does not
always translate into a constituent-centered policy.
Another challenge relates to the quality of the regulation. Kumorotomo
(2007) studied a large number of regional regulations throughout 1999-2004
and found that a significant portion of those regulations is dominated by the
regulations related to the institutional establishment, budget, and local tax.
Only a small share of the local regulations touches upon the community’s
real issues, such as employment, health, and education.

Forestry Management
During the first two decades of Suharto’s regime, the government could be
characterised as a solid and centralised regime. Suharto centralised control
over natural resources throughout the country (Smith et al., 2003). Suharto
also gave privileged access to forests for several politically connected
individuals. Several high-ranked military officers also benefited from non-
collusive corruption by selling their influence to secure access of forests for
business entrepreneurs (Barber and Talbott 2003).
Immediately after the fall of Suharto, the administrative and regulatory
authority of environment management was changed. Under Suharto regime,
the 1967 Basic Forestry Law gave the national government the exclusive
right of forest exploitation in the so-called Forest Estate. The major change
after 1999 was that the district forest departments became part of the district
government, answerable to the head of the district (the bupati), rather than
a division of the central Ministry of Forestry. So that the district government
could initiates proposals to the central government about forest conversion
(Burges et al. 2012), Consequently, the district forest office became central
role in enforcing forest policy to the key gatekeeper for illegal logging.
Under decentralisation, local governments are also entitled to a larger
share of resource revenues. Once given the authority to issue small-scale
logging licenses, district governments quickly recognised the opportunity
to raise revenues. It led to a proliferation of small-scale logging licenses.
However, small-scale logging operations posed problems for forest
management because the regulations for small-scale operators contained no
provisions for replanting. Because of the short duration of these permits,
there were no incentives for the permit holders or their business partners to
follow measures supporting sustainable forest management (Fox et al. 2005).
Following a wave of forest exploitation and as district government
demands for revenues from forest management, the government issued
a regulation that affirmed a trend toward recentralisation in forestry in
2022. Government Regulation No. 34 of 2002 was issued to respond the
uncontrolled forest exploitation during 2000 to 2002, which explicitly states
that only the central government can issue forest harvesting permits, even
at the suggestion of a lower-level government (Moeliono et al., 2008). Based
on this regulation, the Ministry of Forestry has the sole authority to issue
permits for the utilisation of wood products on the recommendation of the
lower levels of government.

In the following decade, through Law No. 23 of 2014 on Regional
Governance, the governments continue to weaken the role of local
governments in allocating logging permits by withdrawing authority over
forest management from district and city governments and shifting it to
provincial and national-level government.
EMPIRICAL STRATEGY
The focus of our research is closely linked with the empirical literature
exploring the connection between institutional quality and environmental
outcomes. Previous studies usually use binary variable contrasting
democracy and autocracy or employ a sort of index to represent the degree
of democracy, civil liberty, and political freedom (Farzin & Bond 2006;
Bhattarai & Hamming 2001; Torras & Boyce 1998). Other studies use a range
of proxies to measure the quality of the institution, such as the strength of
the environmental NGOs (Binder & Neumayer 2005) or electoral system
(Fredrikkson & Millimet 2004). As for the environmental outcomes, past
studies mainly use deforestation, air quality, or carbon emissions as the
representative indicators.
A more recent literature that have the most similarity to ours are
Morjaria (2011) and Su, et al. (2020). Morjaria (2011) exploited the timing
of introduction of multiparty system as the instrument to measure the
impact of political competition on deforestation in Kenya. Su, et al. (2020)
leveraged the party fragmentation index and margin of victory between the
first and second place party to test the relationship between political party’s
competition and air quality in Mexican municipalities. A set of control
variables were employed in those studies, including population, GDP per
capita, density of economic activity, and district-specific fixed-effects.
In this study, fixed effects regression panel data are employed to
capture the relationship between political competition and deforestation at
the district level. This approach is used to capture the time-invariant that
occurs in each district. As GFW data continue to improve methods and
technology to measure the deforestation every year, year fixed-effects are
therefore also employed to measure the effect of methodological difference.
This methodological difference is due to adjustments made by GFW, such
as algorithm adjustments, updated satellite data, and variations in the
availability of satellite imagery1. Our analysis uses the following specification:
1 https://www.globalforestwatch.org/blog/id/data-and-research/menilai-tren-hilangnya-
tutupan-pohon-dalam-kurun-waktu-data-20-tahun-2/

𝑌𝑌𝑑𝑑,𝑡𝑡 = 𝛼𝛼𝑑𝑑,𝑡𝑡 + 𝛽𝛽𝑋𝑋𝑑𝑑,𝑡𝑡 + 𝛾𝛾𝑌𝑌𝑑𝑑,𝑡𝑡 + 𝜃𝜃𝑑𝑑 + 𝜗𝜗𝑡𝑡 + 𝜀𝜀𝑑𝑑,𝑡𝑡
where is logarithmic of deforestation in district d and year t, is the
political competition index, is the vector of control variables and is the
district fixed effects, and is the year fixed effects.
DATA
The analysis was conducted using an unbalanced panel dataset. The number
of districts/cities in our sample is 241 out of a total of 514 districts/cities. We
dropped districts/cities with no primary forest. We addressed the ongoing
process of district proliferation by refining our district framework to the
end of the observation period, building a panel of geographic entities as
observed in 2009 to 2018. This study covers ten years ranging from 2009 to
2018, covering two election cycles. Table 1 provides the summary statistics
of the data used in this study.
Variables Obs Mean Std. Dev. Min Max
Deforestation 1321 2463.623 5755.156 .076 59816.148
Political party 1321 .87 .058 .53125 .9376
fractionalization
Political party polarization 1321 .93 .045 .9259259 .9981689
Political party ideology 1321 .447 .112 0 .6479012
fractionalization
Political party ideology 1321 .954 .121 0 1
polarization
Audit BPK 1321 4.215 1.004 1 5
Price oil exposure 1321 376904.41 458797.5 2 2152481
Environmental 1321 1.260e+10 3.066e+10 0 5.734e+11
expenditure
Natural Resource Revenue 1321 9.828e+10 2.385e+11 2.815e+08 3.622e+12
Sharing
Poverty Rate 1321 10.996 5.906 2.009 45.709
Regional GDP 1321 .183 .544 .009 9.494
Temperature 1321 25.307 1.868 13.17 28.98
Wind Speed 1321 1.138 1.013 .04 4.61
Cloud Amount 1321 81.112 7.32 50.56 91.08
Source: Authors’ own calculation

Deforestation in Indonesia shows a declining trend in recent years,

although previously it had experienced a continuous increase (Figure 1). It
began to decline significantly in 2017 and continues to decrease until today.
Based on the Ministry of Environment and Forestry records, in the 2019-
2020 period, Indonesia’s deforestation rate fell by 75% (Indonesian Ministry
of Environment and Forestry 2021). This deforestation rate is the lowest in
history.
Figure 1. The trend of Deforestation in Indonesia, 2009-2018
Political party fractionalisation and political party polarisation decreased

after the 2014 election (Appendix 1 & 2). The decline in these two indices
could be due to the declining number of political parties competing in 2014.
In 2014, the KPU increased the requirements for political party verification.
As a result, the number of political parties that passed factual verification fell
significantly compared to the elections five years earlier. These two indices
also show relatively high values. It means that there is a diversity of political
parties in the DPRD in Indonesia, but at the same time, there are only a few
major political parties dominating the seats in the local parliament.
As shown in Appendix 3, the relationship between political party
fractionalisation and political party polarisation is not linear. The relationship
between these two indices forms a U-shape curve. This relationship means
that the polarisation of political parties in the DPRD tends to decrease as the
number of parties increases, until at a point, the polarisation increases as
the number of political parties increases. In that part of the curve, the DPRD
consist of many minor political parties and one or two major political parties
dominating the seats in the local parliament.

The Deforestation Rates
The dependent variable used in this study is the deforestation rates (one plus
logarithmic natural of deforestation). We define deforestation as the annual
primary forest loss in hectares and retrieve the data from the Global Forest
Watch (GFW)2. This source is detailed enough to provide primary forest data
at the district level. Primary forest is defined as the mature natural humid
tropical forest that has not been completely cleared and regrown in recent
history. In contrast to tree cover loss which comprises all types of forest, we
only focus on primary forests.
Political Competition
This chapter uses two measures to represent the level of political competition
in the local parliament by adopting the index proposed by Montalvo and
Reynal-Querol (2005) with the value ranging between 0 and 1, namely
fractionalisation and polarisation index:
where FI is the fractionalisation index. PI is the polarisation index. Si,j is
political party’s seat share in the parliament for district i and year j.
𝐹𝐹𝐹𝐹𝑖𝑖,𝑗𝑗 = 1 − ∑𝑁𝑁
𝑖𝑖=1
2
𝑠𝑠𝑖𝑖𝑖𝑖 (1)
0.5−𝑠𝑠𝑖𝑖𝑖𝑖 2
𝑃𝑃𝑃𝑃𝑖𝑖,𝑗𝑗 = 1 − ∑𝑁𝑁
𝑖𝑖=1 ( ) 𝑠𝑠𝑖𝑖𝑖𝑖 (2)
0.5
We constructed the indices using the data of seats number secured by

the political party for the 2009 and 2014 legislative elections at the districts/
cities level. We obtained the dataset from the Indonesia’s General Election
Commission (Komisi Pemilihan Umum/KPU). These two indices represent
political competition in different dimensions. Political party fractionalisation
will represent the degree of fragmentation of political parties in the
parliament. In other words, it showcases the quantity dimension of political
competition. Political party polarisation, on the other hand, explains the
size of each political party’s influence relative to other political parties in
the parliament, representing the quality dimension of political competition
(Dalton 2008).
Control Variables
To minimise the issue of omitted variable bias, we use several control variables
2 Global Forest Watch is the database that provides data and tools for monitoring forests. This data
is retrieved from https://www.globalforestwatch.org/dashboards/

associated with deforestation. First, the BPK (Badan Pemeriksa Keuangan/
Audit Board of the Republic Indonesia) audit opinion is used to represent the
capacity of local governments. The audit conducted by BPK provides 5 levels
of opinion to the financial management of public institutions in Indonesia,
including district governments, wherein the dataset, the higher the number,
the better the opinion (5 is unqualified/Wajar Tanpa Pengecualian). Second,
we measure economic incentives of deforestation use location-specific palm
oil price exposure. The underlying hypothesis is that market participants
observe palm oil’s profitability over recent years and adjust their economic
decisions to convert forest land to oil palm plantations when palm oil prices
increase. This variable is constructed by multiplying yearly global palm
oil prices with the province’s share of oil palm plantation area 3. The larger
the province’s oil palm plantation area, the bigger palm oil price volatility
exposure to that particular province. Third, to control for local government’s
effort in environmental management, we use local government expenditure
for environmental function.
We also include several development and geographical indicators that
could correlate with forest conditions and the rate of deforestation. The
indicators are regional gross domestic product per capita, poverty rate,
temperature, wind speed, and cloud amount.
RESULTS
Empirical Results
The main result from the empirical analysis is provided in Table 2, where
Columns (1-6) show the relationship between political competition
represented by political party fractionalisation and political party
polarisation with deforestation rates. The results of this regression strengthen
the descriptive correlation found in Appendix 4. We found a different
relationship between political party fractionalisation and political party
polarisation with deforestation. A significant and negative relationship is
seen in the political party fractionalisation. An increase in the fractionalisation
index is associated with a slower rate of deforestation. On the other hand, a
positive relationship is observed in political party polarisation, meaning that
the deforestation rate increases along with a higher degree of political party
3 We recognised that government of Indonesia has been imposing moratorium of forest conversion
for oil palm plantation since 2011. However, there are still limitations to its enforcement. This was
revealed in the findings of Gaveau et al. (2021) which showed that based on satellite monitoring,
forest conversion for agricultural land still occurred even after the moratorium despite a declining
trend in recent years. Therefore, price exposure remains a valid control in this case.

polarisation.
Table 2. Main Results
Variables Logarithmic of deforestation
(1) (2) (3) (4) (5) (6)
Political party -4.013*** -4.013*** -3.351** -2.354* -2.354* -3.465**
fractionalisation (-4.24) (-4.24) (-3.30) (-2.12) (-2.12) (-3.25)
Political party 4.135*** 4.135*** 3.814*** 2.357* 2.357* 3.536***
polarisation (4.55) (4.55) (4.07) (2.16) (2.16) (3.40)
Audit BPK -0.0246 -0.0246 0.0131
(-0.88) (-0.88) (0.44)
Logarithmic of price oil 0.0636* 0.0636* 0.0350
exposure (2.02) (2.02) (1.22)
Logarithmic of -0.00872 -0.00872 0.00601
Environmental (-1.57) (-1.57) (0.67)
expenditure
Logarithmic of Natural 0.254*** 0.254*** 0.205**
Resource Revenue (4.20) (4.20) (3.17)
Sharing
Poverty rate -0.0680* -0.0680* -0.0635
(-2.04) (-2.04) (-1.92)
Regional GDP -0.170 -0.170 0.0349
(-1.25) (-1.25) (1.51)
Temperature 0.740*** 0.740*** 0.356
(5.96) (5.96) (1.87)
Wind Speed -0.366 -0.366 -0.0290
(-1.50) (-1.50) (-0.12)
Cloud Amount 0.00198 0.00198 0.0128
(0.26) (0.26) (1.18)
_cons 5.237*** 5.237*** 4.606*** -18.73*** -18.73*** -9.484
(21.97) (21.97) (22.72) (-5.28) (-5.28) (-1.91)
District fixed effects YES YES YES YES YES YES
Island fixed effects NO YES YES NO YES YES
Year fixed effects NO NO YES NO NO YES
N 1321 1321 1321 1321 1321 1321
Notes : t statistics in parentheses * p < 0.05, ** p < 0.01, *** p < 0.001
We provide a placebo test (Table 3) to check whether the future change
in polarisation and fractionalisation index have significant effects on
deforestation. If so, the results in Table 2 might be subject to bias due to

confounding trends. We conducted the test by regressing the deforestation
rates on the lag and forward measure of political competition, controlling
for year-district dummies and time-variants. We found that 1 and 2
years forward and lag measures of political competition have significant
relationship with current deforestation rate. This is understandable as the
independent variables only change once in every five years. As the lag and
forward measures become farther, the estimation become non-significant.
Therefore, we can expect that current deforestation rate is not associated
with the result of past and future election results.

Table 3. Placebo Test for Robustness Check
(1) (2) (3) (4) (5) (6)
Political party -4.831***
Fractionalisation t+1 (-4.42)
Political party 4.809***
Polarisation t+1 (4.49)
Political party -6.286*
Political party 5.540*
Political party -4.779
Political party 3.823
Political party -3.672***
Polarisation t-1 (5.50)
Political party -2.765**
Fractionalisation t-2 (-2.92)
Fractionalisation t-3 (0.27)
_cons 5.373*** 5.996*** 6.204*** 5.169*** 4.928*** 4.894***
(18.63) (14.37) (16.76) (32.04) (59.48) (22.19)
Controls NO NO NO NO NO NO
Island fixed effects YES YES YES YES YES YES
Year fixed effects YES YES YES YES YES YES
N 1124 900 678 1316 1154 1011
Notes : t statistics in parentheses * p < 0.05, ** p < 0.01, *** p < 0.001.

Our finding highlights the relevance of political competition on
deforestation. Our empirical analysis shows that political competition is
associated with deforestation. Furthermore, our approach to measuring
political competition sheds light on the mechanism through which
political competition matters for local environmental outcomes, including
deforestation. We find that both political party fractionalisation and political
party polarisation matters for deforestation yet with a conflicting direction.
These results offer two insights about the measurement of political
competition. First, that the conflicting signs emphasised that the two indices
embody a distinct element of political competition. According to Wang
(2014), party fractionalisation reflects the degree to which electoral support
or legislative representation is divided among various political parties, and
the effective number of parties provides a measure of party fractionalisation
by counting parties weighted by their shares of votes or seats. In contrast,
the polarisation measures how parties are dispersed along an ideological
sphere, indicating the range of ideologicl choices that parties represent.
Dalton (2008) summarised the difference between the two as follows: “Party
fractionalisation tells us about the quantity of party systems; however, party
polarisation shows us about the quality of party systems”.
Second, the findings in this study also prove that conventional methods
for measuring political competition are unable to provide a complete
picture. The literature that uses political competition as an explanatory
variable generally only uses the fractionalisation index or effective number
of political parties (Su, et al. 2020). However, our findings highlight that the
degree of fractionalisation of political parties does not fully represent the
elements of political competition. The fractionalisation index cannot capture
how strong the conflict of ideas occurs in the parliament. Thus, we support
the argument that simply counting the number of parties, although it may
be easier and readily available, could miss the property of party systems that
is of prime interest.
In the following part of this section, we will discuss the interpretation of
our findings for the context of Indonesia. We found that the fractionalisation
of political parties has an inverse relationship with deforestation. It shows
that the DPRD that are less concentrated or is associated with lower rate of
deforestation. It is in line with the argument in the previous literature (Su,
et al. 2020; Rezki 2020; Morjaria 2011) that uses the political fragmentation
index to measure political competition. According to the literature, this

element of political competition created a mechanism to increase the
probability of detection and punishment due to competition among
officials of the ruling party and opposition parties. Therefore, a higher
degree of political competition could encourage the government to reduce
opportunistic behaviour, more efficiently allocate resources, and more
stringent environmental policy.
Interestingly, our study found that political party polarisation has
a significant and positive relationship with the rate of deforestation. This
finding implies that deforestation rate increases along with a more polarised
parliament, contrast with our hypothesis that high level political party
polarisation can intensify ideological debate and strengthen check and
balances over the executive branch.
We suspect that this particular result relates to our approach in
measuring the political party polarisation. In our approach, we use the seat
share of each political party in the parliament to calculate the polarisation
index. The polarisation index with the value of 1 implies that there are two
political parties in the parliament having 50% of the seats each. This approach
assumes that each party represents a mutually exclusive ideological view
and distinct segment of constituent. In fact, two or more parties in parliament
could have similar ideology. We hypothesise that possibility of intersected
ideological view can lead to a nuanced result.
To test our hypothesis, we run our empirical model with different
measure of political party polarisation. Principally, we are interested in
exploring whether the ideological difference of political parties in the
DPRD can be manifested in the debate on the issue of deforestation. In
the new index, instead of using party’s seat share, we use the seat share
of political party’s ideology group. We grouped the political parties in the
local parliament from 2009 to 2018 into three main ideological compasses,
including nationalist, religious, and nationalist religious. The classification
is based on Indonesia political streams classification (Indrayana 2008). The
results are presented in Table 4.

Table 4. Political Party Ideology and Deforestation
(1) (2) (3) (4) (5) (6)
Political party -1.007 -1.007 -1.255 -1.216 -1.216 -1.354
ideology
fractionalisation (-1.01) (-1.01) (-1.27) (-1.27) (-1.27) (-1.37)
Political party 0.808 0.808 1.040 0.736 0.736 0.882

ideology
polarisation (1.25) (1.25) (1.55) (1.06) (1.06) (1.26)
_cons 5.270*** 5.270*** 4.786*** -19.24*** -19.24*** -9.828

(11.09) (11.09) (9.51) (-5.41) (-5.41) (-1.95)
Controls NO NO NO YES YES YES
Island fixed effects NO YES YES NO YES YES
Year fixed effects NO NO YES NO NO YES
N 1321 1321 1321 1321 1321 1321
We found a positive but non-significant relationship between ideological
polarisation of political parties and deforestation. This finding supports the
argument that the political party system in Indonesia differs little on matters
of ideology. As Fionna and Tomsa (2017) noted, political party in Indonesia
tends to be driven primarily by rent-seeking and the quest for patronage
rather than ideological or programmatic debates. Factional divisions exist
but they are usually fluid and based on clientelist loyalties or perceived
opportunities for access to patronage rather than the representation of social
cleavages or competing ideological paradigms.
This result paved our way in proposing the explanation behind the
positive relationship between political party polarisation and deforestation.
We argue that in the context where the ideological view does not matter, a
high degree of party polarisation in parliament lead to a higher probability
of collusive and rent-seeking behaviour. A higher value of polarisation
index implies that the parliament is dominated by two major political party.
Having known that ideology does not have any role in influencing party’s
policy stance, a polarised parliament is therefore an ideal setting for collusion
practices since the negotiation and coalition costs are low.
This argument resonates with findings in the political economy literature

concerning party coalition and rent-seeking behaviour. For instance,
Aytimur (2011) argued that political competition in the context of coalition
governments is not able to eliminate rents. His argument rooted from the
view that coalition formation is stochastic, and this stochastic nature of the
government formation process softens political competition and results
in positive rents in equilibrium. Furthermore, Gottlieb (2015) argued that
the probability of collusion among democratically elected parties on a
representative council is higher when the competing major parties win seats.
Under that circumstance, parties are better able to form a cartel and jointly
engage in private activities that benefit council members at the expense of
citizens.
We extend the analysis by subsampling the estimates for main outcomes
by geography variable that matters in deforestation. Our underlying
hypothesis is that the dominant drivers of the nationwide forest loss are
the major islands of Indonesia with huge initial primary forests. With that
hypothesis, we use heterogeneities of the island for subsampling. However,
our estimation shows that, in general, there is no significant correlation
between political party fractionalisation and polarisation and deforestation
at the island-level other than in Sumatera. The results are presented in the
Table 5.
Table 5. Heterogeneity Analysis, Subsample by Islands
Sumatera Java and Kalimantan Sulawesi Maluku
Bali, Nusa and Papua
Tenggara
(1) (2) (3) (4) (5)
Political party -5.454*** 0.919 -0.682 -0.368 -7.127
fractionalisation (-5.58) (0.05) (-0.23) (-0.31) (-1.55)
Political party -0.946 -48.33 -29.37 1.367 6.171

polarisation (-0.07) (-0.27) (-1.14) (1.23) (0.14)
_cons 15.86 68.27 21.05 -15.00 -3.395

(1.02) (0.42) (0.68) (-1.39) (-0.09)
Controls YES YES YES YES YES
District fixed effects YES YES YES YES YES
Year fixed effects YES YES YES YES YES
N 560 133 276 295 73

CONCLUSION
Our chapter investigates the association between political competition and
deforestation in Indonesian districts. Using panel data over ten years, we
find that political competition is significantly associated with deforestation.
However, our study shows that two measurements of political competition
have an opposite relationship with deforestation. We observe a negative
association between political party fractionalisation and deforestation. This
result supports existing literature on how political competition encourages
the government to reduce opportunistic behaviour and so is associated a
more stringent environmental policy. On the other hand, we find a positive
association between political party polarisation and deforestation. While this
result conflicts with the previous literature, we argue that such a relationship
explains the Indonesian local parliament’s nature where high polarisation
index does not translate into a heightened conflict of ideas. We prove this
argument by showing no association between political ideology polarisation
and deforestation rate, which also confirms the notion that the political party
system in Indonesia differs little on matters of policy and ideology.
This chapter fills the research gap on the institution and environmental
management. First, this study enriches the knowledge about the nexus of
deforestation and institutions in developing countries. Second, it sheds
light on the role of political competition in affecting the government’s
performance. These findings could be helpful for an Indonesian political
context of deforestation and environment management and could be a
starting point to enhance the degree of political competition and reform the
current political system. However, this study has yet to consider political
party coalition and political alignment between local governments and the
legislative (executive and legislative).
REFERENCES
Adrison, V. 2013. “Deforestation in Decentralised and Democratic Indonesia.”
Working Paper No. 01. Institute for Economic and Social Research,
Faculty of Economics Universitas Indonesia (LPEM-FEUI). <https://
www.lpem.org/deforestation-in-decentralized-and-democratic-
indonesia/>
Arnold, L. L. 2008. Deforestation in decentralised Indonesia: What’s law got
to do with it. Law Env’t & Dev. Journal, 4: 75.
Aytimur, R. E. 2011. “Coalition Governments and Political Rents.” Working

Paper.
Barber, C. V. and K. Talbott. 2003. The chainsaw and the gun: The role of the
military in deforesting Indonesia. Journal of Sustainable Forestry, 16 (3-4):
131-160.
Bhattarai, M., and M. Hammig. 2001. Institutions and the environmental
Kuznets curve for deforestation: a crosscountry analysis for Latin
America, Africa and Asia. World development, 29 (6): 995-1010.
Bisson, J., E.S. Guiang, P. Walpole, and D.Tolentino Jr. 2003. Better
governance critical to reversing forest degradation in Southeast Asia. In
Paper submitted to 12th World Forestry Congress.
Blomkamp, E., M.N.Sholikin, F. Nursyamsi, J.M. Lewis, and T. Toumbourou.
2017. Understanding policymaking in Indonesia: In search of a policy
cycle. No. June, 1-45.
Brady, H. E., Verba, S., & Schlozman, K. L. 1995. Beyond SES: A resource
model of political participation. American Political Science Review, 89 (2):
271-294.
Burgess, R., Hansen, M., Olken, B. A., Potapov, P., & Sieber, S. 2012. The
political economy of deforestation in the tropics. The Quarterly journal of
economics, 127 (4): 1707-1754.
Binder, S. ands E. Neumayer. 2005. Environmental pressure group strength
and air pollution: An empirical analysis. Ecological economics, 55 (4): 527-
538.
Cisneros, E., Kis-Katos, K., & Nuryartono, N. 2021. Palm oil and the politics
of deforestation in Indonesia. Journal of Environmental Economics and
Management, 108: 102453.
Dalton, R. J. 2008. The quantity and the quality of party systems: Party
system polarisation, its measurement, and its consequences. Comparative
Political Studies, 41 (7): 899-920.
Farzin, Y. H. and C. A. Bond. 2006. Democracy and environmental quality.
Journal of Development Economics, 81 (1): 213-235.
Fionna, U. and D. Tomsa. 2017. “Parties and factions in Indonesia: The effects
of historical legacies and institutional engineering.” ISEAS Working
Paper No. 2017
Fredriksson, P. G. D.L. Millimet. 2004. Electoral rules and environmental

policy. Economics letters, 84 (2): 237-244.
Fredriksson, P. G., Neumayer, E., Damania, R., & Gates, S. 2005.
Environmentalism, democracy, and pollution control. Journal of
Environmental Economics and Management, 49 (2): 343-365.
Fox, J.J., Adhuri, D.S., Resosudarmo, I. A. P. 2005. Unfinished Edifice
or Pandora’s Box? Decentralisation and Resource Management in
Indonesia. In Resosudarmo, B. P. (Ed). The Politics and Economics of
Indonesia’s Natural Resources. Singapore: ISEAS. 92-108.
Gaveau, D., Locatelli, B., Salim, M., Husnayaen, H., Manurung, T., Descals,
A., and Sheil, D. 2021. “Slowing deforestation in Indonesia follows
declining oil palm expansion and lower oil prices.” Working Paper.
Global Forest Watch. 2021. Primary Forest Loss in Indonesia. Global Forest
Watch. https://www.globalforestwatch.org/dashboards/country/
IDN/
Gottlieb, J. 2015. The logic of party collusion in a democracy: Evidence from
Mali. World Politics, 67 (1): 1-36.
Indrayana, D. 2008. Indonesian Constitutional Reform 1999-2002. Jakarta:
Kompas.
Indonesian Ministry of Environment and Forestry 2021, Deforestasi Indonesia
Turun, Terendah Dalam Sejarah, Indonesian Ministry of Environment
and Forestry. https://www.menlhk.go.id/site/single_post/3640/
deforestasi-indonesia-turun-terendah-dalam-sejarah
Kartodihardjo, H. 2018. Korupsi politik perizinan sawit. Manuscript for
TEMPO Magazine.
Kumorotomo, W. 2007. Memperbaiki Mekanisme Kawal dan Imbang dalam
Sistem Pemerintahan Daerah. Jurnal Kebijakan dan Administrasi Publik,
11 (1): 63-85.
Liddle, R. W. 1978. The 1977 Indonesian Election and New Order Legitimacy.
Southeast Asian Affairs: 122-138
Montalvo, J. G., & Reynal-Querol, M. 2005. Ethnic polarisation, potential
conflict, and civil wars. American economic review, 95 (3): 796-816.
Moeliono, M., Wollenberg, E., Limberg, G. (Eds.). 2009. “Desentralisasi
Tata Kelola Hutan: Politik, Ekonomi dan Perjuangan untuk Menguasai
Hutan di Kalimantan, Indonesia.” CIFOR, Bogor, Indonesia.

North, D. C. 1990. Institutions, Institutional Change and Economic
Performance, Institutions, Institutional Change and Economic
Performance. doi: 10.1017/cbo9780511808678.
North, D. C., and Weingast, B. R. 1989. Constitutions and commitment: the
evolution of institutions governing public choice in seventeenth-century
England. The journal of economic history, 49 (4): 803-832.
Olson, M. 2000. Power and prosperity: Outgrowing communist and capitalist
dictatorships (No. 338.9 Ol84p Ej. 1 025107). Basic Books.
Rezki, J. F. 2020. Political Competition and Economic Performance: Evidence
from Indonesia. Working Paper No. 046. Institute for Economic
and Social Research Faculty of Economics and Business Universitas
Indonesia.
https://www.lpem.org/wp-content/uploads/2020/04/WP-LPEM-
046-Political-competition-and-economic-performance.pdf
Skoufias, E., Narayan, A., Dasgupta, B., & Kaiser, K. 2014. Electoral
accountability and local government spending in Indonesia. Policy
Research Working Paper No. 6782. The World Bank.
Smith, J., Obidzinski, K., Subarudi, S., & Suramenggala, I. 2003. ‘Illegal
logging, collusive corruption and fragmented governments in
Kalimantan, Indonesia.’ International Forestry Review, 5 (3): 293-302.
Su, Z., Pezzola, A., Fidalgo, A., & Cao, X. 2021. Electoral competition, party
system fragmentation, and air quality in Mexican municipalities.
Environmental Politics, 30 (6): 981-1001.
Sulistyo, H. 2002. Electoral politics in Indonesia: A hard way to democracy.
Electoral Politics in Southeast Asia and East Asia, Singapore: Friedrich
Ebert Stiftung, 75-99.
Torras, M., and Boyce, J. K. 1998. Income, inequality, and pollution: a
reassessment of the environmental Kuznets curve. Ecological economics,
25 (2): 147-160.
USAID. 2008. The Role of DPRD in Promoting Regional Autonomy and
Good Governance: Framework, Challenges and New Approaches.
Good Governance Brief
Wang, C. H. 2014. The effects of party fractionalisation and party polarisation
on democracy. Party Politics, 20 (5): 687-699.

APPENDICES
Appendix 1. The trend of Political Party Polarisation in Districts/Cities of
Indonesia 2009-2018
Appendix 2. The Trend of Political Party Fractionalization in Districts/

Cities of Indonesia 2009-2018

Appendix 3. The Relationship between Polarisation and Fractionalisation
Appendix 4. The Relationship between Logarithmic Natural of

Deforestation and Political Competition

Chapter 5
The Role of Socioeconomic Factors in

Voting Preferences in Indonesia’s Local
Elections
Dharra Mahardhika and Ekki Syamsulhakim
INTRODUCTION
It has been argued that voters lean toward picking candidates who have
similar identities with they are than those who have different identities.
For example, in the UK, Heath (2013) found that working class voters are
suspected to be more likely to vote for working class candidates. Nida and
Fahmi (2017), from US survey data in 2012, identified that employees tend to
choose candidates from the same employment level and those voters tend to
choose the same race. From their research in India in 2012, Heath, Verniers, &
Kumar (2015) found that Moslem voters tend to choose Moslem candidates.
On candidate‘s religion, Tilley (2014) explained three things that matter
why religion can influence a person’s choice. First, dominant explanation of
voting behaviour is correlated with religion. Second, religion is correlated
with a particular set of values or at least like policy offerings; and third,
past religious conflicts (history) could shape voting behaviour. Candidates’
religion could influence voter’s decisions directly—i.e., voters choose
candidates based on their beliefs—or indirectly—i.e., voters can conclude
the attribute and belief from candidates’ religion. (Campbell, Green and
Layman, 2011). Only a few research, however, is available describing the
socioeconomic characteristics for such behaviour.
89
As in the case of Indonesia, candidates’ identity is considered as an
important factor in gaining the election. Particularly important are their
religion and gender. Based on a survey data conducted by the Centre for
Strategic and International Studies/CSIS (2017), 58.4% of Indonesians could
hardly accept leaders of different religions. In addition, an Asia Foundation
survey in 2013 found that 38.6% of candidate’s religion becomes an important
factor to elect the Governor, Regent, and Mayor (Strachan 2014). Other
studies elsewhere also find similar results. Ramadlan & Masykuri (2018),
for instance, found similar results in which 62% of the sample community
in their study in Boolang Mongondow Regency, North Sulawesi Province,
Indonesia in 2006 believe that having leaders with the same religion (Islam)
is of importance. Therefore, the role of religion in the election of local heads
is still taken as a determinant factor.
It is also important to look further the reasons why voters’ socio-
economic backgrounds determine their political options. First, since 2007,
Indonesia has conducted direct and concurrent local government head
election (Pilkada langsung dan serentak) as an important instrument
to establish the democratic local government (Martinus, 2013; Suyatno,
2016). The most crucial thing of Pilkada langsung dan serentak is voters’
participation because it can be a reflection of democracy (Akbar, 2017).
Nevertheless, according to Putra (2017), the low participation in Pilkada is
one of the common problems. This confirms the finding by Rochmi (2018)
that the participation rate in pilkada 2018 decreased compared with the 2017
elections, which will have an impact on the success of a democratic country.
An argument for this low and declining participation in elections is that
voters could not find suitable candidates with characteristics as they expect,
particularly related to candidates’ religion and gender.
Second, during the election campaign period, religious issues usually are
typically raised with respect to choosing the “right” regional heads. Indonesia
consists of many different individual and community backgrounds, raising
religious issues during regional elections are very sensitive and prone to
conflics. The candidates’ gender that now becomes one important unwritten
requirement for regional heads is another factor emerging in electing the
“proper” leaders, resulting in a larger number of men becoming regional
leaders than women. Marwah (2016), fore example, mentioned that only 15
female candidate pairs were selected from a total of 56 pairs in simultaneous
regent election in Central Java in 2015.

This chapter, hence, aims to investigate further and measure the
influence of socioeconomic characteristics on the probability of a person
having voting preferences by considering candidate’s religion and gender in
head elections at the local level in Indonesia. In particular, this research’s aim
is twofolds. First, to provide information on the influence of socioeconomic
factors influencing voters’ preferences in local head elections; and second, to
present useful insights and contributions to research in related fields.
Regarding the socioeconomic background of the voters, this chapter
focuses on education, income, occupation, or a combination of these
dimensions which are commonly used to measure socioeconomic status
(Winkleby et al., 1992). They are considered important to measure, as they
determine people’s voting behaviour. For example, using data from the
Election Commission in India from 2004 and 2009, Bhue, Rao, Subramanian,
and Tantri (2014) found that educated voters show a good preference in
voting. When compared to general voters, educated voters will be 20% less
likely to vote for criminals and corruptors and 13% higher in voting for
female candidates.
This chapter employs a pooled logistic regression using the individuals
in the Indonesia Family Life Survey 2007 and 2014 as the sample. Then, we
properly take the individual and community charateristics into account to
measure the influence of socioeconomic characteristics on the probability
of a person having voting preferences by considering religion and gender
using pooled-logistic-regression. We found that the level of education and
voters who live in urban areas—including Moslem-female voters who live
in district and province centre—have negative significance to candidates’
religion and gender.
DATA
The data used in this chapter—i.e., the Indonesian Family life Survey (IFLS)—
are based on a longitudinal survey focused on collecting information on
health and socio-economic indicators of selected household and individuals
living in 13 out of 27 provinces in Indonesia. The selected 13 provinces are
North Sumatra, West Sumatra, South Sumatra, Lampung, DKI Jakarta, West
Java, Central Java, Yogyakarta, East Java, Bali, West Nusa Tenggara, South
Kalimantan, and South Sulawesi (Strauss, Witoelar and Sikoki 2016). The
main reason for using IFLS, dataset instead of other available datasets, is
that information on whether voters consider candidates’ religion and gender
The Role of Socioeconomic Factors in Voting Preferences in 91

in making their choices during elections is only available in the IFLS dataset.
Nevertheless, the main limitation of using IFLS dataset is that the data
represent only 13 out of 27 provinces in Indonesia. IFLS dataset, hence, is
not representative for a whole Indonesia.
Using the book 3B section entitled Partisipasi Masyarakat (Community
Participation) for our independent variable, the questions to be raised
are “What factors do you consider in electing a Bupati/Mayor?”. There
are also some closed-ended questions, and we choose (“Faith/Religion”)
and (Gender) with “yes” or “no” responses. Thus, consider religion and
gender in this chapter clearly indicate voters’ preferences based on equality
of religion and gender consideration during local head elections. Table 1
presents all variables used in this chapter.

Table. 1 Descriptive Statistics
IFLS-4 (2007) IFLS-5 (2014)
Variables Mean Mean
(SD) (SD)
Religion Consideration [1,0] 0.79 0.78
(0.41) (0.42)
Gender Consideration [1,0] 0.41 0.59
(0.49) (0.49)
Level of Education
Elementary School [1,0] 0.41 0.36
(0.49) (0.48)
Junior High School [1,0] 0.17 0.19
(0.38) (0.39)
Senior High School [1,0] 0.25 0.29
(0.43) (0.45)
University [1,0] 0.08 0.10
(0.27) (0.30)
Individual Characteristics
Marital Status [1,0] 0.71 0.74
(0.45) (0.44)
Moslem [1,0] 0.89 0.90
(0.31) (0.30)
Female [1,0] 0.53 0.54
(0.50) (0.50)
Age 39.66 40.37
(15.96) (15.33)
Age (squared) 1828.02 1864.92
(1452.71) (1381.02)
Non-Novice Voter [1,0] 0.12 0.10
(0.32) (0.31)
Per capita expenditure (log) 12.88 13.60
(0.67) (0.54)
Community Characteristics
Living in Urban [1,0] 0.48 0.54
(0.50) (0.50)
Terminal [1,0] 0.25 0.21
(0.43) (0.41)
Market [1,0] 0.40 0.39
(0.49) (0.49)
District Centre [1,0] 0.02 0.03
(0.15) (0.16)
Province Centre [1,0] 0.01 0.01
(0.10) (0.08)
High Belief in the same Religion 0.61 0.62
(0.49) (0.49)
Observations 19492 18014
Source: Author’s own calculation based on the IFLS 2007 and 2014 data using STATA

As shown in Table 1, as many as 79% voters in 2007 and 78% voters
in 2014 considered choosing a candidate in local head elections based on
religion consideration. Meanwhile, 41% of voters in 2007 and 59% of voters in
2014 chose the candidates based on gender in local head elections. Samples
from individuals’ characteristics comprise female (55%) and male (45%)
voters. In addition, 48% of voters in 2007 and 54% of voters in 2014 lived
in urban areas. Highest level of education achieved by samples was mostly
the elementary level (45%). Besides using individual characteristics, we also
use community characteristics. As shown in Table 1, an average of 25% of
voters live near terminal/bus stop and on average of 39% of voters live near
the market.
METHODS
We modify the model used by Putra (2017) as follows:
(1)
This chapter uses 2 models where the dependent variable (y) is a dummy
variable individual preferences when voting local head elections, i.e.,
Consider Religion (Model 1), and Consider Gender (Model 2). Independent
variable is divided into 2 groups, i.e., individual characteristics (Ikit) and
community characteristics (Cjit). Also, this model includes a fixed effect to
control some macro shocks, i.e, year fixed effect (Dt) and island fixed effect
(Smd). While individual characteristics consist of education, sex, religion,
age, and marital status, education is grouped into 5 level, no schooling
until university. Community characteristics consist of household per capita
expenditure, living in urban, near market, terminal, living near provinces or
district centre, and high belief in the same religion.
We estimate the model using pooled-logistic-regression to measure
the influence of socioeconomic characteristics on the probability of a person
having voting preferences by considering religion and gender. Our estimate
is expected to show whether these socioeconomic characteristics associated
with their voting behaviour. The analysis uses various specifications to check
whether the results are robust. As our unit of analysis is at the individual
level, we use restricted-unrestricted mechanism or plug-and-play some
of the community characteristics from the models and see the stability
of the estimated marginal effect of the individual-level characteristics
variables (Appendix 1 and Appendix 2). In addition, we also provide some
heterogeneity analysis to observe if there are differences in the estimates due

to gender of the individuals, island group where the individuals live, as well
as the socio-economic status of the individuals.
RESULTS
Results of marginal effect from Table 2 shows that the level of voter’s
education affects the probability of voter’s religion consideration in choosing
a candidate for local head elections. The pooled logistic results show that
voters who have the highest education in Junior, Senior, and University
tend to have less probability each of 5.15, 7.61, and 7.55 p.p. (percentage
point) in considering candidate’s religion for local head than voters who did
not attend school. This negative effect means that individuals do not have
tendency to choose a leader based on the same religion status as theirs (Nida
and Fahmi 2017).

Table 2. Marginal Effect
Pooled Logistic Regression
VARIABLES
Equality of Religion Gender Consideration
Per capita expenditure (log) -0.00871*** 1.19e-05
(0.00336) (0.00416)
Level of Education
Elementary School [1,0] -0.00249 0.0304***
(0.00857) (0.0104)
Junior High School [1,0] -0.0515*** -0.0512***
(0.00979) (0.0120)
Senior High School [1,0] -0.0761*** -0.0881***
(0.00985) (0.0120)
University [1,0] -0.0755*** -0.0952***
(0.0115) (0.0140)
Marital Status [1,0] 0.0129** 0.00181
(0.00513) (0.00635)
Islam [1,0] 0.311*** 0.241***
(0.00614) (0.00921)
Female [1,0] 0.0527*** 0.0117**
(0.00404) (0.00500)
Age 0.00539*** 0.0104***
(0.000908) (0.00109)
Age (squared) -3.69e-05*** -8.56e-05***
(9.65e-06) (1.13e-05)
Non-Novice Voter [1,0] 0.0441*** 0.0444***
(0.00851) (0.0111)
Living in Urban [1,0] -0.0275*** -0.0291***
(0.00453) (0.00547)
Terminal [1,0] 0.00681 -0.00270
(0.00521) (0.00661)
Market [1,0] -0.0397*** -0.0172***
(0.00440) (0.00551)
District Centre [1,0] 0.0531*** -0.0107
(0.0127) (0.0166)
Province Centre [1,0] -0.0693*** 0.0646**
(0.0211) (0.0290)
Community with High Belief in the same 0.0231*** 0.0140***
Religion (0.00420) (0.00526)
Observations 37,506 37,506

Notes : [1,0] means (1 = Yes; 0 = Otherwise); Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Source : Author’s own calculation based on the IFLS data of 2007 and 2014 using STATA

Meanwhile, marital status, sex, religion (Islam), age, and non-novice
voter affect the probability of voter’s religion consideration in choosing a
candidate for local head elections. Voters who are married tend to have a
1.29 p.p. higher probability to consider candidate’s religion for local head
elections than single or divorced one. Moslem voters significantly have a
higher probability to consider candidate’s religion 31.1 p.p. than non-
Moslem voters. The findings are similar to those by Botterman and Hooghe
(2012) and Nida and Fahmi (2017). Female has a higher probability 5.27
p.p. to consider the candidate’s religion for local head elections than male.
When age increases by 1 year, the probability in considering the religion of
the candidate for local head elections increases by 0.54 p.p., ceteris paribus.
Respondents who are not novice voters or have at least voted once have
a significant influence in considering the candidate’s religion in local head
elections, which is 4.41 p.p. higher than non-novices’ voters or who have
never voted before.
Based on community characteristics, voters who live in urban areas
have significantly 2.75 p.p. less probability of considering the candidate’s
religion in local head elections than those who live in rural. Voters living
near market have less probability (3.97 p.p.) of using religion consideration
in choosing a candidate in local head elections than those living far from
market. Voters who live in province centre have less probability (6.93 p.p.) of
considering the candidate’s religion in local head elections than voters who
do not live in the province centre.
Furthermore, community with “a high level of belief” in the same religion
living in the same area affects the probability of having a preference based
on religion consideration in an election. In addition, voters who live in areas
with a low level of trust in each other (only believe in the same religion) will
have a higher probability (2.31 p.p.) of considering the candidate’s religion
in local head elections than those living in areas with a high level of mutual
trust (heterogeneous).
Regarding gender consideration, Table 2 shows that education has
an influence on the probability of considering candidate’s gender in local
head elections. Voters who have education attainment in Junior, Senior,
and University tend to have less probability—i.e., 5.12, 8.81, and 9.52 p.p.,
respectively—of using gender consideration in choosing a candidate in local
head elections than those who did not attend school.

Moslem voters have higher probability (24.1 p.p.) of using gender
consideration in choosing a candidate in local head elections than non-
Moslem voters. Also, female voters have higher probability (1.17 p.p.) of
considering candidate’s gender for local head elections than male voters.
Meanwhile, non-novice (beginner) voters have higher probability (4.44 p.p.)
of considering candidate’s gender for local head elections than those who
have never voted.
Community characteristics variable significantly affects candidate’s
gender in local head elections. Voters who live in urban areas tend to have
less probability (2.91 p.p.) of considering candidate’s gender in local head
elections than those who live in rural areas. Voters living near market
have less probability (1.72 p.p.) of using gender consideration in choosing
a candidate for local head elections than voters who live far from market.
Also, voters living in province centre will have higher probability (6.46 p.p.)
of considering candidate’s gender for local head elections than voters who
live far from provincial capital.
Various specifications of our model to check whether our results are
robust are presented in Appendices. As shown in Appendix 1, the regression
results are robust because virtually both the marginal effects and associated
t-statistics for the individual level variables do not change when some
community characteristics are restrictive in the models.
HETEROGENEITY ANALYSIS
This chapter also present several heterogeneities analyses that are divided
into 3 sections: (1) By sex, between male and female voters, covering samples
of 17,465 female voters and 20,041 male voters; (2) by Island, which compares
voters from Java Island and Off Java Island, the Java Island including DKI
Jakarta, West Java, Central Java, Special Region of Yogyakarta, and East
Java. Outer Java Island are Bali, West Nusa Tenggara, South Kalimantan,
South Sulawesi, North Sumatra, West Sumatra, South Sumatra, and
Lampung; and, (3) by income group, which classifies voters as Poor, when
Per Capita Expenditure is less than poverty line and Not Poor when Per
Capita Expenditure is more than poverty line.
Table 3 shows more results of male and female voters having the
probability of using religion and gender consideration in choosing a
candidate for local head elections. In general, both male and female voters
have a similar preference in terms of religion consideration when choosing

a candidate for local head elections. Only one variable is found to have the
different relationship as indicated in Table 2 in which voters who live near
terminal have a non-significant result when considering candidate’s religion
for local head elections, but in Table 3 the results of female voters who live
near the terminal are significant (1.64 p.p.), which shows higher probability of
using religion consideration in choosing a candidate for local head elections.

Table 3. Regression Results (by Sex)
Equality of Religion Gender Consideration
VARIABLES
Poor Not Poor Poor Not Poor
Per capita expenditure (log) -0.0777*** -0.00733* -0.105*** -0.00148
(0.0277) (0.00378) (0.0358) (0.00471)
Level of Education
Elementary School [1,0] -0.0187 -6.18e-05 -0.0164 0.0361***
(0.0257) (0.00907) (0.0320) (0.0111)
Junior High School [1,0] -0.0357 -0.0518*** -0.0950** -0.0461***
(0.0309) (0.0103) (0.0398) (0.0126)
Senior High School [1,0] -0.0197 -0.0783*** -0.0765* -0.0850***
(0.0319) (0.0103) (0.0424) (0.0126)
University [1,0] 0.0571 -0.0783*** -0.203** -0.0894***
(0.0483) (0.0120) (0.0795) (0.0145)
Marital Status [1,0] 0.0220 0.0123** -0.00845 0.00276
(0.0184) (0.00534) (0.0244) (0.00658)
Islam [1,0] 0.240*** 0.316*** 0.217*** 0.243***
(0.0255) (0.00636) (0.0439) (0.00947)
Female [1,0] 0.0553*** 0.0524*** -0.0105 0.0135***
(0.0146) (0.00420) (0.0193) (0.00517)
Age 0.00801*** 0.00517*** 0.0121*** 0.0101***
(0.00310) (0.000948) (0.00397) (0.00113)
Age (squared) -5.90e-05* -3.50e-05*** -0.000111*** -8.15e-05***
(3.20e-05) (1.01e-05) (4.00e-05) (1.18e-05)
Non-Novice Voter [1,0] 0.118*** 0.0392*** 0.0744* 0.0408***
(0.0316) (0.00884) (0.0419) (0.0115)
Living in Urban [1,0] -0.0177 -0.0274*** -0.0116 -0.0289***
(0.0168) (0.00472) (0.0220) (0.00567)
Terminal [1,0] 0.0280 0.00534 0.0495* -0.00496
(0.0213) (0.00538) (0.0282) (0.00680)
Market [1,0] -0.0364** -0.0399*** -0.0475** -0.0151***
(0.0166) (0.00457) (0.0222) (0.00569)
District Centre [1,0] 0.0235 0.0542*** -0.186 -0.00829
(0.103) (0.0129) (0.155) (0.0168)
Province Centre [1,0] - -
Community with High Belief in 0.0236 0.0235*** 0.0739*** 0.0107**

the same Religion (0.0163) (0.00435) (0.0224) (0.00541)
Province Centre [1,0] -0.0701*** 0.0615**
(0.0213) (0.0290)
Observations 2,498 35,008 2,498 35,008

The first model shows that voters who have the highest level of education
in Junior, Senior, and University tend to have less probability of 5.15, 7.61,
dan 7.55 p.p. respectively in considering candidate’s religion for local head
election than those who did not attend school. This relationship is statistically
significant in the sample voters who are not poor. Not poor voters who have
the highest level of education in Junior, Senior, and University tend to have
less probability each of 5.18, 7.83, dan 7.83 p.p. in considering candidate’s
religion for local head than not poor voters who did not attend school.
Regarding individual characteristics, i.e., sample voters’ religion and
gender considerations in choosing a candidate for local head elections, the
regeression result shows that not poor voters who are married and not poor
female voters have the same relationship with the first model.
Based on community characteristics, the regression result shows that
voters living in urban areas tend to have less probability of using religion
(2.75 p.p.) and gender (2.91 p.p.) consideration in choosing a candidate,
meaning that this relationship is statistically significant in the samples of
poor voters.
CONCLUSION
Voters mainly consider two factors—i.e., religion and gender—in choosing
local head candidates. As stated earlier, this chapter aims to investigate and
measure the influence of socioeconomic characteristics on the probability of
a person having preferences based on religion and gender considerations
in voting for a candidates in local head elections in Indonesia. This chapter
employs IFLS data of 2007 and 2014 and pooled-logistic model to measure
the influence of individual/community characteristics in considering
religion and gender.
The results found that socioeconomic factors influence voters’ preference
in considering candidates’ religion and gender in local head elections,
especially in 13 selected provinces. The finding also shows that voters’
level of education and those who live in urban have negative significance
to considering candidates’ religion and gender. Moreover, Moslem-female
voters who live in district and province centre, and community who have
“high level of belief” in the same religion tend to consider candidates’
religion and gender in local head elections. It can be concluded that a voter
who has the same social characteristics will have a higher probability of
choosing candidates with the same social characteristics as theirs.

The regression results are robust and unbiased because both the marginal
effects and associated t-statistics for the individual level variables are stable
even when some community characteristics are restrictive in the models.
Meanwhile, the heterogeneity analysis indicates that sex (gender) is not
the main factor for voters’ preferences when considering candidate’s religion
for local head elections. Only one variable shows the different relationship,
i.e., female voters who live near the terminal have significantly 1.64 p.p.
higher probability of considering the candidate’s religion in local head
elections. In addition, different from the first model, female voters who live
in district centre have a higher probability in considering candidate’s gender,
compared to female voters who did not. On the other hand, male voters who
live in district centre have less probability in considering candidate’s gender
in local head elections than male voters who do not live in district centre.
The policy implication of our finding is that voters in general need
to be educated in order to know well that the right choice of local leader
candidates should be based on the candidates’ capability and leadership in
managing the government. This understanding may be achieved through
improving knowledge and democracy education to the society, as the higher
the education, the less likely that voters choose candidates solely based on
the latter’s religion or gender. Attention should also be paid to voters living
in rural areas, as they are more likely to choose candidates from the same
socio-economic characteristics as theirs.
REFERENCES
Akbar, I. 2017. Pilkada Serentak Dan Geliat Dinamika Politik Dan
Pemerintahan Lokal Indonesia. CosmoGov, 2 (1): 95. doi: 10.24198/
cosmogov.v2i1.11852.
Bhue, G. S. et al. 2014. Does Education Affect Voting Preferences? Evidence
from Voting Records in Indian Elections. SSRN Electronic Journal. doi:
10.2139/ssrn.2396784.
Botterman, S. and M. Hooghe. 2012. Religion and voting behaviour in
Belgium: An analysis of the relation between religious beliefs and
Christian Democratic voting. Acta Politica. Nature Publishing Group,
47(1), pp. 1–17. doi: 10.1057/ap.2011.11.
Campbell, D. E., Green, J. C. and Layman, G. C. 2011. The party faithful:
Partisan images, candidate religion, and the electoral impact of party

identification. American Journal of Political Science, 55 (1): 42–58. doi:
10.1111/j.1540-5907.2010.00474.x.
Centre for Strategic and International Studies (CSIS). 2017. Kenaikan Elektoral
& Kepuasan Publik Metodologi. (September). Available at: https://
www.csis.or.id/uploaded_file/event/pemaparan_hasil_survei_opini_
publik_tiga_tahun_pemerintahan_joko_widodo___jusuf_kalla__
kinerja_pemerintahan_dan_peta_politik_nasional__notulen.pdf.
Donald, S. G. and D.S. Hamermesh. 2006. ‘What is discrimination? Gender
in the American economic association, 1935-2004.’ American Economic
Review, 96 (4): 1283–1292. doi: 10.1257/aer.96.4.1283.
Giné, X. and G. Mansuri. 2018. Together we will: Experimental evidence on
female voting behaviour in Pakistan. American Economic Journal: Applied
Economics, 10 (1): 207–235. doi: 10.1257/app.20130480.
Heath, O. 2013. Policy representation, social representation and class voting
in Britain. British Journal of Political Science, 45 (1): 173–193. doi: 10.1017/
S0007123413000318.
Heath, O., G. Verniers, and S. Kumar. 2015. Do Moslem voters prefer Moslem
candidates? Co-religiosity and voting behaviour in India. Electoral
Studies. Elsevier Ltd, 38, pp. 10–18. doi: 10.1016/j.electstud.2015.01.005.
Hendrik, D. 2010. Variabel-variabel yang Mempengaruhi Rendahnnya
Partisipasi Politik Masyarakat dalam Pilkada Walikota dan Wakil Walikota
Padang Tahun 2008. Jurnal Demokrasi, 9: 137–148. Available at: http://
ejournal.fip.unp.ac.id/index.php/jd/article/view/1421%5Cnhttp://
download.portalgaruda.org/article.php?article=101259&val=1511.
Howell, J. 2006. Women’s political participation in China: In whose
interests elections?. Journal of Contemporary China, 15 (49): 603–619. doi:
10.1080/10670560600836614.
Kagema, D. N. 2019. The Aftermath of the 2007/08 Post-Election Violence
in Kenya and the Role of Religion. International Journal of Philosophy and
Theology (IJPT), 7 (1): 70–82. doi: 10.15640/ijpt.v7n1a9.
Martinus, H. 2013. Analisis Perilaku Pemilih pada Pemilihan Gubernur
DKI Jakarta Periode 2012-2017. Humaniora, 4(1): 58. doi: 10.21512/
humaniora.v4i1.3418.
Marwah, S. 2016. Representasi Perempuan dalam Pemilihan Bupati dan
Wakil Bupati Serentak di Jawa Tengah Tahun 2015. Jurnal Politik, 1 (2).

doi: 10.7454/jp.v1i2.17.
Mukarom, Z. 2008. Perempuan dan Politik: Studi Komunikasi Politik tentang
Keterwakilan Perempuan di Legislatif.’ Mediator: Jurnal Komunikasi, 9
(2): 257–270. doi: 10.29313/mediator.v9i2.1125.
Nida, R. and M. Fahmi. 2017. “Candidate Religion and Individual Voting
Behavior in Indonesia.” Working Paper in Economics and Development
Studies 201702.
Putra, H. 2017. Determinants of Vote Buying in Local Head Election in
Indonesia. Jurnal Bina Praja, 9 (2): 205–218. doi: 10.21787/jbp.09.2017.205-
218.
Ramadlan, F. S. and R. Masykuri. 2018. Marketing Isu Agama Dalam
Pemilihan Kepala Daerah di Indonesia 2015-2018. Jurnal Penelitian
Politik, 15 (2): 249. doi: 10.14203/jpp.v15i2.753.
Rochmi, M. N. 2018. Tingkat partisipasi politik Pilkada turun. https://
beritagar.id/artikel/berita/tingkar-partisipasi-politik-pilkada-turun
Strachan, A. L. 2014. “Voting behaviour in Indonesia: Impact of information
and performance.” Research Report
https://www.gov.uk/research-for-development-outputs/voting-
behaviour-in-indonesia-impact-of-information-and-performance-
gsdrc-helpdesk-research-report-1075
Strauss, J., F. Witoelar, and B. Sikoki. 2016. The Fifth Wave of the Indonesia
Family Life Survey: Overview and Field Report: Volume 1, doi: 10.7249/
wr1143.1.
Suyatno, S. 2016. Pemilihan Kepala Daerah (Pilkada) dan Tantangan
Demokrasi Lokal di Indonesia. Politik Indonesia: Indonesian Political
Science Review, 1 (2): 212. doi: 10.15294/jpi.v1i2.6586.
Tilley, J. 2014. “We Don’t Do God”? Religion and Party Choice in Britain.
British Journal of Political Science, 45 (4): 907–927. doi: 10.1017/
S0007123414000052.
Winkleby, M. A. et al. 1992. Socioeconomic status and health: How education,
income, and occupation contribute to risk factors for cardiovascular
disease. American Journal of Public Health, 82 (6): 816–820. doi: 10.2105/
AJPH.82.6.816.

APPENDICES
Appendix 1. Robustness Check of Religion Equality
(1) (2) (3) (4) (5) (6) (7)
Equality of Religion -0.0578** -0.0710** -0.0563* -0.0628** -0.0548* -0.0588** -0.0618**
[1,0]
Per capita (-2.59) (-3.21) (-2.53) (-2.82) (-2.46) (-2.64) (-2.78)
expenditure (log)
Level of Education
Elementary School -0.0187 -0.0490 -0.0195 -0.0132 -0.0192 -0.0203 -0.0198
[1,0] (-0.29) (-0.76) (-0.30) (-0.20) (-0.30) (-0.31) (-0.31)
Junior High School -0.353*** -0.403*** -0.354*** -0.356*** -0.351*** -0.357*** -0.357***
[1,0] (-4.99) (-5.72) (-5.00) (-5.03) (-4.96) (-5.04) (-5.04)
Senior High School -0.503*** -0.576*** -0.502*** -0.501*** -0.496*** -0.505*** -0.506***
[1,0] (-7.14) (-8.31) (-7.13) (-7.12) (-7.05) (-7.18) (-7.19)
University [1,0] -0.499*** -0.573*** -0.499*** -0.489*** -0.485*** -0.505*** -0.504***
(-6.33) (-7.36) (-6.33) (-6.21) (-6.16) (-6.41) (-6.40)
Marital Status [1,0] 0.0855* 0.0981** 0.0849* 0.0832* 0.0859* 0.0863* 0.0854*
(2.51) (2.89) (2.49) (2.44) (2.52) (2.53) (2.51)
Islam [1,0] 2.066*** 2.047*** 2.068*** 2.076*** 2.053*** 2.060*** 2.095***
(46.07) (45.84) (46.13) (46.36) (45.99) (46.00) (46.97)
Female [1,0] 0.350*** 0.344*** 0.350*** 0.349*** 0.351*** 0.350*** 0.349***
(13.00) (12.80) (13.01) (12.98) (13.04) (13.01) (12.97)
Age 0.0357*** 0.0340*** 0.0358*** 0.0357*** 0.0357*** 0.0357*** 0.0356***
(5.93) (5.65) (5.93) (5.93) (5.92) (5.93) (5.91)
Age (squared) -0.000245*** -0.000232*** -0.000245*** -0.000247*** -0.000243*** -0.000245*** -0.000244***
(-3.82) (-3.63) (-3.83) (-3.86) (-3.79) (-3.83) (-3.82)

Non-Novice Voter 0.293*** 0.294*** 0.293*** 0.297*** 0.292*** 0.294*** 0.293***
[1,0] (5.18) (5.21) (5.19) (5.26) (5.17) (5.21) (5.18)
Living in Urban [1,0] -0.182*** -0.176*** -0.212*** -0.178*** -0.187*** -0.200***
(-6.06) (-5.93) (-7.09) (-5.90) (-6.21) (-6.68)
Terminal [1,0] 0.0452 0.00995 -0.0562 0.0574 0.0462 0.0322
(1.31) (0.29) (-1.72) (1.67) (1.34) (0.93)
Market [1,0] -0.263*** -0.283*** -0.251*** -0.265*** -0.265*** -0.263***
(-8.99) (-9.73) (-9.06) (-9.05) (-9.07) (-8.97)
District Centre [1,0] 0.353*** 0.336*** 0.362*** 0.365*** 0.277*** 0.366***
(4.18) (3.96) (4.30) (4.29) (3.44) (4.34)
Province Centre [1,0] -0.460** -0.499*** -0.462*** -0.488*** -0.284* -0.495***
(-3.28) (-3.56) (-3.29) (-3.49) (-2.13) (-3.53)
Community with 0.153*** 0.171*** 0.151*** 0.152*** 0.157*** 0.157***
High Belief in the (5.49) (6.17) (5.41) (5.45) (5.62) (5.64)
same Religion
Constant 0.426 0.607 0.403 0.385 0.377 0.440 0.585

(1.32) (1.89) (1.25) (1.20) (1.17) (1.37) (1.83)
Observations 37506 37506 37506 37506 37506 37506 37506
t statistics in parentheses
* p < 0.05, ** p < 0.01, *** p < 0.001

Appendix 2. Robustness Check of Gender Consideration
Variable (1) (2) (3) (4) (5) (6) (7)
Gender 0.0000521 -0.00957 -0.000263 -0.00189 -0.000301 0.000814 -0.00170
Consideration [1,0]
Per capita (0.00) (-0.53) (-0.01) (-0.10) (-0.02) (0.04) (-0.09)
expenditure (log)
0.132** 0.113* 0.132** 0.134** 0.132** 0.132** 0.131**
Level of Education
Elementary School (2.91) (2.49) (2.92) (2.95) (2.91) (2.93) (2.90)
[1,0] -0.221*** -0.254*** -0.221*** -0.221*** -0.221*** -0.219*** -0.222***
Junior High School (-4.25) (-4.93) (-4.25) (-4.25) (-4.26) (-4.22) (-4.28)
[1,0] -0.382*** -0.431*** -0.382*** -0.380*** -0.383*** -0.380*** -0.383***
Senior High School (-7.34) (-8.42) (-7.34) (-7.31) (-7.35) (-7.30) (-7.36)
[1,0] -0.413*** -0.462*** -0.413*** -0.409*** -0.414*** -0.408*** -0.414***
University [1,0] (-6.78) (-7.68) (-6.78) (-6.72) (-6.82) (-6.71) (-6.80)
Marital Status [1,0] 0.00797 0.0172 0.00813 0.00712 0.00791 0.00754 0.00790
(0.29) (0.62) (0.29) (0.26) (0.28) (0.27) (0.28)
Islam [1,0] 1.060*** 1.048*** 1.059*** 1.066*** 1.062*** 1.063*** 1.071***
(25.37) (25.12) (25.38) (25.52) (25.45) (25.46) (25.76)
Female [1,0] 0.0515* 0.0473* 0.0514* 0.0515* 0.0514* 0.0514* 0.0513*
(2.34) (2.15) (2.34) (2.35) (2.34) (2.34) (2.33)
Age 0.0458*** 0.0445*** 0.0458*** 0.0458*** 0.0458*** 0.0458*** 0.0458***
(9.56) (9.30) (9.56) (9.57) (9.56) (9.56) (9.56)
Age (squared) -0.000376*** -0.000366*** -0.000376*** -0.000376*** -0.000376*** -0.000376*** -0.000376***
(-7.52) (-7.34) (-7.52) (-7.54) (-7.53) (-7.52) (-7.53)

Non-Novice Voter 0.195*** 0.196*** 0.195*** 0.197*** 0.195*** 0.194*** 0.195***
[1,0] (4.00) (4.04) (4.00) (4.04) (4.00) (3.99) (4.01)
Living in Urban -0.128*** -0.130***
-0.135*** -0.129*** -0.126*** -0.135***
[1,0] (-5.32) (-5.45)
(-5.63) (-5.36) (-5.24) (-5.65)
Terminal [1,0] -0.0119 -0.0356 -0.0433 -0.0132 -0.0126 -0.0182
(-0.41) (-1.24) (-1.59) (-0.46) (-0.43) (-0.63)
Market [1,0] -0.0754** -0.0874*** -0.0789*** -0.0754** -0.0750** -0.0749**
(-3.11) (-3.62) (-3.47) (-3.11) (-3.10) (-3.09)
District Centre [1,0] -0.0469 -0.0685 -0.0490 -0.0466 -0.00173 -0.0426
(-0.64) (-0.94) (-0.67) (-0.64) (-0.02) (-0.58)
Province Centre 0.284* 0.258* 0.284* 0.281* 0.261* 0.271*
[1,0] (2.23) (2.02) (2.23) (2.21) (2.14) (2.13)
Community with 0.0614** 0.0749** 0.0622** 0.0607** 0.0611** 0.0595**
High Belief in the (2.66) (3.27) (2.70) (2.63) (2.64) (2.58)
same Religion
Constant -0.803** -0.672** -0.798** -0.808** -0.797** -0.814** -0.738**

(-3.06) (-2.58) (-3.05) (-3.08) (-3.04) (-3.10) (-2.83)
Observations 37506 37506 37506 37506 37506 37506 37506
t statistics in parentheses
* p < 0.05, ** p < 0.01, *** p < 0.001

Appendix 3. Robustness Check of Religion Equality and Gender
Consideration
(1) (2)
VARIABLES Equality of Gender
Religion Consideration
Per capita expenditure (log) -0.00871** 1.19e-05
(0.00339) (0.00417)
Level of Education
Elementary School [1,0] -0.00249 0.0304***
(0.00861) (0.0104)
Junior High School [1,0] -0.0515*** -0.0512***
(0.00981) (0.0120)
Senior High School [1,0] -0.0761*** -0.0881***
(0.00982) (0.0120)
University [1,0] -0.0755*** -0.0952***
(0.0114) (0.0140)
Marital Status [1,0] 0.0129** 0.00181
(0.00515) (0.00635)
Islam [1,0] 0.311*** 0.241***
(0.00617) (0.00944)
Female [1,0] 0.0527*** 0.0117**
(0.00403) (0.00501)
Age 0.00539*** 0.0104***
(0.000928) (0.00109)
Age (squared) -3.69e-05*** -8.56e-05***
(9.92e-06) (1.14e-05)
Non-Novice Voter [1,0] 0.0441*** 0.0444***
(0.00853) (0.0111)
Living in Urban [1,0] -0.0275*** -0.0291***
(0.00452) (0.00546)
Terminal [1,0] 0.00681 -0.00270
(0.00513) (0.00662)
Market [1,0] -0.0397*** -0.0172***
(0.00434) (0.00550)
District Centre [1,0] 0.0531*** -0.0107
(0.0124) (0.0168)
Province Centre [1,0] -0.0693*** 0.0646**
(0.0215) (0.0298)
Community with High Belief in the same Religion 0.0231*** 0.0140***
(0.00422) (0.00525)
Observations 37,506 37,506

Standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1

Appendix 4. Heterogeneity Analysis of Regression Result (by Per capita
expenditure)
Equality of Religion
VARIABLES
Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5
Level of Education
Elementary School [1,0] 0.0160 -0.0108 -0.0407** 0.0233 -0.0125
(0.0143) (0.0166) (0.0176) (0.0254) (0.0353)
Junior High School [1,0] -0.0416** -0.0573*** -0.0971*** -0.0206 -0.0282
(0.0174) (0.0191) (0.0202) (0.0277) (0.0374)
Senior High School [1,0] -0.0632*** -0.0714*** -0.135*** -0.0538** -0.0587
(0.0183) (0.0193) (0.0201) (0.0271) (0.0362)
University [1,0] 0.0120 -0.0721*** -0.141*** -0.0387 -0.0905**
(0.0278) (0.0265) (0.0246) (0.0287) (0.0369)
Marital Status [1,0] 0.0170* 0.00767 0.0201* 0.00112 0.0135
(0.00974) (0.0104) (0.0113) (0.0127) (0.0153)
Islam [1,0] 0.254*** 0.294*** 0.310*** 0.325*** 0.387***
(0.0132) (0.0135) (0.0140) (0.0141) (0.0160)
Female [1,0] 0.0518*** 0.0512*** 0.0523*** 0.0523*** 0.0583***
(0.00768) (0.00816) (0.00885) (0.00982) (0.0123)
Age 0.00528*** 0.00827*** 0.00138 0.00604*** 0.00558**
(0.00168) (0.00181) (0.00204) (0.00234) (0.00277)
Age (squared) -3.52e-05** -6.82e-05*** 2.99e-06 -4.13e-05 -3.87e-05
(1.76e-05) (1.91e-05) (2.20e-05) (2.53e-05) (2.95e-05)
Non-Novice Voter [1,0] 0.0682*** 0.0632*** 0.0139 0.0225 0.0360
(0.0164) (0.0170) (0.0186) (0.0207) (0.0264)
Living in Urban [1,0] -0.0167** -0.0391*** -0.0147 -0.0497*** -0.0312**
(0.00850) (0.00871) (0.00987) (0.0115) (0.0154)
Terminal [1,0] 0.0210* 0.00949 0.00919 -0.0122 0.0103
(0.0109) (0.0110) (0.0115) (0.0122) (0.0143)
Market [1,0] -0.0436*** -0.0393*** -0.0447*** -0.0295*** -0.0417***
(0.00842) (0.00884) (0.00978) (0.0108) (0.0133)
District Centre [1,0] -0.0203 0.0508 0.0959*** 0.0450 0.0592**
(0.0296) (0.0319) (0.0293) (0.0275) (0.0296)
Province Centre [1,0] -0.138** -0.147*** -0.0248 0.0269 -0.0937*
(0.0696) (0.0425) (0.0499) (0.0492) (0.0484)
Community with High 0.0233*** 0.0169** 0.0392*** 0.0171* 0.0210*
Belief in the same Religion (0.00822) (0.00855) (0.00910) (0.0102) (0.0127)
Observations 9,637 8,704 7,925 6,634 4,606

*** p<0.01, ** p<0.05, * p<0.1

Appendix 5. Heterogeneity Analysis of Regression Result (by Per capita
expenditure)
Gender Consideration
VARIABLES
Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5
Level of Education
Elementary School [1,0] 0.0481*** 0.0434** -0.00950 0.0133 0.00328
(0.0173) (0.0209) (0.0241) (0.0295) (0.0411)
Junior High School [1,0] -0.0518** -0.0350 -0.0986*** -0.0633* -0.0314
(0.0212) (0.0242) (0.0272) (0.0325) (0.0438)
Senior High School [1,0] -0.0688*** -0.0717*** -0.149*** -0.104*** -0.0872**
(0.0220) (0.0245) (0.0270) (0.0316) (0.0423)
University [1,0] -0.0546 -0.0655** -0.164*** -0.0741** -0.126***
(0.0374) (0.0330) (0.0310) (0.0337) (0.0427)
Marital Status [1,0] -0.00357 0.00400 -0.00636 -0.00628 0.0276
(0.0125) (0.0134) (0.0139) (0.0153) (0.0175)
Islam [1,0] 0.184*** 0.258*** 0.236*** 0.247*** 0.246***
(0.0208) (0.0211) (0.0201) (0.0208) (0.0228)
Female [1,0] 0.0249** 0.0114 -0.00833 0.00823 0.0198
(0.00984) (0.0104) (0.0109) (0.0119) (0.0142)
Age 0.0107*** 0.0136*** 0.00550** 0.0118*** 0.00962***
(0.00208) (0.00224) (0.00235) (0.00276) (0.00315)
Age (squared) -9.03e-05*** -0.000115*** -3.41e-05 -0.000102*** -7.76e-05**
(2.13e-05) (2.34e-05) (2.46e-05) (2.93e-05) (3.32e-05)
Non-Novice Voter [1,0] 0.0700*** 0.0586** 0.00127 0.0267 0.0500
(0.0215) (0.0229) (0.0241) (0.0270) (0.0323)
Living in Urban [1,0] -0.0123 -0.0437*** -0.0271** -0.0500*** -0.0187
(0.0108) (0.0109) (0.0118) (0.0133) (0.0169)
Terminal [1,0] 0.0469*** -0.0278* 0.0124 -0.0351** -0.0201
(0.0143) (0.0143) (0.0144) (0.0152) (0.0165)
Market [1,0] -0.0327*** -0.0137 -0.00655 -0.00999 -0.0278*
(0.0110) (0.0114) (0.0121) (0.0131) (0.0152)
District Centre [1,0] -0.0457 0.00600 -0.00564 0.000127 -0.0337
(0.0464) (0.0447) (0.0340) (0.0336) (0.0346)
Province Centre [1,0] 0.0255 0.0265 0.172*** 0.0834 0.0287
(0.104) (0.0654) (0.0658) (0.0585) (0.0572)
Community with High 0.0497*** -0.00461 0.0171 -0.00451 0.00477
Belief in the same Religion (0.0107) (0.0110) (0.0113) (0.0124) (0.0146)
Observations 9,637 8,704 7,925 6,634 4,606

*** p<0.01, ** p<0.05, * p<0.1

Chapter 6
Heritability of Cognitive Ability and

Its Relation to Economic Achievement in
Adolescence
Siti Munawaroh and Ni Made Sukartini
INTRODUCTION
Cognitive ability is one of the components to increase individual capacity
and essentially plays a role in child development and their future life (Zahra
and Wisana 2019). According to Glewwe et al. (2017) and Cawley et al.
(2001), cognitive ability is also an indicator of individual achievement in
adult life, such as educational attainment, career, and wage outcomes in the
labour market. In their study, Acosta et al. (2020) reported that individuals
with higher cognitive abilities tend to have better jobs and higher incomes.
This finding implies that childhood cognitive ability could be as predictor
of individual for success in their later life, especially in terms of education
achievement and labour outcome. Therefore, investment on cognitive ability
should be prioritised to improve the development outcome.
Hewi and Saleh (2020) stated that the Programme for International
Student Assessment (PISA) measured in reading, mathematics, and science
can capture students’ cognitive abilities. These cognitive abilities can be seen
in the way students solve problems in each question area and the scores. Based
on international cognitive measurements through PISA, children’s cognitive
scores in Indonesia are still relatively low (Figure 1). Indonesian students’
abilities are still below the average international standard. Evidence can be
investigated in the 2018 PISA score report in which Indonesia ranked 73 out
of 79 countries (OECD 2019). Indonesia’s performance among other Asian
111
countries in literacy, mathematics, and science is still far behind compared
to China, Singapore, and Malaysia. In this regard, developing countries such
as Indonesia urgently need to improve cognitive abilities of children in order
to perform better.
Figure 1. PISA’s Scores in Some Asian Countries in 2018
800
549 569 551 555 591 590
600 487 489 489
393 419 426 440 438
371 379 393 415
400
200
0
The Average Singapura Thailand China Indonesia Malaysia
OECD Score
Reading Mathematics Science
Source: OECD, 2019

Schoon (2010) analysed the relationship between cognitive abilities
measured at age ten and educational attainment as an adult outcome
based on a study in the UK. The results of his study prove that there is a
positive relationship between cognitive abilities and educational outcomes.
The study of Heckman et al. (2006) empirically investigated the role of
cognitive and non-cognitive tests on adult performance such as attending
university and labour outcome. The cognitive test is measured by an index
of Arithmetic and Math score, reading, word knowledge and paragraph
comprehension. The non-cognitive test is measured by subjective report
regarding conscientiousness or diligence, self-control, and perseverance or
persistence. These authors claim that non-cognitive skills strongly influence
schooling decisions and wages in the labour market. In contrast, cognitive
skill is associated with schooling, employment, work experience, and choice
of occupation. Overall, they found that a higher cognitive test score was
strongly related to white-collar occupation.
On the other hand, according to BPS data (2021), the majority of
population aged 15 years and over who have participated in labour market
only have primary school or its equivalence at educational level. When the
data are broken down by education level, 26.11% of Indonesian workers
only graduated from elementary school, and 18.5% graduated from Junior
High School (SLTP). This figure indicates that workers in Indonesia are
dominated by unskilled labour. This also explains the reasons for the low
productivity and low wage of Indonesian labours . Therefore, for developing

countries like Indonesia, it is crucial to improve cognitive ability for gaining
individual better capacity and developing human resources accordingly
(Zahra and Wisana 2019).
The passing down of cognitive abilities from parents to children or their
offspring is also an essential factor in forming quality human capital (Conti
and Kopinska 2018). The study of Black, Devereux, and Salvanes (2009)
estimated the intergenerational transmission of IQ scores. The study results
indicate a hereditary factor—i.e., parents’ intelligence is inherited on to their
children. The authors conclude that a 10 percent increase in father’s IQ will
increase 3.2 % of son’s IQ. In this case, there is a tendency that son is most
likely to inherit his father’s academic intelligence.
Based on the literature review, several factors affect cognitive abilities,
which include among others: socioeconomic background of the family,
such as parents’ education (Woessmann 2004; Burhan et al. 2017; Wang
et al. 2020); parental income (Mayer 2002); educational expenses (Kang
2007); psychological health of parents (Mensah and Kiernan 2010); family
size (Black et al. 2010); location of residence (Alexopoulos 1997); and, child
characteristics, such as birth order (Black, Devereux, and Salvanesz 2011;
Monfardini and See 2016).
According to Burhan et al. (2017), parents’ education level has a positive
influence on children’s cognitive abilities. Parents with higher educational
attainment have a higher awareness of the quality of life, which can indirectly
improve their children’s cognitive abilities. Wang et al. (2020) estimated the
influence of parental education on children’s academic performance, using
data from a Chinese education panel survey. This study found that parental
education plays a positive role in children’s academic performance. Highly
educated parents tend to have higher educational expectations for their
children. Other socioeconomic status background variables, such as parental
income are also strong predictors of cognitive ability. Dahl and Lochner
(2012) used the OLS method to estimate the effect of parental income on
children’s cognitive achievement. The study results found that parental
income has a positive and significant relationship with children’s cognitive
abilities. An increase of USD1,000 in parental income associated with math
test scores and reading ability about six standard deviations above average.
Kang (2007) estimated the effect of education expenditure on children’s
cognitive abilities in South Korea. Education costs are measured through
private tutoring expenses, while cognitive abilities are measured by math
Heritability of Cognitive Ability and 113

Its Relation to Economic Achievement in Adolescence
and language test scores. Using the Instrumental Variable (IV) method in
his study, Kang found that a 10 percent increase in education spending
leads to a 0.56 percentile increase in language test score. Meanwhile, using
data from the Millennium Cohort Study covering 4,781 families in the UK,
Mensah and Kiernan (2010) estimated the influence of parental psychological
health on children’s cognitive abilities. The study results showed a negative
correlation between the psychological health of parents and children’s
cognitive achievement. Parents’ experience of high psychological pressure
will have an impact on their children’s lower cognitive abilities.
Other family socioeconomic backgrounds such as family size and
location of residence also affect cognitive abilities. Black, Devereux and
Salvanes (2010) focused on analysing the effect of family size on cognitive
abilities, using data from the male population of Norway. The estimation
results show a negative correlation between family size and children’s
cognitive abilities. In this study, it is found that the larger the size of the
family, the lower the cognitive ability of the children is. On the other hand,
Alexopoulos (1997) estimated differences in cognitive abilities as measured
by IQ between urban and rural children in Greece. The study shows that due
to better social and school environment, children’s ability scores in urban
areas are higher than in rural areas where education facilities are poor.
Black et al. (2011) and Heiland (2009) also state that the characteristics of
children in the form of birth order are one of the factors that affect cognitive
abilities. The study of Black, Devereux and Salvanesz (2011) found that
birth order has a negative and significant relationship to cognitive ability.
Children born earlier tend to have higher cognitive abilities (Heiland 2009).
It is because the eldest child tend to get better attention from his parents;
whereas children born later have to share everything—including meals and
parents’ attention—with their siblings.
Sundstrup et al. (2020) investigated the role of the individual’s
cognitive ability in early life and its association with labour outcome in
later life. The study utilised 10,171 labour participants in Copenhagen based
on Copenhagen Aging and Midlife Biobank (CAMB) data. These authors
measure labour outcomes in terms of risk of long-term sickness absence,
disability pension, early retirement, and unemployment. The authors claim
that cognitive ability, either low or high, were not associated with the risk of
any of the four labour market outcomes. Following Sundstrup et al. (2020),
this chapter investigates individual outcomes in later life based on their

cognitive test score in early life. The outcome in later life is measured by
school attainment and wage rate in the labour market. The questions to be
investigated are focused on these issues : “Do fathers inherit intelligence to
daughters? Do mothers inherit intelligence to sons?”.
The study of Sundstrup et al. (2020) investigated labour outcomes in
developed countries with a competitive labour market, while this chapter
investigates conditions in developing countries such as Indonesia. This
chapter utilises microdata from the 2000 Indonesia Family Life Survey (IFLS)
to measure individuals in early life, i.e., their cognitive test score, while
individual outcomes in later life are measured based on the 2014 IFLS data.
These IFLS data—both of 2000 and 2014—are employed, yet they are not used
as panel data for the analysis. IFLS data are longitudinal data, consisting of
long data from 1993 to 2014; however, cognitive tests are only available from
the 2000 data. The data from the 2007 IFLS are not used because the time
span between 2000 and 2007 is shorter than the span between IFLS data of
2000 (IFLS-3) and 2014 (IFLS-5). In other words, this chapter investigates the
cognitive ability of individuals at primary school age, i.e., below 13 years.
It used this information as a predictor of individual success in later life, i.e.,
14 years later in 2014 when these individuals reached their teenagers, about
25 years or more. In this age group, typically, most Indonesians already
participate in the labour market or pursue higher education. Therefore, this
chapter uses educational attainment and labour wages to measure individual
success in adult life.
This chapter is organised into three sections. Section 1 consists
of background and literature review. Section 2 describes the data and
methodology as well as the empirical model used to estimate the effect of
interest. Section 3 discusses the estimation results followed by conclusion
and suggestion section.
THE USE OF IFLS DATA

This chapter uses secondary data from the Indonesia Family Life Survey
(IFLS) provided by the RAND corporation. IFLS collect national wide
data, representing about 13 among 33 provinces in Indonesia. The samples
represent approximately 83% of Indonesia’s population consisting of more
than 30,000 individuals living in 13 out of the 27 provinces in Indonesia. The
IFLS survey consisted of five waves, including IFLS-1 carried out in 1993,
IFLS-2 in 1997 and 1998, IFLS-3 in 2000; IFLS-4 in 2007, and IFLS-5 in 2014

(RAND, 2018). The IFLS data used in this chapter are those of 2000 (IFLS-3)
and 2014 (IFLS-5). The IFLS-3 (2000), which is the first micro data providing
children cognitive ability, contains rich longitudinal data. We intentionally
did not use the IFLS-4 data (2007), because we considered the longer impact
of children’s cognitive ability on their later life. The data of IFLS-3 provide
cognitive score of individual in their childhood, and after 14 years, both
the educational and the labour market outcomes of the individuals whose
data are provided in the IFLS-3 can then be investigated in the IFLS-5. We
use IFLS and not other national micro data because only the IFLS data could
provide rich and longitudinal data. Besides, the data are publicly available.
Comprehensiveness regarding education and socioeconomic indicator
is the advantage of IFLS data. Respondents who were interviewed between
surveys were the same subjects to grasp the socio-economic phenomena of
respondents from one survey to the following survey. In addition, the IFLS
data contain cognitive aspects that are asked in a special questionnaire.
Figure 2. Forms of Cognitive Questions
Source: IFLS-3, 2000

The flow of sample selection is carried out in several stages. The first
stage collects the variables needed in IFLS-3 (2000). This data set is selected
for measuring individual cognitive ability in early life. The data sets of
individual and family characteristics are chosen as predictors. Data from
IFLS-5 (2014) are collected to gather information regarding individual
activities in adult life, either pursuing a higher degree or participating in the
labour market. This chapter combines cognitive test scores in early life to
predict adult life outcomes—i.e., educational attainment and labour wage.

The cognitive ability variable in this chapter used IFLS-3 data which contain
questions related to the cognitive ability test aimed at children aged 7 to 14
years. The forms of cognitive ability questions are presented in Figure 2.
The cognitive test consists of 12 verbal questions and five math
questions. The score measurement is based on Raven’s Progressive Matrices
(RPM) test, which gives one when the answer is correct and zeroes if the
answer is wrong. Next, add up the correct answers and then change them to
a scale of 100 (number of correct questions/total number of questions x 100).
There are also several reasons why children did not take the test (Table 1).
The reasons are as follows.
Table 1. Reasons for Not Taking EK1 Test
Reason IFLS-3
Refused 4
Cannot read 2
Unable to answer 0
Not enough time 0
Proxy respondent 0
Other 2
Couldnot be contacted 8
Total 16
Source: Author’s estimations based on the IFLS-3
The second stage is to collect independent variables in 2000, namely
family socioeconomic status and child characteristics. Variables of
socioeconomic status include income, education, number of household
members, the mental health of parents, expenditure for children’s education
costs, and location of residence. In contrast, the variable characteristics of
children include age, gender, and the birth order. The next stage is to collect
the variables used in 2014, namely the variables of educational attainment
and wages in the labour market where the child is estimated to be an adult
and has entered working age.
The missing data is dropped from all the variables collected during the
sample selection stage. The first stage is collecting samples data that consist
of a cognitive test in 2000. This data are combined with several data that
represent both individual and family characteristics. This chapter estimates
determinants of individual cognitive abilities based on their characteristics
and their parent’s educational attainment from this data set. At the second
stage, IFLS data of 2014 were collected to find out individuals with a

cognitive score in 2000 focusing more on individual information regarding
their educational attainment and labour wages. In other words, the second
stage predicts individual success in life based on the information of their
cognitive ability in earlier life. Combining these two steps of collecting data
results in about 603 observations.
THE OLS METHOD AND EMPIRICAL MODEL

The method used in this chapter is the Ordinary Least Squares (OLS) method.
OLS chooses which minimizes the sum of the residual squares for
a given sample (Wooldridge 2018). The OLS method is used to determine
the heritability of cognitive abilities by estimating the relationship between
parental education and children’s cognitive abilities. The equations of the
OLS model are formulated as follows:
(1)
In Model 1, cognitive scores are differentiated into two, namely female
cognitive scores and male cognitive scores. Variable represented by X1 is a
vector of socioeconomic background, consisting of parental age, the natural
logarithm of parental income, parental education divided into father and
mother education, parental psychological health, the natural logarithm of
education expenses, and the number of household members and location.
Variable represented by X2 is a vector of child characteristics, namely a
child’s age and birth order. Notation e is the error term.
The following OLS model equation estimates the relationship between
cognitive and educational attainment and work outcomes. The equation of
the model is as follows:
𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦 𝑜𝑜𝑜𝑜 𝑠𝑠𝑠𝑠ℎ𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑔𝑔𝑖𝑖𝑖𝑖 = 𝛽𝛽0 + 𝛽𝛽1 𝑎𝑎𝑖𝑖𝑖𝑖 + 𝛽𝛽2 𝑏𝑏𝑖𝑖𝑖𝑖 + 𝑒𝑒𝑖𝑖𝑖𝑖 and (2a)
𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 = 𝛽𝛽0 + 𝛽𝛽1 𝑎𝑎𝑖𝑖𝑖𝑖 + 𝛽𝛽2 𝑐𝑐𝑖𝑖𝑖𝑖 + 𝑒𝑒𝑖𝑖𝑖𝑖 (2b)
In this model, years of schooling is the length of education (years),

Lnwage is the natural logarithm of wages for 12 months, and the cognitive
score is the total score of cognitive abilities of children aged 7-14 years. In
Model 2a, variable a is the cognitive score variable. In contrast, b is a control
variable consisting of the child’s age, the natural logarithm of parent’s
income, and the natural logarithm of education expense. On the other hand,
the control variable in Model 2b is c, which comprises children’s age and
years of schooling.

CHILDREN’S COGNITIVE SCORE IN INDONESIA
In this study, cognitive ability was measured by children’s cognitive scores
obtained through the IFLS-3 EK book. Cognitive scores of children in
Indonesia classified according to Raven’s level are as follows:
Table 2. Children’s Cognitive Score in Indonesia in 2000
Level (Nur’aeni, Cognitive Score
Category (Raven, 1960) Frequency
2012) (Percentile)
55
Intelectually superior I 95+
(9.12)
Definitely above the average in 127
II 75-<95
intellectual capacity (21.06)
129
III+ 50-<75
(21.39)
Intellectually average
162
III- 25-<50
(26,86)
78
IV+ 10-<25
Definitely below average in (12.93)
intellectual capacity 23
IV- 5-<10
(3.81)
29
Intelectually defective V <5
(4.80)
Total 603
Source : Research data, processed
Note : The number in brackets is a percentage of the total sample
Table 2 shows that the cognitive scores of children aged 7-14 years in
Indonesia are categoried into several levels. A total of 9.12% occupied Level
I or the intellectually superior category. Level II or definitely above the
average in intellectual capacity accounted for 21.06%; whereas Level III or the
intellectually average category is the most occupied level, making up 48.25%.
Level IV or the below-average intellectual capacity category was occupied
by 16.74% of total children. Finally, as many as 4.80% of children occupy
Level V or the category of intellectually defective. Based on these levels, it
can be concluded that the cognitive score of children at the intellectually
superior and intellectually defective level is only small. Meanwhile, the
average cognitive score of children is at the intellectually average level.

Table 3. Tabulation of Cognitive Score and Years of Schooling
Category of Cognitive Years of schooling
Score 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 18
Intellectually superior 0 0 0 0 0 0 0 0 0 4 1 0 28 1 1 6 13 1
Definitely above the average
2 0 0 0 1 0 3 2 0 13 0 2 59 1 2 14 25 3
in intellectual capacity
Intellectually average (+) 1 1 0 0 0 0 5 4 2 12 0 1 60 1 0 13 29 0
Intellectually average (-) 4 0 0 0 1 1 17 2 1 23 2 2 75 1 4 13 16 0
Definitely below average in
0 1 1 1 2 2 8 2 1 19 1 1 30 0 0 3 6 0
intellectual capacity (+)
Definitely below average in
0 1 0 0 0 1 4 1 0 4 0 0 9 0 0 1 2 0
intellectual capacity (-)
Intellectually defective 1 0 0 2 1 1 8 0 0 3 0 0 5 0 0 3 5 0
Total 8 3 1 3 5 5 45 11 4 78 4 6 266 4 7 53 96 4
Source: Research data, processed
Table 3 presents a tabulation of the categories of cognitive scores
achieved when taking the 2000 cognitive test with educational outcomes
as measured by years of schooling. When viewed by all types, the average
length of schooling is 12 years or equivalent to senior high school. However,
the number of children achieving education is relatively different from each
category. In the intellectually superior, intellectually average, and definitely
below average in intellectual capacity (+), a relatively large number of
students have higher educational attainment. On the other hand, the
definitely below average in intellectual capacity (-) and intellectual defective
only a relatively small number achieve higher education. When viewed from
the number of children to attain education, it can conclude that children
with better cognitive have a better effect on their educational attainment in
adulthood.
Table 4 shows a statistical description of each variable used in the
estimation. It consists of the number of observations, the mean, the standard
deviation, and the minimum and maximum value. As indicated in Table 4,
the mean of children’s cognitive score is 63.41, which belongs to intellectually
average category. The relatively high standard deviation of cognitive scores
indicates a high variation in the observations. The variable of children’s
educational attainment as measured by years of schooling has an average of
11.59, which means that most children’s length of schooling is only up to 11
years. The years of schooling variable has a maximum value of 18, meaning
that the highest educational achievement of children is estimated to be
postgraduate level (S2). On the other hand, the natural logarithm of wages
has an average of 16.00 with a minimum value of 10.60 and a maximum

value of 18.40.
The family socio-economic background variables include the parents’
age, which has an average of 42.34 with a standard deviation of 7.68. The
natural logarithm of parents’ income has an average value of 15.03 with a
standard deviation of 0.967. The natural logarithm of spending on education
has a mean of 13.35, with a standard deviation of 0.95. The maximum length
of education for the father is 16 years, and the maximum for the mother is 17
years. On the other hand, the psychological health of parents has an average
value of 4.62 with a minimum value of 1.75 and a maximum value of 5.
The variable of parental psychological health includes a continuous scale,
meaning that the higher the value, the better the psychological condition of
the parents. Other variables such as the location of residence are also dummy
variables. The location of residence has an average value of 0.52, which
means that respondents who live in cities and villages are comparable. The
household size variable has an average value of 5.14, which means that each
household has an average of five members.
Table 4. Variable Statistics Description
Standard
Variable Observations Mean Min Max
Deviation
Cognitive Score 603 63.41 24.17 0 100
Years of Schooling 603 11.59 3.457 0 18
Wage 603 15,940,000 15,250,000 40,000 98,000,000
Natural Logarithm of Wages 603 16.00 1.283 10.60 18.40
Family Socio-Economic Background
Parent's Age 603 42.34 7.688 25 76
Parent's Income 603 5,160,000 6,300,000 100,000 106,000,000
Natural Logarithm of Parent's Income 603 15.03 0.967 11.51 18.48
Education Expense 603 992,136 1,195,000 19,200 10,500,000
Natural Logarithm of Education
603 13.35 0.958 9.863 16.17
Expense
Father’s Education 603 7.002 4.230 0 16
Mother’s Education 603 6.153 4.065 0 17
Psychological Health of Parents 603 4.619 0.516 1.750 5
Household Size (Number of
603 5.143 1.531 1 11
household members)
Residential Location (1=urban) 603 0.527 0.500 0 1
Child Characteristics
Child's Age 603 10.13 2.197 7 14
Child Gender, male=1 603 0.620 0.486 0 1
Birth Order 603 1.030 0.221 1 4
Variable characteristics of children, namely the age of children in 2000,
had an average of 10.13 with a maximum value of 14. It means that the

average age of children taking the test is ten years, and the oldest is 14 years.
On the other hand, the child’s gender, a dummy variable, has an average
value of 0.62 with a standard deviation of 0.48. From the gender perspective,
boys are slightly more dominant than girls. The birth order variable has a
maximum value of 4, meaning that the highest childbirth order is the fourth .
ESTIMATION RESULTS
Table 5 shows the estimation results explaining the relationship of
socioeconomic status, especially parents’ education, with their children’s
cognitive abilities, as shown in Model 1. Model 1, which was developed
based on Erola et al. (2016) and Nieland (2021) study, claims that parental
socioeconomic status and educational attainment correlate with their
children’s educational achievement. Hence, the question is: Are intelligent
sons associated with intelligent mothers, and conversely, are fathers with
higher education associated with daughters intelligent? On the other hand,
Model 2 explains the relationship of the child’s cognitive abilities with
educational and work outcomes in adulthood. The estimation results related
to the hypothesis are presented in Table 5.
Table 5 shows that parents, especially mothers, passing her intelligence
genes to their children. Mother’s education has a positive relationship
with children’s cognitive scores. These results are supported by Wang’s et
al. (2020) study which states that parental education, especially mothers’
education, significantly impacts their children’s academic achievement.
However, it turns out that fathers tend not to pass their intelligence genes on
to daughters or sons, as evidenced by the estimation results above. Mothers
tend to pass their intelligence genes on to boys and girls, as evidenced by
the estimation results, which show that the mother’s education variable is
significantly positively correlated with the cognitive scores of girls with a
significance level of 1% and boys at a level of 5%. Mothers’ education, as
measured by the length of schooling, can also improve the cognitive scores of
their sons and daughters, but the increase is relatively small. If the mother’s
length of education is increased by one year, it can increase 1.75 cognitive
scores for girls and 1,022 for boys’ cognitive scores.
The natural logarithm of education expenditures also has a positive
and significant relationship with children’s cognitive scores. One percent
increase in spending on education associated with a 3.4 percent increase
of children’s cognitive scores. These results also support research by Kang

(2007) that spending on education causes an increase in children’s test
scores. Parents who allocate their income for the cost of their children’s
education, such as buying school equipment, pocket money, and courses,
allow children’s intellectual development to be better. On the other hand,
children who live in cities have a relatively higher cognitive score of 4.6 than
in villages. These results also support research by Alexopoulos (1997) that
children’s cognitive scores as measured by IQ are indeed higher in urban
areas than in rural areas. It is particularly due to the influence of the school’s
social environment. The community environment in the countryside is very
different from urban areas.As a case in point, libraries and other facilities are
often inaccessible in rural areas (Alexopoulos 1997). Therefore, intellectually
superior individuals are more likely to be found in cities than in rural areas.

124
Table 5. Results of Estimation
Model 1 Model 2
Independent Variables All
Son Cognitive Daughter Cognitive Years of Schooling Wage
Constant -22.76 -33.21 -8.153 -8.368*** 15.11***
(18.72) (24.00) (31.43) (2.354) (0.298)
Parents Age -0.00869 0.00784 0.0369 - -
(0.148) (0.186) (0.227)
Child's Age 3.254*** 3.255*** 3.165*** -0.117* -0.00574
(0.452) (0.611) (0.664) (0.0650) (0.0248)
Residential Location (1=urban) 4.607** 5.840** 2.467 0.105 -
(1.920) (2.499) (3.090) (0.273)
Natural Logarithm of Parent's Income 0.0949 0.187 -0.311 0.809*** -
(1.040) (1.346) (1.721) (0.152)
Natural Logarithm of Education Expense 3.414*** 3.681*** 3.267* 0.504*** -
(1.056) (1.304) (1.876) (0.146)
Father’s Education -0.267 -0.354 -0.164 - -
(0.320) (0.450) (0.474)
Mother’s Education 1.458*** 1.758*** 1.022* - -
(0.346) (0.460) (0.554)
Psychological Health of Parents 1.487 1.760 1.387 - -
Institution, Human Capital and Development in Indonesia

(1.609) (2.313) (2.302)
Household Size (Number of household members) -0.299 -0.613 0.111 - -
(0.632) (0.804) (1.008)
Birth Order -8.066*** -5.747 -12.52*** - -
(2.864) (4.945) (1.959)
Cognitive Score - - - 0.0346*** 0.00821***
(0.00699) (0.00237)
Years of Schooling - - - 0.0367**
(0.0159)
N 603 374 229 603 603
r2 0.211 0.229 0.198 0.187 0.0419
F 16.26 12.39 12.61 29.74 8.471
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Source:Research data, processed
The child’s birth order has a negative and significant effect on cognitive
scores. It shows that the eldest child has a much better cognitive ability than
the child born later in life. These results also support research by Heiland
(2009), Black et al. (2011), and Monfardini & See (2016) that birth order can
affect children’s cognitive abilities. Children with the first birth order receive
special attention from their parents than children born later in life. The eldest
child also benefits more when parents’ financial resources are abundant. The
needs of children with first birth order in terms of education can also be
met and impact better cognitive abilities. The higher the age of the child
also affects the cognitive score. If the child’s age increases by one year, the
cognitive score will increase by 3.25. These results are supported by Schoon
(2010), who states that individuals with high cognitive abilities achieve
academic achievement at a higher age.
Based on the estimation results in model 2, controlling for the child’s
age variable, the natural logarithm of parent’s income, and the natural
logarithm of education expense, the cognitive score has a significant positive
correlation on the educational attainment in question at the level of one
percent. It means that an increase in children’s cognitive scores has the
opportunity to increase their educational attainment. An increase in a child’s
cognitive score by one unit impacts an increase in the length of school by
0.03 years. Statistically, according to Lind et al. (2018), this figure indicates a
very weak correlation. It is also can be seen in Figures 3 and 4 below. Based
on this figure, it can be concluded in the case of Indonesia, and there is
weak evidence that individual cognitive ability in early life has a significant
influence on academic and labour results in their future life.
Figure 3 shows that the cognitive scores of children aged 7-14 years in
Indonesia have a positive relationship with the length of schooling. Figure
4 presents a scatter plot to support the results of research related to the
relationship between cognition and education. Monitoring of children’s
intellectual development by parents is essential to support their children’s
educational needs.

Figure 3. Scatter Plot Cognitive Scores on Years of Schooling

Better cognitive abilities of children with higher educational attainment
will lead to better wages. Based on the estimation results, controlling for the
years of schooling variable, children’s cognitive scores positively influence
labour market outcomes, namely wages. The results of this chapter are also
supported based on research data presented in the form of a scatter plot of
cognitive scores against the natural logarithm of wages. Figure 4 shows that
cognitive scores positively correlate with the natural logarithm of wages in
Indonesia. It means that individuals with good cognitive abilities will lead
to good quality individuals to enter the labour market and impact higher
wages.
Figure 4. Scatter Plot Years of Schooling on the Natural Logarithm of
Wages

CONCLUSION
This chapter analyses the heritability of cognitive abilities and predicts the
relationship between cognitive abilities and later life outcomes, namely
success in educational attainment and labour market outcomes. Based on
the results of the analysis, not all parents pass down intelligence to their
children, indicating that fathers’ education does not necessarily affect the
cognitive abilities of girls and boys. However, mothers’intelligence is
shown to be passed down to their children, in which mother’s high level of
education positively correlates with their offsprings’ better cognitive abilities.
Socioeconomic status in the family, which includes spending on children’s
education, location of residence, and other factors such as birth order and
child’s age, significantly affect children’s cognitive abilities as measured
by cognitive test scores. It means that spending on education, location of
residence in the city, and individuals with the birth initial sequence can help
improve children’s cognitive abilities.
Cognitive ability also explains individual’s success in educational
attainment and labour market outcomes, namely wages, as evidenced by
cognitive scores, which have a positive and significant effect on school
preparation and wages, although the effect is relatively small. An increase in
a child’s cognitive score by one unit will result in an increase in the length of
school by 0.03 years. On the other hand, an increase in children’s cognitive
scores by one unit means an increase in work results in the form of their
wages by 0.008%.
Concerning the inheritance of intelligence and socioeconomic status of
the family, to develop children’s cognitive abilities a policy that supports
families and educational institutions is therefore required, especially for
economically disadvantaged families and educational institutions located
in marginalised areas. On the other hand, policies related to education and
work outcomes are also imperative; they should include skills development
programmes, both with internships and work. In addition, it is also necessary
to provide those living in rural areas with some additional learning
programmes of developing personal abilities, such as the ability to express
opinions, to communicate, to demonstrate self-confidence, and so on.

REFERENCES
Acosta, P., N. Muller, and M. Sarzosa. 2020. Adults Cognitive and
Socioemotional Skills and Their Labour Market Outcomes in Colombia.
Revista de Economía Del Rosario, 23 (1): 109–148.
Alexopoulos, D. S. 1997. Urban vs rural residence and IQ. Psychological
Reports, 80 (3): 851–860. https://doi.org/10.2466/pr0.1997.80.3.851
Black, S. E., P.J. Devereux, and K.G. Salvanes. 2009. Like father, like son?
A note on the intergenerational transmission of IQ scores. Economics
Letters, 105 (1), 138–140. https://doi.org/10.1016/j.econlet.2009.06.022
________. 2010. Small family, smart family? Family size and the IQ scores
of young men. Journal of Human Resources, 45 (1), 33–58. https://doi.
org/10.3368/jhr.45.1.33
________. 2011. ‘Older and wiser? Birth order and IQ of young men.’ CESifo
Economic Studies, 57 (1), 103–120. https://doi.org/10.1093/cesifo/ifq022
BPS. 2021. Penduduk Berumur 15 Tahun Ke Atas Menurut Pendidikan Tertinggi
yang Ditamatkan dan Jenis Kegiatan Selama Seminggu yang Lalu, 2008-2021.
https://www.bps.go.id/statictable/2016/04/05/1909/penduduk-
berumur-15-tahun-ke-atas-menurut-pendidikan-tertinggi-yang-
ditamatkan-dan-jenis-kegiatan-selama-seminggu-yang-lalu-2008-2019.
html
Burhan, N. A. S., M.M. Yunus, M.E.L. Tovar, M. E. L., and N.M.G. Burhan.
2017. Why are cognitive abilities of children so different across
countries? The link between major socioeconomic factors and PISA test
scores. Personality and Individual Differences, 105: 95–106. https://doi.
org/10.1016/j.paid.2016.09.043
Cawley, J., J. Heckman, and E. Vytlacil. 2001. Three observations on wages
and measured cognitive ability. Labour Economics, 8 (4): 419–442. https://
doi.org/10.1016/S0927-5371(01)00039-2
Conti, V. and J. Kopinska. 2018. The role of parental cognitive ageing in the
intergenerational mobility of cognitive abilities. Labour Economics, 51.
https://doi.org/10.1016/j.labeco.2017.11.009
Dahl, B. G. B., and L. Lochner. 2012. The Impact of Family Income on Child
Achievement: Evidence from the Earned Income Tax Credit. American
Economic Review, 102 (5): 1927–1956.

Erola, J., S. Jalonen, S., and H. Lehti. 2016. Parental education, class and
income over early life course and children’s achievement. Research
in Social Stratification and Mobility, 44. https://doi.org/10.1016/j.
rssm.2016.01.003
Glewwe, P., Q. Huang, and A. Park. 2017. Cognitive skills, noncognitive
skills, and school-to-work transitions in rural China. Journal of
Economic Behavior and Organization, 134. https://doi.org/10.1016/j.
jebo.2016.12.009
Heckman, J. J., J. Stixrud, and S. Urzua. 2006. The effects of cognitive
and noncognitive abilities on labour market outcomes and social
behavior. Journal of Labour Economics, 24 (3): 411–482. https://doi.
org/10.1086/504455
Heiland, F. 2009. ‘Does the birth order affect the cognitive development
of a child?’ Applied Economics, 41 (14): 1799–1818. https://doi.
org/10.1080/00036840601083220
Hewi, L., and M. Saleh. 2020. Penguatan Peran Lembaga PAUD untuk
Program International Student Assesment (PISA). Jurnal Program Studi
Pendidikan Guru PAUD STKIP Siliwangi Bandung, 6 (2).
Kang, C. 2007. “Does Money Matter ? The Effect of Private Educational
Expenditures on Academic Performance Does Money Matter ? The
Effect of Private Educational Expenditures on Academic Performance.”
National University of Singapore, Department of Economics, 151–164.
http://www.fas.nus.edu.sg/ecs/pub/wp/wp0704.pdf
Lind, D. A., W.G. Marchal, and S.A. Wathen. 2018. Statistical Techniques in
Business and Economics (17th Ed.). McGraw-Hill Education.
Mayer, S. 2002. The influence of parental income on children’s outcomes. In
Zeitschrift fur Soziologie.
Mensah, F. K. and K.E. Kiernan. 2010. Parents’ mental health and children’s
cognitive and social development: Families in England in the Millennium
Cohort Study. Social Psychiatry and Psychiatric Epidemiology, 45 (11):
1023–1035. https://doi.org/10.1007/s00127-009-0137-y
Monfardini, C., and S.G. See. 2016. Birth order and child cognitive outcomes:
An exploration of the parental time mechanism. Education Economics, 24
(5): 481–495. https://doi.org/10.1080/09645292.2015.1117581

Nieland, G. 2021. Like mother like daughter: Related Columbia College grads
achieve similar jobs, journeys. Columbia Missourian. https://www.
columbiamissourian.com/news/higher_education/like-mother-
like-daughter-related-columbia-college-grads-achieve-similar-jobs-
journeys/article_e511f060-5f83-11ec-bc60-83c7b85637fc.html
OECD. 2019. PISA 2018 Results. OECD Publishing.
RAND. 2018. RAND Indonesian Family Life Survey (IFLS).
Schoon, I. 2010. ‘Childhood cognitive ability and adult academic attainment:
evidence from three British cohort studies.’ Longitudinal and Life Course
Studies, 1 (3): 241–258. https://doi.org/10.14301/llcs.v1i3.93
Sundstrup, E., A.M. Hansen, and E.L. Mortensen, O.M. Poulsen, T. Clausen,
R. Rugulies, A. Møller, and L.L. Andersen. 2020. Cognitive Ability
in Midlife and Labour Market Participation Among Older Workers:
Prospective Cohort Study With Register Follow-up. Safety and Health at
Work, 11 (3). https://doi.org/10.1016/j.shaw.2020.06.006
Wang, W., Y. Dong, X. Liu, Y. Bai, and L. Zhang. 2020. The effect of parents’
education on the academic and non-cognitive outcomes of their children:
Evidence from China. Children and Youth Services Review, 117. https://
doi.org/10.1016/j.childyouth.2020.105307
Woessmann, L. 2004. ‘How Equal are Educational Opportunities? Family
Background and Student Achievement in Europe and the US.’ SSRN
ELibrary.
Wooldridge, J. M. 2018. Introductory Econometrics: A Modern Approach (7th
Ed). Cengage Learning, Inc.
Zahra, L. and I.D.G.K. Wisana. 2019. Parental Migration and Child Cognitive:
Evidence from Indonesian Family Life Survey 2007 and 2014. Department
of Family and Consumer Sciences, Faculty of Human Ecology, IPB
University, 28.

Chapter 7
The Effect of Educational Attainment

on Female Labour Force Participation in
Indonesia
Rizka Isnaini Husna and Eny Sulistyaningrum
INTRODUCTION
Education has been emphasised as one of the keys to encourage female
labour force participation, not only by providing better job opportunities
for women but also by increasing their income (Mammen and Paxson
2000; Cameron, Dowling, and Worswick 2001; Jaumotte 2003; Besamusca,
Tijdens, Keune, and Steinmetz 2015). In addition, previous studies found
that the relationship between educational attainment and female labour
force participation is not linear across all levels of education, but forms a
U-shape relationship (Cameron, Dowling, and Worswick 2001; Klasen and
Pieters 2015; Schaner and Das 2016; Mehrotra and Parida,2017; Klasen et al.
2021). The findings show that female labour force participation is high for
women who are not in school, low for women with primary and secondary
education levels, and high for women with tertiary education level.
However, this relationship differs across countries. Cameron, Dowling,
and Worswick (2001) analysed the relationship between educational
attainment and female labour force participation (FLFP) in Korea, Thailand,
Sri Lanka, Indonesia, and the Philippines for the period of 1975-1976 and
only found the U-shape pattern in Sri Lanka. Examining the driven factors
of FLFP in eight developing countries for the period of 2000-2014, Klasen
et al. (2021) found that the relationship was positive and linear for Brazil
131
and South Africa, U or J-shape for Jordan, India, Indonesia, and the mixture
of both for Bolivia, Vietnam, and Tanzania. Consequently, it is important
to have further empirical studies on the relationship between educational
attainment and FLFP, especially in the case of a particular country.
This study aims to examine the effect of educational attainment,
represented by years of schooling, on FLFP in Indonesia. Investigating
this relationship in Indonesia is important because despite the significant
improvement in female educational attainment, FLFP rate in Indonesia has
remained constant for the past 20 years. Female school enrollment rates for
secondary education in Indonesia experienced a large increase from only
53% in 1999 to 90% in 2018. Female enrollment rate for higher education also
shows an increasing trend from 13.4% in 1999 to 39% in 2018 (Figure 1).
Figure 1. Female School Enrollment Rate (%)
120,00 109,70
104,59
100,00 90,04
80,00
60,00 53,10
38,99
40,00
20,00 13,42
0,00
Primary Secondary Tertiary
Source: World Bank (2022)

Meanwhile, the FLFP rate has been stagnant at around 50-53% of
working age women for the last 20 years (Figure 2). Aside from stagnation,
female labour force participation also far below the male’s, resulting a large
gap between them (Figure 2). Some previous studies had investigated the
source of FLFP stagnation in Indonesia, notable examples include Schaner
and Das (2016) and Cameron, Suarez, and Rowell (2019). However, these
studies did not focus on the effect of educational attainment.

Figure 2. Labour Force Participation Rate
90,00 83,35 82,17
80,00
70,00
60,00 50,55 53,81
50,00
40,00
30,00
20,00
10,00
0,00
Female Male
Source: World Bank (2022)

Using data from 2000, 2007, and 2014 rounds of the Indonesian Family
Life Survey (IFLS), this study finds that the correlation between educational
attainment and FLFP in Indonesia is statistically significant. This study also
confirms the U-shape relationship between educational attainment and FLFP.
The average marginal effect from probit estimation shows that educational
attainment will have a negative effect on the probability of female labour
force participation until reaching the turning point of 7 years of schooling.
By estimating the probit model for each year separately, this study finds that
the turning point did not change much from 2007 to 2014. In addition, from
heterogeneity analysis, this study also finds that the result is consistent for
both Java and off-Java. Lastly, this study provides evidence that women who
have babies typically enter the labour market later than their childless peers.
This study contributes to our understanding of the correlation between
educational attainment and FLFP. This study expands the existing works
by using years of schooling and its square as the main regressors, rather
than dummy variables for the highest educational attainment like most
previous studies. The usage of years of schooling and its square capture the
quadratic relationship between these two variables, along with the turning
point where women enter the labour market. The level of education at which
women enter the labour market is critical to understand how improvements
in education for women can be facilitated.
The remainder of the study is organised as follows. The second section
provides a brief literature review on female labour force participation. The
third and fourth sections describe the data and methodology respectively.
The fifth section presents the main results of the effect of educational
The Effect of Educational Attainment on Female Labour 133

Force Participation in Indonesia
attainment on FLFP in Indonesia. The study is concluded in the last section.
LITERATURE REVIEW
A large number of studies on female labour force participation have been
conducted, especially in the developing countries where women face several
barriers to participate in labour market. Klasen and Pieters (2015) and
Mehrotra and Parida (2017) analysed the causes of low and stagnant female
labour force participation in India for periods 1980s to 2011. They found that
the cause of stagnation in India was a combination between labour supply
and demand factors. From the demand side, structural change in economic
activities from agriculture to construction sector has limited women to
participate in labour force. From the supply side, the rising household income
led to a withdrawal of women from the labour market due to income effect.
Klasen and Pieters (2015) also found that the increase in women’s education
did not make a great contribution to FLFP due to the U-shaped relationship
between women’s educational attainment and FLFP.
In the case of Indonesia, Cameron, Dowling, and Worswick (2001)
conducted an empirical study on relationship between educational attainment
and FLFP within the period of 1975-1976 in five Asian countries, including
Indonesia. The result showed no U-shape relationship between educational
attainment and FLFP. However, they found that intermediate, secondary,
and tertiary levels of education have positive impact on Indonesia women’s
participation.
Schaner and Das (2016) and Cameron, Suarez, and Rowell (2019) also
analysed the main drivers of FLFP in Indonesia for periods 1990s to 2010s.
These two studies found that the main drivers of FLFP in Indonesia are
marital status, educational attainment (especially tertiary level), and the
presence of young children under the age of five. According to Schaner and
Das (2016), the cause of FLFP stagnation in Indonesia is due to the U-shape
relationship between FLFP and educational attainment where an increase in
FLFP at the higher education group has been offset by a decrease in FLFP at
lower education group.
Klasen et al. (2021) also investigated the determinant of FLFP for urban
married women in eight developing and emerging countries, including
Indonesia. Their data covers period from 2000 to 2014. Similar to Schaner
and Das (2016), they confirmed U-shaped relationship between educational
attainment and FLFP in Indonesia. Relative to women who did not finish

primary school, FLFP is lower for women who graduated from primary and
junior high schools; and, FLFP is increasing for secondary and tertiary school
graduates. Meanwhile, household income, household head’s education, and
the number of young children have negative effect on women’s participation.
DATA
The main data in this study are taken from the Indonesian Family Life
Survey (IFLS). IFLS is a longitudinal survey in Indonesia, which has been
conducted six times. The first wave (IFLS 1) was conducted in 1993, the
IFLS 2 in 1997, and the IFLS 2+ in 1998; whereas the IFLS 3, IFLS 4, and
IFLS 5 were conducted in 2000, 2007, and 2014, respectively. Besides IFLS,
there are other data sources which better represent Indonesian population
such as SUSENAS and SAKERNAS. This study uses IFLS since it is publicly
accessible. In addition, IFLS sample is representative of about 83% of the
Indonesian population living in 13 of the nation’s 26 provinces in 1993.
However, IFLS is not a representative for the whole Indonesia, particularly
those in the eastern part of the country.
This study utilises the IFLS 3 (Strauss et al. 2004), the IFLS 4 (Strauss et
al. 2009) and the IFLS 5 (Strauss, Sikoki, and Witoelar 2016) to obtain a pooled
dataset of individuals. The 2000, 2007, and 2014 waves of the Indonesian
Family Life Survey (IFLS) are selected to cover periods of female labour force
participation stagnation.
The samples are limited to married women aged 25 to 54 in each
survey wave. Working women in the age of 25-54 are selected to avoid
the issue of school enrollment and retirement. Married women are chosen
because previous studies have shown that the issue in female labour force
participation is among married women (Cameron, Suarez, and Rowell
2019; Mehrotra and Parida 2017). In addition, previous studies on labour
supply focused on the family context in work decision making (Blundell and
MaCurdy 1999).
The main outcome of interest is female labour force participation. FLFP
is measured as a binary variable which equals one if women participate in
labour force and equals 0 if women did not participate in labour force1. This
1 This variable refers to the definition of labour force participation by International Labour
Organization (2022) which is the number of persons in the labour force as a percentage of the
working-age population. The labour force is the sum of the number of persons employed and
the number of persons unemployed. This study includes people who was, last week, working for
minimum 1 hour, including unpaid family workers, as part of persons employed (Badan Pusat
Statistik 2022).

study also includes female years of schooling and its square as the main
independent variables.
For control variables, this study uses information on individual,
household, and regional characteristics. Individual characteristics contain
(1) age and its square, (2) religion, (3) father’s years of schooling and (4)
mother’s years of schooling. Religion is included to capture the impact of
religiously determined restrictions on women. This study creates dummy
variables for each religion in Indonesia which are Islam, Protestant, Catholic,
Hindu, Buddhist, and other religions. Islam will be used as the base category.
This study also includes father’s and mother’s educational attainment since
parents who are highly educated and participate in the labour force could
give their daughter a positive and modern view of women’s involvement in
the labour force (Stevens and Boyd 1980; Farré and Vella 2013).
Household characteristics comprise eight variables. To capture
household wealth, this study includes (1) household head’s years of
schooling except whenever female respondent is the head and (2) weekly
household expenditure per capita for food and non-food goods and services.
This study also includes (3) a dummy variable equal to one if at least one
male household member is currently engaged in salaried employment as a
proxy of stable income. Household wealth and their security of income are
expected to reduce the need for women to work.
Furthermore, household characteristics are (4) household size which is
the number of people living in the household, (5) the number of children aged
0-4, (6) the number of children aged 5-14, (7) the number of elderly women
and men aged 65+, and lastly (8) the number of adult female household
members aged 15-65 except female respondent in this sample. The number
of children is expected to correlate negatively with women’s participation
in labour force as women tend to exit the labour market to take care of the
children. Meanwhile, the presence of adult females and elderlies is expected
to correlate positively since they are potential babysitters.
For regional characteristic, this study uses provincial unemployment
rate in each survey wave, obtained from Indonesia Central Bureau of
Statistics (Badan Pusat Statistik). Unemployment rate will be used to control
supply side of labour market.
Table 1 shows the summary statistics of all variables, disaggregated by
survey year. Female labour force participation was unchanged during the
study period, with about 60% of total working-age women in this sample.

This result is similar to FLFP macro data from World Bank (2022). Next,
educational attainment for the respondent, their parents, and their household
head increased over time, especially for female respondent whose average
years of schooling increased from 6 years in 2000 to 9 years in 2014. Average
household size, number of children aged 0-4, and number of elderlies per
household remained constant. Meanwhile, the average number of children
aged 5-14 and number of adult females declined over time. Lastly, the
unemployment rate showed a downward trend during the study period,
from 6.2% in 2000 to 5.7% in 2014.
Table 1. Summary Statistics (Sample Mean)
Variable 2000 2007 2014
Female labour force participation 0.641 0.629 0.666
Years of schooling 6.561 7.895 9.009
Age 36.928 36.626 37.051
Religion:
Islam 0.886 0.9 0.904
Protestant 0.041 0.034 0.036
Catholic 0.015 0.016 0.011
Hindu 0.052 0.047 0.048
Buddhist 0.005 0.002 0
Others 0 0.001 0.001
Father’s years of schooling 4.386 5.218 5.923
Mother’s years of schooling 2.876 3.767 4.387
Household Characteristics
Household head’s years of schooling 7.003 8.012 8.733
Log per capita weekly expenditure 54902.992 241897.06 309864.27
Male salaried emp. 0.154 0.058 0.084
Household size 4.929 4.473 4.343
No. of children (aged 0-4) 0.536 0.533 0.505
No. of adult females 0.6 0.46 0.423
No. of elderly 0.172 0.161 0.148
Regional Characteristics
Unemployment rate 6.235 9.123 5.767
Source: Authors’ calculation using IFLS 3 – 5.

METHODOLOGY
The probability of woman i in year t (2000, 2007, and 2014) being in the labour
force is modeled as follows:
𝑃𝑃(𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝑖𝑖𝑖𝑖 = 1) = 𝐹𝐹(𝛽𝛽0 + 𝛽𝛽1 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝑖𝑖𝑖𝑖 + 𝛽𝛽2 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝑖𝑖𝑖𝑖2 + 𝒖𝒖𝒊𝒊𝒊𝒊 𝜹𝜹 + 𝒗𝒗𝒊𝒊𝒊𝒊 𝜹𝜹 + 𝒘𝒘𝒊𝒊𝒊𝒊 𝜹𝜹 + 𝒛𝒛𝒓𝒓𝒓𝒓 𝜹𝜹+∝𝑟𝑟 + 𝜀𝜀𝑖𝑖𝑖𝑖 )
This model is modified from Klasen and Pieters (2015) and Mehrotra and
Parida (2017). This study will estimate the model for each survey year
separately to determine whether the correlation change over time.
The dependent variable is a dummy variable which equals 1
(one) if women participate in labour force and zero otherwise2. The main
independent variables are that are the female years of schooling and
its square The control variables consist of four vector variables.
The vector u consists individual characteristic variables including age and
its square, religion fixed effects, father’s years of schooling, and mother’s
years of schooling. The vector w contains household characteristic variables,
including household head’s years of schooling, the natural log of per capita
weekly expenditure, dummy of having at least having one male with
salaried employment, household size, number of children aged 0-4 and 5-14,
number of adult females aged 15-65, and number of elderly women and men
aged 65+. The vector z includes regional characteristic which is provincial
unemployment rate in each survey wave. The subscripts, i, t, and r indicate
an individual, time index, and province, respectively. The term indicates
province fixed effect, and εit is error term.
This study first estimates the effect of education on FLFP using probit
regression, which is then followed by a calculation of the derivation from
each estimation result to determine a turning point when women enter
the labour market. Lastly, this study conducts heterogeneity analysis by
disaggregating the samples based on respondent region (Java vs Off-Java)
and the presence of babies.
RESULT
Estimation Results
Probit estimation results are shown in Table 2. The table presents the average
marginal effect (AME) for each independent variables, which shows the
change in probability of female participation in labour force due to one-unit
2 Ibid

change in explanatory variable. Meanwhile, for categorical variables, AME
indicates the difference with base category.
Table 2. Estimation Result (Average Marginal Effect)
Dependent variable: FLFP 2000 2007 2014
Years of schooling -0.0110 -0.0255*** -0.0402***
(0.00699) (0.00667) (0.00722)
Years of schooling square 0.000780* 0.00181*** 0.00292***
(0.000459) (0.000413) (0.000409)
Individual characteristics
Age 0.0351*** 0.0375*** 0.0256**
(0.0122) (0.0102) (0.00995)
Age square -0.000391** -0.000405*** -0.000287**
(0.000158) (0.000133) (0.000129)
Father’s years of schooling 0.000537 0.000932 0.00175
(0.00370) (0.00304) (0.00259)
Mother’s years of schooling 0.000443 -0.00314 -0.00305
(0.00413) (0.00341) (0.00286)
Household characteristics
Household head’s years of schooling -0.00491* -0.00567** -0.00414*
(0.00293) (0.00246) (0.00233)
Log per capita weekly expenditure 0.00969 0.0456*** 0.0374***
(0.0152) (0.0118) (0.0122)
Male salaried emp. -0.0330 -0.136*** -0.0699**
(0.0322) (0.0355) (0.0297)
Household size -0.0279** -0.00560 -0.000131
(0.0127) (0.0120) (0.0125)
No. of children (aged 0-4) -0.0567*** -0.0378** -0.0915***
(0.0198) (0.0181) (0.0184)
(0.0155) (0.0154) (0.0157)
No. of adult females 0.0190 0.0306 0.0359*
(0.0190) (0.0186) (0.0189)
No. of elderly 0.0236 0.0109 0.00108
(0.0243) (0.0223) (0.0212)
Regional characteristics
Unemployment rate 0.0794*** 0.0166 0.0360
(0.0267) (0.0314) (0.0659)
Religion fixed effect Y Y Y
Province fixed effect Y Y Y
Observations 2,102 3,058 3,237
Source: Authors’ calculations using IFLS 3 – 5.
Notes:
(1) The signs *, **, *** indicate significance at 0.1, 0.05, and 0.01 significance levels.
(2) Standard errors in parentheses.

This study finds that years of schooling years of schooling has a negative
effect, while years of schooling square has a positive effect, although
the result for 2000 period did not significant. This indicates a U-shaped
relationship between educational attainment and FLFP in Indonesia in 2007
and 2014 periods. This finding supports the previous studies that also found
a U-shape relationship between educational attainment and female labour
force participation (Cameron, Dowling, and Worswick 2001; Klasen and
Pieters 2015; Schaner and Das 2016; Mehrotra and Parida 2017; Klasen et al.
2021).
The U-shaped relationship shows that educational attainment has
negative effect until some point before turning to be positive. Therefore, this
study will calculate the derivation of estimation results for the 2007 and 2014
periods to determine a turning point when women enter the labour market
and whether the turning point changed during the study period.
For 2007 period, attaining the first year of schooling will reduce the
probability of female participation in labour force by -0.0255 + (2*0.00181)
= -0.02188 or 2.19%. Each additional year of schooling will reduce the
probability of female participation in labour force smaller than the first
year. Educational attainment will have a negative effect on the probability
of female participation in labour force until reaching the turning point
-0.0255/-(2*0.00181) = 7.04 or 7 years of schooling, equivalent to first-grade
junior high school. After achieving 7 years of schooling, the additional years
of schooling will increase the probability of female participation in labour
force. The turning point is similar to the mean of female schooling in this
sample which is 7 years of schooling (see Table 1 Column 2). This result is in
line with previous studies by Klasen and Pieters (2015), for instance, which
found the turning point is similar to their mean of schooling.
For 2014 period, attaining the first year of schooling will reduce the
probability of female participation in labour force by -0.0402 + (2*0.00292) =
-0.0344 or 3.4%. Each additional year of schooling will reduce the probability
of female participation in labour force, which is smaller than the first year.
Educational attainment will have a negative effect on the probability of
female participation in labour force until reaching the turning point -0.0402/-
(2*0.00292) = 6.88 or 7 years of schooling, equivalent to first-grade junior
high school. The turning point did not change much from the turning point
for the 2007 period. However, the turning point is slightly lower than their
mean of female schooling in this period (see Table 1 Column 3).

The result from both periods shows that the average woman in Indonesia
has been passing the turning point to entering the labour market. However,
as discussed previously, female labour force participation in Indonesia has
remained stagnant for the last 20 years. This study expects that some other
factors might contribute to the stagnation.
Factors outside educational attainment such as age shows an inverted
U-shape relationship. Women’s participation first increases and then
declines with age. The peak age of participation did not change much
from 45 years in 2000 to 44 years in 2014. Furthermore, household wealth
in terms of household head’s years of schooling has the expected negative
effect, while the household expenditure has a significant positive effect,
except for the 2000 period. The indicator of stable income—i.e., men’s
salaried employment—has a significant negative effect, except for the 2000
period. The effects of these three variables are decreasing from 2007 to 2014,
which shows that women become less responsive to household wealth and
security of income during the study period. The number of children aged
0-4 is associated with lower women’s participation in the labour force with
the magnitude becoming larger over time. This indicates that the role of
women in taking care of young children becomes stronger over years, which
encourages women to leave the labour market.
HETEROGENEITY ANALYSIS
Java vs Off-Java
This study conducts heterogeneity analysis by separating the result between
Java and Off-Java. The result is shown in Table 3. This study confirms a
U-shape relationship between educational attainment and female labour
force participation in Java and Off-Java, except for the 2000 period. The
turning point for educational attainment in Java is increasing from 6 years
of schooling in 2007 to 7 years of schooling in 2014. However, for Off-Java,
the turning point is decreasing from 8 years of schooling in 2007 to 7 years
of schooling in 2014. This result is consistent with the earlier findings by
Cameron, Suarez, and Rowell (2019) in which a strong correlation between
educational attainment (especially upper secondary and tertiary school) and
FLFP for Java-Bali and other islands.

Table 3. Heterogeneity Analysis (Java vs Off-Java)
Dependent variable : 2000 2007 2014
FLFP Java Off-Java Java Off-Java Java Off-Java
Years of schooling -0.0132 -0.00566 -0.0272*** -0.0238** -0.0521*** -0.0291***
(0.00924) (0.0108) (0.00895) (0.0102) (0.0105) (0.00988)
Years of schooling square 0.000943 0.000427 0.00211*** 0.00150** 0.00371*** 0.00218***
(0.000580) (0.000751) (0.000540) (0.000647) (0.000598) (0.000559)
Age 0.0217 0.0519*** 0.0485*** 0.0243 0.0203 0.0302**
(0.0159) (0.0188) (0.0136) (0.0152) (0.0140) (0.0141)
Age square -0.000243 -0.000564** -0.000537*** -0.000246 -0.000227 -0.000332*
(0.000206) (0.000245) (0.000178) (0.000199) (0.000181) (0.000185)
Father’s years of schooling -0.000905 0.00314 -0.000162 0.00146 0.00948** -0.00552
(0.00469) (0.00601) (0.00421) (0.00440) (0.00383) (0.00348)
Mother’s years of 0.00580 -0.00597 -0.00278 -0.00298 -0.00261 -0.00172
schooling (0.00538) (0.00651) (0.00464) (0.00504) (0.00432) (0.00379)
Household head’s years of -0.0107*** 0.000891 -0.00833** -0.00237 -0.00942*** 0.000666
schooling (0.00390) (0.00444) (0.00336) (0.00359) (0.00345) (0.00315)
Log per capita weekly 0.0153 0.00487 0.0580*** 0.0311* 0.0593*** 0.0218
expenditure
(0.0197) (0.0237) (0.0160) (0.0173) (0.0175) (0.0172)
Male salaried emp. -0.0563 -0.000731 -0.123*** -0.169*** -0.0867** -0.0546
(0.0409) (0.0534) (0.0465) (0.0548) (0.0410) (0.0434)
Household size -0.0155 -0.0411** 0.00382 -0.00824 -0.00145 0.00533
(0.0165) (0.0197) (0.0161) (0.0179) (0.0181) (0.0173)
No. of children (aged 0-4) -0.0974*** -0.00795 -0.0555** -0.0313 -0.0960*** -0.0873***
(0.0261) (0.0304) (0.0250) (0.0262) (0.0264) (0.0257)
No. of children (aged 5-14) 0.0230 0.0208 -0.0243 0.0410* -0.00284 0.00301
(0.0209) (0.0234) (0.0219) (0.0217) (0.0226) (0.0218)
No. of adult females 0.0249 0.00949 0.0295 0.0232 0.0459* 0.0158
(0.0244) (0.0301) (0.0251) (0.0274) (0.0266) (0.0268)
No. of elderly 0.00875 0.0441 0.0138 -0.00288 0.0123 -0.00773
(0.0304) (0.0406) (0.0294) (0.0344) (0.0296) (0.0303)
Unemployment rate -0.0108* 0.0677** -0.0287 0.0147 0.0660 0.0320
(0.00564) (0.0281) (0.0190) (0.0303) (0.0971) (0.0640)
Religion fixed effect Y Y Y Y Y Y
Province fixed effect Y Y Y Y Y Y
Observations 1,176 924 1,702 1,355 1,624 1,612
Source: Authors’ calculations using IFLS 3 – 5.
Notes:
(1) The signs *, **, *** indicate significance at 0.1, 0.05, and 0.01 significance levels.

Other variables follow a similar pattern to those presented in Table 2,
especially for Java. The household head’s years of schooling in Java has a
negative effect, consistent with the previous result. Meanwhile, the results of
household head’s years of schooling in Off-Java are insignificant. Household
expenditure has a positive effect and male salaried employment has a
negative effect for both regions, although the effect is disappearing for Off-
Java in 2014. Lastly, the number of young children aged 0-4 is associated
with lower women’s participation in labour force in Java and off-Java.
Having Baby vs Not Having Baby

This study also conducts heterogeneity analysis by disaggregating the
samples between women who have babies and women who do not have
babies. Table 4 presents the result. This study also finds a U-shape relationship
between educational attainment and female labour force participation for
both sub-samples, except in 2000.

Table 4. Heterogeneity Analysis (Having baby vs Not having baby)
2000 2007 2014
Dependent variable : Having Not Having Not Having Not
FLFP Baby Having Baby Having Baby Having
Baby Baby Baby
Years of schooling -0.0251* -0.000190 -0.0224* -0.0209*** -0.0473*** -0.0335***
(0.0128) (0.00831) (0.0116) (0.00801) (0.0124) (0.00863)
Years of schooling 0.00184** -6.88e-05 0.00156** 0.00175*** 0.00337*** 0.00251***
square (0.000819) (0.000543) (0.000688) (0.000515) (0.000679) (0.000503)
Age 0.0211 0.0449*** 0.0210 0.0491*** 0.00635 0.0283**
(0.0219) (0.0153) (0.0186) (0.0125) (0.0174) (0.0125)
Age square -0.000203 -0.000524*** -0.000185 -0.000550*** -4.24e-06 -0.000336**
(0.000292) (0.000193) (0.000255) (0.000159) (0.000234) (0.000159)
Father’s years of -0.00615 0.00542 -0.00179 0.00196 0.00526 -0.00234
schooling (0.00634) (0.00460) (0.00466) (0.00395) (0.00409) (0.00331)
Mother’s years of 0.00692 -0.00693 -0.000773 -0.00720 -0.00339 -0.00172
schooling (0.00709) (0.00510) (0.00525) (0.00441) (0.00440) (0.00378)
Household head’s years -0.00325 -0.00547 0.00345 -0.0118*** -0.00463 -0.00239
of schooling (0.00496) (0.00377) (0.00387) (0.00315) (0.00384) (0.00289)
Log per capita weekly 0.0347 -8.48e-05 0.0628*** 0.0339** 0.0477** 0.0291*
expenditure (0.0277) (0.0181) (0.0190) (0.0149) (0.0203) (0.0151)
Male salaried emp. -0.0192 -0.0603 -0.212*** -0.0814* -0.0362 -0.101***
(0.0566) (0.0390) (0.0580) (0.0439) (0.0481) (0.0362)
Household size -0.0152 -0.0224 -0.0110 -0.00526 0.0125 -0.0133
(0.0228) (0.0157) (0.0206) (0.0142) (0.0206) (0.0155)
No. of children (aged -0.0453 -0.0259 -0.0914**
0-4) (0.0439) (0.0372) (0.0388)
No. of children (aged 0.0143 0.0114 0.0331 0.00700 -0.00788 0.00793
5-14) (0.0285) (0.0191) (0.0258) (0.0189) (0.0254) (0.0195)
No. of adult females -0.000680 0.0276 0.0911*** -0.00498 0.0178 0.0348
(0.0343) (0.0230) (0.0325) (0.0220) (0.0314) (0.0229)
No. of elderly -0.0128 0.0390 0.0601 -0.0194 0.0160 -0.00629
(0.0421) (0.0298) (0.0389) (0.0264) (0.0356) (0.0253)
Unemployment rate 0.00145 -0.0169 -0.0116 0.0275 0.0310 0.159***
(0.0237) (0.0178) (0.0505) (0.0392) (0.0752) (0.0512)
Religion fixed effect Y Y Y Y Y Y
Province fixed effect Y Y Y Y Y Y
Observations 840 1,125 1,362 1,693 1,442 1,792
Source : Authors’ calculations using IFLS 3 – 5.
Notes : (1) The signs *, **, *** indicate significance at 0.1, 0.05, and 0.01 significance levels

In addition, the educational attainment turning point for women who
have babies is slightly higher than for women who do not have babies in all
periods, suggesting that women with babies enter the labour market later
than their childless peers. This result is similar to Schaner and Das (2016)
who found that women face difficulty entering the labour market, especially
the formal sector, after having children.
CONCLUSION
Using data from the Indonesian Family Life Survey (IFLS) of 2000, 2007,
and 2014 periods, this study finds that educational attainment significantly
affects female labour force participation in Indonesia. This study also
confirms the U-shaped relationship between educational attainment and
female labour force participation. The average marginal effect from probit
estimation shows that educational attainment will have a negative effect on
the probability of female labour force participation until reaching the turning
point of 7 years of schooling, equivalent to first-grade junior high school. The
turning point is also similar to the mean of female schooling in this sample.
After achieving 7 years of schooling, the additional years of schooling will
increase the probability of female labour force participation. By estimating
the probit model for each year separately, this study finds that the turning
point did not change much from 2007 to 2014. The result is also consistent
for both Java and Off-Java. Lastly, this study provides evidence that women
who have babies enter the labour market later than their childless peers.
Our findings provide important implications for social welfare policy
in Indonesia. First, since higher education has a positive significant effect on
female labour force participation, policies that focus on increasing women
access to upper secondary and tertiary school would improve women’s
participation in labour force. The government can increase the effectiveness
of the program by providing better information about the importance of
college and the availability of tuition assistance for the under-informed
and underprivileged communities. Several studies have found that simple
interventions by providing more complete information about the return of
education and tuition assistance can improve educational attainment (Jensen
2012; Castleman and Page 2015). The provision of this information can be in
the form of a campaign through text messages or peer mentoring which does
not cost a lot of resources. The government should take this programme into
consideration, since the enrollment rate of higher education, especially in
rural areas, is still low (Badan Pusat Statistik 2020).

Second, this study also finds out that the number of children aged 0-4 has
a negative effect. This indicates that policy which support working mothers
can encourage female labour force participation. Several studies have found
that childcare subsidy policy implemented in some countries, such as
Argentina, Quebec, and Italy can increase female labour force participation
and reduce reservation wages for unemployed women, thereby increasing
the likelihood of finding work (Berlinski and Galiani 2007; Baker, Gruber,
and Milligan 2008; Carta and Rizzica 2018). Therefore, childcare subsidy
policy can be considered by the government to increase female labour force
participation. In addition, the government can also encourage companies
in the formal sector to provide telework or work from home option. By
providing opportunities for women to work from home, this can reduce the
reservation wage for childcare as well as the psychological costs of being
away from children while working (Dettling 2017).
Nonetheless, this study has some limitations, as it focuses only on the
factors that influence female labour force participation from the supply side,
particularly educational attainment. Yet, it does not include factors from the
demand side of the labour market. The cause of the stagnation of female
labour force participation in Indonesia can be due to the lack of employment
opportunities for women, but this is beyond the discussion of this study.
Further study on the effect of educational attainment on female labour force
participation is expected to consider factors from the demand side of labour
market.
REFERENCES
Badan Pusat Statistik. 2020. ‘Angka Partisipasi Kasar (APK) Perguruan
Tinggi (PT) Menurut Daerah Tempat Tinggal, 2011-2019.’
(“Badan Pusat Statistik”) 20 July. https://www.bps.go.id/
dynamictable/2018/07/11/1524/angka-partisipasi-kasar-apk-
perguruan-tinggi-pt-menurut-daerah-tempat-tinggal-2011-2019.html.
________. 2022. ‘Tenaga Kerja: Konsep/Penjelasan Teknis.’ 07 February.
https://www.bps.go.id/subject/6/tenaga-kerja.html .
Baker, Michael, Jonathan Gruber, and Kevin Milligan. 2008. Universal
Child Care, Maternal Labour Supply, and Family Well-Being. Journal of
Political Economy 116 (4): 709–45.
Berlinski, Samuel and Sebastian Galiani. 2007. The Effect of a Large Expansion
of Pre-Primary School Facilities on Preschool Attendance and Maternal

Employment. Labour Economics 14 (3): 665–80.
Besamusca, Janna, Kea Tijdens, Maarten Keune, and Stephanie Steinmetz.
2015. Working Women Worldwide. Age Effects in Female Labour Force
Participation in 117 Countries. World Development 74: 123–41.
Blundell, Richard, and Thomas MaCurdy. 1999. Labour Supply: A Review of
Alternative Approaches. In Handbook of Labour Economics, 3:1559–1695.
Amsterdam: Elsevier.
Cameron, L. A., J. M. Dowling, and C. Worswick. 2001. Education and Labour
Market Participation of Women in Asia: Evidence from Five Countries.
Economic Development and Cultural Change 49 (3): 459–77.
Cameron, Lisa, Diana Contreras Suarez, and William Rowell. 2019. Female
Labour Force Participation in Indonesia: Why Has It Stalled? Bulletin of
Indonesian Economic Studies 55 (2): 157–92.
Card, David. 2001. Estimating the Return to Schooling: Progress on Some
Persistent Econometric Problems. Econometrica 69 (5): 1127–60.
Carta, Francesca, and Lucia Rizzica. 2018. Early Kindergarten, Maternal
Labour Supply and Children’s Outcomes: Evidence from Italy. Journal
of Public Economics 158: 79–102.
Castleman, Benjamin L., and Lindsay C. Page. 2015. Summer Nudging: Can
Personalized Text Messages and Peer Mentor Outreach Increase College
Going among Low-Income High School Graduates? Journal of Economic
Behavior and Organization 115: 144–60.
Dettling, Lisa J. 2017. Broadband in the Labour Market: The Impact of
Residential High-Speed Internet on Married Women’s Labour Force
Participation. Industrial and Labour Relations Review 70 (2): 451–82.
Duflo, Esther. 2001. Schooling and Labour Market Consequences of
School Construction in Indonesia: Evidence from an Unusual Policy
Experiment. American Economic Review 91 (4): 795–813.
Farré, Lídia, and Francis Vella. 2013. The Intergenerational Transmission of
Gender Role Attitudes and Its Implications for Female Labour Force
Participation. Economica 80 (318): 219–47.
Goldin, Claudia. 1994. The U-Shaped Female Labour Force Function in
Economic Development and Economic History. NBER Working Paper
Series no. 4707: 1–40.

International Labour Organization. 2022. Indicator Description: Labour
Force Participation Rate. 07 February. https://ilostat.ilo.org/resources/
concepts-and-definitions/description-labour-force-participation-rate/.
Jaumotte, Florence. 2003. Female Labour Force Participation: Past Trends and
Main Determinants in OECD Countries. OECD Economics Department
Working Papers no. 376: 1–66.
Jensen, Robert. 2012. Do Labour Market Opportunities Affect Young
Women’s Work and Family Decisions? Experimental Evidence from
India. The Quarterly Journal of Economics 127 (2): 753–92.
Klasen, Stephan, TU THI NGOC Le, Janneke Pieters, and Manuel Santos Silva.
2021. What Drives Female Labour Force Participation? Comparable
Micro-Level Evidence from Eight Developing and Emerging Economies.
Journal of Development Studies 57 (3): 1–26.
Klasen, Stephan and Janneke Pieters. 2015. What Explains the Stagnation of
Female Labour Force Participation in Urban India? World Bank Economic
Review 29 (3): 449–78.
Mammen, Kristin and Christina Paxson. 2000. Women’s Work and Economic
Development. Journal of Economic Perspectives 14 (4): 141–64.
Mehrotra, Santosh and Jajati K. Parida. 2017. Why Is the Labour Force
Participation of Women Declining in India? World Development 98: 360–
80.
Schaner, Simone and Smita Das. 2016. “Female Labour Force Participation
in Asia: Indonesia Country Study.” ADB Economics Working Paper
Series, no. 474: 1–42.
Stevens, Gillian and Monica Boyd. 1980. The Importance of Mother : Labour
Force Participation and Intergenerational Mobility of Women. Social
Forces 59 (1): 186–99.
Strauss, John, Kathleen Beegle, Bondan Sikoki, Agus Dwiyanto, Yulia
Herawati, and Firman Witoelar. 2009. “The Fourth Wave of the
Indonesia Family Life Survey (IFLS4): Overview and Field Report.”
RAND Working Paper.
Strauss, John, Kathleen Beegle, Bondan Sikoki, Agus Dwiyanto, Yulia
Herwati, Firman Witoelar, and RAND Corporation. 2004. “The Third
Wave of the Indonesia Family Life Survey (IFLS3): Overview and Field
Report.” RAND Working Paper.

Strauss, John, Bondan Sikoki, and Firman Witoelar. 2016. “The 5th Wave of
the Indonesia Family Life Survey (IFLS5): Overview and Field Report.”
RAND Working Paper.
World Bank. 2022. World Development Indicators. 06 February. https://
databank.worldbank.org/source/world-development-indicators/
preview/on.

Chapter 8
Forestry Agrarian Reform in Balumpewa

Village, Central Sulawesi Province
Kasmiati, Mohamad Shohibuddin, Ganies Oktaviana, and Ahmad Jaetuloh
INTRODUCTION
After experiencing a period of submerging during the New Order period,
Indonesia’s agrarian reform policy was finally strengthened again through
the National Agrarian Reform Programme (Programme Pembaruan Agraria
Nasional or PPAN) launched by Susilo Bambang Yudhoyono (SBY)
government in 2007 (Shohibuddin and Salim 2012). Then, the Jokowi
government continued the programmeme through its inclusion in 2015-2019
National Medium-Term Development Plan (Rencana Pembangunan Jangka
Menengah Nasional/RPJMN) and the 2020-2024 RPJMN as one of the national
goals priority programmes.
The acceleration of the agrarian reform promotes forest lands as the fifth
priority programme of the 2016-2019 National Strategy, simultaneously bring
about the management of forest resources on the hand of the community
people (Kompas.com 2014; BAPPENAS 2014). In 2017, a legal foothold on
the Settlement of Land Tenure in Forest Areas (Penyelesaian Penguasaan
Tanah Dalam Kawasan Hutan or PPTKH) was published through Presidential
Regulation No. 88/2017. In the following year, Presidential Regulation No.
86/2018 on Agrarian Reform entailing forest areas as one of the objects
of agrarian reform was also published. These two regulations are the
government’s instruments to establish the Land Objects of Agrarian Reform
151
(TORA) programme in forest areas which is expected to solve problems
related to land settlement in forest areas.
The implementation of these policies, however, received many criticisms,
especially from civil society organisations. These organisations then
actively proposed TORA based on people participatory mapping processes
(Berdikarionline.com 2017; Wirawan 2018; Ekawati, Salim, and Utami 2019;
Celebespos.com 2020; Susetyo 2021; Tempo.co 2021; Panahpapua.or.id 2021).
They argue that people participation is the pre-request for the programme to
be successful (Carranza 2015; Kantor Staf Presiden Republik Indonesia 2017;
Djauhari, Lubis, and Moenir 2018; Suharti, 2019; Mitchell, Safik and Dodd
2020; Winarti, Joko, and Budiman 2021).
In Sigi District, Central Sulawesi Province, there has been a collaboration
between the local government of Sigi District, GTRA, and civil society
organisations (i.e., YMP, JKPP and KPA)1 together with residents did
participatory work to propose a TORA programme to the central and
provincial governments (JKPP 2017; YMP 2017; Jemali 2017; YMP 2019;
Saputra 2020). In 2017, they proposed 168,000 hectares of forest areas to be
part of their TORA programme. The Inver PTKH Team (Tim Inventarisasi dan
Verifikasi Penguasaan Tanah dalam Kawasan Hutan), however, approved only
3,000 hectares to be processed as the TORA programme of forest areas in Sigi
District. This case shows how significant it can be the gap in size between the
area proposed to be in a TORA programme by local communities and the
final results by the PTKH Inver Team. Even in the case of high community
participation, the final result is mainly determined by the PTKH Inver Team
and this can be far from communities’ expectations.
This TORA scheme is finally implemented under the initiative of the Sigi
District Government of Sulawesi Tengah Province (Savitri 2020), which in
turn has made Sigi District a pilot district for the national TORA programme.
Ironically, the implementation of this forestry agrarian reform in the Sigi
District encounters many obstacles. For example, one village in the Sigi
District proposed a TORA programme for forest areas is the Balumpewa
Village. Residents in this village have participated in various endeavours,
ranging from participatory mapping, allocating budget for agrarian reform
from village funds, to taking direct actions. Unfortunately, the TORA
programme in the Balumpewa Village is hitherto stagnant.
1 YMP stands for Yayasan Merah Putih, JKPP stands for Participatory Mapping Network, and KPA
stands for Consortium for Agrarian Reform

Luthfi (2018) who analyses various TORA programmes in Indonesia,
including that in the Sigi District, argues that conflicting proposals and
decisions produced by the Agrarian Reform Task Force (Gugus Tugas Reforma
Agraria/GTRA) of the Ministry of Agrarian Affairs and Spatial Planning/
National Land Agency (Kementerian Agraria dan Tata Ruang/Badan Pertanahan
Nasional/Kemen ATR/BPN) and by the Forest Zone Stabilisation Unit (Balai
Pemantapan Kawasan Hutan or BPKH) of the Ministry of Environment and
Forestry (Kementrian Lingkungan Hidup dan Kehutanan/KLHK) have
caused issues in the implementation of this programme. In other words, the
stagnation of TORA programme’s implementation is due to the emergence
of institutional issue, namely sectoralism (Fauzi 2011)2. This argument is
supported by the previous study (Larastiti et al. 2020) that examines land
tenure policies in Indonesia.
Using the case of Balumpewa Village, this chapter aims to observe
whether it is only this institutional issue, i.e., the sectoralism, caused the issue
in the implementation of the TORA programme or there are other issues.
This chapter seeks the exact root problems causing agrarian reform at forest
areas to be difficult to implement. It also attempts to capture the dynamics
occuring at village level during the participatory implementation of the
TORA programme supported by villagers and the district-level government.
To achieve these goals, a fieldwork was conducted in 2020.
METHODOLOGY
The first part of this research, conducted from July until October 2020, was
a comprehensive desk study to unpack and review agrarian reform policies
and regulations in Indonesia, from Soekarno’s Era to the era of Jokowi. The
second part of this researcher was conducting a fieldwork on November-
December 2020 in Balumpewa Village, Sigi District, Central Sulawesi
Province. During this fieldwork, direct observations among farmers and
two rounds of Focus Group Discussions (FGDs) with various stakeholders
related to the theme of this research followed by in-depth interviews were
organized.
The first round of FGD is to examine the history of Balumpewa Village,
both from an administrative perspective and based on folklore. We were
2 A part of this dissertation discusses (1) How did Plantation Land become Excluded from the
Land Redistribution Programme and (2) How did Forestry Become Separate from the Agrarian,
and How did Forestland become Excluded from the Land Redistribution Programme? It shows
the sectoralism of regulation and institution in Indonesia. Currently, forests are included in the
agrarian reform programme but are too complicated to implement, so this is the main topic of this
chapter.
Forestry Agrarian Reform in Balumpewa Village, 153

Central Sulawesi Province
reviewing the history of land tenure, tenure systems in the village and
questioning what mechanisms apply in the management of agrarian
resources—studying the community’s livelihood system and the division
of roles or work between men and women in the direction of agrarian
resources. The second FGD focused on examining conflicts related to
agrarian resources. History of forest area determination in Balumpewa
Village. In addition to reviewing the implementation of agrarian reform in
the village and questioned the involved institutions and the role of each, we
also identify the participation process of the local community, the motive
why people want to carry out agrarian reform, and the obstacles they face.
The village government, customary leaders, religious leaders, village-level
agrarian reform implementing committees, village participatory mapping
implementers, women, youth, and ordinary community members attended
both FGDs.
The follow up in-depth interview utilised a set of questionnaires to
see the level of control, access, and distribution of agrarian resources both
within and outside the forest area at the farm household level. In-depth
interviews were also conducted to know the contestation and dynamics
of the application of agrarian reform in Balumpewa Village. Some of
the stakeholders interviewed are the government, non-governmental
organisations (NGOs), and local communities (Table 1).

Table 1. The List of Stakeholders Interviewed
Government NGO Local Community
The Head of Sigi District Mining Advocacy Network Head of Balumpewa village
in Sulawesi Tengah Province
(JATAM)
GTRA Sigi District Yayasan Tanah Merdeka Implementing Committee
(YTM) for Agrarian Reform in
Balumpewa Village
Representative of Sigi Indonesian Forum for the Women's activists of the
County Council Member Environment in Sulawesi agrarian reform Balumpewa
Tengah (WALHI) Village
Sigi District Planning, The Indigenous Peoples' Youth's activists of the
Research and Development Alliance of the Archipelago agrarian reform Balumpewa
Agency (Bappelitbang in Sulawesi tengah Province Village
Kabupaten Sigi (AMAN)
Sulawesi Tengah Provincial Karsa Institute the head of hamlet in the
Forestry Service (Dinas Balumpewa Village
Kehutanan Provinsi
Sulawesi Tengah)
Forest Area Designation Consortium for Agrarian Balumpewa Village
Bureau (Regional Office Reform of Sulawesi Tengah Religious Stakeholders
of Forestry Planning) XVI Province (KPA)
in Palu (Balai Pemantapan
Kawasan Hutan Wilayah
XVI Palu)
Nature Resources and Women Solidarity in Palu Balumpewa Village
Ecosystem Conservation of (Solidaritas Perempuan) Customary Stakeholders
Sulawesi Tengah Province
(BKSDA Sulawesi Tengah)
The district office of the Yayasan Merah Putih (YMP) Balumpewa Village
Ministry of Agrarian and Agrarian Reform
Spatial Planning/National Participatory Mapping
Land Agency of Sigi District Team
(ATR/BPN Kabupaten Sigi)
Regional office of the Sigi District GTRA
Ministry of Agrarian and Volunteers for Survey and
Spatial Planning/National Mapping
Land Agency of Sulawesi
Tengah Province (ATR/
BPN Provinsi Sulawesi
Tengah)
Sociology Student
Association of Tadulako
University Students Group
(HIMASOS)

Socio-Ecological Landscape of Balumpewa Village
Balumpewa is an old village that has been inhabited by the Kaili people,
especially sub-ethnic Inde, long before Indonesia became independent
(Kasmiati 2021). According to local people’s stories, the Church was built
in the Balumpewa area in about the 1900s. The surrounding area hills,
mountains, and valleys were inhabited before the Church was founded. It
proves that the Kaili Inde people in Balumpewa lived there before the 1900s.
Based on the local history, Balumpewa is once referred to as Balumpeva,
representing a woman (a widow) who resisted. Kaili Inde people there call
their ‘village’ Ngata (Ngata Balumpeva). However, in terms of administrative
rules, Ngata Balumpeva was changed and known as Balumpewa Village.
The village’s topographical landscape comprises 50% of the highland
area, and the remaining 50% fall into the mountain category3 (BPS 2020).
Balumpewa is a village that is still very homogeneous in terms of inhabitant
ethnic composition. The Kaili Inde people (Topo Inde) are now also Salvation
Army Christians. The main livelihood of the Balumpewa’s people is from
farming. As a farmer, the land has significance for the people of Balumpewa.
The land is declared as “Tampa mangelo katuva”4 (a place of seeking life).
Therefore, losing land means losing the source of their life itself. Losing
their land is described as living a life on the edge of the nail. This proverb
represents the vulnerability of losing their land. A satire meaning of it is that
a landless one can also be a cultivator yet living under others’ command
and dictation. This way of life is contrary to the egalitarian principle of Topo
Inde’s life where everyone strives to support themselves and their household
by cultivating the land (Kasmiati 2021).
Over the past few decades, the Topo Inde community in Balumpewa
has had to face the real threat of losing access to their ancestral lands due
to unilateral claims on behalf of the state. The land they have worked on
suddenly entered into state control called a protected forest and conservation
area. If a unilateral claim based on the country’s positive laws is upheld, then
at any time, Balumpewa citizens may lose access to their livelihoods.
3 Highlands have a minimum elevation of 500 meters above sea level, while mountains are usually
above 1,000 meters above sea level and have peaked.
4 Terms in Kaili Inde Language

Dominance of State Land in Balumpewa Village
Since hundreds of years ago, the Balumpewa area has been a roaming space,
a place to live, cultivated land, fields, and forest reserves for the Topo Inde
people. However, since 1960s, this area has undergone a process of negaraisasi
(state-izing)5 in line with the policy of designating and determination of the
country’s forest areas. In Sigi District, the state-izing of the land process
began from the Decree (SK) of the Minister of Agriculture Number: 843/
Kpts/Um/11/1980 dated 25 November 1980, concerning the establishment
of the Wera Nature Recreational Park (Taman Wisata Alam or TWA Wera)
covering an area of 250 hectares. This decree became a foothold for the
birth of subsequent regulations that facilitate annexing the living space of
Balumpewa villagers. On 28 October 2014, based on the Decree of the Minister
of Forestry Number: SK.6586/Menhut-VII/KUH/2014, the TWA Wera was
set to be 349.39 hectares. From these two decrees, there is a difference in the
area of TWA Wera which almost reached 100 hectares, between Appointment
Letter in 1980 and Determination Letter in 2014. If referring to the Nature
Conservation Agency (Balai Konservasi Sumber Daya Alam or BKSDA) of
Central Sulawesi Province in the letter Number: S.961/IV.K-22/2018, this
difference occurs due to the change in the status of protected forest areas into
conservation forests.
In addition to being a conservation forest or TWA, part of Balumpewa
Village also becomes a protected forest through the Decree of the Minister
of Forestry No.869/Menhut-II/2014. This decree caused approximately
5 The term land “state-izing” used in this text refers to Fauzi (2014) as “...the institutional practice of
the colonial government (later continued by the post-colonial government) that exerted state power
to control forest resources. In more detail, Fauzi writes about the “nationalisation” of customary
territory... (which includes settlements, agricultural land/cultivation, fallow land, grazing areas,
hunting areas, forests containing plants and animals, coasts and beaches, as well as natural
resources. Other lands on Earth), categorised by the government as “state land” and “state forest,”
then based on authority based on legislation, public officials include part or all of these customary
areas as part of the licenses granted by government agencies. Central and local governments to
companies that extract natural resources and produce plantations/forestry/mining to produce
global commodities or government agencies managing conservation areas (national parks, grand
forest parks, and others). This concept of state-izing can also be traced through Fauzi (1999). Similar
to him, Siscawati et. al (2017) emphasized the state-izing phenomenon was already implemented
during the Dutch colonial period which was called domeinverklaring. In 1960, when The Basic
Agrarian Law (BAL) was legislated, there was an ambiguity in State’s rights on lands and natural
resources. Even though The BAL had a mission to cancel domeinverklaring, the State’s right which
stipulated in Article 2 of the 1960 BAL that “…allowed central government to have supreme
authority to regulate and manage land and natural resources as well as to determine property
relations” remained. Through the BAL, the government also failed to institutionalize customary
tenure systems. Consequently, the government made a blunder declaration by claiming 70% of its
territory as forest zone (under State’s control) without any reference to existing local tenure and
land-use systems.” The ambiguity has been exacerbated by establishment of Forestry Law No. 5
of 1967 which revitalised the principle of domeinverklaring. Brief history of forest land use and
management can be seen in Resosudarmo et al. (2012)

2,050.53 hectares (91.06%) of the Balumpewa Village area to become a forest
area with protected and conservation functions. This process of “state-izing”
led to extended state control in the Balumpewa Village. This process implies
that land tenure between the state and citizens becomes unequal, as shown
in Table 2 below.
Table 2. Land Tenure in Balumpewa Village
Land Status Area (Hectares)

Community-owned land 201.53
State Forest 2,050.83
Total Area 2,252.18
Source: Indonesian: Network for Participatory Mapping (JKPP) Spatial Data, 2019
Officially, the land in Balumpewa Village that people can use for
agricultural or residential activities is less than 10% of the total village area, or
only equal to 201.53 hectares. This kind of unequal situation forces citizens to
cultivate on state lands. The following picture will help us imagine that the
inequality in controlling agrarian resources in Balumpewa Village is sharp6.
The vast area of Balumpewa Village is not able to help the people’s
living conditions be better if the unequal land allocation, as shown in Figure
1, is not resolved first. The highest percentage of land owned by residents is
less than 0.5 hectares. Meanwhile, for land tenure in forest areas, the highest
percentage is at 0.5-2 hectares7. The means that the people’s arable land is
wider in the forest area than in non-forest estate area (Area Penggunaan Lain/
APL). The results of this survey are in line with the macro data that shows
the inequality of land allocation in Balumpewa Village.
6 Shohibuddin (2020) termed this kind of inequality as “allocation inequality,” namely the condition
of land tenure inequality caused by unfair state policy in allocating land between the interests of
the people and for the benefit of sectoral control on a large scale, such as for forest areas, plantation
HGU, and so on. This is distinguished from the “inequality of distribution,” which more closely
reflects the unequal land tenure conditions among the farmers themselves.
7 The data is processed from the survey results of the research team.

Figure 1. Map of Land Tenure in Balumpewa Village
Source: Results of participatory mapping of the Balumpewa village community

Finally, the process of state-izing creates unequal allocations among
the state and citizens (especially villagers) are also between citizens, as can
be seen by the Gini Inequality Index of Agricultural Land Tenure in Dolo
Barat sub-district is 0.5143 and for the Gini Inequality Index of Rice Fields in
Balumpewa Village is very high, with a value of 0.9576. It is bringing people
into a vulnerable situation. Now they live and farm on state land illegally so
that they are constantly overshadowed by evictions, violence, and conflicts
that can occur at any time.
The sharpening of inequality in control, ownership, use, and allocation
of land/forest/other natural resources is one of the root causes of structural
agrarian conflicts (Fauzy 2013). This situation motivates residents to move
from controlling the land through cultivation to participatory efforts to
reclaim their source of livelihood as legal property. Population growth has
an insignificant contribution to the process. The population in Balumpewa
Village did not experience significant growth. The Balumpewa Village
Government data shows that the total number of population of the village
is only around 684 in 2020. Like a traditional village, the marriage process
in this village, especially with parties from outside of the community, rarely
occurs so that population growth is not high.

From Cultivation to Participation for Agrarian Reform
For several years after the state-izing process took place in Balumpewa
Village, the citizens were away from cultivation as they had lived for decades
before. It is an effort to maintain the livelihoods of Balumpewa residents
when the land tenure in their village is uneven, as described in the previous
section. So far, the residents of Balumpewa Village view the cultivation of
land deep in the forest area as normal, as they consider the land currently
included in the forest area as their land cultivated for generations to be
managed for farming (Kasmiati 2021).
So, it is not surprising if the land included in the forest area has become a
garden overgrown with seasonal and annual plants. We can find candlenut,
cacao or coffee trees and acres of land to grow corn and kidney beans. All of
these commodities are a source of livelihood for the people of Balumpewa.
In addition to cultivating as has been going on, anxiety about the status of
the land that has been designated as a forest area overshadows the people’s
days. Therefore, when the Sigi District Government established agrarian
reform as one of the main programmes, the Balumpewa Village Government
and the residents and all traditional leaders in this village saw the agrarian
reform programme as an opportunity to free their arable lands from state
control.
Officially the knowledge about the agrarian reform programme in
Balumpewa village was introduced by the Sigi District Government (Nugraha
2017; YMP 2019; Antaranews.com 2019). Starting in 2016 and establishing
the GTRA also carried out public dissemination activities of participatory
forest and land governance within agrarian reform and agrarian reform
preparation workshops at the district level that invited all villages. In 2017,
the Head of Sigi then wrote to the village head to participate in participatory
mapping training activities represented by three villagers appointed by
the village head. This activity is facilitated by the Sigi district government
and various NGOs that support the implementation of the agrarian reform
programme (Kasmiati 2021) .
Balumpewa residents participated in this mapping training process.
Those trained then return to the village to hold village deliberations, mainly
to discuss budgeting through the Medium-term Village Development Plan
(Rencana Pembangunan Jangka Menengah Desa or RPJMDES) for participatory
mapping activities in Balumpewa Village. In addition, the village-level
Agrarian Reform Acceleration Programme (Programme Percepatan Reforma

Agraria or PPRA) was also formed. The establishment of PPRA is carried
out through deliberation in the village openly. PPRA formed in this village
then held various meetings to determine the time of implementation of
participatory mapping in the village, which is also an essential part of
determining agrarian reform’s subject and object that Balumpewa villagers
will submit.
The process of determining the subject and object of agrarian reform
proposed by the community is carried out through deliberation. In the
rules, deliberation to determine the subject of the object is held twice, but in
practice, it can be adjusted to the needs at the village level. The deliberation
process in determining the subject of the agrarian reform object is carried out
openly involving PPRA, assistants, residents or landowners in the village.
After the spatial data (map) along with the social data (subjects and objects)
for agrarian reform (AR) are completed in a joint deliberation through the
minutes set by the village government, then the data is then submitted to
Balumpewa Village in this case, the GTRA of Sigi District to be proposed
as the Sigi District TORA programme. The capacity of the residents and the
GTRA of Sigi District has only reached here to propose the TORA programme
that has been agreed upon by the residents to the Ministry of Environment
and Forestry for the proposal of TORA programme in forest areas and the
Ministry of ATR/BPN for lands outside the forest area so that they can be
officially released to the community.
Local Institutions Supporting the Participation of Balumpewa

Residents
Participation is not something that happens all at once. Instead, it is dynamic
and related to internal factors within the village and those outsides. This
participation is closely related to the peasant movement through an
organised one. Knowledge flow mechanism regarding AR facilitated by
the Sigi District Government and NGOs that support this issue is present
among the people of Balumpewa Village. Such endeavour resulted in the
local community’s increased awareness of their critical position on the land
they are stepping on that is now under state control on behalf of the forest
area. The awareness and participation of villagers did matter because one of
the determining factors in implementing the agrarian reform programme in
the village is the involvement and support of local community stakeholders
(Ndlovu, Mwale, and Zuwarimwe 2021).

Fortunately, the local community’s new knowledge and awareness are
supported by the local institutions in Balumpewa Village. Local institutions
are one of the pillars of farmers’ participation in various efforts to implement
agrarian reform (Kantor Staf Presiden Republik Indonesia 2017; Purwanto
(ed) 2017; Djauhari, Lubis, and Moenir 2018; Nurfatriani and Alviya,
2019). In Balumpewa village, at least three forms of local institutions have
their respective functions: Customs, Religion, and Village Government.
These three institutional forms have arrangements that bind Balumpewa
community lives.
Figure 2. Local Institutions and Collective Actions of Community to
Participate in Supporting agrarian reform
Source: Processed from the researcher’s primary data

These three institutional forms in their daily functions are different
carriages but in the implementation of agrarian reform trying to be united
in one locomotive, which is then boarded by the community on the way
to realize agrarian reform in Balumpewa Village. These three things are
important because we can imagine that if the stakeholders in the three
institutional systems have different orientations, it will certainly break
down farmers’ perspectives and potentially break up the Balumpewa farmer
movement. The roles of each of these local institutions in agrarian reform
efforts supported from below (by leverage) are as follows:

Customary Institutions
These institutions are the oldest form of regulation in the village. Residents
highly respect the prevailing customary system by fully carrying out all the
rituals that bind their life cycle from birth to death. Although traditional
rituals have economic consequences or require very large costs, residents still
try to carry out various mandatory traditional rituals, for more example, see
Kismarsilah and Mulyadi (2004). With such great respect, if the customary
institutions do not agree on the agrarian reform being pursued, it will
certainly become an obstacle at the village level. However, it turned out that
the traditional stakeholders agreed that the agrarian reform could be carried
out because it departed from the established knowledge that even though
the lands in the entire Balumpewa village were ancestral lands, they did not
mean anything before state law. This fact made the traditional stakeholders
in the village agree to encourage access to land that has become part of the
forest area. Therefore, traditional stakeholders are always present in various
deliberations and activities related to the implementation of agrarian reform
in Balumpewa Village. Customary institutions as a medium of the local
community struggling for their right to land are not only carried out in
Balumpewa Village. This institution has been used as a medium against the
hegemony of market capitalism (Dale 2010; Cotula 2013).
Religious Institutions or Churches

The religion adopted by the Balumpewa residents is Salvation Army
Christianity, which makes the Church an inseparable part of their daily
lives. The Church is the center of social activities, and so far, settlements
have grown around the Church (Kasmiati 2021). In implementing the AR
in Balumpewa, the Church functions as a medium to convey messages.
Every time a service is held, various messages related to agrarian reform
activities are announced by the Church so that every person or household
gets information. Collecting personal or household identities to complete
social data as TORA recipient subjects is also always announced through
the Church and continues until the AR implementing committee completes
the subject at the village level. The data will be submitted to the assistant
committee from the district. Meanwhile, the officers participated in reading
forest regulations and prayed for various activities carried out by the
residents, including actions. Although praying is seen as a fragile form of
involvement, prayers from officers are taken as essential and meaningful for
residents because they are believed to have spiritual support to strengthen

residents that what they are doing is on the right path and is approved by
God.
The Village Government

The village government is an institution that comes later in the lives of the
Balumpewa residents but has power that covers all things in the village,
and has the authority and resources (money) to implement various policies
that bind all community members in the village. The village government
encourages the implementation of agrarian reform as a district government
policy that directly assigns village funds to implement agrarian reform.
The village government also facilitates various activities such as meetings,
committee formation, and various correspondence processes needed to carry
out agrarian reform activities in Balumpewa Village.
The collective action of Balumpewa Village residents to participate in
various activities to encourage the implementation of agrarian reform in
their village is supported by the three local institutions. Without the support
of the Village Government, religious institutions, and customary institutions,
it is not easy to imagine the birth of solid participation at the village level.
However, because all Balumpewa residents respect all important institutions
with the same voice, the village participation is solid.
TORA in Balumpewa Villages: Fall Short of Expectation

Although the Balumpewa’s residents have made participatory efforts
to encourage the implementation of agrarian reform in their village, in
reality, the results of the verification of the Inver PTKH Team are falling
short of expectation. The total proposed TORA in Balumpewa Village is
465.01 hectares. However, after being verified by the Inver PTKH team, the
recommendation for boundary changes to be released from forest areas can
be processed as TORA is only 62.03 hectares. This result indeed disappoints
the villagers, considering the various participatory efforts that the residents
have made by sacrificing much time, energy, and budget to own agricultural
land safely.
Unfortunately, this case happens not only in Balumpewa village but
also in all villages in Sigi District which priorly proposed TORA programme
and Social Forestry (SF) scheme (Oktaviana and Naharoh 2021). SF is a
scheme in which local communities obtain forest management licenses, but
land ownership belongs to the government. For the central government, SF

is part of the Agrarian Reform programme. In the Sigi district, both schemes’
implementation is disappointing. It also happens in several regions, such
as Papua, Sumatra, Kalimantan, and Bali (KPA 2019; Faryadi and Elisabeth
2020; Wicaksono 2020; Beritakalteng.com 2020; RCTI+ 2021; Tribunebali.
co.id 2021).
The rejection of TORA submitted by the Balumpewa villagers is related
to the status of the land submitted by the Balumpewa residents. Almost
one hundred per cent of the proposed status comes from forest areas which
in the verification process must pass through various filters following
(1) Presidential Regulation No. 88/2017 concerning Settlement of Land
Tenure in Forest Areas; and (2) Regulation of the Coordinating Minister for
Economic Affairs No. 33/ 2018 concerning Guidelines for the Inventory and
Verification Team of Land Tenure in Forest Areas.
In principle, the above two regulations are still very conventional.
Namely, the process of releasing forest areas still refers to the old regulations
that tend to maintain the status quo of forest areas through the verification
process based on the function of forest areas, require the length of time of
land management worked, as well as distinguish the type of use of TORA in
the forest area whether for plantations, public facilities, social or residential
facilities. So, the above regulations have not supported the PPTKH because
specific programmes—i.e., such as agrarian reform in forest areas—require
special policies.
Learning from the case in Balumpewa Village, the most widely submitted
TORA proposals came from conservation forest as much as 326.29 hectares,
then from protected forest covering an area of 136.37 hectares, and the rest
came from APL covering an area of 2.35 hectares. The filtration process
refers to the two regulations used when verifying the TORA proposal of the
residents of Balumpewa Village. So, the area of TORA in this village differs
greatly between the community’s proposal and the verification results issued
by the government.
This very wide difference is mainly an implication of the status of
conservation forests. TORA in conservation forests cannot be automatically
processed due to resettlement as “the only solution” provided. Meanwhile,
for protected forests, only a portion of the community proposals can be
processed as TORA. While others require changes to the Spatial Plan (RTRW)
first before it can be further processed to be used as a TORA, it can be seen
clearly in Figure 3, which at the same time shows how the filtering process

works and resulted in the area of forest area that may be released for the AR
concerns to be very small.
Figure 3. Flow of Processing the TORA Proposal in Balumpewa
TORA proposal The first The second
Inver Team
from the community filtration filtration
Results
It can't be Proposed For Social

326.29 Hectares processed at all for Forestry 1,653.04 Status quo of
of Conservation TORA because it hectares forest areas
Forest is a conservation
forest.
Requires review of Need to go
RTRW changes: through the
54.53 hectares political process
136.37 Hectares Processed through
of Protected verification by the Can be changed
Forest Inver Team forest area
boundary: 62.03
hectares
Processed Approved when

according to the Can be further the change of
authority of ATR- redistribution: 64.38 RTRW is
2,35 hektar BPN hectares accepted:
area for other
118,91 hectare
land uses
Source: Processed from the researcher’s primary data

The filtration process, as shown above, causes the TORA proposal of
the residents of Balumpewa Village, after going through verification, to
only leave an area of 64.38 hectares which may be designated as TORA.
Thus, what is happening in Balumpewa Village is a small snapshot of the
conditions in the Sigi District or in other parts of Indonesia that are trying to
implement AR in forest areas. The area of land that residents may manage as
a living space for living and farming is very small compared to the land area
allocated as a forest area for various functions.
The case above illustrates how the regulations or policies hinder the
agrarian reform agenda’s implementation in forest areas, especially in
conservation and protection areas. Therefore, the current regulations
need to be reviewed. More diverse policy alternatives are needed besides
resettlement schemes for conservation forests and for protected forests that
are not used for plantation activities or used for gardening, but the tenure
is still less than 20 years, instead of changing the boundaries, Presidential
Regulation Number 88 of 2017 instead suggests to implementing the Social
Forestry (SF) scheme.
The Jokowi Government’s agrarian reform policy, framing SF as part

of the AR8, although in principle these two things have a different spirit in
which Social Forestry is interested in maintaining the status-quo of forest
area. In contrast, forestry agrarian reform primarily aims to overhaul the
allocation gap between the state and citizens. Later is also expected to resolve
the structural inequality of land tenure between citizens.
The government sees the social forestry scheme as a way out for regions
that apply for TORA in areas that have been designated as forest areas with a
protected function in provinces with a forest area equal to or less than 30% of
the total area of watersheds, islands, and the province will be given access to
forest management through the social forestry programme. The means that
in areas where the forest area is still more than 30%, but the proposed TORA
is arable land whose tenure is still less than 20 (twenty) years in a row, the
settlement pattern is to provide access to forest management through the SF
programme9.
Therefore, the PTKH Inver Team results in Balumpewa Village, in
addition to deciding on a recommendation to change the boundary covering
an area of 62.03 hectares, also provide recommendations for social forestry
covering an area of 1,653.04 hectares. These unequal figures show how the
government’s logic views the forestry agrarian reform programme tends to
maintain the quo status of forest areas. McCarthy et al. (2022) term this kind
of thing as ‘adverse formalization’10.
The release of forest areas is challenging with various strict conditions that
work as a filter for TORA proposals submitted by residents. However, this
does not apply to efforts to provide access to SF. The government seems
very generous to the people by proposing access to SF, broad like the figure
above, which reaches thousands of hectares. The problem then is that apart
from the issue of substance which differs significantly from AR as previously
explained, SF itself contains problems in itself both in terms of policy and
implementation, as shown in Pambudi (2020), Naibaho (2019), and Kasmiati
(2019).
8 The Jokowi administration divides the AR policy into two terms, namely ‘Asset Management’
and ‘Access Management’. This asset arrangement is further divided into two, namely land
redistribution and asset legalization. The forestry AR policy is in the category of land redistribution
(release of forest areas). Meanwhile, social forestry is part of the arrangement of access.
9 Referring to the Presidential Regulation of the Republic of Indonesia Number 88 of 2017 concerning
Settlement of Land Tenure in Forest Areas.
10 ‘adverse formalisation’ where the state claims sovereign control of extensive ‘public lands’ and
embarks on formalisation processes that include local populations into new land-based production
systems on adverse terms, (McCarthy et al. 2022).

Caged in Chaos: AR Participatory Actions Under Agrarian Reform
Regulations
Learning from Balumpewa’s case shows that participatory efforts carried
out by villagers as long as the basic rules related to forestry agrarian reform
are still conventional, efforts to achieve the national target of releasing 4.1
million hectares of forest area as TORA are impossible to achieve. Because
even in areas where the District Government and the Village Government
have a political commitment and are serious about running a forestry
agrarian reform and are supported by the participatory work of citizens,
regulations are ultimately suffocating.
The regulations that serve as the basis for implementing the forestry
agrarian reform have standard and rigid requirements. So, it is very inflexible
in responding to socio-ecological situations at the site level. As a result, even
the most participatory proposals cannot be accommodated because the
requirements for releasing forest areas in the current regulations work as a
filtration tool for TORA proposals for forest areas.
So the Government, in this case, the PTKH Inver Team working at the
village level, should be able to show scientific evidence of the verification
process carried out to residents, as shown by Junarto and Djurdjani (2020)
in their text11. However, when the PTKH Inver Team went to the Village to
carry out socialisation, they only brought a map of the Verification results
and without adequate explanation except on behalf that most of the land that
the Balumpewa residents proposed for TORA was in forest areas, especially
Conservation and Protection forests, so they could not be processed further.
The Presidential Regulation No. 88/2017 concerning Settlement of Land
Tenure in Forest Areas indicates that the types and functions of forests
proposed as TORA is a shield against releasing forest areas.
This is in line with the findings of Dhiaulhaq and McCarthy (2019) that
the rights of indigenous people in state law are limited and bureaucratised.
Therefore legal reforms are needed to provide legal and political
empowerment and political, institutional and structural barriers reduction
11 In this manuscript, Junarto and Djurdjani show how to prove the 20 years of mastery. First, image
change data is used to answer questions regarding spatial pattern analysis resulting in abstraction
and classification of the distribution of community land biophysical appearances in forest areas in
forest biophysical pattern maps. Then, based on the map, scientific questions form what, when, and
where are answered. Meanwhile, other scientific questions related to social criteria are answered
through why, who, and how. This kind of thing should have been shown by the PTKH Inver Team
to residents when conducting socialization in the village, but it was not carried out, only showing
the final result in the form of the PTKH Inver Team’s decision regarding the amount of land that
could be released and suggestions to encourage SF in Balumpewa Village.

to reduce the exclusion and marginalisation of local communities from land
access and control.
CONCLUSION
This chapter explores the obstacles to implementing agrarian reform in forest
areas. We explore Balumpewa Village’s case to show how the dynamics of
the agrarian reform programme are beyond institutional and sectoral issues.
The majority of land in this village is forest. Along the time, the Balumpewa’s
residents have been cultivating the forest. However, since the government
of the Sigi district adopted the agrarian reform programme by president
Jokowi, people in this village are struggling to enforce it. The participation of
the Balumpewa community in implementing agrarian reform programmes
is very high. They take much collective actions, which is supported by
local institutions such as customary institutions, religious institutions or
churches, and the village government. This indicates the enthusiasm and
determination of all components in Balumpewa Village to carry out the
agrarian reform programme.
Nevertheless, the community participation has not been seriously
considered and their efforts were unable to influence the Inver PTKH Inver
Team’s decision-making to change or release the status of the forest area
to be a TORA as requested by their proposals either. The results of the
verification issued by the PTKH Inver Team become a definitive decision.
This ‘filtration process’ becomes the right term used to show how the
regulation (1. Presidential Regulation No. 88/2017 concerning Settlement of
Land Tenure in Forest Areas; and 2. Regulation of the Coordinating Minister
for Economic Affairs No. 33/2018 concerning Guidelines for the Inventory
and Verification of Land Tenure Teams in Forest Areas) is a tool to maintain
the status quo of forest areas.
Implementing forestry agrarian reform has become problematic because
of conventional regulatory and rigid bureaucracy. The community does not
have the opportunity to sue or reject the verification results. So, various
participatory works ended up deadlocked because of rigorous regulations
and generalised the whole situation at the site level. Various participatory
efforts of citizens to propose forest areas as objects of agrarian reform have
stalled. Too many conditions to be met while the offer of solutions from the
state is restricted. The current regulations have not responded to the diversity
of socio-ecological situations in the villages that apply for agrarian reform.

A new and comprehensive mechanism for resolving land tenure patterns in
forest areas is needed as it is supposed to be more flexible and diverse.
More options are needed in the process of releasing conservation and
protected forest areas. Various alternatives need to be presented following
the situation of the ecological and social landscape in the village. For example,
it considers the level of citizens’ needs for land, the possibility of conflict, and
various other situations of the local community. Therefore, it is necessary to
immediately formulate a policy that can catch up with the factual condition
of land tenure in a very varied forest area.
Acknowledgments
This article is part of the Sajogyo Institute research supported by the
Knowledge Sector Initiative (KSI). Our thanks to both of these institutions.
Furthermore, we would like to thank all the local community and agrarian
activists we met during our activities in Sulawesi Tengah, especially in Sigi
District, who have assisted us during the research process.
REFERENCES
Antaranews.com. 2019. Warga 103 desa di Sigi ikut konsolidasi reforma agraria.
[online] https://www.antaranews.com. https://www.antaranews.
com/berita/1052714/warga-103-desa-di-sigi-ikut-konsolidasi-
reforma-agraria [Accessed 20 Mar. 2022].
Ariyo Wicaksono, R. 2020. Perhutanan Sosial Mestinya untuk Pemulihan
Kedaulatan Masyarakat. [online] https://betahita.id/. https://betahita.
id/news/lipsus/5745/perhutanan-sosial-mestinya-untuk-pemulihan-
kedaulatan-masyarakat.html?v=1606983148 [Accessed 20 Mar. 2022].
Bachriadi D. 2017. “Reforma Agraria untuk Indonesia: Kritik atas Reforma
ala SBY.” Working Paper series in kebijakan agraria dan pembangunan
paska Orde Baru, Vol.1 No.07/WP-KAPPOB/I/2017. ARC, Bandung.
BAPPENAS. 2014. Rencana Pembangunan Jangka Menengah Nasional
(RPJMN) 2015-2019. [online] Jakarta: BAPPENAS. https://djsn.go.id/
storage/app/media/RPJM/BUKU%20I%20RPJMN%202015-2019.pdf
[Accessed 16 Mar. 2022].
BPS. 2020. Kecamatan Dolo Barat dalam Angka 2020. Badan Pusat Statistik
Kabupaten Sigi.
Berdikarionline.com. 2017. Ini 7 Alasan Mengapa Programme TORA Jauh

dari Harapan Rakyat. [online]
https://www.berdikarionline.comhttps://www.berdikarionline.
com/ini-7-alasan-mengapa-programme-tora-jauh-dari-harapan-
rakyat/ [Accessed 15 Mar. 2022].
Beritakalteng.com. 2020. Manfaat TORA Sangat Diharapkan Masyarakat
Kotim Kata Dewan ini. https://beritakalteng.com/2020/05/19/
manfaat-tora-sangat-diharapkan-masyarakat-kotim-kata-dewan-ini/
Carranza, D. T. 2015. Agrarian reform and the difficult road to peace in the
Philippine countryside. Norway: The Norwegian Peacebuilding Resource
Centre.
https://reliefweb.int/sites/reliefweb.int/files/resources/Carranza_
NOREF_Agrarian%20reform%20and%20the%20difficult%20road%20
to%20peace%20in%20the%20Philippine%20countryside_Dec2015_
FINAL.pdf
Celebespos.com. 2020. Datangi Kantor BPKH Wilayah XVI Palu, Ribuan
Masyarakat Sigi Gelar Unjuk Rasa
http://celebespos.com/2020/08/11/datangi-kantor-bpkh-wilayah-
palu-ribuan-masyarakat-sigi-gelar-unjuk-rasa/ [Accessed 15 Mar.
2022].
Djauhari, Moh., A.S. Lubis, and N.A. Moenir. 2018. Bunga Rampai: Strategi
Percepatan Tora Dan Perhutanan Sosial Seri: Penguatan Pranata Dan
Prakarsa Lokal Dalam Penyiapan Dokumentasi Dan Perizinan. 1st ed.
http://kpshk.org/wp-content/uploads/2018/07/Strategi-Percepatan-
TORA-dan-PS.pdf [Accessed 16 Mar. 2022].
Cotula, L. 2013. The new enclosures? Polanyi, international investment law
and the global land rush. Third World Quart. 34 (9): 1605–1629.
https://doi.org/10.1080/ 01436597.2013.843847.
Dale, G. 2010. Social democracy, embeddedness and decommodification: on the
conceptual innovations and intellectual affiliations of Karl Polanyi. New
Poli. Econ. 15 (3), 369–393. https://doi.org/10.1080/13563460903290920.
Dhiaulhaq, A. and J.F. McCarthy. 2019. Indigenous Rights and Agrarian
Justice Framings in Forest Land Conflicts in Indonesia. The Asia Pacific
Journal of Anthropology, 1 (21). doi:10.1080/14442213.2019.1670243
Ekawati, F.N.F., M.N. Salim, and W. Utami. 2019. Pemetaan Partisipatif Guna

Pengusulan Tanah Obyek Reforma Agraria (TORA) dalam Kawasan
Hutan di Kabupaten Ogan Komering Hulu. Tunas Agraria, 2(3):24–48.
https://jurnaltunasagraria.stpn.ac.id/index.php/JTA/article/
view/37/56
Faryadi, E. and Elisabeth, A. 2020. Memeriksa Programme TORA dan
Perhutanan Sosial di Jayapura dan Keerom. Wacana: Jurnal Transformasi
Sosial 38: 197–210.
Fauzi N. 1999. Petani & Penguasa: Dinamika Perjalanan Politik Agraria
Indonesia. Yogyakarta, INSIST Press bekerjasama dengan Konsorsium
Pembaruan Agraria (KPA).
_______ . 2011. The Resurgence of Land Reform Policy and Agrarian Movements
in Indonesia. A PhD Dissertation in Environmental Science, Policy and
Management. University of California, Berkeley
_______. 2013. Rantai Penjelas Konflik-Konflik Agraria Yang Kronis,
Sistemik, dan Meluas di Indonesia.’ Bhumi, Jurnal Agraria dan Pertanahan
12, no. 37 (April).
_______. 2014. ‘Masyarakat Hukum Adat Adalah Bukan Penyandang Hak,
Bukan Subjek Hukum, dan Bukan Pemilik Wilayah Adatnya. Wacana
Jurnal Transformasi Sosial, no. 30, (XV): 25–50.
Jemali, V. 2017. Warga di Sigi Terlibat Aktif Memetakan Lahan.
https://www.kompas.id/baca/nusantara/2017/08/29/warga-di-sigi-
terlibat-aktif-memetakan-lahan/
JKPP. 2017. Kab Sigi Target Selesaikan Reforma Agraria Di 125 Desa.
https://jkpp.org/kab-sigi-target-selesaikan-reforma-agraria-di-125-
desa/ [
Junarto, R, Djurdjani. 2020. Pemetaan objek reforma Agraria dalam
kawasan hutan (studi kasus di Kabupaten Banyuasin). Bhumi, Jurnal
Agraria dan Pertanahan 6, no. 2: 219-235.
Kantor Staf Presiden Republik Indonesia. 2017. PELAKSANAAN REFORMA
AGRARIA Arahan Kantor Staf Presiden: Prioritas Nasional Reforma Agraria
Dalam Rencana Kerja Pemerintah Tahun 2017.
http://kpa.or.id/publikasi/download/ac891-strategi-nasional-
reforma-agraria.pdf.

Kasmiati. 2019. ‘Ilusi Perhutanan Sosial: Legalitas Tanpa Kedaulatan, Kasus
Hutan Kuta Tandingan, Desa Parung Mulya Kecamatan Ciampel
Kabupaten Karawang.’ Dalam ‘Menakar Kebijakan Rfeorma Agraria
dan Perhutanan Sosial Laporan Penelitian Tiga Kebijakan Enam Studi
Kasus’, Bogor, Sajogyo Institute.
_________. 2021. “Studi Kebijakan Pendaftaran Tanah dan Kriteria TORA
di Desa Balumpewa, Kecamatan Dolo Barat, Kabupaten Sigi, Sulawesi
Tengah”. Working Paper Sajogyo Institute. [online] sajogyo-institute.
org. Bogor: Sajogyo Institute. Available at: https://sajogyo-institute.
org/studi-kebijakan-pendaftaran-tanah-dan-kriteria-tora-di-desa-
balumpewa-kecamatan-dolo-barat-kabupaten-sigi-sulawesi-tengah.
Kismarsilah and Mulyadi. 2014. Legalitas perkawinan pada suku bangsa Kaili
Inde di Balumpewa Sulawesi Tengah. Master’s thesis in Antropologi.
Universitas Gadjah Mada.
Kompas.com. 2014. “Nawa Cita”, 9 Agenda Prioritas Jokowi-JK.
https://nasional.kompas.com/read/2014/05/21/0754454/.Nawa.
Cita.9.Agenda.Prioritas.Jokowi-JK
KPA. 2019. “Realisasi Redistribusi Tanah dari Pelepasan Kawasan Hutan
Masih Dibawah Satu Persen.” [online] http://kpa.or.id/. Available at:
http://kpa.or.id/media/baca/peristiwa/499/Realisasi_Redistribusi_
Tanah_dari_Pelepasan_Kawasan_Hutan_Masih_Dibawah_Satu_
Persen/ [Accessed 20 Mar. 2022].
Luthfi, A.N. 2018. Reforma kelembagaan dalam kebijakan Reforma Agraria
era Joko Widodo-Jusuf Kalla. Bhumi, Jurnal Agraria dan Pertanahan 4, no.
2, (Nov.): 140-163.
Larastiti C., S. Arofat S., and A. Jaetuloh A. 2020. ‘Enclosure dan Kebijakan
Penguasaan Lahan’, Working Paper Sajogyo Institute No. 01/2020-WP
SAINS, Bogor, Sajogyo Institute.
McCarthy, John F., A. Dhiaulhaq, Suraya Afiff, and Kathryn Robinson. 2022.
Land reform rationalities and their governance effects in Indonesia: Provoking
land politics or addressing adverse formalisation? Geoforum 132 (2022): 92–
102. DOI. 10.1016/J.GEOFORUM.2022.04.008
Mitchell, R., A. Safik, and M.L. Dodd. 2020. Status Report on Indonesian
Land Reform. https://eprints.uai.ac.id/1690/1/ILS0140-21_Laporan-
Penelitian.pdf

Naibaho J. M. 2019. ‘Implementasi Perhutanan Sosial Muara Gembong
Percontohan Yang Tidak Layak Dicontoh’, dalam ‘Menakar Kebijakan
Rfeorma Agraria dan Perhutanan Sosial Laporan Penelitian Tiga
Kebijakan Enam Studi Kasus’, Bogor, Sajogyo Institute.
Ndlovu W., Mwale M., and Zuwarimwe J. 2021. Challenges and Critical
Success Factors for Rural Agrarian Reforms in Limpopo Province,
South Africa. AGRARIS: Journal of Agribusiness and Rural Development
Research. Vol. 7 No. 2 July December 2021, 191-206.
Nugraha, I. (2017). Pemerintah Sigi Komitmen Jalankan Reforma Agraria, Seperti
Apa? [online] https://www.mongabay.co.id. Available at:
https://www.mongabay.co.id/2017/03/25/pemerintah-sigi-
komitmen-jalankan-reforma-agraria-seperti-apa/ [Accessed 20 Mar.
2022].
Nurfatriani, F. and I. Alviya. 2019. Efektivitas Kebijakan Alokasi Lahan 12,7
Juta Ha Untuk Perhutanan Sosial: Studi Kasus Provinsi Nusa Tenggara
Barat Dan Daerah Istimewa Yogyakarta. Jurnal Analisis Kebijakan
Kehutanan, 16(1):47–66.
https://www.researchgate.net/publication/333747797_Efektivitas_
Kebijakan_Alokasi_Lahan_127_Juta_Ha_Untuk_Perhutanan_Sosial_
Studi_Kasus_Provinsi_Nusa_Tenggara_Barat_Dan_Daerah_Istimewa_
Yogyakarta_Effectiveness_of_127_Million_Ha_Land_Allocation_
Policy_.
Oktaviana, G. and L. Naharoh. 2021. Studi Kebijakan TORA di Tiga Jenis
Kawasan Di Desa Bunga, Kecamatan Palolo, Provinsi Sulawesi Tengah.
Working Paper Sajogyo Institute. sajogyo-institute.org. Bogor: Sajogyo
Institute.
https://sajogyo-institute.org/studi-kebijakan-pendaftaran-tanah-
dan-kriteria-tora-di-desa-bunga-kecamatan-palolo-kabupaten-sigi-
sulawesi-tengah/.
Pambudi A.S. 2020. ‘The Development of Social Forestry in Indonesia: Policy
Implementation Review, 2007-2019’, JISDeP – The Journal of Indonesia
Sustainable Development Planning I, no. 1 (April): 57—66.
Panahpapua.or.id 2021. Perhutanan Sosial dan Reforma Agraria tidak Mengubah
Kesenjangan Penguasaan Lahan di Tanah Papua, Masyarakat Sipil Kecewa
Kinerja Pokja PS. https://panahpapua.or.id/berita/perhutanan-sosial-

dan-reforma-agraria-tidak-mengubah-kesenjangan-penguasaan-lahan-
di-tanah-papua-masyarakat-sipil-kecewa-kinerja-pokja-ps .
Purwanto, E. (ed). 2017. Tantangan Perhutanan Sosial dan Peran CSO.
Prosiding Lokakarya “Strategi Penguatan Perhutanan Sosial dan
Peran CSO”, Bogor, 22-23 October 2015. Bogor, Indonesia: Tropenbos
Indonesia.
RCTI+ 2021. Kritik Pemberihan Lahan Konsesi Kepada PP Pemuda Muhammadiyah,
KPA: Seharusnya Diberikan Kepada Masyarakat Yang Berhak - News+ on
RCTI+.
https://www.rctiplus.com/news/detail/nasional/972457/kritik-
pemberihan-lahan-konsesi-kepada-pp-pemuda-muhammadiyah-kpa-
seharusnya-diberikan-kepada-masyarakat-yang-berhak [Accessed 20
Mar. 2022].
Resosudarmo, B.P., A.A. Nawir, I.A.P. Resosudarmo, and N.L. Subiman.
2012. Forest Land Use Dynamic in Indonesia. In A. Booth, C. Manning
and T.K. Wie (eds.), Livelihood, the Economy, and the Environment in
Indonesia. Jakarta: Yayasan Obor, 20-50.
Saputra, N. A. 2020. Siaran Pers Forum Desa Kabupaten Sigi Kawal Reforma
Agraria.
http://kpa.or.id/media/baca2/siaran_pers/200/Siaran_Pers_Forum_
Desa_Kabupaten_Sigi_Kawal_Reforma_Agraria/ [Accessed 15 Mar.
2022].
Savitri L. A. 2020. Reforma Agraria Kehutanan dan Perluasan Privatisasi
Tanah di Merauke. Wacana, jurnal Transformasi Sosial 38:171-196.
Siscawati M., et. al. 2017.” Overview of forest tenure reforms in Indonesia”.
Working Paper 223. Bogor, Indonesia: CIFOR.
Shohibuddin M. 2020. ‘Ketimpangan Agraria di Indonesia: Pendekatan Studi,
Kondisi Terkini dan Kebijakan Penanganan’, Bogor, IPB Press.
Shohibuddin M. and Salim N., (Eds) 2012. ‘Pembentukan Kebijakan Reforma
Agraria, 2006-2007: Bunga Rampai Perdebatan’, Yogyakarta, STPN Press
dan Sajogyo Institute.
Suharti, S. 2019. Kebijakan pengelolaan hutan berbasis masyarakat: dari partisipasi
menuju inklusi. Orasi Pengukuhan Profesor Riset Bidang Ekonomi Sosial
Kehutanan, Bogor: Badan Penelitian Pengembangan dan Inovasi.

Susetyo, P. D. 2021. Problem Mendasar Perhutanan Sosial.
https://www.forestdigest.com/detail/1178/problem-perhutanan-
sosial
Tempo.co. 2021. Pakar Hukum Unair: Reforma Agraria Jokowi Tidak Menyentuh
Redistribusi Tanah.
https://nasional.tempo.co/read/1510479/pakar-hukum-unair-
reforma-agraria-jokowi-tidak-menyentuh-redistribusi-tanah [Accessed
15 Mar. 2022].
Tribunebali.co.id. 2021. Sisa Kasus Reforma Agraria Desa Sumberklampok,
Pemkab Fasilitasi Percepat Realisasi.
https://balitribune.co.id/content/sisa-kasus-reforma-agraria-desa-
sumberklampok-pemkab-fasilitasi-percepat-realisasi [Accessed 20 Mar.
2022].
Winarti, Joko, S. and W. Budiman, W. 2021. An Analysis on Public
Participation in Agrarian Reform Programme in Jenawi Sub-district of
Karanganyar Regency. International Journal of Business Continuity and
Risk Management, 11(2-3): 237–246.
Wirawan, E. 2018. 5 Faktor “Policy Trap” Programme Perhutanan Sosial.
http://staging-point.com/read/2018/07/01/225041/5.Faktor.Policy.
Trap.Programme.Perhutanan.Sosial
YMP. 2017. Susun Roadmap Hutan Adat Sulteng. https://www.ymp.or.id/
susun-roadmap-hutan-adat-sulteng/
______. 2019. Tata Guna Lahan Berkelanjutan di Kabupaten Sigi.
https://www.ymp.or.id/tata-guna-lahan-berkelanjutan-di-
kabupaten-sigi/

Chapter 9
Elements of Maritime Logistics

Transaction Costs in Islands Province of
North Maluku, Eastern Part of Indonesia
Chairullah Amin, Heti Mulyati, Eva Anggraini, Tridoyo Kusumastanto,
and D.S. Priyarsono
INTRODUCTION
Indonesia, as an archipelagic country, faces challenges in realising an
efficient maritime logistics system. According to the World Bank report
(2018), Indonesia’s logistics performance has several weaknesses. Namely,
the distance to export and import is far; the integration of industrial zones
with port infrastructure is still weak; the load time, the number of documents,
and the clearance time are still relatively high. A study conducted by Cho
(2014) mentioned that high logistics costs incurred at container ports are
associated with completing the procedures to export or import the goods.
These costs include document fees, administrative fees for customs clearance
and technical control, custom brokers, terminal handling charges, and inland
connection. Uncoordinated logistics chain is another factor proved causing
logistic costs to increase. The number of container loads returning empty
and low coordination systems could result in expensive administrative costs
(Mohamed-Chérif and Ducruet 2016).
The logistic issues taking environmental and institutional uncertainty
into account play a significant role in creating transaction costs (Huo et al.,
2018). In maritime logistics, port institutions should adapt to the changing
demands and dynamics of the global market. Ports are needed to make
adjustments to meet international standards both in physical infrastructure
177
and institutional infrastructure. Stough (2005) suggests that institutional
reform in port management is fundamental in dealing with the crisis and
rapid globalisation dynamics. Port authorities and maritime stakeholders
in the regions should be sensitive to environmental changes and market
uncertainty fluctuations that are difficult to predict and vulnerable to global
economic volatility.
Transportation costs, inadequate transportation systems, and timeliness
in the supply chain are components of the logistics system that can impact
maritime logistics costs. Higher port efficiencies are associated with lower
maritime transport costs (Lei and Bachmann 2020). In particular, for island
areas, irregular boat schedules, high sea transportation costs, and minimal
transportation modes are obstacles in minimizing logistics costs (Briguglio
1995). Similarly, an organized transportation system and unsatisfactory
timeliness in the supply chain created an inefficient logistics service system
in archipelago regions (Kerr 2005).
The research in this chapter took a case in North Maluku, one of the
island provinces in the eastern part of Indonesia. The efficiency of maritime
logistics costs, therefore, is crucial for the island’s economic development.
This chapter attempts to analyse the factors that affect maritime logistics costs
in these archipelagic area using a transaction cost economic approach. The
islanders of this area live with the availability of minimal port infrastructure
facilities, poor land connectivity, inefficient governance in setting tariffs, and
high transportation costs. Hence, it is assumed that there are high transaction
costs in the maritime logistics system for transporting goods between islands.
The transaction costs in the maritime logistics system include operational
costs during the production process, land transportation rental costs, loading,
and unloading costs at island ports. Other costs are administrative costs, sea
transportation rental fees to the destination island, loading and unloading
costs at the destination island port, land transportation costs to warehouses
or storage areas, packing costs, warehouse costs, and filling costs (Zeng and
Rossetti 2003; Panayides 2006).
This chapter is organised as follows: the next section outlines the
literature review related to transaction cost economics in the maritime
logistics system. The later part discusses data and methodology. Section of
result and discussion describes the interrelationships between transaction
cost variables. Conclusions and policy implications are presented at the end
of this chapter.

TRANSACTION COST ECONOMICS IN MARITIME
LOGISTICS SYSTEM:
A LITERATURE REVIEW
Transaction costs were defined widely as the comparative costs of planning,
adapting, and monitoring task completion under alternative governance
structures (Rao, 2003). In another perspective, North (1990) described
transaction costs as the costs of measuring the valuable attributes of what
has been exchanged, the costs of protecting property rights and policies
and the costs of enforcing agreements. Meanwhile, according to Williamson
(1979), transaction cost economics is the core of studying an organisation to
efficiently govern its transaction structure with the market.
The transaction involves moving goods or services from one stage to
another through a different technology. The transaction costs economics are
closely related to contractual problems to enable costs to arise. In general,
transaction costs linked to contract issues are grouped into the discrimination
price before the contract (ex-ante) and the cost after the warranty (ex-post).
Ex-ante costs include the cost of drafting, negotiating, and securing the
agreement. While the ex-post expenses cover the cost of failure of adaptation
when the transaction deviates from the agreement, the cost of bargaining
in case of deviations after the contract, the fees for designing and carrying
out activities related to the governance structure, and the binding charges to
ensure the effectiveness of commitments (Williamson 1985). The difference in
transaction costs depends on how many sources come from the environment
inside and outside. Asset specificity and uncertainty are crucial variables in
transaction costs. Asset specificity refers to how an asset can be transferred
to another alternative user without sacrificing or reducing its productive
value (Williamson 1987). Transaction cost theory emphasizes the transaction
hazards of the asset specification
Williamson (1979) defines uncertainty as an event that is difficult to
predict. It depicts a particular environmental or individual state instead of
risk’s outcome nature (König 2009). The uncertainty is related to changes
that cannot be anticipated so that the resulting variables will bias each other.
Environmental uncertainty refers to unanticipated changes in circumstances
surrounding an exchange (Noordewier, John and Nevin 1990). Several
forms of uncertainty involve environmental, behaviour, customer supply,
competition, and strategy uncertainty (Cho 2014).
Elements of Maritime Logistics Transaction Costs in Islands 179

Transaction cost theory confirms that environmental uncertainty increases
due to various difficulties related to unusual environmental adaptation
and the obstacles encountered when monitoring contract implementation
performance with business partners or clients. Environmental uncertainty
results in adaptation and evaluation problems, increasing transaction
costs regarding both coordination costs and transaction risks (Grover and
Malhotra 2003). In other words, uncertainty can generate difficulties in
estimating future expenses. Cho (2014) noticed that the maritime industry’s
environmental uncertainty would increase transaction costs, such as
communication, negotiation, and coordination costs. It is in line with a
statement that environmental uncertainty and specific assets rise to different
transaction costs (Huo et al. 2018).
Maritime logistics refer to movement from the origin of goods until
their final destination through sea transportation, which generates long
transactions costs in the so-called maritime logistics system. Zeng and
Rossetti (2003) grouped six cost components from general logistics activities:
transportation, inventory, administration, custom costs, risks and damage,
handling, and packaging. These six components illustrate a series of logistics
activities resulting from transaction costs in the logistics system. For logistics
costs through sea transportation, aside from ships’ cost during shipping
at sea, several other prices from the origin and destination ports include
terminal handling charges (THC), warehouse, container shifting, port
workers, and freight costs from the port to the buyer’s warehouse (Wook
Song and Panayides 2015). Other logistical cost components include illegal
fees charged to the freight forwarders or the shipping liners (Musa 2014).
In the maritime industry, the levels of container port condition in each
region are varied, which makes it vulnerable to the risk of costs incurred to
the users of port services which are not cheap. Shipping service companies
can face high risks in areas with minimal port infrastructure, management
governance systems, and port services that do not comply with eligibility
standards. So the company must provide its own facilities or hire specialist
workers at the port of the region (Cho 2014).
In the maritime logistics industry, environmental uncertainty means
market uncertainty resulting in freight forwarding companies being
vulnerable to unexpected costs. Related parties tend to prefer to use container
services that are small from the risk of uncertainty. In the maritime trading
industry, environmental fate is an external challenge for businesses to

maximise profits due to the pressures of change that can reduce their ability
to survive.
Another factor that can increase maritime logistics transaction costs is
environmental uncertainty from the institutional aspect. North (1990) stated
that the primary role of institutions in a society is to decrease the uncertainty
by establishing a stable structure to human interaction. The institutional
environment regulates law and rules, professional organisations, and
positive institutions, namely behaviour, and habits. From an institutional
perspective, the institutional environment can be interpreted as uncertainty
surrounding an organisation or institution (Cho 2014). For the port industry,
institutional uncertainty involves the practice of enforcing rules misused by
unscrupulous apparatuses and port operators, thus causing costs to be borne
by users of port services and consumers. Port officers commonly charge
the third parties or service users, passengers, shipping companies illegal
fees. Unfair practices often occur at every stage of activities or transactions
while transporting and unloading goods at the port. The imposition of
administrative, shipping, and goods’ distribution costs often does not follow
established rules, such as a fee of USD0.04 per ton is imposed on expedition
company when carrying out unloading goods, and the ship document fee is
USD71 per trip (Amin et al. 2021).
DATA COLLECTION
Data collection was carried out through the focus group discussion (FGD)
method to assist information sharing and transfers between participants
(Asnawi et al. 2020). This method is used to identify and determine key
variables in measuring the causes of transaction costs in the maritime
logistics system. The study adopts the approach used by Cho (2014) that
divided the components of transaction costs into two categories: first,
expenses caused by the asset specificity, including port infrastructure,
availability of port services, and port accessibility, and second, costs that
arise due to environmental uncertainty include maritime uncertainty and
institutional uncertainty. Maritime uncertainty is a condition that causes
search costs to appear in trade by sea due to market uncertainty (Hallikas,
Virolainen and Tuominen 2002). Meanwhile, institutional uncertainty is a
condition of uncertainty regarding institutional regulations, institutional
professionalism, and the actors’ habits or behaviour that can cause transaction
costs in operationalisation and governance (Grewal and Dhawadkar 2002).

This chapter adds indicators of maritime uncertainty caused by climate,
natural disasters, and island economies of scale which often create cargo
uncertainty that can hinder the island’s economic development (Briguglio
1995). The variable logistics costs and traffic volume throughput are used as
parameters for the island’s maritime performance, reflecting the transaction
costs arising from uncertainty and specific asset factors in the port business.
The research also defines the transactions costs in the maritime logistics
system as expenses arising from a series of logistics activities by sea from the
port of origin to the destination port until it ends at the owner’s warehouse.
These transaction costs involve transports costs, loading and unloading
costs, port charges, and the fee of port workers.
Figure 1. Regional Ports of North Maluku Province
FGD participants were representatives of the maritime logistics business

stakeholders from each local port in North Maluku Province, namely port
management, loading and unloading companies, port workers, academician,
entrepreneurs, and local governments. The FGD involved 20 respondents
who were intentionally included to determine the maritime logistics
transaction costs variables. The objective area of the chapter are nine ports
located in North Maluku Province, namely Ternate Island, Tidore, West
Halmahera, Morotai Island, North Halmahera, East Halmahera, Central
Halmahera, South Halmahera, and Sula Islands (see Figure 1).

METHODOLOGY
The chapter used MICMAC (Impact Matrix Cross-Reference Multiplication
Applied to a Classification) prospective analysis method to determine asset
specificity and environmental uncertainty indicators, introduced by Arcade,
Godet, Meunier, and Roubelat (2003). The MICMAC was a structural analysis
based on a hierarchy of issues in various direct functional classifications,
indirect, or potential, which has many sources of information to determine
the principal value of a particular domain (Omran, Khorish and Saleh 2014).
The direct influence is referred to if variable A impacts variable B. The indirect
effect is taken into account if variable A controls variable B and variable B
affects variable C, so variable C is indirectly affected by variable A. While the
potential impact is the control of variable A which is greater than the other
variable B, and there is no direct effect of one variable on another variable or,
in other words, there is no relationship between variables (Luz et al. 2016).
In the first step, each participants defined the problems of the study
in this chapter and identified the internal and external variables that affect
the transactions costs in the maritime logistics system. Based on the FGD’s
elaboration, the determinants of the transaction cost component are asset
specificity consisting of port infrastructure, port services availability, and
accessibility. Uncertainty indicators are maritime uncertainty that comprises
cargo uncertainty, island economy scale, and weather; whereas institutional
uncertainty includes regulations, organisational governance, and employee
behaviour (Table 1). The next stage was analysing the relationship between
variables based on qualifications as follows: 0 = no influence, 1 = weak
influence, 2 = moderate influence, 3 = strong effect, and P = potential effect.
The types of variables were classified into four quadrants according to their
dependencies and influences on the other variables through the insight of
interrelation graphs. The last phase was entering the judgment results into
the MICMAC programme.

Table 1. Identification of Variables of Maritime Logistics Transaction Costs
Dimension Variables Short Label
Asset Specificity 1. Port Infrastructure INFRA
2. Port Services SERVICES
3. Port Accessibility ACCESS
Uncertainty Maritime Uncertainty
4. Cargo Uncertainty CARGOUNCER
5. Economic Scale of Island ECONSCALE
6. Weather WEATHER
7. Traffic volume throughput VOLTHROUG
8. Maritime Logistics Costs LOGMARCOST
Institutional Uncertainty
9. Regulation REGULATION
10. Organisation ORGANIZ
11. Behaviour BEHAVIOUR
The results of the analysis of impact and dependence are displayed in

the quadrant location, which is the variable consisting of primary variable,
autonomous, conflict or output variable according to the level of influence
and dependence (Luz et al. 2016). The advantages of using MICMAC analysis
were identifying variables, describing relationships between variables, and
determining key variables (Godet and Roubelat 1996). In the MICMAC
method, the captured interactions are used to expand and highlight a figure
of the integrated variables, describing them as the detailed interactions
between system elements (Veltmeyer and Sahin 2014).
RESULTS
The MICMAC methods grouped the variables into four quadrants according
to the categories of its dependence and influence (Figure 2). Quadrant I is
the influencing variables that display elements that are very dominant with
less dependence. This element is an essential variable in the system because
it can become a critical factor. Quadrant II contains relay variables that are
influential and highly dependent. This variable is often classified as elements
that depict the system instability. Quadrant III includes the dependent
variables or output variables, represented by high dependence and small
influence. This variable is relatively sensitive to the reversal in influence and
relay variables. Excluded variables in Quadrant IV are called autonomous
variables (Saxena, Sushil and Vrat 1990). According to the FGD results, the
maritime logistics transaction costs are determined by 11 main elements

based on two dimensions, i.e., asset specificity and uncertainty (Table 1).
Various activities in the maritime logistics system generate transaction costs
due to several factors. As shown in Figure 2, the weather, infrastructure,
load uncertainty, behaviour, island economy scale, and regulation indicators
are in the influencing variable (Quadrant II), which has the highest impact
on the system. These six variables are the leading indicator that can trigger
high maritime logistics transactions costs.
Figure 2. Direct Influence and Variable Dependence of Maritime Logistics
Transactions Costs
Meanwhile the port services variables—i.e., throughput volume, and

maritime logistics costs—are in the quadrant dependent variable, which
means that these variables are those affected by other variables. The
organisation and accessibility variables are independent variables in which
they have relatively little effect on the maritime logistics transactions costs in
the islands of North Maluku Province.
Figure 3 displays the infrastructure variables, load uncertainty, and
economies of scale that strongly correlate with other variables (the direction
of the arrow out). Port services, throughput volume, and maritime logistics
costs are the indicators most strongly influenced by other variables in the
maritime logistics system.

Figure 3. Direct Influence of Maritime Logistics Transactions Costs
Variables
The MICMAC prospective analysis method can also describe the

indirect effects between observed variables. As marked by red arrows
in Figure 4, the economies of scale have a powerful indirect effect on the
volume of throughput and maritime logistics costs. It confirms that the
influence of islands’ economies of scale on other variables indirectly has an
enormous impact on maritime logistics costs and throughput volumes. It
also indicates that a robust indirect effect is cargo uncertainty and weather,
while regulations show a relatively weak indirect effect.
Figure 4. Indirect Influence of Island’s Economies of Scale

The island’s economy scale is one of the elements in measuring the
island community’s economic productivity. The greater the port’s density
throughput volume activity, the more efficient the island’s economic scale
is. Likewise, if the island’s economies of scale are small, the port’s volume
throughput density level will also be low. The patterns of consumption
and community production level largely determine the scale of the island’s
economy. If the island’s level of consumption is higher than the production
level, the amount of goods unloaded is relatively more generous than the
number of goods loaded. The effect of island’s economies of scale on other
variables can indirectly impact maritime transport logistics costs and port
throughput volume.
DISCUSSION
The relationship between variables that affect maritime logistics transactions
costs shows port management’s complexity in North Maluku Province. Based
on the prospective MICMAC analysis results of the direct influence between
variables, there are six central elements that trigger high maritime logistics
transactions costs: weather, infrastructure, cargo uncertainty, behaviour,
economic scale of island, and regulation.
Weather is a variable that has a substantial burden on transaction
costs in the maritime logistics system. As the area consists of small islands,
environmental uncertainty due to extreme weather can hamper the
distribution of goods between islands. The loading and unloading goods
at the port with minimal loading infrastructure can be disrupted and run
slowly if the weather is terrible. The berth of the ship may usually take
longer time, which causes the anchor costs to increase. Weather uncertainties
also impact the supply side of goods, considering the vessel used is quite
vulnerable to being affected by sea waves to disrupt logistics activities. The
amount of goods available for sale to consumers on the island is limited,
impacting the selling price accordingly. For example, the commodities most
vulnerable to fluctuations due to the weather are fish and vegetables, as they
are perishable. If the weather is extreme, areas such as Ternate island, are the
most susceptible. Most of their food needs, such as vegetables, are transported
by public boat or ferry from Bitung, North Sulawesi, and then distributed
to the surrounding islands. If the sea waves are high, the transportation of
goods to be delivered to the destination island will be hampered . Delay in
delivering goods to end customers will increase logistics costs if the goods
are damaged, especially perishable goods.

Another push variable on the maritime logistics transactions costs is
port infrastructure. Generally, the facilities owned by ports on the isle adjust
to the volume capacity of the goods. A port that serves many throughputs
has better facilities than a port that serves a smaller cargo. For north Maluku
Province, Ternate, North Halmahera, and south Halmahera ports have better
facilities than other ports. However, an adequate level of facilities does not
guarantee whether the port’s logistics costs are low. The better the port
facilities serve the loading and unloading activities, the higher the logistics
costs. The cost of leasing loading and unloading equipment such as cranes,
trucks, and operator costs leads to more significant economic transaction costs
at these ports. For example, the total price of loading and unloading goods at
the port of Ternate is USD83 per container includes stevedoring fees, loading
and unloading financing services, and crane equipment services. Based on
the economic theory of transaction costs, container and crane facilities at the
port, including asset specificity associated with high prices in loading and
unloading service transaction activities.
At ports with minimal infrastructure facilities—such as in Jailolo port in
West Halmahera district and Weda in Central Halmahera district—maritime
logistics costs tend to be smaller, yet that does not necessarily mean low
transaction costs. Minimal port facilities can instead be a source of other
expenses outside the agreed transaction costs. Opportunism actions taken by
labour workers and port officials may incur additional transaction costs. For
example, labourers arbitrarily set expensive and unreasonable loading rates.
The goods owners often complain, particulary when they want to send goods
to the island district. Additional charges, such as searching costs, must be
prepared because the information on labour costs is unknown. Confidential
unloading tariff information is opportunism intentionally carried out by
unscrupulous port officers and port workers’ labour organisations. Therefore,
opportunism is an action that can hamper organisational efficiency due to
individual interests, which is not in line with corporate goals (Williamson
1979).
Opportunism also often occurs when the ship owner or forwarder
wishes to arrange a sailing or unloading permit where administrative costs
are higher than the official rate. For instance, a port officer requires an
administrative charge of USD24.2 per trip and a ship pass fee of USD69.17
per trip. In another case, the officials collected fees from unloading goods at
USD69.17 per ton. The goods fee is a joint agreement between the expedition
company and the port manager as additional service fees that are not

regulated in the regulations.
In addition to behavioural factors, cargo uncertainty has intense pressure
on the maritime logistics system’s inefficiency. Inequality in economic
development between the western and eastern regions in Indonesia has led
to more trade flows to industrial centers in the west of the area than to the
eastern regions. This fact impacts the logistics costs of shipping goods from
west to east, which become very expensive. The high cost of sea transportation
to the east part of Indonesia is due to ships returning with empty cargo,
so transportation costs for transporting goods are charged when leaving.
For example, the cost of sea transportation from the port of Tanjung Perak,
Surabaya, to the seaport of Ternate is USD928 per container, while the return
costs from Ternate to Surabaya are only USD500 per container (Amin et al.
2021). The uncertainty of the amount of cargo transported reflects market
uncertainties in the maritime business, which are difficult to estimate and
thus vulnerable to high-risk costs. Ship companies and cargo forwarders
have difficulty estimating freight costs and additional service costs while at
the destination port. This is due to the ship waiting times that can be longer.
Therefore, the higher the uncertainty of the amount of cargo, the higher the
transaction costs for maritime logistics.
Furthermore, the economies of scale on the island are other factors
that determine the maritime logistics costs and the throughput volumes.
Their natural characteristics severely limit the economies of scale in the
islands. It makes the islands vulnerable to economic uncertainties such as
small domestic markets, limited natural resources, and human resources,
uneconomic infrastructure, remoteness, far from market centers, irregular
transportation, difficulties in producing scale economics in the production
and service provision activities (Briguglio 1995; Kerr, 2005; Fernandes and
Pinho 2017). These various vulnerabilities can lead to uncertain conditions in
its economic activities, leading to high transaction costs such as coordination
costs, information costs, and negotiation costs. Ship agents or forwarders
will incur coordination and negotiation costs with local entrepreneurs or
traders to ensure the volume capacity of goods to be loaded.
The island’s economies of scale are relatively limited and small amounts
of goods can impact the weak ability of the island to control its market. As
a result, the community will obtain relatively small income. The island
community’s economic life has a high dependence on the nearby region.
For instance, most food needs in Ternate island must be supplied from the

cities of Manado in North Sulawesi and Surabaya in East Java. The volume
of goods unloaded is greater than the number of goods loaded, reflecting
that the level of island community’s consumption and demand for goods
are more significant than its ability to offer its products to other regions.
This fact is consistent with Malik (2016) that the level of imports of small
islands is relatively higher than exports. It is depicted by the volume of
goods unloaded that is more significant than the volume of goods loaded.
Simultaneously, ports located in the continental area explain that goods
loaded are relatively more generous than goods unloaded.
Table 2. Information of Cargo Throughput of Local Ports
Average of Cargo Throughput
(2010 – 2018)
Local Ports (Year) Container Ports Sea Toll-Way
Unload (Ton) Load (Ton)

Ternate 390,401.00 71,051.00 Yes No
Tidore 19,582.00 8,019.00 Yes Yes
South Halmahera 29,323.00 15,264.00 Yes No
West Halmahera 21,338.00 8,706.00 No No
Central Halmahera 46,478.00 228,975.00 No No
East Halmahera 10,449.00 4,390,944.00 Yes Yes
North Halmahera 108,346.00 147,266.00 Yes Yes
Morotai Island 50,479.00 6,120.00 Yes Yes
Sula Islands 46,579.00 15,324.00 No No
Source: Local Ports Offices 2019
Table 2 illustrates the ports located on small islands, namely Ternate,
Tidore, Morotai, Sula, Bacan in South Halmahera whose volume of unloading
goods is more excellent than the loading volume. Meanwhile, at ports in
Halmahera island’s land area, namely Central Halmahera, East Halmahera,
and North Halmahera, the number of goods loaded is greater than the
volume unloaded. The three districts in Halmahera island have a superiority
of natural resources in the mining sector. The volume of loading mining
products in nickel and gold via ships is greater than the loading volume in
other districts that do not rely on mining products.
Furthermore, regarding the port accessibility variable, the distance from
the port to the owner’s warehouse is generally close enough to smooth the
accessibility to the port. Most harbours and warehouses are located in regency
cities, so transporting goods by truck is relatively easy. Truck transportation
in a single trip is USD10.37 to USD20.75, with a payload capacity of eight

tons. For cargo with destinations outside the city, the transportation costs
are counted based on the distance. Additional logistics costs are a little bit
expensive if the goal is on another island. The additional costs are higher
because they have to rent a small boat. In the northern Maluku island province,
especially Halmahera island, improved land accessibility had resulted in the
port’s declining volume of cargo. This is because the goods—most of which
are only local commodities, such as copra, nutmeg, cloves—are directly sold
by the owner of the goods to Bitung North Sulawesi and transported by
truck via the ferry transportation mode. Likewise, when returning, they use
the same truck carrying goods in the form of food and other food needs,
which are then distributed to retailers and consumers in the market.
Available service facilities at the port are part of asset specificity, which
will result in high transaction costs if they do not meet acceptable service
standards. Generally, service facilities in the ports of the islands are still
insufficient and not fully utilised. It is related to the condition of inadequate
port infrastructure and the relatively low ship arrival density level. Only the
main ports have better service facilities, such as Ternate, South Halmahera,
and North Halmahera. Meanwhile, other ports’ availability of services adjusts
to the volume of goods and vessels served. For example, many ports in the
islands do not provide scout services. As a result of the level of ship visits,
it is not so dense that the ship is directly anchored at the dock. Warehouse
services in some ports are also not utilised well, such as in Morotai, central
Halmahera, and Sula islands.
Most goods unloaded from the ship are directly transported by truck
to the owner’s shop or warehouse. While in the ports of Ternate and north
Halmahera, warehouse services are ineffective because of unsatisfactory
warehouses capacities, so goods are left in containers in the stacking field.
Likewise, as happened in the port of Sula islands, due to the unavailability
of cranes or forklifts, the loading and unloading services require labour
costs accordingly. Hence, the labour costs are highly unreasonable and very
expensive, which can reach USD17.29 per ton higher than the current price of
USD6.22 per ton (Table 3). If the unloaded goods are related to government
project work, then it is certain that the cost of unloading can be more
expensive. In this case, throughput volume, infrastructure, accessibility, the
economy of scale, load uncertainty, regulation, organisation, and behaviour
enormously boost the port services variable. Therefore, based on economic
theory, transaction costs for services are part of the asset specificity that can
impact the port’s high transaction costs.

Table 3. Maritime Logistics Costs of Local Ports
Load and Unload Land Transport
Sea Freight Cost
Local Ports Cost Cost
From T.Perak Port (USD*)/ Trip (USD*)/Ton/M3
Ternate 7.49 20.75 58.79
Tidore 6.26 6.91 38.91
South Halmahera 6.15 4.15 62.69
West Halmahera 4.34 10.37 -
Central Halmahera 4.15 6.91 58.79
East Halmahera 7.99 10.37 47.55
North Halmahera 6.02 10.37 82.14
Morotai Island 7.88 10.37 38.91
Sula Islands 6.22 10.37 41.50
*Currency on 17 March 2021 (IDR14.456/USD1)
Source : Local Ports Offices 2019.
Meanwhile, regulatory indicators related to the rental service for loading
and unloading equipment at the port often trigger complaints between
the owners of the goods and the loading and unloading entrepreneurs
and labourers. The set fees are very detrimental to the owner of the goods
because they do not follow the cost standards set out in the regulations. For
example, in Bacan Island, South Halmahera Port, the rental fee for loading
and unloading equipment was USD24.21 per hour. According to the rules,
the rental price for a crane car mechanic tool for a capacity of 15 tons to 25 tons
is only USD4.70 per hour. The inconsistent implementation of regulations in
the port service business has resulted in the high cost of logistics for the
maritime transport sector in the islands of North Maluku Province. The
weak regulatory enforcement factor related to port service tariffs is part of
the maritime institutional uncertainty leading to high transaction costs.
Furthermore, maritime logistics costs are the dependent variable that
gets the most influence from other maritime logistics systems. This variable
can be the leading indicator in measuring the economic transaction costs of
a region. Maritime logistics costs arise from a series of logistics activities by
sea from the port of origin to the destination port to the owner’s warehouse
covering sea transportation costs, stevedoring fees, document fees,
administrative fees for customs clearance and technical control, charges for
terminal handling, and transportation costs on land.
Specifically for the island ports, the logistics costs for maritime transport
vary depending on the density throughput volume served. Ports that have

large throughput volumes and good unloading facilities can create potential
opportunities for high transaction costs. Under the economic theory of
transaction costs, asset specificity and uncertainty are factors that construct
transaction costs (Williamson 1979, 1987; Sawant 2012). Ternate harbour,
which has a loading dock and unloading facility, sets a stevedoring fee
of USD65.71 per container; whereas at the port of north Halmahera the
stevedoring price is USD59.69 per container with the use of a car crane. The
ports with minimal loading and unloading facilities still use human physical
support. The labour costs from ship to the truck until the goods owner’s
warehouse are not the same in each area. The inequality in facilities that
causes differences in applying tariffs at each island indicates that the island
ports’ management has not effectively decrease economic transaction costs.
MICMAC measurement results show that the organisational governance
of port services has a low influence on the maritime logistics transactions
costs.Yet, poor port organisation management can impact the emergence of
transaction costs in each line of port service activities. Problems such as the
low level of discipline of the harbormaster employees affect the monitoring
of loading and unloading activities. In several cases, the indiscipline of
the working hours of the dock workers and the inaccuracy of activity data
related to requests for labour from loading and unloading companies to port
labor associations required intense supervision from officers. For example,
the company of loading and unloading asked ten port workers, but only
seven were given, so it impacted the effectiveness of loading and unloading
time and goods’ costs to and from the ship.
Unsupported communication technology services in island areas also
result in coordination and negotiation costs. This communication service is
essential as a medium of information between goods owners, local small
industries, entrepreneurs, and local farmers to determine the amount of
cargo so that the ship will not return empty. The certainty of freight quantity
depends on the amount of the island’s superior commodity production
quantities. The leading commodities are fishery and plantation products,
such as tuna, skipjack, nutmeg, cloves, and copra. The island must develop
the raw material processing industry for its superior commodities into
finished products that have added economic value for sale to markets
outside the island territory. The ship’s cargo capacity can increase if the
production scale of the processing industry on the islands develops. With
the availability of information system service regarding the certainty of
the amount of cargo to be unloaded and loaded, the ship can estimate the

transportation costs and other additional service costs while at the port. For
goods with large quantities and weight, the freight cost per unit tends to be
cheaper. The availability of the goods information system can also develop
island industrial centers at the sub-district and regency levels close to local
ports.
Information openness and transparency related to the applicable
standard of port service rates, mostly the loading and unloading costs of
goods, are also significant for the public and port stakeholders. It is necessary
to prevent transaction costs in the form of illegal levies detrimental to the
owners of goods and consumers of port services. Organisational management
to serve goods delivery services also needs a better arrangement. The
conventional goods order system has encumbered the cargo companies
with searching and delivery costs, especially if the goods ordered or sent
are on another island. Information technology service systems are essential
to lessen transaction costs.
It would be significantly beneficial if the islands have processing
industry development centers to facilitate administrative service processes,
delivery of goods, ship activities while in port, goods documents, ship
documents, bill of loading, and information related to the certainty of the
number of goods. As such, the issue of uncertainty in the amount of cargo
can therefore be handled.
The long logistics chain between islands makes distribution costs quite
expensive (Figure 5). For example, instead of being directly shipped to
Tanjung Perak Port in Surabaya, the catch fish commodities from the Kayoa
Islands must be transported to Ternate Island or Bacan Island first and then
they are subsequently shipped to the port of Tanjung Perak in Surabaya or
the port of Bitung. The inter-island logistics distribution lines in Figure 5
show that Ternate port and three other supporting ports, namely Tobelo
port, Bacan port, and Sanana port, have a role as a regional logistics center
for the North Maluku province. The four ports may improve the industrial
centers to distribute logistics services and the region’s needs and surrounding
islands.
All of the island’s only commodities can be collected and processed in
the industrial area adjacent to the four ports and then exported outside. If the
production volume of superior island products can be increased on a large
scale and then processed into finished goods with economic added value,
sea transportation costs can be reduced because the cargo ship is no longer

empty when it returns to its base. After all, it is already filled with products
produced by the island industry.
Figure 5. Existing Condition of the Inter-Island Maritime Logistics
Network, North Maluku Province
Source: data analysis result, 2020.

The high cost of port logistics is a strong reason for traders when setting
selling prices in the market. The internal port party considers that transaction
costs are not the main factor in the high price of goods; traders and business
distributors of goods outside the port sometimes take large profits to
increase the selling price. Therefore, forming a regional logistics task force
is needed to anticipate the price games carried out by local entrepreneurs or
distributors on the main island for the island served by sea toll ship services.
This logistics task force is vital to investigate the price of goods in the market
or shops that use sea toll ship services.
CONCLUSION
The efficiency of maritime logistics costs is crucial for the islands’ economic
development in North Maluku, one of the island provinces in the eastern part
of Indonesia. The islanders of this area have been living with the available
minimal port infrastructure facilities, poor land connectivity, inefficient

governance in setting tariffs, and high transportation costs. Having analysed
the factors that affect maritime logistics costs in these islands area using
a transaction cost economic approach, it can be concluded that in the
maritime logistics system, transporting goods between islands requires high
transaction costs.
As mentioned previously, this chapter used MICMAC (Impact Matrix
Cross-Reference Multiplication Applied to a Classification) prospective
analysis method that determines asset specificity and environmental
uncertainty indicators. Data collection was carried out through the focus
group discussion (FGD) method to assist information sharing and transfers
between participants. The FGD method addresses to identify and determine
key variables in measuring the causes of transaction costs in the maritime
logistics system.
The limited knowledge of the FGD participants in the study has derailed
the attempt to further investigate the problems that cause high transaction
costs in the maritime logistics system. Nonetheles, it is still possible to develop
other variables that cause maritime logistics transaction costs, especially for
areas with large-scale ports. Taking into account these possible weaknesses,
this chapter discovered 11 elements considered crucial for maritime logistics
transactions costs in the Islands Province of North Maluku. Based on the
MICMAC analysis, some key variables that caused high maritime logistics
transactions costs are revealed—i.e., port infrastructure, cargo uncertainty,
bad weather, behaviour, island economies of scale, and inconsistent
regulation. The dependent variables are mainly the port services, throughput
volume, and maritime logistics costs. Improving port infrastructure and
developing island capacity-based economic activities can significantly
reduce maritime logistics transaction costs, which will positively boost the
economic growth of the islands in North Maluku Province, Eastern part of
Indonesia.
REFERENCES
Amin, C. et al. 2021. ‘Impact of maritime logistics on archipelagic economic
development.’ The Asian Journal of Shipping and Logistics, (xxxx). https://
doi: 10.1016/j.ajsl.2021.01.004.
Arcade, J. et al. 2003.’ Structural Analysis with The MICMAC Method &
Actors’ Strategy with MACTOR Method.’ The Millenium Project, 3.0,
0–48.

Asnawi, R. et al. 2020. ‘Analysis of Key Variables for Rice Farming
Sustainability in the Downstream of Sekampung Watershed : an
Application of Micmac Method.’ Plant Archives, 20(2), 7895–7904.
Briguglio, L. 1995. ‘Small Island Developing States and Their Economic
Vulnerabilities.’ World Development, 23(9), 1615–1632.
Cho, H. S. 2014. ‘Determinants and effects of logistics costs in container ports:
The transaction cost economics perspective.’ Asian Journal of Shipping
and Logistics, 30(2), 193–215. https://doi: 10.1016/j.ajsl.2014.09.004.
Fernandes, R. and Pinho, P. 2017. ‘The distinctive nature of spatial
development on small islands.’ Progress in Planning, 112, pp. 1–18. doi:
10.1016/j.progress.2015.08.001.
Godet, M. and Roubelat, F. 1996. ‘Creating the Future: The Use and
Misuse of Scenarios.’ Long Range Planning, 29(2),164–171. https://doi:
10.1016/0024-6301(96)00004-0.
Grewal, R. and Dhawadkar, R. 2002. ‘The Role of the Institutional Environment
in Marketing Channels.’ Journal of Marketing, 66(3), 82–97. https://doi:
10.1509/jmkg.66.3.82.18504.
Grover, V. and Malhotra, M. K. 2003. ‘Transaction cost framework in
operations and supply chain management research: Theory and
measurement.’ Journal of Operations Management, 21(4), 457–473. https://
doi: 10.1016/S0272-6963(03)00040-8.
Hallikas, J., Virolainen, V. M. and Tuominen, M. 2002. ‘Risk analysis and
assessment in network environments: A dyadic case study.’ International
Journal of Production Economics, 78(1, SI), 45–55. https://doi: 10.1016/
S0925-5273(01)00098-6.
Huo, B. et al. 2018. ‘Environmental uncertainty, specific assets, and
opportunism in 3PL relationships: A transaction cost economics
perspective.’ International Journal of Production Economics, 203, 154–163.
https://doi: 10.1016/j.ijpe.2018.01.031.
Kerr, S. A. 2005. ‘What is small island sustainable development about?.’
Ocean and Coastal Management, 48(7–8), 503–524. https://doi: 10.1016/j.
ocecoaman.2005.03.010.
König, F. 2009. The Uncertainty-Governance Choice Puzzle Revisited. (1st Ed.).
https://doi: 10.1007/978-3-8349-9449-3.

Lei, Q. and Bachmann, C. 2020. ‘Assessing the role of port efficiency as a
determinant of maritime transport costs: evidence from Canada.’
Maritime Economics and Logistics, 22(4), 562–584. https://doi: 10.1057/
s41278-019-00135-5.
Luz, M. et al. 2016. ‘Structural Analysis of Strategic Variables through
MICMAC Use : Case Study Structural Analysis of Strategic Variables
through MICMAC Use : Case Study.’ (July). https://doi: 10.5901/
mjss.2016.v7n4p.
Malik, A. 2016. ‘Reply to Lenzen , 2008 , JCLEPRO 16 , 2018 e 2035 : input
e output analysis for island economies.’ Journal of Cleaner Production,
112(2016), 4890–4895. https:// doi: 10.1016/j.jclepro.2015.06.034.
Mohamed-Chérif, F. and Ducruet, C. 2016. ‘Regional integration and maritime
connectivity across the Maghreb seaport system.’ Journal of Transport
Geography, 51, 280–293. https://doi: 10.1016/j.jtrangeo.2015.01.013.
Musa, S. 2014. ‘Corruption and firm behaviour: Evidence from African ports.’
Journal of International Economics, 94(2), 277–294. https://doi: 10.1016/j.
jinteco.2014.08.010.
Noordewier, T. G., John, G. and Nevin, J. R. 1990. ‘Performance
Outcomes of Purchasing Arrangements in Industrial Buyer-Vendor
Relationships.’ Journal of Marketing, 54(4), 80–93. https://doi:
10.1177/002224299005400407.
North, D. C. 1990. Institutions, Institutional Change and Economic Performance,
Individuals, Institutions, and Markets. New York: Cambridge University
Press. https://doi: 10.1017/cbo9781139175302.016.
Omran, A., Khorish, M. and Saleh, M. 2014. ‘Structural Analysis with
Knowledge-based MICMAC Approach.’ International Journal of Computer
Applications, 86(5), 39–43. https:// doi: 10.5120/14985-3290.
Panayides, P. M. 2006. ‘Maritime logistics and global supply chains: Towards
a research Agenda.’ Maritime Economics and Logistics, 8(1), 3–18. https://
doi: 10.1057/palgrave.mel.9100147.
Rao, P. . 2003. The Economics of Transaction Costs. New York: Palgrave
Macmillan.
Sawant, R. J. 2012. ‘Asset Specificity And Corporate Political Activity in
Regulated Industries.’ Academy of Management Review, 37(2), 194–210.

Saxena, J. P., Sushil and Vrat, P. 1990. ‘Impact of indirect relationships in
classification of variables—a micmac analysis for energy conservation.’
Systems Research, 7(4), 245–253. https://doi: 10.1002/sres.3850070404.
Stough, R. R. 2005. Institutional Barriers to Ports Infrastruktur and Harbor
Development. International Association of Traffic and Safety Sciences, 29(2),
30–40. https://doi: 10.1016/S0386-1112(14)60131-X.
Veltmeyer, J. and Sahin, O. 2014. ‘Modelling climate change adaptation
using cross-impact analysis : an approach for integrating qualitative
and quantitative data.’ In Proceedings - 7th International Congress on
Environmental Modelling and Software: Bold Visions for Environmental
Modeling, iEMSs 2014.
Williamson, O. E. 1979. ‘Transaction-Cost Economics : The Governance of
Contractual Relations.’ Journal o f Law and Economics, 22(2), 233–261.
Williamson, O. E. 1985. The Economic Institutions of Capitalism. New York:
Free Press.
Williamson, O. E. 1987. ‘Transaction cost economics; The comparative
contracting perspective.’ Journal of Economic Behaviour and Organisation,
8, 617–625.
Wook Song, D. and Panayides, P. M. 2015. Maritime Logistics; A Guide to
Contemporary Shipping and Port Management. New Delhi: Kogan Page
Limited.
World Bank. 2018. Logistics Performance Index. World Bank, Washington DC.
Zeng, A. Z. and Rossetti, C. 2003. ‘Developing a framework for evaluating
the logistics costs in global sourcing processes.’ International Journal of
Physical Distribution & Logistics Management, 33(9), 785–803. https://doi:
10.1108/09600030310503334.

Chapter 10
Institutional Access and Poverty of

Capture Fisheries Households in
Indonesia
Silviana Taniu, Dyah Wulan Sari, and Dias Satria
INTRODUCTION
The fisheries sector has long been developing and has become an important
sector in economic activity in many countries, where there are millions of
people in the world who depend on this sector for their livelihoods. Data
from the Food and Agriculture Organization of the United Nation (FAO
2018) states that fish or fishery products are the most traded type of food in
the world today.
The fishery sector has become one of the important economic sectors
globally. There are 35% of world fish production entered international trade
in various forms for human consumption or non-edible purposes. 60 million
tonnes (live weight equivalent) of total fish and fish products exported in
2016 represented a 245% increase since 1976. During the same period, world
trade in fish and fish products also grew significantly in terms of value, with
exports increasing from USD 8 billion in 1976 to USD 143 billion in 2016.
In the last 40 years, this figure has grown much faster than exports from
developed countries (FAO 2018).
Béné, Devereux, and Roelen (2015), define fisheries in terms of the
people involved, species or types of fish, territorial waters or seabed, fishing
methods, class of boat, the purpose of the activity, or a combination of these.
As an economic activity, fisheries are characterised by their operational scale
201
ranging from small to large scale. There are approximately 40 million people
involved in the fishing sector, and almost 90% operate small-scale fishing
activities, and most live in low- and middle-income countries. This includes
self-employed fishers (who own the boat and fishing gear), crew members
(who are contracted by the boat owner to operate the vessel), as well as
households involved in fishing).
Nonetheless, some other literature describes fishing communities as
‘the poorest of the poor, since their income depends on catches which vary
widely, is uneven and unpredictable (Bailey and Jentoft 1990; Béné and Friend
2011; Jentoft and Eide 2011). Yet, in many places, small-scale fishermen are
considered poorer but happier than those with other occupations, although
fishing is considered a job with a lower social status in society (Pitchford et
al. 2007; Fromentin et al. 2014; Anna et al. 2019).
The relationship between fisheries and poverty in developing countries
has often been debated by experts. Béné, Devereux, and Roelen (2015),
explained that the debate focused on the economic aspect, where the
lives of fishermen are low-income, and the biological aspect, namely the
overexploitation of the sea. The debate about poverty (income) of small-scale
fishermen has been going on for a long time, since FAO revealed that 20% of
those working in the global fisheries sector were estimated to earn less than
USD 1 per day, in 2005.
The relationship between fishing and income poverty is complex. In
many cases fishing communities may be prosperous, or even richer (in terms
of income) than other agricultural communities (Béné and Friend 2011;
Jentoft and Eide 2011). However, fishermen’s incomes depend on the size
of their catch which is usually highly variable, uneven, and unpredictable.
Their level of poverty is often more closely related to the severe degree of
geographic and/or political isolation that characterizes many full-time or
mixed fishing communities.
Fishery resources are mobile, and easily moving from one place to
another is a challenge faced by fishermen. Acheson (1981); Stanford et al.,
(2013) explained that poor fishermen depend entirely on the sea, so they
need modern fishing facilities and other supporting facilities to survive
long on the water. Islam et al., (2014) explain that fishermen who use boats
with larger engine capacities earn large incomes. This reflects an unequal
distribution of benefits, as only those who can invest in fishing facilities and
who spend more on fishing operations can derive maximum benefits.

Béné and Friend (2011) explain that the cause of poverty in small-scale
fisheries is not only the low productivity of fishermen but also the economic,
political, and institutional marginalisation of fishing communities in general.
As a result, the majority of fishermen do not have any access to institutions
(e.g., efficient credit markets or decent labour markets) or if they have, the
access to these institutions is transactionally or economically too costly for
the poor. Without this access, fishers remain unable to achieve the minimum
level of investment that would allow them to generate greater financial
returns and pull themselves out of the low-income productivity and poverty
levels in which they are trapped.
Formal institutions are essential for fishermen to facilitate various needs
of life and support the sustainability of fishing businesses. Satria and Li,
(2017); Haque et al., (2015), explained that fishermen face the problem of
uncertainty related to climate change, so they usually seek loans for several
different purposes, such as filling gaps in their consumption needs, resulting
from the impact of significant income variations due to weather uncertainty,
or for investment purposes such as catching more fish. In this case, formal
and informal credit institutions are important for fishermen. According to
Parapphurathu et al. (2019), Kolade and Harpham (2014), Amarasinghe and
Bavinck (2017), Twumasi et al. (2021), Khan, Alam, and Islam (2012), Islam,
Yew, and Viswanthan (2014), the participation of fishermen in institutions
such as cooperatives or fishing communities can help increase fishermen’s
income.
In addition to the several studies on fishermen’s livelihoods previously
described, Béné et al., (2016) summarized 202 articles published between
2003-2014 and carried out an appropriate evaluation protocol to evaluate
the existing evidence on how and to what extent capture fisheries and
aquaculture contribute to improved nutrition, food security, and economic
growth and poverty in developing countries. The conclusion is that although
studies on fisheries make an undeniable contribution to nutrition and
food security, studies on the relationship between fisheries/aquaculture
and poverty alleviation are complex and unclear, especially studies at the
national and household levels.
INDONESIAN MARINE AND FISHERIES SECTOR

Indonesia is the largest archipelagic country in the world which is
geographically located between two continents, Asia and Australia, and two
Institutional Access and Poverty of Capture Fisheries 203

Households in Indonesia
oceans, the Indian and the Pacific. A dynamic area in the world arena, both
economically and politically. In addition to this strategic position, Indonesia’s
sea area, starting from the territorial sea, the contiguous zone, the EEZ, to the
continental shelf holds very abundant wealth. The total area of Indonesian
waters is 6.32 million km2 or 2/3 of the total area of Indonesia, with a coastline
of 90,093 km and the number of islands 17,504, which shows the enormous
potential of fisheries resources, especially capture fisheries in Indonesia.
In 2014 the growth of the fisheries and marine sector was at 7.55% while
the national GDP growth was at 5.02% (Pusat Data Statistik dan Informasi,
Ministry of Marine Affairs and Fisheries of The Republic of Indonesia 2014).
In addition, Indonesia is an important player in global fisheries production.
With 6.5 million tonnes per year (7% of global production), it is the second-
largest capture fisheries producer after China.
To improve the welfare of fishermen and optimise capture fisheries
resources, the Indonesian government through the Ministry of Marine
Affairs and Fisheries has created a programmeme to increase the institutional
capacity of coastal communities in the form of ship grants, fishing gear grants,
business financing grants. These grants are channeled through cooperatives
and fishing groups (Joint Business Groups). This programme was started
in 2010 by distributing 1000 30 GT vessels throughout Indonesia. This
programme continues every year in which varied sizes and types of boats
and several types of fishing gear have been distributed to fishermen. The
vessel includes engines, equipment, and licensing documents. However, the
effectiveness of this government programmeme is still questionable, due to
delays between implementation and results. For example, at the beginning
of this programme, many boats did not meet specifications when given to
fishermen. The size and type of vessel and fishing gear must be adapted to
the oceanographic conditions and the type of fish being the target of catching.
A recent study measuring the welfare of fishermen in Indonesia (Anna et
al. 2019) found that fishermen are more optimistic in life and have a positive
attitude towards changes in economic status in the future. This is interesting
considering that most Indonesian fishermen are small-scale fishermen with
low productivity. According to data from the Central Statistics Agency in
2013, the number of poor people in Indonesia reached 28.07 million, of which
25.14% are estimated to live in coastal areas.
Many poverty studies have been conducted that analyse individual,
household, and aggregate poverty, using poverty measurement with a

monetary and multidimensional approach. However, this chapter tries to
fill this gap by measuring the poverty of captured fisheries households in
Indonesia, with a monetary and multidimensional approach. In addition,
using survey data on household income in agriculture/capture fisheries
by the Central Bureau of Statistics (BPS) 2013 and dividing the sample for
poverty analysis in households in Java and outside Java, this chapter aims to
analyse the relationship between institutional access and poverty in capture
fisheries households in Indonesia.
MULTIDIMENSIONAL POVERTY
Measurement of poverty has evolved since its discovery. In general, the
method of calculating the poverty rate uses a monetary approach in which it
is estimated by the value of money (income, expenditure, or consumption).
The advantage of this method is that the measurement is clear, quantitative
in nature so that it can measure the depth of poverty. However, according to
Hulme, Moore, and Shepherd (2001), such an approach will play a significant
role in future research but must be complemented by a multidimensional
approach to poverty that is qualitative in nature and allows subjective
assessment by the poor.
From the outset economists have recognised some basic limitations of
the income method, namely patterns of consumer behaviour may not be
uniform. Therefore, reaching a poverty line level of income does not guarantee
that a person will meet his or her minimum needs. Price differences that can
reduce the accuracy of the poverty line, and the ability to convert a certain
amount of income into certain functions vary according to age, gender,
health, location, climate, quality of affordable services, such as water, health
and education which are often not provided through the market (Akire and
Santos 2014).
According to Anshory Yusuf (2018), using income or expenditure as a
basis for measuring welfare is indeed quite common. However, measuring
welfare using the income basis is still incomplete in capturing the various
dimensions of welfare itself. If welfare is calculated based on income alone,
it has not fully captured the non-income dimensions of real welfare. People
who may not be categorised as poor because their income exceeds the
poverty line may have exceptionally low education, do not have access to
proper water and sanitation, live in houses of limited quality, and do not
have permanent jobs. This question is to be answered with a concept called

multidimensional poverty.
To measure poverty with a multidimensional approach needs to use
factors that are not related to money and are more varied. As a case in point,
Amartya Sen has developed a capability approach to measure poverty
(Budiantoro et al. 2013). In Sen’s view, development is the expansion of
human capabilities, not the maximisation of human needs represented by
income. The dimensions of education, health, quality of life, democracy, and
people’s freedom of economic access are especially important to be used as
benchmarks to measure an individual’s poverty.
On the other hand, the United Nations Development Programme
(UNDP) and the Oxford Poverty and Human Initiative (OPHI) since 2010
have developed the Multidimensional Poverty Index (MPI) that aims to
map an individual’s poverty using clearer indicators, namely education,
health, and living standards (Alkire et al. 2011). The indicators in each of
these dimensions are child mortality and nutrition (in the health dimension),
the length of schooling, and school participation indicators (in the education
dimension), and fuel for cooking, toilets, water, electricity, and assets ( in
the standard of living). The indicators in each of these dimensions allow
for further development and may differ between countries but remain
within the intended three-dimensional corridor. Furthermore, according
to Alkire et al. (2011), the Multidimensional Poverty Index (MPI) can
also measure households experiencing deprivation in one of the existing
indicators/dimensions. However, this does not necessarily mean the factor
that makes the household multidimensionally poor. Based on the MPI
standards, households could be considered multidimensionally poor if they
experience deprivation in one third (1/3) of the weighed indicators at the
same time.
Empirical studies that measure poverty using both monetary and
multidimensional approaches have been conducted on various objects and
sectors, such as that conducted by Kim (2019) that analysed the transition
in monetary and multidimensional child poverty in Ethiopia, India, Peru,
and Vietnam. Meanwhile, Banks et al, (2021) compared monetary and
multidimensional poverty rates among people with and without disabilities
in Cam Le district, Vietnam, and Tanahun, Nepal.
Using the Demographic and Health Survey (DHS) and Alkire and Foster’s
multidimensional poverty measure with four dimensions: assets, health,
schools, and empowerment, Batana (2013) estimated multidimensional

poverty among women in fourteen Sub-Saharan countries, (Benin, Burkina
Faso, Guinea, Madagascar, Mali, Niger, Senegal, Ghana, Kenya, Malawi,
Nigeria, Tanzania, Uganda, and Cameroon).
In the agricultural sector, Oqutu and Qaim (2019) used data from
a survey of farm households in western Kenya to evaluate the effect of
commercialization on income poverty and multidimensional poverty. Yu
(2013) estimated multidimensional poverty in China by applying the Alkire-
Foster methodology to China’s health and nutrition survey data. The five
dimensions include: income, the standard of living, education, health,
and social security, and compare multidimensional poverty with income
poverty, HDI and GDP. From the fisheries sector, Solaymani and Kari (2014)
use survey data of fishermen living in rural areas in 13 Malaysian states and
apply the Alkire and Foster multidimensional poverty index to measure
multidimensional poverty. The dimensions measured are income, education
level, health, insurance, environment, access to clean water, electricity, and
road access.
Several studies measuring poverty were carried out in Indonesia,
including those conducted by Sumarto and De Silva (2014), which analysed
the degree and dynamics of multidimensional household welfare in Indonesia
for the periods of 2004 and 2013. Hanandita and Tampubolon (2016) used
data from the National Socio-Economic Survey (Susenas), to analyse the level
and pattern of multidimensional poverty in Indonesia from 2003 to 2013.
Anshory Yusuf and Sumner (2017) carried out poverty measurements not
only focusing on the dimensions of education, health and, assets/household
living standards but also on the employment dimension (informality,
unemployment, and underemployment) to obtain a multidimensional
measure of poverty that is compatible with inclusive growth.
This chapter adopts the multidimensional approach refers to the Alkire-
Foster multidimensional poverty index method which is measured from the
dimensions of education, health, and living standards (Alkire and Foster
2011a), (Alkire and Foster 2011b), (Alkire et al. 2011). Using an employment
dimension with an indicator of informality (job status that generates the
main income as fishing labor), following Anshory Yusuf and Sumner
(2017) who used the employment dimension (informality, unemployment,
and underemployment) to obtain multidimensional poverty, and also
following the study by Sumarto and Silva (2014) that included informality
and unemployment as part of the dimensions of household assets and living

standards, this chapter then analyses the magnitude of the overlap between
consumption and multidimensional poverty.
Dimensions, weights, and indicators in measuring multidimensional
poverty are as presented in Table 1.
Table 1. Dimensions, Weights, and Multidimensional Poverty Indicators
Capture Fisheries Households in Indonesia
Dimension Weight Indicator
1/8 - Have no family members who have completed 9 years

of education
Education (1/4)
1/8 - Have at least one school-age child (up to grade 9) who
dropped out of school
1/8 - Having at least one malnourished family member
Health (1/4)
- Having one or more children who died
1/20 - No electricity
Standard of living 1/20 - No access to safe drinking water

1/20 - No access to improved sanitation
and Household
Assets 1/20 - Own a house with a dirt floor
(1/4) 1/20 - Do not have a motorized vehicle and only have one
of the following items: Bicycle, Motorcycle, Radio,
Refrigerator, Telephone, or Television
1/8 - The job status that generates the main income is as a
Employment (1/4) labor fisherman
1/8 - Using traditional boats to fish
Source: Alkire et al. (2011), Modified.
The stages of measuring multidimensional poverty based on the Alkire-
Foster method (Alkire and Foster 2011a) are as follows:
–– Choose the unit of analysis (i.e., the household)
–– Choose the dimension of poverty
–– Choose an indicator from each indicator
–– Determine the cut off deprivation of each indicator
–– Determine the weight (wi) of each dimension/indicator
–– Calculating the deprivation score (∑) experienced by the household:
∑ ci = ωiI1 + ωiI2 + ωiI3 +… + ωdId
Where I1 = 1 if the household lacks the indicator i and = 0 if the household
is not deprived at the indicator of i

To determine whether households are poor or non-poor should use
total deprivation score. A household is multidimensionally poor if the total
deprivation score (∑ci) is more than or equal to 1/3. However, if the score is
less than 1/3, then the household is defined as not poor, and its deprivation
score is changed by zero (deprivation sensor). If a household is deprived in
all indicators, then ∑ ci = 1, if the household is not deprived in all indicators,
then ∑ ci = 0.
METHODOLOGY
The Logit model is used to analyse the relationship between institutional
access and poverty of capture fisheries households in Indonesia. Poverty
status (dependent variable) is measured in two monetary approaches
(income) and a multidimensional approach (multidimensional poverty
index). The association model for institutional access and household poverty
of capture fisheries is defined as follows:
ln is the logit or natural logarithm (ln) dari odd ratio , where Pi
describes the probability of poor households and 1 - Pi indicates the

probability of a non-poor household.
β1 is an important coefficient, if it has a negative sign, institutional access
(access to credit, utisation of cooperatives, membership of fishermen’s
groups (KUB), access to boat assistance, fishing gear and capital from the
government, as well as access to marketing) can reduce the likelihood of
households becoming poor. Institution designates a binary variable and has
a value of one if the household has access to institutions in carrying out its
capture fisheries business, otherwise it will be zero.
Xj (household characteristic factors including gender, age, education
of the head of the household, and household size) were determined as all
determinants that impact household poverty status. β0 and βj are the intercept
and the important determinant coefficient that affect the poverty status of
the household, whereas v is short-term error.

Table 2. List of Variables Used in the Model
Variable Description
Poverty Household Poverty Value 1 if the household is poor, and zero if non-
Status (monetary and poor
Multidimensional)
Households Value 1 if gender of the household head is Male,
Gender
Characteristic and zero if not
Age Age of the household head
Value 1 if the household head did not finish
Education
primary school, and zero if finished primary school
Household size Number of household members
Institutions Value 1 if household receives credit access, and
Credit Access
Access zero if not
Utilization of Value 1 if household receives utilisation of
Cooperative Facilities cooperative facilities, and zero if not
Member of Value 1 if household is a member of Kelompok
Kelompok Usaha Usaha Bersama (KUB), and zero if not
Bersama/KUB
Access to production Value 1 if household has access to production
facilities facilities, and zero if not
Value 1 if household has access to market, and zero
Access to market
if not
Value 1 if household receives boat grant, and zero
Receive boat grant
if not
Receive fishing gear Value 1 if household receives fishing gear grant,
grant and zero if not
Receive financing Value 1 if household receives financing grant, and
grant zero if not
DATA
The dataset utilised for this chapter is the household sample data for capture
fisheries in the 2013 survey of household incomes for agriculture/capture
fisheries in Indonesia by the Central Bureau of Statistics (BPS). This survey
is part of the agricultural census, which is conducted every ten years. The
number of households selected as the sample in this survey is 28,779. The
poverty analysis focused on capture fisheries households nationally and
divided the sample for poverty analysis on households in Java and outside
Java. This helps to demonstrate the consistency and robustness of the
estimation results, as well as to test whether there are significant differences
in the characteristics of household poverty in Java, Outside Java, and
Nationally.

Capture Fisheries Household Poverty in Indonesia (Monetary and
Multidimensional)
Capture fisheries households in Indonesia are poorer in monetary terms
(having income below the poverty line) than multidimensionally. As shown
in Figure 1, based on monetary approach, 19.91% of the sample capture
fisheries households are poor and 80.09% of them are non-poor. For the
measurement of poverty based on a multidimensional approach, 17.26% are
poor households and 82.74% are non-poor households.
Figure 1. Capture Fisheries Household Poverty in Indonesia, 2013
Moneter Multidimensional
19,91 17,26
80,09 82,74
Not Poor Poor Not Poor Poor
Source : Processed from household income survey of agriculture, Central Bureau of

Statistics Republic of Indonesia (BPS), 2013.
The fishing business in the sea is strongly influenced by the seasons.
Usually, in November – January every year the West season occurs with
high waves, which makes fishermen unable to operate. The ability to carry
out fishing operations, especially during the West season, must be supported
by a stable condition of the ship/boat to be able to deal with fierce waves.
However, most of the boats/boats owned by fishing households are not
designed to be able to withstand big waves. The ability of the ship/boat
and the distance from the fishing operation area to the base of the ship/boat
affect the length of days of fishing operations in one trip.
The results of an agricultural survey by the Central Bureau of Statistics
Republic of Indonesia (BPS) 2013 show that 60-93% of fishing households
at sea carry out fishing operations for only 1 day (one-day fishing). This
indicates that the fishing operation area is close and that their ship/boat
capabilities are limited. The proximity of the fishing operation area will

affect the catch. This is because these places are saturated or experiencing
overfishing conditions due to a large number of ships/boats operating in the
same area.
In addition, capture fishing activities are still carried out using
traditional means. This can be seen from the lack of fishing tools that aim to
facilitate their fishing operations. Only a small percentage of marine capture
fisheries households in all ship/boat classifications use echo sounders/fish
finder/GPS and FADs in their fishing operations. This situation confirms
the finding by Stanford et al. (2013) that poor fishermen—including those
who use simple traditional boats/fishing gear and crew/labourers—depend
entirely on the sea for their income although some have other livelihoods,
besides
For labour fishermen, income is determined based on a profit-sharing,
and therefore they rarely receive fixed wage or salary. In this revenue-sharing
system, what is shared is the income after being deducted for operational
costs plus the cost of selling the proceeds, including the cost of fuel, oil, ice
cubes, salt, food costs for crew members, and levies or permits. The cost
of repairing the ship and fishing gear is usually borne by the shipowner.
The results or income that is divided is the result of the sale of caught fish
that have been sold by the shipowner and then the profit-sharing calculation
is carried out, whereas the calculation time for the profit-sharing is usually
conducted according to the agreement, it can be either monthly or quarterly.
Institutional Access and Household Poverty of Capture Fisheries in

Indonesia
Based on the methodology as discussed previously, a regression model was
used to analyse the relationship between institutional access and poverty
in capture fisheries households in Indonesia. Summary statistics of all the
variables used in the model are presented in Table 2.

Table 2. Statistic Summary of Regression Variables
Java Outside Java National
Variables
Mean SD Mean SD Mean SD
Poverty_Moneter 0.14 0.35 0.20 0.40 0.20 0.40
Poverty_Multidimensi 0.08 0.28 0.18 0.39 0.17 0.38
Household Characteristic
Gender of household head 0.96 0.17 0.95 0.20 0.96 0.20
Age of household head 46.75 11.26 44.78 11.91 45.10 11.89
Education of household head 0.39 0.48 0.34 0.47 0.35 0.48
Household size 4.15 1.53 4.62 1.85 4.55 1.81
Institutional Access
Credit access 0.07 0.25 0.04 0.21 0.05 0.22
Cooperative benefits 0.03 0.17 0.02 0.16 0.03 0.18
Kelompok Usaha Bersama/KUB 0.27 0.44 0.25 0.43 0.26 0.44
Production facilities 0.82 0.38 0.69 0.46 0.71 0.45
Market 0.92 0.26 0.80 0.39 0.83 0.38
Boat grant 0.01 0.10 0.02 0.14 0.02 0.14
Fishing gear grant 0.03 0.17 0.02 0.14 0.02 0.15
Financing grant 0.03 0.18 0.03 0.18 0.03 0.18
Number of Observation 4728 24051 28779
Notes : SD stands for standard deviation; Min stands for minimum value; Max stands for
maximum value.
Source : Processed from survey of agriculture household’s income, BPS
In terms of demographic characteristics, capture fisheries households
are typically headed by men, while the number of households headed by
women is classified as exceedingly small and financially poorer. The average
age of the household head is 45 years, and the education level of the household
head is still extremely low, where although most of them could complete
elementary school education, there are still many of them who could not.
The average number of capture fisheries household member is more than
four , while the highest number of family member in one household is 20.
Regarding their fishing business, only 5% of the sample fishery
households get access to credit from banks. Very few households have
utilised the cooperative’s facilities (3%), but 25% of these households have
become members of a joint business group (Kelompok Usaha Bersama/KUB).
In addition, 71% of households have easy access to production facilities in
the form of boats/boat engines and fishing aids, and 82% of them have easy
access to markets.

In general, very few capture fisheries households have access to government
and non-government grant programmes, comprising those who receive
ship/boat and fishing gear grants (2%) and financing grants (3%). There
is a significant difference between capture fisheries households in Java and
outside Java in which fewer capture fishery outside Java have access to
credit, access to production facilities, and ease in marketing their products.
RESULTS
As explained earlier, this chapter focused on capture fisheries households
nationally, and the samples for poverty analysis on households were divided
into those living in Java and outside Java. The logistic regression estimation
results from the institutional access model and household poverty of capture
fisheries are presented in Table 3.

Table 3. Estimation Regression Results
Java Outside Java National
Variables Multidimen- Multidimen- Multidimen-

Monetary Monetary Monetary
sional sional sional
Coefficients Coefficients Coefficients
Coefficients Coefficients Coefficients
Gender of -0.2989 -0.1716 -0.1806 *** 0.0997 -0.2094 *** 0.0185

household head 0.2325 0.2720 0.0829 0.0826 0.0781 0.0788
Age of -0.0045 -0.0164 *** -0.0132 *** -0.0159 *** -0.0126 *** -0.0175 ***
household head 0.0039 0.0048 0.0015 0.0014 0.0014 0.0014
Education of 0.1190 0.8731 *** 0.2181 *** 0.7891 *** 0.2024 *** 0.7809 ***
household head 0.8920 0.1102 0.0352 0.0349 0.0327 0.0331
0.2633 *** 0.0479 0.3185 *** -0.0413 *** 0.3170 *** -0.0204 ***
Household size
0.0257 0.0324 0.0088 0.0093 0.0083 0.1111
Receive credit -0.1737 -0.1861 -0.2337 *** -0.0287 -0.2325 *** -0.0797
access 0.1757 0.2186 0.0794 0.0785 0.0722 0.0733
Receive -0.3895 ** -0.7448 *** -0.1178 -0.3269 *** -0.2051 *** -0.5003 ***
cooperative
benefits 0.2244 0.3362 0.1055 0.1167 0.0946 0.1089
-0.2669 *** -0.5760 *** -0.2257 *** -0.1540 *** -0.2351 *** -0.1856 ***
Member of KUB
0.1048 0.1438 0.0397 0.4040 0.0371 0.0387
Access to -0.2112 *** 0.0756 -0.2558 *** -0.1785 *** -0.2609 *** -0.2012 ***
production
facilities 0.1097 0.1428 0.0366 0.0373 0.0346 0.0358
Access to -0.8656 0.0289 -0.1045 *** -0.3641 *** -0.1095 *** -0.4007 ***
market 0.1687 0.2131 0.0425 0.0421 0.0410 0.0409
Receive boat -0.4637 -0.8598 0.1396 0.1325 0.1199 0.1437
grant 0.4765 0.7287 0.1120 0.1135 0.1083 0.1111
Receive fishing -1.1891 *** 0.4072 -0.2908 *** 0.0878 -0.4101 *** 0.0783
gear grant 0.0395 0.3091 0.1214 0.1102 0.1146 0.1029
Receive -0.0940 0.1947 -0.1019 -0.0307 -0.1045 -0.0064
financing grant 0.2320 0.2756 0.0948 0.0943 0.0876 0.0888
-2.2371 -0.1947 -1.8746 0.5149 -1.8715 0.5072
Constant
0.3557 0.4334 0.1202 0.1200 0.1131 0.1147
Number of
4728 4728 24051 24051 28779 28779
Observation
LR chi2 (12) 143.26 105.72 1578.09 749.16 1784.91 867.25
Prob > chi2 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
Pseudo R2 0.0364 0.0374 0.0639 0.0321 0.0621 0.0328
Source : Processed from survey pendapatan rumah tangga pertanian (Badan Pusat
Statistik) with STATA.
Note : Numbers at the second line are standard errors.

The sign of the coefficient in the three models is the same for the variables
of gender, age, education of the head of the household and household size.
Male household heads are less likely to be poor (monetary) than women
in the Outer Java and National models. The variables of age and education
of the head of household have a significant on poverty status (monetary
and multidimensional) in the Outer and National Java models, but only
significant on multidimensional poverty status in the Javanese model.
The household size variable is significant on poverty status (monetary
and multidimensional) in the Outer Java and National models, but only
significant on monetary poverty status in the Java model.
Credit access is significant for monetary poverty status in the Outer Java
and National models. Households that have access to credit from banks/
non-banks have a lower probability of being poor than households that
have never received credit. This finding is consistent with those by Dartanto
and Nurkholis (2013), Haque et al. (2015), and Satria and Li (2017). Formal
and informal credit institutions are especially important for small-scale
fishermen, where these credits are needed by fishermen for fishing business
operations, including repairing boats and fishing gear. In addition, credits
are required to fill the gap in their consumption needs, resulting from the
impact of significant shortage of income due to weather uncertainty.
The function of cooperatives, apart from being able to provide credit
to their members, is to accommodate fish caught by fishermen and also
to provide cheap production facilities for fishermen so that they are not
entangled in moneylenders. The existence of these moneylenders in the
fishing community offers credit facilities in the form of boats, fishing gear,
and capital. However, the process that follows in the relationship between
moneylenders and fishermen tends to be detrimental to fishermen. The
variable of cooperative utilisation is significant to multidimensional poverty
status in all models and significant to monetary poverty status in the
National and Java models. This explains that capture fishery households that
use cooperative facilities have a lower probability of being poor than those
that do not. This finding confirms the previous studies by Parapphurathu et
al. (2019), Kolade and Harpham (2014), Amarasinghe and Bavinck (2017),
Twumasi et al. (2021), Khan, Alam, and Islam (2012), Islam, Yew, and
Viswanthan (2014).
In addition to cooperatives, within the fishing community itself, small
groups of fishermen who engaged in fishing business activities have been

formed based on the results of an agreement or deliberation of all members
based on the interests of mutual needs and desires to be carried out and
accounted for together. This fishing group is known as a joint venture group
or KUB. The benefits of KUB are as a place to do business together, raise
capital together, as an institution that will partner with fishing companies,
and especially with the existence of KUB, it can facilitate guidance from the
government. The fisherman group membership variable (KUB) is significant
in monetary poverty status and is multidimensional in all models, both
national, outside Java, and Java. This explains that capture fishery households
that are members of the fishermen’s group (KUB) have a lower probability of
being poor than those who are not.
Access to production facilities is significant for monetary poverty status
in all models, both national, outside Java, and Java, but only significant in
the multidimensional poverty status in the national and outside Java models.
This explains that fisherman households that do not experience difficulties in
accessing production facilities such as boats, fishing gear, and other fishing
facilities have a lower probability of being poor than those who experience
difficulties. This can be understood because the condition of fishery resources
is mobile, or it is easy to move from one place to another, so to catch fish,
fishermen need adequate facilities.
According to Stanford et al. (2013), despite poverty alleviation efforts,
the number of poor fishermen continues to grow, and there are two main
types of poor fishermen, namely fishermen who use simple traditional boats/
fishing gear and crew/laborers. Some poor fishermen depend entirely on the
sea for their income and others have diverse livelihoods.
Market access is significant to the monetary and dimensional poverty
status in the national and outside Java models, but not significant to the
poverty status in the Java model. It can be said that fishing households that
do not experience difficulties in marketing their caught fish have a lower
probability of being poor than those who experience difficulties. The market
or fish auction place is especially important for fishermen. The catches of
fishermen obtained will only have more value if they are not only used for
their consumption, but also to meet various needs of daily life.
In the view of Satria and Li (2017), fishermen have a higher level of
transaction risk compared to other agricultural sectors. This is the main
reason why some investors are reluctant to engage directly with fisheries
markets in developing countries, and as a result, has hampered the economic

development of coastal areas. Transactions in the fisheries sector can
work in certain circumstances through the introduction of intermediaries
that facilitate market exchange. The wide range of services provided by
intermediaries has been identified as an important bridge in the fishery value
chain. This is because fishermen often do not have market access, especially
those who live on small islands (Imron 2003).
Meanwhile, the perishable nature of fish is a big problem faced
by fishermen. In such conditions, the role of moneylenders becomes
exceptionally large in the lives of fishermen. In small islands, the existence of
moneylenders does not only function as buyers of marine products, but also
sells various needs of fishermen, both for fishing gear and daily necessities,
such as sugar, coffee, tea, and cigarettes. Dependence on moneylenders
makes fishermen’s position weak. The weak position of fishermen is often
exploited by moneylenders to buy fish at cheap prices, and they will sell them
at soaring prices. On the other hand, moneylenders sell daily necessities to
fishermen at unaffordable prices.
Furthermore, in their fishing effort, fishermen are faced with the
problems of lacking facilities and highly depending on the season. Therefore,
fishermen cannot go to sea anytime, especially due to the wave season, which
can last for more than one month. As a result, in addition to the limited
catch, with the simplicity of the fishing gear owned, in certain seasons no
catch can be obtained. To overcome this, fishermen will try to maximize their
income by increasing the capacity of the means to catch fish. However, this
is often not possible due to limited capital. Only fishermen who have large
capital will be able to buy more varied fishing gear with a higher level of
exploitation than those with small capital. The absence of fishing gear puts
them in a position only as fisherman labourers.
Of the three grant programmes for fishermen, namely boat grants,
fishing gear grants, and government/non-government business financing
grants, only fishing gear grants have a significant impact on monetary
poverty status in all models, national, outside Java, and Java. This finding
confirms the study by Sumarto et al. (2005) who explained that safety net
programmes to help the poor and the new poor overcome the impact of the
upcoming economic crisis, including food security, job creation, education,
health, and community empowerment.

CONCLUSION
Observing the survey data set of agricultural household incomes and
using a monetary and a multidimensional approach to measure poverty in
Indonesian capture fisheries fishermen’s households, we find that Indonesian
capture fisheries fishermen’s households are poorer in monetary terms or
have a per capita income below poverty line. Then a regression model was
used to analyse the relationship between institutional access and poverty in
capture fisheries households in Indonesia, and we divided the sample for
poverty analysis in households in Java and outside Java.
We find that the sign of the coefficients in the three models is the same
for the variables of gender, age, education of the head of the household, and
household size and is an important demographic determinant of poverty
status. Likewise, with institutional access variables—i.e., access to credit,
use of cooperatives, membership of fishermen’s groups (KUB), access to
production facilities, and market access—the sign of the coeffients the three
models remains the same. However, for grant programmes for fishermen,
namely boat grants, fishing gear grants, and government/non-government
business financing grants, only fishing gear grants are significant for
monetary poverty status in all models, national, outside Java, and Java.
The estimation results of this chapter confirm that increasing the
capacity of formal institutions such as cooperatives and fishermen’s groups
(KUB) is particularly important to ensure the sustainability of fishermen’s
household businesses through credit services on easy terms and assisting
with the marketing of their catches. In addition, many cases of poverty
alleviation are carried out by making fishing communities as objects. This is
done, for example, in the form of boats and fishing gear that do not refer to
the needs of fishermen, but packages that have been determined from above
and tend to be uniform across regions. This top-down aid system resulted
in ineffective assistance. For this reason, the policy format for assistance
to fishermen must be based on dialogue with local fishing communities to
adjust the target of assistance according to their needs.
REFERENCE
Acheson, J.M., 1981. Anthropology of fishing. Annual review of anthropology,
10(1):275-316.
Ankrah Twumasi, M. Y. Jiang, B. Addai, Z. Ding, A.A. Chandio, P. Fosu,
D. Asante, A. Siaw, F.O. Danquah, B.A. Korankye, and G. Ntim-Amo.

2021. The impact of cooperative membership on fish farm households’
income: The case of Ghana. Sustainability, 13 (3): 1059.
Amarasinghe, O., and M. Bavinck. 2017. Furthering the implementation of
the small-scale fisheries guidelines: strengthening fisheries cooperatives
in Sri Lanka. In The small-scale fisheries guidelines. Springer, Cham. 379-
399.
Alkire, S. and J. Foster. 2011a. Counting and multidimensional poverty
measurement. Journal of public economics, 95 (7-8):476-487.
________. 2011b. Understandings and misunderstandings of multidimensional
poverty measurement. The Journal of Economic Inequality, 9 (2):289-314.
Alkire, S., J.M. Roche, M.E. Santos, and S. Seth. 2011. Multidimensional
poverty index 2011: Brief methodological note.
https://ophi.org.uk/wp-content/uploads/MPI_2011_Methodology_
Note_4-11-2011_1500.pdf
Alkire, S. and M.E. Santos. 2014. Measuring acute poverty in the developing
world: Robustness and scope of the multidimensional poverty index.
World Development, 59 : 251-274.
Anna, Z., A.A. Yusuf, A.S. Alisjahbana, and A.A. Ghina. 2019. Are fishermen
happier? Evidence from a large-scale subjective well-being survey in a
lower-middle-income country. Marine Policy, 106:103559.
Bailey, C. and S. Jentoft. 1990. Hard choices in fisheries development. Marine
policy, 14 (4): 333-344.
Banks, L.M., M.Pinilla-Roncancio, H. Walsham, S. Neupane Van Minh, V.Q.
Mai, S. Neupane, K. Blanchet, and H. Kuper. 2021. Does disability
increase the risk of poverty ‘in all its forms’? Comparing monetary and
multidimensional poverty in Vietnam and Nepal. Oxford Development
Studies, 49 (4): 386-400.
Batana, Y.M. 2013. Multidimensional measurement of poverty among
women in Sub-Saharan Africa. Social Indicators Research, 112 (2): 337-362.
Béné, C. and R.M. Friend. 2011. Poverty in small-scale fisheries: old issue,
new analysis. Progress in Development Studies, 11 (2): 119-144.
Béné, C., S. Devereux, and K. Roelen. 2015. Social protection and sustainable
natural resource management: initial findings and good practices from
small-scale fisheries. FAO Fisheries and Aquaculture Circular, (C1106): I.

Béné, C., R. Arthur, H. Norbury, E.H. Allison, M. Beveridge, S. Bush, L.
Campling, W. Leschen, D. Little, D. Squires, and S.H. Thilsted. 2016.
Contribution of fisheries and aquaculture to food security and poverty
reduction: assessing the current evidence. World Development, 79 : 177-
196.
Budiantoro, S., V. Fanggidae, W. Saputra, A. Maftuchan, and D.R.P.
Artha. 2013. “Multidimensional Poverty Index (MPI): Konsep dan
Pengukurannya di Indonesia.” Working Paper. Perkumpulan Prakarsa.
https://repository.theprakarsa.org/publications/667/
multidimensional-poverty-index-mpi-konsep-dan-pengukurannya-di-
indonesia
Badan Pusat Statistik. 2015. Analisis Tematik ST 2013 Subsektor Analisis
Kesejahteraan Rumah Tangga Usaha Perikanan. Jakarta: BPS.
Dartanto, T. and Nurkholis. 2013. The determinants of poverty dynamics
in Indonesia: evidence from panel data. Bulletin of Indonesian Economic
Studies, 49 (1): 61-84.
FAO. 2018. The State of World Fisheries and Aquaculture 2018 - Meeting the
sustainable development goals. Rome. License: CC BY-NC-SA 3.0 IGO.
Fromentin, J.M., S. Bonhommeau, H. Arrizabalaga, and L.T. Kell. 2014. The
spectre of uncertainty in management of exploited fish stocks: The
illustrative case of Atlantic bluefin tuna. Marine Policy, 47 : 8-14.
Haque, C.E., C.J. Idrobo, F. Berkes, and D. Giesbrecht. 2015. Small-scale
fishers’ adaptations to change: The role of formal and informal credit in
Paraty, Brazil. Marine Policy, 51 : 401-407.
Hanandita, W. and G. Tampubolon. 2016. Multidimensional poverty in
Indonesia: Trend over the last decade (2003–2013). Social Indicators
Research, 128 (2): 559-587.
Hulme, D., K. Moore, and A. Shepherd. 2001. Chronic poverty: meanings and
analytical frameworks. Working Paper (2). Chronic Poverty Research
Centre.
Imron, M., 2003. Kemiskinan dalam masyarakat nelayan. Jurnal Masyarakat
dan Budaya, 5 (1): 63-82.
Islam, G.M.N., K.M. Noh, S.F. Sidique, and A.F.M. Noh. 2014. Economic
impact of artificial reefs: A case study of small-scale fishers in

Terengganu, Peninsular Malaysia. Fisheries Research, 151 : 122-129.
Islam, G.M.N., T.S. Yew, and K.K. Viswanathan. 2014. Poverty and livelihood
impacts of community-based fisheries management in Bangladesh.
Ocean & coastal management, 96 : 123-129.
Jentoft, S. and A. Eide. (eds.) 2011. Poverty mosaics: realities and prospects in
small-scale fisheries. Springer Science & Business Media.
Kolade, O. and T. Harpham. 2014. Impact of cooperative membership on
farmers’ uptake of technological innovations in Southwest Nigeria.
Development Studies Research. An Open Access Journal, 1 (1): 340-353.
Khan, M.A., M.F. Alam, and K.J. Islam. 2012. The impact of co-management
on household income and expenditure: An empirical analysis of
common property fishery resource management in Bangladesh. Ocean
& Coastal Management, 65 : 67-78.
Kim, H. 2019. Beyond monetary poverty analysis: The dynamics of
multidimensional child poverty in developing countries. Social Indicators
Research, 141 (3): 1107-1136.
Ogutu, S.O. and M. Qaim. 2019. Commercialization of the small farm sector
and multidimensional poverty. World Development, 114 : 281-293.
Parappurathu, S., C. Ramachandran, K.K. Baiju, and A.K. Xavier. 2019.
Formal versus informal: Insights into the credit transactions of small-
scale fishers along the southwest coast of India. Marine Policy, 103 : 101-
112.
Pitchford, J.W., E.A. Codling, and D. Psarra. 2007. Uncertainty and
sustainability in fisheries and the benefit of marine protected areas.
Ecological Modelling, 207 (2-4): 286-292.
Pusat Data Statistik dan Informasi. 2014. Kelautan dan Perikanan dalam Angka
Tahun 2015. Jakarta: Kementrian Kelautan dan Perikanan Republik
Indonesia.
Satria, D. and E. Li. 2017. Contract engagement in the small-scale tuna-
fishing economies of East Java. Bulletin of Indonesian Economic Studies,
53 (1): 27-54.
Stanford, R.J., B. Wiryawan, D.G. Bengen, R. Febriamansyah, and J. Haluan.
2013. Exploring fisheries dependency and its relationship to poverty: A
case study of West Sumatra, Indonesia. Ocean & coastal management, 84

: 140-152.
Solaymani, S. and F. Kari. 2014. Poverty evaluation in the Malaysian fishery
community. Ocean & Coastal Management, 95 : 165-175.
Sumarto, S. and I. De Silva. 2014. Beyond the Headcount: Examining the
Dynamics and Patterns of Multidimensional Poverty in Indonesia.
Sumarto, S., A. Suryahadi, and W. Widyanti. 2005. Assessing the impact of
Indonesian social safety net programmemes on household welfare and
poverty dynamics. The European journal of development research, 17 (1):
155-177.
Yu, J., 2013. Multidimensional poverty in China: Findings based on the
CHNS. Social indicators research, 112 (2): 315-336.
Yusuf, A.A. and A. Sumner. 2017. “Multidimensional poverty in Indonesia:
How inclusive has economic growth been? “Working Paper in Trade and
Development (No. 2017-09). Arndt-Corden Department of Economics.
Australian National University.
Yusuf, A.A., 2018. Keadilan untuk Pertumbuhan. Unpad Press.


Buku Institusi

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Buku Institusi

Uploaded by

Copyright:

Available Formats

Institution, Human Capital and Development

IRSA Book Series On Regional Development No. 20

Hak Cipta dilindungi Undang-Undang

iv Institution, Human Capital and Development in Indonesia

Diah Wulan SARI

vi Institution, Human Capital and Development in Indonesia

Ghifari Ramadhan FIRMAN

Gumilang Aryo SAHADEWO

Harnum Yulia SARI

I Gede Sthitaprajna VIRANANDA

Raka Rizky FADILLA

Rizka Isnaini HUSNA

viii Institution, Human Capital and Development in Indonesia

Indonesian government aspires to escape from the middle-income trap

x Institution, Human Capital and Development in Indonesia

Bandung, June 2022

Institutional and Human Capital Aspects

Source: BPS via CEIC.

2 Institution, Human Capital and Development in Indonesia

Institutional and Human Capital Aspects of 3

(0,50) 0,00 0,50 1,00 1,50 2,00

2000 2010 2020

4 Institution, Human Capital and Development in Indonesia

0 0,5 1 1,5 2 2,5 3 3,5 4

2000 2010 2020

Institutional and Human Capital Aspects of 5

Source: World Development Indicator

6 Institution, Human Capital and Development in Indonesia

Institutional and Human Capital Aspects of 7

INSTITUTION, HUMAN CAPITAL, AND

8 Institution, Human Capital and Development in Indonesia

Institutional and Human Capital Aspects of 9

10 Institution, Human Capital and Development in Indonesia

Institutional and Human Capital Aspects of 11

12 Institution, Human Capital and Development in Indonesia

Institutional and Human Capital Aspects of 13

Productivity Impact of Political

DEVELOPMENT OF BATAM FTZ

16 Institution, Human Capital and Development in Indonesia

Productivity Impact of Political Dynamics in A Decentralised Indonesia: 17

Decentralisation Reform and Revitalisation of Batam FTZ

18 Institution, Human Capital and Development in Indonesia

Source : BKPM, 2021

Productivity Impact of Political Dynamics in A Decentralised Indonesia: 19

Source: peraturan.bpk.go.id, 2021

20 Institution, Human Capital and Development in Indonesia

DATA AND ESTIMATION STRATEGY

Productivity Impact of Political Dynamics in A Decentralised Indonesia: 21

22 Institution, Human Capital and Development in Indonesia

After Policy Intervention

As indicated in the statistics of variables used (Table 1), the total

Productivity Impact of Political Dynamics in A Decentralised Indonesia: 23

Parallel Trend Test

24 Institution, Human Capital and Development in Indonesia

Productivity Impact of Political Dynamics in A Decentralised Indonesia: 25

RESULTS AND DISCUSSION

26 Institution, Human Capital and Development in Indonesia

Post 0.222*** 0.233*** 0.227*** 0.226***

Post*Treat -0.392** -0.423*** -0.303*** -0.309***

Constant 10.225*** 6.739*** 5.671*** 3.611***

Province FE Yes No Yes No

Productivity Impact of Political Dynamics in A Decentralised Indonesia: 27

28 Institution, Human Capital and Development in Indonesia

Treat 2.436*** 2.328*** -1.463* -1.448* 1.379* 1.337* -2.003*** -1.03

Post 0.222* 0.233* 0.227* 0.226*

PostTreat -0.392 -0.423 -0.303* -0.309*

Constant 10.225* 6.739* 5.671* 3.611*

Treat 2.436* 2.328* -1.463* -1.448* 1.379* 1.337* -2.003*** -1.03

PostTreat -0.368 -0.375* -0.271* -0.175 -0.279* -0.267* -0.242 -0.106

Mach. 0.066* 0.090* 0.064* 0.077* 0.051* 0.058* 0.057* 0.061*

Constant 7.962* 7.812* 9.651* 8.645* 3.277* 3.172* 4.633* 3.723*

Treat 1.02 -0.7 1.763* -2.071 0.57 -2.570*** 0.6

PostTreat -0.229 -0.346 -0.513* 0.051 -0.012 0.469 1.275*

Constant 4.110* 2.644* 3.901* 3.700* 3.680* 3.374*** 2.439

Social Protection 0.361 0.627* 0.262 0.213 0.124