Decentralisation and Citizen Happiness: A Multilevel Analysis of Self-rated Happiness in Indonesia

RESEARCH PAPER

Decentralisation and Citizen Happiness: A MultilevelAnalysis of Self-rated Happiness in Indonesia

Sujarwoto Sujarwoto • Gindo Tampubolon

� Springer Science+Business Media Dordrecht 2014

Abstract The effect of decentralisation reform on citizen happiness in the context of a

developing country remains under-investigated. We study this effect by examining the

experience of Indonesia. Data comes from the Indonesian Family Life Survey 2007 which

comprises face-to-face interviews with the adult population in Indonesia (Nindividuals =

29,024, Ndistricts = 262). Multilevel analysis is used to examine the effect of fiscal and

political decentralisation on citizen happiness, and shows that fiscal decentralisation is

significantly associated with citizen happiness, while political decentralisation is not; these

are robust to the inclusion of individual and district variables related to happiness. The

findings suggest that decentralisation increases the happiness of citizens through the

improved capacity of districts to deliver public services, rather than through the improved

opportunities of citizens to engage in direct political participation.

Keywords Decentralisation � Self-rated happiness �Multilevel analyses � Indonesia

1 Introduction

Over the last three decades, decentralisation has become a key development reform across

developing countries to promote citizen welfare and wellbeing (World Bank 2008). Yet

how and to what extent this reform affects citizens’ happiness remains under-investigated.

Although a few studies examining the relationship between decentralisation and citizen

happiness can be found in the field of happiness research, most are either based on cross-

country data or have been conducted within the context of a single developed country (Frey

Electronic supplementary material The online version of this article (doi:10.1007/s10902-014-9518-3)contains supplementary material, which is available to authorized users.

S. Sujarwoto (&) � G. TampubolonInstitute for Social Change, University of Manchester, Manchester, UKe-mail: [email protected]

123

J Happiness StudDOI 10.1007/s10902-014-9518-3

http://dx.doi.org/10.1007/s10902-014-9518-3

and Stutzer 2000; Bjørnskov et al. 2008; Hessami 2010; Diaz-Serrano and Rodriguez-Pose

2012).

This study contributes to the existing literature on decentralisation and subjective

wellbeing in three ways. Firstly, it uses a large sample of district governments as a unit of

analysis. The locus of decentralisation reform is generally found at district level of

administration, that is, below provincial government level (World Bank 2008). Using

district government as the unit of analysis is thus more relevant when scrutinising the

effects of decentralisation than using either country or provincial government, since it is

able to capture the dynamic of reform within those units which are directly affected by it.

Secondly, using district as the unit of analysis enables us to control for unobserved het-

erogeneities that in cross-country studies might lead to biased estimates (such as differ-

ences between cultures and overall institutional settings) (Helliwell 2006; Helliwell and

Huang 2008). A number of studies show that the performance of decentralisation is

embedded in the political, institutional, economic and social context of a country

(Rondinelli et al. 1989; Litvack et al. 1998). District government studies have the

advantage, since such units are considerably more similar within a single country than

across countries. Thirdly, the multilevel model used in this study is able to account for the

clustering of citizens within district by separating their variance in happiness from the

district variance in happiness. Using this model is thus the most appropriate to test

hypotheses about the effects of varying district characteristics on citizen happiness.

The Indonesian radical decentralisation in 2001 makes it an interesting case. Firstly, it

has brought political freedom to the districts by giving local citizens the right to select their

mayor and parliament through direct election. Secondly, districts now have greater

authority to deliver public services, as most of the responsibility for managing them has

been handed over to the district government. This has been supported by substantial

financial transfer: around 30 % of total national expenditure has been devolved to districts

(World Bank 2008), making district councils more independent to initiate policies to meet

the demands of local citizens. There is still however lack of empirical research examining

the implications of Indonesian decentralisation on citizen happiness.

The paper is structured as follows: Sect. 2 presents overview of previous studies and

describes the theoretical link between decentralisation and citizen happiness. Next, we look

at how decentralisation reform has played out in Indonesia. Section 4 presents the data,

variables and multilevel analyses that our research uses and applies. Section 5 reports the

results of the estimation, while Sect. 6 discusses and concludes.

2 Decentralisation and Citizen Happiness

Government is a key actor in determining the welfare and wellbeing of its citizens through

the use of its instrumental powers, such as formal institutions, social resources and public

policies (Kim and Kim 2012). However, only a very limited number of studies have so far

attempted to assess the impact of decentralisation reform on citizen happiness (Frey and

Stutzer 2000; Mullen 2011). Of those that do, most use countries as the unit of analysis

(Veenhoven 2000; Radcliff 2001; Bjørnskov et al. 2008; Hessami 2010; Rodriguez-Pose

and Maslauskaite 2011). Only one study, that of Frey and Stutzer (2000), investigates the

relationship between decentralisation and happiness within a single country context.

Examining how it applies to 6,000 individuals in 26 Swiss cantons, they find that local

autonomy and local direct democracy increase happiness. They suggest that this might be

S. Sujarwoto, G. Tampubolon

123

attributed to two factors: local citizens feeling more closely connected to political out-

comes, and political participation bringing benefits to those involved.

When it comes to cross-country studies, the findings those which examine decentrali-

sation and subjective wellbeing present mixed results. Bjørnskov et al. (2008) investigate

the relationship between decentralisation and life satisfaction in 60,000 individuals from 66

countries using data obtained by the World Values Survey 1997–2007. They find that a

country with greater revenue and decentralised spending is likely to be happier than one

with less. However, they also find that political decentralisation only enhances life satis-

faction when the government concerned also spends more on public services. This finding

suggests that to increase citizen happiness, political decentralisation needs to be accom-

panied by more resources being allocated to public services, which are delivered efficiently

by a government with the capacity to do so. Rodriguez-Pose and Maslauskaite (2011)

compare data from the 1999 and 2000 European Values Study to examine the association

between decentralisation, governance and happiness in ten Central and Eastern European

countries, and their findings are twofold. Firstly, they show that people living in a country

with greater political and fiscal decentralisation are likely to be more satisfied with life than

those living in countries with less. They also found that corruption within a decentralised

country is detrimental to the life satisfaction of its citizens. Hessami (2010) concurs, while

at the same time examining the effect of government size on citizen happiness. Referring to

twelve European Union countries covered by the European Barometer Survey 1990–2000,

he finds that happiness follows an inverse U-shape on the size of public sector, while the

effect of public sector size on wellbeing depends positively on the extent of decentrali-

sation and negatively on the level of corruption.

Other studies report that the degree to which decentralisation depends on the particular

political and economic characteristics of the country concerned. Veenhoven (2000) finds

that political and private freedoms exert a positive influence on life satisfaction only in

countries that already enjoy well-established democracies. Using the five waves of the

World Values Survey 1981–2005, Bjørnskov et al. (2010) find that the more efficient

performance of better economic and judicial institutions has a significant influence on

happiness in low income countries, while happiness in medium and high income countries

improves when political institutions perform better. Other studies (such as Pacek and

Radcliff 2008) go slightly further to uncover a positive relationship between a number of

features of the welfare state and subjective wellbeing.

Bjørnskov et al. (2008) explain mechanisms by which decentralisation leads to happi-

ness. Firstly, they find that it encourages local governments to provide citizens with the

public goods and services that they need. It entails a shift in political decision-making from

central to local government, with the implication that local government has a greater

potential to tailor its policies specifically to the demands of local citizens (Oates 1972).

This improved matching of public goods and service delivery to local needs necessarily

increases happiness levels of the beneficiaries. Secondly, decentralisation provides greater

political freedom, encouraging more participation in a local government, providing citizens

with increased opportunities to supply information about the goods and services they need

to those with the power to provide them (Seabright 1996). Decentralisation also facilitates

citizens taking a more active role in monitoring the performance of elected and admin-

istrative officials. Furthermore, when government is brought closer to the people, citizens

are more likely to demand quality of services, and to be more motivated to demand

improvements if that quality declines (Grindle 2007). This increased potential for public

complaint in turn provides civil servants with an enhanced incentive to orientate their

behaviour towards doing their job well. All of these contribute to local government

Decentralisation and Citizen Happiness

123

becoming more accountable and responsive, with the increased satisfaction of citizens

which results being reflected in their higher levels of overall happiness.

It has been found however that in some circumstances decentralisation can decrease

happiness. Bjørnskov et al. (2008) explain two ways this might happen. Firstly, decen-

tralisation often leads to loss of economic scale, which particularly affects the provision of

basic infrastructure and services, the need for which is universal and unlikely to change

across territories. Secondly, decentralisation does not guarantee the eradication of cor-

ruption and conflict, particularly when the mechanism to ensure local political account-

ability is weak (Bardhan and Mookherjee 2000). There is no reason why local corruption

within decentralised government should be any less invidious than at central level. There is

also the potential for local conflict to become more accentuated as power and economic

resources are distributed at local government level (Heller 2001). Local governments often

reflect a wide range of social, political, and economic conflicts, which can create division

within communities and have a consequent detrimental effect on happiness. Earlier cross-

country studies (for example Welsch 2008; Rodriguez-Pose and Maslauskaite 2011) pro-

vide evidence of the negative effects of corruption and civil conflict on happiness in

decentralised countries.

The extent to which decentralisation increases happiness thus depends on an array of

local conditions: Firstly, political freedom appears to be facilitated by decentralisation,

creating more opportunities for citizens to participate in the local political process, a

situation which tends to result in local government becoming more responsive to local

needs and concerns. However, when decentralisation is attended or followed by local

conflict, unhappiness is likely to ensue. Secondly, decentralisation is likely to increase

happiness when it means that the newly-acquired resources of government at local level are

more likely to match local needs, and citizens get what they ask for. However, when

decentralisation is followed by corruption, happiness is likely to decrease. Whereas some

of these hypotheses have been tested and supported in cross-country research, this study

differs in that it examines several district governments within one newly-decentralised

country.

3 Democracy and Political Transition in Contemporary Indonesia

Indonesia is a unitary state, meaning that subnational government is the creation of central

government. Up to the point of decentralisation, government had been highly-centralised,

and in the control of an authoritarian regime for over 32 years (Smoke and Lewis 1996).

During this time, districts had limited authority to decide local policies, and little ability to

respond to local preferences. Local democracy and citizen participation in politics were

restricted, as political parties and local leaders had to toe the party line. Mayors (or bupati)

were appointed and controlled by a centralised authority with lack of consideration for

local aspirations.

Economic and political crises in 1998 triggered political reform in Indonesia. Dissat-

isfaction with the regime forced it from power, and the first free and fair national, central

elections of parliament and president were held in 1999. More than forty new political

parties participated in this direct election which, coupled with new media openness, sup-

ported a burgeoning country-wide democracy. From then onwards, citizens experienced

increased freedom to make their voice heard through the press, over which the government

has continued to relax its control. By 2009, Indonesia had come to be termed a liberal


123

democracy (Freedom House 2009)1 by Freedom House, an international organisation

which monitors democracy worldwide.

In 2001, Indonesia embraced decentralisation, a move which transformed the country’s

political system. Decentralisation has given every district the power to perform the key

functions of state, including the provision of health, education, environmental and infra-

structure services. Districts can now perform any function not explicitly reserved for

central or provincial government. This increase in autonomy has been supported by the

substantial transfer of financial resources from the centre to the district: more than a third

of the national budget is now under district control (Fig. 1), a fiscal reform which in 2001

was accompanied by the reassignment of more than 2.5 million civil servants to the

districts (World Bank 2008). These changes have made districts more independent in

mobilising resources, promoting their own interests, and initiating local policies aimed at

meeting local citizen demands.

Further reforms in 2005 allowed citizens to elect their own mayor and parliament

through direct local elections: by the end of 2006, more than half of all districts had

conducted direct elections (Fig. 2). So far, 166 districts out of 305 have conducted direct

elections, voting in seven governors and 159 mayors. This is a significant step forward for

the Indonesian local political system, one which gives local citizens the freedom and

responsibility to choose leaders that they believe will best represent their aspirations.

After 5 years of decentralisation, the performance of district governments was found to

vary widely. World Bank (2006) found a number had forged ahead with reform and

become the loci for an innovative form of government and better public services. In other

districts however, the new political context has been charged with being responsible for a

new lack of accountability and failure to respond to local needs (Asia Foundation 2002).

Some scholars have questioned the depth of democratic participation in Indonesian elec-

tions. Observers of these elections have echoed a much wider body of comparative liter-

ature located in Southeast Asia, which notes the influence of political elites, patronage and

‘money politics’ in local elections (Aspinall and Mietzner 2010). Some evidence also

suggests that the political accountability mechanisms of a number of districts are weak

(Kaiser and Hofman 2003). In addition, local corruption, violence and ethnic conflict have

been widespread across districts following decentralisation (World Bank 2008). Decen-

tralisation has thus brought not only new freedoms but also a number of problems for local

citizens. This study aims to understand whether and to what extent these local political and

financial changes are associated with citizen happiness.

4 Methods

4.1 Indonesian Family Life Survey 2007 and Official Statistics

To examine the effects of political and fiscal decentralisation on citizen happiness, we

assembled district and individual data from various sources. The data possesses a

1 In 2009, Freedom House classified 119 out of 193 countries as electoral democracies. There is a sig-nificant difference between an ‘electoral democracy’ and a ‘liberal democracy’. In the former, reasonablyopen elections can coexist with serious defects in the implementation of individual rights, the rule of law andother preconditions crucial for a free society. Liberal democracies, by contrast, enjoy free, fair and com-petitive elections, but also a large catalogue of civil liberties. According to Freedom House, only 89 of the119 electoral democracies were classed as liberal democracies in 2009. Interestingly, this includedIndonesia.


123

multilevel structure, with individuals nested within districts. Data on individuals is taken

from IFLS 2007, while district data comes from nationally-representative surveys and

official statistics.

IFLS is an on-going longitudinal survey that began in Indonesia in 1993. It represents

83 % of the country’s non-institutionalised, adult population and above living in 13

provinces and 262 districts (Frankenberg and Thomas 2000), and brings together a rich set

of information on individuals and households, the communities they live in, and the

facilities available to them. Its participation rate is high, at 92 % (Thomas et al. 2012).

Although IFLS has a panel structure, the happiness, social trust and local governance

modules were introduced only in its fourth wave; this study thus uses the cross-sectional

structure of the data instead. It puts questions on happiness, social trust and local gover-

nance to citizens aged 15 years or older (this provides a sample of 29,060 respondents). On

average, there are about 100 respondents in each district, and respondents who migrate to

other regions (about 14 %) are also included in the analysis.

The IFLS data was linked to a number of other surveys and official statistical datasets

using district codes. Firstly, we linked it with the government Village Potential Statistics

Fig. 1 Trends of central-local government spending in Indonesia, 1994–2007. Source: World Bank 2008

Fig. 2 Local direct elections in Indonesia, 2005–2009. Source: The Ministry of Home Affairs 2010


123

(Podes) 2006.2 Podes contains detailed information about the incidence of local conflict

and violence, and the number of community groups within districts, calculating aggregates

at village and urban neighbourhood levels to measure their distribution. Next, we linked

the IFLS data with the local and national election database of the Indonesian Ministry of

Home Affairs. This contains information about those districts which by 2007 had already

implemented direct elections. Thirdly, we linked the IFLS data to district fiscal data.

Collected by the Ministry of Finance, this dataset provides detailed information ranging

from each district’s own revenue source, to balancing funds and general allocation funds

deriving from central government, and sectoral development expenditure (Indonesian

Ministry of Finance 2008). We use fiscal data from 2006 (the year prior to the IFLS

survey), as district development spending in the Indonesian budgeting system takes at least

1 year to take effect. Since price levels of consumer goods and services in Indonesia vary

across the country (Strauss et al. 2004), we deflated the amount of district spending with

the consumer price index for urban and rural regions. We took the rural inflation rate to be

5 % higher than urban inflation (Resosudarmo and Jotzo 2009), a calculation which pro-

duces real spending adjusted with urban and rural inflation. The Consumer Prices Index

2006 data is retrieved from the government Central Bureau of Statistics.

4.2 Variables

Table 1 presents summary statistics for the key variables used, which at district level are

derived from administrative reference data rather than the IFLS survey: this enhances

accuracy or at least reduces measurement error. The use of independent administrative data

also assuages concerns arising from the use of the same surveys to calculate both aggregate

and individual variables in the estimation (Deaton 2001). Only the proxy for district

corruption is calculated from the IFLS survey. To reduce measurement error further, we

excluded respondents from our calculation when constructing the aggregate for district

corruption. A detailed definition and explanation of each variable can be seen in supple-

mentary data online 3.

4.2.1 Happiness and Decentralisation Measures

According to IFLS, happiness is defined as the degree to which an individual judges the

overall quality of his/her life to be favourable (Veenhoven 1984). In the IFLS survey,

respondents were asked: ‘‘Taken all together, how would you say things are these days:

‘very happy’, ‘pretty happy’, ‘not too happy’ or ‘very unhappy’?’’ This item has been

widely used and been validated during use by previous studies (Frey and Stutzer 2002;

Krueger and Schkade 2008; Oswald and Wu 2011). Frey and Stutzer (2002) summarise a

number of ways to validate happiness data, and demonstrate that self-rated happiness is

strongly correlated with objective wellbeing. Stable wellbeing numbers are found over

time (Krueger and Schkade 2008), and a strong match has been found between subjective

and objective wellbeing (across the United States, by Oswald and Wu 2011). The majority

of respondents (85 %) fell into one of the ‘happy’ categories.

One way to define decentralisation is in terms of the degree of political and fiscal

autonomy experienced at district level. Political autonomy measures the discretionary

power in political decision-making awarded to each district, while fiscal autonomy refers to

2 Podes census is conducted by the Indonesian Central Bureau Statistics every 2 years. We use Podes 2006as there is no Podes census in 2007.


123

fiscal self-reliance and spending power (Schneider 2003). Following Schneider (2003), we

use direct local election as a proxy to measure political decentralisation, while fiscal

decentralisation is measured in terms of the share of central government transfer of total

district revenue, and total district spending on public services. Districts resulting from

direct local elections are almost certain to have more autonomy in decision-making than

those created by the former centralised regime. By 2007, more than half (69 %) of

Indonesia’s districts had a directly-elected mayor; the remaining districts continued to be

led by a mayor elected under the previous regime who had started their term before the

direct election bill came into effect. This unusual political condition provides the oppor-

tunity to examine the effect on happiness of two different systems of local government

running concurrently.

Table 1 Descriptive statistics of analytic sample

Percent Mean SD Range

Citizen happiness Happy

District (n = 262)

Direct election 70 46 % 0/1

Shared of fiscal transfer on total revenue 86 9 % 44/98 %

Log total spending for public services 32.35 0.46 31.30/34.26

Inefficiency on public spending 48 10 % 16/69 %

Local corruption 23 27 % 0/100 %

Local conflicts and violence 75 40 % 0/1

Leadership 13 34 % 0/1

Bureaucrats capacity 30 26 % 10/80 %

Community social capital 30 20 % 0/89 %

Log gross domestic product 15.92 1.25 13.34/18.80

Geographical areas (1,000 km2) 258.7 352.4 1.1/2010.0

Individual (n = 29,024)

Age (years) 37 17 15/90

Female 52 50 % 0/1

Years schooling (years) 8 5 0/18

Unemployed 3 17 % 0/1

Divorced 2 15 % 0/1

Widowed 6 23 % 0/1

Religiosity Daily

Poor health 14 % 50 % 0/1

Social ties 99 % 38 % 0/1

Loneliness 5 % 21 % 0/1

No child 17 % 24 % 0/1

Household size 4 2 1/22

Household expenditure (IDR1,000) 624 630 28/1,360

Remote islands 5 7 % 0/1

Migrate 14 35 0/1

Source: IFLS 2007 & official statistics


123

Fiscal decentralisation is measured by the proportion of total district revenue which

comes from central government transfer, total district spending on public services, and

efficiency or otherwise of its budget allocation (Schneider 2003). The sum that each district

receives from central government is an indication of its fiscal self-reliance, while the

amount it spends and the way it spends it is a measure of district capacity to deliver the

services that its citizens need. Most districts in Indonesia are reliant on central government

funds; at 86 %, the average proportion of central government fiscal transfer of total rev-

enue is quite large. Log spending for public services also varies across districts, with a

range of between 31.30 and 34.26.

Central government transfer does not necessarily lead to increased wellbeing, particu-

larly if its allocation is not carried out efficiently. For example, some districts may receive

a small amount of local revenue, thus their share of central government subsidy will be

large despite being nothing like sufficient. The efficiency of local budget allocation may

thus matter for citizen wellbeing, since it reflects the financial capacity of the district to

deliver services. Following the World Bank (2008), we use the ratio of total spending on

civil service salaries to total district development expenditure—which at 48 % is relatively

large—to measure the efficiency of the latter. Most districts in Indonesia manage their

development expenditure inefficiently.

4.2.2 Local Corruption and Conflict

Earlier qualitative studies (Kaiser and Hofman 2003; Choi 2004; Nordholt and Van-

Klinken 2005) have found that direct local elections in Indonesia give rise to two concerns

regarding corruption in local politics, which we measure by the perceived degree of

nepotism and ‘money politics’ operating specifically in mayoral elections. Firstly, the

increasing issue of a mayoral candidate being deemed to be a putra daerah (‘son of the

region’) has proved to be the embryo of the rise of primordialism and nepotism (Kaiser and

Hofman 2003). Secondly, the issue of politik uang (or ‘money politics’) gained prominence

following the inception of direct local elections; today, this is generally accepted to be

endemic particularly during the local election process. Districts with higher incidence of

primordialism, nepotism and money politics are, not surprisingly, more corrupt, with clear

evidence of the mayor, bureaucrats and local parliament officials tending to misuse public

money in their own interests (Choi 2004; Nordholt and Van-Klinken 2005).

This study constructs the aggregate score of those respondents who choose their

mayor on the basis of being of the same ethnicity and distributing money during their

mayoral campaign to measure primordialism, nepotism and the culture of money politics.

Those districts with a higher share of respondents who chose a leader because they come

from the same ethnic background and hand out money during their campaign are con-

sidered to be more corrupt. On average, 23 % of respondents voted for a candidate on

this basis in 2007.

To address whether local conflict was detrimental to citizen happiness, a dummy

variable to indicate conflict and violence within each district is constructed. Nordholt

and Van-Klinken (2005) report that decentralisation in Indonesia was followed by an

increasing amount of local conflict and violence, and indicate the inability of local

democracy to guarantee effective local leadership. They demonstrate in some cases the

weak capacity of the newly-created district authority to manage both their responsi-

bilities and resources. In 2006, the percentage of incidence of conflict and violence at

local level was very high, with 75 % of districts reporting occurrences of unrest.


123

4.2.3 District leadership, Capacity of Bureaucrats, and Social Capital

Decentralisation in Indonesia introduced a new style of leadership at district level. To

capture this, we constructed the dummy variable of a mayor elected from a new political

party (that is, not from one of the three parties of the previous regime: Golkar, Partai

Persatuan Pembangunan, and Partai Demokrasi Indonesia Perjuangan). Thirteen per cent

of mayors elected since 2006 fall into this category.

Grindle (2007) discusses the capacity of local bureaucrats (such as bupati) and finds a

relationship between this and effective decentralisation. Since the end-point of district

public services is located in village and urban neighbourhoods in the Indonesian post-

decentralisation administrative system, we examined the proportion of village and urban

neighbourhood heads who have attained graduate education or above 30 % and took this as

our proxy to measure the capacity of local bureaucrats.

Prior studies have also examined the beneficial effect of community social capital on

decentralisation performance. Putnam (1993) found that social groups not only exert

pressure on district governments to provide better services, but also present models of the

services that will best enhance citizen wellbeing. At the same time, such groups provide a

channel for the personal and social support that increases happiness. Following Putnam

(1993), we use the density of community groups active in a district to measure social

capital. This provides information about, among others, kelompok pengajian and kelompok

kebaktian (religious groups), karang taruna (youth groups), persatuan kematian (funeral

groups), and kelompok wanita (women’s groups), all active community groups found

within villages or urban neighbourhoods. We calculated the aggregate number for each

active social group within a district, and found that community social capital varies across

districts, with range between 0 and 89 %.

4.2.4 Control Variables

We include a range of characteristics as control variables, addressing economic

development and geographic coverage of services, and individual and personal cir-

cumstances. A number of earlier studies use gross domestic product (GDP) as a

standard variable to capture economic development (Easterlin 1974; Diener et al. 1993),

and debate the positive and negative effects of GDP on happiness both within and

between developed countries (Easterlin 1974; Blanchflower and Oswald 2004). Next,

the geographical size of each district is included as a variable. Indonesia is an archi-

pelago of over 17,500 islands covering an area of 9.8 million km2 thus plays an

important role in determining citizens’ access to public services and potentially in its

relationship with happiness.

We have also included a range of individual and household characteristics—widely

used in single-level studies—as control variables. For example, happiness may differ

across age—Blanchflower and Oswald (2008) found that age has U-shape relationship

with subjective wellbeing, which is at its minimum in middle age. We thus include age

as a variable: most respondents were middle-aged, with the average age being 37 years.

Gender, being a parent or not, and marital status are included: being male, not having

children and being divorced and widowed are often negatively related to happiness (Lane

2000; Argyle 2001; Graham 2009). Half the respondents were women, of which 6 %

were widowed and 2 % divorced. A dummy variable of employment status was con-

structed to address the issue of whether being unemployed has a detrimental effect on


123

happiness (Clark and Oswald 1994). The proportion of unemployed in the IFLS survey is

low (3 %).3

Health status is also used as a control, to determine whether respondents at the lower

end of distribution of self-rated health are less likely to be happy. Self-rated health predicts

chronic disease in many settings (Frankenberg and Jones 2004), and all respondents in the

IFLS survey report on this. The survey asks, ‘‘In general, would you say that you are ‘very

healthy’, ‘sufficiently healthy’, ‘less than healthy’, or ‘unhealthy’?’’ We combine the first

two categories as ‘good health’, and the last two as ‘poor health’, and find that 14 % of

respondents report having poor health.

Social ties and religiosity are also included, following earlier studies which found these

to be variables of individual happiness (Helliwell 2006; Abdel-Khalek 2006). Helliwell

(2006), for example, finds social ties to be significant, and we measure these by using

household size and individual willingness to help neighbours as proxy indicators. Several

studies have also found that people who identify as religiously devout tend to enjoy not

only better mental health but also happiness (Abdel-Khalek 2006). To determine whether

religiosity matters to happiness, we thus construct a variable to indicate whether respon-

dents carry out daily prayer; the majority of respondents report that they do.

The other personal variable that strongly affects happiness is loneliness; Lane (2000)

finds that loneliness has a depressing effect on happiness, and to control for this we

construct a dummy variable indicating whether respondents’ spouse lives away from home.

We find 5 % of respondents report in the affirmative. Previous studies also show that

education may contribute to happiness by enabling individuals to better adapt to changing

environments. On the other hand, education also tends to raise aspiration levels which may,

if unmet, decrease happiness (Graham 2009). To address this, we construct the number of

years a respondent attended school as a measure of education. Most respondents were

educated to secondary high school level (representing 9 years in school).

Household expenditure is included to determine whether household income is associ-

ated with happiness (household and individual income are consistent predictors of sub-

jective wellbeing) (Blanchflower and Oswald 2004). We use household expenditure as a

proxy of income, as information regarding the latter is often biased and difficult to assess in

developing countries, particularly in subsistence farming households. Income data is

typically prone to under-reporting and measurement error, with the contribution of own

production and in-kind transfers often overlooked: household expenditure is thus a more

accurate measure of household economic resources, both in developing and developed

countries (Deaton and Zaidi 2002; Jorgensen 2002). This study refers to household

expenditure adjusted with the 2007 consumer price index data for urban and rural areas.

The average real monthly household expenditure in 2007 was IDR575,000 (USD57.5).

Migration is included to control whether it makes respondents less happy than those who

are settled in one district. To address the effect of geographic location, we construct a

dummy variable representing respondents living in remote islands, with respondents living

on the mainland (a more developed region) used as a reference group. We find 0.5 % of

respondents live on remote islands.

3 The percentage of unemployment in IFLS 2007 is lower compare with the World Bank 2007 unem-ployment data (9.1 %). One explanation could explain these differences are that both IFLS and World Bankuse difference measure of unemployment. In the IFLS survey respondent is asked during the past week, didyou do work for pay. This question is used to measure unemployed status. We checked this number with theIndonesian Socio economic survey (Susenas) 2007 which also used the same question. The result is same.


123

4.3 Multilevel Analysis

This study uses multilevel analysis to account for the effects of both political and fiscal

decentralisation on citizen happiness. This is more appropriate than the use of Ordinary

Least Squared (OLS) regression by aggregating data at individual level (Frey and Stutzer

2002) or ignoring the nested structure of data (Bjørnskov et al. 2008); if ignored, this

nesting of individuals within large district units can lead to the underestimation of standard

errors of the effect of district characteristics (Snijders and Bosker 1999). One consequence

of failing to recognise hierarchical structures is that standard errors of the regression

coefficient will be underestimated, leading to an overstatement of statistical significance.

Standard errors of the coefficient of higher-level predictors will be the most affected by

ignoring the grouping which occurs in OLS estimation. Multilevel analyses on the other

hand combine the regression and the variance component models to account for the

nested structure of the data. The variance component models can be written thus:

Yij = b0 ? lj ? eij, eij * N(0,re2), lj * N(0,rl

2), where Yij is the value of Y for individual

in j district, b0 is the overall mean of Y (across all districts), and lj and eij refer to district

and individual level residuals respectively. Residuals at both levels are assumed to follow

normal distribution, where zero means eij * N(0,re2), lj * N(0,rl

2). The total variance is

partitioned into the between-district variance (rl2) and the within-district variance (re

2).

These variances capture the effects of unobserved heterogeneities variables that are

independent of the covariates in the model (Rabe-Hesketh and Skrondal 2012). The effect

of unobserved heterogeneities in cross-country study of decentralisation is larger compare

to cross-districts study since such units are considerably more similar within a single

country than across countries.

In this study, we have data in J districts, and a different number of individuals Nj in each

district. A multilevel model can be thus written:

Yij ¼ b0 þ b1Xij þ lj þ eij; eij�N 0; r2e

� �; ljN 0; r2

l

� �

where Yij is the happiness of each citizen in j district, b0 is the overall mean of citizen

happiness across all districts, b1Xij represent the slope coefficients of independent vari-

ables, lj and eij refer to district and individual level residuals respectively, and re2 and rl

2

are variances within and between districts respectively. This model accounts for the

clustering of individuals by separating individual variance in happiness from that of district

variance in happiness.

We carried out multilevel analyses with generalised linear latent and mixed models

commands (GLLAMMs) using Stata 11.2. Rabe-Hesketh and Skrondal (2012) explain

that GLLAMMs are a class of multilevel latent variable models for (multivariate)

responses of mixed type, including continuous responses, counts, duration/survival data,

dichotomous, ordered and unordered categorical responses and rankings. In this ana-

lysis, GLLAMM is used with an ordinal probit link, as the dependent variable (self-

rated happiness) is ordinal. To test the robustness of the results we also estimate the

model using the linear mixed-effect regression command (xtmixed). The results of

xtmixed do not differ substantially from those obtained by GLLAMMs (see Supple-

mentary data online 1). For each of the models, the estimated coefficient, standard

errors, district and individual variances, intra-class correlation and log likelihood as an

indicator of model fit are reported. All models were estimated using maximum likeli-

hood estimation.


123

5 Results

Table 2 presents the regression results of the OLS and multilevel analyses, and shows the

standard error for the OLS coefficient is lower than the multilevel regression coefficient.

Standard errors for the district variables are the most affected in the OLS which is to be

expected as, it ignores the nested structure of the data.

To examine the degree of variance in citizen happiness at individual and district levels,

before regressing happiness on any predictors we estimated an empty model (which

includes only variance components; not reported in the tables). This revealed that the

individual variance in citizen happiness to be explained is 0.049, and the unexplained

district variance comes to 0.007. Thus the intra-class correlation (ICC) is 0.007/

(0.007 ? 0.049) = 0.125. This means that 12.5 % of the variation occurs between district

units, which if ignored can lead to inefficient and biased estimates.

Model A and B present regression results before and after individual control variables

are included. Fiscal decentralisation is positively associated with citizen happiness, as

indicated by the significant association of district spending with public services, and of

central government fiscal transfer with happiness (0.384, p \ 1 % and 0.014, p \ 1 %

respectively). In contrast, the association between local direct elections and happiness

appears not to be significant (0.015, p [ 5 %). Inefficiency in district budgeting (as

indicated by the high proportion of the total development budget allocated to civil service

salaries) is negatively related to happiness (-0.100, p \ 1 %). The higher capacity of local

bureaucracy (as indicated by the education level attained by village/neighbourhood heads)

increases citizen happiness, but the association is not significant. Local corruption is

detrimental to happiness: those living in districts with a higher tendency to nepotism and

‘money politics’ during mayoral elections are likely to be less happy (-0.200, p \ 1 %).

Living within regions of conflict and violence make citizens less happy than living in

regions which are more peaceful and safe (-0.127, p \ 5 %). The ineffectiveness of new

mayoral leadership in enhancing happiness is shown by the insignificance of this variable

(-0.031, p [ 10 %).

Cross-level interaction between higher district public service spending on unemploy-

ment and poor health is estimated to account for the effect of district capacity to deliver

services on happiness levels of unemployed and unhealthy citizens (Table 3). We found

higher district spending on public services is significant for the happiness of the unem-

ployed people (0.427, p \ 5 %). This indicates that high spending on people who are

unemployed increases their happiness. However, the effect of the interaction between

higher public services spending and poor health (0.017, p [ 5 %) is not significant.

Community social capital (as represented by social groups active at district level) is

positively associated with citizen happiness (0.25, p \ 5 %). In all models, district GDP is

negatively associated with citizen happiness, with those living in richer districts being less

happy than those living in areas that are less rich. The size of geographical area seems not

to matter for the happiness of local citizens.

The lower part of the panel shows individual coefficients. Happiness is likely to

decrease with age, and all models show that women tend to be happier than men. Family

and social ties are shown to be important variables; all models show these to have a

positive and significant association with happiness. Being divorced and widowed are

associated with being less happy (as opposed to being married and being single). Likewise,

being separated and being without children are likely to decrease happiness (Lane 2000).

Having more education and more household expenditure make people happier: the sig-

nificance of both variables is shown in all models. As might be expected, being


123

Table 2 Results of OLS and multilevel regression of citizen happiness

Multilevel Model

OLS Model A Model B

Coef. SE Coef. SE Coef. SE

District (n = 262)

Direct election 0.022 0.063 0.039 0.038 0.015 0.043

Shared of fiscal transferon total revenue

0.014*** 0.002 0.008*** 0.003 0.014*** 0.003

Log total spending forpublic services

0.350*** 0.052 0.320*** 0.081 0.384*** 0.092

Inefficiency of publicspending

-0.252*** 0.010 -0.167*** 0.062 -0.100** 0.013

Local corruption -0.230*** 0.051 -0.287** 0.072 -0.200** 0.083

Local conflicts andviolence

-0.112*** 0.031 -0.101** 0.046 -0.127** 0.052

Leadership -0.007 0.030 -0.085 0.050 -0.031 0.055

Bureaucrats capacity 0.181 0.076 0.156 0.133 0.167 0.128

Community socialcapital

0.229*** 0.078 0.416*** 0.114 0.248** 0.130

Log gross domesticproduct

-0.055** 0.0231 -0.079** 0.036 -0.082** 0.041

Geographical areas(1,000 km2)

0.023 0.051 -0.023 0.073 0.001 0.082


Age -0.011*** 0.001 -0.006*** 0.001

Age2 -0.000 0.011 -0.000 0.011

Female 0.056** 0.023 0.058** 0.023

Years schooling 0.023*** 0.003 0.031*** 0.003

Unemployed -0.281*** 0.083 -0.262*** 0.082

Divorced -0.650*** 0.067 -0.685*** 0.067

Widowed -0.198*** 0.050 -0.241*** 0.050

Religiosity 0.144*** 0.029 0.123*** 0.029

Poor health -0.249*** 0.021 -0.247*** 0.021

Social ties 0.108*** 0.028 0.123*** 0.028

Loneliness -0.237*** 0.050 -0.234*** 0.050

No child -0.271*** 0.053 -0.271*** 0.053

Household size 0.011** 0.006 0.011** 0.006

Household expenditure(IDR1,000)

0.225*** 0.019 0.222*** 0.019

Remote islands -0.036 0.263 -0.037 0.253

Migrate -0.038 0.031 -0.035 0.031

j1 6.457*** .806 1.739** 1.035 6.897*** 1.242

j2 7.898*** .806 3.142*** 0.034 8.370*** 1.242

j3 11.030*** .808 6.120*** 1.035 11.538*** 1.244

Variances

Individual 0.028*** 0.004 0.033*** 0.005


123

unemployed has a depressing effect (Clark and Oswald 1994). Not surprisingly, poor

health harms happiness. Religiosity (indicated by daily prayer) and social ties (indicated by

willingness to help ones neighbours) are likely to increase happiness (Lim and Putnam

2010). Living on a remote island and migrating at a young age are likely to decrease

happiness, but the associations found are not significant.

The variances at district level are significant across all specifications. The estimation of

these goes some way to ensuring that remaining estimates (from district spending on public

services to cross-level interaction between high spending and unemployed status) are

robust against unobserved district heterogeneities. Those single level studies which ignore

unobserved heterogeneities at either district or higher geographical levels may not be as

robust. This is worth bearing in mind when comparing these results with current literature.

6 Discussion and Conclusion

The question of what contributes to individual happiness has long been of interest to social

scientists in developed countries. However, this has rarely been explored in the context of

developing countries. Using the condition of radical decentralisation in Indonesia, we

examine the effect of political and fiscal decentralisation on citizen happiness. The main

results show that fiscal decentralisation is significantly associated with citizen happiness,

while political decentralisation is not. In all specifications, the share of central government

balancing funds and district spending allotted to public services have significant associa-

tion with citizen happiness. In contrast, null findings are found regarding the association of

direct local election with citizen happiness.

These contrasting results seem to signal that decentralisation in Indonesia increases

citizen happiness through the better capacity of district government to deliver public

services, rather than through the enhanced opportunities it provides for direct political

participation. This finding is supported by the variables which indicate district capacity to

deliver public goods and services. The capacity of district government to finance these

goods and services is significant for citizen happiness, as is local bureaucrat capacity as

indicated by the education of village/neighbourhood heads. Cross-level interaction

between high local spending on public services for the unemployed shows a positive

association with happiness, thus indicating its ameliorating effect on unemployment. In

contrast, budget allocation inefficiency and district corruption are detrimental to happiness.

Both the positive relation of fiscal decentralisation and the negative relation of local

corruption with happiness confirm the findings of Bjørnskov et al. (2008) and Rodriguez-

Pose and Maslauskaite (2011). However, these studies were examining the association

Table 2 continued

Multilevel Model

OLS Model A Model B

Coef. SE Coef. SE Coef. SE

District 0.005*** 0.000 0.006*** 0.000

ICC 15 % 16 %

Log likelihood -9053.41 -12,859.77 -9,010.14

Source: IFLS 2007 & official statistics, * p \ 0.10; ** p \ 0.05; *** p \ 0.001


123

Table 3 Results of multilevel regression for cross level interaction

Model C

Coef. SE

District (n = 262)

Direct election 0.015 0.730

Shared of fiscal transfer on total revenue 0.014*** 0.003

Log total spending for public services 0.369*** 0.096

Inefficiency of public spending -0.101** 0.016

Local corruption -0.201** 0.020

Local conflicts and violence -0.125** 0.052

Leadership -0.033 0.055

Bureaucrats capacity 0.162 0.129

Community social capital 0.245** 0.131

Log gross domestic product -0.082** 0.041

Geographical areas (1,000 km2) 0.002 0.083


Age -0.006*** 0.001

Age2 -0.000 0.012

Female 0.058*** 0.023

Years schooling 0.031*** 0.003

Unemployed -0.366*** 0.094

Divorced -0.685*** 0.067

Widowed -0.240*** 0.050

Religiosity 0.123** 0.029

Poor health -0.250*** 0.022

Social ties 0.123 0.028

Loneliness -0.235*** 0.050

No child -0.271*** 0.053

Household size 0.011** 0.006

Household expenditure (IDR1,000) 0.222*** 0.019

Remote islands -0.039 0.253

Migrate -0.034 0.031

Cross level interaction

Spending for public services 9 unemployed 0.427** 0.191

Spending for public services 9 poor health 0.017 0.053

j1 6.646*** 1.340

j2 8.120*** 1.340

j3 11.289*** 1.342

Variances

Individual 0.033** 0.001

District 0.006*** 0.000

ICC 16 %

Log likelihood -9,007.539

Source: IFLS 2007 & official statistics, * p \ 0.10; ** p \ 0.05; *** p \ 0.001


123

between fiscal decentralisation and happiness at cross-country level: our study establishes

the same association cross-district within a developing country. Its findings are thus

unique, in that they show that fiscal decentralisation and corruption matter for citizen

happiness not only between countries but also within a country.

Moreover, the null result of the relationship between direct local elections and citizen

happiness contradicts Frey and Stutzer (2000), whose findings show the positive effect of

direct election on happiness in well-established democratic countries. Such differing

results may however merely signal the extent to which the conditions of direct democracy

vary between a newly-decentralised country and one where democracy is well-established.

Decentralisation in Indonesia has been accompanied by an increasing amount of local

conflict, which has not only rendered local democracy less effective (Choi 2004; Nordholt

and Van-Klinken 2005), but which is also in itself detrimental to happiness. Under such

conditions, the widening participation of political parties and local direct elections may not

guarantee effective local leadership which can provide better local policies and services

which improve wellbeing. This study has demonstrated the impact on happiness of inef-

fective new local leadership.

The positive connection between community social capital and happiness shows that the

benefits of such capital do not only accrue to developed countries but also to developing

countries. This study finds a prominent role played by active social groups within village/

neighbourhood programmes, an indication that these groups facilitate their members’

participation in and access to the benefits that such programmes provide (programmes

which have themselves been established through decentralisation). In fact, the benefits

which these social groups bring (measured by various development outcomes such as

poverty reduction, child health, mental and physical health) has been widely documented

(see, among others, Grootaert 1999; Shiffman 2002; Miller et al. 2006); in Indonesia, we

find these benefits to pertain not only to development outcomes such as these but also to

citizen happiness.

The negative association between district GDP and happiness contrasts with a number

of cross-country analysis findings (Di Tella et al. 2003; Clark and Senik 2011), and needs

further investigation. For example, recent studies on Indonesian economic development

indicate that despite high growth rate between 2005 and 2011, economic inequality has

deepened. Asian Development Bank (2012) and Euromonitor (2012) report that Indonesia

has experienced the highest increases in income inequality levels worldwide (rising from a

Gini coefficient of 0.33 in 2005 to 0.47 in 2011); districts with the highest GDP also have

the highest economic inequality (World Bank 2008). Future research can usefully examine

the link between local GDP, economic inequality, and subjective wellbeing; gathering data

on district economic inequality may yield a better understanding of these negative

relationships.

This study leaves a number of limitations which need to be addressed. Firstly, because

of its cross-sectional design we have to be cautious about the possible causality of asso-

ciations. The estimated coefficient should be viewed as a measure of association, rather

than of causation. Almost every variable in our model could be considered endogenous, as

happiness affects almost all aspects of social, political, and economic life, for both indi-

viduals and institutions (Graham 2009). The causal effect of decentralisation and happiness

is something which future research, using available panel data on subjective well-being and

the most appropriate method, should seek to establish. Secondly, due to the lack of

availability of district corruption data, this study uses respondents’ answers as proxy of

corruption rather than examining specific corruption data from a separate source. We tried

to minimise measurement error by excluding respondents from calculations when


123

constructing the aggregate for the proxy data. Nevertheless, additional data collection and

analysis are needed to ensure a more reliable measure of district corruption. Kaufmann

et al. (2006) for example propose objective measures of corruption, such as the audit of

district budgets; this could reveal information about administrative malfeasance at district

level. Thirdly, self-rated happiness (which this study uses) is only one measure of sub-

jective wellbeing; Diener (1984, 1999) identifies a number of others—level of life satis-

faction, positive affect (i.e. satisfaction with past and current life) and negative affect (i.e.

anxiety and depression)—and future study may also examine the effect of decentralisation

on these. In addition, single item self-report scales used in this study may also limit the

interpretation of results and the reliability of the measures used. Stock et al. (1982), for

example, explain that single-item scales tend to be less reliable over time than multi-item

scales (although we note that the temporal reliability of single-item measures has been

moderately high).

Despite these limitations, we have been able to present a number of implications

which are valid for both subjective wellbeing and decentralisation literature, and also for

the practice of decentralisation in terms of development. Firstly, most subjective well-

being studies have tended to analyse happiness in terms of it being a personal issue

(Layard 2005; Graham 2009). This study shows that happiness is much more than that: it

is strongly associated with contextual factors such as government policies and reforms.

The results show that the degree of citizen happiness varies across districts, and is

significantly associated with local politics and local socio-economic conditions. From an

empirical perspective this study therefore suggests that it is essential to consider a range

of district social, economic, and political variables, and examine their effects on well-

being. By substantially enlarging the scope of potential questions put to respondents,

multilevel analysis provides the most appropriate basis for developing a detailed con-

textual description of how district performance can be enhanced to improve citizen

wellbeing.

Secondly, the findings support the fiscal decentralisation hypothesis, namely that fiscal

decentralisation can improve citizen wellbeing through the capacity of districts to provide

better policies and public services (Coase 1960; Oates 1972). In the context of a newly-

democratic developing country, the quality of district administration seems more important

for citizen wellbeing than the provision of local democracy and the freedoms it implies. As

we have already noted, citizens report being happier when their district authority proves

itself more capable of providing better public services. Moreover, the negative association

between district GDP and happiness presents an important contrast with previous analyses

of the macroeconomic implications of decentralisation (see for example Davoodi and Zou

1998; Iimi 2005). Our findings also suggest that the ultimate goal of decentralisation

should not necessarily be to increase economic growth, but more importantly to improve

citizen wellbeing, and to do so through the provision of better policies and services.

Whether or not increased economic growth follows depends on the capacity of the district

to first devise the best policies for its citizens and then implement (or in the case of

services, deliver) them effectively. The importance of the role of districts in improving

wellbeing in a decentralisation context is thus clearly paramount and should not be

underestimated.

Acknowledgments We would like thank to Directorate of Higher Education, Ministry of Education andCulture, Indonesia for providing funding of this study and the Indonesian Family Life Survey team forproviding the survey available. We would also like thank to Adi Pierewan, Devi Femina and the anonymousreviewers for their help in improving this manuscript. Thanks Frances Hunt editing this manuscript.


123

References

Abdel-Khalek, A. (2006). Happiness, health, and religiosity: Significant relations. Mental Health, Religionand Culture, 9(1), 85–97.

Argyle, M. (2001). The psychology of happiness. East Sussex: Routledge.Asian Development Bank. (2012). Regional development in Indonesia after decentralisation. Jakarta: Asian

Development Bank.Aspinall, E., & Mietzner, M. (2010). Problems of democratisation in Indonesia: Election, institutions, and

society. Singapore: ISEAS.Bardhan, P., & Mookherjee, D. (2000). Capture and Governance at local and national levels. American

Economic Review, 90(2), 135–139.Bjørnskov, C., Dreher, A., & Fischer, J. (2008). On decentralization and life satisfaction. Economics Letters,

99(1), 147–151.Bjørnskov, C., Dreher, A., & Fischer, J. (2010). Formal institutions and subjective wellbeing, Revisiting the

cross-country evidence. European Journal of Political Economy, 26(1), 419–430.Blanchflower, D., & Oswald, A. (2004). Wellbeing over time in Britain and the USA. Journal of Public

Economics, 88(1), 1359–1386.Blanchflower, D., & Oswald, A. (2008). Is well-being U-Shaped over the life cycle? Social Science and

Medicine, 66(8), 1733–1749.Choi, N. (2004). Local elections and party politics in Indonesia. Contemporary Southeast Asia, 26(2),

280–301.Clark, A., & Oswald, A. (1994). Unhappiness and unemployment. The Economic Journal, 104(424),

648–659.Clark, A. & Senik, C. (2011). Will GDP growth increase happiness in developing countries? IZA Dis-

cussion Paper No. 5595.Coase, R. (1960). The problem of social cost. Journal of Law and Economics, 3(1), 1–44.Davoodi, H., & Zou, H. (1998). Fiscal decentralization and economic growth: A cross-country study.

Journal of Urban Economics, 43(2), 244–257.Deaton, A. (2001). Health, inequality, and economic development. Working Paper 8318, National Bureau of

Economic Research.Deaton, A., & Zaidi, S. (2002). Guidelines for constructing consumption aggregates for welfare analysis.

Washington DC: World Bank.Di Tella, R., MacCulloch, R., & Oswald, A. (2003). The macroeconomics of happiness. Review of Eco-

nomics and Statistics, 85(4), 793–809.Diaz-Serano, L., & Rodriguez-Pose, A. (2012). Decentralization, happiness and the perception of institu-

tions. Kyklos, 65(2), 179–193.Diener, E. (1984). Subjective wellbeing. Psychological Bulletin, 95, 542–575.Diener, E., Sandvik, E., Seidlitz, L., & Diener, M. (1993). The relationship between income and subjective

wellbeing: relative of absolute? Social Indicators Research, 28(1), 195–223.Diener, E., Suh, E. M., Lucas, R. E., & Smith, H. L. (1999). Subjective wellbeing. Psychological Bulletin,

125(2), 276–302.Easterlin, R. (1974). Does economic growth improve the human alot? In P. David & M. Reder (Eds.),

Nations and households in economic growth. New York: Academic Press.Euromonitor. (2012). Income inequality rising across the globe. Paris: Euromonitor International.Foundation, Asia. (2002). First Indonesian rapid decentralisation appraisal-synopsis of findings. Wash-

ington DC: Asia Foundation.Frankenberg, E., & Jones, N. (2004). Self-rated health and mortality: does the relationship extend to a low

income setting? Journal of Health and Social Behavior, 45(4), 441–452.Frankenberg, E., & Thomas, D. (2000). The Indonesian Family Life Survey: study design and results from

wave 1 and 2. Santa Monica: RAND.Freedom House. (2009). Freedom in the world 2009. Washington, DC: Freedom House.Frey, B. S., & Stutzer, A. (2000). Happiness, Economy and Institutions. Economic Journal, 110(446),

918–938.Frey, B. S., & Stutzer, A. (2002). Happiness and Economics: How the economy and institutions affect

human wellbeing. Princeton: Princeton University Press.Graham, C. (2009). Happiness around the world. Oxford: Oxford University Press.Grindle, M. (2007). Going local: Decentralization, democratization, and the promise of good governance.

New Jersey: Princeton University Press.


123

Grootaert, C. (1999). Social capital, household welfare, and poverty in Indonesia. World Bank PolicyResearch Working Paper Number 2148. Washington, DC: World Bank.

Heller, P. (2001). Moving the state: The politics of democratic decentralization in Kerala, South Africa, andPorto Alegre. Politics and Society, 26(1), 131–163.

Helliwell, J. (2006). Wellbeing, social capital and public policy: What’s new? The Economic Journal,116(510), C34–C45.

Helliwell, J., & Huang, H. (2008). How’s your government? International evidence linking good govern-ment and wellbeing. British Journal of Political Science, 38, 595–619.

Hessami, Z. (2010). The size and composition of government spending in Europe and its impact onwellbeing. Kyklos, 63(3), 346–382.

Iimi, A. (2005). Decentralization and economic growth revisited: An empirical note. Journal of UrbanEconomics, 57(1), 449–461.

Indonesian Ministry of Finance. (2008). Sistem informasi keuangan daerah. Jakarta: Indonesia.Jorgensen, D. (2002). Did we lose the war on poverty? In D. Jorgensen (Ed.), Econometrics volume 3:

Economic growth in the information age. Cambridge, Massachusetts: The MIT Press.Kaiser, K., & Hofman, B. (2003). Decentralisation, democratic transition, local governance in Indonesia.

World Bank: Jakarta.Kaufman, D., Kray, A., & Mastruzzi, M. (2006). Measuring corruption: Myths and realities. Washington,

DC: World Bank.Kim, S., & Kim, D. (2012). Does government make people happy? Exploring new research directions for

government’s roles in Happiness. Journal of Happiness Studies, 13(1), 875–899.Krueger, A., & Schkade, D. (2008). The reliability of subjective wellbeing measure. Journal of Public

Economics, 92(8), 1–16.Lane, R. (2000). The Loss of Happiness in Market Democracies. New Heaven: Yale University Press.Layard, R. (2005). Happiness: Lessons from a new science. New York: The Penguin Press.Lim, C., & Putnam, R. (2010). Religion, social networks, and life satisfaction. American Sociological

Review, 75(6), 914–933.Litvack, J., Ahmad, J. &, Bird, R. (1998). Rethinking Decentralization in Developing Countries. Wash-

ington, D.C.: Sector Studies Series, Poverty Reduction and Economic Management, World Bank.Miller, D., Schiffer, R., Lam, S., & Rosenberg, R. (2006). Social capital and health in Indonesia. World

Development, 34(6), 1084–1098.Mullen, R. D. (2011). Decentralization, local governance, and social wellbeing in India. New York:

Routhledge.Nordholt, H., & Van-Klinken, G. (2005). New actors, new identities: Post-Suharto etnic violence in

Indonesia. In D. F. Anwar, H. Houvier, G. Smith, & R. Tol (Eds.), Violent internal conflicts in AsiaPacific: Histories, political economies and policies. Jakarta: KITLV-Jakarta.

Oates, W. (1972). Fiscal federalism. New York: Harcourt Brace Jovanovich.Oswald, A., & Wu, S. (2011). Wellbeing across America. The Review of Economics and Statistics, 93(4),

1118–1134.Pacek, A., & Radcliff, B. (2008). Assessing the welfare state: The politics of happiness. Perspective on

Politics, 6(2), 267–277.Putnam, R. (1993). Making democracy work: Civic traditions in modern Italy. Princeton: Princeton Uni-

versity Press.Rabe-Hesketh, S., & Skrondal, A. (2012). Multilevel and longitudinal modeling using Stata. Texas: Stata

Press.Radcliff, B. (2001). Politics, markets and life satisfaction: The political economy of human happiness.

American Political Science Review, 95(4), 939–956.Resosudarmo, B., & Jotzo, F. (2009). Working with nature against poverty: Development, resources, and the

environment in Eastern Indonesia. Singapore: ISEAS.Rodriguez-Pose, A. &, Maslauskaite, K. (2011). Can policy make us happier? Individual characteristics,

socio-economic factors and life satisfaction in Central and Eastern Europe. Cambridge Journal ofRegions, Economy and Society, 1–20 doi:10.1093/cjres/rsr038.

Rondinelli, D., McCoullough, J., & Johnson, R. (1989). Analyzing decentralization policies in developingcountries: A political-economy framework. Development and Change, 20(1), 5–27.

Schneider, A. (2003). Decentralization: Conceptualization and measurement. Studies in ComparativeInternational Development, 38(3), 32–56.

Seabright, P. (1996). Accountability and decentralisation in government: An incomplete contracts model.European Economic Review, 40(1), 61–89.

Shiffman, J. (2002). The construction of community participation: Village family planning groups and theIndonesian state. Social Science and Medicine, 54(8), 1199–1214.


123

http://dx.doi.org/10.1093/cjres/rsr038

Smoke, P., & Lewis, B. (1996). Fiscal decentralization in Indonesia: A new approach to an old idea. WorldDevelopment, 24(8), 1281–1299.

Snijders, T., & Bosker, R. (1999). Multilevel analysis: An introduction to basic and advanced multilevelmodeling. London: Sage.

Stock, W. A., Okun, M. A., Haring, M. J. &, Witter, R. A. (1982). Reporting reliability: A case study of lifesatisfaction research. Paper presented at the 90th Annual Convention of the American PsychologicalAssociation. Washington, DC.

Strauss, J., Beegle, K., Dwiyanto, A., Herawati, Y., Pattinasarany, D., Satriawan, E., et al. (2004). Indo-nesian living standards: Before and after the financial crisis. Santa Monica: RAND Corporation.

Thomas, D., Witoelar, F., Frankenberg, E., Sikoki, B., Strauss, J., Sumantri, C., et al. (2012). Cutting thecosts of attrition: results from the Indonesia Family Life Survey. Journal of Development Economics,98(1), 108–123.

Veenhoven, A. (1984). Conditions of happiness. Reidel, Dordrecht: Boston and Lancaster.Veenhoven, R. (2000). Freedom and happiness: A comparative study in forty-four nations in the early,

1990S. In E. Diener & E. Suh (Eds.), Culture and subjective wellbeing. Cambridge MA/London: TheMIT Press.

Welsch, H. (2008). The social cost of civil conflict: Evidence from survey of happiness. Kyklos, 61(2),320–340.

World Bank. (2006). Making services work for the poor: Nine case studies from Indonesia. INDOPOV:World Bank.

World Bank. (2008). Spending for development: Making the most of Indonesia’s new opportunities.Washington DC: World Bank.


123

Documents

Decentralisation and Citizen Happiness: A Multilevel Analysis of Self-rated Happiness in Indonesia