Social vulnerability to natural hazards

in Indonesia: driving factors and policy

implications

Tiodora Hadumaon Siagian, Purhadi

Purhadi, Suhartono Suhartono &
Hamonangan Ritonga

Natural Hazards
Journal of the International Society
for the Prevention and Mitigation of
Natural Hazards

ISSN 0921-030X

Nat Hazards
DOI 10.1007/s11069-013-0888-3

1 23
 Author's personal copy
Nat Hazards
DOI 10.1007/s11069-013-0888-3

ORIGINAL PAPER

Social vulnerability to natural hazards in Indonesia:

driving factors and policy implications

Tiodora Hadumaon Siagian • Purhadi Purhadi • Suhartono Suhartono •

Hamonangan Ritonga

Received: 22 February 2012 / Accepted: 29 September 2013

 Springer Science+Business Media Dordrecht 2013

Abstract Indonesia is located in the Pacific Ring of Fire and situated at the joining point
of four major world tectonic plates. Regions of Indonesia are highly prone to various
natural hazards such as earthquakes, tsunamis and volcanic eruptions. Some recent major
natural hazard events are the 2004 tsunami in Aceh and Nias and the 2010 Mount Merapi
volcanic eruptions in Central Java. In parallel with advancement in knowledge of the
existing hazards, the importance of social aspects of vulnerability in mitigating natural
hazards has been acknowledged by the Indonesian government. However, to date, there is
no institutionalized effort for assessing social vulnerability to natural hazards that would
cover all the districts of Indonesia. Accordingly, no comprehensive profile of social vul-
nerability is available as basis information for developing strategies to prevent larger risk
and losses and reduce vulnerability of communities in Indonesia. Only a few studies have
been conducted in Indonesia on this field. This study attempts to fill this gap by quantifying
the social vulnerability of Indonesian districts to natural hazards, determining its driving
factors and mapping its variations. The social vulnerability index (SoVI) approach is
utilized in this study. Three main driving factors affecting social vulnerability in Indonesia
are found: ‘socioeconomic status and infrastructure,’ ‘gender, age and population growth’
and ‘family structure.’ The combination of SoVI with thematic map utilizing ArcView GIS
can be used to identify districts with relative high social vulnerability level. The results can
support the prevention, mitigation, preparedness, response and recovery programs of the
impacts of natural hazards in Indonesia.

T. H. Siagian (&)  P. Purhadi  S. Suhartono

Department of Statistics, Faculty of Mathematics and Natural Sciences, Institut Teknologi Sepuluh
Nopember (ITS), Surabaya, Indonesia
e-mail: [email protected]

T. H. Siagian
Directorate for Social Resilience Statistics, BPS-Statistics Indonesia, Jakarta, Indonesia

H. Ritonga
Institute of Statistics, BPS-Statistics Indonesia, Jakarta, Indonesia

123
 Author's personal copy
Nat Hazards

Keywords Social vulnerability  Natural hazards  Factor analysis  Indonesia

1 Introduction

In the last few years, Indonesia has experienced several catastrophic hazard events, such as
the Indian Ocean tsunami that hit Aceh and Nias in 2004, the earthquake causing tsunami
in southern Java in 2006, the earthquake in Padang in 2009 and the Mount Merapi volcano
eruption in 2010. Natural hazard impact is often associated with vulnerability. The term
‘vulnerability’ is derived from the Latin word vulnerare which means ‘to be wounded.’
Vulnerability has various concepts and definitions that depend on the area of application.
Early definitions of vulnerability referred to the potential for loss when a natural hazard
occurs (Cutter 1996). Wisner et al. (2004:11) defined vulnerability as the characteristics of
a person or group in terms of their capacity to anticipate, cope with, resist and recover from
the impact of a natural hazard. Adger (2006) views vulnerability as the state of suscep-
tibility to harm from exposure to stresses associated with environmental and social change
and from the absence of capacity to adapt. Flanagan et al. (2011) consider vulnerability as
the degree to which persons or things are likely to be negatively affected. Many other
definitions of vulnerability are discussed by Birkmann (2006a) and Adger (2006).
The Hyogo Framework for Action highlights the importance of disaster prevention,
mitigation, preparedness and vulnerability reduction (UN/ISDR 2008). Informed decision
making on effective strategy and measures to reduce vulnerability can be done through
vulnerability assessment. Vulnerability assessment is increasingly being considered as an
effective guiding tool to reduce risk and losses of the impact of natural hazards and to promote
a culture of disaster resilience (Birkmann 2006a). Previously, many vulnerability assess-
ments put more emphasis on the biophysical vulnerability and physical aspects leaving the
social dimension of vulnerability poorly addressed (Cutter et al. 2003; Gall 2007; Flanagan
et al. 2011). Cutter et al. (2003) argued that social vulnerability was largely disregarded,
mainly because it is difficult to quantify. Research on social vulnerability has largely focused
on qualitative assessment methodologies rather than quantitative risk modeling (Dwyer et al.
2004). While the Indonesian government has acknowledged the importance of social vul-
nerability assessment in hazard mitigation through the Act No. 24/2007 on disaster man-
agement, only few studies on this field have been conducted in Indonesia so far.
In the aftermath of the devastating earthquake and tsunami in Aceh and Nias in 2004,
several social vulnerability assessments have been carried out in Indonesia (Post et al.
2007; Birkmann et al. 2008; Wigati 2008; Utami 2008; Hizbaron et al. 2011). However,
these assessments have limited coverage and scope. For example, Birkmann et al. (2008)
assessed socioeconomic vulnerability in Padang city, Utami (2008) measured social vul-
nerability to volcanic disaster of people living around Merapi volcano, and Hizbaron et al.
(2011) carried out social vulnerability assessment in seismic-prone areas of Bantul district.
To date, there is no systematic and coordinated effort for profiling the social vulnerability
to natural hazards throughout the country, and the efforts in preventing damage and losses
are not supported by adequate information about social vulnerability. This study attempts
to fill this gap by measuring social vulnerability to natural hazards at district level, iden-
tifying its driving factors and determining its variation among districts.
This paper is structured as follows. Following the motivation for the study, given in this
section, the next section reviews the concepts and definitions of social vulnerability, its

123
 Author's personal copy
Nat Hazards

indicators and assessments. The third section details the source of data and methodology
used in the study. The fourth section discusses the findings. The fifth section provides
recommendations for using the results to derive policy implications. A conclusion is given
in the final section.

2 Social vulnerability

Social vulnerability has various concepts and definitions. Adger (1999) views social vul-
nerability as the exposure of groups or individuals to unexpected changes and disruption to
livelihoods. Social vulnerability can be considered as a result of social inequalities and
place inequalities (Cutter et al. 2003). Cutter and Emrich (2006) defined social vulnera-
bility as the limitation of a community to the impact of natural disasters that influence its
ability or resilience in the effort to recover from the impacts. Social vulnerability is a
preexisting condition of existing communities, irrespective of type of hazards (Cutter et al.
2009). Holand et al. (2011) consider the social vulnerability as two distinct parts: socio-
economic vulnerability and built environment vulnerability.
Social vulnerability can encompass various aspects and features. Hence, Birkmann
(2006a) argued that the concept of social vulnerability should not be restricted to social
fragilities, but should also cover social inequalities regarding income, age or gender.
Societal conditions and exposure to natural hazards can influence vulnerability of com-
munities and places (Gall 2007). Hence, social vulnerability can explain why some
communities experience and suffer from a hazard event differently than others. Moreover,
social vulnerability plays a crucial role in all phases of the disaster cycle, especially in the
response and recovery phase (Solangaarachchi et al. 2012).
Key tools for identifying and measuring vulnerability are vulnerability indicators and
criteria (Birkmann 2006b). Social vulnerability assessments utilize different sets of indi-
cators because the indicators used in one setting or context may not be appropriate in
another (Rygel et al. 2006). Common individual characteristics including age, race, gender,
income, ethnicity, dwelling, employment, disabilities and level of education influence
social vulnerability (Cutter et al. 2003; Rygel et al. 2006; Holand et al. 2011). Social
vulnerability is influenced by income distribution, access to and diversity of economic
assets and informal social security (Adger 1999).
Quantification of social vulnerability is considered crucial for hazard mitigation planning
and to better understanding on disaster risk (Tate 2012; Solangaarachchi et al. 2012). One
widely known measure in quantifying the social aspects of vulnerability is social vulnera-
bility index (SoVI) approach developed by Cutter et al. (2003). It can be used to identify
location of socially vulnerable groups (Fekete 2009). Originally, SoVI was applied in the
context of the USA and constructed based on eleven factors of vulnerability (personal wealth,
age, density of buildings, the reliance on a single sector of the economy, stock and rental
housing, race African American, ethnicity Hispanic, native American tribes of Asia,
employment and infrastructure dependence), which were applied in an additive model. Later,
SoVI was applied in various vulnerability assessments worldwide (see Utami 2008; Fekete
2009; Flanagan et al. 2011; Dunno 2011; Solangaarachchi et al. 2012; Holand et al. 2011).
Nevertheless, SoVI approach can identify only variations in relative social vulnerability
(Wood et al. 2010); it cannot be used to detect the root causes of social vulnerability.
A systematic social vulnerability assessment needs a proper conceptual model which fits
the context, target and structure of the research. Such a model of vulnerability should
essentially be based upon existing data and can be updated (King and MacGregor 2000).

123
 Author's personal copy
Nat Hazards

This study utilizes the hazards of place model proposed by Cutter et al. (2003), which
describes the interaction between biophysical vulnerability (exposure) and social vulner-
ability. The interconnection between these two components forms the overall place vul-
nerability, which in turn influences the initial conditions of risk-mitigation capabilities
(Toscano 2011).

3 Study area

Indonesia, an archipelago country, is located between 6080 North and 11150 South lati-
tude and between 94450 West and 141050 East longitude (BPS 2010b). It lies between the
Asian and Australian continent. It is bounded by the South China Sea in the North and the
Pacific Ocean in the North and East, and the Indian Ocean in the South and West.
Indonesia is administratively divided into provinces, and each province is divided into
districts. As of December 2010, there were 33 provinces and 497 districts (BPS 2010a).
Total land area of Indonesia is approximately 1.9 million km2, and the coastline length is
approximately 104,000 km (BPS 2012). A map of the study area can be seen in Fig. 1.
Some selected sociodemographic characteristics of Indonesia are provided in Table 1.
According to the 2010 census, total population of Indonesia was 237.6 million people,
spread over five main islands and four archipelagos. Indonesia is the fourth most populous
country in the world after China, India and the USA. The annual population growth rate
during 2000–2010 was 1.49 percent. The elderly constitute 5.04 percent of the population,
and 9.54 percent of its population is under the age 5. Indonesia’s Human Development
Index (HDI) is in the upper-medium human development category. However, there are
significant differences in the HDI values between districts in Indonesia, and still 13.3
percent of the Indonesian population lived below the poverty line.
Indonesia has a long history of major natural disaster events. In 1883, the Krakatau
volcano erupted, setting a tsunami that hit western Java and southern Sumatra. In 2004, a
big earthquake triggered the tsunami that struck Aceh and Nias. The earthquake and
tsunami in 2004 killed around 165,708 people, and the loss it caused was assessed by over
Rp. 48 trillion (BNPB 2010b). In the last few years, the incidence of various natural
hazards has increased in Indonesia.
Indonesia is located in the Pacific Ring of Fire and situated at the meeting point of four
major world tectonic plates: the Asian plate, Indian Ocean plate, Australian plate and Pacific
Ocean plate. This location makes Indonesia highly prone to various natural hazards, such as
earthquakes, tsunamis and volcanic eruptions. Due to its topography with steep hills, Indo-
nesia is also prone to debris flows, landslides and slope failure. Data from the past two decades
indicate that there are six dominant natural hazards in Indonesia: earthquake, tsunami,
landslide/soil movement, volcanic eruption, flood and drought (BNPB 2010a).

4 Data and methodology

4.1 Data source

Based on review on the current literature on vulnerability assessment (Cutter et al. 2003;
Cutter and Emrich 2006; Rygel et al. 2006; Birkmann 2006b; Utami 2008; Cutter et al.
2009; Wood et al. 2010), thirteen variables were selected in the first indicator set. Other
potential indicators of social vulnerability, such as medical services and ‘special needs’

123
 Author's personal copy
Nat Hazards

1 Aceh 10 Kepulauan Riau 19 Nusa Tenggara Timur 28 Gorontalo

2 Sumatera Utara 11 DKI Jakarta 20 Kalimantan Barat 29 Sulawesi Barat
3 Sumatera Barat 12 Jawa Barat 21 Kalimantan Tengah 30 Maluku
4 Riau 13 Jawa Tengah 22 Kalimantan Selatan 31 Maluku Utara
5 Jambi 14 DI Yogyakarta 23 Kalimantan Timur 32 Papua Barat
6 Sumatera Selatan 15 Jawa Timur 24 Sulawesi Utara 33 Papua
7 Bengkulu 16 Banten 25 Sulawesi Tengah
8 Lampung 17 Bali 26 Sulawesi Selatan
9 Bangka Belitung 18 Nusa Tenggara Barat 27 Sulawesi Tenggara

Fig. 1 The location of the study area

Table 1 Selected socio demo-

Characteristics Value Percentage
graphic characteristics of Indo-
nesia, 2010
Population 237.6 million –
Annual growth rate 1.49 percent –
Population density 124 people/km2 –
Person over 65 12.0 million 5.04
Children under 5 22.7 million 9.54
Adult illiterate 16.8 million 7.09
Poor people 31.0 million 13.3
Human Development Index 72.27 –
Source (BPS 2012)

123
 Author's personal copy
Nat Hazards

populations, are excluded from this study due to limited data availability at district level.
After testing for multicollinearity, ten variables were retained to create a social vulnera-
bility index. They are listed in Table 2.
Data source for this study is the 2010 Survei Sosial Ekonomi Nasional (SSN) and the
2010 Sensus Penduduk (SP) provided by BPS-Statistics Indonesia (BPS). Brief description
of the data sources used in this study is provided below:
• SSN is the annual socioeconomic surveys, which collects information on education,
health and nutrition, housing condition, income/expenditure, welfare, sociocultural
activities, travelling and the like.
• SP is the population census, which collects characteristics of Indonesian population
such as gender, age, marital status, educational attainment, migration, occupation,
religion, etc. The population census is taken every 10 years, most recently in 2010.

4.2 Construction of social vulnerability index

We apply the SoVI approach to measuring social vulnerability to natural hazards at district
level in Indonesia. Construction of SoVI involves the following steps:
a. Selection of variables that represent the component of social vulnerability.
b. Standardization of the variables by converting the values to z scores. This step is
necessary because the variables are recorded in incompatible units.
c. Data screening (checking for missing values).
d. Detecting extreme multicollinearity (confounding) among the variables. In the
application of factor analysis method, the variables have to be intercorrelated, but the
correlation should not too high.
e. Performing the factor analysis method on the selected variables. Factor analysis aims
to reduce data and identify the smallest number of common or latent factors for a set of
variables (Sharma 1996). Three main processes in factor analysis method are:
• Calculation of initial factor loadings using principal component technique.
• Factor rotation using varimax rotation to minimize the complexity of the factors by
maximizing the variance of loading on each factor.
• Calculation of factor score and choosing the number of factors.

Table 2 Social vulnerability

Indicator Variables
indicators and selected variables
used in the study
Age X1 Percentage of children under 5
X2 Percentage of elderly people
Gender X3 Percentage of female
X4 Percentage of female-headed household
Income X5 Percentage of poor people
Education X6 Percentage of illiterate people
X7 Percentage of population aged 15 and
above with low education attainment
Family structure X8 Family size
Infrastructure X9 Percentage of households without electric
lighting
Population growth X10 Population growth

123
 Author's personal copy
Nat Hazards

f. Study of the output, i.e., the correlation matrix, the Kaiser–Meyer–Olkin (KMO)
measure of sampling adequacy and the Bartlett’s test statistics. The appropriateness of
factor analysis is examined by the KMO measure of sampling adequacy, and the
Bartlett’s test statistics is used to confirm that the variables are pairwise correlated
(Sharma 1996).
g. Interpretation of the common factors and determination of the direction of each
resulting factors of social vulnerability.
h. Calculation of SoVI scores
i. The SoVI scores are then classified into five groups of social vulnerability using a
classification scheme based on its standard deviation from the mean to identify the
range of vulnerability within districts of Indonesia. Scores greater than 1.5 St.Dev
indicate high social vulnerability, and scores less than -1.5 St.Dev indicate low social
vulnerability.
j. Creation of a thematic map based on the classification of the SoVI scores using
ArcView GIS to displays geographical variations in social vulnerability.

5 Results and discussion

Table 3 presents the descriptive statistics of the selected variables. Most people in all
districts have low educational attainment (primary school or less). About 50 percent of the
population was female. High standard deviation of variables X5, X6, X7 and X9 indicates
that the data are spread out over a large range of values. No values are missing in the
dataset; its sample size (the number of districts) is 497.
Factor analysis can be performed by a number of statistical packages. We utilized
MINITAB to perform factor analysis method to the ten variables to determine significant
factors that affect certain people and places to be more vulnerable than others. An
examination of the correlation matrix indicated that all variables are intercorrelated but the
correlation is not too high. The KMO measure of sampling adequacy, 0.751, indicates that
the data are appropriate for factor analysis. Sharma (1996) noted that KMO measure
greater than 0.70 is appropriate. The Bartlett’s test statistic is highly significant
(p \ 0.000), implying that the data are appropriate for factoring. The high extracted
communalities (C0.5) indicate that the extracted factors represent the selected variables
well. The eigenvalue-greater-than-one rule and the scree plot can be used to determine the
number of factors that adequately explain the correlations between the variables (Sharma
1996). Figure 2 shows that based on these rules, three factors should be retained.
By factor analysis of the ten variables, we extracted three common factors that explain
72.5 percent of the district-level variance of the relative level of social vulnerability. These
factors are: ‘socioeconomic status and infrastructure’, ‘gender, age and population growth’
and ‘family structure,’ which accounted for 26.4 percent, 23.5 percent and 22.5 percent of
the total variance explained, respectively (see Table 4).
The direction of these common factors was determined with respect to their known
influences on vulnerability, which are identified from the existing literature. Positive
directionality was given to factors that increase vulnerability and negative directionality to
factors that decrease it (Cutter et al. 2003). Various schemes of weighting of the index can
be considered: equal weight scheme (Cutter et al. 2003; Fekete 2010), pareto-ranking
scheme (Rygel et al. 2006) and weighting according to the contribution to the total vari-
ance explained (Wood et al. 2010; Solangaarachchi et al. 2012). No agreed methodology

123
 Author's personal copy
Nat Hazards

Table 3 Descriptive statistics of the selected variables and number of observations

Selected variables Mean St.Dev Min Max

Percentage of children under 5 (X1) 10.05 1.8 5.53 17.17

Percentage of elderly people (X2) 4.76 2.3 0.00 13.45
Percentage of female (X3) 49.48 1.6 43.83 54.03
Percentage of female-headed household (X4) 10.61 3.9 0.51 27.43
Percentage of poor people (X5) 15.51 9.4 1.67 49.58
Percentage of illiterate people (X6) 9.26 11.2 0.08 86.20
Percentage of population aged 15 and above 50.45 15.2 10.42 97.03
with low education attainment (X7)
Family size (X8) 4.86 0.6 3.50 7.06
Percentage of households without electric lighting (X9) 13.49 19.1 0.00 100.00
Population growth (X10) 1.99 2.1 -2.88 18.65
St.Dev standard deviation, Min minimum, Max maximum

Fig. 2 Scree plot of ten variables

exists for assigning weight in constructing the index (Nardo et al. 2005; Rygel et al. 2006).
In our case, we chose unequal weights for scoring. The SoVI scores are constructed by
weighting each factor score with its percentage of variance explained divided by the total
variance explained by all extracted common factors. In this way, factor with higher var-
iance contributes more on the SoVI scores. The SoVI scores for the district were calculated
as follows:
SoVI ¼ ð0:365  Factor 1Þ þ ð0:324  Factor 2Þ þ ð0:311  Factor 3Þ
The analysis revealed that the SoVI scores range from 1.83 (most vulnerable) to -1.24
(least vulnerable). District with the highest vulnerability is Sumba Tengah located in Nusa
Tenggara Timur province. Table 5 lists the top ten districts with highest SoVI scores. Eight
of these districts located in the eastern Indonesia except two districts that have struck by
the enormous tsunami in 2004, i.e., Nias Barat and Pidie Jaya.
Emrich (2005) compared the unweighted SoVI and the weighted SoVI in the US
metropolitan and found that the weighted SoVI did not significantly change the rank of

123
 Author's personal copy
Nat Hazards

Table 4 Vulnerability factors, the eigenvalues, the percentage of variance explained by the factor and
variables of each factor
Factor Variable Eigenvalue Percentage
variance
explained

1 Socioeconomic status Percentage of illiterate people 3.828 26.4

and infrastructure Percentage of population aged 15 and
above with low education attainment
Percentage of households without electric
lighting
Percentage of poor people
2 Gender, age and Percentage of female-headed household 2.001 23.5
population growth Percentage of female
Percentage of elderly people
Population growth
3 Family structure Family size 1.421 22.5
Percentage of children under 5 years

places. However, in our case, we found that the unequal weighted SoVI significantly
changes the rank of places of districts in Indonesia. Comparison result of the unweighted
SoVI and the unequal weighted SoVI revealed that the pattern of social vulnerability only
remains relatively constant among the top ten districts with highest SoVI scores (see
Table 6). Therefore, based on this result, we continued the analysis using the unequal
weighted SoVI.

5.1 Driving factors of social vulnerability

5.1.1 Factor 1: Socioeconomic status and infrastructure

This is the most significant factor that explains 26.4 percent of total variations in social
vulnerability. Four variables (X5, X6, X7 and X9) are loaded on this factor. These variables
are positively loaded. Socioeconomic status and infrastructure are normally associated
especially in developing countries such as Indonesia, where the provision of infrastructure
is not equally distributed and leads to discrepancy in socioeconomic status. This factor
addresses an important aspect in socioeconomic vulnerability. As the wealthy people have
better access to resources, they are less vulnerable than poor people to hazard impacts
(Fothergill and Peek 2004 cited in Cutter et al. 2009: 20). For example, poor people are
more likely to live in poor-quality housing and suffer from malnourishment (Rygel et al.
2006). Poor people with low social–economic status cannot afford to buy emergency
supplies and take longer time to recover from the impact of natural hazards compared to
people with high social–economic status, and their everyday survival exhausts all their
physical and mental resources. The four variables loaded in this factor are interlinked. In
Indonesia, education and installation of electric supply are relatively expensive. As a
result, most poor people cannot afford for high education and do not have electricity supply
at home. People with low education are likely to become poor people because they tend to
be less competitive in the job market.

123
 Author's personal copy
Nat Hazards

Table 5 Ten districts with highest in the unequal weighted SoVI scores
No Region Province District SoVI

1 Eastern Indonesia Nusa Tenggara Timur Sumba Tengah 1.83

2 Eastern Indonesia Papua Paniai 1.75
3 Eastern Indonesia Nusa Tenggara Timur Sumba Barat Daya 1.73
4 Western Indonesia Sumatera Utara Nias Barat 1.72
5 Eastern Indonesia Papua Deiyai 1.68
6 Eastern Indonesia Nusa Tenggara Barat Lombok Timur 1.65
7 Eastern Indonesia Papua Intan Jaya 1.65
8 Eastern Indonesia Nusa Tenggara Timur Sabu Raijua 1.59
9 Western Indonesia Aceh Pidie Jaya 1.53
10 Eastern Indonesia Sulawesi Tenggara Buton 1.38

Table 6 Ten districts with highest in the unweighted SoVI scores

No Region Province District SoVI

1 Eastern Indonesia Nusa Tenggara Timur Sumba Tengah 5.50

2 Eastern Indonesia Nusa Tenggara Timur Sumba Barat Daya 5.25
3 Western Indonesia Sumatera Utara Nias Barat 5.24
4 Eastern Indonesia Papua Paniai 5.13
5 Eastern Indonesia Nusa Tenggara Barat Lombok Timur 4.89
6 Western Indonesia Aceh Pidie Jaya 4.71
7 Eastern Indonesia Papua Deiyai 4.69
8 Eastern Indonesia Nusa Tenggara Timur Sabu Raijua 4.64
9 Eastern Indonesia Papua Intan Jaya 4.55
10 Western Indonesia Aceh Pidie 4.25

5.1.2 Factor 2: Gender, age and population growth

The second factor is comprised of four variables, which explain 23.5 percent of total
variations in social vulnerability. Four variables are loaded on this factor. Three variables
are positively loaded (X2, X3 and X4), and one variable is negatively loaded (X10). In
accord with the literature, women are more likely to be single parents or primary caregivers
and have lower income, fewer financial resources and less autonomy than men (Tapsell
et al. 2010; Wood et al. 2010). The elderly are vulnerable when a disaster occurs (King and
MacGregor 2000; Myers 2007). The elderly are more likely in need of physical and
economic support, and they tend to be more reluctant to be evacuated when a disaster
occurs (Rygel et al. 2006). It may be difficult to evacuate old people during disasters due to
their potential health and mobility constraint (Wood et al. 2010). As an indicator of
community instability, rapid population growth often contributes to higher social vulner-
ability (Myers 2007). Districts experiencing rapid population growth, mostly in urban
areas, may not have had enough time to adjust to the recent population increase (Cutter
et al. 2003).

123
 Author's personal copy
Nat Hazards

5.1.3 Factor 3: Family structure

The third factor is nearly as important as the other two, as it explains 22.5 percent of
variations. Two variables (X1 and X8) are positively loaded on this factor which highlights
districts with bigger family size and high percentage of children under 5. Literature stated
that large families increase social vulnerability (Cutter et al. 2009). Families with high
number of dependents, especially children, often face difficulty in finding outsource care
for their family members, and they often have limited financial resource, which makes
them difficult to recover from natural hazards’ impact. While a large family size has not
become a problem anymore in developed countries, it still remains a challenge for the
Indonesian government to deal with through its family planning program.

5.2 District-level variation in social vulnerability

Identifying districts with relative high social vulnerability level to the impact of natural
hazards is of obvious importance. By classifying the SoVI scores based on standard
deviations classification scheme, the districts are grouped into five categories. Group 1 is
labeled as highly vulnerable, group 3 was labeled as moderately vulnerable, and group 5
was labeled as lowly vulnerable. Figure 3 shows the spatial patterns of social vulnerability
at the district level. It also shows the location of the top ten districts with highest SoVI
scores.
As can be seen from Fig. 3, the majority of districts in Indonesia were classified as
having a moderate level of social vulnerability. Among 51 districts that were included in
the high level of social vulnerability, 39 of them are located in the eastern Indonesia. This
finding confirms the fact that there is uneven regional development between western
Indonesia (Java, Bali, Sumatera and Kalimantan) and eastern Indonesia (Sulawesi, Nusa
Tenggara archipelago, Maluku archipelago and Papua). Hence, it was not surprising to find
that Sumba Tengah in Nusa Tenggara Timur Province has the highest SoVI score. In 2010,
in Sumba Tengah, poor people constitutes 34.05 percent: 71.49 percent of its population
has low education, and 77.34 percent of households live without electricity. However,
caution must be taken to interpret this result. Ecological fallacy would be committed by
concluding all people living in the district with high SoVI scores have high social vul-
nerability, since the SoVI is a relative and aggregate (district-level) index in regional
assessment for comparative purposes only (Wood et al. 2010).
We assess statistically whether there are any differences between the five district
groups. An unbalanced one-way multivariate analysis of variance (MANOVA) technique
can be used for this purpose (Rencher 2002: 168). Although Wilks’ lambda is the most
common multivariate statistic used for testing the null hypothesis of equality among
group’s means, this paper used the Pillai’s criterion value due to violation on the
assumption of multivariate homogeneity of covariance. The results of the unbalanced one-
way MANOVA in Table 7 indicate a statistically significant difference at the 0.05 alpha
level between five groups based on the set of ten variables. In other words, the five groups
of districts are significantly different with respect to social vulnerability.
To obtain a better insight into the association between the variables and the level of
social vulnerability, descriptive statistics for two dominant variables within the groups are
presented in Table 8. The within-group means of variable X7 (percentage of population
aged 15 and above with low education attainment) and X9 (percentage of households
without electric lighting) increase as the level of social vulnerability increases. Similar
pattern is also found for variables X1, X3, X4, X5, X6 and X8, but not X2 and X10.

123
 Author's personal copy
Nat Hazards

Fig. 3 District-level variation in social vulnerability in Indonesia, 2010

6 Policy implications

In the past, disaster management in Indonesia was focused on emergency response (rescue
and relief) and recovery (rehabilitation and reconstruction) phase. However, the devas-
tating earthquake and tsunami in Aceh and Nias in 2004 and the earthquake in Java Island
in 2006 have raised the awareness of the importance of disaster prevention and risk
reduction. In 2007, the Indonesian government passed the Act No. 24 on Disaster Man-
agement. In 2008, Presidential Regulation Number 8 gave a legal mandate to the National
Agency for Disaster Management (BNPB) and the Local Disaster Management Agencies
(BPBD) to organize pre- and post-disaster management activities. BPBD is established at
provincial and district level. However, only 22 of 497 districts in Indonesia have so far had
BPBD (BNPB 2010b).
The Act No. 24/2007 has shifted disaster management paradigm from a responsive
orientation to preventive orientation. In the National Disaster Management Plan
2010–2014, disaster reduction has been put as one of its important policies. At present,
only a few disaster risk reduction programs are planned and programmed (BNPB 2010b).
Taking this situation into account, the results of this study can be used to develop further
risk reduction programs.

6.1 Implications on vulnerability and risk mapping

The Act No. 24/2007 states that BNPB and BPBD have responsibility for compiling,
identifying and informing vulnerability maps. The SoVI map can support their duties in

123
 Author's personal copy
Nat Hazards

Table 7 Result of the unbalanced one-way MANOVA

Effect Value F Hypothesis Df Error Df Sig.

Five groups 1.131 19.170 40.000 1,944.000 0.000

Pillai’s trace

Table 8 Descriptive statistics

Variable Group of social Mean St.Dev N
vulnerability

Percentage of population aged 15 and above Group 1 62.7749 12.85083 51

with low education attainment (X7) Group 2 54.7326 14.21136 76
Group 3 50.1580 14.34072 193
Group 4 46.1946 14.39084 161
Group 5 37.0300 15.19067 16
Total 50.4457 15.18977 497
Percentage of households without electric Group 1 40.8194 27.93197 51
lighting (X9) Group 2 22.9878 25.10085 76
Group 3 8.2242 11.63709 193
Group 4 7.7409 8.28386 161
Group 5 2.7719 1.63676 16
Total 13.4945 19.09267 497

compiling data and analysis on vulnerability. Moreover, the evidence discussed in the
previous section indicates that social vulnerability level varies among the districts, which
means the capacity for preparedness and response is uneven between districts. When
resources are limited, the SoVI map can be used to prioritize those districts with relatively
high-level social vulnerability. By doing so, resources can be used more effectively to
reduce the preexisting vulnerability.

6.2 Implications on early warning system

Vulnerability maps may be good for communication tool; they can motivate people to take
action (Fekete 2011). Besides strong technical foundation, effective early warning system
requires good knowledge of the risks. Communities often do not understand their risk. The
SoVI map provides information on districts with relatively high social vulnerability. By
integrating the SoVI map in the early warning system, communities who live in those
districts will be aware of their risks and are encouraged to conduct actions. Consequently,
it will enhance communities’ preparedness for natural hazards.

6.3 Implications on integration programs

The economic and social impact of disasters can be reduced more effectively by integrating
the disaster risk reduction programs with other development programs. This study has
found three main driving factors affecting social vulnerability: ‘socioeconomic status and
infrastructure’, ‘gender, age and population growth’ and ‘family structure.’ Hence, it is
suggested that the mitigation, preparedness, response and recovery programs have to be

123
 Author's personal copy
Nat Hazards

integrated with other programs on infrastructure development, poverty reduction, gender

and family planning.

7 Conclusions

This study is considered as an initial effort for supporting disaster prevention, mitigation,
preparedness and vulnerability reduction in Indonesia. The SoVI approach is utilized to
quantify social vulnerability of Indonesian districts to natural hazards and to determine its
driving factors. The results have shown that social vulnerability is varied significantly
among districts in Indonesia. Three driving factors affecting social vulnerability have been
identified, i.e., ‘socioeconomic status and infrastructure’, ‘gender, age and population
growth’ and ‘family structure.’
This study only uses ten variables to assess social vulnerability to natural hazards due to
constraint in the availability of data. By incorporating some more potential variables,
driving factors to social vulnerability may be better captured and understood. Finally, it is
important to return to the point that this study applied SoVI approach at district level.
Using this approach to assess Indonesian subdistricts’ social vulnerability to natural haz-
ards will be a further challenge.

Acknowledgments The authors are very grateful to two anonymous reviewers for valuable comments on
an earlier version of the paper. We also would like to thank Nicholas Longford and Neysa Setiadi whose
constructive comments, suggestions and corrections greatly improved the quality of this paper.

References

Adger WN (1999) Social vulnerability to climate change and extremes in coastal Vietnam. World Dev
27(2):249–269
Adger WN (2006) Vulnerability. Glob Environ Chang 16:268–281
Birkmann J (2006a) Measuring vulnerability to promote disaster-resilience societies: Conceptual frame-
works and definitions. In: Birkmann J (ed) Measuring vulnerability to natural hazards: towards disaster
resilience societies. United Nations University Press, Tokyo, pp 9–54
Birkmann J (2006b) Indicators and criteria for measuring vulnerability: Theoretical bases and requirements.
In: Birkmann J (ed) Measuring vulnerability to natural hazards: towards disaster resilience societies.
United Nations University Press, Tokyo, pp 55–77
Birkmann J, Setiadi NJ, Gebert N (2008) Socio-economic vulnerability at the local level in context of
tsunami early warning and evacuation planning in the city of Padang, West Sumatera. Paper for
International Conference on Tsunami Warning, November 12–14, 2008, Nusa Dua, Indonesia
BNPB (2010a) National Action Plan for Disaster Risk Reduction (NAP-DRR) 2010-2012 Available from
http://www.bnpb.go.id/website/file/pubnew/99.pdf. Accessed 23 Feb 2012
BNPB (2010b) National Disaster Management Plan 2010-2014. http://www.bnpb.go.id/website/file/pubnew/
100.pdf. Accessed 23 Sept 2012
BPS (2010a) Data strategies BPS. BPS-Statistics Indonesia, Indonesia
BPS (2010b) Statistical year book of Indonesia 2010. BPS-Statistics Indonesia, Indonesia
BPS (2012) Trends of the selected socio-economic indicators of Indonesia. BPS-Statistics Indonesia,
Indonesia
Cutter SL (1996) Vulnerability to environmental hazards. Prog Hum Geogr 20(4):529–539
Cutter SL, Emrich CT (2006) Moral hazard, social catastrophe: the changing face of vulnerability along the
hurricane coast. Ann Am Acad Polit 604:102–112
Cutter SL, Boruff BJ, Shirley WL (2003) Social vulnerability to environmental hazards. Soc Sci Q
84(2):242–261
Cutter SL, Emrich CT, Webb JJ, Morath D (2009) Social vulnerability to climate change variability hazards:
a review of the literature. Final Report to Oxfam America. http://adapt.oxfamamerica.org/resources/
Literature_Review.pdf. Accessed 1 Oct 2010

123
 Author's personal copy
Nat Hazards

Dunno CH (2011) Measuring social vulnerability to natural hazards: an examination of the United States
Virgin Islands. Unpublished PhD Dissertation, the University of South Carolina at Greensboro
Dwyer A, Zoppou C, Nielsen O, Day S, Robert S (2004) Quantifying social vulnerability: a methodology for
identifying those at risk to natural hazards. Geoscience Australia Record, vol 2004/014. Geoscience
Australia, Canberra
Emrich CT (2005) Social vulnerability in US metropolitan areas: improvements in hazard vulnerability
assessment. Dissertation at the Department of Geography, University of South Carolina
Fekete A (2009) Validation of a social vulnerability index in the context to river-floods in Germany. Nat
Hazards Earth Syst Sci 9:393–403
Fekete A (2010) Assessment of social vulnerability for river-floods in Germany. University of Bonn,
Dissertation
Fekete A (2011) Spatial disaster vulnerability and risk assessments: challenges in their quality and
acceptance. doi:10.1007/s11069-011-9973-7
Flanagan BE, Gregory EW, Hallisey EJ, Heitgerd JL, Lewis B (2011) A social vulnerability index for
disaster management. J Homel Secur Emerg Manag 8(1):1–22
Gall M (2007) Indices of social vulnerability to natural hazards: a comparative evaluation. Dissertation at
the Department of Geography, University of South Carolina
Hizbaron DR, Baiquni M, Sartohadi J, Rijanta R, Coy M (2011) Assessing social vulnerability to seismic
hazard through spatial multi criteria evaluation in Bantul District, Indonesia. http://www.tropentag.de/
2011/abstracts/full/869.pdf. Accessed 25 May 2012
Holand IS, Lujala P, Rød JK (2011) Social vulnerability assessment for Norway: a quantitative approach.
Nor J Geogr 65(1):1–17
King D, MacGregor C (2000) Using social indicators to measure community vulnerability to natural
hazards. Aust J Emerg Manag 15(3):52–57
Myers CA (2007) Population change and social vulnerability in the wake of disaster: the case of Hurricanes
Katrina and Rita. Unpublished MA Thesis at the Department Sociology. Louisiana Tech University
Nardo M, Saisana M, Saltelli A, Tarantola S, Hoffman A, Giovannini E (2005) Hand book on constructing
composite indicators: methodology and user guide. OECD, Paris
Post J, Zosseder K, Strunz G, Birkmann J, Gebert N, Setiadi N, Anwar HZ, Harjono H, Siagian T (2007)
Risk and Vulnerability assessment to tsunami and coastal hazards in Indonesia: conceptual framework
and indicator development. http://hydroteam.de/Padang_Post_Zosseder_et%20al.pdf. Accessed 11 Jun
2012
Rencher AC (2002) Methods of multivariate analysis, 2nd edn. Wiley, USA
Rygel L, O’Sullivan D, Yarnal B (2006) A method for constructing a social vulnerability index: an
application to hurricane storm surges in a developed country. Mitig Adapt Strat Glob Chang
11(3):741–764
Sharma S (1996) Applied multivariate techniques. Wiley, Canada
Solangaarachchi D, Griffin AL, Doherty MD (2012) Social vulnerability in the context of bushfire risk at the
urban-bush interface in Sydney: a case study of the Blue Mountains and Ku-ring-gai local council
areas. doi:10.1007/s11069-0120334-y
Tapsell S, McCarthy S, Faulkner H, Alexander M (2010) Social vulnerability and natural hazards. CapHaz-
Net WP4 Report, Flood Hazard Research Centre-FHRC, Middlesex University, London. Available at
http://CapHaz-net.org/outcomes-results/CapHaz-Net_WP4_Social_Vulnerability.pdf. Accessed 12 Jan
2012
Tate E (2012) Uncertainty analysis for a social vulnerability index. Ann Assoc Am Geogr. doi:10.1080/
00045608.2012.700616
Toscano MB (2011) Social vulnerability in Louisiana’s upper industrial corridor: spatial distribution and
linkages with cumulative pollution-a ZIP code level analysis. Unpublished MSc Thesis, Louisiana
State University and A & M College
UN/ISDR (2008) Indicators of progress: guidance on measuring the reduction of disaster risks and the
implementation of the Hyogo framework for action. United Nations secretariat of the International
Strategy for Disaster Reduction (UN/ISDR), Geneva, Switzerland
Utami P (2008) Measuring social vulnerability in volcanic hazards: the case study of Merapi volcano.
Unpublished MSc Thesis, University of Bristol, UK
Wigati M (2008) Improving flood hazard and vulnerability assessment based on social assessment in
Bogowonto River. Unpublished MSc Thesis. Gajah Mada University
Wisner B, Blaikie P, Cannon T, Davis I (2004) At Risk: Natural hazards, people’s Vulnerability and
Disasters, 2nd edn. Routledge, London
Wood NJ, Burton CG, Cutter SL (2010) Community variations in social vulnerability to Cascadia-related
tsunamis in the US. Pacific Northwest. Nat Hazards 52:369–389

123