![Jogoyudan Location and Mt Merapi and Surrounds (modified from Voight et al. [50]).](https://www.researchgate.net/profile/Aaron-Opdyke/publication/374904835/figure/fig2/AS:11431281216471748@1704750401976/Jogoyudan-Location-and-Mt-Merapi-and-Surrounds-modified-from-Voight-et-al-50_Q320.jpg)
Content uploaded by Aaron Opdyke
Author content
All content in this area was uploaded by Aaron Opdyke on Jan 08, 2024
Content may be subject to copyright.
International Journal of Disaster Risk Reduction 100 (2024) 104076
Available online 21 October 2023
2212-4209/© 2023 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/).
Contents lists available at ScienceDirect
International Journal of Disaster Risk Reduction
journal homepage: www.elsevier.com/locate/ijdrr
The long-term impact of humanitarian housing interventions
following the 2010 Merapi eruption
Tatiana Skwarko a,Ivy He a,Sarah Cross a,Aaron Opdyke a,*,Tantri Handayani b,
Jim Kendall c,Andreas Hapsoro d,Gregg McDonald c,Yunita Idris e,f
aSchool of Civil Engineering,The University of Sydney,Sydney,NSW,2006,Australia
bDepartment of Civil and Environmental Engineering,Universitas Gadjah Mada,Yogyakarta,55284,Indonesia
cHabitat for Humanity International,Asia-Pacific Area Office,Makati City,1229,Philippines
dHabitat for Humanity Indonesia,National Office,Jakarta,11480,Indonesia
eTsunami and Disaster Mitigation Research Center (TDMRC), Universitas Syiah Kuala,Banda Aceh,23111,Indonesia
fDepartment of Civil Engineering,Universitas Syiah Kuala,Banda Aceh,Aceh,23111,Indonesia
ARTICLE INFO
Keywords:
Humanitarian
Shelter
Housing quality
Indonesia
Disasters
ABSTRACT
There is strong evidence of the short-term impact of humanitarian interventions after disasters,
however,relatively less is known about what,if any,impact this support has on long-term recov-
ery outcomes in communities.This research examined long-term housing outcomes following as-
sistance provided after the 2010 Merapi eruption in Yogyakarta,Indonesia.We surveyed 316
households who received and did not receive housing assistance following damaging lahar flows
in the community of Jogoyudan to assess housing quality through a multi-dimensional measure.
Ordinary Least Squares (OLS)regression was used to evaluate the relationship of humanitarian
assistance on housing quality outcomes,controlling for respondent and household demographics
as well as the impacts of the disaster.Findings indicate that post-disaster housing assistance was
positively correlated with higher long-term housing quality.Land tenure was found to be the
strongest predictor of housing quality,explaining nearly a fifth of variance in housing quality.
Livelihoods and construction abilities were also found to be significant predictors of long-term
housing quality.Our results demonstrate the value of post-disaster housing programs in raising
the living standard of recovering communities and the institutional,economic,and knowledge
systems that support housing markets.
1.Introduction
There is a paucity of research on long-term impacts of humanitarian interventions implemented after disasters [1]–a part of
broader gaps over evidence on the effectiveness of humanitarian assistance [2]. Within the shelter and housing sector,there is a grow-
ing recognition of the need to close the humanitarian-development gap to make better use of increasingly scarce funding amidst deep-
ening need [3]. Recent criticisms have drawn to light issues with traditional conceptualisation of disaster cycles [4]and there is a
need to refocus on root causes of vulnerability [5].
While there is no shortage of studies that focus on emerging challenges in post-disaster housing [6,7], practitioners continue to
call for greater study of the long term trajectories of housing interventions implemented to support communities after disasters [8].
Recent efforts have started to respond to this need [9], but there remains a need to build knowledge of what becomes of housing solu-
*Corresponding author.
E-mail address:aaron.opdyke@sydney.edu.au (A.Opdyke).
https://doi.org/10.1016/j.ijdrr.2023.104076
Received 7 June 2023;Received in revised form 3 October 2023;Accepted 20 October 2023
International Journal of Disaster Risk Reduction 100 (2024) 104076
2
T.Skwarko et al.
tions in recovery.Building stronger evidence of housing interventions will allow for more effective assistance to vulnerable communi-
ties.
This research aimed to answer the question:How did the long-term housing quality compare between households that re-
ceived post-disaster shelter assistance and those that did not after the 2010 Merapi eruption in Indonesia? By developing an
understanding of the impact of housing assistance on household recovery post-disaster,this research aims to contribute to the devel-
opment of an economically viable,autonomous,and locally driven aid approach to recovery.Aligning with the Sendai Framework for
Disaster Risk Reduction,this approach will encourage the prioritisation of contextually appropriate reconstruction that is safer and
more resilient to disaster risk in the long-term.
2.Background
There is currently a dearth of high-quality evidence linking humanitarian interventions to desired long-term outcomes [9–11].
Barakat [12]notes that whilst the built environment in high income countries can be impacted during disasters,the destruction in
lower income countries is amplified.This destruction disproportionately impacts housing,with it often representing the largest share
of economic loss [13]. Aside from constituting a basic human right,housing is a key component of the social and economic fabric of
settlements [14]. This is pertinent when considering low-and middle-income countries,where the home is not only the social centre
for family and friends [12], but is also often the workplace [15], and therefore linked to livelihood generation [16].Barakat and Zyck
[17]note that ‘the numerous secondary impacts of housing destruction,compensation,and reconstruction lead one to wonder which elements
of society,culture,the economy and politics are not influenced by housing’.The loss or destruction of housing therefore has broad reaching
implications for recovery,leading affected people to often prioritise housing reconstruction in the wake of a disaster [18].
Whilst a disaster can represent an opportunity for broader development and improvements in safety [19], reconstruction too often
reproduces the same pre-disaster vulnerabilities [5,13]. The link is therefore clear between ineffective housing recovery and overall
risk.Reconstruction and development can contribute to a perpetuating cycle of vulnerability,where short term societal or economic
pressures are exacerbated.The result is an acceptance of long-term disaster risk [20], for example,where vulnerable groups have no
choice but to live in cheap but unsafe settlement locations [5]. This in turn results in increased vulnerability,which eventuates in dis-
asters [21]. There is therefore a need to ensure housing reconstruction by affected populations following disasters is supported by ex-
ternal authorities to contribute to meaningful disaster risk reduction for the vulnerable.
2.1.Models of housing reconstruction delivery
Approaches to post-disaster recovery can be broadly categorised as ‘top-down’or ‘bottom-up’.The traditional models are gener-
ally built around the former,where an externally driven reconstruction program is financed by donors,and carried out by large con-
tractors [22]. This model,herein called ‘donor-driven’,is popular due to the speed and theorised efficiency of reconstruction efforts,
however this is achieved through a one-size-fits-all approach.The ratio of decision making responsibility weighs heavily in favour of
external authorities,as shown in Fig.1.The main criticisms of this model are that it often disregards the actual needs of affected pop-
ulations in favour of donor or implementing agency priorities [23,24], fails to recognise the diversity and skill sets of communities,
and reverts to a paternal approach to relief [25]. Another criticism is that risk management processes from aid agencies often results
in direction of funding towards land-tenured owners,disadvantaging renters or informal settlements,and often resulting in those
most vulnerable being left unassisted [26,27]. These traditional recovery models often overemphasise relief activities,and regularly
focus more on the symptoms of disasters than on tackling the causes [21]. The complexity of not only vulnerability,but also of the
networks that make up affected communities,necessitates a more multi-dimensional perspective of recovery [28].
Bottom-up approaches place the agency of reconstruction efforts with impacted communities and focuses on active participation
throughout the process.This approach is characterised by the use of locally available materials,labour and technical knowledge
[29,30]. It can be financially,materially or technically assisted by donors and implementing organisations,with the key difference
compared with the donor-driven model being that design agency ultimately lies with households [29,31]. Within the owner-driven
model,the level of assistance can vary from community-led approaches supported by external assistance,down to fully autonomous
recovery.However,the key distinction on this continuum,shown in Fig.1,is the level of choice or decision-making of households
[32]. For the purpose of this research,models where affected people have primary agency will be defined as self-recovery [33].
Fig.1.Choice continuum in post-disaster housing assistance.
International Journal of Disaster Risk Reduction 100 (2024) 104076
3
T.Skwarko et al.
2.2.Improving housing assistance
There are multiple reasons for a transition away from predominantly donor-driven models towards a greater focus on self-
recovery,the primary being the inevitability of the process.Despite housing assistance often comprising some of the highest expendi-
tures in recovery efforts [13], formal international humanitarian assistance rarely reaches more than 30 %of an affected population,
and more commonly reaches less than 10 % [25,34]. These statistics are based on previously studied humanitarian responses,and do
not consider the increasing scale of humanitarian needs.The combination of rapidly expanding vulnerable populations with growing
severity of natural hazards attributed to climate change is increasingly placing rebuilding capacity beyond the limits of the aid com-
munity [35]. Whether humanitarian approaches embrace self-recovery or not,the reality is that the vast majority of impacted people
are left to re-build autonomously [36]. Self-recovery recognises that survivors are not passive,and will begin rebuilding long before
formal aid is able to reach impacted communities [37].
However,the literature suggests that the inevitability of self-recovery as process should not be viewed negatively.Many studies
[13,18,28,29,38–41]stridently support self-recovery,noting that disaster survivors should be the primary actors in any reconstruc-
tion activity.Past housing assistance in Sri Lanka [13], India [33], and others [25,29,42,43]found that self-recovery of housing re-
sulted in higher completion and occupancy rates,resulted in earlier completion of construction,and increased pride,dignity and sat-
isfaction amongst households.
The economic argument for self-recovery has not fundamentally changed since Davis [40]stated that ‘the cost of temporary donor
provisions using western technology may be higher than that of permanent housing’.This is primarily due to the reliance on expensive im-
ported materials,technology,labour,combined with transport costs [36]. Conversely,a self-recovery housing model relies on local or
vernacular technology and materials,building on resources,labour and knowledge already available within the community [21].
Studies show [13,17,38,43]that this results in more houses being rebuilt,and more households receiving assistance.The greater
reach does not necessarily imply lower quality results.When studying beneficiaries of housing rebuilding grants in India after the
2004 Tsunami,Andrew et al. [43]found that despite the lower monetary assistance received,beneficiaries'homes were generally
larger and more culturally appropriate than those in provisioned housing.
Additionally,when considering the broader implications for recovery beyond initial housing reconstruction,a self-recovery ap-
proach has been shown to have significant impacts on the long-term risk reduction of households and communities.The active partic-
ipation of communities allows for a transferral of knowledge,which can result in replicability of the construction process beyond the
initial project [42,44]. This transferral of knowledge contributes to the building of local capacity beyond the initial recovery period,
which can increase livelihood opportunities and income security,a key factor in reducing socio-economic vulnerability.Finally,the
self-recovery approach has also been found to increase beneficiaries’perception of their own recovery to the current disaster [43],
and their ability to recover in the future.The more holistic approach of the self-recovery process has been shown better able to ad-
dress disaster risk than other models,by looking beyond the structural recovery of a community.This leaves impacted groups better
placed to recover from future disasters.However,without with the right support,self-recovery can result in the same or increased vul-
nerability as pre-disaster [17,43].
2.3.Housing quality in recovery
The mantra of ‘Build Back Better’(BBB)often captures a desire to move towards more meaningful risk management in responses
[13,45]. However,the meaning of ‘better’remains contested as a result of poor operationalisation [19]. Furthermore,the attainment
of structural safety in recovery is often entangled in the affordability and practicality of such solutions [45]. Flinn [19]draws atten-
tion to the challenges of achieving this at scale and need to think about ‘better’as more than just building code-defined measures of
safety.Much earlier work from Goodman [69]appropriately notes that the critical goal of post-disaster reconstruction is to achieve a
‘decent home and a suitable living environment’.Broadly,a‘quality’house could be described as meeting the physical and psycho-
logical needs of the occupants [46]. However,both these needs are relative to the geographic,cultural,demographic,or economic
context it is being applied to [47]. Even within the same household,the definition of housing quality can vary with time.Definitions
need to acknowledge that a house is more than shelter –it is a ‘home’[48]. This recognises the full spectrum of health,livelihood,and
well-being that housing can provide,and its emotional significance to a family,and to the community.The ability to assess housing
recovery against a wider range of indicators is necessary and a persistent gap in literature [18].
3.Methods
This research aimed to identify whether households who received post-disaster housing assistance displayed materially different
long-term recovery outcomes.To date,there have been limited examples of studies showing the long-term impacts of housing assis-
tance [9,49]. This research examines recovery in the context of the 2010 Merapi eruption and subsequent lahar flows in Yogyakarta,
Indonesia.
3.1.Case context:2010 Merapi eruption in Indonesia
Mt Merapi,located in Central Java about 30 km from Yogyakarta,is the country's most active volcano [51]. The regular cycle of
eruptions is characterized by pyroclastic flows and destructive lahars –a rapidly flowing mixture of volcanic debris,ash and water.
Despite this hazard,the fertile nature of the land flanking Mt Merapi draws large populations of Javanese dependent on its natural re-
sources.As a result,it is also one of the most densely populated volcanoes in the world [52], with about 1.6 million people living in its
immediate surrounds,shown in Fig.2.
International Journal of Disaster Risk Reduction 100 (2024) 104076
4
T.Skwarko et al.
Fig.2.Jogoyudan Location and Mt Merapi and Surrounds (modified from Voight et al. [50]).
In October 2010,following a period of reduced activity,Mt Merapi produced its largest and most explosive eruption in over a cen-
tury,resulting in 398 casualties [53]. After the initial eruption,heavy rains remobilised the ash and pyroclastic deposits,resulting in
frequent and destructive lahar flows lasting into 2011 [52]. The combined impacts of the initial eruption and subsequent lahars re-
sulted in the damage or destruction of almost 4,000 homes.It is estimated that more than 1 million were displaced from their homes
[54].
3.2.Community and project selection
The target community selected for this research,Kampung Jogoyudan,is a riverside community within Yogyakarta,shown in Fig.3
and Fig.4.The term kampung is the Indonesian term for a ‘village’.Urban kampungs are often characterised by high density settle-
ments,located in hazard-prone areas,and with limited access to basic infrastructure [55]. Yogyakarta is divided into fourteen district-
level subdivisions called kemantren.The population density of Kemantren Jetis,the district in which Jogoyudan is located,was nearly
four times the average density across Yogyakarta,based on the 2020 National Census, [56].
In the 2010 Merapi disaster,the lower region of Jogoyudan sustained significant damage due to the destructive lahar flows in the
wake of the eruption.Houses in the area were submerged in up to 1 m of lahar sedimentation [57], shown in Fig.5.However,large
areas of the community,located away from the riverbed,were left untouched.For the purpose of this research,the households within
the Jogoyudan community were categorised first based on the level of damage sustained during the disaster,as classified in Table 1,
and second based on assistance they received following the disaster.This created four distinct sub-groups of interest: (1)damaged and
assisted, (2)damaged and unassisted, (3)undamaged and assisted,and (4)undamaged and unassisted.
Following the 2010 Merapi eruption,Habitat for Humanity Indonesia assisted with the recovery effort in Jogoyudan through the
rehabilitation or rebuilding of 165 houses,as well as the provision of 12 public water facilities with the community implementing
subsequent household connections.Design for rebuilt houses was supported by Habitat for Humanity,and approximately 50 %of the
labour was sourced from within the community,which had the added benefit of developing local construction skills.They also as-
sisted with construction of an early childhood development centre and five drainage points.More broadly,they provided community
training sessions in construction,disaster risk reduction,and household financial management.No houses received financial assis-
tance from Habitat for Humanity Indonesia.The 165 households who were recipients of direct housing assistance formed the ‘assisted
recovery’target group of interest in this study and the remainder formed the control group.
3.2.1.Survey sampling strategy
Population estimates based on administrative levels in Indonesia,combined with previous Habitat for Humanity Indonesia knowl-
edge,were used to determine sample sizes.The relevant administrative levels were:
•Rukun Warga or Community Unit (RW): 7 RWs in Jogoyudan;
•Rukun Tettangga or Neighbourhood Unit (RT): approximately 40–50 families per RT,and four RTs per RW.
Using an average of 45 families per RT,an estimated population of 1260 families resided in Jogoyudan.The majority of the RTs
were similar in population density,demographic,and socioeconomic level,and were considered part of the target group for this
study.However,two RTs predominantly consisted of apartment blocks constructed by the Indonesian government prior to the 2010
eruption which were excluded from the analysis.Additionally,there were five RTs which predominantly consisted of hotels and
larger businesses,which would indicate a higher socioeconomic background than the rest of the community.For this reason,they
International Journal of Disaster Risk Reduction 100 (2024) 104076
5
T.Skwarko et al.
Fig.3.Map of Jogoyudan and lahar impacts.
were excluded from analysis.20 RTs were thus identified as the population of interest,or an estimated 900 families.Once the target
group of 165 households was removed,the remaining households was considered as the control group comprising 735 households.
Using a 95 %confidence level,allowing for a 6 %margin of error and 10 %non-response rate,we targeted 206 households who
did not receive housing assistance and 107 household who did receive housing assistance.The 6 %margin of error was used based on
a 10 %relative margin of error in previous field studies using the survey tool in other national contexts.
3.3.Data collection
Data was collected using a Housing Quality Assessment (HQA)tool developed by Habitat for Humanity International to assess the
quality of house improvements funded by micro-finance organisations who received consulting services from Habitat's Terwilliger
Center for Innovation in Shelter (TCIS). The HQA tool seeks to measure the quality of physical structures,workmanship,structural de-
International Journal of Disaster Risk Reduction 100 (2024) 104076
6
T.Skwarko et al.
Fig.4.Kampung Jogoyudan,a riverside urban village in Yogyakarta after the 2010 Merapi Disaster.Image Credit:Habitat for Humanity Indonesia,November 2010.
Fig.5.Destructive lahar flows during the 2010 Merapi disaster.Image Credit:Habitat for Humanity Indonesia,November 2010.
Table 1
Damage scales for assessment of housing impact.
Level Damage Description
0 None Not Impacted
1 No Visible Damage No permanent visible structural damage to structure.May present temporary cosmetic impacts,such as silt on floors.
2 Minor Minor cracks in walls/columns or cladding/plaster.
3 Major Large cracks in walls/columns or cladding/plaster.May show signs of roof damage due to supporting structure movement.
4 Extensive Failure of walls or columns.May be signs of partial roof collapse in portions of structure.
5 Fully Destroyed Total or near total collapse of building.
sign against relevant building standards,disaster resilience,and functionality of a home.It was initially developed in 2016 for use in
Sri Lanka,and a new edition was revised in 2019 for use in Nepal.Under this research,the research team refined the tool for use in the
Indonesian context.By validating and streamlining the survey,it was made applicable to a broader range of contexts and projects.By
extending its use and facility of administration,this research aimed to increase its usage as a post-implementation tool and show how
it can be used to carry out impact evaluations to inform programming strategies.The survey was administered by civil engineering
students from Universitas Gadjah Mada,with questions asked in Indonesian.
3.3.1.Survey development
The HQA tool was designed in KoboToolbox,a survey platform used extensively in humanitarian assistance.The KoboCollect app
allowed the survey to be administered by enumerators on smartphones.Surveys captured extensive data on the design and construc-
tion of existing structures,including observations of both structural,such as material quality and connections,and non-structural
characteristics,such as ventilation and cladding.The structural questions were sectioned based on the predominant structural typol-
International Journal of Disaster Risk Reduction 100 (2024) 104076
7
T.Skwarko et al.
ogy of the building and roof,such that material specific quality indicators can be assessed.This allowed for assessment of the follow-
ing typologies:
•load bearing masonry;
•reinforced concrete frame (columns,beams,precast concrete);
•light frame (bamboo,timber,steel)
The survey also captured information on the quality the roof structure,floor and roof coverings.Overall,this data was used to as-
sess the change in technical safety and quality of the housing in the intervening period since the disaster.Questions were included to
identify household size,access to water,sanitation and cooking facilities,privacy and security.Finally,respondents were asked about
their quality of life,their own perceptions of the safety of their house,and their vulnerability to future disasters.This multidimen-
sional assessment aimed to move past structural safety as a principal indicator of housing quality,and allow for wider outcomes [58].
In addition to the data described above,questions were asked about access and types of assistance,motivations for construction,
and perceptions of recovery support.These questions were open ended and designed to support findings of relationships between as-
sistance types and overall recovery outcomes.They also provided insight into participants'individual recovery experiences.Open-
ended responses were captured using enumerator field notes.The full list of survey questions can be accessed in attached supplemen-
tary material.
All potential candidates were advised that participation in the survey was voluntary,and that they could opt out at any time with-
out negative repercussions.Information was provided at the beginning of the survey that outlined the aims,risks and benefits of the
study,and consent was clearly sought and recorded before commencing.Participants were advised that the survey was discussing the
Merapi Disaster in 2010,and that recollections of the event may be trigger negative emotions.They were further informed that their
data would be anonymized and responses would be kept confidential to protect their privacy.
3.4.Data analysis
Survey data was first converted into a measurable housing quality score,which was used as the dependent variable for analysis.
Ordinary Least Squares (OLS)regression was then used to evaluate the impact of assistance on housing quality using statistical confi-
dence.Demographic indicators were used as controls.Three separate models were developed –the first for the entire surveyed popu-
lation,Model 1,and then two separate models for damaged and undamaged populations,Model 2 and Model 3.
3.4.1.Housing quality indicator
The structural quality of buildings was assessed based on different building attributes.The survey logic was designed so that an
initial triage question was used to filter the subsequent questions based on the primary structural systems that included (1)load bear-
ing masonry, (2)reinforced concrete frames,and (3)light timber framing.Attributes were assessed for compliance with the Building
Code of Indonesia (such as column spacing)or general best construction practice.Additional indicators were included in the overall
housing quality score included access to electricity,water supply,and sanitation.
Most questions were binary yes or no,where the desirable response was scored 1.For example,when assessing the structural sys-
tem,we asked and observed,“Is there a continuous foundation under all load-bearing walls?”Some questions were scaled,such as “Is the
water source regularly tested? and “Does the testing show that the water is clean and safe to drink?”Responses in this example were ‘never’
(0), rarely (0.25), sometimes (0.75), and always (1). These follow up questions were used to provide more detail on the response,with
an example being the quality of the water source.A summary of all housing quality categories and number qualities assessed is shown
in Table 2.Each category was equally weighted.For the structural system,roof structure,and roof cover if multiple construction ty-
pologies were present,the minimum score of the assessed sub-categories was taken.This was considered appropriate from a safety
perspective.Each category score was then translated to a value out of 1,giving each house a maximum possible score of 7.As the
housing quality metric could vary across studies or contexts,housing quality scores were standardised in analysis.
3.4.2.Ordinary least squares regression
The dependent variable being investigated was housing quality.Additional control variables in the model captured individual char-
acteristics of respondents that included gender and age.Household characteristics included family status,household size,household
structure,income,employment,and land tenure.The impacts of the disaster were captured by type of reconstruction activity and
construction abilities.The relationship between housing assistance and housing quality was identified using ordinary least squares re-
gression.The general linear regression model is given as
Y=𝛽1x1+𝛽2x2+𝛽3x3+𝛽4x4+𝜖
where Y =standardised housing quality,and x1,x2,x3,x4=household demographics,assistance,damage,and reconstruction group-
ings as defined above.The relative impact of each variable on housing quality is defined by β1,β2,β3,β4,and the model error is given as
ϵ.Three models were developed –the first which focused on all households in the target community and two additional models for
damaged and undamaged households.We developed these latter two models to disaggregate effects within those directly impacted by
the disaster,allowing us to compare assisted verses self-recovery outcomes more directly.
To test for the overall fit of the models,linearity was first assessed on a scatterplot of the studentized residuals against the unstan-
dardised predicted values.Homoscedasticity was then visually assessed using scatterplots,where the spread of residuals was con-
stant.All partial regression plots between each independent variable and housing quality were similarly assessed for linearity.Multi-
International Journal of Disaster Risk Reduction 100 (2024) 104076
8
T.Skwarko et al.
Table 2
Housing quality components.
Category Sub-Categories Survey Items Qualities Assessed
Water –6 Distance to water
Time to collect water
Type/quality of water source
Sanitation –10 Access to toilet
Distance/quality of toilet
Access to handwashing
Electricity –3 Access to electricity
Safety of electricity
Structural System Load Bearing Masonry (LBM)28 Foundation build quality
Bonding of masonry
Reinforced Concrete (RC)25 Foundation build quality
Column/beam spans
Reinforcing
Light Framing (LF)30 Foundation build quality
Cracking/Rusting/Bowing/Rotting
Roof Structure Steel Structure (SS)8 Reinforcing
Diagonals
Cracking/Rusting/Bowing
Timber Structure (TS)6 Connections/Splices/Fasteners/Anchors
Cracking/Rotting/Insects
Bamboo Structure (BS)8 Connections
Cracking/Rotting/Insects
Concrete Structure (CS)5 Leaking,cracking,sagging
Reinforcing
Roof Cover Light roof cover (eg.Sheeting)14 Roof Slope/Overhangs/Capping
Rafter/purlin spacing/Overlaps
Fasteners
Leaking/Rusting
Insulation
Heavy roof cover (eg.tiling)9 Roof Slope/Overhangs/Capping
Rafter/purlin spacing/Overlaps
Fasteners
Leaking/Sagging/broken tiles
Insulation
Flooring –12 Spacing of structural members
Anchors/Fasteners
Cracking/Rusting/Bowing/Rotting
collinearity was assessed using variance inflation factor (VIF)values,where a VIF value greater than 10 indicates a collinearity prob-
lem.All variables had VIF values below the threshold,with an average of 1.67,indicating multicollinearity was not an issue.
4.Results
Summary statistics of those households surveyed are first presented followed by results of the three OLS models produced to assess
the relationship between housing quality and assistance.Differences across assisted and unassisted subgroups are then compared.
4.1.Summary statistics
Of the 316 responses,132 households (42 %) received assistance post-disaster compared to 184 households (58 %) who did not.
The mean housing quality was 0.12 higher for those that received assistance among those who experienced some level of damage,
while housing quality was 0.30 higher for those who received assistance,but whose home was not damaged,show in Table 3.Under
all components except sanitation,households who received assistance had equal or higher housing quality.Households who were not
impacted by the disaster and who did not receive assistance had the lowest mean housing quality of any group.
A summary of the sample demographics comparing the overall and disaggregated assisted and unassisted groups is included in
Table 3.The average age of the respondents was similar across assistance and damage groups (54), as was the average household size
(3.8). The predominant family structure was a married couple (81.3%) with a male head of house (82.6%). Across all groups,the av-
erage household income fell in the IDR 1–2,000,000 per month bracket.According to the Indonesian Bureau of Statistics,the poverty
line for Yogyakarta City was IDR 533,423 per person per month in 2020 [59]. Dividing each household's reported income bracket by
household size,an estimated 58.5% (upper bound of income brackets asked)to 86.6%of households (a lower bound of income
brackets asked)of residents were living below the poverty line.This percentage was slightly higher among the assisted group,ranging
between 61.3%and 89.6%. The average unemployment rate (13 %) was double the national unemployment rate (6.5%) for August
2021 [56].
Of the 41.8%of households who received housing assistance,96 %received technical assistance and 98 %received material as-
sistance,or some combination of the two.Only two households reported receiving financial assistance.Of the 132 households who
International Journal of Disaster Risk Reduction 100 (2024) 104076
9
T.Skwarko et al.
Table 3
Summary statistics for households in Jogoyudan disaggregated by damage and post-disaster assistance.
All Undamaged Damaged
Unassisted Assisted Sub-Total Unassisted Assisted Sub-Total
Sample Size Count 316 133 36 169 51 96 147
Housing Quality Mean (s.d.) 5.18 (0.57)5.00 (0.60)5.30 (0.42)5.08 (0.58)5.20 (0.60)5.32 (0.47)5.28 (0.52)
Water Mean (s.d.) 0.73 (0.13)0.73 (0.13)0.78 (0.11)0.74 (0.13)0.73 (0.14)0.72 (0.12)0.72 (0.13)
Sanitation Mean (s.d.) 0.95 (0.13)0.53 (0.24)1.00 (0)0.98 (0.08)0.95 (0.15)0.90 (0.18)0.92 (0.17)
Electricity Mean (s.d.) 0.98 (0.1)0.97 (0.11)0.97 (0.12)0.97 (0.11)0.96 (0.11)0.99 (0.05)0.98 (0.08)
Structural System Mean (s.d.) 0.60 (0.22)0.53 (0.24)0.61 (0.20)0.54 (0.23)0.59 (0.19)0.7(0.16)0.66 (0.18)
Roof Structure Mean (s.d.) 0.58 (0.20)0.52 (0.21)0.58 (0.19)0.53 (0.21)0.62 (0.20)0.65 (0.16)0.64 (0.18)
Roof Cover Mean (s.d.) 0.55 (0.17)0.54 (0.18)0.52 (0.14)0.54 (0.17)0.56 (0.18)0.56 (0.16)0.56 (0.16)
Flooring Mean (s.d.) 0.78 (0.13)0.78 (0.14)0.78 (0.13)0.78 (0.13)0.78 (0.15)0.79 (0.09)0.79 (0.12)
Housing Quality (Standardised)Mean (s.d.) 0.00 (1.00)−0.26 (1.07)0.22 (0.75)−0.16 (1.03)0.05 (1.08)0.26 (0.84)0.19 (0.93)
Age Mean (s.d.) 54.5(11.9)54.5(11.9)53.9(13.2)54.4(12.1)55.25 (10.7)54.1(12.2)54.5(11.7)
Gender Female (%) 52.2 49.6 61.1 52.1 47.1 55.2 52.4
Male (%) 47.8 50.4 38.9 47.9 52.9 44.8 47.6
Family Status Unmarried (%) 18.7 14.3 27.8 17.2 15.7 22.9 20.4
Married (%) 81.3 85.7 72.2 82.8 84.3 77.1 79.6
Household Structure Female-headed (%) 17.4 12.8 16.7 13.6 11.8 27.1 21.8
Male-headed (%) 82.6 87.2 83.3 86.4 88.2 72.9 78.2
Household Size Mean (s.d)3.8(1.8)3.8(1.6)3.6(1.7)3.8(1.6)4.16 (2.2)3.7(1.9)3.9(2.0)
Covered Area Square meters,mean (s.d)54.8(32.5)68.8(34.7)44.0(22.5)63.5(34.0)53.5(36.2)40.2(20.3)44.8(27.5)
Land Tenure Tenured (%) 87.3 78.2 91.7 81.1 92.2 95.8 94.6
Untenured (%) 12.7 21.8 8.3 18.9 7.8 4.2 5.4
Household Income 0-1,000,000 IDR (%) 33.5 30.8 36.1 32.0 33.3 36.5 35.4
1-2,000,000 IDR (%) 43.0 41.4 41.7 41.4 35.3 50.0 36.1
2-3,000,000 IDR (%) 13.9 16.5 8.3 14.8 17.6 10.4 29.9
3-5,000,000 IDR (%) 6.3 6.0 8.3 6.5 11.8 3.1 27.9
5-7,500,000 IDR (%) 2.5 4.5 5.6 4.7 0.0 0.0 0.0
7.5–10,000,000 IDR (%) 0.6 0.8 0.0 0.6 2.0 0.0 0.7
Employment Construction (%) 6.6 3.0 5.6 3.6 13.7 8.3 10.2
Retail (%) 26.6 21.1 27.8 22.5 33.3 30.2 31.3
Hospitality (%) 10.4 9.8 16.7 11.2 9.8 9.4 9.5
Unemployed (%) 13.3 15.8 16.7 16.0 11.8 9.4 10.2
Assistance Unassisted (%) 58.2 100.0 0.0 78.7 100.0 0.0 34.7
Assisted (%) 41.8–100.0 21.3–100.0 65.3
Assistance Type Technical (%) 40.2–91.7 19.5–97.9 63.9
Material (%) 40.8–97.2 20.7–97.9 63.9
Financial (%) 0.6–0.0 0.0–1.0 0.7
Other (%) 0.9–2.8 0.6–2.1 1.4
Damage 0–No damage (%) 53.8 100.0 100.0 100.0– – –
1–No visible damage (%) 10.1–––51.0 7.3 22.5
2–Minor damage (%) 14.9–––37.3 29.2 32.0
3–Major damage (%) 6.3–––5.9 17.7 13.6
4–Extensive damage (%) 9.5–––3.9 29.2 20.4
5–Fully destroyed (%) 5.4–––2.0 16.7 11.6
Construction Activity 0-No Construction (%) 58.2 95.5 94.4 0.0 35.3 5.2 15.6
1-Repaired (%) 7.0 1.5 0.0 0.0 29.4 5.2 13.6
2-Partial rebuild (%) 13.9 0.8 2.8 0.0 27.5 29.2 28.6
3-New house (%) 20.9 2.3 2.8 0.0 7.8 60.4 42.2
Construction Abilities 1-Not at all (%) 33.2 42.1 41.7 42.0 17.6 26.0 23.1
2-Not very (%) 21.8 20.3 11.1 18.3 31.4 22.9 25.9
3–Neutral (%) 16.1 10.5 27.8 14.2 17.6 18.8 18.4
4-Slightly Confident (%) 18.7 19.5 5.6 16.6 25.5 18.8 21.1
5-Very Confident (%) 10.1 7.5 13.9 8.9 7.8 13.5 11.6
**Continuous data is reported as mean (standard deviation), and categorical data is reported using frequencies.
had received assistance,112 households (85 %) reported receiving assistance from Habitat for Humanity Indonesia while others in
this group reported receiving housing assistance through other means.Noting the 10-year period between the disaster and the survey,
it is possible that individual recollections or understandings of the assistance type may have varied from actual assistance,however
there were no other organisations known to be working in the area.
Overall,46.2%of households reported some level of damage to their homes during the 2010 Merapi disaster.Community-wide,
the most common reported level of damage was minor damage (14.9%). The rest of the damage levels were,in decreasing order,no
visible damage (10.1%), extensive damage (9.5%), major damage (6.3%) and fully destroyed (5.4%). 21.3%of households who re-
International Journal of Disaster Risk Reduction 100 (2024) 104076
10
T.Skwarko et al.
ported no damage to their homes received housing assistance following the Merapi disaster.The reason for this was that assistance
was also targeted at households who were living in substandard housing.
4.2.Model results
A summary of the three developed models,including for the overall population,Model 1,as well as damaged and undamaged sub-
groups,Model 2 and 3,is shown in Table 4.The unstandardised coefficients (B)represent the standard deviation (SD)change in hous-
ing quality based on a one-unit change in respective independent variables.The standardised coefficient
(𝛽)
represents a standard de-
viation change in housing quality based on one standard deviation change in respective independent variables.A comparison of βcan
therefore be used to assess the relative impact of independent variables.These will be referenced in text with their 95 %Confidence
Intervals (CI)as (
𝛽, 95%CI ).
We will discuss the results from the combined Model 1,noting where relevant differences may have oc-
curred for undamaged,Model 2,and damaged,Model 3 subgroups.
Within Model 1,Land tenure (0.17,95 %CI:0.06,0.28)was the strongest predictor of housing quality,with those having a more
secure form of tenure resulting in 0.17 standard deviation higher housing quality for the overall population.In practice,this equates
to a household moving from the 50th percentile to the 67th percentile if they had more secure land tenure,holding all other factors
constant.While land tenure was also a significant predictor among Model 3:damaged households (0.24,95 %CI:0.07,0.40), it was
not among the Model 2:undamaged group.Within Model 1,the second strongest predictor of housing quality was whether a house-
hold received assistance (0.16,95 %CI:0.02,0.30), with the same trend among the Model 2:undamaged group.Assistance was not a
significant predictor for the Model 3:damaged group,suggesting that those whose homes were damaged but did not receive support
also found ways to rebuild effectively.Our findings cannot however comment on whether those who received assistance would have
been able to recover without the support they received.Employment in retail was also significant amongst the Model 1:Combined group
(0.12,95 %CI:0.01,0.24)and correlated with higher housing quality,likely due to higher incomes as compared to other livelihoods.
This was significant in the Model 3:damaged group,but not among the Model 2:undamaged group.The last variable to show a signifi-
cant relationship with housing quality was self-reported construction abilities (0.12,95 %CI:0.00,0.23), which followed a similar
trend for the Model 2:undamaged group,but was not significant for the Model 3;damaged group.
In addition to the above variables,there were a few additional variables within the Model 2:undamaged and Model 3:damaged
groups that had unique trends.Family status (married households)was correlated with lower housing quality among the Model 2:un-
damaged group (−0.32,95 %CI:0.59,−0.05), but unmarried households were relatively rare among this group.Household size
(−0.19,95 %CI:0.37,−0.01)was negatively correlated and household income (0.19,95 %CI:0.01,0.37)was positively correlated
with housing quality in the Model 3:damaged group,likely due to increased financial pressures placed on larger and poorer house-
holds.
5.Discussion
This section discusses the role of assistance on post-disaster recovery processes.We first discuss the impact of assistance on hous-
ing quality followed by a discussion of other important factors that emerged from our analysis.
Table 4
Summary of ordinal linear regression results predicting housing quality for overall community,and undamaged and damaged subgroups.
Variables Model 1:Overall Model 2:Undamaged Model 3:Damaged
B(S.E)β β 95 %C.I B (S.E)β β 95 %C.I B (S.E)β β 95 %C.I
Age 0 (0.01)0.01 (-0.11,0.13)0(0.01)−0.06 (-0.23,0.11)0.01 (0.01)0.14 (-0.04,0.31)
Gender −0.11 (0.12)−0.05 (-0.17,0.07)−0.24 (0.17)−0.12 (-0.28,0.05)−0.08 (0.17)−0.04 (-0.22,0.14)
Family Status −0.32 (0.22)−0.12 (-0.29,0.04)−0.86 (0.37)** −0.32** (-0.59, -0.05)−0.04 (0.27)−0.02 (-0.25,0.22)
Household Structure 0.22 (0.23)0.08 (-0.09,0.26)0.29 (0.41)0.10 (-0.17,0.37)0.17 (0.28)0.08 (-0.17,0.32)
Household Size −0.02 (0.03)−0.03 (-0.15,0.09)0.07 (0.05)0.12 (-0.05,0.28)−0.09 (0.04)** −0.19** (-0.37, -0.01)
Covered Area 0 (0)0.07 (-0.05,0.19)0(0)0.10 (-0.07,0.27)0(0)0.08 (-0.09,0.25)
Land Tenure 0.52 (0.17) *** 0.17*** (0.06,0.28)0.33 (0.2)0.12 (-0.03,0.28)0.96 (0.34)*** 0.24*** (0.07,0.40)
Household Income 0.06 (0.06)0.06 (-0.06,0.18)0(0.08)0.00 (-0.17,0.17)0.19 (0.09)** 0.19** (0.01,0.37)
Employment:Construction −0.1(0.23)−0.03 (-0.14,0.09)−0.12 (0.43)−0.02 (-0.17,0.13)−0.27 (0.27)−0.09 (-0.26,0.08)
Employment:Retail 0.28 (0.13)** 0.12* (0.01,0.24)0.32 (0.2)0.13 (-0.03,0.29)0.16 (0.18)0.08 (-0.09,0.25)
Employment:Hospitality −0.18 (0.19)−0.06 (-0.17,0.06)0.08 (0.26)0.02 (-0.13,0.18)−0.49 (0.27)*−0.15 (-0.33,0.02)
Employment:Unemployed −0.01 (0.18)0.00 (-0.13,0.12)0.01 (0.25)0.00 (-0.17,0.18)−0.07 (0.28)−0.02 (-0.20,0.16)
Assistance 0.33 (0.14)** 0.16** (0.02,0.30)0.36 (0.2)* 0.14* (-0.02,0.31)0.24 (0.21)0.13 (-0.09,0.34)
Damage 0.04 (0.06)0.06 (-0.14,0.25)– – – −0.02 (0.08)−0.03 (-0.25,0.19)
Construction Activity −0.01 (0.08)−0.01 (-0.20,0.19)−0.14 (0.17)−0.07 (-0.24,0.09)0.03 (0.1)0.04 (-0.19,0.27)
Construction Abilities 0.08 (0.04) ** 0.12** (0.00,0.23)0.1(0.06)* 0.14* (-0.02,0.29)0.07 (0.06)0.10 (-0.07,0.27)
Model 1:Overall N =316,R2=0.13,p<0.001.
Model 2:Undamaged N =169,R2=0.16,p=0.02.
Model 3:Damaged N =147,R2=0.18,p=0.04.
*** = p<0.01, ** = p<0.05,*=p<0.1.
International Journal of Disaster Risk Reduction 100 (2024) 104076
11
T.Skwarko et al.
5.1.The impact of assistance on housing quality
The primary aim of assistance in Jogoyudan was to reduce vulnerabilities by providing safe and decent housing.When considering
that the mean difference in standardised housing quality for households who received assistance was higher among both damaged
and undamaged households,0.48 and 0.21 respectively,this appears to have been successfully achieved.Assistance was statistically
significant in Model 1:Combined which held constant both levels of damage,household incomes,and other factors that play a large
role in shaping housing quality.We observed that some who self-recovered without assistance were able to achieve higher housing
quality,but others were not.Households who did not receive assistance and who were not impacted by the disaster can be considered
a baseline,unsupported level of development in Jogoyudan.This group had the lowest mean housing quality of any group studied (M:
0.26,SD:1.07).
Taking a broad view of the impact of assistance in Jogoyudan,households who received assistance were generally satisfied with
the quality of their housing construction and were more satisfied than others within the community.48 %of households who re-
ceived assistance reported an increase in quality-of-life post-disaster,while 31 %responded no change.When compared to house-
holds who did not receive assistance,only 13 %reported an increase in quality since the disaster,with the vast majority (72 %) re-
porting no change.A comparison of satisfaction with construction quality is shown in Fig.6.Similar trends were found when asking
households about improvements in their quality of life,shown in Fig.7.
5.2.Beyond assistance:What else impacts recovery?
Our models show that assistance impacted housing quality over the long term,however there were two other variables that were
consistently correlated with housing quality –land tenure and construction abilities.
Fig.6.Satisfaction with housing construction quality between assisted and unassisted groups.
Fig.7.Changes in quality of life between assisted and unassisted groups.
International Journal of Disaster Risk Reduction 100 (2024) 104076
12
T.Skwarko et al.
5.2.1.Land tenure
In all three models,land tenure was correlated with higher housing quality.Land tenure has previously been raised as an impor-
tant component of disaster vulnerability [28,60]. To further demonstrate the long-term impact of tenure,only 46 %of those house-
holds with insecure tenure carried out construction work in the last 10 years,compared to 76 %with formal tenure.When asked if
land tenure had impacted ongoing rebuilding decisions,85 %of households surveyed responded that this played a significant role in
their housing construction investments.Across our results,land tenure was the single largest predictor of housing quality.This high-
lights a reticence among households to carry out construction activities where land protection is not in place.Untenured households
have limited control over their property [61], and no legal recourse against forced development or eviction.The impact of this is evi-
dent within the sub-group with the lowest housing quality,those unassisted and not impacted by disaster,where 1 in 5 households
did not have secure tenure.Faced with such insecurity,untenured households may be less willing to spend their limited resources on
improving building safety or disaster resilience,which ultimately contributes to lower housing quality.
This uncertainty extends beyond households,with humanitarian organisations often defining beneficiary eligibility based on land
tenure.Organisations need to remove the exclusionary power of land tenure as a determinant of eligibility for assistance and better
consider tenure-specific initiatives.Technical assistance for example is a lower cost alternative,and without the long-term risks of
abortive investments if forced relocation or development occurs.It also has the added benefit of being able to be proactively imple-
mented,which removes some of the complexities of delivering assistance programs in post-disaster or post-conflict environments.Ad-
dressing informal tenure directly is a complex challenge,as it requires coordinated action by public authorities to formally recognise
land occupants [62]. Organisations can assist by providing legal advice and support to communities to support this process.
5.2.2.Construction abilities
Confidence in construction abilities had the second largest correlation with housing quality after assistance.A one standard deviation
increase in confidence resulted in a 0.12 standard deviation increase in housing quality when all other factors were held constant.
This is noteworthy because of all the variables assessed,it is the only one most often targeted through humanitarian intervention.
To assess the impact of the disaster on construction abilities,households were asked whether they felt more confident than they
did pre-disaster.Amongst those whose homes were not damaged,81 %of assisted households reported an increase in confidence in
their construction abilities,compared to 73 %of unassisted households.For households whose homes were damaged,77 %of as-
sisted households reported an increase in construction abilities,compared to 84 %who did not receive assistance.Conceptually this
aligned with field observations,where households who were impacted had no option but to carry out construction activity to recover.
It also makes sense that households not impacted by the disaster,and who did not receive assistance,would be the least likely to re-
port an increase in construction ability because they did not participate in reconstruction activities.This is particularly relevant con-
sidering the relatively high percentage in this group who had unsecure tenure and were therefore unwilling to carry out construction
activities.Households who received assistance unrelated to disaster impacts were more likely to report an increase in construction
abilities compared to those who received assistance because of the disaster.
Furthermore,we found that households who self-recovered,the subgroup that were impacted and did not receive assistance,had
the largest increase in construction abilities.This supports that engagement in broader recovery processes may have led to better long
term outcomes [25], and anecdotally that safer construction techniques are learnt and replicated by non-beneficiaries [32,44]. Noting
that our results show that increasing construction abilities can positively impact housing quality in the long term,investing in com-
munity education could have a significant impact in reducing disaster vulnerability over the long term.
5.3.Limitations
While efforts were made to triangulate data sources through reviewing assistance reports from Habitat for Humanity Indonesia,
we did rely on household self-reporting and recollection for the damage assessments and assistance received.In almost all cases
(98.6%of households surveyed), there was not a change in residence since the disaster,giving confidence that at least the same indi-
viduals were residing in the housing assessed.It is possible that some inconsistencies in the data exist given the time (over 10 years)
which has passed since the disaster.We do recognise the subjectivity in the damage assessments as a limitation,despite providing de-
tailed damage descriptions to respondents adapted from Jenkins et al. [52]. We were also missing pre-disaster data on specific hous-
ing conditions;however,this is not unexpected or uncommon in disaster studies.Whilst it may be difficult to develop a baseline view
of pre-disaster housing quality,future studies could attempt to integrate community perspectives to better control for pre-existing in-
equalities in housing when assessing longer term outcomes.The research was also only carried out in one location,and the geographi-
cal context may have limitations on the wider applicability of the findings.This could be validated by extending the study to addi-
tional locations.
6.Conclusion
This research sought to understand the long-term impact of humanitarian assistance on housing quality,drawing on an evaluation
of housing programs implemented in Jogoyudan located in Yogyakarta,Indonesia over 10 years after the 2010 Merapi eruption.We
found that housing assistance was a significant predictor of improved housing quality across the entire community.However,the
strongest predictor of housing quality was land tenure,accounting for nearly one-fifth of the variation in housing quality.Other sig-
nificant predictors included access to more stable livelihoods and the construction ability of a household.Our results demonstrate the
potential of humanitarian assistance to accomplish more than just short-term benefit and the potential to lead to longer term transfor-
mation of housing.
International Journal of Disaster Risk Reduction 100 (2024) 104076
13
T.Skwarko et al.
Identifying land tenure as an important factor that shapes long-term housing outcomes is not new [63,64], but our identification
of the extent to which this influences longer term housing quality is a significant contribution of this work.Donors and organisations
often focus on physical housing units,but our results show the importance of attention to tenure security.This security is often a
larger hurdle that households themselves are unable to overcome independently.Shifting the timescales of evaluation,it is not diffi-
cult to see the benefit of funding land initiatives over just product-focused programs.
Our findings also identified construction abilities as a predictor of housing quality –a likely by-product of the incremental and
owner-driven nature of housing systems in informal settlements [65]. We found that households whose homes were damaged and
who recovered with no assistance –those who self-recovered –had the highest self-assessed construction abilities and were most
likely to report an increase in those abilities since the disaster.This has important implications for the growing self-recovery literature
[25,37,41,66]. Our case also highlights an example where assistance was not just given to those who homes were directly damaged by
the disaster,but also those more broadly whose housing was substandard.While this is not always possible for organisations,it show-
cases an example where collectively lifting the living standard of communities can lead to benefits in recovery.This recognition of set-
tlement-scale thinking better recognises the role of vulnerability in creating disaster risk [67,68].
Our research also found that households who were unimpacted by the disaster and did not receive assistance had the lowest mean
housing quality of all groups assessed,were characterised by the highest percentage of insecure tenancies and the lowest reported
construction abilities.Whilst the case for self-recovery is clear post-disaster,this finding further highlights that communities still re-
quire support to enable this process.There is therefore a need to shift away from narrowly targeted post-disaster assistance towards
more broad programming strategies aimed at reducing inequalities to better shape long-term outcomes.
Declaration of competing interest
The authors declare the following financial interests which may be considered as potential competing interests.Tatiana Skwarko,
Ivy He,Sarah Cross,Aaron Opdyke,Tantri Hadayani,Andreas Hapsoro,and Yunita Idris report financial and administrative support
were provided by Habitat for Humanity International.
Data availability
Data will be made available on request.
Acknowledgements
We would like to acknowledge Habitat for Humanity International for funding this research.Habitat for Humanity International
and Habitat for Humanity Indonesia provided input to contextualise data,however the study design,data collection,analysis,writ-
ing,and decision to both commence the study and submit this article for publication were taken independently by authors not affili-
ated with the funding organisation.The views and opinions expressed are those of the authors and do not necessarily reflect the offi-
cial policy or position of Habitat for Humanity International or Habitat for Humanity Indonesia.We would also like to thank Shof-
farisna Ithmaanna,Tri Budi Utami,Ayumna Uzlifati,Atilla Famega M,Ferian Yudha Pratama,Muhammad Irfan Tian Syamsuddin,
Muhammad Rifki Febrianto,and Bagas Muhammad for their assistance in collecting the survey data.
Appendix A.Supplementary data
Supplementary data to this article can be found online at https://doi.org/10.1016/j.ijdrr.2023.104076.
References
[1]T.Schilderman,E.Parker,Still Standing?:Looking Back at Reconstruction and Disaster Risk Reduction in Housing,Practical Action Publishing,2014.
[2]J.Puri,A.Aladysheva,V.Iversen,Y.Ghorpade,T.Brück,Can rigorous impact evaluations improve humanitarian assistance? J.Dev.Effect.9(2017)519–542,
https://doi.org/10.1080/19439342.2017.1388267.
[3]K.Georgieva,N.Shah,H.Ibrahim,B.Jafar,W.Macnee,T.Manuel,L.Mohohlo,D.Sriskandarajah,M.Wallstrom,Too important to fail -addressing the
humanitarian financing gap,High-Level Panel on Humanitarian Financing Report to the Secretary-General,New York,2016,pp.1–3.
[4]L.Bosher,K.Chmutina,D.van Niekerk,Stop going around in circles:towards a reconceptualisation of disaster risk management phases,Disaster Prev.Manag.
30 (2021)525–537,https://doi.org/10.1108/DPM-03-2021-0071.
[5]B.Wisner,P.Blaikie,T.Cannon,I.Davis,At Risk :Natural Hazards,People’s Vulnerability and Disasters,second ed., Routledge,Abingdon,UK,2004,https://
doi.org/10.4324/9780203714775.
[6]C.B.Shrestha,B.Parajuli,S.Malani,Unravelling constraints of urban housing reconstruction,Prog.Disaster Sci.11 (2021)100178,https://doi.org/10.1016/
j.pdisas.2021.100178.
[7]V.Sukhwani,H.Napitupulu,D.Jingnan,M.Yamaji,R.Shaw,Enhancing cultural adequacy in post-disaster temporary housing,Prog.Disaster Sci. (2021)
100186,https://doi.org/10.1016/j.pdisas.2021.100186.
[8]A.Opdyke,B.Goldwyn,A.Javernick-Will,Defining a humanitarian shelter and settlements research agenda,Int.J.Disaster Risk Reduc.52 (2021)101950,
https://doi.org/10.1016/j.ijdrr.2020.101950.
[9]E.Maly,M.Vahanvati,T.Sararit,People-centered disaster recovery:a comparison of long-term outcomes of housing reconstruction in Thailand,India,and
Japan,Int.J.Disaster Risk Reduc. (2022)103234,https://doi.org/10.1016/j.ijdrr.2022.103234.
[10]I.Davis,C.Parrack,Taking the long view,in:D.Sanderson,A.Sharma (Eds.), State Humanit.Shelter Settl.2018,International Federation of Red Cross and Red
Crescent Societies,2018,pp.9–14.
[11]C.Parrack,Long Term Impacts of Shelter Programmes,Evid.Aid., 2018.https://evidenceaid.org/long-term-impacts-of-shelter-programmes/. (Accessed 1
May 2023).
[12]S.Barakat,Housing Reconstruction after Conflict and Disaster,Overseas Development Institute,London,2003.
[13]M.Lyons,Building Back better:the large-scale impact of small-scale approaches to reconstruction,World Dev.37 (2009)385–398,https://doi.org/10.1016/
j.worlddev.2008.01.006.
International Journal of Disaster Risk Reduction 100 (2024) 104076
14
T.Skwarko et al.
[14]E.Sunarti,E.Gunawan,S.Widiyantoro,G.I.Marliyani,R.Ida,Critical point on housing construction,resilience and family subjective welfare after disaster:
notes from the Lombok,Indonesia,earthquake sequence of July-August 2018,Geomatics,Nat.Hazards Risk 12 (2021)922–938,https://doi.org/10.1080/
19475705.2021.1910576.
[15]B.Simons,V.Maynard,E.Wagemann,Supporting the recovery of home-based enterprises,in:J.Twigg,E.Babister (Eds.), Roadmap Res.Collab.Res.Framew.
Humanit.Shelter Settl.Assist., InterAction,Washington,DC,2021,pp.89–96.https://www.interaction.org/blog/roadmap-for-research/.
[16]E.Nabong,A.Opdyke,M.K.Alinsunurin,S.Merril,G.McDonald,P.Jensen,Exploring the role of shelter and livelihoods recovery,in:J.Twigg,E.Babister
(Eds.), Roadmap Res.Collab.Res.Framew.Humanit.Shelter Settl.Assist., InterAction,Washington,DC,2021,pp.97–104.https://www.interaction.org/blog/
roadmap-for-research/.
[17]S.Barakat,S.A.Zyck,Housing reconstruction as socio-economic recovery and state building:evidence from southern Lebanon,Hous.Stud.26 (2011)133–154,
https://doi.org/10.1080/02673037.2010.512750.
[18]J.Twigg,E.Lovell,H.Schofield,L.M.Morel,B.Flinn,S.Sargeant,T.Dijkstra,V.Stephenson,A.Albuerne,T.Rossetto,D.D’Ayala,Self-recovery from
Disasters,2017,p.48.
[19]B.Flinn,Defining ‘better’better:why building Back better means more than structural safety,J.Humanit.Aff.2(2020)35–43,https://doi.org/10.7227/
JHA.032.
[20]A.Maskrey,Disaster Mitigation:a Community Based Approach,Oxfam,Oxford,1989.
[21]T.Schilderman,Adapting traditional shelter for disaster mitigation and reconstruction:experiences with community-based approaches,Build.Res.Inf.32
(2004)414–426,https://doi.org/10.1080/0961321042000250979.
[22]H.Schofield,E.Lovell,B.Flinn,J.Twigg,Barriers to urban shelter self-recovery in Philippines and Nepal:lessons for humanitarian policy and practice,J.Br.
Acad.7(2019)83–107,https://doi.org/10.5871/jba/007s2.083.
[23]G.Karunasena,R.Rameezdeen,Post-disaster housing reconstruction:comparative study of donor vs owner-driven approaches,Int.J.Disaster Resil.Built
Environ.1(2010)173–191,https://doi.org/10.1108/17595901011056631.
[24]Y.Chang,S.Wilkinson,R.Potangaroa,E.Seville,Donor-driven resource procurement for post-disaster reconstruction:constraints and actions,Habitat Int.35
(2011)199–205.
[25]C.Parrack,B.Flinn,M.Passey,Getting the message across for safer self-recovery in post-disaster shelter,Open House Int.39 (2014)47–58,https://doi.org/
10.1108/OHI-03-2014-B0006.
[26]A.Mukherji,From Tenants to Homeowners:housing renters after disaster in Bhuj,India,hous,For.Stud.30 (2015)1135–1157,https://doi.org/10.1080/
02673037.2015.1008423.
[27]Global Shelter Cluster,Humanitarian Rental Market Interventions:A Review of Best Practices,2020.https://sheltercluster.org/resources/documents/
reportrentalmarketsfinal.
[28]W.G.Peacock,N.Dash,Y.Zhang,Sheltering and housing recovery following disaster,in:Handb.Disaster Res., Springer,New York,New York,NY,2009,pp.
258–274.
[29]V.Maynard,E.Parker,J.Twigg,The Effectiveness and Efficiency of Interventions Supporting Shelter Self-Recovery Following Humanitarian Crises:an
Evidence Synthesis Protocol,Oxfam,2016,https://doi.org/10.21201/2016.605179.
[30]M.Vahanvati,M.Mulligan,A new model for effective post-disaster housing reconstruction:lessons from Gujarat and Bihar in India,Int.J.Proj.Manag.35
(2017)802–817,https://doi.org/10.1016/j.ijproman.2017.02.002.
[31]A.Opdyke,A.Javernick-Will,M.Koschmann,Assessing the impact of household participation on satisfaction and safe design in humanitarian shelter projects,
Disasters 43 (2019)926–953.
[32]L.Harriss,C.Parrack,Z.Jordan,Building safety in humanitarian programmes that support post-disaster shelter self-recovery:an evidence review,Disasters 44
(2020)307–335,https://doi.org/10.1111/disa.12397.
[33]J.E.Duyne Barenstein,Housing Reconstruction in Post-earthquake Gujarat:a Comparative Analysis,Overseas Development Institute,London,2006.
[34]L.M.Morel,Shelter Assistance:Gaps in the Evidence,Care International UK,2018.
[35]C.for R.on the E.of D,Cred,The Human Cost of Natural Disasters,Centre for Research on the Epidemiology of Disasters CRED,Brussels,Belgium,2015.
[36]United Nations Disaster Relief Office,Shelter After Disaster,Guidelines for Assistance -World,1982.https://reliefweb.int/report/world/shelter-after-disaster-
guidelines-assistance. (Accessed 13 April 2021).
[37]S.Sargeant,A.Finlayson,T.Dijkstra,B.Flinn,H.Schofield,L.M.Morel,J.Twigg,E.Lovell,V.Stephenson,B.R.Adhikari,The influence of the physical
environment on self-recovery after disasters in Nepal and the Philippines,Int.J.Disaster Risk Reduc.50 (2020)101673,https://doi.org/10.1016/
j.ijdrr.2020.101673.
[38]J.Bredenoord,P.van Lindert,Pro-poor housing policies:Rethinking the potential of assisted self-help housing,Habitat Int.34 (2010)278–287,https://
doi.org/10.1016/j.habitatint.2009.12.001.
[39]I.Davis,What have we learned from 40 years’experience of Disaster Shelter? Environ.Hazards 10 (2011)193–212,https://doi.org/10.1080/
17477891.2011.597499.
[40]I.Davis,Emergency shelter,Disasters 1 (1977)23–39,https://doi.org/10.1111/j.1467-7717.1977.tb00006.x.
[41]J.Twigg,The evolution of shelter “self-recovery”:adapting thinking and practice for post-disaster resilience,J.Br.Acad.9s8 (2021)5–22,https://doi.org/
10.5871/jba/009s8.005.
[42]I.Ahmed,An overview of post-disaster permanent housing reconstruction in developing countries,Int.J.Disaster Resil.Built Environ.2(2011)148–164,
https://doi.org/10.1108/17595901111149141.
[43]S.A.Andrew,S.Arlikatti,L.C.Long,J.M.Kendra,The effect of housing assistance arrangements on household recovery:an empirical test of donor-assisted and
owner-driven approaches,J.Hous.Built Environ.28 (2013)17–34,https://doi.org/10.1007/s10901-012-9266-9.
[44]A.Opdyke,A.Javernick-Will,M.Koschmann,Household Construction Knowledge Acquisition in Post-Disaster Shelter Training,2018,https://doi.org/
10.1016/j.ijdrr.2018.02.038.
[45]S.Mannakkara,S.Wilkinson,Re-conceptualising “Building Back Better”to improve post-disaster recovery,Int.J.Manag.Proj.Bus.7(2014)327–341,https://
doi.org/10.1108/IJMPB-10-2013-0054.
[46]H.Ren,N.Yuan,H.Hu,Housing quality and its determinants in rural China:a structural equation model analysis,J.Hous.Built Environ.34 (2019)313–329,
https://doi.org/10.1007/s10901-018-9629-y.
[47]R.J.Lawrence,Housing quality:an agenda for research,Urban Stud.32 (1995)1655–1664,https://doi.org/10.1080/00420989550012294.
[48]P.Oliver,The cultural context of shelter provision,in:Disasters Small Dwell,Pergamon Press,Oxford,1981,pp.39–42.
[49]W.G.Peacock,S.Van Zandt,Y.Zhang,W.E.Highfield,Inequities in long-term housing recovery after disasters,J.Am.Plann.Assoc.80 (2014)356–371,
https://doi.org/10.1080/01944363.2014.980440.
[50]B.Voight,E.K.Constantine,S.Siswowidjoyo,R.Torley,Historical eruptions of Merapi volcano,central Java,Indonesia,1768–1998,J.Volcanol.Geoth.Res.
100 (2000)69–138,https://doi.org/10.1016/S0377-0273(00)00134-7.
[51]P.Jousset Surono,J.Pallister,M.Boichu,M.F.Buongiorno,A.Budisantoso,F.Costa,S.Andreastuti,F.Prata,D.Schneider,L.Clarisse,H.Humaida,S.Sumarti,
C.Bignami,J.Griswold,S.Carn,C.Oppenheimer,F.Lavigne,The 2010 explosive eruption of Java’s Merapi volcano—a‘100-year’event,J.Volcanol.Geoth.
Res.241–242 (2012)121–135,https://doi.org/10.1016/j.jvolgeores.2012.06.018.
[52]S.F.Jenkins,J.C.Phillips,R.Price,K.Feloy,P.J.Baxter,D.S.Hadmoko,E.de Bélizal,Developing building-damage scales for lahars:application to Merapi
volcano,Indonesia,Bull.Volcanol.77 (2015)75,https://doi.org/10.1007/s00445-015-0961-8.
[53]R.Gertisser Subandriyo,N.Aisyah,H.Humaida,K.Preece,S.Charbonnier,A.Budi-Santoso,H.Handley,S.Sumarti,D.S.Sayudi,I.G.M.A.Nandaka,H.E.
Wibowo,An overview of the large-magnitude (VEI 4)eruption of Merapi,in:R.Gertisser,V.R.Troll,T.R.Walter,I.G.M.A.Nandaka,A.Ratdomopurbo (Eds.),
Merapi Volcano Geol.Eruptive Act.Monit.High-Risk Volcano,Springer International Publishing,Cham,2023,pp.353–407 https://doi.org/10.1007/978-3-031-
15040-1_12,2010.
[54]S.Jenkins,J.-C.Komorowski,P.J.Baxter,R.Spence,A.Picquout,F.Lavigne,Surono,the Merapi 2010 eruption:an interdisciplinary impact assessment
International Journal of Disaster Risk Reduction 100 (2024) 104076
15
T.Skwarko et al.
methodology for studying pyroclastic density current dynamics,J.Volcanol.Geoth.Res.261 (2013)316–329,https://doi.org/10.1016/
j.jvolgeores.2013.02.012.
[55]M.S.Roychansyah,A.Diwangkari,Kampung Oriented Development Model,A Rapid Appraisal of Local Communities,2009,p.16.
[56]Badan Pusat Statistik (BPS), Data Sensus Penduduk 2020 –Badan Pusat Statistik Republik Indonesia,2021 Jakarta.https://sp2010.bps.go.id/index.php/
site?id=3400000000&wilayah=DI-Yogyakarta.
[57]V.Ariyanti,T.Gaafar,S.De La Sala,J.Edelenbos,P.Scholten,Towards liveable volcanic cities:a look at the governance of lahars in Yogyakarta,Indonesia,
and Latacunga,Ecuador,Cities 107 (2020)102893,https://doi.org/10.1016/j.cities.2020.102893.
[58]F.Kelling,The Wider Impacts of Humanitarian Shelter and Settlements Assistance,InterAction,Washington,D.C., 2020.
[59]D.I.Bps Provinsi,Yogyakarta,Garis Kemiskinan Menurut Kabupaten/Kota (Rupiah), 2019-2021,Badan Pus.Stat.Provinsi Yogyak,2021.https://
yogyakarta.bps.go.id/indicator/23/168/1/garis-kemiskinan-menurut-kabupaten-kota.html. (Accessed 14 November 2021).
[60]J.Y.Lee,S.Van Zandt,Housing tenure and social vulnerability to disasters:a review of the evidence,J.Plann.Lit.34 (2019)156–170,https://doi.org/
10.1177/0885412218812080.
[61]J.Y.Lee,S.Van Zandt,Housing tenure and social vulnerability to disasters:a review of the evidence,J.Plann.Lit.34 (2019)156–170,https://doi.org/
10.1177/0885412218812080.
[62]A.Durand-Lasserve,H.Selod,The formalization of urban land tenure in developing countries,in:S.V.Lall,M.Freire,B.Yuen,R.Rajack,J.-J.Helluin (Eds.),
Urban Land Mark., Springer Netherlands,Dordrecht,2009,pp.101–132,https://doi.org/10.1007/978-1-4020-8862-9_5.
[63]A.Reale,J.Handmer,Land tenure,disasters and vulnerability,Disasters 35 (2011)160–182,https://doi.org/10.1111/j.1467-7717.2010.01198.x.
[64]I.R.Jahn,L.Guadagno,E.Gandia,V.Bonnefoy,Overcoming land tenure Barriers in shelter and other reconstruction activities in post-disaster settings,in:K.
Sudmeier-Rieux,M.Fernández,I.M.Penna,M.Jaboyedoff,J.C.Gaillard (Eds.), Identifying Emerg.Issues Disaster Risk Reduct.Migr.Clim.Change Sustain.Dev.,
Springer International Publishing,Cham,2017,https://doi.org/10.1007/978-3-319-33880-4.
[65]G.Lizarralde,The Invisible Houses:Rethinking and Designing Low-Cost Housing in Developing Countries,Routledge,New York,NY,2015.
[66]B.Flinn,H.Schofield,L.M.Morel,The case for self-recovery,Forced Migr.Rev.55 (2017)12–14.
[67]M.Keen,D.Sanderson,K.Osborne,R.Deo,J.Faith,A.Ride,Area-based approaches and urban recovery in the Pacific:lessons from Fiji,Solomon Islands and
Vanuatu,Environ.Urbanization (2022)https://doi.org/10.1177/09562478211072668,09562478211072668.
[68]D.Sanderson,Implementing area-based approaches (ABAs)in urban post-disaster contexts,Environ.Urbanization 29 (2017)349–364,https://doi.org/
10.1177/0956247817717422.
[69]J.L.Goodman,Causes and indicators of housing quality,Social Indicators Research 5 (1) (1978)195–210.https://doi.org/10.1007/BF00352929.
