Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Livelihood resilience in the face of recurring floods: an empirical evidence from Northwest Ethiopia

Livelihood resilience in the face of recurring floods: an empirical evidence from Northwest Ethiopia Background: The recent trend of increasing incidents of floods in Ethiopia is disrupting the livelihoods of a significant proportion of the country’s population. This study assesses the factors that shape the resilience and the vulnerability of rural households in the face of recurring floods by taking the case of Dembia district of Northwest Ethiopia as one of the flood-prone areas in the country. Results: The data for the study were collected through a survey of 284 households, two focus group discussions, and 12 key informant interviews. Principal Component Analysis and simple linear regression were used for the analysis. The former served both for data reduction and identification of the dominant factors that explain resilience to recurring flood hazards while the latter was used to check the relationship between resilience and vulnerability. Findings indicate that access and use of livelihood resources such as size of farmlands, availability of farm oxen, credit as well as ability to draw help from social networks were found to be the most important factors that determine the resilience of households to floods. Similarly, the coping strategies employed by households were found to be constrained mainly by the scale and impact of the recent floods and lack or shortage of basic infrastructural and social facilities. Conclusions: The results confirmed that most of the traditional coping strategies employed by households failed to effectively help households offset the impacts of flooding. Given the livelihood context of smallholder farming system in the studied area, context specific institutional interventions such as the integrated use of both safety nets and cargo nets may help communities to overcome livelihood predicaments associated with the recurrent flood disasters. This implies that policy should focus more on addressing the factors that expose people to flood disasters and shape their resilience, rather than focusing on short-term emergency responses which seems to be the norm in much of the flood affected areas in the country. Keywords: Flood disaster, Resilience index, Vulnerability index, Dembia, Northwest Ethiopia Background Reduction [UNISDR] (2015). The frequency and severity It is widely recognized that environmental hazards fre- of flooding are also increasing in many parts of the quently affect the livelihoods of many people around the world associated with population pressure, urbanization world. The effects of these hazards cannot be expected and climate change (Hirabayashi et al., 2013; Jongman to be similar as people and nations differ in terms of et al., 2014). This is evident when one considers the their level of development, which largely determines number of people affected by flooding in recent decades. their response to specific disasters. For instance, flooding accounts much of the loss event Flooding is one of the most frequent and destructive worldwide between 1980–2014 more than any other sin- environmental hazards that occur annually worldwide gle disaster (Munich RE, 2015) and tops the list of nat- (United Nations International strategy for Disaster ural disasters by economic damages in 2014 (Guha-Sapir et al., 2015). Flooding is also the leading disaster agent * Correspondence: zerihun.berhane@aau.edu.et in the world in terms total number of reported disasters Center for African and Oriental Studies, Addis Ababa University, P.O. Box: from 1900–2014 (see Additional file 1: Figure A) while it 1176, Addis Ababa, Ethiopia is the second largest natural hazard, next to drought, in Full list of author information is available at the end of the article © The Author(s). 2017 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 2 of 19 terms of total number of affected persons between 1960 Table 2 Flood Disaster Statistics in Ethiopia between 1960 and 2013 and 2015 (see Additional file 1: Figure B). In Ethiopia, despite being given relatively less attention Year Occurrence Total deaths Affected Homeless Total affected as compared to drought, flooding has long been recog- 1968 1 1 10,000 6000 16,000 nized as one of the major environmental hazards that 1976 1 0 50,000 20,000 70,000 often develop into a disaster affecting the lives and liveli- 1977 1 7 16,000 0 16,000 hoods of people for many years. In fact, flooding and its 1978 1 9 1000 0 1000 damages are considered as a perennial phenomenon in 1981 1 0 20,000 0 20,000 the highlands (Disaster Prevention and Preparedness 1985 2 9 8000 20,000 28,000 Commission [DPPC], 1994). The country’s proneness to non-drought disasters such as floods has been limited in 1988 2 45 47,240 0 47,240 the past in terms of frequency and scope (DPPC, 1997a). 1990 1 0 350,000 0 350,000 However, the historical records on flood data suggests 1993 2 2 30,000 4800 34,800 that Ethiopia faced 47 major floods since 1900, which af- 1994 1 4 43,000 0 43,014 fected close to 2.2 million people (You and Ringer, 1995 1 27 93,875 0 93,875 2010). In this regard, many of the flood disasters oc- 1996 2 40 90,000 25,000 115,000 curred since 1980 (World Bank, 2010) (see also Table 1). This coupled with climate change and variability is likely 1997 2 326 65,000 0 65,022 to increase flooding as one of the major extreme events 1999 6 48 22,255 125,000 147,255 in the future posing a growing threat to many liveli- 2000 2 69 30,000 0 30,000 hoods (Intergovernmental Panel on Climate Change 2001 3 5 39,500 0 39,500 (IPCC), 2014; Savage et al., 2015). Flooding as a recur- 2002 1 22 4000 0 4000 rent environmental hazard is particularly felt in areas 2003 1 119 110,000 0 110,000 where people are already vulnerable to any adverse cli- matic event as a result of weakened resilience. For in- 2005 4 211 242,418 0 242,418 stance, an estimated 210,600 people were affected by 2006 7 951 434,050 0 434,146 flooding only within three months (November, 2015– 2007 2 17 245,386 0 245,386 January, 2016) (United Nations Office for the Coordin- 2008 3 45 115,595 810 116,440 ation of Humanitarian Affairs [UN-OCHA], 2016 p.1) 2010 2 19 80,700 0 80,700 A complete national and regional disaggregated data 2011 1 0 40,200 0 40,200 on flood disasters is limited in Ethiopia (see Table 2). However, the available literature indicates that some 2013 1 0 51,500 0 51,500 areas in the country are far more frequently affected Source: Authors’ computation from EM-DAT: OFDA/CRED International Disaster Database-www.emdat.be than others, to the extent of being labeled as ‘flood- prone areas’ (World Meteorological Organization (WMO), 2003; Nederveen et al., 2011; UN-OCHA, Tigray; North Gondar, North and South Wello, and 2016). These areas include central and western zones of Oromia zones of Amhara region which are often affected by flash flooding as well as those that are affected by Table 1 Total damage due to natural disasters between 1900 riverine floods, which include almost all the major river and 2013 in Ethiopia basins and the Tana Basin (DPPC 1994; 1997b; Nederveen Year Type Total damage ('000 US$) et al., 2011; UN-OCHA, 2016). 1906 Earthquake 6750 The Amhara region as indicated above is one of the 1969 Drought 1000 flood-prone areas in the country where severe and fre- 1973 Drought 76,000 quent floods affect a considerable number of people in recent years. In this regard, the limited available data on 1994 Flood 3500 the effects of floods in the region indicate that riverine 1998 Drought 15,600 floods were recorded in 1966, 1967, 1974, and 1975. 1999 Flood 2700 Severe flash floods have also been recorded in 1993 and 2005 Flood 5000 1996, with 72,569 people being affected. And a severe 2005 Flood 1200 flooding in 2006, has affected 107,286 people, displacing 2006 Flood 3200 37,982, damaging crops on 18,000 ha of land in six zones (Disaster Prevention and Preparedness Agency [DPPA], 2013 Flood 2200 2007; Nederveen et al., 2011; UN-OCHA 2016). More- Source: Authors’ computation from EM-DAT: OFDA/CRED International Disaster Database-http://www.emdat.be/ over, seven districts in the region, which are all found Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 3 of 19 around Lake Tana, are particularly well known for being Frankenberger and Nelson, 2013). Following Maru et al. frequently affected by both flash and riverine floods. (2014), this study argues that there is a need to combine One of these areas is Dembia district in North Gondar the two concepts since both are concerned with features zone, which is highly affected by Megech, Derema and that affect people’s ability to cope with and respond to Gumero rivers that frequently overflow their banks change. affecting the nearby settled plains (DPPA, 2007) (see In dealing with resilience, it is important to define “re- Additional file 1: Table A). silience to whom” and “resilience of what” (Cutter, 2016 Flooding in Dembia district, has become all too com- p.1). Accordingly, livelihood resilience as the building mon in recent years, and remains the most serious chal- block of this study is conceptualized as “the capacity of lenge to peoples’ livelihood with its short and long-term all people across generations to sustain and improve effects. As a result, some people were forced into desti- their livelihood opportunities and well-being despite en- tution (UN-OCHA, 2006; DPPA, 2007; You and Ringer, vironmental, economic, social, and political distur- 2010; Kreft et al., 2016; UN-OCHA, 2016). When this, bances” (Tanner et al., 2015 p.1). However, one of the coupled with the increasing flooding scenario predicted main deficiencies in the literature so far has been the by the reports of the IPPC (2007; 2012; 2014) amplify failure to identify the root causes of vulnerability as an the magnitude of the problem. Furthermore, the prob- initial step to understanding resilience owing to discip- lem of flooding would particularly be worse for countries linary perspectives and focus limited dimensions (Cutter, like Ethiopia with the majority of its population sub- 2016). This in turn resulted in lack of working defini- jected to poverty and vulnerability to climatic shocks tions, key indicators, and valid measurements for the (Berhanu and Fekadu, 2015; Ethiopian Panel on Climate concepts of vulnerability and resilience in the literature Change [EPCC], 2015; Savage et al., 2015). This in turn, (Alfani et al., 2015; Bahadur et al., 2015; Razafindrabe justifies the need to study flooding as a livelihood prob- et al., 2015; Cutter, 2016). lem since it creates downward pressures on livelihoods. Most studies conducted on natural disasters and their The understanding of flooding as a livelihood shock also effects on peoples’ livelihoods in different parts of needs an analysis of resilience of livelihood systems in Ethiopia focused mainly on drought and overlooked the face of the recurring flood disasters. flooding and its impacts (Woldemariam, 1986; Rahmato, The concept of resilience has recently been widely 1991; Sharp et al., 2003; Rahmato, 2007). The few avail- promoted in many fields such as engineering, psych- able studies on floods also focus on issues such as risk ology, and ecology, very recently resilience has become perceptions and risk management strategies (Moges, widely used by humanitarian and development actors 1978; Bekele, 2003; WMO, 2003; Nederveen et al., working across diverse thematic areas including, disaster 2011). Although there are recent studies that looked into risk reduction, climate change, ecosystem management, resilience and vulnerability in Ethiopia, flooding and its and food and nutrition security (Frankenberger et al., impact on livelihoods has not been investigated (Deressa 2012; Constas and Barrett, 2013; Maxwell et al., 2013; et al., 2008; Simane et al., 2014; Mengistu et al., 2015). Hoddinott, 2014; Razafindrabe et al., 2015). Building re- This is a key gap in the existing empirical studies given silience of households, communities, and systems has that flooding is a major natural hazard that affects the also been considered as a crucial policy objective among livelihoods of thousands of smallholder farming commu- various development frameworks including, the Sendai nities every year across the country (see also Table 2). Framework for Disaster Risk Reduction (United Nations This study therefore addresses the gap in the literature [UN], 2015a), the Paris Agreement on Framework by looking into the root causes of vulnerability and Convention on Climate Change (UN, 2015b), and the measuring livelihood resilience of smallholder farmers to Sustainable Development Goals (UN, 2015c). Resilience flood hazards. Linking livelihood approaches to resili- harbors different meanings in different contexts. In dis- ence thinking is imperative to enhance the understand- aster risk reduction, it is broadly viewed as a concept ing of livelihood dynamics and to explore how that deals with a system’s capacity to anticipate, to cope, households maintain and improve their livelihoods in to absorb, adapt to, and recover from the adverse impact the face of natural disasters (Scoones, 2009; Sallu et al., of hazards and reduce vulnerability (Razafindrabe et al., 2010). In view of this, the study contributes to the disas- 2015; Tanner et al., 2015). The concept of vulnerability ter risk reduction literature by providing empirical evi- is often contrasted with resilience; however, it is an dences on the determinants of vulnerability and interlinked function of exposure, sensitivity, and adaptive resilience to the recurring flood hazards. The study also capacity (Adger, 2006; IPCC, 2014). Being a common in- adds to the conceptual and methodological debates sur- dicator, adaptive capacity, can be taken as a desirable rounding vulnerability and resilience by focusing on the characteristic of a system that minimizes vulnerability least studied hazard in Ethiopia and developing and while enhances resilience at all levels (Engle, 2011; applying context-specific indices. This would further Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 4 of 19 contribute to the application of relevant measurements and scoring exercises. These data were obtained between in relation to capturing the multidimensional nature of March-May, 2015. both vulnerability and resilience. Finally, the study high- lights the synergy between the vulnerability and resili- Sampling and sample size ence , which need to be fostered, if the objective of In selecting the sample households for the survey, a achieving Sustainable Development Goals (SDGs) in multistage sampling procedure was employed. In stage rural parts of developing countries is to be addressed in one, eight Kebeles were selected from the 40 rural years to come (Fig. 1). Kebeles in the district purposively as they are frequently hit by seasonal flooding. In stage two, two Kebeles (Tana Methods Weyna and Gur-Amba) out of the eight flood prone Research design Kebeles were selected purposively using pre-defined cri- A quantitative-dominant, qualitative mixed research de- teria. The criteria include, the physical proximity to sign was employed, where the quantitative data and flood hazard source particularly to the nearby rivers (lo- qualitative information were collected concurrently. This cation and exposure) and the severity and frequency of approach helped the study to assess how vulnerability flood-disasters. After selecting the two Kebeles, a list of and resilience are conceived in local contexts, examine the households in 26 villages (15 in Tana Weyna and 11 locally-specific impacts of flooding, and factors that in Gura-Amba) was recompiled and used as a sampling shape the resilience of households in the face of this frame to select the households. Thus, a final sample of disaster. 256 households out of the 971 households were selected using systematic random sampling technique. Data sources Quantitative data and qualitative information for this For the qualitative interviews, both KIIs and FGD study were obtained from both primary and secondary participants were selected purposively using criteria that sources. A cross-sectional survey of 284 farm households includes being born in a particular village or lived there was supplemented by qualitative information from 12 for not less than two decades; have a first-hand experi- Key Informant Interviews (KIIs), two Focus Group Dis- ence of flood disasters; and being knowledgeable about cussions (FGDs), field observations, and Participatory the local environment, weather patterns and climate. Rural Appraisal (PRA) tools including problem ranking This was meant to capture the spatio-temporal perspectives Fig. 1 Location of the study district The map shows Dembia district of the Amhara region, Northwestern Ethiopia. It is located at 12°18'30''N and 37°17'30''E (see Fig. 1). It has an area of the 148,968 ha from which plain land accounts for about 87%, mountain and hills 5%, valleys and gorges 4.8% and water bodies 3.2%. The altitude of the district ranges from 1850 to 2000 m.a.s.l. Therefore, it is predominantly classified as Mid-land agro-ecology and the slope ranges from 2 to 4%. The district on average receives an annual rainfall between 700 mm to 1160 mm. Belg (the short rains February-April) and Meher (the long rains June-September). The average yearly minimum and maximum temperature is 18 °C and 28 °C respectively. Based on the recent Central Statistical Authority’s (CSA) population projection, the district had an estimated total population of 307,967 (CSA, 2013). Out of this total population, the majority, about 90%, are rural residents with an average agricultural household size of six persons. Source: Authors’ based on Ethio-Geographic Information System (GIS) and (CSA, 2007) Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 5 of 19 of the studied households and communities about flood di- The LVI measurement largely fits to the study context sasters based on recall. and target population (i.e., smallholder communities in sub Saharan Africa) and similar sample size based Data collection instruments on primary data obtained through a cross-sectional A structured survey questionnaire was designed and survey. The LVI also helps to capture the key factors piloted in order to generate information on house- that reflect the vulnerability situation of smallholder holds’ socio-economic and demographic characteris- farming communities in the face climate induced en- tics, livelihood asset profiles, livelihood activities, and vironmental hazards. Similar to the LVI used in Hahn income portfolios. The questionnaire also consisted et al. (2009), this study employed seven key variables, questions related to households’ vulnerability situa- which relate to socio-demographic characteristics tions, including indicators relating to exposure, sensi- (SDC) (household size, dependency ratio, age, gender tivity, and adaptive capacity. Moreover, questions of household head and education), livelihood strat- pertaining to absorptive and transformative capacities egies (LS), health status (HS), food security status of resilience were added while adaptive capacity indi- (FSS), access to water (AW), social network (SN), and cators were used as common indicators for both vul- flood disaster (FD) and its impact. Moreover, following nerability and resilience. Both interview guides and Madhuri et al., (2014) and in line with the Sustainable discussion checklists were designed to gather qualita- Livelihood Framework (SLF) (Birkmann, 2006; Scoones, tive information to supplement the household survey. 2009) this study further included natural capital (NC) that Smallholder farmers, community members, govern- mainly refers to ownership of land and size of farmland. ment and non-governmental organization representa- tives, and leaders of Community Based Organizations Calculating the LVI (CBOs) were considered as key informants and FGD The dimensions of vulnerability were systematically participants. Accordingly, 12 KIIs, two mixed FGDs combined with equal weights to create an index on a consisting of 20 people (12 men and eight women), scale of 0 to 1. As in the case of the computation of the and two PRA exercises were carried out with the same life expectancy index of the Human Development Index FGD participants. (HDI), the computation of each indicator of the vulner- ability index followed the process of standardization Approaches to measuring vulnerability and resilience (Hahn et al., 2009). In terms of measurement, Deressa and Hassan (2009) S −S a min documented the two commonly used approaches (i.e., I ¼ ð1Þ S −S max min econometric and indicator based) to measure vulnerabil- ity to disasters, including flooding. In the earlier case, Where, I is the standardized value of each indicator. the use of econometric method such as regression ana- S the original sub-component for household a, S is a min lysis is commonly employed to construct the Livelihood the minimum value of the indicator across all house- Vulnerability Index (hereafter referred to as LVI). The holds, and S is the maximum value of the indicator max drawback of this technique is, however, the challenge as- across all households. After each indicator was stan- sociated with testing various econometric assumptions dardized, the average value of each component was concerning the standard errors, hypotheses, confidence calculated using equation 2: intervals and imputing causality without making strin- gent assumptions (Etwire et al., 2013). In the latter case, I i a¼1 M ¼ ð2Þ it involves the selection of indicators that the researcher finds to largely account for the vulnerability (Deressa Where M is the one of the eight components for and Hassan, 2009). In this approach, the subjectivity of household a, I i indicates the sub-components indexed the variable selection process is considered as a limita- by i, which builds each major component, and n is the tion (Etwire et al., 2013). Although this is a major number of sub-components of each major component. limitation of the indicator based approach, recently dif- After obtaining values for each of the eight components, ferent scholars used this approach to construct LVI in the household level LVI was obtained by combining different contexts, including Ethiopia (Etwire et al., 2013; these components using equation 3: Limsakul et al., 2014; Madhuri et al., 2014; Simane et al., 2014). Similarly, this study adapted indicator based ap- w M i M a i¼1 i proach to develop LVI of smallholder farm households LV I ¼ ð3Þ a P i¼1 i in the study district. LVI developed by Hahn et al. (2009) was applied to as- sess the vulnerability of households in the study area. Which can be further expressed as: Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 6 of 19 w SDC þ w LS þ w HS þ w FSS þ w AW þ w SN þ w FD þ w NC SDC a LS a HS a FSS a AW a SN a FD a NC a LV I ¼ ð4Þ w þ w þ w þ w þ w þ w þ w þ w SDC LS HS FSS AW SN FD NC Where LVI , is the Livelihood Vulnerability Index for AC is adaptive capacity for household a a a household a, which equals the weighted average of eight ABC is absorptive capacity for household a major components, w .The weights of each major com- TC is transformative capacity for household a M a ponent are given by the number sub-component that Using the FAO’s Resilience Index Measurement and make up each major component, which are used to Analysis (RIMA) model (Food and Agricultural Organization guarantee that all sub-components have equal contribu- [FAO], 2012; Alinovi et al., 2015) equation 5 can be further tion to the total LVI (Sullivan, 2002; Hahn et al., 2009). expressed as: The LVI value ranges between 0 and 1, where 0 denotes the least vulnerable while 1 implies the most vulnerable LRI ¼fIðÞ FA ; ABS ; A ; SSN ; S ; AC ð6Þ a a a a a a a (Etwire et al., 2013; Madhuri et al., 2014). Resilience is a multidimensional concept that blends Where: relevant evidence as to how people really withstand IFA refers to income and food access; ABS = access to shocks (Almedom, 2009). Though the concept of resili- basic services; ence has been popular in development studies including, A = assets; SSN = social safety nets; S = stability; AC = poverty, vulnerability, and food security, it has been adaptive capacity. Since the indicators used in RIMA challenging to find a sound measure to resilience and have been applied to measure household’s resilience cap- how to quantify resilience remains controversial (Alfani acity to food insecurity (FAO, 2012; Alinovi et al., 2015), et al., 2015; Béné et al., 2016). However, some empirical in this study, the RIMA components were contextual- studies have attempted to measure resilience using a ized and subsumed to into the three resilience capacity composite index as proxy indicator (Amaya, 2014; Alfani indicators to measure households’ resilience to flood et al., 2015; Alinovi et al., 2015; Béné et al., 2016; Smith disasters. Accordingly, IFA, A, and AC indicators were et al., 2015). The current understanding of the resilience taken as part of adaptive capacity along with other indi- entails three interrelated capacities (adaptive, absorptive, cators; S was captured under absorptive capacity indica- and transformative), which are relevant to its measure- tors using sensitivity to flood disasters as a proxy ment (Amaya, 2014; Frankenberger et al., 2014; Bahadur indictor in addition to others; and SSN and ABS were et al., 2015; Béné et al., 2016). included under transformative capacity . Resilience being a context-specific concept, the dimen- As this index was a rough approximation of resilience sions and indicators may change depending on the con- and scale-sensitive, which may not be useful for inter- text. In assessing resilience to flood disasters, most household comparative analysis, a composite index using studies use ex-post resilience indicators as opposed to ex Principal Component Analysis (PCA) was constructed. ante measurements partly because the debate in the re- PCA is a multivariate statistical technique mostly used silience literature regarding the possibility of measuring for data reduction (i.e., larger number of variables into resilience in the absence of a hazardous event is unset- smaller numbers of components) and express the data tled (Keating et al., 2014). Therefore, the SLF was as a set of new orthogonal variables called principal adapted and built a resilience index using five capacity components (PCs) (Abdi and Williams, 2010; Abson dimensions: social, economic, institutional, infrastruc- et al., 2012; Schürer and Penkova, 2015). In this study, ture, and community capacities with each having specific PCA was used both for data reduction and identification indicators. These indicators are then aggregated by equal of the dominant factors that explain household’s resili- weighting into the three components–adaptive, absorp- ence to flood disasters. tive, and transformative capacitates to obtain a multidi- There are number of ways that can be used to retain mensional livelihood resilience index (LRI), following principal component score. In order to obtain PCs, the similar steps used in the LVI computation as given by study used Kaiser criterion of extracting factors with equations 1 to 4 (Amaya, 2014; Frankenberger et al., eigenvalues greater than one, which is one of the fre- 2014; Suman, 2014; Smith et al., 2015). Thus, LRI con- quently used technique (Abdi and Williams, 2010; Mooi structed is expressed as: and Sarstedt, 2011; Abson et al., 2012; Schürer and Penkova, 2015). Thus, the heaviest loading of principal LRI ¼fAðÞ C ; ABC ; TC ð5Þ a a a a component expressed in terms of the variables as an Where, index for each household that captured largest amount LRI is the resilience index for household a of information (Abson et al., 2012). The individual a Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 7 of 19 resilience score using PCA was computed using equa- at peak flows in the main rainy season starting from July tion 7 as follows: through August with water volume showing declines only in late September. As a result, the rivers regularly RS ¼ f XaðÞ −a =ðÞ s þ …… inundate many nearby villages with water staying on the a a1 1 1 þ fðÞ a −a =ðÞ s ð7Þ plains for several weeks. However, some severe floods aN N N have occurred in the past that are associated with heavy Where RS is the resilience score for household a; rainfall in the highlands. For instance, the floods of f is the component loading generated by PCA for the 1995/96, 2001, and 2006 were mentioned by FGD partic- first variable; ipants as the most severe flood disasters. Secondary re- th a is the a household’s value for the first variable; cords obtained from DPPA corroborate this information a1 a and s are the mean and standard deviation respect- and show that severe floods also occurred much earlier 1 1 ively of the first variable overall the households. in 1973/74 and 1982/83 in the district (DPPC, 1997a). After extracting the principal components, a simple A change in the severity of floods was also noted by linear regression was applied to check the relationship experts and other study participants. People felt that between resilience and vulnerability as used in a similar flooding is more severe and frequent than in the past. study (Madhuri et al., 2014). Apart from this, following Most of them came to understand that the population Nguyen and James (2013) a dichotomous response items pressure and the associated farmland expansion have were used to capture subjective indicators of household brought people close to the rivers which made them resilience to flooding. These indicators were quantified more vulnerable to flooding. This view partly agrees with and integrated by using an exploratory factor analysis the major assertion in the literature which relates to (Costello and Osborne, 2005; Child, 2006; Preacher causes of people’s vulnerability both to socio-economic et al., 2013). and contextual factors compared to the mere exposure to floods (Handmer, 2003; Cannon, 2006). Results and discussions In contrast, some participants stated that rivers have The nature of flooding and effects on livelihoods begun inundating farmlands and villages by changing Natural hazards such as floods and droughts often ex- their natural courses. For instance, one expert men- pose poor communities to vulnerabilities that can be tioned Megech River as one of such rivers that have investigated from two dimensions (1) external dimen- changed its natural course since the 2006 flooding. The sions or vulnerability context which can be expressed as river (Megech) is now flowing in a new channel which is the exposure to circumstances beyond people’s control, too narrow and shallow, causing the river to meander including shocks, trends and seasonality (2) internal di- and spread out onto the plains easily overflowing its mensions which refers mainly to socio-economic sys- bank, flooding several villages in Tana Weyna Kebele. tems, access and use of resources to the extent to which Two points stand out from the above findings, (1) peoples’ livelihood is affected by the exposure to external riverine flooding is the major type of flood in the study factors (Blaikie et al., 2014; IFRCS, 2015). area (2) the nature of the flooding in the area is showing In view of this therefore, the nature of flooding in the a marked change in terms of its severity having major study area in terms of its cause, magnitude, severity, fre- consequences on the lives and livelihoods of people in quency and duration is discussed as a major component the area. This finding is consistent with evidences from of the vulnerability context of people. Alongside this, by other studies in Ethiopia that suggest increasing fre- drawing together the findings from the household sur- quency of flood hazards. In this regard, Maxwell et al. vey, the FGDs and the interviews with key informants (2013) in their study of Tsaeda Amba district in Tigray on the effects of flooding on the livelihoods of people is (Northern Ethiopia) find that there is an increasing ten- discussed with the perceptions of people towards flood- dency of run-off and flooding due to environmental deg- ing as a livelihood threat. radation. Similarly, a study by Tesso et al. (2012) Flooding in Dembia district is a seasonal phenomenon. indicates frequent flooding as a major environmental The district is situated bordering the biggest lake in hazard that erode the coping capacities used by vulner- Ethiopia–Lake Tana. Several rivers that spring from able communities such as kinship support network in neighboring districts drain into Lake Tana traversing the North Shewa Zone of Oromia region. Focusing on river- district. According to the information obtained from the ine floods, a recent study by Hallegatte et al. (2016) that district Disasters Prevention and Preparedness Desk, the assessed the socioeconomic resilience to floods in 90 major cause of flooding in the area can be attributed to countries also found that, for poor people, a major risk the over-flow of five major rivers namely, Megech, associated with flood hazards is the loss of wellbeing. Derema, Nededit, Gumara, and Senzelit during the rainy Other factors that contribute to and aggravate the flood- seasons. According to key informants, these rivers reach ing in the area were also revealed by FGDs and key Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 8 of 19 informant interviews. The soil type of the district was the tributary rivers sending huge amounts of water into mentioned as one such factor. According to informants the nearby plains and beyond. As a result, thousands from the district Agriculture and Rural Development Of- have lost their assets and were dislocated from their fice, the black clay soil [which is the dominant soil type homesteads. Flooding has shown an increase in its inten- in the district] aggravates flooding as it is poor in its sity in the flood prone villages since then particularly drainage capacity and gets saturated and sticky with after the river Megech has changed its course and begun even a small amount of rainfall. It also fails to absorb flowing in a shallow bank crossing major settlement additional water flowing from rivers, contributing to areas, farm and grazing lands. flooding and water logging. The findings from the household survey indicate that Although the nature of the watershed and soil type in crop damage is most the common type of economic loss the area can be mentioned as factors that influence the experienced by households in the study villages. Accord- occurrence of flooding in the district, it is hardly pos- ingly, nearly all surveyed households (98.3%) reported sible to attribute the cause and the occurrence of flood- that they have experienced loss of crops due to flooding ing only to these factors. In fact, all of the district in the last five years before the survey. Through the agricultural experts interviewed about the cause of problem ranking and scoring exercises, participants of flooding mentioned that flooding in the district is partly the FGDs also indicated that crop damage is the fore- attributable to the following human activities that played most impact of flooding in economic terms. The loss of a greater role in determining flood damage. standing crops such as teff was substantial during the floods in 2006 and 2009. During the FGDs in both vil- 1. Deforestation: This made the highlands barren by lages, it was noted that farmers were compelled to exposing the top soil to heavy erosion and increasing change the cropping pattern from teff and wheat in to the run-off of rain water from these areas to low finger millet in recent years. In addition, almost all par- areas. Periodic changes in the amount and intensity ticipants and key informants indicated that farmers in of rainfall aided by the lack of vegetation cover in the study villages have begun to rely more on secondary the highlands also help in aggravating the run-off crops such as chick-peas, field peas, and faba beans and and the flooding in the study area. other leguminous crops which grow by using the 2. Traditional and subsistence-oriented farming system residual moisture left in the soil in the dry seasons. in the highlands was mentioned as a factor that However, the overall production of cereals and pulses causes and accentuates the rate of soil erosion and has gone down in recent years owing to the loss of soil run-off in the study area. According to the opinions fertility as a result of sedimentation which creates suffo- of the district Agriculture and Rural Development cation to such crops. In addition, the humidity of the experts, some irresponsible local farming practices soil resulting from flooding creates a favorable condition such as tilling hilly lands have increased the problem for pests such as Cut Worm (Agnotis Segetum) that re- of run-off and thereby contributed to increase duces the productivity of the crops. In relation to this, flooding in low lying areas. one of the participants of FGDs in Tana Weyna Kebele 3. Lack of integrated conservation activities and disclosed that: watershed management was also mentioned as contributing factor to the rise in the frequency of …flooding is making the cultivation of crops a flooding as well as the increasing human challenging task. During the rainy season, it washes vulnerability in the area. away crops that we grow spending so much labor and time and when we plant secondary crops, Korache In general, the district’s geographic location, topog- [Cut-worm] destroys it. raphy and soil type aggravated by the effects of human intervention such as deforestation, traditional cultivation The loss of primary crops and the declining product- practices and lack of sustainable water-shed measures ivity of secondary crops suggest that exposure to food were found to cause or exacerbate flooding in the study insecurity is inevitable for the affected households. villages. The effect of flooding on the food security of house- holds is also amplified by the loss of production as a Effects of flood disaster on livelihoods result of the time spent on recovery and rehabilitation Flooding has been affecting the study villages for years. in the aftermath of the flooding. Flooding has also According to the district agriculture and rural develop- increased the vulnerability of households to food inse- ment office, the study villages experienced one of the curity as attested by the increasing relief grain re- worst floods in 2001 caused by the heavy rainfall in the quests made by the District Agricultural and Rural highlands that increased the volume of Lake Tana and Development Office. Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 9 of 19 Households’ vulnerability as measured in LVI However, the lack of access to alternative income earn- The LVI that measures the vulnerability of households ing activities in the district, coupled with the severity of to flood disasters indicates that most households are the recent floods. These floods were mentioned to have highly vulnerable to flooding with a mean value being adverse impacts on most farmers. Lack of human cap- around 0.5. The LVI shows the inter-household differ- ital, particularly labor was reported to be a major factor ences in terms of exposure, sensitivity, and adaptive cap- that heightens households’ vulnerability situation. acity. Accordingly, the major contributing factor to the Focus group discussants agreed that the degree of ex- high vulnerability of households to flooding in the study posure to flooding is mainly determined by the physical area is found to be exposure with a mean index value of proximity of farmlands and settlement areas (villages). 0.65 followed by sensitivity with a mean index value of Poor asset holdings mainly farmlands and oxen were re- 0.56 out of 1 (Fig. 2). Thus, most households are highly ported to be sources of social vulnerability. In the FGDs exposed to flooding and more sensitive to flood-related and KIIs, it was repeatedly noted that physical exposure risks such as gully erosion resulting in the loss of both to floods (physical vulnerability) was the major factor farm and pasture lands (see Fig. 3a). Studied households that puts studied households’ livelihoods at risk. In view are also found to have relatively low adaptive capacity this, it was vividly indicated the “better-off” households with a mean index of 0.53 out of 1. This implies that the in terms of asset holdings were highly affected by flood studied households have limited capacities in terms of disaster, which resulted in to the loss of assets accumu- offsetting flood disasters by employing long-lasting lated over time. The major floods that occurred in the methods such as constructing flood dykes, which is only 2006, 2008, and 2012 rainy seasons were mentioned as reported to have been used by 31.08 percent of house- blatant examples of such phenomenon. This, however, holds (see Table 3). Instead, as field observation shows does not mean that asset holding did not contribute to many households largely rely on coping strategies mainly the resilience of households, it only confirms the fact plastering the basement of their huts with daub, which that not all households in the study area were exposed may not help to withstand more severe flood hazards, to floods to the same extent and therefore were not af- frequenting the area in recent years (see Fig. 3b). More- fected in similar ways. This view strengthens the evi- over, data from the household survey highlights that dence that exposure to flood events is a necessary but other frequently employed coping strategies include not sufficient factor in determining the vulnerability of changing crops (86.01%), relying on informal social livelihoods. For instance, participants of FGDs in Tana transfers (83.89%), and borrowing seeds (80.93%) (see Weyna Kebele, noted that the extent of flood damage on Table 3). standing crops, depends more on the proximity of a The results of qualitative interviews and discussions farmland to the river Megech as opposed to the asset corroborated the findings from the LVI. holding of the household. Accordingly, it was stated that Accordingly, participants of FGDs mentioned that households whose farmlands are located near to the households with adequate labor can engage in dyke con- river were exposed to more flood hazards both in the struction and timely drain their farmlands. Moreover, it short and long rains. was highlighted that such households are able to engage To establish the relationship between the resilience in both on-farm and off-farm activities and maintain and the vulnerability of households in the study area, their household income during times of extreme floods. Ordinary Least Square (OLS) regression was used with LVI as an explanatory variable and the resilience index obtained using PCA as a dependent variable. The result shows that the LVI decreases livelihood re- silience index by 6.73 points, statistically significant at less than 1%. The first component of the PCA, which captures the largest variability of the sub-components is considered for capturing the resilience of surveyed households, which is composed of adaptive, absorptive, and transformative capacities (Frankenberger et al., 2014; Béné et al., 2016; Smith et al., 2015). The first component indicates the dominance of adaptive capacity over other components. The relationship between the two indices is to be expected as resilience is often taken to be the flip side of vulnerability (IPCC, 2001; 2007). In Fig. 2 The three components of vulnerability Source: Authors’ own this study, adaptive capacity was taken as joint compo- construction from household survey (April 2015) nent shared between the LRI and LVI indices, however, Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 10 of 19 Fig. 3 a Gullies created by flooding. b House built with raised platform, plastered with mud to protect from floods. Source: Field observation in Debmia district (May 2015) absorptive and transformative capacities as the other significant determinants of resilience of households components of LRI that positively contribute towards (Table 5). Thus, those with higher educational levels and households’ overall resilience status seems to be rele- having relatively adequate farmlands are likely to have gated as the PCA only extracts the first component– more resilience. Most importantly, engaging in trade as adaptive capacity. Therefore, we argue that the compari- the highest form of diversified livelihood strategies is son between the two indices need further analysis that likely to increase LRI by 0.042 points, statistically signifi- captures the multidimensionality of both vulnerability cant at less than 1%. and resilience. The factor analysis results on the dichotomous re- sponse items also show that three of the six statements Households’ resilience capacity as measured by LRI express smallholders’ subjective resilience. The first Relying on PCA, factors that have eigenvalue greater component represents 22.8% of the variance and relates than 1 were chosen as resilience indicators. Accordingly, to greater reliance on social networks that contributes to the results from the PCA indicate that five of the 13 adaptive and absorptive capacities, for example in terms components have higher than one eigenvalues and rep- of borrowing seeds (Table 6). Here, crop damage being resent 62.7% of the total variance (Table 4 and Fig. 4). the most common type of economic loss experienced by Most of these variables belong to adaptive capacity indi- households in the study villages reflects the crucial im- cators and include household and demographic charac- portance of social capital in a household’s resilience. teristics (age, household size, education, and supply of At the time of disasters and soon after, people largely labor). Next to household and demographic characteris- count on their kinship networks, mutual aid, self-help tics, livelihood diversification, which mainly belongs to groups and indigenous organizations secure help and absorptive capacity, describes the resilience of house- support (Haines et al., 1996; Aldrich, 2012). However, as holds towards flood disasters in the study area. the frequency and severity of co-variate shocks such as Following Nguyen and James (2013) those factor flooding increases, the role of social networks begins to scores with the highest eigenvalue were used as a wane. This process came out in the FGDs where partici- dependent variable for further analysis in the exploratory pants have mentioned the severity of flooding in recent multiple regression. The result indicates that human and years as the main obstacle for relying less on kinship natural capital endowments mainly education and land networks and neighbors. Moreover, flooding has affected holding size as well as engagement in more diversified the majority people in neighboring villages so much so activities mainly trade seems to be positive and that it was impossible to rely on kinship networks. For instance, one key informant explained that since the heavy flooding of 2006, the frequency and severity of Table 3 Coping strategies for flood disasters Coping strategy Number Percent Table 4 Principal components of resilience indicators of Borrowing seeds 236 80.93 households Selling household assets 236 12.71 Component Eigenvalue Difference Proportion Cumulative Changing crops 236 86.01 Comp1 2.68293 .994965 0.2064 0.2064 Constructing flood dykes 232 31.03 Comp2 1.68796 .3136 0.1298 0.3362 Informal social transfers 236 83.89 Comp3 1.37436 .0426096 0.1057 0.4419 Relocating to higher grounds 234 58.97 Comp4 1.33175 .258619 0.1024 0.5444 Source: Authors’ own construction from household survey (April 2015) Comp5 1.07313 .0815582 0.0825 0.6269 Note: This is a multiple response item and therefore the percentage does not add up to 100 percent. Source: Authors’ own construction from household survey (April 2015) Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 11 of 19 Fig. 4 Resilience spider diagram of the major components the LRI. Source: Authors’ own construction from household survey (April 2015) floods are increasing in all villages as a result of which not necessarily signify clear boundaries as they are only households have to “rely on relief grains to sustain their used to facilitate the analysis process. In addition, they lives”. This opinion was also verified by data obtained do not show some causes of vulnerability such as ill- from the District Agriculture and Development Office nesses, divorces and similar idiosyncratic shocks that that showed an increase in relief grain recipients. contribute to the weakening of resilience. In general, resilience was understood as a state of hav- Through FGDs and interviews, it was possible to iden- ing strength to quickly recover from the damages caused tify major factors that limit the resilience and coping by flooding. A key component of livelihood resilience for capacity of households in the face of flood disasters. many participants of FGDs and key informants was ar- Accordingly, participants and informants have identified ticulated as the ability to regain pre-disaster level of liv- a range of factors that determine the resilience of house- ing without sustaining crippling damage to household holds, by focusing mainly on the major flood disasters assets that could push people further into poverty. that occurred in the study villages in the past ten years. Moreover, during the focus group discussions it was Since the majority of factors relate with livelihood indicated that flooding as a livelihood problem does not resources and access to them, attempt was made to as- affect households equally in the study villages. This im- sess the household livelihood situations by using a com- plies that the resilience of households is understood bination of qualitative and quantitative methods. Below, more in relative terms which further indicate the need the major factors that determine the resilience of house- to set some locally specific indicators in order to differ- holds are discussed. entiate households in terms of their level of resilience. In this regard, the FGDs made with farmers in the study Natural capital: land villages yielded some useful locally specific indicators In any rural community land is a basic productive re- that helped to measure the level of resilience of source, and access to it determines the wellbeing of a households. given livelihood. According to the findings of this study, Accordingly, the participants identified the location of however, farmland location, and fertility were indicated farmland, critical asset holdings such as a pair of oxen, to be more important than a mere access to land in de- the ability to draw help form relatives in other villages, termining the resilience of households in the face of and time taken to recover from the impact of the floods flood disasters. The FGDs and interviews made with the as some of the major indictors of the livelihood resili- study households indicated that the qualities as well as ence of households faced with flood- disasters in the the location of farmland are the key factors that limit or study villages (Table 7). The categories were also used in enhance the resilience of households to flood-induced the household survey to differentiate sample heads of shocks. In terms of location, the proximity of farmlands households roughly in to three groups namely, those to rivers was mentioned to have a significant role in with high resilience, those with medium resilience and determining the vulnerability and resilience of house- households with poor resilience or more vulnerable to holds more than the size and fertility of farmlands. This flooding. These three categories only show the level of finding was also supported by data obtained from the resilience of households in comparative terms and do household survey, in which farmland ownership was not Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 12 of 19 Table 5 Exploratory OLS on factors that determine LRI found to have a significant effect on the resilience of households as almost all of the households (92.8%) own Explanatory variables farmlands and those who do not own their own land Sex of household head 0.00495 were found to be equally distributed among the (0.0128) respondents. Age of household head 0.000716 This however cannot be taken to mean that access to (0.000569) land does not have a role in determining the resilience Educational status of household head 0.00545 of households. In fact, it could be argued that access to (0.00212) land may indirectly determine resilience. The detailed discussions with participants of FGDs and key infor- Household size 0.00129 mants also indicated that farmers with no land holdings (0.00429) are less resilient to the effects of flooding as compared Supply of labor 0.00657 to those who have land or can access land through vari- (0.00432) ous mechanisms. This, as mentioned by focus group dis- Incidence of illness dummy −0.00180 cussants and informants, was to be expected since the (0.0112) landless would lose their income largely drawn from ** wage labor on farms of other farmers during flooding Land size in ha 0.0379 and are likely to be affected even by moderate flooding (0.0120) as they lose the daily wages they earn from certain activ- Availability of farm oxen 0.0133 ities like weeding. Most participants of the FGDs also (0.0133) noted shortage of farmland in their respective villages. Social networks 0.0321 This problem, according to an informant from the (0.0168) Dwaro, have forced farmers, particularly the young ones *** to encroach the wetlands found on the shores of Lake Engagement in trade 0.0425 Tana for planting horticultural crops such as spices. This (0.0124) finding corroborates with results from other studies that Exposure to flood hazards −0.0166 reported small landholdings, land degradation, and (0.00996) population pressure as the major causes of vulnerability *** _cons 0.283 to disasters in other parts of Ethiopia (Rahmato, 2007; (0.0344) Tesso et al., 2012; Maxwell et al., 2013). N 214 2 Economic capital: financial asset R 0.144 Economic capital generally refers to the financial re- Standard errors in parentheses * ** *** sources that, in times of shocks could be used to reduce p < 0.05, p < 0.01, p < 0.001 Source: Authors’ own construction from household survey (April 2015) vulnerability and enhance recovery (Mayunga, 2007). Notes: The major forms of economic resource that were identi- The exploratory OLS model result has passed all the diagnostic tests such as multi-collinearity tests, omitted variables test, heteroscedasticity test and fied by the study households as having direct influence diagnostic plots to check the normality and linearity assumptions. on the resilience or coping capacity of households are discussed as follows. Livestock holding Focus group discussants in the two study villages have identified the size and type of livestock owned by a Table 6 Principal components/correlation of dichotomous household as a factor that determines the resilience of response items households. According to the focus group discussants, Component Eigenvalue Difference Proportion Cumulative households who own a large number of livestock tend to be more resilient to the effects of flooding as they use Comp1 1.36742 .30489 0.2279 0.2279 the animals as a buffer stock. This gave them the finan- Comp2 1.06253 .054492 0.1771 0.4050 cial capacity to quickly regain their livelihood, as they Comp3 1.00804 .086046 0.1680 0.5730 would sell their livestock and use the money to buy Comp4 .921993 .048510 0.1537 0.7267 seeds, rent-in farmlands for planting secondary crops Comp5 .873483 .106951 0.1456 0.8722 when the flood waters recede. Comp6 .766532 . 0.1278 1.0000 An interesting insight is also gained from the FGDs re- Source: Authors’ own construction from household survey (April 2015) garding the type of livestock and its role in the resilience Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 13 of 19 Table 7 Factors that affect the resilience of households and communities in the face of flooding in the study villages Factors Relatively resilient households Households with Medium resilience Households with poor resilience / more vulnerable households Time to recover from the 2- 3 months 6 months More than 6 months impacts of major floods Size of farmland 8-10 kada (2.0-2.5 ha) 4-8 kada (1.0- 2 ha) Less than 4 kada (1.0 ha) but mostly landless Livestock holding - Minimum 4 farm oxen - 2 cows - minimum 2 farm oxen - 1 farm oxen or none - 2 donkeys & 1 or 2 mules - 1 or 2 cows - no cows - 1 donkey - no pack animals Exposure to flooding Farm plots and homesteads Farm plots and homesteads Farm plots and homesteads located far from river banks located far from river banks located near the river banks or on the way where major rivers usually break their banks Availability of social Have relatives in other districts or occasionally draw some help Largely depend on relief grains at capital villages and are able to send their from relatives in other villages times of severe floods or resort to cattle to these places before the in the form of seeds or food taking loans from other households coming to the rainy season on grains at times of flooding regular basis. Source: FGDs and key informant interviews (April 2015) of the household. Accordingly, the participants of FGDs During the FGDs and interviews, it was also mentioned mentioned that possession of farm oxen often enhances that households with no oxen, land and other assets the resilience of households, since it gives the advantage were excluded from receiving loans as they were unable of draining flood water from farmlands so as to lessen to furnish collateral. In relation to this, a young inform- crop damage or failure. ant from Tana Weyna Kebele disclosed: “we are not However, focus group discussion participants and key given credit; they [ASCI] only give it to household heads informants alike agreed that flooding, with increased who own land”. This exacerbates their vulnerability to volume of water and duration, affected livestock and re- the effects of flood-induced shocks. versed the situation in recent years, in which those with During the discussions, it was also indicated that those more livestock were affected the most, since they lost who have better access to credit were in a better position their livestock during the floods through drowning and to withstand the aftermath shocks of flooding, as they in the aftermath through various diseases and lack of can replace their lost assets. Participants of the FGD fodder, which in turn affected their productivity. In view from the Gura Amba Kebele mentioned that there was of this one key informant said the following: good access to credit services as opposed to those in Tana Weyna Kebele. This difference in accessing credit A decade ago, farmers in our village used to keep could probably be explained by the differences in the de- many cattle. In fact, some farmers used to own as gree of physical proximity to the main credit provider much as 60 heads of cattle. Currently however we are i.e. Amhara Saving and Credit Institution. Some infor- having problems even to keep our farm oxen as the mants from Tana Weyna Kebele have also asserted that grazing fields are now covered with weeds and the credit service was not made available to farmers living in cattle are starving as they no longer find those fine most villages as the staffs of ASCI avoid remote villages grasses that used to grow in the fields. since there is a need to make frequent visits in attempt- ing to ensure repayments. Access to credit Generally, it can be argued that those households with Access to credit services was the other form of financial economic capital in the form of livestock and credit are capital, identified by household heads participated in the in a better position to withstand and recover from the study, as having effect on the resilience of households. effects of flooding as such assets contribute to their According to the household survey 36.9 percent of the resilience through creating more opportunities for liveli- respondents were able to have access to credit. And out hood diversification that enable households to manage of these, only 24 percent of them were able to receive and cope with flooding in more sustainable ways. loans from formal rural credit services (Amhara Saving Among those not having access to credit and economic and Credit Institution[ASCI]). This indicates that there assets, their resilience level is found to be very low. For is lack of access to credit, which is crucial in helping instance, among the 130 households, who reported hav- households to quickly recover from the effects of flood- ing no access to credit, only 9 (6.92 percent) were found induced shocks to replace lost assets and income. to have LRI above 0.5. Similarly, all the of the landless Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 14 of 19 households were found to be non-resilient (see Additional flood hazards and loss of assets disaggregated by resili- file 1: Table B). These results indicate the important role ence status. As indicted in the Table, the resilient and that these and similar economic assets play in determining non-resilient households provided more or less similar the resilience capacities of rural communities. assessments on their loss due to flood hazards except for The FGD participants in both villages mentioned that flood exposure. Further, looking at the educational status more resilient households have the capacity to engage in of households as one component of human capital that both on-farm and off-farm diversification activities and determines the resilience of households, the results from keep a relatively good stock of animals in neighboring dis- the survey showed that there the resilient groups are tricts that enables them to further off-set livelihood shocks better than the non-resilient ones. However, this differ- during major flood disasters like that of the 2006, 2008, ence is not statistically significant as a two-sample t-test and 2012 kremet floods. Diversification of income sources with equal variances gives a result of Pr (|T| > |t|) = is stressed in the literature as an important strategy of en- 0.1455. Thus, the evidences from the household survey hancing the resilience of vulnerable communities and it seems to concur with the qualitative information that “stands as the primary measure of household vulnerability underlines the importance of the degree of exposure to and resilience (Tesso et al., 2012 p 884; Nguyen and flood hazards and the associated human activities such James, 2013). Thus, given the benefits of diversification, as land use changes. The finding on the prominent role households that diversify their income sources are likely of exposure concurs with Doocy et al. (2013) that pro- to build their resilience to flood disasters in the future. vides a historical review of flood events worldwide from 1980 to 2009 and asserts that human vulnerability to Human capital floods is increasing among other things, mainly due to Human capital as referring to the level of education, health population growth, urbanization, and land use changes. conditions and availability of skilled labor was repeatedly The major components of the LRI for the studied mentioned as an important factor that shape the resilience households is provided in Table 9. As shown in the of households and communities to disaster-induced shocks Table, the mean LRI for all households is 0.44, which is in the literature (Adger, 2000; Mayunga, 2007). In this below the minimum threshold value–0.5. This indicates study, the availability of labor in the household was found that most households are not resilient enough to in the to be the most important form of human capital that face of the increasing flood hazards in the area. More- contribute to household resilience in the face of recurring over, from the sub-components of the LRI, one can see floods. that the studied households seem to have relatively The qualitative data obtained from interviews and higher absorptive capacity than adaptive or transforma- FGDs have also indicated that the availability of labor in tive capacities, a further indication of their vulnerability. a household play a determining role in enhancing the re- With the view of providing a more illustrative repre- silience of households. For instance, in explaining the sentation of studied households’ resilience capacity, we role of labor in household livelihood resilience an in- constructed a quadrant following the Andersen and formant in Gura Amba Kebele noted that “a farmer with Cardona (2013). The quadrant represents income per no asset can live by the sweat of his brow as long as he capita on the x-axis and LRI on the y-axis. Households is healthy and capable to work”. This clearly shows the falling in the right side of the mean values include, rich, value of labor in in terms of determining the resilience but not resilient groups, highly resilient, and extremely capacity of households. resilient groups (Fig. 5). Households in the left side of Table 8 provides a summary statistics of the responses the threshold include, poor, but resilient, highly vulner- of surveyed households with regards to exposure to able, and extremely vulnerable groups. In terms of the Table 8 Reported exposure to floods and loss of assets due to flooding Resilient group Non-resilient group Loss/damage to housing 63.41% (n = 26) Loss/damage to housing 69.66% (n = 124) Exposure to flood hazards 63.41% (n = 26) Exposure to flood hazards 56.74% (n = 101) Loss of crops due to flood hazards 95.12% (n = 39) Loss of crops due to flood hazards 99.44% (n = 177) Loss of livestock due to flood hazards 73.17% (n = 30) Loss of livestock due to flood hazards 73.6% (n = 131) Ownership of at least an ox for farming 78.05% (n = 32) Ownership of at least an ox for farming 79.78% (n = 142) Education (no. years of schooling) 3.43 (n = 41) Education (no. years of schooling) 2.68 (n = 178) Source: Authors’ own construction from household survey (April 2015) Notes: Resilient and non-resilient groups were identified based on the LRI index values, where households having an LRI value of 0.5 and above were taken as resilient groups while those with LRI below this threshold were considered to non-resilient Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 15 of 19 Table 9 Components of livelihood resilience index (LRI) The above quadrant is informative in terms of offering data as to where to focus development intervention ef- Variable Obs. Mean index Std. Dev. Min Max forts. In this regard, it is imperative to invest on various Adaptive capacity 222 0.55 0.07 0.17 0.73 livelihood resilience schemes that enhances the capacity Absorptive capacity 233 0.65 0.13 0.15 0.67 of highly and extremely resilient groups while focusing Transformative capacity 236 0.49 0.17 0.11 0.83 on reducing the number of highly and extremely vulner- LRI 219 0.44 0.07 0.18 0.62 able groups. Apart from this, it is also important to work Source: Authors’ own construction from household survey (April 2015) on empowering poor-but resilient households and rich but not resilience households. This is particularly im- portant given the overwhelming evidence, which indi- y-axis, the quadrant construction was based on the cates the likelihood of a shift in the global pattern and mean value of LRI, which was aggregated and/or com- intensity of flood hazards associated with climate change posed from adaptive capacity index, absorptive capacity (Few, 2003). index, and transformative capacity index. The LRI value ranges between 0.1-0.99 (the lowest being 0.18 and the Conclusions highest value stands at 0.62). The quadrant below and Focusing mainly on the vulnerability and resilience of above the mean and/or threshold value divide was based rural households in one of the flood prone areas in on 0.44 LRI value. The quadrant with the mean value Ethiopia- Dembia district, the study attempted to show above 0.44 consists of poor but resilient, highly resilient, that the nature of flooding in the study area has mark- and extremely resilient groups. While, the quadrant with edly changed over the past decade. The floods have be- the mean value below the mean includes, rich but not come more frequent and severe owing to a number of resilient, highly vulnerable, and extremely vulnerable factors that derive from both climatic and topographic groups. The average monthly income of households is conditions such as, periodic changes in the amount of about 10.26 USD, which reflects the level of poverty and rainfall, the nature of watershed system and soil type of depravation among the study communities as this would the area. In addition, certain human activities including mean that the average daily income of households is deforestation, increased settlement on flood plains, and only 0.34 USD. As can be shown from Fig. 5, even by traditional systems of cultivation were found to aggra- taking this low income level as a threshold, 31.9% of vate flood hazards in the area. households were found to be vulnerable. When roughly The findings of the study highlight the importance of extrapolated to the district level using CSA (2013) fig- access and use of livelihood resources such as size of ures, this proportion would mean that 88,417 people are farmlands, access to income diversifying options, credit vulnerable to flood hazards in the district out of the as well as ability to draw help from social networks in 277,170-rural population. terms of determining the resilience of households facing Fig. 5 Resilience typologies by income of households. Source: Authors’ own computation based on household survey (April 2015) Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 16 of 19 frequent flood hazards. The scale and impact of the considered as non-response cases (9.8% of the total recent floods and lack of basic infrastructural and social sample size). facilities are also found to have hampered the use of The size of a sample in purposive sampling is often robust coping strategies by affected communities and determined on the basis of “theoretical saturation” (the households. point in data collection when new data no longer pro- Given the livelihood context of smallholder farming vide additional insights to the research questions) (May, system in the studied area, which is highly vulnerable to 2002; Patton, 2002). environmental hazards and persistently challenged by The concern of livelihood approach is to understand population pressure and land degradation, it is highly how different in different places live (Scoones, 2009). likely that the size of farm land will remain to be a major Apart from being an analytical tool, SLF takes vulnerabil- determining factor of the resilience capacity of the stud- ity as a comprehensive concept covering livelihood assets ied households. Despite this, however, context specific and their access, and vulnerability context elements (i.e., institutional interventions such as the integrated use of shocks, seasonality, and trends) as well as institutional both safety nets and cargo nets may off-set livelihood structure and processes (Birkmann, 2006). predicaments. The safety nets can be implemented in To capture adaptive capacity, we used labor, education, the form of public works that are relevant to minimizing asset (income)/consumption/per capita, household size, exposure to the recurring flood hazards, particularly natural capital, and social capital. Absorptive capacity is through construction and maintenance of flood dykes. captured through access to credit, asset, diversification, The cargo nets can be put in place in the form of flood disaster exposure indices. The transformative cap- targeted microfinance, flood insurance schemes, or acity is measured by using access to services, infrastruc- agricultural input subsidization. These interventions will ture, and formal safety nets. strengthen both the absorptive and adaptive capacities of Very recently, FAO proposed RIMA-II, which is an in- households and communities in the short-term while direct measure of resilience that adopts regression analysis enhancing their transformative capacity in the long- allowing for making causal inference. However, RIMA-II term. These imply that policy should focus more on is more suitable for assessing the dynamic nature of addressing the factors that expose people to flood disas- household resilience to measurable outcomes such as food ters and shape their resilience, rather than focusing on insecurity, which requires the use of panel data. short-term emergency responses, which seems to be the Adaptive capacity (AC) indicators include: IFA (income norm in much of the flood affected areas in Ethiopia. and consumption per capita), A (availability of labor, own- ership of asset, and natural capital (land)), AC (educational Endnotes status). Other indicators included are: social capital (infor- Resilience as a concept has been highly promoted as mal transfers and participation in festive work groups) and a uniting policy instrument that links humanitarian and household size; Absorptive capacity (ABC) indicators in- development approaches to address peoples’ chronic vul- clude, S Access to credit, asset ownership, diversification of nerability to recurrent shocks and disasters (Choularton income, and flood index (flood duration, flood severity, et al., 2015). These view is also shared by the Sendai exposure to flood disasters, frequency of flood disasters, Framework for Disaster Risk Reduction (UN, 2015a) and and losses sustained due to flood disasters including crops, the UN’s Paris Agreement on Framework Convention on damage to housing, and loss of livestock); and Transforma- Climate Change (UN, 2015b). tive capacity (TC) indicators include, SSN (access to formal Kebele is the lowest administrative unit in Ethiopia. safety net (Productive Safety Net Program)) and ABS These criteria were used to account for variations in (access to services, access to infrastructure). the degree of flood-hazard exposure as all of the eight There are two major types of factor analysis tech- Kebeles are not equally affected by the flood disasters. niques (These are namely, Confirmatory Factor Analysis Hence, the two Kebeles were selected out of the eight (CFA) and Exploratory Factor Analysis (EFA). The former Kebeles to ensure the representativeness of the sample CFA helps to check hypotheses and uses path analysis drawn from the Kebeles. diagrams to denote variables and factors. The latter, EFA The overall sample size was determined by using the attempts to discover multifaceted patterns by exploring sample size determination equation that takes into the dataset and testing predictions (Costello and Osborne, account the desired confidence level (95%), the error 2005; Child, 2006). As for the rotation techniques. There margin (5%), and the prevalence of the issue under are two types, viz, orthogonal rotation and oblique investigation (p= 0.5). The required sample size was rotation. The first, orthogonal rotation (e.g.,Varimax and determined using Kothari (2004) sample size determin- Quartimax) consists of uncorrelated factors whereas ation formula. 28 households did not respond the major oblique rotation (e.g., Direct Oblimin and Promax) in- modules of the structured household survey and were cludes correlated factors. The interpretation of factor Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 17 of 19 analysis is based on rotated factor loadings, rotated eigen- Alinovi, L, E. Mane, and D. Romano. 2009. Towards the measurement of household resilience to food insecurity: An application to Palestinian households. European values, and scree test. In reality, researchers often apply Commission and the United Nations Food & Agriculture Organization more than one extraction and rotation technique relying Working Paper. Available at http://www.fsnnetwork.org/sites/default/files/ on pragmatic reasoning rather than theoretical reasoning measuring_household_resilience_to_food_insecurity.pdf. Accessed 20 Nov (Preacher et al., 2013). Thus, for the sake of brevity in in- Almedom, A.M. 2009. A call for resilience index for health and social system in terpretation, this study used varimax method of rotation Africa. Issues in Brief. The Federick S Pardee Center for the study of the variables that helps to reduce the number of variables with Longer‐Range Future. Number 10, October 2010. Retrieved from http://www. bu.edu/pardee/policy-010-resilience-index/. a high loading on a factor. Amaya, L.C. 2014. Disaster resilience to food insecurity metrics: A case study in rural Owing to the complex nature of the concept and the Costa Rica. PhD Dissertation. USA: Texas State University. lack of an exact equivalent of the word resilience in the Andersen, L.E., and M. Cardona. 2013. Building Resilience against adverse shocks: What are the determinants of vulnerability and resilience? Development local Amharic dialect, it was necessary to first obtain Research Working Paper Series, 2. farmers own subjective meanings of the term through Bahadur, A., E. Lovell, E. Wilkinson, and T. Tanner. 2015. Resilience in SGDs: Developing group discussions. Accordingly, the participants of FGDs Indicators for target 1.5 that is fit for the purpose,Briefing Paper,1–6. London: Oversee Development Institute. in both villages, agreed that the concept has a positive Bekele, L. 2003. Community perceptions and responses to Awash Floods in connotation in the sense that it matched with certain the Wonji environs, Ethiopia. Ethiopian Journal of Development Research terms like ‘ability’, ‘capacity’, ‘strength’ and ‘resistance’. 25(2): 1–33. Béné, C., D. Headey, L. Haddad, and K. von Grebmer. 2016. Is resilience a useful concept in the context of food security and nutrition programs? Some Additional file conceptual and practical considerations. Food Security 8(1):123–138. Berhanu, W., and B. Fekadu. 2015. Climate variability and household adaptation strategies in Southern Ethiopia. Sustainability 7: 6353–6375. Additional file 1: Statistics on the incidence and effects of major natural Birkmann, J. 2006. Measuring vulnerability to promote disaster-resilient societies: disasters. (DOC 314 kb) Conceptual frameworks and definitions. Measuring vulnerability to natural hazards 1: 9–54. Acknowledgements Blaikie, P., T. Cannon, I. Davis, and B. Wisner. 2014. At risk: natural hazards, We would like to thank the study participants for volunteering information on people’s vulnerability and disasters. London: Routledge. which this study is based. The constructive comments from two reviewers are Cannon, T. 2006. Vulnerability analysis, livelihoods and disasters. Risk 21: 41–49. sincerely acknowledged. Central Statistics Authority (CSA). 2007. Summary and statistical report of population and housing census: Population size by age and sex. Addis Authors’ contributions Ababa: Federal Democratic Republic of Ethiopia (FDRE). The corresponding author, ZBW is an Assistant Professor of Development Central Statistics Authority (CSA). 2013. Federal Democratic Republic of Ethiopia Studies at the Center for African and Oriental Studies (CAfOS), Addis Ababa (FDRE). Population projection of Ethiopia for all regions: At Wereda level from University. He designed the data collecting instruments, performed the 2014–2017. Addis Ababa: Federal Democratic Republic of Ethiopia (FDRE). statistical analysis, and drafted the manuscript. BEA is a PhD candidate at the Child, D. 2006. The essentials of factor analysis, 3rd ed. New York: Continuum Center for Environment and Development, Addis Ababa University. He International Publishing Group. participated in reviewing pertinent literature, collecting the data, performing Choularton, R., T. Frankenberger, J. Kurtz, and S. Nelson. 2015. Measuring shocks statistical analysis, and drafting the manuscript. Both authors contributed to and stressors as part of resilience measurement, Resilience Measurement and approved the final manuscript. Technical Working Group. Technical Series, 5. Constas, M., and C. Barrett. 2013. Principles of resilience measurement for food Competing interests insecurity: metrics, mechanisms, and implementation plans, Paper presented at The authors declare that they have no competing interests. Expert Consultation on Resilience Measurement Related to Food Security, Rome. Author details Costello, A.B., and J.W. Osborne. 2005. Best practices in exploratory factor analysis: Center for African and Oriental Studies, Addis Ababa University, P.O. Box: Four recommendations for getting the most from your analysis. Practical 1176, Addis Ababa, Ethiopia. Center for Environment and Development, Assessment Research and Evaluation 10(7): 1–9. Addis Ababa University, P.O. Box: 1176 Addis Ababa, Ethiopia. Cutter, S. L. 2016. Resilience to what? Resilience for whom? The Geographical Journal, n/a-n/a. http://doi.org/10.1111/geoj.12174. Received: 21 July 2016 Accepted: 24 February 2017 Deressa, T.T., and R. M. Hassan. 2009. Economic impact of climate change on crop production in Ethiopia: evidence from cross-section measures. Journal of African Economies 18(4): 529–554. Deressa, T., R. Hassan, and C. Ringler. 2008. Measuring Ethiopian farmers’ References vulnerability to climate change Across Regional States. In International Food Abdi, H., and L.J. Williams. 2010. Principal component analysis. Wiley Interdiscip Policy Research Institute (IFPRI) Discussion Paper 00806. Rev Comput Stat 2: 433–459. Disaster Prevention and Preparedness Agency. (DPPA). 2007. Regional summary of Abson, D.J., A.J. Dougill, and L.C. Stringer. 2012. Using principal component multi-agency flood impact assessment of 2006. Addis Ababa: Early Warning analysis for information-rich socio-ecological vulnerability mapping in Department. Southern Africa. Applied Geography:1–10. doi:10.1016/j.apgeog.2012.08.004. Disaster Prevention and Preparedness Commission. 1997b. Worst case scenario Adger, W.N. 2000. Social and ecological resilience: are they related? Progress in for drought, flood, influx of refugees and epidemics and the present Human Geography 24(3): 347–364. response system. June 1997: Addis Ababa: Federal Democratic Republic of Adger, W.N. 2006. Vulnerability. Global Environmental Change 16: 268–281. doi:10. Ethiopia (FDRE). 1016/j.gloenvcha.2006.02.006. Aldrich, D. P. 2012. Building Resilience: Social Capital in Post Disaster Recovery. Disaster Prevention and Preparedness Commission (DPPC). 1994. Non-drought Chicago: University of Chicago Press. disaster propensity in Ethiopia. Final report April, 1994. Addis Ababa: Federal Alfani, F., A. Dabalen, P. Fisker, and V. Molini. 2015. Can we measure resilience? a Democratic Republic of Ethiopia (FDRE). proposed method and evidence from countries in the Sahel. A Proposed Method Disaster Prevention and Preparedness Commission. (DPPC). 1997a. Flood vulnerability and Evidence from Countries in the Sahel, World Bank Policy Research Working in Ethiopia and needs for preparedness. June, 1997: Addis Ababa: Federal Paper, 7170. Democratic Republic of Ethiopia (FDRE). Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 18 of 19 Doocy,S.,A. Daniels,S.Murray,and T.D.Kirsch. 2013. The human impact of floods: A Keating, A., K. Campbell, R. Mechler, E. Michel‐Kerjan, J. Mochizuki, and C. Egan. historical review of events 1980–2009 and systematic literature review. PLOS Currents 2014. Operationalizing resilience against natural disaster risk: opportunities, Disasters 5(1). doi:10.1371/currents.dis.f4deb457904936b07c09daa98ee8171a. barriers and a way forward. Engle, N.L. 2011. Adaptive capacity and its assessment. Global Environmental Kothari, C.R. 2004. Research methodology: methods and techniques, 2nd ed. New Change 21: 647–656. Delhi: New Age International Publisher Ltd. Kreft, S., D. Eckstein, L. Dorsch, and L. Fischer. 2016. Global Climate Risk Index Ethiopian Panel of Climate Change (EPCC). 2015. First Assessment Report. In 2016. Who suffers most from extreme weather events? Weather-related Loss Working Group II Agriculture and Food Security. Addis Ababa: Ethiopian events in 2014 and 1995 to 2014. Briefing Paper,1–32. Germany: Academy of Sciences. Germanwatch e.V. Bonn. Etwire, P.M., R.M. Al-Hassan, J.K.M. Kuwornu, and Y. Osei-Owusu. 2013. Application of livelihood vulnerability index in assessing vulnerability to climate change Limsakul, A., W. Katasaenee, W. Paengkaew, A. Kammuang, D. Tipmanee, and P. and variability in Northern Ghana. Journal of Environment and Earth Science Sompongchaiyakul. 2014. Vulnerability index to climate change and its 3(2): 157–170. application for community-level risk assessment in Thailand. Environment Asia Few, R. 2003. Flooding, Vulnerability and coping strategies: local responses to a 7(2): 108–116. global threat. Progress in Development Studies 3(1): 43–58. Madhuri, H.R. Tewari, and P.K. Bhowmick. 2014. Livelihood vulnerability index Food and Agricultural Organization (FAO). 2012. Measuring resilience: A concept analysis: an approach to study vulnerability in the context of Bihar: original note on the resilience tool. Rome: The United Nations Food and Agricultural research, Jamba. Journal of Disaster Risk Studies 6(1): 1–13. Organization. Available at http://www.fao.org/resilience/resources/resources- Maru, Y. T., S.M. Stafford, A. Sparrow, P.F. Pinho, and O.P. Dube. 2014. A linked detail/en/c/317275/. vulnerability and resilience framework for adaptation pathways in remote disadvantaged communities. Global Environmental Change. .doi:10.1016/j. Frankenberger, T., and S. Nelson. 2013. Background paper for the expert gloenvcha.2013.12.007. consultation on resilience measurement for food security. In Paper Maxwell, D., B. Vaitla, G. Tesfay, and N. Abadi. 2013. Resilience, food security presented at Expert Consultation on Resilience Measurement Related to dynamics, and poverty traps in Northern Ethiopia, Analysis of a Biannual Panel Food Security, Rome. Dataset, 2011–2013. Medford: Feinstein International Center, Tufts University. Frankenberger, T., T. Spangler, S. Nelson, and M. Langworthy. 2012. Enhancing resilience to food security shocks in Africa. In Discussion Paper. May, T. 2002. Qualitative research in action. London: Sage Publications. Frankenberger, T.R., M.A. Constas, S. Nelson, and L. Starr. 2014. Current Mayunga, S.J. 2007. Understanding and applying the concept of community disaster approaches to resilience programming among Non-Governmental resilience: A capital-based approach: a draft working Paper prepared for the Organizations. In 2020 Conference Paper, vol. 7, 1–42. summer academy for social vulnerability and resilience building, 22–28 July 2007, Munich, Germany. Guha-Sapir, D., P. Hoyois, and R. Below. 2015. Annual Disaster Statistical Review Mengistu, A., A. Argaw, and T. Seid. 2015. Resilience of Ecosystems to climate 2014: The numbers and trends. Centre for Research on the Epidemiology of change. American Journal of Environmental Protection 4(6): 325–333. Disasters (CRED). (http://reliefweb.int/report/world/annual-disaster-statistical- Moges, A .1978. Flood risks and vulnerability in different regions of Ethiopia. review-2014-numbers-and-trends). Accessed 5 Feb 2016. Disaster Preparedness Planning Program. Relief and Rehabilitation Hahn, M.B., A.M. Riederer, and S.O. Foster. 2009. The livelihood vulnerability index: A Commission. Addis Ababa: Relief and Rehabilitation Commission (RRC). pragmatic approach to assessing risks from climate variability and change—A case study in Mozambique. Global Environmental Change 19(1): 74–88. Mooi, E., and M. Sarstedt. 2011. A concise guide to market research: The process, Haines, V. A., J.S. Hurlbert, and J.J. Beggs. 1996. Exploring the determinants of data, and methods using IBM SPSS Statistics. Heidelberg: Springer. support provision: Provider characteristics, personal networks, community Munich RE. 2015. Loss events worldwide 1980–2014 (1–10). Munich: Munich RE. contexts, and support following life events. Journal of Health and Social Nederveen, S., M. Abebe, F.V. Steenbergen, A. Tena, and G. Yohannes. 2011. Flood Behavior 37(3):252–264. based farming practices in Ethiopia: status and potential, Overview Paper Spate Hallegatte, S., M. Bangalore, and A. Vogt-Schilb. 2016. Assessing socioeconomic Irrigation 3. resilience to floods in 90 countries. In World Bank Policy Research Working Nguyen, K., and H. James. 2013. Measuring household resilience to floods: A case Paper 7663. Washington DC: World Bank Group. study in the Vietnamese Mekong River Delta. Ecology and Society 18(3):13. http://www.ecologyandsociety.org/vol18/iss3/art13/. Handmer, J. 2003. We are all vulnerable. Australian Journal of Emergency rd Management 18: 55–60. Patton, M.Q. 2002. Qualitative research and evaluation methods (3 ed.). Hirabayashi, Y., R. Mahendran, S. Koirala, L. Konoshima, D. Yamazaki, S. Watanabe, Thousand Oaks: Sage. and S. Kanae. 2013. Global flood risk under climate change. Nature Climate Preacher, K.J., G. Zhang, C. Kim, and G. Mels. 2013. Choosing the optimal number Change 3(9): 816–821. of factors in exploratory factor analysis: A model selection perspective. Hoddinott, J. 2014. Understanding resilience for food and nutrition security. 2020 Multivariate Behavioral Research 48(1): 28–56. Conference Paper 8. Addis Ababa: International Food Policy Research Rahmato, D. 1991. Famine and survival strategies: a case study from Northeast Institute (IFPRI). Ethiopia. Nordic Africa Institute. Rahmato, D. 2007. Development intervention in Wollaita, 1960s-2000s: a critical Intergovernmental Panel on Climate Change (IPCC). 2001. Climate Change 2001: review. Forum for social studies monograph 4. Addis Ababa: Forum for Social The Scientific Basis. IPCC Third Assessment Report. Studies. Intergovernmental Panel on Climate Change (IPCC). 2007. Climate Change 2007. Impacts, adaptation and vulnerability: Working Group II contribution to the Razafindrabe, B., M.B. Cuesta, R. He, K. Rañola Jr., S. Yaota, Inoue, and A. Santos-Borja. Fourth Assessment Report of the Intergovernmental Panel on Climate 2015. Flood risk and resilience assessment for Santa Rosa-Silang sub-watershed in the Laguna Lake region, Philippines. Environmental Hazards 14(1): 16–35. Change (IPCC). Cambridge: Cambridge University Press. Intergovernmental Panel on Climate Change (IPCC). 2012. Managing the Risks of Sallu, S.M., C. Twyman, and L.C. Stringer. 2010. Resilient or vulnerable livelihoods? extreme events and disasters to advance climate change adaptation. Assessing livelihood dynamics and trajectories in rural Botswana. Ecology and Cambridge: Cambridge University Press. Society 15(4): 1–24. Intergovernmental Panel on Climate Change (IPCC). 2014. Climate Change 2014. Savage, M., A. Mujica, F. Chiappe, and I. Ross. 2015. Climate finance and water Mitigation of climate change. Contribution of Working Group III to the Fifth security: Ethiopia case study. Oxford Policy Management (OPM) Client Report: Assessment Report of the Intergovernmental Panel on Climate Change. 1–30. June 2015. Available at http://www.opml.co.uk/publications/climate- Cambridge, UK: Cambridge University Press. finance-and-watersecurity-ethiopia-case-study. Schürer, K., and T. Penkova. 2015. Creating a typology of parishes in England and International Federation of Red Cross and Red Crescent Societies (IFRCS). 2015. Wales: Mining 1881 census data. Historical Life Course Studies 2: 38–57. World Disasters Report 2015: Focus on local actors, the key to humanitarian Scoones, I. 2009. Livelihoods perspectives and rural development. The Journal of effectiveness. Geneva: International Federation of Red Cross and Red Crescent Societies. Available at https://ifrc-media.org/interactive/wp-content/ Peasant Studies 36(1): 171–196. uploads/2015/09/1293600-World-Disasters-Report-2015_en.pdf. Accessed 10 Sharp, K., S. Devereux, and Y. Amare. 2003. Destitution in Ethiopia’s Northeastern Feb 2016. highlands (Amhara National Regional State). In Institute of Development Jongman, B., S. Hochrainer-Stigler, L. Feyen, J.C. Aerts, R. Mechler, W.W. Botzen, Studies at the University of Sussex. and P.J. Ward. 2014. Increasing stress on disaster-risk finance due to large Simane, B., B.F. Zaitchik, and J.D. Foltz. 2014. Agroecosystem specific climate floods. Nature Climate Change 4(4): 264–268. vulnerability analysis: application of the livelihood vulnerability index to a Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 19 of 19 tropical highland region. Mitigation and Adaptation Strategies for Global Change 21(1): 39–65. Smith, L., T. Frankenberger, B. Langworthy, S. Martin, T. Spangler, S. Nelson, and J. Downen. 2015. Ethiopia Pastoralist areas resilience improvement and market expansion (PRIME) Project Impact Evaluation. In Baseline Survey Report Vol. 1, U.S. Agency for International Development, Feed the Future Feedback,1–155. Sullivan, C. 2002. Calculating a water poverty index. World Development 30: 1195–1210. doi:10.1016/S0305-750X(02)00035-9. Suman, S.L. 2014. Climate change resilience and vulnerability of farmers in Nepal. Doctoral Dissertation. Japan: Hiroshima University. Tanner, T., D. Lewis, D. Wrathall, R. Bronen, N. Cradock-Henry, S. Huq, and M.A. Rahman. 2015. Livelihood resilience in the face of climate change. Nature Climate Change 5(1): 23–26. Tesso, G., B. Emana, and M. Ketema. 2012. Analysis of vulnerability and resilience to climate change induced shocks in North Shewa, Ethiopia. Agricultural Sciences 3: 871–888. United Nations (UN). 2015a. Sendai Framework for Disaster Risk Reduction 2015–2030 (http://www.unisdr.org/we/coordinate/sendai-framework). Accessed 3 Mar 2016. United Nations (UN). 2015b. Adoption of the Paris Agreement. UNFCCC. Report No. FCCC/CP/2015/L.9/Rev.1, (http://unfccc.int/resource/docs/2015/cop21/eng/ l09r01.pdf). Accessed 3 Mar 2016. United Nations (UN). 2015c. Sustainable Development Goals (http://www.un. org/sustainabledevelopment/sustainable-development-goals/). Accessed 3 Mar 2016. United Nations International strategy for Disaster Reduction (UNISDR). 2015. Global Assessment Report on Disaster Risk Reduction. United Nations Office for the Coordination of Humanitarian Affairs (UN-OCHA). 2016. Humanitarian funding analysis: Flood and impacts in Ethiopia. In UN Office for the Coordination of Humanitarian Affairs, Crisis Briefing. United Nations Office for the Coordination of Humanitarian Affairs. (UN-OCHA). 2006. Situation Report on floods in Ethiopia.No.3 21st August 2006. Woldemariam, M. 1986. Rural vulnerability to famine in Ethiopia, 1958–1977. New Delhi: Vikas publishing House LTD. World Bank. 2010. Ethiopia: Economics of adaptation to climate change. Washington DC: The World Bank Group. World Meteorological Organization (WMO). 2003. Integrated Flood Management Case Study1Ethiopia: Integrated Flood Management. In WMO/GWP Associated Programme on Flood Management, Technical Report, 1–15. You, GJY., and C. Ringler. 2010. Hydro-Economic Modeling of Climate Change Impacts in Ethiopia. IFPRI Discussion Paper No. 960. Washington DC: International Food Policy Research Institute (IFPRI). Submit your manuscript to a journal and benefi t from: 7 Convenient online submission 7 Rigorous peer review 7 Immediate publication on acceptance 7 Open access: articles freely available online 7 High visibility within the fi eld 7 Retaining the copyright to your article Submit your next manuscript at 7 springeropen.com http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Geoenvironmental Disasters Springer Journals

Livelihood resilience in the face of recurring floods: an empirical evidence from Northwest Ethiopia

Loading next page...
 
/lp/springer-journals/livelihood-resilience-in-the-face-of-recurring-floods-an-empirical-8QZhCW4bJ7
Publisher
Springer Journals
Copyright
Copyright © 2017 by The Author(s).
Subject
Environment; Environment, general; Earth Sciences, general; Geography, general; Geoecology/Natural Processes; Natural Hazards; Environmental Science and Engineering
eISSN
2197-8670
DOI
10.1186/s40677-017-0074-0
Publisher site
See Article on Publisher Site

Abstract

Background: The recent trend of increasing incidents of floods in Ethiopia is disrupting the livelihoods of a significant proportion of the country’s population. This study assesses the factors that shape the resilience and the vulnerability of rural households in the face of recurring floods by taking the case of Dembia district of Northwest Ethiopia as one of the flood-prone areas in the country. Results: The data for the study were collected through a survey of 284 households, two focus group discussions, and 12 key informant interviews. Principal Component Analysis and simple linear regression were used for the analysis. The former served both for data reduction and identification of the dominant factors that explain resilience to recurring flood hazards while the latter was used to check the relationship between resilience and vulnerability. Findings indicate that access and use of livelihood resources such as size of farmlands, availability of farm oxen, credit as well as ability to draw help from social networks were found to be the most important factors that determine the resilience of households to floods. Similarly, the coping strategies employed by households were found to be constrained mainly by the scale and impact of the recent floods and lack or shortage of basic infrastructural and social facilities. Conclusions: The results confirmed that most of the traditional coping strategies employed by households failed to effectively help households offset the impacts of flooding. Given the livelihood context of smallholder farming system in the studied area, context specific institutional interventions such as the integrated use of both safety nets and cargo nets may help communities to overcome livelihood predicaments associated with the recurrent flood disasters. This implies that policy should focus more on addressing the factors that expose people to flood disasters and shape their resilience, rather than focusing on short-term emergency responses which seems to be the norm in much of the flood affected areas in the country. Keywords: Flood disaster, Resilience index, Vulnerability index, Dembia, Northwest Ethiopia Background Reduction [UNISDR] (2015). The frequency and severity It is widely recognized that environmental hazards fre- of flooding are also increasing in many parts of the quently affect the livelihoods of many people around the world associated with population pressure, urbanization world. The effects of these hazards cannot be expected and climate change (Hirabayashi et al., 2013; Jongman to be similar as people and nations differ in terms of et al., 2014). This is evident when one considers the their level of development, which largely determines number of people affected by flooding in recent decades. their response to specific disasters. For instance, flooding accounts much of the loss event Flooding is one of the most frequent and destructive worldwide between 1980–2014 more than any other sin- environmental hazards that occur annually worldwide gle disaster (Munich RE, 2015) and tops the list of nat- (United Nations International strategy for Disaster ural disasters by economic damages in 2014 (Guha-Sapir et al., 2015). Flooding is also the leading disaster agent * Correspondence: zerihun.berhane@aau.edu.et in the world in terms total number of reported disasters Center for African and Oriental Studies, Addis Ababa University, P.O. Box: from 1900–2014 (see Additional file 1: Figure A) while it 1176, Addis Ababa, Ethiopia is the second largest natural hazard, next to drought, in Full list of author information is available at the end of the article © The Author(s). 2017 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 2 of 19 terms of total number of affected persons between 1960 Table 2 Flood Disaster Statistics in Ethiopia between 1960 and 2013 and 2015 (see Additional file 1: Figure B). In Ethiopia, despite being given relatively less attention Year Occurrence Total deaths Affected Homeless Total affected as compared to drought, flooding has long been recog- 1968 1 1 10,000 6000 16,000 nized as one of the major environmental hazards that 1976 1 0 50,000 20,000 70,000 often develop into a disaster affecting the lives and liveli- 1977 1 7 16,000 0 16,000 hoods of people for many years. In fact, flooding and its 1978 1 9 1000 0 1000 damages are considered as a perennial phenomenon in 1981 1 0 20,000 0 20,000 the highlands (Disaster Prevention and Preparedness 1985 2 9 8000 20,000 28,000 Commission [DPPC], 1994). The country’s proneness to non-drought disasters such as floods has been limited in 1988 2 45 47,240 0 47,240 the past in terms of frequency and scope (DPPC, 1997a). 1990 1 0 350,000 0 350,000 However, the historical records on flood data suggests 1993 2 2 30,000 4800 34,800 that Ethiopia faced 47 major floods since 1900, which af- 1994 1 4 43,000 0 43,014 fected close to 2.2 million people (You and Ringer, 1995 1 27 93,875 0 93,875 2010). In this regard, many of the flood disasters oc- 1996 2 40 90,000 25,000 115,000 curred since 1980 (World Bank, 2010) (see also Table 1). This coupled with climate change and variability is likely 1997 2 326 65,000 0 65,022 to increase flooding as one of the major extreme events 1999 6 48 22,255 125,000 147,255 in the future posing a growing threat to many liveli- 2000 2 69 30,000 0 30,000 hoods (Intergovernmental Panel on Climate Change 2001 3 5 39,500 0 39,500 (IPCC), 2014; Savage et al., 2015). Flooding as a recur- 2002 1 22 4000 0 4000 rent environmental hazard is particularly felt in areas 2003 1 119 110,000 0 110,000 where people are already vulnerable to any adverse cli- matic event as a result of weakened resilience. For in- 2005 4 211 242,418 0 242,418 stance, an estimated 210,600 people were affected by 2006 7 951 434,050 0 434,146 flooding only within three months (November, 2015– 2007 2 17 245,386 0 245,386 January, 2016) (United Nations Office for the Coordin- 2008 3 45 115,595 810 116,440 ation of Humanitarian Affairs [UN-OCHA], 2016 p.1) 2010 2 19 80,700 0 80,700 A complete national and regional disaggregated data 2011 1 0 40,200 0 40,200 on flood disasters is limited in Ethiopia (see Table 2). However, the available literature indicates that some 2013 1 0 51,500 0 51,500 areas in the country are far more frequently affected Source: Authors’ computation from EM-DAT: OFDA/CRED International Disaster Database-www.emdat.be than others, to the extent of being labeled as ‘flood- prone areas’ (World Meteorological Organization (WMO), 2003; Nederveen et al., 2011; UN-OCHA, Tigray; North Gondar, North and South Wello, and 2016). These areas include central and western zones of Oromia zones of Amhara region which are often affected by flash flooding as well as those that are affected by Table 1 Total damage due to natural disasters between 1900 riverine floods, which include almost all the major river and 2013 in Ethiopia basins and the Tana Basin (DPPC 1994; 1997b; Nederveen Year Type Total damage ('000 US$) et al., 2011; UN-OCHA, 2016). 1906 Earthquake 6750 The Amhara region as indicated above is one of the 1969 Drought 1000 flood-prone areas in the country where severe and fre- 1973 Drought 76,000 quent floods affect a considerable number of people in recent years. In this regard, the limited available data on 1994 Flood 3500 the effects of floods in the region indicate that riverine 1998 Drought 15,600 floods were recorded in 1966, 1967, 1974, and 1975. 1999 Flood 2700 Severe flash floods have also been recorded in 1993 and 2005 Flood 5000 1996, with 72,569 people being affected. And a severe 2005 Flood 1200 flooding in 2006, has affected 107,286 people, displacing 2006 Flood 3200 37,982, damaging crops on 18,000 ha of land in six zones (Disaster Prevention and Preparedness Agency [DPPA], 2013 Flood 2200 2007; Nederveen et al., 2011; UN-OCHA 2016). More- Source: Authors’ computation from EM-DAT: OFDA/CRED International Disaster Database-http://www.emdat.be/ over, seven districts in the region, which are all found Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 3 of 19 around Lake Tana, are particularly well known for being Frankenberger and Nelson, 2013). Following Maru et al. frequently affected by both flash and riverine floods. (2014), this study argues that there is a need to combine One of these areas is Dembia district in North Gondar the two concepts since both are concerned with features zone, which is highly affected by Megech, Derema and that affect people’s ability to cope with and respond to Gumero rivers that frequently overflow their banks change. affecting the nearby settled plains (DPPA, 2007) (see In dealing with resilience, it is important to define “re- Additional file 1: Table A). silience to whom” and “resilience of what” (Cutter, 2016 Flooding in Dembia district, has become all too com- p.1). Accordingly, livelihood resilience as the building mon in recent years, and remains the most serious chal- block of this study is conceptualized as “the capacity of lenge to peoples’ livelihood with its short and long-term all people across generations to sustain and improve effects. As a result, some people were forced into desti- their livelihood opportunities and well-being despite en- tution (UN-OCHA, 2006; DPPA, 2007; You and Ringer, vironmental, economic, social, and political distur- 2010; Kreft et al., 2016; UN-OCHA, 2016). When this, bances” (Tanner et al., 2015 p.1). However, one of the coupled with the increasing flooding scenario predicted main deficiencies in the literature so far has been the by the reports of the IPPC (2007; 2012; 2014) amplify failure to identify the root causes of vulnerability as an the magnitude of the problem. Furthermore, the prob- initial step to understanding resilience owing to discip- lem of flooding would particularly be worse for countries linary perspectives and focus limited dimensions (Cutter, like Ethiopia with the majority of its population sub- 2016). This in turn resulted in lack of working defini- jected to poverty and vulnerability to climatic shocks tions, key indicators, and valid measurements for the (Berhanu and Fekadu, 2015; Ethiopian Panel on Climate concepts of vulnerability and resilience in the literature Change [EPCC], 2015; Savage et al., 2015). This in turn, (Alfani et al., 2015; Bahadur et al., 2015; Razafindrabe justifies the need to study flooding as a livelihood prob- et al., 2015; Cutter, 2016). lem since it creates downward pressures on livelihoods. Most studies conducted on natural disasters and their The understanding of flooding as a livelihood shock also effects on peoples’ livelihoods in different parts of needs an analysis of resilience of livelihood systems in Ethiopia focused mainly on drought and overlooked the face of the recurring flood disasters. flooding and its impacts (Woldemariam, 1986; Rahmato, The concept of resilience has recently been widely 1991; Sharp et al., 2003; Rahmato, 2007). The few avail- promoted in many fields such as engineering, psych- able studies on floods also focus on issues such as risk ology, and ecology, very recently resilience has become perceptions and risk management strategies (Moges, widely used by humanitarian and development actors 1978; Bekele, 2003; WMO, 2003; Nederveen et al., working across diverse thematic areas including, disaster 2011). Although there are recent studies that looked into risk reduction, climate change, ecosystem management, resilience and vulnerability in Ethiopia, flooding and its and food and nutrition security (Frankenberger et al., impact on livelihoods has not been investigated (Deressa 2012; Constas and Barrett, 2013; Maxwell et al., 2013; et al., 2008; Simane et al., 2014; Mengistu et al., 2015). Hoddinott, 2014; Razafindrabe et al., 2015). Building re- This is a key gap in the existing empirical studies given silience of households, communities, and systems has that flooding is a major natural hazard that affects the also been considered as a crucial policy objective among livelihoods of thousands of smallholder farming commu- various development frameworks including, the Sendai nities every year across the country (see also Table 2). Framework for Disaster Risk Reduction (United Nations This study therefore addresses the gap in the literature [UN], 2015a), the Paris Agreement on Framework by looking into the root causes of vulnerability and Convention on Climate Change (UN, 2015b), and the measuring livelihood resilience of smallholder farmers to Sustainable Development Goals (UN, 2015c). Resilience flood hazards. Linking livelihood approaches to resili- harbors different meanings in different contexts. In dis- ence thinking is imperative to enhance the understand- aster risk reduction, it is broadly viewed as a concept ing of livelihood dynamics and to explore how that deals with a system’s capacity to anticipate, to cope, households maintain and improve their livelihoods in to absorb, adapt to, and recover from the adverse impact the face of natural disasters (Scoones, 2009; Sallu et al., of hazards and reduce vulnerability (Razafindrabe et al., 2010). In view of this, the study contributes to the disas- 2015; Tanner et al., 2015). The concept of vulnerability ter risk reduction literature by providing empirical evi- is often contrasted with resilience; however, it is an dences on the determinants of vulnerability and interlinked function of exposure, sensitivity, and adaptive resilience to the recurring flood hazards. The study also capacity (Adger, 2006; IPCC, 2014). Being a common in- adds to the conceptual and methodological debates sur- dicator, adaptive capacity, can be taken as a desirable rounding vulnerability and resilience by focusing on the characteristic of a system that minimizes vulnerability least studied hazard in Ethiopia and developing and while enhances resilience at all levels (Engle, 2011; applying context-specific indices. This would further Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 4 of 19 contribute to the application of relevant measurements and scoring exercises. These data were obtained between in relation to capturing the multidimensional nature of March-May, 2015. both vulnerability and resilience. Finally, the study high- lights the synergy between the vulnerability and resili- Sampling and sample size ence , which need to be fostered, if the objective of In selecting the sample households for the survey, a achieving Sustainable Development Goals (SDGs) in multistage sampling procedure was employed. In stage rural parts of developing countries is to be addressed in one, eight Kebeles were selected from the 40 rural years to come (Fig. 1). Kebeles in the district purposively as they are frequently hit by seasonal flooding. In stage two, two Kebeles (Tana Methods Weyna and Gur-Amba) out of the eight flood prone Research design Kebeles were selected purposively using pre-defined cri- A quantitative-dominant, qualitative mixed research de- teria. The criteria include, the physical proximity to sign was employed, where the quantitative data and flood hazard source particularly to the nearby rivers (lo- qualitative information were collected concurrently. This cation and exposure) and the severity and frequency of approach helped the study to assess how vulnerability flood-disasters. After selecting the two Kebeles, a list of and resilience are conceived in local contexts, examine the households in 26 villages (15 in Tana Weyna and 11 locally-specific impacts of flooding, and factors that in Gura-Amba) was recompiled and used as a sampling shape the resilience of households in the face of this frame to select the households. Thus, a final sample of disaster. 256 households out of the 971 households were selected using systematic random sampling technique. Data sources Quantitative data and qualitative information for this For the qualitative interviews, both KIIs and FGD study were obtained from both primary and secondary participants were selected purposively using criteria that sources. A cross-sectional survey of 284 farm households includes being born in a particular village or lived there was supplemented by qualitative information from 12 for not less than two decades; have a first-hand experi- Key Informant Interviews (KIIs), two Focus Group Dis- ence of flood disasters; and being knowledgeable about cussions (FGDs), field observations, and Participatory the local environment, weather patterns and climate. Rural Appraisal (PRA) tools including problem ranking This was meant to capture the spatio-temporal perspectives Fig. 1 Location of the study district The map shows Dembia district of the Amhara region, Northwestern Ethiopia. It is located at 12°18'30''N and 37°17'30''E (see Fig. 1). It has an area of the 148,968 ha from which plain land accounts for about 87%, mountain and hills 5%, valleys and gorges 4.8% and water bodies 3.2%. The altitude of the district ranges from 1850 to 2000 m.a.s.l. Therefore, it is predominantly classified as Mid-land agro-ecology and the slope ranges from 2 to 4%. The district on average receives an annual rainfall between 700 mm to 1160 mm. Belg (the short rains February-April) and Meher (the long rains June-September). The average yearly minimum and maximum temperature is 18 °C and 28 °C respectively. Based on the recent Central Statistical Authority’s (CSA) population projection, the district had an estimated total population of 307,967 (CSA, 2013). Out of this total population, the majority, about 90%, are rural residents with an average agricultural household size of six persons. Source: Authors’ based on Ethio-Geographic Information System (GIS) and (CSA, 2007) Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 5 of 19 of the studied households and communities about flood di- The LVI measurement largely fits to the study context sasters based on recall. and target population (i.e., smallholder communities in sub Saharan Africa) and similar sample size based Data collection instruments on primary data obtained through a cross-sectional A structured survey questionnaire was designed and survey. The LVI also helps to capture the key factors piloted in order to generate information on house- that reflect the vulnerability situation of smallholder holds’ socio-economic and demographic characteris- farming communities in the face climate induced en- tics, livelihood asset profiles, livelihood activities, and vironmental hazards. Similar to the LVI used in Hahn income portfolios. The questionnaire also consisted et al. (2009), this study employed seven key variables, questions related to households’ vulnerability situa- which relate to socio-demographic characteristics tions, including indicators relating to exposure, sensi- (SDC) (household size, dependency ratio, age, gender tivity, and adaptive capacity. Moreover, questions of household head and education), livelihood strat- pertaining to absorptive and transformative capacities egies (LS), health status (HS), food security status of resilience were added while adaptive capacity indi- (FSS), access to water (AW), social network (SN), and cators were used as common indicators for both vul- flood disaster (FD) and its impact. Moreover, following nerability and resilience. Both interview guides and Madhuri et al., (2014) and in line with the Sustainable discussion checklists were designed to gather qualita- Livelihood Framework (SLF) (Birkmann, 2006; Scoones, tive information to supplement the household survey. 2009) this study further included natural capital (NC) that Smallholder farmers, community members, govern- mainly refers to ownership of land and size of farmland. ment and non-governmental organization representa- tives, and leaders of Community Based Organizations Calculating the LVI (CBOs) were considered as key informants and FGD The dimensions of vulnerability were systematically participants. Accordingly, 12 KIIs, two mixed FGDs combined with equal weights to create an index on a consisting of 20 people (12 men and eight women), scale of 0 to 1. As in the case of the computation of the and two PRA exercises were carried out with the same life expectancy index of the Human Development Index FGD participants. (HDI), the computation of each indicator of the vulner- ability index followed the process of standardization Approaches to measuring vulnerability and resilience (Hahn et al., 2009). In terms of measurement, Deressa and Hassan (2009) S −S a min documented the two commonly used approaches (i.e., I ¼ ð1Þ S −S max min econometric and indicator based) to measure vulnerabil- ity to disasters, including flooding. In the earlier case, Where, I is the standardized value of each indicator. the use of econometric method such as regression ana- S the original sub-component for household a, S is a min lysis is commonly employed to construct the Livelihood the minimum value of the indicator across all house- Vulnerability Index (hereafter referred to as LVI). The holds, and S is the maximum value of the indicator max drawback of this technique is, however, the challenge as- across all households. After each indicator was stan- sociated with testing various econometric assumptions dardized, the average value of each component was concerning the standard errors, hypotheses, confidence calculated using equation 2: intervals and imputing causality without making strin- gent assumptions (Etwire et al., 2013). In the latter case, I i a¼1 M ¼ ð2Þ it involves the selection of indicators that the researcher finds to largely account for the vulnerability (Deressa Where M is the one of the eight components for and Hassan, 2009). In this approach, the subjectivity of household a, I i indicates the sub-components indexed the variable selection process is considered as a limita- by i, which builds each major component, and n is the tion (Etwire et al., 2013). Although this is a major number of sub-components of each major component. limitation of the indicator based approach, recently dif- After obtaining values for each of the eight components, ferent scholars used this approach to construct LVI in the household level LVI was obtained by combining different contexts, including Ethiopia (Etwire et al., 2013; these components using equation 3: Limsakul et al., 2014; Madhuri et al., 2014; Simane et al., 2014). Similarly, this study adapted indicator based ap- w M i M a i¼1 i proach to develop LVI of smallholder farm households LV I ¼ ð3Þ a P i¼1 i in the study district. LVI developed by Hahn et al. (2009) was applied to as- sess the vulnerability of households in the study area. Which can be further expressed as: Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 6 of 19 w SDC þ w LS þ w HS þ w FSS þ w AW þ w SN þ w FD þ w NC SDC a LS a HS a FSS a AW a SN a FD a NC a LV I ¼ ð4Þ w þ w þ w þ w þ w þ w þ w þ w SDC LS HS FSS AW SN FD NC Where LVI , is the Livelihood Vulnerability Index for AC is adaptive capacity for household a a a household a, which equals the weighted average of eight ABC is absorptive capacity for household a major components, w .The weights of each major com- TC is transformative capacity for household a M a ponent are given by the number sub-component that Using the FAO’s Resilience Index Measurement and make up each major component, which are used to Analysis (RIMA) model (Food and Agricultural Organization guarantee that all sub-components have equal contribu- [FAO], 2012; Alinovi et al., 2015) equation 5 can be further tion to the total LVI (Sullivan, 2002; Hahn et al., 2009). expressed as: The LVI value ranges between 0 and 1, where 0 denotes the least vulnerable while 1 implies the most vulnerable LRI ¼fIðÞ FA ; ABS ; A ; SSN ; S ; AC ð6Þ a a a a a a a (Etwire et al., 2013; Madhuri et al., 2014). Resilience is a multidimensional concept that blends Where: relevant evidence as to how people really withstand IFA refers to income and food access; ABS = access to shocks (Almedom, 2009). Though the concept of resili- basic services; ence has been popular in development studies including, A = assets; SSN = social safety nets; S = stability; AC = poverty, vulnerability, and food security, it has been adaptive capacity. Since the indicators used in RIMA challenging to find a sound measure to resilience and have been applied to measure household’s resilience cap- how to quantify resilience remains controversial (Alfani acity to food insecurity (FAO, 2012; Alinovi et al., 2015), et al., 2015; Béné et al., 2016). However, some empirical in this study, the RIMA components were contextual- studies have attempted to measure resilience using a ized and subsumed to into the three resilience capacity composite index as proxy indicator (Amaya, 2014; Alfani indicators to measure households’ resilience to flood et al., 2015; Alinovi et al., 2015; Béné et al., 2016; Smith disasters. Accordingly, IFA, A, and AC indicators were et al., 2015). The current understanding of the resilience taken as part of adaptive capacity along with other indi- entails three interrelated capacities (adaptive, absorptive, cators; S was captured under absorptive capacity indica- and transformative), which are relevant to its measure- tors using sensitivity to flood disasters as a proxy ment (Amaya, 2014; Frankenberger et al., 2014; Bahadur indictor in addition to others; and SSN and ABS were et al., 2015; Béné et al., 2016). included under transformative capacity . Resilience being a context-specific concept, the dimen- As this index was a rough approximation of resilience sions and indicators may change depending on the con- and scale-sensitive, which may not be useful for inter- text. In assessing resilience to flood disasters, most household comparative analysis, a composite index using studies use ex-post resilience indicators as opposed to ex Principal Component Analysis (PCA) was constructed. ante measurements partly because the debate in the re- PCA is a multivariate statistical technique mostly used silience literature regarding the possibility of measuring for data reduction (i.e., larger number of variables into resilience in the absence of a hazardous event is unset- smaller numbers of components) and express the data tled (Keating et al., 2014). Therefore, the SLF was as a set of new orthogonal variables called principal adapted and built a resilience index using five capacity components (PCs) (Abdi and Williams, 2010; Abson dimensions: social, economic, institutional, infrastruc- et al., 2012; Schürer and Penkova, 2015). In this study, ture, and community capacities with each having specific PCA was used both for data reduction and identification indicators. These indicators are then aggregated by equal of the dominant factors that explain household’s resili- weighting into the three components–adaptive, absorp- ence to flood disasters. tive, and transformative capacitates to obtain a multidi- There are number of ways that can be used to retain mensional livelihood resilience index (LRI), following principal component score. In order to obtain PCs, the similar steps used in the LVI computation as given by study used Kaiser criterion of extracting factors with equations 1 to 4 (Amaya, 2014; Frankenberger et al., eigenvalues greater than one, which is one of the fre- 2014; Suman, 2014; Smith et al., 2015). Thus, LRI con- quently used technique (Abdi and Williams, 2010; Mooi structed is expressed as: and Sarstedt, 2011; Abson et al., 2012; Schürer and Penkova, 2015). Thus, the heaviest loading of principal LRI ¼fAðÞ C ; ABC ; TC ð5Þ a a a a component expressed in terms of the variables as an Where, index for each household that captured largest amount LRI is the resilience index for household a of information (Abson et al., 2012). The individual a Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 7 of 19 resilience score using PCA was computed using equa- at peak flows in the main rainy season starting from July tion 7 as follows: through August with water volume showing declines only in late September. As a result, the rivers regularly RS ¼ f XaðÞ −a =ðÞ s þ …… inundate many nearby villages with water staying on the a a1 1 1 þ fðÞ a −a =ðÞ s ð7Þ plains for several weeks. However, some severe floods aN N N have occurred in the past that are associated with heavy Where RS is the resilience score for household a; rainfall in the highlands. For instance, the floods of f is the component loading generated by PCA for the 1995/96, 2001, and 2006 were mentioned by FGD partic- first variable; ipants as the most severe flood disasters. Secondary re- th a is the a household’s value for the first variable; cords obtained from DPPA corroborate this information a1 a and s are the mean and standard deviation respect- and show that severe floods also occurred much earlier 1 1 ively of the first variable overall the households. in 1973/74 and 1982/83 in the district (DPPC, 1997a). After extracting the principal components, a simple A change in the severity of floods was also noted by linear regression was applied to check the relationship experts and other study participants. People felt that between resilience and vulnerability as used in a similar flooding is more severe and frequent than in the past. study (Madhuri et al., 2014). Apart from this, following Most of them came to understand that the population Nguyen and James (2013) a dichotomous response items pressure and the associated farmland expansion have were used to capture subjective indicators of household brought people close to the rivers which made them resilience to flooding. These indicators were quantified more vulnerable to flooding. This view partly agrees with and integrated by using an exploratory factor analysis the major assertion in the literature which relates to (Costello and Osborne, 2005; Child, 2006; Preacher causes of people’s vulnerability both to socio-economic et al., 2013). and contextual factors compared to the mere exposure to floods (Handmer, 2003; Cannon, 2006). Results and discussions In contrast, some participants stated that rivers have The nature of flooding and effects on livelihoods begun inundating farmlands and villages by changing Natural hazards such as floods and droughts often ex- their natural courses. For instance, one expert men- pose poor communities to vulnerabilities that can be tioned Megech River as one of such rivers that have investigated from two dimensions (1) external dimen- changed its natural course since the 2006 flooding. The sions or vulnerability context which can be expressed as river (Megech) is now flowing in a new channel which is the exposure to circumstances beyond people’s control, too narrow and shallow, causing the river to meander including shocks, trends and seasonality (2) internal di- and spread out onto the plains easily overflowing its mensions which refers mainly to socio-economic sys- bank, flooding several villages in Tana Weyna Kebele. tems, access and use of resources to the extent to which Two points stand out from the above findings, (1) peoples’ livelihood is affected by the exposure to external riverine flooding is the major type of flood in the study factors (Blaikie et al., 2014; IFRCS, 2015). area (2) the nature of the flooding in the area is showing In view of this therefore, the nature of flooding in the a marked change in terms of its severity having major study area in terms of its cause, magnitude, severity, fre- consequences on the lives and livelihoods of people in quency and duration is discussed as a major component the area. This finding is consistent with evidences from of the vulnerability context of people. Alongside this, by other studies in Ethiopia that suggest increasing fre- drawing together the findings from the household sur- quency of flood hazards. In this regard, Maxwell et al. vey, the FGDs and the interviews with key informants (2013) in their study of Tsaeda Amba district in Tigray on the effects of flooding on the livelihoods of people is (Northern Ethiopia) find that there is an increasing ten- discussed with the perceptions of people towards flood- dency of run-off and flooding due to environmental deg- ing as a livelihood threat. radation. Similarly, a study by Tesso et al. (2012) Flooding in Dembia district is a seasonal phenomenon. indicates frequent flooding as a major environmental The district is situated bordering the biggest lake in hazard that erode the coping capacities used by vulner- Ethiopia–Lake Tana. Several rivers that spring from able communities such as kinship support network in neighboring districts drain into Lake Tana traversing the North Shewa Zone of Oromia region. Focusing on river- district. According to the information obtained from the ine floods, a recent study by Hallegatte et al. (2016) that district Disasters Prevention and Preparedness Desk, the assessed the socioeconomic resilience to floods in 90 major cause of flooding in the area can be attributed to countries also found that, for poor people, a major risk the over-flow of five major rivers namely, Megech, associated with flood hazards is the loss of wellbeing. Derema, Nededit, Gumara, and Senzelit during the rainy Other factors that contribute to and aggravate the flood- seasons. According to key informants, these rivers reach ing in the area were also revealed by FGDs and key Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 8 of 19 informant interviews. The soil type of the district was the tributary rivers sending huge amounts of water into mentioned as one such factor. According to informants the nearby plains and beyond. As a result, thousands from the district Agriculture and Rural Development Of- have lost their assets and were dislocated from their fice, the black clay soil [which is the dominant soil type homesteads. Flooding has shown an increase in its inten- in the district] aggravates flooding as it is poor in its sity in the flood prone villages since then particularly drainage capacity and gets saturated and sticky with after the river Megech has changed its course and begun even a small amount of rainfall. It also fails to absorb flowing in a shallow bank crossing major settlement additional water flowing from rivers, contributing to areas, farm and grazing lands. flooding and water logging. The findings from the household survey indicate that Although the nature of the watershed and soil type in crop damage is most the common type of economic loss the area can be mentioned as factors that influence the experienced by households in the study villages. Accord- occurrence of flooding in the district, it is hardly pos- ingly, nearly all surveyed households (98.3%) reported sible to attribute the cause and the occurrence of flood- that they have experienced loss of crops due to flooding ing only to these factors. In fact, all of the district in the last five years before the survey. Through the agricultural experts interviewed about the cause of problem ranking and scoring exercises, participants of flooding mentioned that flooding in the district is partly the FGDs also indicated that crop damage is the fore- attributable to the following human activities that played most impact of flooding in economic terms. The loss of a greater role in determining flood damage. standing crops such as teff was substantial during the floods in 2006 and 2009. During the FGDs in both vil- 1. Deforestation: This made the highlands barren by lages, it was noted that farmers were compelled to exposing the top soil to heavy erosion and increasing change the cropping pattern from teff and wheat in to the run-off of rain water from these areas to low finger millet in recent years. In addition, almost all par- areas. Periodic changes in the amount and intensity ticipants and key informants indicated that farmers in of rainfall aided by the lack of vegetation cover in the study villages have begun to rely more on secondary the highlands also help in aggravating the run-off crops such as chick-peas, field peas, and faba beans and and the flooding in the study area. other leguminous crops which grow by using the 2. Traditional and subsistence-oriented farming system residual moisture left in the soil in the dry seasons. in the highlands was mentioned as a factor that However, the overall production of cereals and pulses causes and accentuates the rate of soil erosion and has gone down in recent years owing to the loss of soil run-off in the study area. According to the opinions fertility as a result of sedimentation which creates suffo- of the district Agriculture and Rural Development cation to such crops. In addition, the humidity of the experts, some irresponsible local farming practices soil resulting from flooding creates a favorable condition such as tilling hilly lands have increased the problem for pests such as Cut Worm (Agnotis Segetum) that re- of run-off and thereby contributed to increase duces the productivity of the crops. In relation to this, flooding in low lying areas. one of the participants of FGDs in Tana Weyna Kebele 3. Lack of integrated conservation activities and disclosed that: watershed management was also mentioned as contributing factor to the rise in the frequency of …flooding is making the cultivation of crops a flooding as well as the increasing human challenging task. During the rainy season, it washes vulnerability in the area. away crops that we grow spending so much labor and time and when we plant secondary crops, Korache In general, the district’s geographic location, topog- [Cut-worm] destroys it. raphy and soil type aggravated by the effects of human intervention such as deforestation, traditional cultivation The loss of primary crops and the declining product- practices and lack of sustainable water-shed measures ivity of secondary crops suggest that exposure to food were found to cause or exacerbate flooding in the study insecurity is inevitable for the affected households. villages. The effect of flooding on the food security of house- holds is also amplified by the loss of production as a Effects of flood disaster on livelihoods result of the time spent on recovery and rehabilitation Flooding has been affecting the study villages for years. in the aftermath of the flooding. Flooding has also According to the district agriculture and rural develop- increased the vulnerability of households to food inse- ment office, the study villages experienced one of the curity as attested by the increasing relief grain re- worst floods in 2001 caused by the heavy rainfall in the quests made by the District Agricultural and Rural highlands that increased the volume of Lake Tana and Development Office. Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 9 of 19 Households’ vulnerability as measured in LVI However, the lack of access to alternative income earn- The LVI that measures the vulnerability of households ing activities in the district, coupled with the severity of to flood disasters indicates that most households are the recent floods. These floods were mentioned to have highly vulnerable to flooding with a mean value being adverse impacts on most farmers. Lack of human cap- around 0.5. The LVI shows the inter-household differ- ital, particularly labor was reported to be a major factor ences in terms of exposure, sensitivity, and adaptive cap- that heightens households’ vulnerability situation. acity. Accordingly, the major contributing factor to the Focus group discussants agreed that the degree of ex- high vulnerability of households to flooding in the study posure to flooding is mainly determined by the physical area is found to be exposure with a mean index value of proximity of farmlands and settlement areas (villages). 0.65 followed by sensitivity with a mean index value of Poor asset holdings mainly farmlands and oxen were re- 0.56 out of 1 (Fig. 2). Thus, most households are highly ported to be sources of social vulnerability. In the FGDs exposed to flooding and more sensitive to flood-related and KIIs, it was repeatedly noted that physical exposure risks such as gully erosion resulting in the loss of both to floods (physical vulnerability) was the major factor farm and pasture lands (see Fig. 3a). Studied households that puts studied households’ livelihoods at risk. In view are also found to have relatively low adaptive capacity this, it was vividly indicated the “better-off” households with a mean index of 0.53 out of 1. This implies that the in terms of asset holdings were highly affected by flood studied households have limited capacities in terms of disaster, which resulted in to the loss of assets accumu- offsetting flood disasters by employing long-lasting lated over time. The major floods that occurred in the methods such as constructing flood dykes, which is only 2006, 2008, and 2012 rainy seasons were mentioned as reported to have been used by 31.08 percent of house- blatant examples of such phenomenon. This, however, holds (see Table 3). Instead, as field observation shows does not mean that asset holding did not contribute to many households largely rely on coping strategies mainly the resilience of households, it only confirms the fact plastering the basement of their huts with daub, which that not all households in the study area were exposed may not help to withstand more severe flood hazards, to floods to the same extent and therefore were not af- frequenting the area in recent years (see Fig. 3b). More- fected in similar ways. This view strengthens the evi- over, data from the household survey highlights that dence that exposure to flood events is a necessary but other frequently employed coping strategies include not sufficient factor in determining the vulnerability of changing crops (86.01%), relying on informal social livelihoods. For instance, participants of FGDs in Tana transfers (83.89%), and borrowing seeds (80.93%) (see Weyna Kebele, noted that the extent of flood damage on Table 3). standing crops, depends more on the proximity of a The results of qualitative interviews and discussions farmland to the river Megech as opposed to the asset corroborated the findings from the LVI. holding of the household. Accordingly, it was stated that Accordingly, participants of FGDs mentioned that households whose farmlands are located near to the households with adequate labor can engage in dyke con- river were exposed to more flood hazards both in the struction and timely drain their farmlands. Moreover, it short and long rains. was highlighted that such households are able to engage To establish the relationship between the resilience in both on-farm and off-farm activities and maintain and the vulnerability of households in the study area, their household income during times of extreme floods. Ordinary Least Square (OLS) regression was used with LVI as an explanatory variable and the resilience index obtained using PCA as a dependent variable. The result shows that the LVI decreases livelihood re- silience index by 6.73 points, statistically significant at less than 1%. The first component of the PCA, which captures the largest variability of the sub-components is considered for capturing the resilience of surveyed households, which is composed of adaptive, absorptive, and transformative capacities (Frankenberger et al., 2014; Béné et al., 2016; Smith et al., 2015). The first component indicates the dominance of adaptive capacity over other components. The relationship between the two indices is to be expected as resilience is often taken to be the flip side of vulnerability (IPCC, 2001; 2007). In Fig. 2 The three components of vulnerability Source: Authors’ own this study, adaptive capacity was taken as joint compo- construction from household survey (April 2015) nent shared between the LRI and LVI indices, however, Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 10 of 19 Fig. 3 a Gullies created by flooding. b House built with raised platform, plastered with mud to protect from floods. Source: Field observation in Debmia district (May 2015) absorptive and transformative capacities as the other significant determinants of resilience of households components of LRI that positively contribute towards (Table 5). Thus, those with higher educational levels and households’ overall resilience status seems to be rele- having relatively adequate farmlands are likely to have gated as the PCA only extracts the first component– more resilience. Most importantly, engaging in trade as adaptive capacity. Therefore, we argue that the compari- the highest form of diversified livelihood strategies is son between the two indices need further analysis that likely to increase LRI by 0.042 points, statistically signifi- captures the multidimensionality of both vulnerability cant at less than 1%. and resilience. The factor analysis results on the dichotomous re- sponse items also show that three of the six statements Households’ resilience capacity as measured by LRI express smallholders’ subjective resilience. The first Relying on PCA, factors that have eigenvalue greater component represents 22.8% of the variance and relates than 1 were chosen as resilience indicators. Accordingly, to greater reliance on social networks that contributes to the results from the PCA indicate that five of the 13 adaptive and absorptive capacities, for example in terms components have higher than one eigenvalues and rep- of borrowing seeds (Table 6). Here, crop damage being resent 62.7% of the total variance (Table 4 and Fig. 4). the most common type of economic loss experienced by Most of these variables belong to adaptive capacity indi- households in the study villages reflects the crucial im- cators and include household and demographic charac- portance of social capital in a household’s resilience. teristics (age, household size, education, and supply of At the time of disasters and soon after, people largely labor). Next to household and demographic characteris- count on their kinship networks, mutual aid, self-help tics, livelihood diversification, which mainly belongs to groups and indigenous organizations secure help and absorptive capacity, describes the resilience of house- support (Haines et al., 1996; Aldrich, 2012). However, as holds towards flood disasters in the study area. the frequency and severity of co-variate shocks such as Following Nguyen and James (2013) those factor flooding increases, the role of social networks begins to scores with the highest eigenvalue were used as a wane. This process came out in the FGDs where partici- dependent variable for further analysis in the exploratory pants have mentioned the severity of flooding in recent multiple regression. The result indicates that human and years as the main obstacle for relying less on kinship natural capital endowments mainly education and land networks and neighbors. Moreover, flooding has affected holding size as well as engagement in more diversified the majority people in neighboring villages so much so activities mainly trade seems to be positive and that it was impossible to rely on kinship networks. For instance, one key informant explained that since the heavy flooding of 2006, the frequency and severity of Table 3 Coping strategies for flood disasters Coping strategy Number Percent Table 4 Principal components of resilience indicators of Borrowing seeds 236 80.93 households Selling household assets 236 12.71 Component Eigenvalue Difference Proportion Cumulative Changing crops 236 86.01 Comp1 2.68293 .994965 0.2064 0.2064 Constructing flood dykes 232 31.03 Comp2 1.68796 .3136 0.1298 0.3362 Informal social transfers 236 83.89 Comp3 1.37436 .0426096 0.1057 0.4419 Relocating to higher grounds 234 58.97 Comp4 1.33175 .258619 0.1024 0.5444 Source: Authors’ own construction from household survey (April 2015) Comp5 1.07313 .0815582 0.0825 0.6269 Note: This is a multiple response item and therefore the percentage does not add up to 100 percent. Source: Authors’ own construction from household survey (April 2015) Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 11 of 19 Fig. 4 Resilience spider diagram of the major components the LRI. Source: Authors’ own construction from household survey (April 2015) floods are increasing in all villages as a result of which not necessarily signify clear boundaries as they are only households have to “rely on relief grains to sustain their used to facilitate the analysis process. In addition, they lives”. This opinion was also verified by data obtained do not show some causes of vulnerability such as ill- from the District Agriculture and Development Office nesses, divorces and similar idiosyncratic shocks that that showed an increase in relief grain recipients. contribute to the weakening of resilience. In general, resilience was understood as a state of hav- Through FGDs and interviews, it was possible to iden- ing strength to quickly recover from the damages caused tify major factors that limit the resilience and coping by flooding. A key component of livelihood resilience for capacity of households in the face of flood disasters. many participants of FGDs and key informants was ar- Accordingly, participants and informants have identified ticulated as the ability to regain pre-disaster level of liv- a range of factors that determine the resilience of house- ing without sustaining crippling damage to household holds, by focusing mainly on the major flood disasters assets that could push people further into poverty. that occurred in the study villages in the past ten years. Moreover, during the focus group discussions it was Since the majority of factors relate with livelihood indicated that flooding as a livelihood problem does not resources and access to them, attempt was made to as- affect households equally in the study villages. This im- sess the household livelihood situations by using a com- plies that the resilience of households is understood bination of qualitative and quantitative methods. Below, more in relative terms which further indicate the need the major factors that determine the resilience of house- to set some locally specific indicators in order to differ- holds are discussed. entiate households in terms of their level of resilience. In this regard, the FGDs made with farmers in the study Natural capital: land villages yielded some useful locally specific indicators In any rural community land is a basic productive re- that helped to measure the level of resilience of source, and access to it determines the wellbeing of a households. given livelihood. According to the findings of this study, Accordingly, the participants identified the location of however, farmland location, and fertility were indicated farmland, critical asset holdings such as a pair of oxen, to be more important than a mere access to land in de- the ability to draw help form relatives in other villages, termining the resilience of households in the face of and time taken to recover from the impact of the floods flood disasters. The FGDs and interviews made with the as some of the major indictors of the livelihood resili- study households indicated that the qualities as well as ence of households faced with flood- disasters in the the location of farmland are the key factors that limit or study villages (Table 7). The categories were also used in enhance the resilience of households to flood-induced the household survey to differentiate sample heads of shocks. In terms of location, the proximity of farmlands households roughly in to three groups namely, those to rivers was mentioned to have a significant role in with high resilience, those with medium resilience and determining the vulnerability and resilience of house- households with poor resilience or more vulnerable to holds more than the size and fertility of farmlands. This flooding. These three categories only show the level of finding was also supported by data obtained from the resilience of households in comparative terms and do household survey, in which farmland ownership was not Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 12 of 19 Table 5 Exploratory OLS on factors that determine LRI found to have a significant effect on the resilience of households as almost all of the households (92.8%) own Explanatory variables farmlands and those who do not own their own land Sex of household head 0.00495 were found to be equally distributed among the (0.0128) respondents. Age of household head 0.000716 This however cannot be taken to mean that access to (0.000569) land does not have a role in determining the resilience Educational status of household head 0.00545 of households. In fact, it could be argued that access to (0.00212) land may indirectly determine resilience. The detailed discussions with participants of FGDs and key infor- Household size 0.00129 mants also indicated that farmers with no land holdings (0.00429) are less resilient to the effects of flooding as compared Supply of labor 0.00657 to those who have land or can access land through vari- (0.00432) ous mechanisms. This, as mentioned by focus group dis- Incidence of illness dummy −0.00180 cussants and informants, was to be expected since the (0.0112) landless would lose their income largely drawn from ** wage labor on farms of other farmers during flooding Land size in ha 0.0379 and are likely to be affected even by moderate flooding (0.0120) as they lose the daily wages they earn from certain activ- Availability of farm oxen 0.0133 ities like weeding. Most participants of the FGDs also (0.0133) noted shortage of farmland in their respective villages. Social networks 0.0321 This problem, according to an informant from the (0.0168) Dwaro, have forced farmers, particularly the young ones *** to encroach the wetlands found on the shores of Lake Engagement in trade 0.0425 Tana for planting horticultural crops such as spices. This (0.0124) finding corroborates with results from other studies that Exposure to flood hazards −0.0166 reported small landholdings, land degradation, and (0.00996) population pressure as the major causes of vulnerability *** _cons 0.283 to disasters in other parts of Ethiopia (Rahmato, 2007; (0.0344) Tesso et al., 2012; Maxwell et al., 2013). N 214 2 Economic capital: financial asset R 0.144 Economic capital generally refers to the financial re- Standard errors in parentheses * ** *** sources that, in times of shocks could be used to reduce p < 0.05, p < 0.01, p < 0.001 Source: Authors’ own construction from household survey (April 2015) vulnerability and enhance recovery (Mayunga, 2007). Notes: The major forms of economic resource that were identi- The exploratory OLS model result has passed all the diagnostic tests such as multi-collinearity tests, omitted variables test, heteroscedasticity test and fied by the study households as having direct influence diagnostic plots to check the normality and linearity assumptions. on the resilience or coping capacity of households are discussed as follows. Livestock holding Focus group discussants in the two study villages have identified the size and type of livestock owned by a Table 6 Principal components/correlation of dichotomous household as a factor that determines the resilience of response items households. According to the focus group discussants, Component Eigenvalue Difference Proportion Cumulative households who own a large number of livestock tend to be more resilient to the effects of flooding as they use Comp1 1.36742 .30489 0.2279 0.2279 the animals as a buffer stock. This gave them the finan- Comp2 1.06253 .054492 0.1771 0.4050 cial capacity to quickly regain their livelihood, as they Comp3 1.00804 .086046 0.1680 0.5730 would sell their livestock and use the money to buy Comp4 .921993 .048510 0.1537 0.7267 seeds, rent-in farmlands for planting secondary crops Comp5 .873483 .106951 0.1456 0.8722 when the flood waters recede. Comp6 .766532 . 0.1278 1.0000 An interesting insight is also gained from the FGDs re- Source: Authors’ own construction from household survey (April 2015) garding the type of livestock and its role in the resilience Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 13 of 19 Table 7 Factors that affect the resilience of households and communities in the face of flooding in the study villages Factors Relatively resilient households Households with Medium resilience Households with poor resilience / more vulnerable households Time to recover from the 2- 3 months 6 months More than 6 months impacts of major floods Size of farmland 8-10 kada (2.0-2.5 ha) 4-8 kada (1.0- 2 ha) Less than 4 kada (1.0 ha) but mostly landless Livestock holding - Minimum 4 farm oxen - 2 cows - minimum 2 farm oxen - 1 farm oxen or none - 2 donkeys & 1 or 2 mules - 1 or 2 cows - no cows - 1 donkey - no pack animals Exposure to flooding Farm plots and homesteads Farm plots and homesteads Farm plots and homesteads located far from river banks located far from river banks located near the river banks or on the way where major rivers usually break their banks Availability of social Have relatives in other districts or occasionally draw some help Largely depend on relief grains at capital villages and are able to send their from relatives in other villages times of severe floods or resort to cattle to these places before the in the form of seeds or food taking loans from other households coming to the rainy season on grains at times of flooding regular basis. Source: FGDs and key informant interviews (April 2015) of the household. Accordingly, the participants of FGDs During the FGDs and interviews, it was also mentioned mentioned that possession of farm oxen often enhances that households with no oxen, land and other assets the resilience of households, since it gives the advantage were excluded from receiving loans as they were unable of draining flood water from farmlands so as to lessen to furnish collateral. In relation to this, a young inform- crop damage or failure. ant from Tana Weyna Kebele disclosed: “we are not However, focus group discussion participants and key given credit; they [ASCI] only give it to household heads informants alike agreed that flooding, with increased who own land”. This exacerbates their vulnerability to volume of water and duration, affected livestock and re- the effects of flood-induced shocks. versed the situation in recent years, in which those with During the discussions, it was also indicated that those more livestock were affected the most, since they lost who have better access to credit were in a better position their livestock during the floods through drowning and to withstand the aftermath shocks of flooding, as they in the aftermath through various diseases and lack of can replace their lost assets. Participants of the FGD fodder, which in turn affected their productivity. In view from the Gura Amba Kebele mentioned that there was of this one key informant said the following: good access to credit services as opposed to those in Tana Weyna Kebele. This difference in accessing credit A decade ago, farmers in our village used to keep could probably be explained by the differences in the de- many cattle. In fact, some farmers used to own as gree of physical proximity to the main credit provider much as 60 heads of cattle. Currently however we are i.e. Amhara Saving and Credit Institution. Some infor- having problems even to keep our farm oxen as the mants from Tana Weyna Kebele have also asserted that grazing fields are now covered with weeds and the credit service was not made available to farmers living in cattle are starving as they no longer find those fine most villages as the staffs of ASCI avoid remote villages grasses that used to grow in the fields. since there is a need to make frequent visits in attempt- ing to ensure repayments. Access to credit Generally, it can be argued that those households with Access to credit services was the other form of financial economic capital in the form of livestock and credit are capital, identified by household heads participated in the in a better position to withstand and recover from the study, as having effect on the resilience of households. effects of flooding as such assets contribute to their According to the household survey 36.9 percent of the resilience through creating more opportunities for liveli- respondents were able to have access to credit. And out hood diversification that enable households to manage of these, only 24 percent of them were able to receive and cope with flooding in more sustainable ways. loans from formal rural credit services (Amhara Saving Among those not having access to credit and economic and Credit Institution[ASCI]). This indicates that there assets, their resilience level is found to be very low. For is lack of access to credit, which is crucial in helping instance, among the 130 households, who reported hav- households to quickly recover from the effects of flood- ing no access to credit, only 9 (6.92 percent) were found induced shocks to replace lost assets and income. to have LRI above 0.5. Similarly, all the of the landless Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 14 of 19 households were found to be non-resilient (see Additional flood hazards and loss of assets disaggregated by resili- file 1: Table B). These results indicate the important role ence status. As indicted in the Table, the resilient and that these and similar economic assets play in determining non-resilient households provided more or less similar the resilience capacities of rural communities. assessments on their loss due to flood hazards except for The FGD participants in both villages mentioned that flood exposure. Further, looking at the educational status more resilient households have the capacity to engage in of households as one component of human capital that both on-farm and off-farm diversification activities and determines the resilience of households, the results from keep a relatively good stock of animals in neighboring dis- the survey showed that there the resilient groups are tricts that enables them to further off-set livelihood shocks better than the non-resilient ones. However, this differ- during major flood disasters like that of the 2006, 2008, ence is not statistically significant as a two-sample t-test and 2012 kremet floods. Diversification of income sources with equal variances gives a result of Pr (|T| > |t|) = is stressed in the literature as an important strategy of en- 0.1455. Thus, the evidences from the household survey hancing the resilience of vulnerable communities and it seems to concur with the qualitative information that “stands as the primary measure of household vulnerability underlines the importance of the degree of exposure to and resilience (Tesso et al., 2012 p 884; Nguyen and flood hazards and the associated human activities such James, 2013). Thus, given the benefits of diversification, as land use changes. The finding on the prominent role households that diversify their income sources are likely of exposure concurs with Doocy et al. (2013) that pro- to build their resilience to flood disasters in the future. vides a historical review of flood events worldwide from 1980 to 2009 and asserts that human vulnerability to Human capital floods is increasing among other things, mainly due to Human capital as referring to the level of education, health population growth, urbanization, and land use changes. conditions and availability of skilled labor was repeatedly The major components of the LRI for the studied mentioned as an important factor that shape the resilience households is provided in Table 9. As shown in the of households and communities to disaster-induced shocks Table, the mean LRI for all households is 0.44, which is in the literature (Adger, 2000; Mayunga, 2007). In this below the minimum threshold value–0.5. This indicates study, the availability of labor in the household was found that most households are not resilient enough to in the to be the most important form of human capital that face of the increasing flood hazards in the area. More- contribute to household resilience in the face of recurring over, from the sub-components of the LRI, one can see floods. that the studied households seem to have relatively The qualitative data obtained from interviews and higher absorptive capacity than adaptive or transforma- FGDs have also indicated that the availability of labor in tive capacities, a further indication of their vulnerability. a household play a determining role in enhancing the re- With the view of providing a more illustrative repre- silience of households. For instance, in explaining the sentation of studied households’ resilience capacity, we role of labor in household livelihood resilience an in- constructed a quadrant following the Andersen and formant in Gura Amba Kebele noted that “a farmer with Cardona (2013). The quadrant represents income per no asset can live by the sweat of his brow as long as he capita on the x-axis and LRI on the y-axis. Households is healthy and capable to work”. This clearly shows the falling in the right side of the mean values include, rich, value of labor in in terms of determining the resilience but not resilient groups, highly resilient, and extremely capacity of households. resilient groups (Fig. 5). Households in the left side of Table 8 provides a summary statistics of the responses the threshold include, poor, but resilient, highly vulner- of surveyed households with regards to exposure to able, and extremely vulnerable groups. In terms of the Table 8 Reported exposure to floods and loss of assets due to flooding Resilient group Non-resilient group Loss/damage to housing 63.41% (n = 26) Loss/damage to housing 69.66% (n = 124) Exposure to flood hazards 63.41% (n = 26) Exposure to flood hazards 56.74% (n = 101) Loss of crops due to flood hazards 95.12% (n = 39) Loss of crops due to flood hazards 99.44% (n = 177) Loss of livestock due to flood hazards 73.17% (n = 30) Loss of livestock due to flood hazards 73.6% (n = 131) Ownership of at least an ox for farming 78.05% (n = 32) Ownership of at least an ox for farming 79.78% (n = 142) Education (no. years of schooling) 3.43 (n = 41) Education (no. years of schooling) 2.68 (n = 178) Source: Authors’ own construction from household survey (April 2015) Notes: Resilient and non-resilient groups were identified based on the LRI index values, where households having an LRI value of 0.5 and above were taken as resilient groups while those with LRI below this threshold were considered to non-resilient Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 15 of 19 Table 9 Components of livelihood resilience index (LRI) The above quadrant is informative in terms of offering data as to where to focus development intervention ef- Variable Obs. Mean index Std. Dev. Min Max forts. In this regard, it is imperative to invest on various Adaptive capacity 222 0.55 0.07 0.17 0.73 livelihood resilience schemes that enhances the capacity Absorptive capacity 233 0.65 0.13 0.15 0.67 of highly and extremely resilient groups while focusing Transformative capacity 236 0.49 0.17 0.11 0.83 on reducing the number of highly and extremely vulner- LRI 219 0.44 0.07 0.18 0.62 able groups. Apart from this, it is also important to work Source: Authors’ own construction from household survey (April 2015) on empowering poor-but resilient households and rich but not resilience households. This is particularly im- portant given the overwhelming evidence, which indi- y-axis, the quadrant construction was based on the cates the likelihood of a shift in the global pattern and mean value of LRI, which was aggregated and/or com- intensity of flood hazards associated with climate change posed from adaptive capacity index, absorptive capacity (Few, 2003). index, and transformative capacity index. The LRI value ranges between 0.1-0.99 (the lowest being 0.18 and the Conclusions highest value stands at 0.62). The quadrant below and Focusing mainly on the vulnerability and resilience of above the mean and/or threshold value divide was based rural households in one of the flood prone areas in on 0.44 LRI value. The quadrant with the mean value Ethiopia- Dembia district, the study attempted to show above 0.44 consists of poor but resilient, highly resilient, that the nature of flooding in the study area has mark- and extremely resilient groups. While, the quadrant with edly changed over the past decade. The floods have be- the mean value below the mean includes, rich but not come more frequent and severe owing to a number of resilient, highly vulnerable, and extremely vulnerable factors that derive from both climatic and topographic groups. The average monthly income of households is conditions such as, periodic changes in the amount of about 10.26 USD, which reflects the level of poverty and rainfall, the nature of watershed system and soil type of depravation among the study communities as this would the area. In addition, certain human activities including mean that the average daily income of households is deforestation, increased settlement on flood plains, and only 0.34 USD. As can be shown from Fig. 5, even by traditional systems of cultivation were found to aggra- taking this low income level as a threshold, 31.9% of vate flood hazards in the area. households were found to be vulnerable. When roughly The findings of the study highlight the importance of extrapolated to the district level using CSA (2013) fig- access and use of livelihood resources such as size of ures, this proportion would mean that 88,417 people are farmlands, access to income diversifying options, credit vulnerable to flood hazards in the district out of the as well as ability to draw help from social networks in 277,170-rural population. terms of determining the resilience of households facing Fig. 5 Resilience typologies by income of households. Source: Authors’ own computation based on household survey (April 2015) Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 16 of 19 frequent flood hazards. The scale and impact of the considered as non-response cases (9.8% of the total recent floods and lack of basic infrastructural and social sample size). facilities are also found to have hampered the use of The size of a sample in purposive sampling is often robust coping strategies by affected communities and determined on the basis of “theoretical saturation” (the households. point in data collection when new data no longer pro- Given the livelihood context of smallholder farming vide additional insights to the research questions) (May, system in the studied area, which is highly vulnerable to 2002; Patton, 2002). environmental hazards and persistently challenged by The concern of livelihood approach is to understand population pressure and land degradation, it is highly how different in different places live (Scoones, 2009). likely that the size of farm land will remain to be a major Apart from being an analytical tool, SLF takes vulnerabil- determining factor of the resilience capacity of the stud- ity as a comprehensive concept covering livelihood assets ied households. Despite this, however, context specific and their access, and vulnerability context elements (i.e., institutional interventions such as the integrated use of shocks, seasonality, and trends) as well as institutional both safety nets and cargo nets may off-set livelihood structure and processes (Birkmann, 2006). predicaments. The safety nets can be implemented in To capture adaptive capacity, we used labor, education, the form of public works that are relevant to minimizing asset (income)/consumption/per capita, household size, exposure to the recurring flood hazards, particularly natural capital, and social capital. Absorptive capacity is through construction and maintenance of flood dykes. captured through access to credit, asset, diversification, The cargo nets can be put in place in the form of flood disaster exposure indices. The transformative cap- targeted microfinance, flood insurance schemes, or acity is measured by using access to services, infrastruc- agricultural input subsidization. These interventions will ture, and formal safety nets. strengthen both the absorptive and adaptive capacities of Very recently, FAO proposed RIMA-II, which is an in- households and communities in the short-term while direct measure of resilience that adopts regression analysis enhancing their transformative capacity in the long- allowing for making causal inference. However, RIMA-II term. These imply that policy should focus more on is more suitable for assessing the dynamic nature of addressing the factors that expose people to flood disas- household resilience to measurable outcomes such as food ters and shape their resilience, rather than focusing on insecurity, which requires the use of panel data. short-term emergency responses, which seems to be the Adaptive capacity (AC) indicators include: IFA (income norm in much of the flood affected areas in Ethiopia. and consumption per capita), A (availability of labor, own- ership of asset, and natural capital (land)), AC (educational Endnotes status). Other indicators included are: social capital (infor- Resilience as a concept has been highly promoted as mal transfers and participation in festive work groups) and a uniting policy instrument that links humanitarian and household size; Absorptive capacity (ABC) indicators in- development approaches to address peoples’ chronic vul- clude, S Access to credit, asset ownership, diversification of nerability to recurrent shocks and disasters (Choularton income, and flood index (flood duration, flood severity, et al., 2015). These view is also shared by the Sendai exposure to flood disasters, frequency of flood disasters, Framework for Disaster Risk Reduction (UN, 2015a) and and losses sustained due to flood disasters including crops, the UN’s Paris Agreement on Framework Convention on damage to housing, and loss of livestock); and Transforma- Climate Change (UN, 2015b). tive capacity (TC) indicators include, SSN (access to formal Kebele is the lowest administrative unit in Ethiopia. safety net (Productive Safety Net Program)) and ABS These criteria were used to account for variations in (access to services, access to infrastructure). the degree of flood-hazard exposure as all of the eight There are two major types of factor analysis tech- Kebeles are not equally affected by the flood disasters. niques (These are namely, Confirmatory Factor Analysis Hence, the two Kebeles were selected out of the eight (CFA) and Exploratory Factor Analysis (EFA). The former Kebeles to ensure the representativeness of the sample CFA helps to check hypotheses and uses path analysis drawn from the Kebeles. diagrams to denote variables and factors. The latter, EFA The overall sample size was determined by using the attempts to discover multifaceted patterns by exploring sample size determination equation that takes into the dataset and testing predictions (Costello and Osborne, account the desired confidence level (95%), the error 2005; Child, 2006). As for the rotation techniques. There margin (5%), and the prevalence of the issue under are two types, viz, orthogonal rotation and oblique investigation (p= 0.5). The required sample size was rotation. The first, orthogonal rotation (e.g.,Varimax and determined using Kothari (2004) sample size determin- Quartimax) consists of uncorrelated factors whereas ation formula. 28 households did not respond the major oblique rotation (e.g., Direct Oblimin and Promax) in- modules of the structured household survey and were cludes correlated factors. The interpretation of factor Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 17 of 19 analysis is based on rotated factor loadings, rotated eigen- Alinovi, L, E. Mane, and D. Romano. 2009. Towards the measurement of household resilience to food insecurity: An application to Palestinian households. European values, and scree test. In reality, researchers often apply Commission and the United Nations Food & Agriculture Organization more than one extraction and rotation technique relying Working Paper. Available at http://www.fsnnetwork.org/sites/default/files/ on pragmatic reasoning rather than theoretical reasoning measuring_household_resilience_to_food_insecurity.pdf. Accessed 20 Nov (Preacher et al., 2013). Thus, for the sake of brevity in in- Almedom, A.M. 2009. A call for resilience index for health and social system in terpretation, this study used varimax method of rotation Africa. Issues in Brief. The Federick S Pardee Center for the study of the variables that helps to reduce the number of variables with Longer‐Range Future. Number 10, October 2010. Retrieved from http://www. bu.edu/pardee/policy-010-resilience-index/. a high loading on a factor. Amaya, L.C. 2014. Disaster resilience to food insecurity metrics: A case study in rural Owing to the complex nature of the concept and the Costa Rica. PhD Dissertation. USA: Texas State University. lack of an exact equivalent of the word resilience in the Andersen, L.E., and M. Cardona. 2013. Building Resilience against adverse shocks: What are the determinants of vulnerability and resilience? Development local Amharic dialect, it was necessary to first obtain Research Working Paper Series, 2. farmers own subjective meanings of the term through Bahadur, A., E. Lovell, E. Wilkinson, and T. Tanner. 2015. Resilience in SGDs: Developing group discussions. Accordingly, the participants of FGDs Indicators for target 1.5 that is fit for the purpose,Briefing Paper,1–6. London: Oversee Development Institute. in both villages, agreed that the concept has a positive Bekele, L. 2003. Community perceptions and responses to Awash Floods in connotation in the sense that it matched with certain the Wonji environs, Ethiopia. Ethiopian Journal of Development Research terms like ‘ability’, ‘capacity’, ‘strength’ and ‘resistance’. 25(2): 1–33. Béné, C., D. Headey, L. Haddad, and K. von Grebmer. 2016. Is resilience a useful concept in the context of food security and nutrition programs? Some Additional file conceptual and practical considerations. Food Security 8(1):123–138. Berhanu, W., and B. Fekadu. 2015. Climate variability and household adaptation strategies in Southern Ethiopia. Sustainability 7: 6353–6375. Additional file 1: Statistics on the incidence and effects of major natural Birkmann, J. 2006. Measuring vulnerability to promote disaster-resilient societies: disasters. (DOC 314 kb) Conceptual frameworks and definitions. Measuring vulnerability to natural hazards 1: 9–54. Acknowledgements Blaikie, P., T. Cannon, I. Davis, and B. Wisner. 2014. At risk: natural hazards, We would like to thank the study participants for volunteering information on people’s vulnerability and disasters. London: Routledge. which this study is based. The constructive comments from two reviewers are Cannon, T. 2006. Vulnerability analysis, livelihoods and disasters. Risk 21: 41–49. sincerely acknowledged. Central Statistics Authority (CSA). 2007. Summary and statistical report of population and housing census: Population size by age and sex. Addis Authors’ contributions Ababa: Federal Democratic Republic of Ethiopia (FDRE). The corresponding author, ZBW is an Assistant Professor of Development Central Statistics Authority (CSA). 2013. Federal Democratic Republic of Ethiopia Studies at the Center for African and Oriental Studies (CAfOS), Addis Ababa (FDRE). Population projection of Ethiopia for all regions: At Wereda level from University. He designed the data collecting instruments, performed the 2014–2017. Addis Ababa: Federal Democratic Republic of Ethiopia (FDRE). statistical analysis, and drafted the manuscript. BEA is a PhD candidate at the Child, D. 2006. The essentials of factor analysis, 3rd ed. New York: Continuum Center for Environment and Development, Addis Ababa University. He International Publishing Group. participated in reviewing pertinent literature, collecting the data, performing Choularton, R., T. Frankenberger, J. Kurtz, and S. Nelson. 2015. Measuring shocks statistical analysis, and drafting the manuscript. Both authors contributed to and stressors as part of resilience measurement, Resilience Measurement and approved the final manuscript. Technical Working Group. Technical Series, 5. Constas, M., and C. Barrett. 2013. Principles of resilience measurement for food Competing interests insecurity: metrics, mechanisms, and implementation plans, Paper presented at The authors declare that they have no competing interests. Expert Consultation on Resilience Measurement Related to Food Security, Rome. Author details Costello, A.B., and J.W. Osborne. 2005. Best practices in exploratory factor analysis: Center for African and Oriental Studies, Addis Ababa University, P.O. Box: Four recommendations for getting the most from your analysis. Practical 1176, Addis Ababa, Ethiopia. Center for Environment and Development, Assessment Research and Evaluation 10(7): 1–9. Addis Ababa University, P.O. Box: 1176 Addis Ababa, Ethiopia. Cutter, S. L. 2016. Resilience to what? Resilience for whom? The Geographical Journal, n/a-n/a. http://doi.org/10.1111/geoj.12174. Received: 21 July 2016 Accepted: 24 February 2017 Deressa, T.T., and R. M. Hassan. 2009. Economic impact of climate change on crop production in Ethiopia: evidence from cross-section measures. Journal of African Economies 18(4): 529–554. Deressa, T., R. Hassan, and C. Ringler. 2008. Measuring Ethiopian farmers’ References vulnerability to climate change Across Regional States. In International Food Abdi, H., and L.J. Williams. 2010. Principal component analysis. Wiley Interdiscip Policy Research Institute (IFPRI) Discussion Paper 00806. Rev Comput Stat 2: 433–459. Disaster Prevention and Preparedness Agency. (DPPA). 2007. Regional summary of Abson, D.J., A.J. Dougill, and L.C. Stringer. 2012. Using principal component multi-agency flood impact assessment of 2006. Addis Ababa: Early Warning analysis for information-rich socio-ecological vulnerability mapping in Department. Southern Africa. Applied Geography:1–10. doi:10.1016/j.apgeog.2012.08.004. Disaster Prevention and Preparedness Commission. 1997b. Worst case scenario Adger, W.N. 2000. Social and ecological resilience: are they related? Progress in for drought, flood, influx of refugees and epidemics and the present Human Geography 24(3): 347–364. response system. June 1997: Addis Ababa: Federal Democratic Republic of Adger, W.N. 2006. Vulnerability. Global Environmental Change 16: 268–281. doi:10. Ethiopia (FDRE). 1016/j.gloenvcha.2006.02.006. Aldrich, D. P. 2012. Building Resilience: Social Capital in Post Disaster Recovery. Disaster Prevention and Preparedness Commission (DPPC). 1994. Non-drought Chicago: University of Chicago Press. disaster propensity in Ethiopia. Final report April, 1994. Addis Ababa: Federal Alfani, F., A. Dabalen, P. Fisker, and V. Molini. 2015. Can we measure resilience? a Democratic Republic of Ethiopia (FDRE). proposed method and evidence from countries in the Sahel. A Proposed Method Disaster Prevention and Preparedness Commission. (DPPC). 1997a. Flood vulnerability and Evidence from Countries in the Sahel, World Bank Policy Research Working in Ethiopia and needs for preparedness. June, 1997: Addis Ababa: Federal Paper, 7170. Democratic Republic of Ethiopia (FDRE). Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 18 of 19 Doocy,S.,A. Daniels,S.Murray,and T.D.Kirsch. 2013. The human impact of floods: A Keating, A., K. Campbell, R. Mechler, E. Michel‐Kerjan, J. Mochizuki, and C. Egan. historical review of events 1980–2009 and systematic literature review. PLOS Currents 2014. Operationalizing resilience against natural disaster risk: opportunities, Disasters 5(1). doi:10.1371/currents.dis.f4deb457904936b07c09daa98ee8171a. barriers and a way forward. Engle, N.L. 2011. Adaptive capacity and its assessment. Global Environmental Kothari, C.R. 2004. Research methodology: methods and techniques, 2nd ed. New Change 21: 647–656. Delhi: New Age International Publisher Ltd. Kreft, S., D. Eckstein, L. Dorsch, and L. Fischer. 2016. Global Climate Risk Index Ethiopian Panel of Climate Change (EPCC). 2015. First Assessment Report. In 2016. Who suffers most from extreme weather events? Weather-related Loss Working Group II Agriculture and Food Security. Addis Ababa: Ethiopian events in 2014 and 1995 to 2014. Briefing Paper,1–32. Germany: Academy of Sciences. Germanwatch e.V. Bonn. Etwire, P.M., R.M. Al-Hassan, J.K.M. Kuwornu, and Y. Osei-Owusu. 2013. Application of livelihood vulnerability index in assessing vulnerability to climate change Limsakul, A., W. Katasaenee, W. Paengkaew, A. Kammuang, D. Tipmanee, and P. and variability in Northern Ghana. Journal of Environment and Earth Science Sompongchaiyakul. 2014. Vulnerability index to climate change and its 3(2): 157–170. application for community-level risk assessment in Thailand. Environment Asia Few, R. 2003. Flooding, Vulnerability and coping strategies: local responses to a 7(2): 108–116. global threat. Progress in Development Studies 3(1): 43–58. Madhuri, H.R. Tewari, and P.K. Bhowmick. 2014. Livelihood vulnerability index Food and Agricultural Organization (FAO). 2012. Measuring resilience: A concept analysis: an approach to study vulnerability in the context of Bihar: original note on the resilience tool. Rome: The United Nations Food and Agricultural research, Jamba. Journal of Disaster Risk Studies 6(1): 1–13. Organization. Available at http://www.fao.org/resilience/resources/resources- Maru, Y. T., S.M. Stafford, A. Sparrow, P.F. Pinho, and O.P. Dube. 2014. A linked detail/en/c/317275/. vulnerability and resilience framework for adaptation pathways in remote disadvantaged communities. Global Environmental Change. .doi:10.1016/j. Frankenberger, T., and S. Nelson. 2013. Background paper for the expert gloenvcha.2013.12.007. consultation on resilience measurement for food security. In Paper Maxwell, D., B. Vaitla, G. Tesfay, and N. Abadi. 2013. Resilience, food security presented at Expert Consultation on Resilience Measurement Related to dynamics, and poverty traps in Northern Ethiopia, Analysis of a Biannual Panel Food Security, Rome. Dataset, 2011–2013. Medford: Feinstein International Center, Tufts University. Frankenberger, T., T. Spangler, S. Nelson, and M. Langworthy. 2012. Enhancing resilience to food security shocks in Africa. In Discussion Paper. May, T. 2002. Qualitative research in action. London: Sage Publications. Frankenberger, T.R., M.A. Constas, S. Nelson, and L. Starr. 2014. Current Mayunga, S.J. 2007. Understanding and applying the concept of community disaster approaches to resilience programming among Non-Governmental resilience: A capital-based approach: a draft working Paper prepared for the Organizations. In 2020 Conference Paper, vol. 7, 1–42. summer academy for social vulnerability and resilience building, 22–28 July 2007, Munich, Germany. Guha-Sapir, D., P. Hoyois, and R. Below. 2015. Annual Disaster Statistical Review Mengistu, A., A. Argaw, and T. Seid. 2015. Resilience of Ecosystems to climate 2014: The numbers and trends. Centre for Research on the Epidemiology of change. American Journal of Environmental Protection 4(6): 325–333. Disasters (CRED). (http://reliefweb.int/report/world/annual-disaster-statistical- Moges, A .1978. Flood risks and vulnerability in different regions of Ethiopia. review-2014-numbers-and-trends). Accessed 5 Feb 2016. Disaster Preparedness Planning Program. Relief and Rehabilitation Hahn, M.B., A.M. Riederer, and S.O. Foster. 2009. The livelihood vulnerability index: A Commission. Addis Ababa: Relief and Rehabilitation Commission (RRC). pragmatic approach to assessing risks from climate variability and change—A case study in Mozambique. Global Environmental Change 19(1): 74–88. Mooi, E., and M. Sarstedt. 2011. A concise guide to market research: The process, Haines, V. A., J.S. Hurlbert, and J.J. Beggs. 1996. Exploring the determinants of data, and methods using IBM SPSS Statistics. Heidelberg: Springer. support provision: Provider characteristics, personal networks, community Munich RE. 2015. Loss events worldwide 1980–2014 (1–10). Munich: Munich RE. contexts, and support following life events. Journal of Health and Social Nederveen, S., M. Abebe, F.V. Steenbergen, A. Tena, and G. Yohannes. 2011. Flood Behavior 37(3):252–264. based farming practices in Ethiopia: status and potential, Overview Paper Spate Hallegatte, S., M. Bangalore, and A. Vogt-Schilb. 2016. Assessing socioeconomic Irrigation 3. resilience to floods in 90 countries. In World Bank Policy Research Working Nguyen, K., and H. James. 2013. Measuring household resilience to floods: A case Paper 7663. Washington DC: World Bank Group. study in the Vietnamese Mekong River Delta. Ecology and Society 18(3):13. http://www.ecologyandsociety.org/vol18/iss3/art13/. Handmer, J. 2003. We are all vulnerable. Australian Journal of Emergency rd Management 18: 55–60. Patton, M.Q. 2002. Qualitative research and evaluation methods (3 ed.). Hirabayashi, Y., R. Mahendran, S. Koirala, L. Konoshima, D. Yamazaki, S. Watanabe, Thousand Oaks: Sage. and S. Kanae. 2013. Global flood risk under climate change. Nature Climate Preacher, K.J., G. Zhang, C. Kim, and G. Mels. 2013. Choosing the optimal number Change 3(9): 816–821. of factors in exploratory factor analysis: A model selection perspective. Hoddinott, J. 2014. Understanding resilience for food and nutrition security. 2020 Multivariate Behavioral Research 48(1): 28–56. Conference Paper 8. Addis Ababa: International Food Policy Research Rahmato, D. 1991. Famine and survival strategies: a case study from Northeast Institute (IFPRI). Ethiopia. Nordic Africa Institute. Rahmato, D. 2007. Development intervention in Wollaita, 1960s-2000s: a critical Intergovernmental Panel on Climate Change (IPCC). 2001. Climate Change 2001: review. Forum for social studies monograph 4. Addis Ababa: Forum for Social The Scientific Basis. IPCC Third Assessment Report. Studies. Intergovernmental Panel on Climate Change (IPCC). 2007. Climate Change 2007. Impacts, adaptation and vulnerability: Working Group II contribution to the Razafindrabe, B., M.B. Cuesta, R. He, K. Rañola Jr., S. Yaota, Inoue, and A. Santos-Borja. Fourth Assessment Report of the Intergovernmental Panel on Climate 2015. Flood risk and resilience assessment for Santa Rosa-Silang sub-watershed in the Laguna Lake region, Philippines. Environmental Hazards 14(1): 16–35. Change (IPCC). Cambridge: Cambridge University Press. Intergovernmental Panel on Climate Change (IPCC). 2012. Managing the Risks of Sallu, S.M., C. Twyman, and L.C. Stringer. 2010. Resilient or vulnerable livelihoods? extreme events and disasters to advance climate change adaptation. Assessing livelihood dynamics and trajectories in rural Botswana. Ecology and Cambridge: Cambridge University Press. Society 15(4): 1–24. Intergovernmental Panel on Climate Change (IPCC). 2014. Climate Change 2014. Savage, M., A. Mujica, F. Chiappe, and I. Ross. 2015. Climate finance and water Mitigation of climate change. Contribution of Working Group III to the Fifth security: Ethiopia case study. Oxford Policy Management (OPM) Client Report: Assessment Report of the Intergovernmental Panel on Climate Change. 1–30. June 2015. Available at http://www.opml.co.uk/publications/climate- Cambridge, UK: Cambridge University Press. finance-and-watersecurity-ethiopia-case-study. Schürer, K., and T. Penkova. 2015. Creating a typology of parishes in England and International Federation of Red Cross and Red Crescent Societies (IFRCS). 2015. Wales: Mining 1881 census data. Historical Life Course Studies 2: 38–57. World Disasters Report 2015: Focus on local actors, the key to humanitarian Scoones, I. 2009. Livelihoods perspectives and rural development. The Journal of effectiveness. Geneva: International Federation of Red Cross and Red Crescent Societies. Available at https://ifrc-media.org/interactive/wp-content/ Peasant Studies 36(1): 171–196. uploads/2015/09/1293600-World-Disasters-Report-2015_en.pdf. Accessed 10 Sharp, K., S. Devereux, and Y. Amare. 2003. Destitution in Ethiopia’s Northeastern Feb 2016. highlands (Amhara National Regional State). In Institute of Development Jongman, B., S. Hochrainer-Stigler, L. Feyen, J.C. Aerts, R. Mechler, W.W. Botzen, Studies at the University of Sussex. and P.J. Ward. 2014. Increasing stress on disaster-risk finance due to large Simane, B., B.F. Zaitchik, and J.D. Foltz. 2014. Agroecosystem specific climate floods. Nature Climate Change 4(4): 264–268. vulnerability analysis: application of the livelihood vulnerability index to a Weldegebriel and Amphune Geoenvironmental Disasters (2017) 4:10 Page 19 of 19 tropical highland region. Mitigation and Adaptation Strategies for Global Change 21(1): 39–65. Smith, L., T. Frankenberger, B. Langworthy, S. Martin, T. Spangler, S. Nelson, and J. Downen. 2015. Ethiopia Pastoralist areas resilience improvement and market expansion (PRIME) Project Impact Evaluation. In Baseline Survey Report Vol. 1, U.S. Agency for International Development, Feed the Future Feedback,1–155. Sullivan, C. 2002. Calculating a water poverty index. World Development 30: 1195–1210. doi:10.1016/S0305-750X(02)00035-9. Suman, S.L. 2014. Climate change resilience and vulnerability of farmers in Nepal. Doctoral Dissertation. Japan: Hiroshima University. Tanner, T., D. Lewis, D. Wrathall, R. Bronen, N. Cradock-Henry, S. Huq, and M.A. Rahman. 2015. Livelihood resilience in the face of climate change. Nature Climate Change 5(1): 23–26. Tesso, G., B. Emana, and M. Ketema. 2012. Analysis of vulnerability and resilience to climate change induced shocks in North Shewa, Ethiopia. Agricultural Sciences 3: 871–888. United Nations (UN). 2015a. Sendai Framework for Disaster Risk Reduction 2015–2030 (http://www.unisdr.org/we/coordinate/sendai-framework). Accessed 3 Mar 2016. United Nations (UN). 2015b. Adoption of the Paris Agreement. UNFCCC. Report No. FCCC/CP/2015/L.9/Rev.1, (http://unfccc.int/resource/docs/2015/cop21/eng/ l09r01.pdf). Accessed 3 Mar 2016. United Nations (UN). 2015c. Sustainable Development Goals (http://www.un. org/sustainabledevelopment/sustainable-development-goals/). Accessed 3 Mar 2016. United Nations International strategy for Disaster Reduction (UNISDR). 2015. Global Assessment Report on Disaster Risk Reduction. United Nations Office for the Coordination of Humanitarian Affairs (UN-OCHA). 2016. Humanitarian funding analysis: Flood and impacts in Ethiopia. In UN Office for the Coordination of Humanitarian Affairs, Crisis Briefing. United Nations Office for the Coordination of Humanitarian Affairs. (UN-OCHA). 2006. Situation Report on floods in Ethiopia.No.3 21st August 2006. Woldemariam, M. 1986. Rural vulnerability to famine in Ethiopia, 1958–1977. New Delhi: Vikas publishing House LTD. World Bank. 2010. Ethiopia: Economics of adaptation to climate change. Washington DC: The World Bank Group. World Meteorological Organization (WMO). 2003. Integrated Flood Management Case Study1Ethiopia: Integrated Flood Management. In WMO/GWP Associated Programme on Flood Management, Technical Report, 1–15. You, GJY., and C. Ringler. 2010. Hydro-Economic Modeling of Climate Change Impacts in Ethiopia. IFPRI Discussion Paper No. 960. Washington DC: International Food Policy Research Institute (IFPRI). Submit your manuscript to a journal and benefi t from: 7 Convenient online submission 7 Rigorous peer review 7 Immediate publication on acceptance 7 Open access: articles freely available online 7 High visibility within the fi eld 7 Retaining the copyright to your article Submit your next manuscript at 7 springeropen.com

Journal

Geoenvironmental DisastersSpringer Journals

Published: Mar 9, 2017

References