Are Non-poor Households always Less Vulnerable? The Case of Households exposed to Protracted Civil War in Southern Sudan Luka Biong deng southern Sudan Centre for Census, Statistics and Evaluation and The World Bank Juba, Southern Sudan



Download 189 Kb.
Page1/3
Date16.03.2019
Size189 Kb.
#83664
  1   2   3



Are Non-poor Households always Less Vulnerable?

The Case of Households exposed to Protracted Civil War in Southern Sudan


Luka Biong DENG

Southern Sudan Centre for Census, Statistics and Evaluation

and

The World Bank

Juba, Southern Sudan

c/o P.O. Box 30577, 00100 Nairobi, Kenya

lukabiongus@hotmail.com

ldeng@worldbank.org

2006

Technical Abstract: Civil wars in Africa are not only endemic but they stand now as the leading contributory causes of vulnerability of rural communities. Understanding vulnerability during civil war is critical for humanitarian response and the post-conflict rehabilitation planning. The dearth of understanding of vulnerability has made existing studies to make sweeping generalization by either unwittingly equating the dynamics of vulnerability during civil with those in the context of other risk events or projecting people in the “war zones” as all vulnerable. This paper is an attempt to gain a nuanced understanding of such dynamics of vulnerability. It has been shown in this paper that it is not always true during civil war that non-poor are necessarily less vulnerable than poor households. The sweeping generalization that people caught up in civil war are all vulnerable is not supported by the findings of this paper. The paper has shown that the ‘standard’ pattern of vulnerability to drought is similar to that in the context of exogenous counter-insurgency warfare, while a different pattern of vulnerability to endogenous shocks has been identified.
Non-technical Abstract:

This paper is an attempt to link civil war, as one of the risk events, to household vulnerability. It discusses the household vulnerability framework and provides description of empirical context. Although livelihood strategies adopted by households exposed to prolonged conflict are not discussed, the paper presents the outcomes of these copying strategies in terms of level of vulnerability. The level and dynamics of vulnerability during civil war are analyzed in this paper in terms of indicators related to susceptibility (level of exposure), sensitivity (level of depletion of assets and poverty) and resilience (level of consumption). The main conclusion of this paper is that non-poor households are not always less vulnerable, particularly in the context of civil wars.




1. Introduction
Civil wars have become pronounced and endemic to many African countries since the end of the Cold War. It is estimated that 20 per cent of sub-Saharan Africa’s population now live in countries that are at war with themselves and low-intensity conflict has become endemic to many other African countries (Elbadawi and Sambanis, 2000). In comparison to other regions, Africa has the highest incidence of intense civil wars, this trend has been increasing during the last two decades, while it has fallen or remained static in other regions. This upsurge of civil wars in Africa has a considerable negative impact on the socio-economic structures and increased vulnerability of the rural communities.
The risk of civil wars in much of Africa stands now as the leading contributory cause of vulnerability during the last two decades and has overtaken the long dominant role of ecological risk. The main causes of African famines during the last two decades, as noted by Devereux (2000), have changed from being drought to civil wars. For example, during the 1990s Von Braun et al. (1998) identify only one famine out of a total of eight famines in Africa that was mainly caused by drought, while the rest were mainly triggered by civil wars.
This paper is an attempt to gain a nuanced understanding of household vulnerability in the context of civil war. The dearth of understanding of vulnerability and livelihood strategies adopted by households during civil war has made existing studies in risk management literature unwittingly equating these behaviours with risk-related behaviours in the context of other risk events, or even ruled out any rational risk behaviour in the context of civil war. In most studies on civil wars, the households in the ‘war zones’ are perceived as unable to take any rational behaviour in confronting the effects of civil war and subsequently projected as victims and vulnerable. By describing people in the ‘war zones’ as all vulnerable and victims, these studies have failed to heed the subtle and complex risk behaviours which occur in the context of civil war. In fact, there are few attempts in rural development literature that have been made to link risk events in a systematic way to livelihood strategies and vulnerability. This paper is an attempt to link civil war, as one of the risk events, to household vulnerability. It is organized as follows: section 2 discusses the household vulnerability framework and section 3 provides description of empirical context. Although livelihood strategies adopted by households exposed to prolonged conflict are not discussed, section 4 presents the outcomes of these strategies in terms of level of vulnerability. The level of vulnerability during civil war is analyzed in terms of indicators related to susceptibility, sensitivity and resilience. General conclusions are presented in section 5.
2. Household Vulnerability Framework
The term ‘vulnerability’ is defined as exposure to risk (susceptibility), the intensity with which the risk event is experienced (sensitivity), and the capacity to resist downward movement in well-being (resilience) as a result of the occurrence of a risk event (Deng, 2003, Chamber, 1989). In most poverty studies, the outcome of a risk event is measured in terms of vulnerability. Vulnerability in return is measured in terms of indicators related to the level of well-being and welfare loss. Vulnerability, as an outcome of a risk event, is undoubtedly conditioned by the livelihood strategies adopted by households to reduce their susceptibility to such a risk event or to mitigate its potential impacts. As vulnerability is ex ante and ex post risk outcome, its measurement becomes complex. This implies that risk outcomes should be seen not as the last phase of vulnerability, but rather as a broad concept that includes the entire process of increasing vulnerability. In order to understand the dynamics of household vulnerability, a general vulnerability household model is discussed generally and its relevance is assessed in the context of drought and civil war.
There is a common consensus and cumulative empirical evidence in the risk, poverty and rural development literature that unambiguously suggest the empirical regularity of a negative association between vulnerability and wealth, including assets and income. It is commonly argued that poor households are more susceptible to risk, less resilient with a higher sensitive asset-base than are non-poor households because they have fewer tools (assets) at their disposal to defend against risk events and thus suffer proportionally greater welfare losses for given levels of risk. Specifically, Swift (1989), Chambers (1989), Moser (1998) and Siegel and Alwang (1999) presented arguments or frameworks that closely linked vulnerability to asset ownership and generally suggested that the more assets people have, the less vulnerable they are.
This commonly held view and the sweeping generalisation of a negative association between vulnerability and the level of asset ownership have not been questioned, as the findings from a huge body of empirical studies were convincingly overwhelming. There are, however, a limited but growing thinking and empirical studies that have started questioning this sweeping generalisation about vulnerability and asset ownership. In particular, Glewwe and Hall (1998) question this general assertion about vulnerability, using the case study of Peru that shows how subsistence and poor farmers in rural areas were less vulnerable to economic shocks than were high-income earners in the urban areas. In the context of civil war, there is also different thinking (Duffield, 1993; Keen, 1994), which has questioned this general assertion of vulnerability in the context of the ‘asset transfer economy’ and has suggested instead a positive association between wealth and vulnerability in the Dinka community in Sudan.
The underlying problem in the risk and rural development literature is that, while risk is conceptually and theoretically emphasised as an important determinant in household risk management strategies and vulnerability in particular, it is seldom incorporated in the empirical analysis. Risk events and their characteristics are generally assumed as exogenous factors and subsequently have no any bearing in explaining the differential vulnerability among households. While it is a common fact that assets play an important role in determining the level of household vulnerability, assets are not the only determinant of vulnerability, as other factors such as type of risk and its characteristics, are equally important in determining the level of vulnerability.
General Model1:
In most risk management literature, vulnerability is generally defined in terms of the expected value of welfare E(xit+1) output (for ith household in the tth time period) in relation to poverty or the minimum level of survival (x*), given the state of nature in the initial period, that includes conditioning variables (xit, ) that are specific to households, such as assets and livelihood strategies adopted by a household ( A it ) and exogenous variables or shocks (risk events), such as rainfall, prices or civil war (Zt) (Siegel and Alwang, 1999). Thus a household is defined as vulnerable if:

E(xit+1)  x* (1)


In fact the household welfare output (xit) represents the socio-economic status of a household at a given point of time. Generally in an axiomatic approach the probability of being vulnerable or poor is defined as:
Pr xit+1  x* xit, A it, Zt  (2)
Given the distribution of xi ( (xi)), then the probability of being vulnerable (poor) is defined as:

Pr.= ( (xi) dx = vit = Vulnerability index of household i. (3)
According to this general vulnerability model, a household is defined as vulnerable when its exposure to a risk event (Zt) is likely to change its future socio-economic status (E(xit+1)) to below the minimum level of survival (x*). As the socio-economic status of household is largely assumed to be determined and conditioned by the level of household asset ownership (A it), the non-poor households, according to this model, are less vulnerable than the poor households regardless of risk event. In analysing household vulnerability in rural Africa, the World Bank adopted similar vulnerability model that narrowly focuses on the minimum poverty level and variations in income, and generally suggest that the welfare outcomes of realised risk for the poor can be a downward spiral and poverty trap, whereas, for the non-poor, welfare losses can be much less severe (World Bank, 2001:49).
Though it is true that households with more assets are likely to sustain the impact of risk events, it is not always true that all households are equally exposed to the same risk events, as some risk events and shocks tend to be specific to some households rather than others. Taking into account the nature and characteristics of a risk event, it is arguable that a household that finds itself above the poverty line (x*), but with a high exposure to risk events, should not be considered less vulnerable than a household whose consumption level is certain, but at or even below the poverty line.
The risk events (Zt), such as rainfall and civil war, play an important role in determining the level of vulnerability. That is, depending on type, nature, characteristics, severity and probability of risky events (Zt), households at time t=1 whose initial wellbeing (xit,) is either below or above the poverty line (x*) might both be considered vulnerable. While it might be reasonable to assume risk events, such as rainfall and price (economic shocks), are exogenous variables, such an exogeneity assumption is questionable in the context of civil war. It is instead hypothesised in this paper that risk event (Zt), such as counter-insurgency warfare2, is not only an endogenous variable but is triggered by household asset ownership (Ait).
In the case of generic risk events, such as drought, it is most likely that households with more assets are more able to sustain its effects than those households with fewer assets in the rural context. In the case of civil war - particularly idiosyncratic risk events, such as counter-insurgency warfare, that is primarily triggered by household assets - it makes households with more assets more likely to experience welfare loss and subsequently more vulnerable than households with fewer initial assets. The level of vulnerability of non-poor households will even become precarious with the repeated, sequential and concomitant occurrence of multiple risk events, such as drought and counter-insurgency warfare.
This definition of vulnerability that includes the dimensions of a household’s level of well-being (xit,) and risk events (Zt) is visually represented and set out in Figures 1 and 2, which respectively reflect exposure to drought and civil war. In Figure 1 the solid line represents actual consumption paths, whereas the dotted line represents the counterfactual with no risk-averse action by household or no welfare losses due to shocks and effective social protection mechanisms. The arrows that link risk events with different types of households generally symbolise the level of potential and actual effects of risk events. Two types of households, namely poor (xpt) and non-poor (xnpt), have been identified in order to show their level of vulnerability in the face of risk events, such as drought and counter-insurgency warfare.
Vulnerability Dynamics in Drought Context:
In confronting risk events, such as drought, the non-poor households (xnpt) are better able to sustain the effects of drought than are poor households (xpt), as shown in Figure 1. In the absence of effective household risk management strategies and informal insurance mechanisms, the welfare outcomes of realised risk for the poor can be a downward spiral and a poverty trap, whereas, for the non-poor welfare losses can be much less severe, as shown in Figure 1. The livelihood condition of the poverty trap is experienced by poor households because of their low-level of assets and if coupled with the absence of effective social protection, they will become more risk-averse in confronting drought, and subsequently adopt low-risk and low-return livelihood activities, rather than high-risk and high-return livelihood risk management strategies. With repeated and sequential shocks, such as drought, the poor households will experience considerable and relatively high welfare losses and the steep downward spiral of vulnerability.
Figure 1: Hypothetical Outcomes of Drought for Poor and Non-poor Households



Drought
xi


Drought with recovery


xnpt


Effective Social Protection Intervention

Minimum Level of Survival


x*


Risk Averse Poverty Trap



xpt


Drought and Downward Spiral



Time (t)
On the other hand the non-poor households with their strong asset-base, are likely to adopt high-risk and high-return livelihood activities to confront the effects of drought and they are likely to recover relatively quickly after drought occurs. With effective social protection intervention, the poor households are more likely to bear risk and take high-risk and high-return livelihood activities that will lessen welfare loss from drought and encourage quick recovery, as shown with a dotted arrow in Figure 1. However, with a repeated and sequential occurrence of drought, both poor and non-poor households will experience considerable welfare losses, that will push them below the minimum level of survival and they will become vulnerable. These hypothetical livelihood outcomes of drought for poor and non-poor households have been validated by the findings and evidence from research in the field of food security and coping strategies, particularly in the 1980s.
Vulnerability Dynamics in the Context of Civil War:
As civil war, particularly war between government and rebels, tends to be generic, the relevant risk event for the analysis of rural livelihoods and vulnerability is counter-insurgency warfare that is used to sustain the war efforts. As counter-insurgency warfare is specific, and its occurrence is hypothesised to be triggered largely by assets, then it is likely that the non-poor households (xnpt) will experience a higher welfare loss, and are thus more vulnerable to the recurrent shocks of counter-insurgency warfare than the poor households (xpt), as shown with solid arrows in Figure 2. With repeated occurrence of counter-insurgency warfare, the non-poor households are likely to experience a downward spiral of vulnerability and their socio-economic status may decline below the minimum level of survival and end up even lower than that of poor households,. Although the poor households will experience welfare loss as a result of counter-insurgency warfare, the negative change in their socio-economic status will be less severe than that of non-poor households.
These hypothetical livelihood outcomes of civil war, particularly of counter-insurgency warfare, are not adequately validated by in-depth analysis in the context of civil war. The patchy and limited research assessing vulnerability in the context of civil war tends to make generic statement about the level of vulnerability at the meso-community level rather than at micro-household level. For example some researchers have even generally concluded (de Waal, 1993; Duffield, 1993; Keen, 1994; Deng, 1999) that in the context of ‘asset transfer economy’, vulnerability is associated more with wealth than poverty.
Figure 2: Hypothetical Outcomes of Civil War for Poor and Non-poor Households


Counterinsurgency Warfare
xi





xnpt

Minimum Level of Survival
x*


Risk Averse Poverty Trap



xpt


Poor and Downward Spiral



Non-poor and Downward Spiral

Time (t)
Comparing these hypothetical livelihood outcomes of civil war with those of drought, it becomes very clear that the dynamics of vulnerability are quite different and largely shaped and conditioned by the nature of risk to which households are exposed. While non-poor households exposed to counter-insurgency warfare may experience a rapid decline in their socio-economic status, falling below a minimum survival level, the non-poor households in the context of drought may experience limited welfare loss and recover quickly to the pre-drought socio-economic status. With poor households, though they all experience welfare loss and a decline in their socio-economic status, those exposed to drought are hypothesised to experience greater welfare losses than those exposed to counter-insurgency warfare.



3. Empirical Context:
Sudan is geographically the largest country in Africa and a land of extraordinary cultural diversity. It is justifiably considered as a microcosm of Africa because of its central location reflecting within its borders all the racial, ethnic, religious and cultural diversity of the continent. Since its independence in 1956, Sudan has been embroiled in a devastating civil war in Southern Sudan and wasted more than 40 years of its independence in two major civil wars (1955-1972, 1982-2005).
The costs of the current civil war in Sudan have been exceptionally heavy, especially on the civilian population that have been denied protection, social and economic development. This resulted in estimated death toll of 2 million and about 4.5 million have been uprooted from their homelands (World Bank, 2003). Since the second civil war erupted in 1982 and coupled with recurrent drought, Sudan has been experiencing chronic food shortages, resulting in widespread famines most notably in 1984, 1988, 1992 and 1998. The Southern Sudan, that was not prone to famines during the first civil war in 1952-72 and post-civil war period in 1972-82, has experienced two major famines in 1988 and 1998 with excess famine mortality of more than 300,000 persons during the second civil war (Deng, 1999).
The second civil war started in 1982 in Southern Sudan and gradually reached most rural areas by 1990. There are three major regions in Southern Sudan namely: Bahr el Ghazal, Equatoria and Upper Nile. Bahr el Ghazal region in particular had been the battleground during the second civil war, particularly counter-insurgency warfare. This counter-insurgency campaign was being waged by the government militia, mainly composed of northern Arab pastoralists who live just to the north of the internal frontier of Bahr el Ghazal region, and later on during the 1990s by the southern militias composed of the major ethnic groups (Nuer and Dinka).
The situation in Bahr el Ghazal region worsened during the 1990s when political divisions erupted in 1991 within the forces of the main rebel movement (Sudan People’s Liberation Army (SPLA)) that resulted in a splinter group (mainly composed of Dinka and Nuer). This splinter group joined the government forces to further intensify counter-insurgency warfare in Bahr el Ghazal region, a stronghold of the SPLA. Unlike the raids of the Arab militia that are exogenous and occur during the dry season, the counter-insurgency warfare that was waged by the splinter group (Dinka militia) was all-year round and emerged from within the Dinka communities. For easy reference we term the counter-insurgency warfare that emanated from within the targeted community as endogenous (Dinka militia), while exogenous counter-insurgency warfare is associated with activities of government militias from outside the community (Arab militia).
The unique characteristics of the Bahr el Ghazal region and its experience in the 1990s made it ideal for conducting vulnerability study on communities who have had different experiences and exposure to different types of counter-insurgency warfare. The period covered by the study is the 1990s with the pre-war period used as a baseline to gauge changes and trends in level of vulnerability. The years 1988 and 1998 when famines occurred in the Bahr el Ghazal region were used as benchmark years to relate events and to help in recollection of data, as they are vividly remembered by the research communities.
The primary data were collected during the May 2000 – March 2001 fieldwork period in three villages in Abyei, Gogrial, and Cuiebet counties of Bahr el Ghazal region. The three sample areas were purposively selected to represent different context of risk events and the actual selection of sample villages was guided by a set of predetermined criteria, such as population concentration, geographical spread, representation of various socio-economic groups.
Abyei area, being located at the northern end of Bahr el Ghazal region, represents communities exposed mainly to exogenous counter-insurgency warfare (Arab militias), while Gogrial area, being located in the central Bahr el Ghazal region, represents communities primarily exposed to endogenous counter-insurgency warfare. The village in Cuiebet area, being located to the southern end of Bahr el Ghazal region, represents communities who had been mainly exposed to drought. Cuiebet sample area is also included for comparison, but it does not represent the main research focus and it is used interchangeably, or on some occasions not considered in the analysis, when it is not necessary.
The research fieldwork basically used various approaches to collect the necessary qualitative and quantitative data for assessing vulnerability. Because of the lack of secondary socio-economic household data in Southern Sudan, questionnaire-based household surveys and community surveys were used as the most relevant and best-suited field methods to investigate changes and trends in the level of vulnerability. In fact the use of a hybrid approach consisting of the two approaches - sample surveys and participatory methods - is necessary as each method provides a separate emphasis but complements each other within an overall research design in investigating the changes and trends in vulnerability (Ellis, 2000:198). The Statistical Package for Social Sciences was used for quantitative data analysis (Puri, 1996; Rodeghier, 1996; Norusis, 1997; Nachimas and Nachimas, 1996; SPSS, 1998).
The stratification of communities according to wealth is complex as the understanding of poverty or wealth varies considerably across cultures and disciplines. Despite this apparent variation, it is assumed in most research that ‘the poor’ can be identified by using a single measure (Ellis, 2000). Many economists favour a universal or ‘objective’ measure, such as minimum level of consumption or income, while a ‘subjective’ measure that is based on the community definition of poverty or wealth is preferred by participatory approaches (Bevan and Joireman, 1997). The most commonly used poverty measures, such as household survey wealth and consumption poverty3, provided systematic differences in wealth ranking except the personal and community wealth ranking measures (Bevan and Joireman, 1997:320). One major disadvantage of the personal wealth ranking measure is that it is subjective with head of household making an overall judgment of the wealth status of the household. Besides this major disadvantage and particularly in the context of communities who have been recipients of food aid, is that respondents may be tempted to conceal their real wealth status. Despite these disadvantages, the personal wealth ranking measure takes account of economic and non-economic capital as well as relative deprivation, which can be aggregated across research communities (Bevan and Joireman, 1997:324).
Given its relevancy and appropriateness in the context of civil war, we used personal wealth ranking measure in the study by specifically asking household heads to describe the wealth status of their households before war and during war with pre-coded possible answers (poor, modal and rich). Instead of using community wealth ranking, we asked households about the community perception of their wealth status. Besides these two questions about the wealth status of the household, we specifically asked household heads about the number of livestock owned as a proxy indicator of wealth.

4. Empirical Results:

The aim of this section is to present the empirical results and to understand better the perception of communities about their sources of risk events, level of vulnerability, types of households and members of household that were most vulnerable during civil war, and to provide possible explanations for their vulnerability. In particular there seems to be a common consensus in the risk literature of linking vulnerability to assets, with non-poor households seen as less vulnerable than poor households in the face of any risk event. Besides linking vulnerability to assets, it is important to understand which household members were most affected and whether household vulnerability is gendered in the context of civil war.


Sources of Risk: Community Perception:
Based on the community survey data collected through participatory rural appraisal methods, the research communities identified various sources of risk events during the 1990s as shown in Table 2. These sources of risk include: conventional war between the Government of Sudan (GOS) and the rebel movement (SPLA), counter-insurgency warfare (Arab militia, Dinka militia, Nuer militia), drought, floods and diseases. While the communities were exposed to a combination of these risk events, the degree and level of exposure to different types of risk varied across the research communities.


Download 189 Kb.

Share with your friends:
  1   2   3




The database is protected by copyright ©sckool.org 2022
send message

    Main page