Household SI’sThursday, 01 April 20041

Household Vulnerability in Dynamic Environments: Frameworks for Quantifying Vulnerability to Extreme Climatic Events

Andrew Dougill, Evan Fraser, Matt Chadwick and Mark Reed

Introduction: The Research Challenge

Many communities, especially the rural poor in the developing world, depend directly on natural resources for their livelihood (Rennie and Singh, 1996). Therefore for their livelihoods to be sustainable there is a fundamental need that their use of natural resources is sustainable. Consequently, any assessment of household sustainability needs to include both social and environmental analysis. Methods to achieve this integrated analysis increasingly use sustainable rural livelihoods analytical approaches (e.g. Scoones, 1998; Carney, 1998), which have developed from participatory work of development economists such as Sen (1980) and Chambers (1987, 1997). Livelihoods approaches are now mainstream within all key development bodies (see Lindenberg, 2002 for recent review), including international organisations (e.g. UNDP, FAO), national donor agencies (e.g. DFID, USAID and CIDA) and major development NGO’s (e.g. CARE, Oxfam).

The methodological frameworks provided by livelihoods approaches (e.g. Figure 1) display the threats to livelihood sustainability, arising from external shocks and trends (either environmental or socio-economic), but mediated by the “vulnerability context” of affected households. Difficulties however arise in using these largely qualitative frameworks in unpacking this vulnerability context box into a quantifiable assessment of household vulnerability to external shocks. This failing of recommended development research approaches is particularly apparent when examining household vulnerability to climate change (Adger, 1999; Kelly and Adger, 2000; IPCC, 2001), notably the shock imposed on households by extreme events, such as either droughts or floods.

Figure 1: The DFID Sustainable Livelihoods Framework (DFID, 2002)

Much existing literature on vulnerability to climate extremes, especially droughts (e.g. Davies, 1996; del Ninno et al., 2001; Gray, 2002) describes community ‘coping strategies’ by using in-depth social survey methods, similar to those of livelihoods analysis. These approaches are good at producing lists of potential management options assigned after a major event, but struggle to quantify household vulnerability before an event. This paper provides temporal analysis of the information gained using such livelihoods analysis methods before-, during- and after- both a major drought in Southern Africa and an extreme flood in rural Bangladesh. We aim to use this information to demonstrate that households not only react to, and cope with, these extreme events, but that they also anticipate (and plan for) the effects of droughts or floods. This anticipation, or ‘preparedness’, is a key element in affecting household vulnerability to such climatic shocks and is therefore a component that should always be addressed in livelihoods analysis in dynamic environments, such as drylands or floodplains. We aim to discuss the manner in which analysis of household preparations for extreme events can be used as part of vulnerability assessments that provide a useful sustainability indicator at a household level.

Purpose of this Paper

This paper will contribute to debates on vulnerability to climate change by discussing the various ways in which household level sustainable livelihoods analysis can be used to provide both qualitative and quantitative information on how households anticipate and react to extreme climatic events. A research framework that focuses on environmental resilience and socio-economic resilience is formulated and discussed in terms of its ability to predict household vulnerability and therefore better inform policy decisions on institutional support to drought- and flood-prone areas. Discussions are based on the temporal analysis of livelihoods information taken from two distinct case studies (Southern African drought of 1999 and Bangladeshi floods of 1998) to examine if methodological issues are common across different climatic shocks.

In achieving integrated socio-environmental analysis, improved models of environmental change for dynamic environments are essential. For example, models highlighting the non-equilibrium functioning of dryland environments must be applied in drought-prone semi-arid rangelands (e.g. Behnke et al., 1993; Walker and Abel, 2002). We argue that research frameworks need to provide understanding of how extreme events affect not only the natural resource base, but also social capital asset changes, which influence people’s long-term adaptation mechanisms and thus their socio-economic resilience. This paper aims to evaluate how sustainable livelihoods analysis can be used in conjunction with theoretical concepts, such as the Panarchy framework proposed by Gunderson and Holling (2002), to better understand the factors which influence (and could quantify) household sustainability. This builds on the Global Environmental Change paper by Fraser et al. (2003) that has outlined the data required to assess ‘environmental sensitivity’ and ‘social resilience’ by providing case study analysis based on how livelihoods analytical approaches can provide information on these variables. The temporal analysis (before, during and after extreme events) enables us to examine how changes in communities, due to long-term environmental changes, policy changes and/or market changes can affect household vulnerability to changing climatic conditions.

Objectives

The general aims of this paper are met by investigation of the following research objectives:

  • To evaluate the ability of, and problems with, ‘static’ sustainable livelihoods analytical approaches to quantify vulnerability to extreme climatic events, as required for a quantitative assessment of household sustainability.
  • To analyse the extra information available from ‘dynamic’ assessments of environmental sensitivity and social resilience from analysis of case studies through time including, pre-, during- and post- an extreme climatic event for a drought event in Southern Africa and a flood event in Bangladesh.
  • To examine whether the Panarchy framework of Gunderson and Holling (2002) offers a methodological tool capable of improving assessments of household vulnerability to extreme climatic events.
  • To propose ‘best-practice’ methodological approaches that can build on sustainable livelihoods approaches by providing more detailed assessments of the ‘vulnerability context’ in terms of assessments of social resilience and environmental resilience to extreme climatic events.

Methodological Framework

Sustainable livelihoods analytical approaches (based on Scoones, 1998) were repeated three times in each case study region. This includes analysis before, during and after the 1999 drought event in Southern Africa when many crops failed in semi-arid mixed farming regions (Dougill et al., 2002); and studies before, during and after the 1998 Bangladesh flood when the equivalent of 68 % of the country was inundated by flood waters (DMB, 1999). Assessments were all conducted at the household level as most people experience and respond to extreme climatic events as members of a household first and foremost. Only by improved understanding of household vulnerability to extreme events can better community, sub-national and national sustainability assessments and policy support be provided. In developing world rural situations the key element in defining a household is the sharing of assets of which the key asset in terms of livelihood security is food, and therefore we define a household as a unit where food consumption is shared.

The nature of the studies conducted in each region differed in accordance with wider research project goals and individual research decisions made in each case. Research in Southern Africa looked specifically at issues on how land use policies either side of a national border (South Africa – Botswana) influence rural livelihoods, whilst also attempting detailed environmental sustainability assessments by developing quantitative participatory nutrient budget analyses for a drought year and subsequent wet year. During the drought of 1999, over 70 semi-structured interviews were conducted across 6 villages in the Molopo area focusing on livelihoods dynamics and recent changes (see Twyman et al., 2004 for details). Subsequently, given the farmers’ concerns over the long-term viability of small-scale agriculture, 15 case study households were chosen in 3 of the study villages for follow-on interviews in November 1999, where recent ‘drought-coping’ strategies were discussed along with quantitative assessments of field nutrient budget models (see Dougill et al., 2002 for details). Finally, return interviews were conducted for the 15 case study households in July 2000 to allow feedback on environmental sustainability analyses and the opportunity for more informal discussions on specific agricultural management issues and household livelihood concerns.

Research in Bangladesh, …

  • Bangladesh rural communities and adaptive strategies to the 1998 flood using livelihoods interview approaches in 1998 (pre and during-flood) and 1999 (post-drought) (based on key bits of Chadwick, 2004 PhD thesis – Matt this to me looks like just a summary overview of very basic methods for your main study villages?).

Analytical Approaches [a1]

Case study results are presented around a series of key themes that address the research objectives identified. In this way, we aim to address theoretical questions on how to measure vulnerability rather than just a presentation of case study material. The key themes that underpin the analysis are to provide:-

  • Descriptions of inter-household differences in each case studies and analysis of potential explanations in terms of capital assets;
  • Evaluation of changing livelihood interview responses through time in relation to extreme rainfall events to assess whether these methods capture simply information on household ‘coping strategies’ or more detailed information on social resilience to external shocks.
  • Identification of how different forms of support by key stakeholder groups (village, Government, NGO and aid institutions) affect household decision-making.
  • Application of Panarchy framework and its components of system wealth, connectedness and diversity to rural households to examine if this can provide a more robust measure of household vulnerability to extreme rainfall events.

The first three of these themes are considered briefly in turn and illustrated with example findings from each of the case study regions / events before a more detailed discussion of how the panarchy framework could improve household vulnerability assessments is provided.

1. Household livelihood assets

2. Changes in livelihoods through time

3. Institutional support through extreme event

Discussion: Panarchy and a New Framework to Study Vulnerability

A range of scholars have proposed preliminary methods to assess vulnerability. This includes Homer-Dixon’s process tracing (Homer-Dixon, 1995), Holling’s Panarchy (Holling, Gunderson, & Peterson, 2002), Alcamo’s security diagrams (Alcamo, 2002; Alcamo & Endejan, 1999; Alcamo, Endejan, Kaspar, & Rosch, 2001), and Kasperson’s regions at risk framework (Kasperson, Kasperson, & Turner, 1995). Although the details of these approaches vary, there are a number of similarities. First, all suggest that these are complex adaptive systems that evolve through time. Any framework must, therefore, be dynamic allowing for comparisons across time. Second, all differentiate between endogamous conditions (i.e. local economic factors and biophysical indicators) and exogamous forces (such as fluctuations in the global market or climatic variation). Third, all stress the need to include social variables and environmental factors. These similarities match much of the focus of sustainable livelihoods analytical approaches there appears suitable scope for basing vulnerability assessments on their use. The main constraint has rightly been identified as the difficulties in using SLA to understand the temporal dynamism of livelihoods (Ashley and Carney, 2000), but this is something our studies aim to overcome by repeating and extending studies through time around an extreme event, or livelihood shock.

These proposals each build on an extensive theoretical literature that discusses vulnerability, resilience, fragility and adaptability as a series of related concepts through which to understand complex human-environmental relations (Fraser, Mabee, & Slaymaker, 2003). For example, Turner et al. define vulnerability as the “…degree to which a system, subsystem, or system component is likely to experience harm due to exposure to a hazard, either a perturbation or stress/stressor “ (Turner et al., 2003). This builds on two basic previous models of risk: (1) the risk: hazard model (ref ??), which looks at the hazard event and the sensitivity of exposed entity, but failed to consider how systems amplify or attenuate the hazard and variation within sub-systems that led to heterogeneous impacts, and the role of institutions in shaping responses; and (2) the pressure and release model (ref ??) that directs our attention towards the conditions that make a situation unsafe. This emphasizes the distinctiveness of the exposed groups (e.g. gender and ethnicity) but fails to understand the coupled nature of human-environmental systems or provides any understanding of how a hazard develops or moves through a landscape. Drawing on these two intellectual traditions, vulnerability assessments could include: entitlements; coping through diversity and resilience.

Differential levels of vulnerability (the ability to cope with a disturbance) between groups has a great deal to do with entitlements (legal, physical and customary rights to command access to necessities of life) so entitlements help explain why some areas are at risk. Social units also have different coping mechanisms, and this may link to a large number of social institutions as well as the basic endowments of particular groups. Diversification of these entitlements and endowments is a strategy to reduce risk. Resilience comes from ecology and refers to the ability to bounce back after a disturbance and maintain certain key functions throughout a disturbance. This can be referred to as the amount of stress that can be maintained. (Turner et al., 2003). However, Adger shows that resilience itself is a contested term (Adger, 2000) and poses a key question: are societies that depend on ecosystems less resilient?

There are two major components to these definitions/ approaches: the first is the impact of landuse on the ecosystem that determine whether current land uses are unsustainable. This is the ecological component of vulnerability. The second is to understand whether people have the ability to adapt and respond. This is the social component. Blakie et al. confirm the distinction between social and ecological by distinguishing between environmental and social resilience, suggesting that vulnerability is a social phenomena that refers to the capacity of human systems to adapt, resist and recover from natural hazards (Blaikie, Cannon, Davis, & Wisner, 1994).

One way to understand the social component is to assess the “architecture of entitlements” (P. Kelly & N. Adger, 2000) or the social, institutional and econmic factors that shape adaptation options. In this way, vulnerability is a function of entitlements (P. M. Kelly & W. N. Adger, 2000) and there are three key components: poverty (as a proxy for marginalization) inequality (as a proxy for social welfare) and institutional capacity (as a way of understanding the role of society in providing entitlements). This builds on the work of Sen (Dreze & Sen, 1989; Sen, 1980a, 1980b, 1987) who suggests that in order to understand vulnerability we need to evaluate all the various strategies that a community or household uses to obtain food. Sen identified three strategies or food entitlements: (1) direct entitlements whereby a family produces their own food; (2) indirect entitlements whereby a family works for a wage and then obtains food from the market; or (3) a transfer entitlement whereby food is obtained through charity or gifts (such as money sent from family members abroad). Problems in the food system arise when a person's or a community's entitlement is disrupted and they cannot switch to another method of obtaining sustenance. The entitlement theory is useful because it disaggregates the reasons why a person or group may become vulnerable to hunger. It also focuses our attention on food systems where people do not have the ability to switch to alternative entitlements if their prime entitlement is disrupted.

To understand the ecological component it [e2]is possible to apply principles from landscape ecology’s Panarchy framework. Landscape ecology is relevant because most ecosystems experience disturbances such as forest fires, windstorms and pest outbreaks (Attwill, 1994). While the timing and nature of these disturbances are impossible to anticipate, the impact of a disturbance depends on three key general characteristics of the ecosystem: the wealth present in the system, the connectivity of individuals within the system as well as the connectivity between systems, and the diversity associated with the system (Gunderson, Holling, & Peterson, 2002; Gunderson & Holling, 2002; Holling, 2001). For example, the impact of a forest fire will be dictated by the amount of fuel available for the fire to consume (wealth), the forest’s linkages to other resources (connectivity), and the age and species distribution of the flora and fauna being destroyed (diversity). A fire in a large, mature forest that is densely planted and only made up of a few species will have greater impact than a fire in a remote, small, and poorly stocked forest; thus, the former system is more vulnerable to the threat of fire than the latter (Fraser & Mabee, Accepted April 2004; Fraser, Mabee, & Figge, Submitted Feb. 2004).

While these three characteristics were first used to describe ecological systems, there is some evidence to suggest they can be effectively applied to human networks as well. The Irish Potato Famine, for instance, resulted when a large number of communities depended entirely on an agro-ecosystem that was biologically wealthy, closely connected and had low diversity (Fraser, 2003). Forest fires that occur in close proximity to human settlement, such as the wildfires in California or British Columbia, have much greater economic impact than do remote disturbances. This framework has also been used to draw parallels between the development of human networks, such as the Hindu caste system, and the evolution of ecosystems (Berkes & Folke, 2002).

Neither framework, however, provides an adequate way of examining the totality of human-environmental relations. However, by combining a socially-driven framework like Entitlements with Panarchy, a tool of landscape ecology, we have ways of exploring both the social and environmental characteristics of vulnerability. Entitlements and Panarchy can be coupled by first assessing the extent to which communities depend on the natural environment for livelihoods, and the Entitlement options available to a community if the environment changes. Once a dependency on the environment has been ascertained it is necessary to then assess the potential, connectivity and resilience of the specific ecosystems in question to determine the vulnerability to external shocks and disturbances (Fraser, 2003). This sort of synthesis builds on work done on resource dependency, which refers to communities and individuals whose societal structure and the stability of livelihoods directly relate to resource production in the local economy (Adger, 2000). Both the landscape ecology framework and the Entitlement framework provide insight into different aspects of human vulnerability. Assessing Entitlements provides understanding on how communities may become vulnerable over time due to socio-economic forces, while Panarchy provides insight into the characteristics of an ecosystem vulnerable to disruption.