« A conceptual framework to assess vulnerability : …A conceptual framework to assess vulnerability. Application to global change stressors on South Indian farmers. St ephanie Aulong

« A conceptual framework to assess

vulnerability : Application to global change

stressors on South Indian farmers »

Stéphanie AULONG

Robert KAST

DR n°2011-03

A conceptual framework to assess vulnerability. Application to global

change stressors on South Indian farmers.

Stephanie Aulong∗ Robert Kast‡

January 6, 2011

Abstract

The objectives of the paper are (1) to apply Fussels (2007) conceptual framework of vulner-ability to a concrete ongoing research and (2) to discuss on the resulting choice of an adequatevulnerability approach. The research aims at assessing the vulnerability of South Indian farmersto global change at two periods of time: medium term (2030-2040) to account for rapid global eco-nomic changes, and long term (2045-2065) to account for climate change and variability. The termvulnerability is defined in so many ways that its use has become controversial. Fussel proposedan original conceptual framework of vulnerability based on a common and transversal terminologyunderstandable whatever the scientific domain of concern. This conceptual framework relies on thedescription of six dimensions of the vulnerability concept. The first four dimensions describes thevulnerable situation and the last two dimensions explain the factors of vulnerability. Fussel arguesthat with this set of dimensions, it is possible to class any conceptual approach of vulnerabilityfound in the literature. After the six dimensions were adapted to South Indian farmers vulner-ability, the use of a cross-scale integrated approach of vulnerability appears clearly as the mostappropriate. Therefore, the use of the classical risk-hazard approach of vulnerability was dismissedas it focuses mainly on social systems and biophysical discrete and regional hazards. Our vulnera-ble situation fits with the so-called integrated approach of vulnerability. Integrated approaches arewidely used in the context of global change and climate change. They can address continuous aswell as discrete hazards and view the vulnerable system as a dynamic one. Among the integratedapproaches is the one proposed by the Intergovernmental Panel on Climate Change (IPCC). Thisapproach has been enlarged and applied to global change hazards: climate change and variabilityplus global economic changes, as the impacts of the last ones are often more severe at least in themedium term (Eakin and Bojorquez-Tapia, 2008; Belliveau et al., 2006; ?). The combination of thethree concepts of sensitivity, adaptive capacity and hazard exposure brings the dynamic dimensionof vulnerability. As a conclusion, we found the Fussel’s transversal terminology particularly func-tional in a context of multidisciplinary research where communication and cross-understanding areof major importance. Going through the description of the six dimensions was also useful to argueon the choice of an appropriate approach of vulnerability. Finally, and as highlighted by Fussel,this conceptual framework gives sense and scientific robustness to the IPCC integrated approachof vulnerability that is now more and more developed in the applied research community.

Acknowledgements: This study has been supported by French Research National Agency (ANR)under the VMCS2008 program (SHIVA project n ANR-08-VULN-010-01).SHIVA website: http://www.shiva-anr.org.Keywords: Adaptive capacity, Agriculture, Global change, India, Sensitivity, Vulnerability con-cept

1 Introduction

The assessment of farmers’ vulnerability to globalchange is currently at the heart of agriculturalpolicies, particularly when food production de-pends on irrigated agriculture and when agri-

culture is still a bedrock of countries’ economy.Then, climate change poses the risk of further de-pressing the agricultural sectors economic perfor-mance. Authors have described the consequencesin various countries, particularly exposed to global

∗BRGM Water Economics Unit, 1039 rue de Pinville, 34 000 Montpellier, France. Email: [email protected]‡IFP, Indian Institute of Pondicherry, Pondicherry, India.

1

change such as South developing countries (Adger,1999; Leichenko and O’Brien, 2002; O’Brien et al.,2004; Acosta-Michlik and Espaldon, 2008; O’Brienet al., 2009). Among these countries, Mediter-ranean ones are prone to vulnerabilities. They areundergoing rapid social and environmental pres-sures like population growth, urbanization and cli-mate change with negative implications on naturalresources (water resources in particular) (Iglesiaset al., 2007). Consequences on farming systemssustainability are straight forward: crops yieldsare decreasing with some years important lost ofproduction as a consequence of droughts. Crop di-versity is suggested as an adaptation strategy butwith no stable empirical evidence (Reidsma andEwert, 2008).The questions we try to answer (partially) in thispaper, came from an applied study of the vul-nerability of Indian farmers confronted to globalchange. We shall refer to that study (SHIVAANR-research project) to give some idea of thedifficulties that were encountered. Though, con-clusions remain true for farmers under stressors ofglobal change in other countries. Reading the lit-erature on the subject (mainly written by geogra-phers and public decision makers: Brooks (2003),O’brien et al. (2000), Alwang et al. (2001), Adger(2006), Smit et al. (2006), Fussel (2007), Eakin etal. (2009)) the main questions that arise are: whatare we talking about? And what are we measur-ing when dealing with vulnerability? Otherwisestated: what is the relevant concept of vulnerabil-ity that is at stake?Within multidisciplinary SHIVA research project,for example, hydrologists understand the termvulnerability as the natural system sensitivityto a well-known hazards such as a pollution ora specific climate events (sea-level rise, flood,etc.); geographers better see in vulnerability theamount of harm caused by the natural events; andeconomists view the concept either as a damagefunction of hazards or as a synonym of the re-silience or the capacity to cope with any shocks.And finally, we wanted to use IPCC definition ofvulnerability which is again an additional defini-tion of the term. A concept only makes sense ina given context. Thus, we needed first a clear in-terpretation of IPCC definition in the context ofglobal change and second a clear understanding ofthe concepts of vulnerability in each disciplines.To these conditions, we would be able to sharea common vocabulary on vulnerability and paral-lel concepts as adaptive capacity, sensitivity, risk,

hazard, etc.. In this search, we find Fussels pa-per (2007) particularly enlightening: it gives themeans necessary to describe different contexts inwhich a concept of vulnerability may be put for-ward (or defined). More precisely, the vulnerabil-ity we may try to measure in a given context maydiffer from the vulnerability concept relevant inanother context. The context mainly depends onthe system under consideration. Then, the otherelements of Fussels paper allow to make it moreprecise, so that at the end, the relevant conceptemerges from its own. In the first part of the pa-per, we present Fussels analysis and show that ityields a complete grid to answer the two previousquestions for most works done on vulnerability.In the second section we insist on elements of dy-namics that are mentioned but not developed inFussel’s conceptual framework, along with exam-ples developed in our study SHIVA. In the lastpart of the paper, we show that vulnerability con-cept used in our context are linked to behaviors.Indeed we show how, in some contexts, vulner-ability is related to the ability/inability to makedecisions. Through this approach, the relevanceand sensitivity to dynamics may be more easilygrasped.In our study, adaptive capacity and sensibility fac-tors are two expressions of the dynamic of the sys-tem under consideration. Here, given we considerhuman systems, part of the system is a group ofdecision makers. A decision maker reacts to haz-ards and uncertainties about social, economicaland environmental factors. These uncertain phe-nomena have their own dynamics that the deci-sion maker figures out through information. Infor-mation arrivals at some future dates are, in turn,the states from which the decision maker choosesamong possible actions: thats the decision processdynamics. If the level of information is too low,for instance, the decision maker is unable to plananything and is inhibited in making decisions. Forexample, the level of information about climatechange is certainly too low at the individual levelto have any prospective vision of the future. Atthe national level, however, this is not the caseand decisions can be made, plans can be designedthat will adapt to future information arrivals andnational or regional programs may be developedto help individuals to adapt. Among those indi-viduals, some may be more able to react to suchplans, hazards and socio-economic uncertainties,this is what we try to measure by adaptive capac-ity factors. Sensitivity, as well, is an indicator of

2

the possible impact of information on individualdecision making.

2 Fussel conceptual frameworkof vulnerability: applicationto Indian farmers’ vulnerabil-ity to global change

Fussel’s conceptual framework combines a nomen-clature of vulnerable situations and a terminologyof vulnerability concepts aiming at characterizingthe vulnerability concepts employed in the mainschools of research. The nomenclature of vulnera-ble situations is described through four dimensionsof a given vulnerable situation. The first funda-mental dimension is the system of analysis. It ad-dresses either natural systems (ecosystem, aquifer,plant, animal, soil, etc.) or human systems (apopulation, an economic sector, a country, etc.).As an example, within SHIVA study, we are con-cerned with the farmers of South India. By SouthIndia we mean a geographic area that shares thethree following important characteristics to un-derstand farmers’ vulnerability to global change.(i) Food production relies mainly of groundwa-ter irrigation. (ii) Farmers have to compose withsemi-arid climatic conditions: with rainfall rang-ing from 600 to 1100 mm in this area, farming isgreatly reliant on monsoon quantity and quality.And (iii) farmers are constrained by hard rocksgeological context which limits groundwater stor-age in the aquifers (low recharge capacity). Thefocused area is presented in Figure 1. By farmers,we mean the operational landholders within theiroperational land unit (or the farming system). Ac-cording to the focused area described above, weare then analyzing rural farmers and their farm-ing system relying on monsoon and groundwaterirrigation.The second dimension of vulnerable situations isthe attribute(s) of concern. The attributes of con-cern are ”the valued attributes of the vulnerablesystem that are threatened by its exposure to ahazard”. Attributes can again be natural (biodi-versity, timber production, etc.) or socioeconomic(income, health, etc.). Going on with our SHIVAstudy, we are interested in how farmers’ liveli-hood sustainability is impacted by global change.This is to say that we are considering a set of im-pacted attributes (income, water resources, soil,etc.) that defines farmers’ livelihood resources.Our approach is based on Sustainable Livelihood

Approach (SLA) (Orr and Mwale, 2001; Scoones,1998). The SLA defines households capacities ac-cording to five capital assets: human, physical,natural, financial and social assets. The analy-sis of farmers’ capital assets and their evolutionfurther allows to discuss the farmers’ livelihoodstrategy trends.The third dimension is the hazard which Fusseldefines as ”a potentially damaging influence onthe system of analysis” and ”some influence thatmay adversely affect a valued attribute of a sys-tem”. The hazard of interest is discrete (pertur-bations) or continuous (stressors). A fine descrip-tion of the hazard under consideration is impor-tant as it introduce a temporal dimension in theanalysis. As a consequence, a dynamic perspec-tive of the system may be required. This is in-deed the case in our research where we are study-ing the effects of the multiple stressors of globalchange in the future. As we focus on one side,on climate change and variability, keeping asideextreme events (storms, unusual floods, etc.), andon another side, on likely future economic changes,we are looking at continuous hazards in the termsof Fussel. Thus, vulnerability assessment must bedynamic as it translates a succession of adaptationprocedures in reaction to ongoing changes (envi-ronmental and economic).At last, the fourth dimension of vulnerable situ-ations is the temporal reference which can be apoint in time or a period of interest. This di-mension introduces a kind of temporal limits tothe dynamics we want to give to the vulnerabilityconceptualization. It is not clear in Fussel’s frame-work how to tackle this issue within quantitativeassessment. Continuing on SHIVA study example,two periods of time are considered in addition tothe assessment of current vulnerability, in order totake into account the different dynamics of climateand economic changes (2020-2030 and 2045-2065).But the dynamic view of the system is restricted tothe comparison of these three snapshots of vulner-ability states. A real dynamic conceptualizationwould imply a marginal approach of vulnerabilityquantification.As a result, these four dimensions of Fussel’s vul-nerable situations nomenclature give a completeand precise answer to our first question What arewe talking about?. Fussel’s second step of his ap-proach offers a terminology to identify and charac-terize vulnerability concepts. This additional de-scription of vulnerable concepts allows answeringto our second question What are we exactly mea-

3

suring as vulnerability?. Vulnerability conceptsare described through a two dimensions matrixof vulnerability factors. These two dimensions arethe sphere or scale of and the knowledge domain.Sphere is internal when referring to ”propertiesof the vulnerable system” (income, soil quality,etc.) or external when referring to ”somethingoutside the system” (groundwater policies, marketprices, pest attacks, bad monsoon). If the sphereof concern refers to both internal and external fac-tors, then Fussel speaks of ”cross-scale factors”.Knowledge domain is split in two categories ofvulnerability factors: socioeconomic factors andbiophysical ones. Socioeconomic factors relate to”characteristics of social groups” (income, ground-water policies, market prices) whereas biophysi-cal ones relate to ”system properties investigatedby physical sciences” (soil quality, bad monsoon).When factors are of both domains, Fussel speaksof ”integrated vulnerability factors”.Going on our SHIVA example, internal sphere fac-tors of vulnerability are identified through the de-scription of the farming system itself with empha-sis on its performance. In our approach, perfor-mance is assessed by farmers’ capacity to adapt tochanges which is itself based a Sustainable Liveli-hood Approach (SLA) (Orr and Mwale, 2001;Scoones, 1998). Thus, internal sphere factors referto the five capital resources described by Scoones(see Section 2). Without going deeper in our con-ceptualization, we can already see that some cap-ital resources are of the socioeconomic knowledgedomain whereas others are from the biophysicalknowledge one. On the other side, we use thePorter’s diamond framework adapted from a firmto a farming system in the line of Vandermeulenframework (2009), in order to identify the exter-nal factors that influence the development of thefarming structure. This conceptual framework al-lows a concise description of external forces, tak-ing into account altogether the context of farm-ing system strategy that can be natural (rainfall,temperature, altitude, water availability, etc.) oreconomic (agricultural policies, subsidies, invest-ments, etc.); the evolution of farming productsdemand and consumption; the evolution of up-stream and downstream industries (e.g. fertil-izer or machinery industries, sales cooperatives,regional trade); and the factor conditions (land,labour, information, infrastructures, etc.). Again,factors that we take into account in our analy-sis are of both knowledge domains (socioeconomicand biophysical).

Fussel’s matrix of vulnerability factors, in its ba-sic conception, offers a set of factors combinationsthat is sufficient to depict any vulnerability ap-proaches found in the literature (Fussel, 2007, p.160). In the classical risk-hazard approach, forexample, vulnerability concept is defined as thelevel of damage or loss due to the severity of thehazard of concern. Our project’s geographers canthen recognize their usual approach of vulnerabil-ity. It can also be denoted as dose-response orexposure-effect relationship. In this case, vulnera-bility is equivalent to sensitivity as in our hydrolo-gists’ conceptualization: the properties of the vul-nerable system are only considered to estimate theimpacts of the hazard and the vulnerable system isexclusively restricted to physical systems, naturalor not (equipment, production assets like land).Vulnerability concept in risk-hazard approach isthen denoted as internal biophysical vulnerability.In the second main approach, which is the politi-cal economy one, vulnerability concept focused onpeople livelihood and well-being and their abilityto cope with and adapt to stresses. Here SHIVAeconomists will find their own traditional concep-tualization of vulnerability. Internal as well as ex-ternal factors are generally analyzed, but knowl-edge domain of concern is exclusively socioeco-nomic. This approach is classed as a cross-scalesocioeconomic vulnerability approach. There ex-ists cross-scale integrated approaches of vulnera-bility that takes into account all four kinds of vul-nerability factors. They combine risk-hazard andpolitical economy approaches as factors of vulner-ability can come from the properties of the sys-tem as well as from its external environment. Inaddition, vulnerability factors can be both of so-cioeconomic (e.g. market failure) and of biophys-ical domain (e.g. storm). Our SHIVA study isin the line of these cross-scale integrated conceptsof vulnerability as the above description of factorsshows. That is why the IPCC definition totallyfits with the context of vulnerability assessmentwithin SHIVA study.

3 Passways to characterize thedynamics of vulnerability

Public decision makers as well as experts and sci-entific studies cannot avoid considering the gen-eral evolution of the system under consideration.In most cases, plan makers have limited their vi-sion of the evolution to some perspective theywished to impose (e.g. 5 years plans in France,

4

India and the USSR). This limitation has lead tomany inabilities to adapt such plans to hazards,social movements or new instruments. Dynamicsdoesn’t just mean we take the temporal referenceinto account. It amounts to describe in details dif-ferent processes and their interactions (e.g. a haz-ard stressor and socio-economic processes). Fur-thermore, when decision makers are part of thesystem (as is our case), dynamics concerns deci-sion making processes that respond to externalhazards. This is why a prospective approach isneeded in order to introduce flexibilities, optionsand more generally to integrate the possibility toadapt to new information. Even in the case of awell-known natural hazard, the ways to react toit may depend on other parallel processes such asnew technology, social changes or other hazards.A prospective vision of the system evolution in-cludes the different factors processes affecting theattributes, the different relevant hazards amongthe spheres (internal or external, and in our case,integrated). Obviously, the temporal reference isfundamental in a prospective approach as it limitsthe different dynamics that interact and are takenunder consideration. Leichenko et al. (2002) sug-gest that

”traditional vulnerability indicatorsmay be insufficient in capturing thenature of global change, including itsmany dimensions and its diverse ef-fects at different scales of analysis.[...] one possible strategy for ad-dressing dynamic vulnerability via amulti-scale method of analysis wouldbe to combine macro-level vulnerabil-ity mapping using dynamic indicators[...], with local-level survey-based in-vestigations of how changing economicpolicies are affecting farmer and insti-tutional response to climate variabil-ity.”

The authors then proposed to assess changes in aset of vulnerability indicators during one periodof time, but they didn’t applied the methodol-ogy in a concrete case study. That’s what wetried to do within SHIVA project (i) by down-scaling global change effects, either climatic andeconomic, and (ii) by incorporating elements ofdynamics in adaptive capacity and sensitivity in-dexes. However, dynamics is limited to some largetime intervals.First of all, in a pilot phase, we try to identify themajor hazards to which a farmer has been exposed

and is sensitive. Results show that droughts, badmonsoons, pest attacks, low groundwater table,input prices increase and output prices decreasewere the major hazards encountered. Faced tothem, farmers use to adjust their strategy (or takemanagement decisions). As a matter of fact, thesehazards perceived by farmers can easily be linkedto climate and economic changes. Thus, we in-troduce a prospective vision of the hazards un-der concern, global change, with a particular at-tention to those hazards perceived locally (Bel-liveau et al., 2006). To forecast these changes,we use medium and long term IPCC scenariosof climate change. IPCC scenarios are derivedfrom the future simulation of gas emissions whichare themselves simulated according to hypothe-ses about countries’ future behavior in terms ofenvironmental concern and market exchanges (so-called SRES scenarios). These scenarios are thestart of our prospective approach of global changehazards. Though, additional work was achievedto downscale SRES and IPCC scenarios over theSHIVA area of concern. Only to this condition,differentiate impacts of global change on farmerscan be geographically observed. For the climatechange part of downscaling, mathematical GlobalClimate Models (GCMs) exist and downscaling ofmodels is now a common work in research projectsdealing with climate change impacts assessment.From GCM time frames comes the temporal di-mension of our study.Besides, the downscaling of SRES/economic sce-narios is clearly less ”trivial”. We consider onemost probable scenario in India: the A2 scenariothat is a continuation of what already happens,that is limited sustainable development and verymoderate environmental consciousness. Then, thedownscaling of this scenario in South India canbe achieved through a prospective research on theeconomic factors that will influence the farmingsystems within the time frames considered (2020-2030; 2045-2045). These economic external fac-tors are identified when describing the externalsphere factors (according to Fussel’s terminology,see section 2). National and regional plans, strate-gies and vision for these factors are then ana-lyzed together with experts points of view in or-der to evaluate future trends. Within this eco-nomic prospective approach, we had to take intoaccount some adjustment (or adaptation) made bygovernment facing global change. Indeed, at na-tional and likely regional scales, decision makersare informed of global change hazards and their

5

potential impacts. Plans usually integrate generalmeasures to limit negative effects of changes (col-lective adaptation capacity). As an example, wa-ter resource management plans include measuresas increasing storage capacity, more efficient use ofwater or more efficient irrigation systems. Then,even if, at the very local scale, the farmers are notinformed of hazards, they will be encouraged tochange some elements of their system.Though, the farmer (the decision maker) has tofaced many other hazards than those derived fromglobal change: health hazards, personal hazards,production hazards, institutional hazards, etc..And finally, farmers’ decisions are taken in thiscontext of multiple hazards. That is why it isessential to describe farmers within a broadly en-vironment and apply prospective analysis on themajor driving forces influencing the system. As anexample, food production is expected to remain apriority for Indian States with consequence suchthat a shift from subsistence to economic farmingsystems, and likely an important decrease of thefarmers population. As a consequence, in AndhraPradesh, farmers population in 2030 is expected todecrease by 30%. Livestock production won’t besupported by government whereas more efficientirrigation will be subsidized and cash crops pricesshould be sustained by government. Energy priceshould grows up at international level with im-pacts on Indian farmers as local energy shouldn’tremain free. The construction of new Hyderabadairport has impacts on farming systems and jobopportunities. Analyzing farmers local and re-gional environment thus allows to built relevanthypothesis on national and regional economic haz-ards that are likely to impact farmers and farm-ing local economies. These hazards are also de-scribed as ”driving forces” within SHIVA studyand we use Portal’s diamond conceptual frame-work to identify main forces. The incorporationof some of these elements into vulnerability index(which combines sensitivity and adaptive capacityindexes) contributes to improve the dynamics ofthe concept. In the line of Leichenko et al. (2002)and Belliveau et al.(2006), we thus identified someindicators that could be ”modeled” or ”simulate”at future dates. As an example, sensitivity indexincludes indicators such as farming income evolu-tion within the household, in order to grasp pasttrends; or the percentage of the cashcrops area, toget an idea of farmers’ dependence to governmentsupport. Besides, adaptive capacity index incor-

porates indicators as education, health, infrastruc-ture and information availability, operational area,production means, etc., for which hypotheses canbe done for their evolution in the future (see Fig-ure 3 of adaptive capacity index). Indeed, fu-ture vulnerability not only relies on climatic ornon-climatic conditions found to influence farm-ers. As we try to explain, farm-level factors arealso likely to evolve and change the farmers’ ca-pacity to adapt to hazards. For some indicators,projection scenarios should bring more dynamicsinto the adaptive capacity concept.

4 Vulnerability concept ob-served from a decision-makingapproach

According of what we have described above, inorder to better understand the dynamics of vul-nerability, it might be helpful to analyze the sys-tem and its own dynamics. In our case wherepart of the system is a decision maker, that isan active entity, this dynamics is described by thedecision process itself, which responds to hazardprocesses. In this context, the vulnerability con-cept could be approached by an analysis of deci-sion makers ability/inability to react to hazards.Otherwise stated, vulnerability would amount toa limitation of the ability to take risks in orderto mitigate and/or compensate the risks that arefaced. It could be related to constraints that re-duce the range of optimal decisions. It could berelated as well to something much more difficultto grasp: a change in the decision makers pref-erences. This could be the case if, for instancein front of an unexpected hazard, family safetypassed in front of the timing to plant. There isindeed a close relationship between factors of at-tributes that are integrated in the vulnerabilitymeasures and the variations of constraints limitingthe decisions range. Following this point of view,it may be enlightening to integrate more questionsabout the decision process into the inquiries, as-suming the decision maker is dynamically consis-tent (preferences are not assumed to change) in or-der to grasp the capacity to make up a prospectiveview (in opposition with the submission to habits),the relative importance of constraints (e.g. mone-tary vs working time), the relative importance ofrelevant events (subjective probability).In August 2009, a survey of 153 farmers strati-

1Surveys and data management were piloted by the Center of Economic and Social Studies, CESS, Hyderadab.

6

fied by operating area size was carried out in asmall pilot watershed1. The questionnaire aimedat characterizing the household composition andidentity, the household sources of income andother financial determinants, the farming systemin terms of cropping patterns, water uses andtechnical assets, plus additional questions aboutperceptions of climatic, political and economicevents and shocks. Following IPCC definition ofvulnerability as a function of exposure, sensitiv-ity and adaptive capacity, but merging exposureand sensitivity on the basis that working at in-dividual level exposure and sensitivity were in-timately related, we built a composite index ofvulnerability made of adaptive capacity and sen-titivity/exposure indices. Adaptive capacity in-dicators identification is based on Scoones 1998sustainable livelihood approach (SLA - 5 capitalassets). The choice of indicators for each assetsis based on a past research using SLA for wa-tershed management in India (Reddy and Ready,2004) plus discussions with local experts. Sen-sitivity indicators integrate components of haz-ards perception, natural or economic. Farmersare asked to assess the number of shocks that oc-curred during the last 5 years and if they suf-fered from these shocks the year just before thesurvey. This is to integrate the hypothesis thatrecent shocks are more likely to impact currentfarmers’ decisions. Additionally, sensitivity in-dex is composed of livelihood sensitivity indica-tors about past trends of farming income, irri-gation costs, government programs participation,impacts of shocks on ability to buy food, and soon. Adaptive capacity and sensitivity indicatorsare then organized into a hierarchic matrix accord-ing to Analytic Hierarchic Process (AHP) method(Figures 3 and 4). After indicators are (i) val-ued through survey responses, and standardizedto a single [0,1] scale, and (ii) weighted by ex-perts through pairwise comparison of indicators,Compromise Programming (CP) is used to aggre-gate indicators value and weight (see, Eakin et al.2006 for methodological development). ThroughCP, a distance to the anti-ideal point is calculatedfor each farmer and both indexes of adaptive ca-pacity and sensitivity separately. Here, the anti-ideal point is represented by maximum vulnera-bility (the value of vulnerability score is then 1).The greater the distance to the anti-ideal point,the higher the capacity to adapt, and the lower

the sensitivity. Formally, the distance to the anti-ideal point for a farmer is calculated as follows:

di =

J∑j

wpi (1 − xpij)

1/p

(1)

where di is the distance to the anti-ideal point ofthe ith farmer, wj the weight of the jth indicator,xij the standardized score of the ith farmer for thejth indicator, and p a constant metric indicatinghow compensated a decrease/increase in one indi-cator can be by the increase/decrease in anotherindicator. Here we took p = 1, meaning a perfectcompensation.Fuzzy logic is finally used to compile both dis-tances and obtain a final vulnerability score perfarmer2. At the end, farmers are ranked accord-ing to their vulnerability score. The closest toa score of one, the more vulnerable the farmeris; conversely, the closest to zero, the less vulner-able he is. We found 15% of Gajwel watershedfarmers belonging to the high vulnerability classwith average score of 0.652 and 56% belonging tomedium vulnerability class (score of 0.484) (Fig-ure 5). We found no farmer in the ”very highvulnerability” class. Analyzing the origins of vul-nerability from adaptive capacity point of view, weobserve that3: (i) high vulnerable farmers partic-ularly suffer from a lack of information on weatherforecast and agricultural new practices and inno-vations; (ii) their sources of income are less diver-sified and their total income (household income)is lower; (iii) savings are less important; (iv) thiscategory of farmers also suffers from the lack ofpublic infrastructures (access to roads and urbanfacilities) (Figure 6). By the way, crop diversitydoes not seem to be a good strategy to ease farm-ers’ adaptive capacity in this local context: in-deed, this is not a significant variable to explainbetween classes vulnerability variability. Sensitiv-ity indicators contribution to vulnerability is lesstrivial (Figure 7). What can be said is that somefarmers who appears as more sensitive to marketshocks, are yet attached to the ”very low” vulner-ability class. Their sensitivity is compensated bya high adaptive capacity score, showing their abil-ity to take good decisions when faced to sensitivesituations. Participation to government programsappears clearly as a sign of higher vulnerability.At last, the vulnerable farmers are those who areaffected by hazards in such a way that it affects

2We use FisPro software developed by INRA and CEMAGREF, France.3Most vulnerable farmers are located at the periphery of the star whereas less vulnerable ones are in the center.

7

their ability to buy food. Indicators on past haz-ards perceptions doesn’t bring the expected con-tribution to vulnerability score, at least in the con-text of a small watershed as Gajwel one (80 km2).This will be verified over larger basins (700 km2).The results presented show the current farmers’vulnerability scores. Future vulnerability scoreswill be computed based on prospective simula-tion of a set of indicators. Particularly, hazardswill be simulated in order to grasp their influenceon future vulnerability. Then, perceptions willbe replaces by corresponding simulated indicators.In the same way, hypotheses on irrigation coststrends, participation to government programs, etc.will be tested. On the adaptive capacity side, dy-namics will be also introduced through the simula-tion of a number of indicators for which hypothesiscan be made. This will be possible for indicatorson infrastructures and services availability. Thisis all the more important that they are signifi-cantly contributing to vulnerability scores. Localexperts’ knowledge will be used to simulate theseindicators. Among the ways we investigate to im-prove the dynamics of the vulnerability concepts,is the use of farmers’ typology approaches: typolo-gies are a way to study the system of concern andinvestigate its own dynamics. We first use officialtypology of Indian farmers based on operationalland size. But we also built our own farmers’ ty-pology using again SLA to range farmers accord-ing to their current livelihood strategy4. We iden-tify 6 groups of homogeneous farmers that werelinked to 3 specific livelihood strategies. Farmingstrategies are then discussed with experts in orderto forecast their evolution. Finally, we also studyfarmers according to their geographic location (ru-ral/urban villages), as we find that infrastructuresand services are significantly influencing farmers’

vulnerability. Then, discussing with new experts,we hope to develop new hypotheses that could dif-ferentiate between and within farmers’ strategies.Figure 8 shows an idea of farmers’ current vulner-ability according to the typology used. Variabilityalready clearly appears between groups, and ori-gins of vulnerabilities can be identified thanks tovulnerability indicators (in the same line of Fig-ures 6 and 7).

5 Conclusion

The adaptive capacity figured out in Shiva, is aproxy for the Indian farmers’ capacity to takerisks. Indeed, the system we study is not lim-ited to farmers only, but extends to parts of theirenvironment as was explained in the previous sec-tions. As a result, the vulnerability measures dy-namics is constrained by the dynamic of the envi-ronment (bio-physical and socio-economical) andcannot grasp the much faster dynamic of the deci-sion making processes. This is why our dynamic islimited to two periods, whilst farmers may adapttheir behaviors to changes from year to year. Fur-thermore, information contained in the enquiriescould let us follow the farmers dynamics but is in-tegrated here in the indexes. Obviously, such adynamic could only be described with some moreinformation, notably the farmers attitudes towardrisks. In our enquiries, these attitudes can beguessed from some answers, but were not soughtfor. Indeed, it would have not make sense to gofurther in details without a theory to rely on. Sucha theory, relating vulnerability to risk attitudes,and hence its dynamic to the decision processes offarmers according the their attitudes toward risks,will be developed in a forthcoming paper.

References

L. Acosta-Michlik and V. Espaldon. Assessing vulnerability of selected farming communities in thephilippines based on a behavioural model of agent’s adaptation to global environmental change.Global Environmental Change-Human And Policy Dimensions, 18(4):554–563, 2008.

W. N. Adger. Social vulnerability to climate change and extremes in coastal vietnam. World Devel-opment, 27(2):249, 1999.

W. N. Adger. Vulnerability. Global Environmental Change, 16(3):268, 2006.

J. Alwang, P. B. Siegel, and S. L. Jorgensen. Vulnerability: A view from different disciplines. page p.42, 2001.

4The typology is based on statistical cluster analysis over the Gajwel farmers’ sample.

8

S. Belliveau, B. Smit, and B. Bradshaw. Multiple exposures and dynamic vulnerability: Evidencefrom the grape industry in the okanagan valley, canada. Global Environmental Change, 16(4):364 –378, 2006.

N. Brooks. Vulnerability, risk and adaptation: A conceptual framework. Working Paper 38, November2003.

H. C. Eakin and L. A. Bojorquez-Tapia. Insights into the composition of household vulnerability frommulticriteria decision analysis. Global Environmental Change, 18(1):112 – 127, 2008.

H. C. Eakin and A. L. Luers. Assessing the vulnerability of social-environmental systems. AnnualReview of Environment and Resources, 31:365–394, 2006.

H. C. Eakin and M. B. Wehbe. Linking local vulnerability to system sustainability in a resilienceframework: two cases from latin america. Climatic Change, 93(3-4):355–377, 2009.

H.-M. Fussel. Vulnerability: A generally applicable conceptual framework for climate change research.Global Environmental Change, 17(2):155–167, 2007.

A. Iglesias, L. Garrote, F. Flores, and M. Moneo. Challenges to manage the risk of water scarcity andclimate change in the mediterranean. Water Resources Management, 21:775–788, 2007.

R. M. Leichenko and K. L. O’Brien. The dynamics of rural vulnerability to global change: The caseof southern africa. Mitigation and Adaptation Strategies for Global Change, 7(1):1–18, 2002.

K. L. O’Brien and R. M. Leichenko. Double exposure: assessing the impacts of climate changewithin the context of economic globalization. Global Environmental Change-Human And PolicyDimensions, 10(3):221–232, 2000.

K. L. O’Brien, R. M. Leichenko, U. Kelkar, H. Venema, G. Aandahl, H. Tompkins, A. Javed, S. Bhad-wal, S. Barg, L. Nygaard, and J. West. Mapping vulnerability to multiple stressors: climate changeand globalization in india. Global Environmental Change-Human And Policy Dimensions, 14(4):303–313, 2004.

K. L. O’Brien, T. Quinlan, and G. Ziervogel. Vulnerability interventions in the context of multiplestressors: lessons from the southern africa vulnerability initiative (savi). Environmental Science &Policy, 12(1):23–32, 2009.

A Orr and B. Mwale. Adapting to adjustment: Smallholder livelihood strategies in southern malawi.World Development, 29(8):1325 – 1343, 2001.

V. R. Reddy and M. G. Ready. Participatory watershed development in india: Can it sustain rurallivelihoods? Development and Change, 35(2):297–326, 2004.

P. Reidsma and F. Ewert. Regional farm diversity can reduce vulnerability of food production toclimate change. Ecology and Society, 13(1), 2008.

I. Scoones. Sustainable rural livelihoods. a framework for analysis. IDS Working Paper 72, IDS,University of Sussex, Brighton, UK., 1998.

B. Smit and J. Wandel. Adaptation, adaptive capacity and vulnerability. Global EnvironmentalChange-Human And Policy Dimensions, 16(3):282–292, 2006.

V. Vandermeulen, X. Gellynck, Guido Van Huylenbroeck, J. Van Orshoven, and K. Bomans. Farmlandfor tomorrow in densely populated areas. Land Use Policy, 26(4):859 – 868, 2009.

9

Figure 1: Location of the SHIVA project study area with 3 pilot watersheds.

Figure 2: Distribution of holding area within the Gajwel mandal (small pilot watershed). Marginal:lower than 1 ha; small: 1-2 ha; semi-medium: 2-4 ha; medium: 4-10 ha; large: above 10 ha.

10

Figure 3: Matrix for farmers’ adaptive capacity index.

Figure 4: Matrix for farmers’ sensitivity index.

11

Figure 5: Results of farmers’ vulnerability assessment.

Figure 6: Origins of farmers’ vulnerability: adaptive capacity indicators.

12

Figure 7: Origins of farmers’ vulnerability: sensitivity indicators.

Figure 8: Vulnerability results according to farmers’ typologies.

13

Documents de Recherche parus en 20111

DR n°2011 - 01 : Solenn LEPLAY, Sophie THOYER

« Synergy effects of international policy instruments to reduce

deforestation: a cross-country panel data analysis »

DR n°2011 - 02 : Solenn LEPLAY, Jonah BUSCH, Philippe DELACOTE, Sophie

THOYER

« Implementation of national and international REDD

mechanism under alternative payments for environemtal

services: theory and illustration from Sumatra »

DR n°2011 - 03 : Stéphanie AULONG, Robert KAST

« A conceptual framework to assess vulnerability. Application to

global change stressors on South Indian farmers »

1 La liste intégrale des Documents de Travail du LAMETA parus depuis 1997 est disponible sur le site internet :

http://www.lameta.univ-montp1.fr

Contact :

Stéphane MUSSARD : [email protected]