The costs and risks of coping with drought: livelihood impacts and farmers' responses in Burkina Faso

CLIMATE RESEARCHClim Res

Vol. 19: 119–132, 2001 Published December 4

1. INTRODUCTION

In 1997 rural families of Burkina Faso experiencedone of the driest years in decades, which resulted inwidespread crop losses and food deficits. Over 250 000people faced high levels of food insecurity as a result ofthis drought.1 This paper analyzes the responsesenacted by families of the Central Plateau during theyear that followed the drought. In particular, we exam-ine how food procurement strategies, productionadjustments and livelihood outcomes were shaped byhousehold resource access profiles and livelihood port-

folios (Blaikie et al. 1994, Scoones 1998). The case inquestion represents the livelihood and productiondilemmas faced by farming households in areas char-acterized by high levels of climatic risk and low naturalresource endowment, where absolute income loss maybe lower than in better endowed areas but the human

© Inter-Research 2001 · www.int-res.com

*E-mail: [email protected]

The costs and risks of coping with drought:livelihood impacts and farmers’ responses in

Burkina Faso

Carla Roncoli1,*, Keith Ingram1, Paul Kirshen2

1Department of Crop and Soil Sciences, The University of Georgia, 1109 Experiment Street, Griffin, Georgia 30223-1797, USA

2Civil and Environmental Engineering Department and Fletcher School of Law and Policy, Anderson Hall, Tufts University,Boston, Massachusetts 02155, USA

ABSTRACT: This paper analyzes the responses enacted by families of the Central Plateau in BurkinaFaso during the year that followed a severe drought in 1997. We illustrate the agro-ecological andsocio-economic contexts that shape livelihood options and constraints in an area characterized byhigh levels of climatic risk and low natural resource endowment. A description of farmers’ percep-tions and official accounts identifies key criteria whereby farmers formulate evaluations and predic-tions of a season. We document how food procurement and management practices are shaped byhousehold resource access profiles and livelihood portfolios. Livelihood diversification, encompass-ing migration, non-farm work and social support networks, in addition to livestock production, isshown to be a critical dimension of adaptation. Livelihood and production adjustments entail costsand risks for most, but also gains for those who have the resources needed to take advantage of dis-tress sales and high prices of agricultural commodities. Household livelihood and risk managementincreasingly hinge on efforts by household members who traditionally have had marginal access toresources, especially women. The research points to the need for closer integration of drought pre-paredness efforts, farmers’ understanding of climate-crop interactions and interventions that bolsterthe capacity of resource-limited households to respond. Affordable grain, locally adapted seed vari-eties, labor saving technology and flexible credit are among the most needed inputs.

KEY WORDS: Rainfed agriculture · Burkina Faso · Coping strategies · Food security · Householdlivelihood · Drought · Climate variability · Risk management

Resale or republication not permitted without written consent of the publisher

1http://www.fews.org/bf980262/sh980626.html#bf. In thispaper we use the term ‘drought’ to translate the commonFrench term sécheresse used by educated local peoplewithout entertaining the debate of its meteorological,hydrological or agronomic definitions. The Moré term warérefers to both a dry year and a dry period during the rainyseason, including failure of expected rains event, such asplanting rains

Clim Res 19: 119–132, 2001

costs may be higher because many households operatenear the margins of subsistence (Scoones 1998).

The identification of factors that shape vulnerabilityand the documentation of the type, range andsequence of coping strategies can inform policies andprograms to reduce exposure to risk, to assess needand prioritize interventions, to alleviate the impacts ofstress and to enable those affected to recover in timelyfashion (Corbett 1988, Campbell 1990, Downing 1990,Adams et al. 1998, Benson 1998). But some have dis-puted the concept of ‘coping’ as being too static or toosimplistic, not accounting for the dynamic and diversenature of rural livelihoods. They warn against the dan-ger of evaluating risk and need on the basis of decon-textualized observation of household behavior (Lam-bert 1994, Davis 1996). Such behavior, in fact, mayindicate different conditions in different places ortimes. Fungible resources may be alternatively allo-cated to consumption, production or reproduction(Downing 1990, Lambert 1994). For example, animalsales may be a stage in resource accumulation; that is,animals bought with revenue from crop sales may besold to hire labor for farm operations or as starting cap-ital for dry season enterprises. Grain may be bought forfood, but also used to sow fields, entertain labor par-ties, host funeral celebrations or brew beer for sale.

Recent reviews of coping strategy research havepointed to the need for contextualized analysis of vari-ation between and within groups in the ability to with-stand and recover from food crises (Campbell 1990,Adams et al. 1998, Hussein & Nelson 1998). Short-termsurvival may undermine long-term viability or mayweigh unequally upon social categories, economic sec-tors and natural resources. At times of shortage,women may endure the costs of others’ survival,depriving themselves in order to feed husbands andchildren or using up their petty trade capital to buyfood (Whitehead 1981, Cutler 1984, Jiggins 1986, Cor-bett 1988, Davis 1993).

Research on household coping strategies stemmingfrom the major West African droughts of the mid-1970sand mid-1980s documents the predicaments of house-hold decision making at times of food crisis, whenscarce resources must be allocated to the competing, attimes contradictory, demands of production and repro-duction (Adams et al. 1998). Households may elect toforego reproductive needs in order to hold on to theirproductive viability by reducing food consumption andsocial obligations, or they may be forced to depletetheir productive assets and buffer stocks (Devereux1993, Lambert 1994, Fafchamps et al. 1998). Resource-limited farmers may choose to grow lower yieldingdrought-resistant crops and varieties, foregoinggreater returns to land and labor (Dercon 1996). Anappreciation of these trade-offs should inform policy

approaches aimed at bolstering the coping capacity ofrural livelihoods to avoid reinforcing the marginaliza-tion of certain groups or perpetuating unproductive orunsustainable resource use (Adams et al. 1998).

An analysis of the differing abilities of households toabsorb and recover from climate shocks has beenrecently highlighted as key to a definition of sustain-ability (Davis 1996, Scoones 1998). Evidence from WestAfrica, and especially recent research carried outwithin the sustainable livelihood framework of analy-sis, points to diversification as a crucial mechanismwhereby households cope with high levels of incomeand production uncertainty (Painter et al. 1994, Davis1996, Ellis 1998, Brock 1999). Agricultural diversifica-tion allows farming households to diffuse risks by mix-ing crops and varieties in the same field, by staggeringplanting and scattering crops in different soils, fieldsand locations, and by enacting water conservationmeasures (Sanders et al. 1990, Carter 1997). Economicdiversification, or the ability to spread risk over spaceand sectors, broadens households’ access to resourcesand livelihood options among activities with differentsensitivities to climatic shock. Inter-regional and rural-urban linkages and non-farm income sources are keyfactors that farming households can draw upon in cop-ing with climate shocks (Reardon et al. 1988, Reardon& Matlon 1989, Reardon & Taylor 1996). Finally, socialdiversification and involvement and investment insocial networks mediate access to key resources andprovide support during time of crisis, especially for therural poor (Bratton 1987, Berry 1989, Carter 1997).

The long-term implications of livelihood diversifica-tion and other risk management strategies for sustain-able development need to be assessed in light of theircontext specificity and their differentiated nature(Hussein & Nelson 1998). The case study presentedhere illustrates the diversified matrix of responsesenacted by differently endowed households and bydifferently positioned household members, as well asthe tensions and negotiations they set in motion. It alsobridges the immediate aftermath of the drought, inwhich households strategize to cope with food insecu-rity, and the longer-term reverberations of climateshock that stem from the impacts such strategies haveon agricultural practices during the subsequent rainyseason. Finally, we suggest how scientific information,policy approaches and development interventions cancontribute to a better integration of equity and sustain-ability concerns into drought preparedness efforts.

2. CONTEXTS OF COPING

The findings presented here emerged in the courseof ethnographic research undertaken as part of the

120

Roncoli et al.: Coping with crisis in Burkina Faso

Climate Forecasting and Agricultural Resources(C FAR) project, a multidisciplinary initiative fundedby the US National Oceanic Atmospheric Administra-tion’s Office of Global Programs and jointly imple-mented by the University of Georgia and Tufts Uni-versity between 1997 and 2000. This project aims fora better understanding of how both regional andnational institutions and farming households of theSudan-Sahel region of West Africa can use seasonalprecipitation forecasts to improve resource productiv-ity in farming systems. The ethnographic componentof the project explored the role of rainfall in produc-tive and livelihood strategies in 3 different agro-ecological sites with different climatic profiles. Thispaper focuses on research carried out in the CentralPlateau of Burkina Faso (Fig. 1).

The Central Plateau research site is thevillage of Bonam (2800 inhabitants),located about 100 km northeast of Oua-gadougou, the capital of Burkina Faso,and 20 km from Boulsa, the administra-tive center of the Namentenga Province.A 1997 UNDP report placed NamentengaProvince among the most economicallyand ecologically disadvantaged in thecountry (PNUD 1997). This marginalitycan be attributed to a combination of fac-tors, including scarce and irregular rain-fall, infertile and degraded soils, under-developed transportation infrastructureand a dearth of non-farm employment op-portunities.

Rain is life in Bonam, where rainfedfarming is the main source of livelihoodfor local Mossi households and scatteredhamlets of pastoralist Fulani who havesettled in the area. As in the rest of theSudan-Sahel region, rain falls during a

single season from May to October, with 90% of thetotal during July to September. Rainfall is character-ized by a high degree of spatial and temporal variabil-ity (Fig. 2). The long-term (1964 to 1998) mean annualrainfall is 674 mm, but even within a few years, it canregister values close to the long-term maximum andminimum and average. For instance, Boulsa received1050 mm of rainfall in 1994, but only 491 mm in 1997,the second lowest rainfall in 35 yr, and 617 mm in 1998,which is close to the long-term average.2

During normal years, the farming season begins withthe first rains in May, when farmers begin planting val-ley bottom fields with flood-resistant varieties ofsorghum and rainfed rice. Farmers then proceed toplant a constellation of fields in various locations withdrought-resistant varieties of sorghum and millet,intercropped with cowpea. Maize is also grown in rel-atively fertile areas of compound farms and small plotsof peanut, and Bambara nuts are planted by women inmarginal soils. The construction of a dam in Bonam in1997 by a United Nations-funded project has enabledsome farmers to begin growing dry season irrigatedcrops for both consumption and marketing.

Besides agriculture, livestock, trade and crafts pro-vide the main sources of cash income available locally.Labor migration to Côte d’Ivoire, which began underFrench colonial administration, continues to attractyoung men and plays a crucial role in local livelihoods,but more recently irrigated schemes in southern Burk-ina Faso have also been attracting migrants. Most

121

Fig. 1. Map of Burkina Faso

Fig. 2. Yearly rainfall at Boulsa, Burkina Faso, 1964–1998. Data from Direction Provinciale d’Agriculture

2Data provided by the Direction Provinciale de l’Agricultureof Namentenga

Clim Res 19: 119–132, 2001

households have relatives working as farm laborers orsharecroppers in Côte d’Ivoire, with brothers often tak-ing turns to migrate or to stay at home. Women sellsauce ingredients or cooked foods as their mainincome-generating activity. Non-farm income-gener-ating activities are mostly carried out during the dryseason, when people have more time and money andthere is greater demand for goods in local markets.

Bonam is situated on what used to be an importantroad linking the Sahelian livestock production areaswith important market town of Puytenga, whichattracts traders from neighboring countries. But theroad is now in disrepair and Bonam remains in a mar-ginal position with respect to market and transporta-tion networks. The village hosts a market that follows a3 d cycle, serving a network of smaller villages in thesurrounding areas, as well as some traders, who buysmall surpluses of grain and other farm products afterthe harvest to sell to wholesalers in Boulsa andPuytenga. Several non-governmental organizations(NGOs) have been active in the area with variousinterventions, including drilling bore-holes, healtheducation, literacy classes, micro-credit for women,distribution of transportation and plowing equipment,soil and water conservation, and food aid.

We began working in the area in January and Feb-ruary 1998, shortly after the 1997 harvest, whichoccurred between October and December and was anunmitigated disaster for most crops. Low yields meanta much greater dependence on grain imports and mar-ket purchases for an area that in good years exportsmodest surpluses. We returned in June and July 1998,when farmers were working hard at planting their newcrops, grain was scarce and costly on the market, andthe hunger season was reaching its peak. As theyentered the new farming season, local families werestretched to the limits of their resources and yet facedseveral months before harvesting the new crops. For-tunately, rains were adequate and the new harvestbrought much needed relief to the area. Our thirdstudy period was in January 1999, just as families werebeginning to recover and, in some cases, reconstitutetheir depleted assets.

The research design combined a variety of qualita-tive and quantitative methods of data collection. Togather background information and enable the com-munity to become acquainted with our presence, wefirst carried out focus groups with different socialgroups (farmers, herders, women) and open-endedinterviews with resource persons. Then we conducted3 rounds of a household survey on production andlivelihood strategies in January and June 1998 andJanuary 1999, as well as group semi-structured inter-views with sub-samples of selected households onagricultural decision-making and farm histories. We

also developed case studies of extra-sample house-holds, representing specific livelihood portfolios andtrajectories, such as the case of the wealthy farmer-trader we will refer to in the course of this paper. Thesurvey sample comprised 18 randomly selected com-pounds, 10% of all compounds in Bonam. Because acompound may include more than 1 household (de-fined as those who farm and eat together from a com-mon granary) the original sample comprised 25 house-holds, led by either the compound head or 1 of hisbrothers or sons. After the 1997 drought, 4 householdsmigrated to Côte d’Ivoire and southern Burkina Faso,leaving 21 households with a complete data set. Thissample size allowed for frequent and personalizedinteraction, enabling us to collect reliable data and tocontextualize our interpretations.

Sample households were ranked according to wealthper (adult) capita, by calculating the value of animalsowned and other liquid assets, such as plows, donkeycarts, mopeds, bicycles and radio-cassette players, atthe average selling prices obtained in 1998.3 Wealthendowments for the 2 resulting strata, both above andbelow the sample average wealth, were skewed.Higher and lower ranking households had an averagewealth per capita that is respectively twice and halfthat of the sample average (Table 1).4

122

Higher rank Lower rank(n = 8) (n = 13)

Average per household 131648 33625(CFA Fr)

% of sample average 186 47

Table 1. Average per (adult) capita wealth of 21 households ofthe Central Plateau, Burkina Faso. CFA Fr: Communauté

Financière Africaine francs

3Wealth is an imprecise measure because (1) ownership isembedded in kinship relationships in ways that may spanindividuals or households; (2) wealth is a dynamic variablewith frequent turnover and transactions, especially duringthe dry season, when animals are sold or slaughtered for reli-gious and social celebrations; and (3) possessions are onlyone aspect of the complex mix of material, social and culturalfactors that affect household prestige and position in the lo-cal social organization and household access to key re-sources

4We also ranked households with the help of 2 research assis-tants, and 3 local informants ranked households by usinglocal indicators of wealth and well-being. The most salientdifferentiating criteria referred to material possessions, espe-cially animals; labor force, especially male; and livelihoodsecurity, especially food availability. The 2 resulting cate-gories translate into something like ‘the haves’ (sen taré) and‘the have nots’ (sen pataré) and included the same set ofhouseholds as those produced on the basis of wealth percapita

Roncoli et al.: Coping with crisis in Burkina Faso

Notably, assets made up a greater proportion ofwealth owned by higher ranking households (75%)than wealth owned by lower ranking households(57%). This may have been due to their ownership ofexpensive items, such as motorcycles, but under-reporting of cattle they had entrusted to Fulani care-takers may also have been a factor. Motorcycles are astatus symbol as well as an economic advantage, asthey enable their owners to travel and transport thingsbetween markets and to have access to services andinformation in town. Six of the 8 higher ranking house-holds owned plows and donkey carts, and the two thatdid not own them had access to them within their com-pounds. Only 1 lower ranking household owned aplow and donkey cart, notably the most affluentamong them, which is also related to the chief. Owner-ship of plows and other assets partly reflects the betteraccess of higher ranking households to credit pro-grams. Credit access is often through personal andpolitical connections. These technologies enablehouseholds to expand and diversify their land holdingsand complete farming operations more rapidly, anadvantage that is vital in coping with a shortened rainyseason.5 One farmer evaluated the value of plows ashaving 2 additional men in the household.

3. READING THE RAINS

For Bonam farmers, drought is not a new phenome-non. They can recall several droughts and subsequentfamines that occurred in the past, including those ofthe mid-1970s and mid-1980s. They are usuallyreferred to by the name of places where people went tobuy grain or by events associated with them, such asthe descent of camels from the northern regions. Butthey maintain that in recent decades climate hasshifted toward lower annual rainfall levels, shorteningof the rainy season, and greater frequency of dry peri-ods during the season (Roncoli et al. 1999). The 1997rainy season exhibited all these characteristics, andseveral elders who were interviewed emphasized thatit was one of the worst years they had ever seen

Farmers’ accounts largely converge with the actualmeteorological data as they stressed the failure ofplanting rains as the main problem (Fig. 3). Despite afew scattered showers in May and early June, therewere no planting rains (sigri saaga) during June andfirst half of July.6 A few such rains fell in late July,enabling farmers to begin sowing their fields, but bythen it was getting too late to sow sorghum and millet,the main staple crops. As early August approached,which is generally recognized as the limit to viableplanting, farmers resorted to sowing after no morethan a drizzle or even at the mere sight of clouds, hop-ing for rain. But left in dry soil, much of the seed was

123

5The importance of timely planting and scarcity of plows,which are owned by only one-third of households, meansthat renting plows is uncommon, unlike elsewhere in thecountry (Hansen & Reenberg 1998). Households that do notown plows most often opt for planting by hand rather thanwaiting for plows to become available after plow-owninghouseholds have finished planting

Fig. 3. Daily rainfall at Boulsa, Burkina Faso, 1997

6Sigri saaga rains are rains that fall in the early part of therainy season and deposit at least 20 mm of rain (Roncoli et al.2000b)

Clim Res 19: 119–132, 2001

eaten by birds and insects. Because crops repeatedlyfailed to become established, farmers had to replanttheir fields as many as 8 times, sometimes replantingwith shorter duration varieties as the season pro-gressed. A farmer lamented that it was as if the seedhad been cooked; another declared that is was as if acruel joke were being played at their expense. Severalsaid that, had they known that the season was going tobe this bad, they would have not bothered to keepplanting but would have migrated to Côte d’Ivoire withtheir families.

The rains continued in very aberrant ways, fulfillingfarmers’ premonitions. August, which is normally thewettest month, was again relatively dry. But Septem-ber brought torrential rain that farmers recall damag-ing crops and flooding valley bottom fields they hadplanted as a measure against the drought. Accordingto farmers, premature end of the rains compoundedthe other problems, resulting in the failure of mostcrops, especially those planted late.7 Among staplegrain crops, sorghum suffered more than millet, whichis more resistant to water deficit, but millet was alsodamaged by pests during grain filling.

In retrospect, farmers recalled numerous hints andsigns that had announced an unfavorable season. Forinstance, rains began in the north rather than in thesouth, as is normal. Temperatures during the previousdry season were milder than usual in the early part andcooler than usual in the later part of the season. Violentwinds and persistent fog also announced perturba-tions. Another commonly relied upon forecast indica-tor, the flower and fruit production of certain localtrees, was also poor, giving reasons for apprehension.8

But most of all, the nature of rainfall at the onset of theseason, and especially whether it conforms to ordiverges from an idealized pattern, and the number ofplantings required to establish crops are consideredthe most trustworthy predictors of crop performance(Roncoli et al. 2000). By the end of planting time it wasobvious to farmers that the harvest would be very pooreven in the best of circumstances. In other words,farmers knew that a crisis loomed several monthsbefore the harvest and half a year before they beganadopting survival strategies that are usually consid-

ered famine signals, such as eating wild fruits andskipping meals.

No advance warning had been available before therainy season, since in 1997 the Meteorological Servicestill deemed scientific forecasts to be too experimental.The irregularity of rainfall was noted by the FamineEarly Warning System (FEWS) and a shortfall in pro-duction deficit was predicted.9 But the severity of thesituation was recognized only gradually. By October,FEWS observers noted poor crop conditions, especiallyin Namentenga and the rest of the Central Plateau. InNovember, one-third of the grain crop was estimatedto be lost in the central, northwestern and northernprovinces. Eastern provinces, which are not droughtprone, were affected as well. In early December thePresident appealed to the international community forfood aid. It was not until March, when the governmentupdated production figures, that the situation wasassessed to be far more severe than expected. It wasestimated that crop losses would reach 75% in someareas of the Samantenga, Séno and Bam provinces, butNamentenga was also mentioned among those suffer-ing from food and water shortages.

From March onward, FEWS bulletins describe the sit-uation as very serious, indicating that households wereresorting to consuming famine foods and to migratingelsewhere. They also noted that price increases wereplacing grain outside of the purchasing power of popu-lations of affected areas, especially lower incomehouseholds. Prices had been increasing early in theyear, but initially did so gradually. This was becausemore of the grain produced in the southwest was mar-keted domestically, and the government and NGOsmade grain available at subsidized prices through ce-real banks. Limited supplies of seed were also distrib-uted. But by April and May, when most households hadexhausted their stocks, prices were 50% above Mayprices in 1996 and 1997. They would have increasedmuch further if government and NGOs had not againintervened, releasing grain at 140 Communauté Finan-cière Africaine franc (CFA Fr) kg–1.10 Some food aidalso reached Bonam through NGOs, the Catholic mis-sion and government programs, but it was far from suf-ficient to fill the need. By August 1998, Bonam house-holds were paying up to 240 CFA Fr kg–1.11

124

7Farmers’ accounts diverge from meteorological data ob-tained in Boulsa that shows rains in early November. Theymight have been referring to the lack of significant rain inthe second half of October, a time when late planted cropsneeded rain to mature, as having the same effect as the endof the rains. But it could also be due to the spatial variabilityof late rains

8The most common are the karité (French) or taanga (Moré)and the raisinier (French) or sibga (Moré). The respective sci-entific names are Butyrospermum parkii and Anogeissusleiocarpus

99FEWS is a USAID funded program. Monthly bulletins can befound at http://www.fews.org/fewspub.html

10The CFA franc was valued at 692 CFA Fr = 1 US$ in 199811The highest price recorded in the FEWS bulletin was 171

CFA Fr kg–1. Price monitoring is complicated by the fact thatsellers in local markets shift units of measure according tosupply. At times of plenty, they use a bowl that will hold2.75 kg of sorghum or millet; when grain is scarce they use asmaller one, containing 2.5 kg. But prices are generally citedand recorded ‘by the bowl’ (assiette)

Roncoli et al.: Coping with crisis in Burkina Faso

4. STRATEGIES FOR SURVIVAL

Soon after the harvest, having fully realized theseverity of the situation, families began discussing,negotiating and adopting coping strategies. In somecases the household head gathered all adults in thehousehold to exhort them to be thrifty in using theavailable grain and to do their best to obtain additionalmoney or food. Those who had relatives employed intowns or farming in Côte d’Ivoire sent word that helpwould be needed. Production of staple grains(sorghum, millet and maize) had been only a fraction ofthat in normal years (Table 2). With annual grainrequirements estimated at 200 kg per adult and 67 kgper child or elderly person (Carter 1997), averagehousehold production was barely one-quarter of theper capita demand.12 Notably, higher ranking house-holds did not produce more per capita than lower rank-ing ones. This result partly reflects the larger size anddependency ratio of higher ranking households be-cause they included greater numbers of wives andchildren. It might also be due to their greater relianceon cash crops and non-farm income.13 But cash crops,such as peanut, cowpea, rice and sesame, also failed. Awealthy farmer complained of having lost investmentsof 20 000 CFA Fr for cowpea seed and 30 000 CFA Frfor peanut seed because of the drought.

Higher ranking households, however, compensatedfor the production shortfall by buying more grainthroughout the year. Even though they also had tomanage their resources carefully and reduce house-hold size by sending sons to Côte d’Ivoire, betweenwhat they produced and what they bought, they wereable to meet consumption requirements (Table 3). Onthe other hand, in lower ranking households produc-tion and purchases satisfied only 82% of their con-sumption requirements. Household food supply wasprobably supplemented by small purchases bywomen, transfers from relatives and food aid, whichare difficult to document.

Regardless of wealth status, most households tried tominimize grain purchases in July and August, whenprices are the highest. This is not only because of sea-sonal increase but also because at this time money isvery scarce and households tend to buy by the bowlrather than by the bag, which affords a better value.Many saved their own grain stocks to be consumedduring the farming season, especially during periods ofhigh labor demand. Higher ranking bought one-thirdand lower ranking households one-fourth of all grainpurchased during the year shortly after the harvest andthe remainder shortly before the onset of the nextfarming season. There was no substantial difference in

125

Households Higher rank Lower rank8 13

Jan 1998Consumers hh–1 8.8 6.1Expenditures (CFA Fr hh–1) 57237.50 21173.08Grain purchased

kg hh–1 390.00 166.15kg consumer–1 44 27

Jul 1998Consumers hh–1 7.1 5.8Expenditures (CFA Fr hh–1) 82950.00 51369.23Grain purchased

kg hh–1 514.38 305.77kg consumer–1 72 53

Jan 1999Consumers/hh 7.9 6.1Expenditures (CFA Fr hh–1) 53268.75 24111.54Grain purchased

kg hh–1 316.50 116.15kg consumer–1 40 19

Total yearExpenditures (CFA Fr hh–1) 193456 98961.54Grain purchased

kg hh–1 1221 658.08kg consumer–1 157 110CFA Fr kg–1 158 150

Table 3. Grain purchases by 21 households (hh) in the CentralPlateau, Burkina Faso (1998 and 1999). CFA Fr: Communauté

Financière Africaine francs

Higher rank Lower rank

1997Number of producers 6.5 4.8Number of consumers 8.8 6.1Grain productionkg hh–1 386.60 342.80kg producer–1 59.48 71.42kg consumer–1 43.93 56.20

1998Number of producers 4.9 4.5Number of consumers 7.1 5.8Grain productionkg hh–1 2015.40 1329.40kg producer–1 411.31 295.42kg consumer–1 283.86 229.21

Table 2. Grain production by 21 households (hh) in the Central Plateau, Burkina Faso (1997 and 1998)

12Adult men and women count as 1 ‘consumer’, children andelderly count as one-third

13The proportions of farmers from higher ranking and lowerranking households who planted these crops were similarbut wealthier farmers tend to plant larger areas for marketpurposes, while resource-limited farmers plant small plotsfor home consumption. Because farmers do not estimate cul-tivated areas and often reported total loss of yields, we can-not verify this hypothesis

Clim Res 19: 119–132, 2001

grain prices paid by higher and lower ranking house-holds. But grain purchases absorbed a much greatershare (63%) of the total budget of lower rankinghouseholds, limiting their ability to meet other needs orto invest in inputs or assets (Fig. 4). Higher rankinghouseholds spent almost half of their total budget ongrain but were able to buy livestock and assets as well(Fig. 5).

Soon after the harvest, household heads began tocontrol grain allocation and utilization to ensure thatthe meager supplies they produced or purchasedlasted as long as possible. Food management strate-gies entailed a combination of control, conflict, compli-ance and cooperation among men and women, youngand old within the household (McMillan 1986, Kevane2000, Thorsen 2000). These strategies were as follows:(1) reducing the number of consumers by sendingyoung men into migration and children to stay with rel-

atives; (2) reducing the number of women cooking andcharging 1 woman to cook a common meal rather thanallowing each woman to cook separately; (3) supervis-ing women more strictly in how they handle grain forcooking; (4) using a smaller container to measure grainor extending the time between grain allocations towomen; (5) reducing the amount of grain used inpreparing daily meals, i.e., cooking just enough withno provision for leftovers; (6) reducing the number ofmeals served, by eliminating the morning or middaymeal; (7) relying on women’s contributions of grainfrom their own fields or bought with their own money;and (8) consuming bean, cowpea or peanut whenavailable, or famine foods, especially at the peak ofhunger during the last 2 wk before the new maize har-vest.14 There was considerable variation in the adop-tion and combination of strategies, but some regularityin sequencing. For instance, households lessened thenumbers of consumers or of cooks before reducing thedaily ration, and cooked less before eliminating mealsor eating famine foods.

In addition to the 4 households in our original samplewho went to Côte d’Ivoire, 18% of men, 8% of womenand 15% children in the remaining households leftBonam between January and June 1998. It was onlyafter the 1998 harvest that households began reconsti-tuting their numbers (Table 3). Migration as a way toreduce need was particularly important for wealthierhouseholds because they were also larger. Poorerhouseholds may be less able to spare members fromperforming tasks and duties that need to be taken careof in the dry season. High travel costs and increasingethnic tension and harassment against Burkinabémigrants in Côte d’Ivoire can also be a deterrent. Yetgifts by migrant relatives contributed significantly tograin purchases by lower ranking households, ac-counting for 27% of what they spent (Fig. 6). Higherranking households also derived income from migra-tion but not all was spent on grain (Fig. 7). For instance,a wealthy farmer received 100 000 CFA Fr, which heinvested in cattle. Another received 650 000 CFA Fr,which he used to buy a moped.

Earnings of migrant workers became available onlyafter December and January, when crops are har-vested and marketed in Côte d’Ivoire. Most migrantsreturn at the end of the dry season, when grain pricesare higher. To take advantage of the lower grain prices

126

Fig. 4. Expenditures by lower ranking households

Fig. 5. Expenditures by higher ranking households

14Most famine foods are prepared by cooking the youngleaves of bushes or trees that become available in May andJune. Among the most common are pimpirga, lelongo,kalyanga, hadga, kalquempoodo, katempongo, konkuega,keglega, pelga kilogo and kesega. Wild leaves that growaround the compound and animal pens (sugda, kienba,zinba and vampoonse) are also used

Roncoli et al.: Coping with crisis in Burkina Faso

shortly after the harvest in Burkina Faso and the higherdemand for meat to be consumed at harvest festivitiesand Christian holidays, most households began sellinglivestock early in the year. By January 1998, 56% ofhouseholds had sold goats, 36% had sold sheep and32% had sold cattle. By June 1998, 39% of small rumi-nants and 38% of cattle had been sold.15 While smalllivestock is routinely sold when cash is needed, cattlesales confer a much higher degree of stress becauselow prices due to distress sales and poor animal healthmay mean that households are unable to reconstitutetheir stock after the crisis (Fafchamps et al. 1998). Ofthe households that owned cattle in January 1998,most had sold 1 or more head by June. Some of thegoats and sheep that were sold belonged to women; insome cases they were volunteered by the owner tohelp the family, and in others they were seized andsold by the household head. But few men or womenprovided explicit information about them becauserelying on women’s property to feed the household isconsidered to reflect poorly on the household head.

Both higher and lower ranking households sold live-stock, but the wealthy ones had more animals to selland earned a lot more. In their case, livestock salesaccounted for twice as large a proportion of the foodbudget as in poor households (Figs 6 & 7). Rich farmerswith trading income both bought and sold animals aspart of their herd management, taking advantage oflow prices to buy younger animals to raise. During theyear, for instance, a farmer-trader sold 12 bullocks and9 cows and bought 7 younger bullocks. As the dry sea-son progressed, animal prices began falling to a recordlow because of distress sales and poor health statuscaused by lack of grass and water. One farmer com-plained that ‘goats were selling at the price of chick-ens.’ Low livestock prices created a windfall for tradersfrom urban markets and neighboring countries whooperate in the Puytenga market. They were able to buyemaciated animals, fatten them and then resell them atgreat profit.

Trading and other non-farm income sources playedkey roles in helping households meet their needs,regardless of wealth status. Consistent with data fromother research conducted in Burkina Faso (Delgado1989), about one-quarter of all money spent on foodwas derived from such activities. Trading and skilledwork, such as tailoring, were among the most remu-nerative as well as prestigious. Men from the black-smith caste do iron work, which can also be quite prof-itable (1 higher ranking household was among them),and their wives make and sell pottery. A wealthy

household sent sons to do gold panning in a miningarea and another received income from a son who wasemployed in the construction of the rice perimeter. Afew lower ranking ones derived revenues from butch-ery and the sale of firewood, and a few others practiceddivination and traditional medicine, among whom 1man earned about 20 000 CFA Fr by capitalizing on thearea’s reputation in Ouagadougou for witchcraft.

However, food shortage made most local trading andnon-farm activities less profitable because people hadless money to invest in them and to buy their products.The high price of agricultural commodities and scarcityof water greatly reduced women’s earnings from tradi-tional dry season activities such as food processing andbeer brewing. Even village grain traders were not ben-efiting from high grain prices. Because they gain fromsmall price increments on a high turnover of grainbought and sold, in a situation where nobody had grainto sell or money to buy it, they were losing businessand profits.

127

Fig. 6. Cash sources for grain purchases, lower ranking households

Fig. 7. Cash sources for grain purchases, higher ranking households

1556% of the farmers had sold goats, 36% sold sheep and 32%sold cattle

Clim Res 19: 119–132, 2001

Kin also helped households, especially lower rank-ing ones, to buy grain (Figs 6 & 7). This was particu-larly the case for 1 household related to the chief and afew others who had relatives employed in town. Somewomen went to visit their paternal kin or marrieddaughters in less affected areas and came back withfood. The key role of kinship as a moral framework thatvalidates appeals for assistance at times of crisis isreflected in household livelihood decisions. Despitefood insecurity, both higher and lower ranking house-holds did not fail to invest in building and consolidat-ing kinship networks and social relationships throughmarriage transactions, baptism receptions and funeralceremonies (Figs 4 & 5). Three of the higher rankinghouseholds and 1 among lower ranking ones were ableto take new wives despite the shortage of food andmoney. Marriage not only increases household laborbut also expands affinity ties on which to draw in timeof need. Funeral celebrations, which occur during thedry season and require hosting and feeding hundredsof guests for 3 d are especially salient institutionalvenues in which lineage identity is affirmed and with itthe social claims to land, labor and resources it medi-ates (Berry 1989).

When all other resources and strategies were ex-hausted, a few poor households also resorted to bor-rowing, pawning or mortgaging crops. This meant thatthey had to repay loans by selling crops shortly afterharvest, when prices are lowest. Therefore, they wereless able to take advantage of the good yields obtainedin 1998 to recover from distress sales during the year.Women borrowed from their own relatives or fromother petty traders or beer brewers, or secured loanson behalf of their households. It is believed that womenare more successful at borrowing because lenders aremoved to pity for a mother and her children andbecause the women are considered more likely torepay the loan, being thriftier, more easily intimidatedor less able to run away in case of default. The percep-tion of women as more financially responsible ex-tended to their husbands. One man entrusted his wifewith money he intended to save to buy food during thefarming season. Another deposited with his wife half ofwhat he earned as a sort of savings account to be ableto accumulate enough for emergencies or investments.

Women also sold cloth, utensils and jewelry they hadaccumulated for their daughters’ weddings to tradersfrom Puytenga market who toured the countryside insearch of profitable deals, knowing that people weredesperate for money. Many stories circulated in Bonamand surrounding areas about rural women providingsexual services to Boulsa or Puytenga traders in ex-change for food, but obviously, they could not be di-rectly verified. Rather than condemnation, they gener-ally evoked words of commiseration and commentaries

on mothers’ self-denying dedication to their children’swell-being. Informal interviews with women suggestthat women’s key roles in ensuring household liveli-hood has increased bargaining power vis-à-vis theirhusbands. Women are more often consulted in case ofcrop or livestock sales or major household purchases.Wives whose relatives help at times of need may beable to visit them more often or stay longer. Pettytraders whose earnings help buy food for the house-hold may be able to take more time off farm work to gothe market. Husbands also increasingly involve wivesin farming decisions, such as whether to make a newfield and what proportion of each crop to plant. Al-though marriage transactions are still negotiated bymen, women may derail a marriage they disapprove ofby protesting that they had to sell the items they are re-sponsible to provide to their marrying daughters.These shifts in gender relationships seems to be morecommon for older women who have produced childrento adulthood than for younger wives. They also tend tooccur more often in poorer monogamous householdsthan among wealthier polygynous ones. This may bedue to their being less dependent on women’s contri-butions as well as to the greater influence of Islam, es-pecially among households involved in trade.

5. AGRICULTURAL ADJUSTMENTS

The fact that most households entered the farmingseason with a depleted financial base and labor forcehad serious implications for their ability to managerisks and resources effectively. As usual, farmersstrove to minimize cash investments in agriculture, butin some cases they were unable to do so because manyhad consumed all their seed before planting. The per-centage of farmers who purchased seed almost dou-bled from 43% in 1997 to 76% (62.5% of higher rank-ing and 85% of lower ranking households) in 1998. Inthe case of staple grains, this was a deviation from thecommon practice of carefully selecting and storingseed from season to season. Most of those who boughtseed obtained it from fellow farmers, but some resortedto planting grain they had bought in the market forconsumption. Planting seed purchased in the market isrisky because one does not know from where it comes,for how long and in what conditions it has been stored,or even the variety. A bag of sorghum may contain dif-ferent varieties that traders have bought on variousmarkets. Several Bonam farmers unknowingly planteda sorghum variety that had come from the much wettersouthwestern part of the country and that was mal-adapted to the shorter rainy season of the CentralPlateau. Consequently their yields would have beenpoor even if the 1998 rains had been good.

128

Roncoli et al.: Coping with crisis in Burkina Faso

The 1997 drought stimulated interest in short dura-tion improved sorghum varieties, although nonewere available locally. But short and medium dura-tion varieties have been imported from the driernorthern provinces through spontaneous farmer-to-farmer exchange. Women are instrumental in the dif-fusion of new varieties that they may come acrosswhile helping other women cook, attending marketsor funerals, or visiting relatives in other areas.Medium duration (70 to 90 d) varieties, such as pis-wopoi, kurbulli, fibmiugu and yadega, have largelyreplaced longer duration (120 to 150 d) varieties thatwere grown in the past, such as belko, gambré andbalinga. But this shift entails costs and risks becauselonger duration varieties, when planted early andable to develop fully, are more productive, less vul-nerable to weeds, pests and water stresses, moredurable when stored, and considered more palatableand nutritious. A 50 d variety known as pisnu (whichmeans 50 in Moré) is also planted in small manuredplots together with or shortly after maize, but itrequires higher soil fertility and more careful weed-ing than other varieties. These requirements limitwidespread adoption, but most farmers also dislikeits poor taste and cooking properties, and their wivesresent the fact that it makes such a poor dish that itembarrasses even the best cooks.16

The desire to shorten the hunger season led farmersto increase maize planting because maize matures sev-eral weeks before staple grains. Households growingmaize doubled from 33% in 1997 to 76% of (higherranking and lower ranking) households in 1998, andsome of those who did farm maize in 1997 expandedtheir fields. Hunger even pushed many households torisk planting maize 2 or 3 wk earlier than they nor-mally would. Because of maize’s vulnerability to waterdeficit, farmers usually wait to plant it after a 10 to 12 ddry period that usually occurs between mid-June andearly July, coinciding with the visible phase of the 8thlunar cycle of the year. But unexpectedly in 1998 thisdry period stretched into a full month, with no rainfallbetween 8 June and 12 July, so that all the maize thathad been planted before wilted and died. Once itbegan raining again in mid-July, it did so heavily, trig-

gering such a proliferation of weeds that farmers couldbarely keep up removing them and had no time toreplant maize.

The inability to either save or buy seed also meantthat few farmers were able to grow peanut, a commoncash crop and cooking ingredient as well as an impor-tant stage in crop rotation to restore soil fertility. Only4 (19%) households, 2 higher ranking and 2 lowerranking ones, planted peanut in 1998, compared with43% the year before. Some were unable to plantpeanut because they lacked seed or cash to buy it,while others were constrained by lack of labor. Thosefew who had means to invest in peanut farmingbenefited from the price increase that resulted fromshortage in local supply. A farmer-trader invested60 000 CFA Fr in seed and 43 500 CFA Fr in labor, andborrowed a tractor from a relative to prepare his landat no cost. The favorable rains of 1998 resulted in goodyields, and he harvested a total of 6400 kg for a netprofit of 643 200 CFA Fr at harvest prices. In practicehis net profit may have been greater because he hadnot sold all the grain immediately after harvest and theprice of peanut was likely to increase during the year.

Among farmers interviewed, some expanded andothers reduced their land holdings in 1998. Expansionwas mostly in lowland fields, and sometimes more ele-vated fields that had performed poorly in 1997 wereabandoned because of the drought. Increasing ex-ploitation of valley bottom fields has been one of themain ways in which farmers have responded todrought in recent years. Competition for these landshas increased to such an extent that farmers declaredthat they would continue farming valley bottom fieldseven if abundant rainfall was expected because uncul-tivated land may be occupied by distant relatives whomay be impossible to dislodge. The risk of flood dam-age to crops during an occasional wet season wasdeemed preferable to the loss of an invaluable re-source that insures them against the much more fre-quent occurrence of drought. Land tenure is stilldefined by kin group affiliation, and some lower rank-ing households controlled the best valley land, havingsettled in an area on the margin of the village that inthe past was less desirable because of dense vegeta-tion and wild animals. But the ability to put land intocultivation is predicated upon labor, plows and otherresources, so some households were unable to farm allof the land at their disposal. Prosperous, prominentfarmers, on the other hand, could use their influence inthe community and their ability to assist with loans orplow services to borrow land from others. The farmer-trader’s peanut field mentioned above was developedon borrowed land.

Loss of labor to migration was one of the main con-straints stated by farmers as a reason for having re-

129

16The advantage offered by pisnu’s shorter duration stimu-lated farmer experimentation. In a nearby village, farmersselected a 50 d variety with better taste and texture that wasobtained by the extended household of Islamic clerics whoinhabit the village from their former Koranic teacher in thenorthern town of Kaya. Availability of labor from over 100Koranic students residing and studying with them enabledthem to plant the more labor-intensive pisnu on most of theirlands. Consequently their village was spared from thehunger that plagued the surrounding area following the1997 drought

Clim Res 19: 119–132, 2001

duced their cultivated area in 1998, in addition to lossof land to the reservoir and soil hardening. Nutritionalstress added to migration in depleting household laborforces. In some cases, people were too weak fromhunger to be able to work for more than a few hourseach day. Lack of food and money also made it difficultto mobilize extra-household labor by way of labor par-ties. While 33% of households had engaged labor par-ties during the 1997 season, in 1998 only 3 (14%) wereable to do so. This meant that family members, includ-ing women, had to work harder in the household fields,and had less time and energy to develop personalfields. Women reported that the increasing importanceof women’s work in the production of household grainis resulting in their greater involvement in decisionsabout surplus sales.

Labor constraints hindered the implementation ofsoil and water conservation practices such as strawmulching, stone barriers, grass hedges and zai17.While farmers recognize these measures to be effec-tive in mitigating climatic risk and in improving yields,they are also very labor intensive, requiring up to 200 hha–1 (Sanders et al. 1990). Their feasibility was alsolimited by soil hardening and by shortage of strawcaused by the previous year’s drought, and by the cre-ation of the reservoir and irrigated perimeter, whichtook over an area where animals used to graze andpeople cut grass for mulch. But 2 higher rankinghouseholds were able to organize communal labor par-ties to transport and spread straw in his fields, and thefarmer-trader mobilized people to dig zai holes andbuild stone barriers in exchange for grain.

Production in 1998 was generally satisfactory andmet household consumption requirements (Table 2).Significantly, unlike in the previous year, per capitaproduction by higher ranking households was greaterthan for lower ranking ones. This difference may havebeen due to the risks and trade-offs lower rankinghouseholds were forced to engage in as a result of foodinsecurity, as well as a reflection of greater priority onfood production prompted by the experience of a yearof hunger and high grain prices. Because of their largersize and greater social obligations, higher rankinghouseholds were eager to ensure their food self-suffi-ciency. This emphasis may have been reflected incloser attention and greater intensity of labor deployedon staple grain fields rather than on expansion of culti-vated area or cash investments, which remained rela-tively low in all cases.18

Among all sample households, only 1 ended the yearfollowing the 1997 harvest failure having gained ratherthan lost wealth. A farmer-trader included as a sepa-rate case study also had a positive balance. Bothhouseholds in question benefited from non-farmincome from skilled work and trade as well as from

transfers from migrant sons and urban-based relatives.Despite having to buy large quantities of high pricegrain, they were able to invest in cattle, buy a mopedand marry a new wife each. One sent children toattend high school in the capital city, and the othermade a 40 d pilgrimage to Mecca that enhanced hisprestige in the local Muslim community. Both statedtheir intention to invest surplus income from the good1998 harvest into trading of non agricultural items as alow-risk intermediate venture between livelihoodneeds and livestock investments.

6. CONCLUSION

Counteracting the view of African rural communitiesas inherently vulnerable, the ability of livelihood sys-tems to cope with or recover from climate shocks hasbeen cited as evidence of the rationality and resilienceof rural livelihoods (Torry 1984, Mortimer 1989). But anappreciation of the resourcefulness of African farmersshould not overlook the possibility that coping mayentail painful trade-offs, additional risks and unequallydistributed costs, as the case presented here illustrates.Furthermore, it should not be taken as a reason forneglecting crisis prevention or preparedness or for dis-engaging from development commitments (Gray &Kevane 1993), but as a resource to draw upon.

Farmers’ observations of climatic and agronomicinteractions during planting time can help predict har-vest failure and food insecurity long before their occur-rence, as they did in 1997, and with sufficient advancenotice to plan needed interventions. They should beincorporated into seasonal rainfall forecasting andfamine early systems, but more micro-level researchand closer integration of meteorological, agriculturaland social sciences are called for to understand betterhow information can be disseminated and applied. Thediversity of users’ needs and the disparity of the meansat their disposal must be taken into account to deviserelevant and realistic applications (Ingram et al. 2002).

For farmers to be able to respond optimally to aforecasted climate shock as well as for families to with-stand its consequences, they need more than informa-tion. It is necessary to integrate science and de-

130

17Zai (meaning ‘scratching’) is a fertility restoration and waterconservation technique that has been adopted from thedrier northern provinces and promoted by extension andNGOs since the early 1990s. Seed is planted in holes (20 to40 cm apart) filled with organic matter, with the dug earthamassed downhill in the form of a half-moon barrier to re-tain rainwater

18Higher ranking households invested an average of 9375CFA Fr compared with 7810 CFA Fr spent by lower rankinghouseholds

Roncoli et al.: Coping with crisis in Burkina Faso

velopment interventions in ways that help improve thelivelihood options and the productive capacity of farm-ing households, especially those with limited re-sources. An appreciation of the centrality of thewomen’s work and income for household food securityshould also guide these efforts beyond a household-focused approach toward one that considers differen-tial vulnerabilities and entitlements.

The research in Bonam suggests that access to labor-saving technologies that accelerate land preparationand planting, such as plows, and timely availability oflocally adapted seed varieties of food and cash cropsare key elements of agricultural adaptation to climatevariability. At the same time, access to credit that is nottied to specific crops or inputs and improved marketaccess can increase the flexibility of farmers’ responsesand facilitate livelihood diversification. But, becausehuman labor remains such a key element in local pro-duction systems, donors, government and the privatesector should also be working together to ensure thataffordable grain is available to rural families at times offood insecurity, such as that which followed the 1997drought.

Acknowledgements. Support for this research was providedby the Office of Global Program of the US National Oceanicand Atmospheric Administration. We acknowledge theinvaluable collaboration of our institutional partners in Burk-ina Faso, Plan International, especially Country Representa-tive Felipé Sanchez and Agricultural Program Manager EricMamboué, and the National Meteorological Service, espe-cially Director Dr. Frédéric Ouattara. We also thank Ian Flit-croft, Christine Jost and Gerrit Hogenboom for help at differ-ent stages of the research.

LITERATURE CITED

Adams AM, Cekan J, Sauerborn R (1998) Towards a concep-tual framework of household coping: reflections from ruralWest Africa. Africa 68:263–283

Benson C (1998) Drought and the Zimbabwe economy,1980–93. In: O’Neill H, Toye J (eds) A world withoutfamine? New approaches to aid and development.Macmillan, Houndsmills, p 241–273

Berry S (1989) Social institutions and access to resources.Africa 59:41–55

Blaikie P, Cannon T, Davis I, Wisner B (1994) At risk: naturalhazards, people’s vulnerability, and disasters. Routledge,London

Bratton M (1987) Drought, food and the social organization ofsmall farmers in Zimbabwe. In: Glantz M (ed) Droughtand hunger in Africa: denying famine a future. CambridgeUniversity Press, Cambridge, p 214–244

Brock K (1999) Implementing a sustainable livelihoods frame-work for policy-directed research: reflections from prac-tice in Mali. Institute for Development Studies, WorkingPaper no. 90. University of Sussex, Brighton

Campbell DJ (1990) Strategies for coping with severe fooddeficits in rural Africa: a review of the literature. FoodFoodways 4:143–162

Carter M (1997) Environment, technology, and the socialarticulation of risk in West African agriculture. Econ DevCult Change 45:557–590

Corbett J (1988) Famine and household coping strategies.World Dev 16(9):1099–1112

Cutler P (1984) Famine forecasting: prices and peasantbehavior in Northern Ethiopia. Disasters 10:181–188

Davis S (1993) Are coping strategies a cop-out? IDS Bull 24:60–72

Davis S (1996) Adaptable livelihoods: coping with food inse-curity in the Malian Sahel. Macmillan, London

Delgado CL (1989) The changing economic context of mixedfarming in savanna West Africa: a conceptual frameworkapplied to Burkina Faso. Q J Int Agric 28:3–4

Dercon S (1996) Risk, crop choice, and savings: evidence fromTanzania. Econ Dev Cult Change 44:485–513

Devereux S (1993) Goats before ploughs: dilemmas of house-hold response sequencing during food shortages. IDS Bull24:52

Downing T (1990) Monitoring and responding to famine:lessons from the 1984–85 food crisis in Kenya. Disasters14:204–229

Ellis F (1998) Household strategies and rural livelihood diver-sification. J Dev Stud 35:1–38

Fafchamps M, Udry C, Czukas K (1998) Drought and savingin West Africa: are livestock a buffer stock? J Dev Econ 55:273–305

Gray L, Kevane M (1993) For whom is the rural economyresilient? Initial effects of drought in Western Sudan. DevChange 24:159–176

Hansen TS, Reenberg A (1998) Approaching local limits tofield expansion—land use pattern dynamics in semi-aridBurkina Faso. Dan J Geogr 98:56–70

Hussein K, Nelson J (1998) Sustainable livelihoods and liveli-hoods diversification. Institute for Development StudiesWorking Paper no. 69. University of Sussex, Brighton

Ingram K, Roncoli C, Kirshen P (2002) Opportunities and con-straints for farmers of West Africa to use seasonal precipi-tation forecasts with Burkina Faso as a case study. AgricSyst (in press)

Jiggins J (1986) Women and seasonality: coping with crisisand calamity. IDS Bull 17:9–18

Kevane M (2000) Extrahousehold norms and intrahouseholdbargaining: gender in Sudan and Burkina Faso. In: SpringA (ed) Women farmers and commercial ventures. LynneRienner, Boulder, p 89–112

Lambert RJ (1994) Monitoring local food security and copingstrategies: lessons from information collection and analy-sis in Mopti, Mali. Disasters 18(4):332–343

McMillan D (1986 ) Distribution of resources and products inMossi households. In: McMillan D, Hansen A (eds) Food inSub-Saharan Africa. Lynne Rienner, Boulder, p 260–273

Mortimore M (1989) Adapting to drought: farmers, famines,and desertification in West Africa. Cambridge UniversityPress, New York

Painter T, Sumberg J, Price T (1994) Your terroir and my‘action space’: implications of differentiation, mobility, anddiversification for the approche terroir in Sahelian WestAfrica. Africa 64:447–464

PNUD (1997) Rapport sur le dévéloppement humain durable:Burkina Faso. Programme des Nations Unies pour leDévéloppement, Ouagadougou

Reardon T, Matlon P (1989) Seasonal food insecurity and vul-nerability in drought-affected regions of Burkina Faso. In:Sahn D (ed) Seasonal variability in Third World agricul-ture: the consequences for food security. Johns HopkinsUniversity Press, Baltimore, p 118–136

131

Clim Res 19: 119–132, 2001

Reardon T, Taylor E (1996) Agroclimatic shock, incomeinequality, and poverty: evidence from Burkina Faso.World Dev 24:901–914

Reardon T, Matlon P, Delgado C (1988) Coping with house-hold-level food insecurity in drought-affected areas ofBurkina Faso. World Dev 16:1065–1074

Roncoli C, Bahadio B, Boena S (1999) The role of rainfall infor-mation in farmers’ decisions: ethnographic research in theCentral Plateau (Burkina Faso). Technical Report, ClimateForecasting and Agricultural Resources Project. Univer-sity of Georgia, Athens

Roncoli C, Ingram K, Kirshen P (2001) Reading the rains: localknowledge and rainfall forecasting among farmers ofBurkina Faso. Soc Nat Resour 15 (in press)

Sanders J, Nagy J, Ramaswamy S (1990) Developing new

agricultural technologies for the Sahelian countries: theBurkina Faso case. Econ Dev Cult Change 39:1–22

Scoones I (1998) Sustainable rural livelihoods: a frameworkfor analysis. Institute for Development Studies WorkingPaper no. 72. University of Sussex, Brighton

Thorsen D (2000) Dividing or sharing responsibilities? Intra-household budgeting among Bissa and Mossi of BurkinaFaso. Paper presented at the African Studies AssociationAnnual Meetings, Nashville, 15–19 November 2000

Torry W (1984) Social science research on famine: a criticalevaluation. Hum Ecol 12:227–252

Whitehead A (1981) ‘I’m hungry, mum’: the politics of domesticbudgeting. In: Young K, Wolkowitz C, McCullagh R (eds) Ofmarriage and the market: women’s subordination in inter-national perspective. CSE Books, London, p 88–111

132

Submitted: February 14, 2000; Accepted: February 13, 2001 Proofs received from author(s): June 22, 2001