7/31/2019 Booth Etal 1999

1/49

Experiencing poverty in Africa:perspectives from anth ropology

Background Paper No. 1(b) for theWorld Bank Poverty Status Report 1999

David Booth, Melissa Leach and Alison Tierney

Final Draft, Apr 1999

7/31/2019 Booth Etal 1999

2/49

2

Summary

The value of a multi-disciplinary approach to the understanding of poverty and thedesign of poverty-redu ction strategies is now widely accepted . How ever, this paperargues, current expectations abou t the potential contribution to poverty analysis fromdisciplines other than economics remain rather too slanted towards what arepresumed to be the special strengths of PRA-based PPAs: capturing poor peoplesperceptions, identifying their priorities and describing their coping strategies.Properly understood as centring on the observation and interpretation of behaviour,anthropological enquiry has relevant things to say at all the three levels that concerna poverty statu s report: 1) wh o are the poor? 2) why are they poor? and 3) what canbe done to reduce poverty?

Key find ings und er these headings are:

While anthropological work can help to enrich statistical poverty profiles, a moreimportant contribution may be in documenting the variable, fluid, complex andcontested categorisations and relationships that constitute the reality that poverty-redu ction efforts mu st contend w ith on the ground .

Documented responses to structural change are sufficiently diverse and affectedby the particularities of local structures, including notably gender relations, thatmu ltiple paths of impoverishment or d is-impoverishment remain m ore likely thanhomogeneou s national or regional trends.

Anthropological studies help to remind us that the primary stakeholders in anti-poverty operations are, of necessity, active participants in constructing their ownfuture, while the activities of states and development agencies are not alwaysemp owering of poor peop le.

The main imp lications for the policy and p ractice of poverty red uction are: The complexities which remain uncaptured by statistical and sociological

categorisations of the poor are a source of un certainty. Since it is known that th ey

are numerous and important, planning for poverty reduction will generallybenefit from a strong learning-process orientation.

Diversity of social response does not mean that no generalisations are possible, orthat those that remain robust (e.g. the gains from d econtrolling rur al markets, orlegislating on womens rights) are un importan t. But it does strengthen the viewthat an ti-poverty action needs to be bu ilt at least partly from the bottom up .

Development interventions, including anti-poverty strategies, are likely to benefitfrom an app roach that is mor e institutionally self-aware. This imp lies placing

poor peoples own efforts at the centre, and reflecting more self-critically onpossible side-effects of the exercise of govern men tal and agency power.

7/31/2019 Booth Etal 1999

3/49

3

1 Introduction: on asking the right questionsThe scope of current interest in the status of poverty in Africa is indicated by thethree questions: who are the poor? why are they poor? and what can be done

about it? At each of these levels of enqu iry the profile of pov erty, its causes andthe implications for policy and practice there is growing recognition of thevalue of a mu ltidisciplinary ap proach. This is usu ally expressed in a rath erbipolar way, in terms of the limitations of single-stranded work based on theanalysis of household survey data, and the need to integrate this with morequalitative eviden ce reflecting poor peop les own experience. In this simp leform, the argument forms the now widely-recognised case for including anelement of participatory assessment in any policy-oriented exercise in povertyanalysis.First, there is much to be gained from comparing conventional consumption-poverty measures not only with a wider range of indicators of deprivation, butalso with poor peoples own perceptions of who the poor are, and hence whatpoverty is, or what matters about it. Second, analysis of causes is more robust ifit is informed by the poors own vision of the dynamics of poverty and theprocesses leading to imp overishment and d isimp overishment. Third, the designof policy interventions needs to take into accoun t and build u pon poor p eoplesown coping stra tegies.At all three levels, the argument runs, it is possible to use the results of

participatory assessments and other grass-roots case-study methods to check,qualify and enrich th e find ings from the m ore established analytical app roaches.Independ ently of this, there is value in a p articipatory ap proach to the analysis ofpoverty and assessment of poverty-reduction efforts, since enabling the voice ofthe poor to be articulated and heard by stakeholders at all levels is the key tochanges in policy and practice (Carvalho and White 1997, Robb 1998, Hollandand Blackburn 1998: Part 2).One of the ways in w hich th e above is basically sound but p otentially misleading,concerns the relationship betw een the d ifferent alternative trad itions of enqu iry.

Some w orthw hile clarification has been contributed by d istinguishing the part ofthe argument that refers specifically to participatory methods or systematicstakeholder consultations where the focus is primarily on the links betweenknowledge, learning and action and the part that is about triangulatingbetween different sources of data and styles of analysis to achieve more robustand sophisticated understanding.For the latter purpose, it has been pointed out, the relevant distinction is notbetween survey-based and participatory m ethods, nor even between quan titativeand qualitative data, but between app roaches to data collection that are m ore or

less contextual. The relationship between contextual and non-contextual methodsis best seen as a continu um , with PRA-based exercises at one extreme, along w ith

7/31/2019 Booth Etal 1999

4/49

4

not only ethn ographic research bu t also field stud ies in farming systems, urbanand rural sociology and political science, to the extent that they employ anintensive case-study rath er than su rvey ap proach (Booth et al. 1997).This point that PPAs, in the narrow sense of PRA-based enquiries, are not theonly source of complementary understandings that can contribute to improvedaction on pov erty is widely app reciated in principle, if not always in practice. Itprovid es the p rincipal reason for d eliberately includ ing a review of ethnographicor anthropological sources among the background papers for the 1999 PovertyStatus Report.A fur ther clarification along these lines is also need ed, however. This may helpto explain why it was felt that this paper needed to be written as a separatecontribution to the discussion of how the poor perceive poverty in Africa.Notwithstanding the recognition now given to a multidisciplinary, rather thanmerely twin-track, approach, the language used to express the need for adiversity of persp ectives still tends to be skewed tow ard s the presu med strengthsof PRA-based enquiries. From the persp ective of anthrop ology and otherfieldwork-based disciplines, there is too exclusive a stress in these formulationson perceptions, as opposed to the practices that embody them; on visions ofpoverty, as opposed to experiences ofprocess; and on coping strategies fromwithin the range of relevant social responses.1

There is a well-established terminology in anth rop ology covering one of the mainthings involved here. What distinguishes the anthropological app roach issustained attention to both subtleties of meaning and belief (the emic) andpattern s of observed behaviour and events (the etic). Ethnograph ic work focuseson the often difficult business of interpreting the frequently contradictoryrelationship between these two dimensions of social experience.2 In these terms,what anthropological and other contextual methods have to contribute to theunderstanding of poverty is currently expressed too much as a contribution toemic understanding and not enough as an alternative perspective on etic issuesand on the crucial interface between the em ic and th e etic.A corollary of this is that if one interrogates anthropology just for answers topoverty-statu s questions of the type formulated above in effect, the qu estions towh ich PRA has trad itionally offered answers the results mu st be expected to bed isapp ointing. For certain, they will fail to do justice to the contribut ion that

1 We refer to the presumed strengths of PRA, because it is not even the case of a best-practice

PPA exercise either that it gives direct access to the perceptions of the poor (only to what som eof the poor say, in a particular context), or that it is entirely limited to recording the testimonyof the participants, as op posed to observing their actions and collecting indep endent evidenceon what happens and w hat people do.2 Emic and etic are roughly translated as the actors as against the observers perspectiverespectively, the analogy being with phonem ic as against ph onetic issues in language. Aclassic source on these m atters is Geertz s (1973, 1974) discussion of wh at is inv olved in thickdescription.

7/31/2019 Booth Etal 1999

5/49

5

anthrop ology an d method ologically cognate d isciplines are capable of making inthis area.This was strikingly confirmed by the initial search carr ied out for this paper . Itmight have been expected that anthropologists would have a great deal to sayconcerning how people defined by outsiders as poor perceive themselves; thevarious ways in which well-being, vulnerability, insecurity and their causes areunderstood; and how the nature and causes of poverty as locally-perceived have changed over time. How ever, it seems that what anthrop ological stud ieshave to say directly about these specific issues is both limited and, from theperspective of the d iscipline, rath er problematic.The focused anthropological literature on poverty in Africa turns out to beminimal, and w hat there is does not easily deliver up the visions of poor peop leabout their condition. This is partly a matter of intellectual focus. The treatm entof issues to do with p overty and well being is diffused across a range of specialistliteratures: in early works on kinship, political and marriage systems, religionand economy; in more recent research into rural production systems, foodsecurity, gend er, health, urban hou sing, identity and eth nicity, and so on. Tosome extent the entire corpu s of African ethnography is relevant. How ever, toassemble fragments of insight from across these scattered sources is not only amonumental task but also a rather questionable one, requiring as it doesextracting them from their theoretical and substantive contextsand forcing theminto engagement with an external discourse on another su bject.3

Another kind of issue has to do w ith the seriousness with w hich an thropology asa discipline approaches the interpretation of its material, including the spokentestimony of informan ts, poor and n on-poor. Obviously, no anthrop ologicalwork is raw ethnography from which one could expect to glean authenticperceptions in the sense of being uninterpreted and thus untainted by thetheoretical, political and personal concerns of the researcher.4 Under theinfluence of wh at is called the post-modern turn , recent an thropological work inAfrica and elsewhere has become m ore rather th an less conscious of this essentialpoint about th e nature of fieldw ork.As a result, few researchers today would pretend to provide an authoritativeaccount of how well being is defined by the x people or y group . A more likelyfocus is on the relations between power, knowledge and discourse, drawingattention to how different people might describe such issues in different ways

3 For example, recent monographs which deal with long-term socio-economic change (such as

Caplan 1997, Davison 1996, Moore and Vaughan 1994) in what is a generally relevant way arein fact more about the changing social contexts in which accounts of socio-economy areprod uced, than p roviding an account of local perceptions/ experiences of the kind this review

might have been expected to d raw on.4 The same of course applies to PRA and its use for poverty assessment, on which relevantd iscussion is prov ided by Mosse (1994) and Mick Moore et al. (1998)

7/31/2019 Booth Etal 1999

6/49

6

dep ending on who they are talking to and in wh at context. Statements aboutpoverty, wealth and well being are seen, in this way, as idioms through whichshifting relations of power whether within communities, or with the state andinternational agencies may be affirm ed, negotiated or contested .This is a view that wh en taken to extremes, can seem precious even in academiccircles. How ever, we w ould argu e that, even and perhaps especially in a policycontext, the heightened self-awareness that anthrop ologists bring should be seenas a particular strength of that trad ition. In this persp ective, the bad n ews is thatthe concepts of poverty to be found in anthropological work are those expressedby particular local people in particular social contexts, as interpreted by theanthropologist armed, or disarmed, with a more or less relevant analyticalperspective. The good news is that the conven tions of the discipline areincreasingly alert to the implications of this fact, possibly more so than those ofother social sciences where the same caveats app ly. In recent anth rop ology, thecommon charge of subjectivity is thus fairly typically answered in HenriettaMoores injunction that the politics of positionality and location should berecognised and ad dressed (Moore 1996: 2). In the best cases this entails that theresearchers theoretical framework and personal stance are made explicit, whichmeans that biases are more open to inspection.The problem, then, is not that anthropology has little to contribute to theillumination of the basic status-report questions who, why and what can bedone but, on the contrary, that a potentially important contribution may bemissed if the issues are posed in terms th at are overly influenced by the valu able,but d ifferent, contribution of PRA-based PPAs. This paper is not, therefore, anaccount of what anthropological studies tell us about the views of the Africanpoor on p overty and its causes. Nor d oes it preten d to reflect the voice of thepoor in any significant way . What it offers, instead , is a set of answ ers to thequestions who, why and what to do which generates in turn some importantcaveats regarding th e u se other evidence on these subjects for policy p urp oses.That may not seem very mu ch. How ever, we argu e that it is a potentially crucialcontribution. Moreover, it is imp ortant to app reciate that anthrop ology is acorrective or source of enrichment not just to survey analysis but also and n o lessimportan tly to PPA work. The best PPAs to date have und oubtedly been thosethat have benefited from the guidance of staff trained in an throp ology as w ell asPRA, or which have incorporated some form of review of relevantanth rop ological stud ies. The greatest single threat to the cred ibility of PPAs atpresent lies in the tendency for their practitioners to view PRA as a simple tool-kit for consulting the poor, rather than as a philosophy of triangulation andfitness to purpose implying an obligation to draw optimally on the insights ofboth local know ledge and relevant specialist work. High lighting the specificcontributions from anthrop ology may be of some h elp in this regard.

7/31/2019 Booth Etal 1999

7/49

7

The findings of the paper boil down to three sets of messages, correspondingbroadly to the core status-report qu estions iden tified at the beginning. Thesections of the pap er develop these in turn. Section 2 examines the imp licationsof anthropological evidence for the treatment of poverty concepts and theconstruction of poverty profiles, including the use of data from wealth rankingexercises at commu nity level. Section 3 ad dr esses the wh y question, drawingparticularly on work by anthropologists and social historians on peoplesresponses to sources of long-term change. Section 4 turns to the implications ofthese find ings and other an thropological evidence for the design of anti-povertyinterven tions. Section 5 sum s up .Key points and policy implications emerging from the p aper a re: Who are the poor? While anthropological work can help to enrich statistical

poverty profiles, a more important contribution may be in documenting thevariable, fluid, complex and contested categorisations and relationships thatconstitute the reality that poverty-redu ction efforts mu st contend with on theground . Such complexities are a source of un certainty, wh ich means thatplanning for p overty red uction n eeds a stron g learning-process orientation.

Why are they poor? Documented responses to structural change are

sufficiently diverse and so affected by the particularities of local structures,including n otably gend er relations, that mu ltiple pa ths of disimp overishmentor re-impoverishment remain more likely than homogeneous national orregional trends. This does not mean that established generalisations (e.g.about decontrolling rural markets, or legislating on womens rights) areunimportan t. But it does mean that effective anti-poverty action needs to bebuilt at least partly from the bottom u p.

What can be done? Anthropological studies help to remind us that the

primary stakeholders in anti-poverty operations are, of necessity, activeparticipants in constructing their own future, while the activities of states anddevelopmen t agencies are not always emp owering. It follows thatdevelopment interventions, including the implementation of anti-povertystrategies, are likely to benefit from an approach that is more institutionallyself-aware, as well as process-oriented and partially decentralised.

2 Poverty concep ts and profiles

In the light of our introductory discussion, the reader will not expect from thissection anything oth er than heavily interpreted an d positioned accounts of howpoverty is experienced, and who the poor are considered to be, among somemor e-or-less poor p eople in Africa. What use can this be expected to be? There

seem to be two main avenues, both of which depend on appreciating the

7/31/2019 Booth Etal 1999

8/49

8

difference of character and function between anthropological research andconventional survey analysis.

Vigdis Broch-Due (1995) emphasises that the concepts of poverty employed inmost d evelopm ent analysis are very thin, focussing on material and measu rableelements, such as income and nu trition. In contrast thick ethnograp hic workreveals far more complex, mu lti-layered pictures. The variou s concepts anddefinitions of poverty and wealth, or more broadly, ill being and well being,which policy agencies use have emerged in specific cultural and historicalcontexts. The same is true of the variety of terms which are found amongstAfrican peop les and languages. Material deprivation lack of food and income,poor health may be imp ortant everywhere. But ideas emp loyed by Africanpeoples in particular contexts select aspects of material life and group them withother attributes in distinctive ways, and this forms the natural focus ofanthrop ological work in or arou nd the topic of poverty.

There are two possibilities here. One is that, despite the difference in their focusand function, survey work and anthropological findings can be harnessedtogether in a way that imp roves the quality of both. This follows the nowfamiliar suggestion tha t questionnaire design and app roaches to survey an alysiscan be given greater purchase on reality by reference to the findings of contextualstud ies. Conversely, case stud ies benefit from being located , prospectively orretrosp ectively, in statistical d istributions (Calvalho and White 1997, Booth et al.1997). The other type of contribut ion possibly even more importan t is inprovid ing a basis for groun ded jud gements about the uses and limits of statisticalpoverty profiles in elaborating anti-poverty strategies.

A few of the themes developed in the following su b-sections may be candidatesfor atten tion of the first sort. All of them , on the other hand , contributesomething to the general point that the necessarily thin categories used insurveys, and h ence widely adopted in policy d iscussions about p overty, becomehighly misleading when they are interpreted as sufficiently describing the socialreality with which p overty-redu ction efforts have to contend on the groun d. Thisdanger of reification applies not on ly to the hard-core stat istical categories thebottom quintile, or those under the official poverty line but also, we suggest, tothe poor as they emerge from well-being rankings using PRA methods.Probably, it applies even to some of the functional groups that have beensuggested by Hanmer et al. (1997) as providing a more secure basis for policy-relevant p overty analysis.

Social categories and mu ltiple mean ings

Fairly obviously, local conceptions of poverty do not follow the same principlesof un iversality and consistency as app ly to survey-based pover ty analysis. Forexample, in some cases, poverty is associated exclusively with particularcommun ities or categories of peop le. Focusing on East Africa, Broch-Due

7/31/2019 Booth Etal 1999

9/49

9

contrasts general perceptions of the poor between small-scale pastoralistsocieties such as the Turkana, and the savanna societies. The former tend toconceive of wealth largely in terms of cattle, and :

contrast the prosperous life of their nomadic camps with the poor life offoragers, farmers, fisher folks and others who they conceive of as existingon the fringes of society. Irrespective of income or nu trition, the cultu ralconstructions surrounding cattle herders place them at the centre ofsociety and portray them as rich, while others wh o are a t the periph ery,in th e bush, are regard ed as poor (1995: 3).

Pastora list societies in general tend to define p overty in terms of lack of livestock,this being their key resource.

Iliffe ident ifies two different p atterns among pastora lists. Some societies, such asTuareg and Moors in West Africa and Tswana in southern Africa, incorporatedpoor peop le into openly inegalitarian societies. In contrast, the East Africanherd smen excluded the poor from ostensibly egalitarian ones (1987: 65). Forexample, among the Tuareg in southern Algeria izzagaren cultivators lived in ahierarchical society dominated by camel-owning warrior-nobles (ihaggaren).Among th e Maasai, by contrast, Waller (sum marised in Iliffe 1987: 68) argu es thatrich and poor complemented each other because rich (il karsisi) and poor (ilaisinak) had alw ays coexisted to mu tua l benefit with in Maasai society. Thegroups were linked particularly through demand for labour, which the poorperformed . But poor Maasai wou ld frequently leave pastoralism for farminglivelihoods, simultaneously shedding their socially-defined status as poor andtheir social identity as Maasai.

Certain group s stigmatised in the pre-colonial period have remained so. Forinstance, cultivator ex-slaves of Fulbe continued to maintain that they were notequal to Fulbe despite being championed by Guinean nationalists (Derman1973). For some, reluctance to integrate w ith mod ern economic processes,implying remaining in a state of poverty, has been in part a deliberate strategyaimed at retaining cultural autonomy in the context of the modern state; asargu ed for examp le for the San in Botswana (Lee 1979).

We are not saying that m ore individu al, material issues concerning food, incomeand health are not significant in these societies. We are certainly suggesting thatmulti-dimensional notions around well being and ill being can coexist, and beselectively drawn upon in particular contexts of self-definition or comparison an asp ect that an thropological work h as been at p ains to highlight.

Indeed, the words which capture notions like well being frequently havemu ltiple meanings which the same peop le may employ at different times. Forexample, in Kissi (Republic of Guinea) kende connotes well both in a generalsense of well being, fertility and prosperity, as when people ask a co kende? (are

7/31/2019 Booth Etal 1999

10/49

10

you well?) as part of standard greetings, or carry out a community sacrifice forkendea (well being). But it can also be used to refer specifically to physicalhealth.

In Malinke, the normal word for both poor and indigent in the nineteenthcentury was fangantan, meaning lacking pow er/ wealth. This fitted with cultura ltrad itions of centralised states and militarism, where w ar leaders ep itomised th eopposite, fanga (Iliffe 1987). The word is still used , frequ ently with the sam econnotations, when people relate settlement foundation histories, but also torefer simply to material wealth/ poverty, and even to p hysical strength of anendu ring or tempora ry natu re, or to express tiredness.

Multiple meanings in African language words are no surprise; we take it forgranted that poor can mean many things in different contexts in English.How ever it is a basic point worth noting wh en interpreting the word s people useor are quoted as using (in participatory poverty assessments, for example).Voices from among the poor is an expression that may need to be used moreoften in substitu tion for the usual voice in the singu lar.

Poverty and social relations

In discussions about the multi-dimensionality of poverty, it is commonplace thatthe ways people experience material conditions are mediated through socialrelations and institutions. Yet anth rop ological stud ies emp hasise that well beingis frequently also conceived or defined in terms of social relations and kinnetworks, and a p ersons place in them .

For example, Bledsoe (1980) describes for the Kpelle in Liberia how in broadterm s wea lth is peop le. Both mater ial prosper ity and social power are linkedwith having large numbers of dependent clients and kin who variously provideservices, labour an d p olitical allegiance. Similarly, among the Mende in SierraLeone, both w ealth and pow er are associated with big p eople, whether men orwomen, who have proven their ability to attract and support dependents(Richards 1986). Sup por t from kin or a patron is, concomitantly, seen asnecessary to avoid insecurity that might lead to destitu tion. Being held w ell bykin or a p atron w as an essential part of notions of well being am ong m aterially-poor Mend e villagers in the late 1980s (Leach 1994). Richard s describes howwh en one man died of starvation in Mogbuam a, the village he worked in, otherswere quick to explain tha t he lacked a patron (1986). In this sense, poor was asocio-political category, albeit one which could be occupied very temporarily inthe highly dynamic Mende world of shifting patron-client ties and reversiblefortunes.

Tierney (1997: 256) presents verbatim statements which illustrate generationaldifferences in local concepts of poverty in rural Tanzania in 1992: whereas anelderly man identifies the poor as those who d o not have p eople to live with an d

7/31/2019 Booth Etal 1999

11/49

11

eat with, a man in his twenties reinterprets the same p oint and describes the pooras those who do not have access to edu cation. Whilst the id iom shifts, neither theold man nor the young one d efines poverty in material terms, but rather in termsof networks, contacts and opp ortunities. This is not to say that m aterial povertyis unimportant far from it but to recognise that local perceptions of povertyhighlight other elements of the problem as well.

The centrality of sup portive kin n etworks to well being is an overarching th emeidentified by Iliffe (1987). Iliffe argues that, historically, poverty was associatedprimarily with the incapacitated and unprotected, with those who weredisabled, or who lacked sup port within kin and pa tron-client networks. Inseveral African languages the common word for poor implies lack of kin andfriends. One examp le is umphawi in the Chewa language of Malawi (1987: 7).Iliffe argues that, in the savannas, many societies distinguished ordinarycomm oners from a category of destitute that was identified w ith lack of normalsocial relations, and hence lack of sup por t other than char ity. For examp le,nineteenth-century Hausa distinguished talakwa (commoners) characterised bytalautsi (poverty, humility, meekness) from those suffering matsiata (distress,pover ty, anxiety, care) (1987: 42).

The significance of wealth in people in many areas also related to the relativeabun dan ce of land compared w ith scarce labour. Iliffe argues that the weakhousehold, bereft of male labour , has probably been the most comm on sou rce ofpover ty throu ghout Africas recoverable history (Iliffe 1987: 5). Stru cturalpoverty resulting from land scarcity as is more common in Europe hasdeveloped only slowly in the twentieth century, and combines with lack oflabour to p rod uce specific patterns of pov erty (cf. Broch-Due 1995). Yet thissimplified perspective overlooks both pockets of historically-important landscarcity, and the price which subordination to landholders could carry even inareas of land abu nd ance.

The workings of kinship and family continue to be crucial to the ways povertyand w ealth are defined and experienced. This has been picked up in a number ofparticipatory poverty assessments as well as in anthropological work (see, forexample, Nor ton et al. 1994, May an d N orton 1997: 98-100). Yet in materialterms, kin networks can be both a buffer against poverty, and a cause of itthrough the creation of forms of inequality around gender, generation or otherpositions (Broch-Du e 1995).

Life cycles, gend er and the flu idity of iden tity

Anthrop ological work focuses attention on how experiences of poverty and wellbeing vary over the course of peop les lifetimes. The developm ental cycle ofdom estic groups (Goody 1971) was a key concept wh ich d rew attention to howopportunities and vulnerabilities might shift through the processes ofestablishing marriage, having children, children growing u p, and ageing. There

7/31/2019 Booth Etal 1999

12/49

12

are links here with economists ideas about the importance of household size,dep enden cy ratios and so on. How ever anthropologists pay attention not only tochanges in economic status within developmental cycles, but also to howmembers at different ages and life stages may be differentially linked into widerkin networks and social relations, and hence be more or less secure andsupported.

Feminist anthropology in interaction with feminist work in other disciplines has not only deconstructed the household, but emphasised theinterdepend encies between d ifferent categories of household-member an d otherinstitutions, includ ing the state. These interrelationsh ips all provide settingswhere needs and rights may be negotiated and contested, in what Moore (1994:101-6) term s a system of redistribution. A large num ber of anth ropologicalstudies show how womens and mens experiences alter in the course of theirlifetimes in r elation to their changing positions in su ch systems (e.g. Guyer 1984,1997; Leach 1994; Moore and Vaughan 1994; Linares 1991; Whitehead 1984).

Equally, anthropological perspectives on child malnutrition and survival locatechildrens experiences in relation to their identity amidst broader social relations.Howard (1994), for example, shows how the selective survival of Chaggachildren in Tanzania depends on their gender and perceived clan alignment asrelated to birth order, affecting their status in the eyes of kin and particularly ofthe grand mothers mainly responsible for care post-weaning. In Sierra Leone,Bledsoe (1990) locates child fosterage arrangements and the circumstanceswhen they lead to malnutrition and deprivation in dynamics of patron-clientage and political alliance among Mende adults.

Current thinking on poverty acknowledges that certain groups of people arepar ticularly significant, both in term s of und erstanding the natu re of the p roblemand in defining priorities for policies aimed at red ucing poverty. Partly on thebasis of poverty assessments that include some element of iteration betweensurveys and contextual data-sources, female-headed households, the disabled,the elderly, casual labourers, street children and excluded ethnic groups havebeen recognised as being more vu lnerable than other sections of the pop ulation.

There is quite a strong case for taking these and other functional groups ratherthan the categories of poor and very poor defined by poverty lines as the focusfor monitoring of trends (Hanm er et al. 1997). Howev er, it is as well to point ou tthat these categories hardly less than the strictly statistical ones involve fairlyheroic abstraction from th e social relations in which poor people live. They areon the whole not groups in the proper sense of the term, and most are notsocially recognised categories. This is no doubt w ell un derstood in p rinciple, butthe danger of reification can manifest itself almost imperceptibly as analyticallanguage becomes the language in wh ich p olicy options are d iscussed.

7/31/2019 Booth Etal 1999

13/49

13

There is a general anthropological attitude to categorisation which has merit inthis context. When a pop ulation is d ivided into categories (such as the fun ctionalgroups indicated above) it is considered important to remain as aware of thelinks between the categories as of the contents of each box. Abstraction fromsuch relationships is legitimate but poses dangers when the original thickethnographic context gets forgotten, which is not only possible but likely, giventhe need for relatively simple, generalisable formulas to guide policy andpractice. The danger ap plies, we believe, not only to statistical poverty-linecategories and sociological concepts of the above typ e, but also to th e categoriesof the poor which emerge from community-level wealth or well-being rankingswh en ( as happ ens) these kind s of findings get abstracted away from the p rocesswh ich generated them.

Several types of example are relevant in addition to those mentioned already.They relate to the importance of the short-run micro-dynamics of poverty, theissue of multiple identities, and additional aspects of the embeddedness ofpoverty in social relations. One type of dynam ic issue has just been men tioned ,the sort relating to life cycles, but there are others. For examp le, the shiftingcircumstances of rural producers, the role of unpaid family workers and thecond itions faced by imm igrants, all raise again th e question about people m ovingback and forth betw een social categories, and betw een d ifferent levels or kind s ofpoverty.

The complexities of an individuals identity are well represented by the case ofthe female household head . While a woman may be locally regarded, and m aysee herself, in terms of female headship, this is unlikely to be her only iden tity, interms of how she sees herself, how she is perceived by others, and how she leadsher life. It is also likely that the notion of female-headed hou seholds as thepoorest of the poor, which has been a cornerstone of recent developmentthinking, is itself questionable.

Chant (1997) has explored this question in a way that is relevant here, eventhough the stud y is not concerned w ith Africa. She argues that a simpleclassification wh ich focuses on female-head ed hou seholds as amongst the p oorestis un satisfactory in several ways. It exclud es those female-head ed househ oldsthat are relatively wealthy; it fails to represent the advantages of being a femalehousehold head (includ ing greater personal autonom y, wh ich an ind ividu al maychoose, even if it means greater financial constraints); and it tends to ignore theimportance of interactions between, as well as within, households.

Links between social groups are revealed as important in another way inGrinkers (1994) study of the farmers and foragers of northeastern (ex) Zaire.Whereas in the past the foragers (pygmy hunter-gatherers) were treated, byanthropologists and development personnel alike, as a separate social group,Grinker shows that the links between the p ygmies and th e neighbouring farmers

7/31/2019 Booth Etal 1999

14/49

14

are central to und erstanding the lives of each group . Economic partnershipsbetween farmer an d forager men are fund amen tal to the livelihoods of both.

Recent work on disability emphasises how local perceptions and experiences ofdisability (e.g. Ingstad 1997, on research carried out in Botswana) are linked,throu gh ideas abou t misfortun e generally, to man y other aspects of life, such aswitchcraft, the breaking of taboo, and on-going relationships with deadancestors. The integration of a biomed ical explan ation of d isability with notionsabout cosmology shows how the categorisation of a person as disabled leavesout significant elements of a disabled persons experience, and strategies forcoping with disadvantage.

Porters detailed (1996) analysis of the lives of working children in northernTanzania provid es insights into the impact of the market on childrens economiccontribution to the household. By remind ing us that child ren are social actors intheir own right, Porter shows that children manoeuvre effectively within theirsocial netw orks to maximise their opportu nities for income generation. Portersaccount reveals the kind of social change taking place in rural areas childrensstrategies are increasingly conditioned by the market which may help toexplain why some children end up living on the streets in urban centres (1996:16).

A general p rinciple to emerge from an throp ological work is to question whethera category of the population which is associated with poverty is alwaysexperiencing hard ship, and , wh en they do, how th ey cope with it. An elderlyperson, for example, can draw on a variety of different social connections inord er to susta in a livelihood , provided that the social connections are there. Thekey to whether the person experiences poverty is not so much to do with beingelderly or not, but rather to d o with wh ether or not s/ he has family members, orother forms of social sup port, to turn to. Anthrop ologys emph asis on mu ltipleaspects of an individuals identity, and on the links between categories, shouldnot be regarded as a pedantic insistence on a level of reality which a disciplinehas made its own although it can be that too! Rather, diversification and theuses of social relationships are keys to h ow th e poor cope with h ard ship, as wellas to mu ch else in d evelopm ent.

Discourses around poverty

The various w ays people d efine and attribute mean ing to poverty can be seen interm s of d iscourse. Discourse per spectives in anth rop ology examine howpeople use par ticular types of langu age and imagery to represent them selves andothers in par ticular ways. The focus is on how these images are un derlain by,and reprod uced through, pow er relations, and on w hat their social, political andeconomic effects are rather than w hether or not th ey are tru e.

7/31/2019 Booth Etal 1999

15/49

15

Broch-Due (1995: 4) highlights some of the general insights which discourseanalysis of pov erty genera tes. First, it serves to emp hasise that poor is one ofman y social identities which an individual may p ossess. Broch-Due shows w hatthis identity means, how it is constructed and how it changes in differentcontexts. Second, the idea of discourse can help explain how p eople organisethemselves under a banner such as poor, highlighting the real effects of suchsocial constru ctions in opening or closing access to material resources. Third, itposes the questions how and why particular discourses are legitimated whileothers are p ushed into the background: The p ower to d efine reality is a crucialaspect of power and one of the major means by which certain groups can beforegrounded and empowered while others are silenced and suppressed (ibid.).And fourth , it p oints to the p ractical route of asking how counter-discourse m ightbe mobilised to effect change.

This sort of analysis linking what people say with what they do, and withrelationships of power, is relevant, also, to important issues in the policyprocesses of poverty red uction. In this context, the theme is taken up again inSection 4.

Putting m easuremen t in its place: reification and un certainty

Anthrop ological findings and analysis are imp ortant to the d ebate about povertyconcepts and p rofiles in Africa. But we have suggested th at this is not becausethey provide a privileged insight into the perceptions of the poor, still less intowhat poor peop le think. It is more because of the subjects trad itionalpreoccupation with observing and interpreting the complexities of behaviourthat su rrou nd issues of social categorisation in this field as in many others.

This is valuable, to a limited extent, because it may permit better abstractions tobe made for statistical purposes, a point that has been well made in recentd iscussions. But pr obably mor e imp ortan t is the role of anth rop ological insightin moderating and qualifying the policy conclusions that may too quickly andincautiously be m ade from statistical data.

The view is often expressed that for anthropological perspectives to be relevantto policy and planning, they must contribute to measuring poverty moreaccurately, and that their findings must be translated into a quantifiable value.Thus, references to anthropologists lagging behind on measurement (Bevan1996: 4) imp ly that they might catch up even tua lly, and that th ey at least ought totry. In contrast, the general attitude running through much of theanthropological literature conveys the message that the obligation to measureand quantify is a diversion, and that instead it is the quality of interaction thatmatters. The position w e have been taking accords some legitimacy to both ofthese points of view.

7/31/2019 Booth Etal 1999

16/49

16

Our em phasis is dr iven by a practical concern. Anthrop ological work can andshould contribute to improving the hand le that statistical data-collection has onthe complexities of local reality. Good statistics are imp ortan t to good policy.But no less important, and probably more so, is the contribution thatanthropological work can make to policy that is sound in recognising thediversity and comp lexity of the real social world, and knowing the limits of thesimplified formulas that (necessarily) form the stuff of national and globalstrategising for poverty redu ction.

There are many practical implications of a healthy respect for social complexity.The most importan t have to d o with th e way, in the context of any kind of socialengineering, complexity translates into uncertainty, and thus into theinevitability of actions hav ing unanticipa ted consequences. In the project andprogramme literature, there is a long-established argument leading from socialcomp lexity and un certainty to th e critique of blueprint planning and ad vocacyof learning-pr ocess designs. The same logic applies, with equal or greater force,to the implementation of national anti-poverty strategies. The more serious theybecome, the more African efforts to eradicate extreme poverty will need torespond creatively to unexpected difficulties, and maybe some unexpectedopportunities too, in handling the actual social relations in which poverty isembedded.

3 Poverty processes and responses to change

Throughout Section 2, in which concepts of poverty were reviewed, theanthropological evidence suggested the multi-dimensional nature of localperceptions of pov erty. As a genera l observat ion this is increasingly accepted ,and even comm onp lace. May and Nor ton (1997), for examp le, examine thecomponents of peoples experience of poverty in South Africa, and advocate amu lti-dimensional concept of poverty for u se by d evelopm ent p ersonnel, so thatun derstan dings of the term reflect mor e closely the exper ience of the poor . Butthere are two parts to this argument, one of which has received moreacknowledgement than the other.

PPA-based discussions about the meaning of poverty move back and forth, aslocal concepts often d o, between w hat social science would call definitional issuesand explan atory issues respectively. Thus, poverty is often treated as meaning alack of economic or social assets, because these things are seen as keys to thecausal process that make people poor or vulnerable. For our pu rposes it makessense to distinguish these analytical steps more clearly and identify two distinctparts of the multidimensionality issue. The point then is that PPAs andanthropological studies serve to alert us not just to possible complications in themeaning of poverty, but also to the role of a wid er ran ge of causal p rocesses than

has been recognised by m ainstream p overty analysis.

7/31/2019 Booth Etal 1999

17/49

17

Although PPAs have done a great deal recently to widen the range of asset andaccess issues considered in policy debates (see Booth et al. 1997 for a survey),mu ch of this insight der ives from anthrop ological literatur e. Because of theemphasis placed on long-term observation and the use of archival and well asfieldwork sources, anthropological research is better placed than PRA fieldworkper se to shed light on many kinds of issues that are relevant here. This formsone of the two them es of this section. The other concerns th e likely limits to thegeneralisability of observed processes of impoverishment, disimpoverishment,vulnerability and coping.

Discussion of these issues is assisted by an initial basic distinction drawn fromIliffe. In his ambitious review of the history of Africa s poor, Iliffe distinguishestwo kinds of poverty. Structural poverty is the long term poverty of individualsdue to their personal or social circumstances, and conjunctural poverty istemporary hardship into which ordinarily self-sufficient people may be thrownby a crisis (1987: 4).

Iliffe argues that the greatest twentieth-century changes have been inconjunctural poverty, which at the beginning of the century was causedpr incipally by climatic and p olitical insecurity. The direct effects of these factorsdiminished through broad increases in wealth, diversified sources of income,improved infrastructure and governance, wider markets, and improvedmed icine, so that structural and conjun ctural poverty converged. Conjunctura lpoverty arising from mass famine and war has, however, re-emerged since the1960s. At the same time, structural poverty und erwent shifts linked tocolonialism and its aftermath, and processes such as commoditisation,opp ortunities for (male) wage labour, taxation, the growth of land markets, newfood and cash crops, the rise of national policies and planning, and externalpolicies such as stru ctura l ad justm ent. Certain new categories of poor werecreated by economic chan ge and colonial rule.

Our review of anthropological findings begins with a range of issues that areespecially relevant to forms of conjun ctura l pov erty. We then add ress stud ies oflong-term str uctu ral change an d implications for assessing the imp acts of policychan ge since the 1970s.

Cosmologies an d social ecologies

In general, anthropological work draws attention to the embeddedness ofpeoples understandings of economic and social issues within broader, culturalund erstandings of how the world works and their place within it. From thisperspective, misfortune, whether to the person or the community and whethermanifested in terms of material suffering, ill-health or social problems, may betriggered by phenomena and events which lie well outside the purview ofconventional dev elopment analysis. This is a d imension that is probablyunderstated in PPA field reports, given the sensitivity and openness to

7/31/2019 Booth Etal 1999

18/49

18

misinterpretation of these kinds of issues, but whose importance to practical anti-poverty w ork can easily be und erestimated .

Witchcraft, sorcery, malicious spirits, or the breaking of rules regardingreproductive activities, can all trigger problems for peoples well being and thesuccess of economic activities (e.g. agriculture) on which they depend (Gottleib1992). The un predictability of these phenomena, and of the actions of otherpeople that may trigger them, can contribute to vu lnerability and insecurity.

For example, Howard (1994) describes how for the Chagga of Tanzania thepresence of poverty and even more the occurrence of kuvimba (to swell), theSwahili term people commonly used to describe one of the symptoms ofkwashiorkor, tend to be taken as evidence of violation of Chagga culturalprecepts regarding cosmological balance, particularly those that deal withmarriage and reprod uction (1994: 246). On similar lines, Fairhead and Leach(1996) describe how among Kissi peoples in Guinea, breaking ancestrally-setrules concerning the timing and placing of sexual activities can result in maa, apu tting off track or out of synchron y of the social and ecological worlds, w hoseconsequences can include, simultaneously, ill-health, failures in social relations,infertility, failures of economic activities, and ill-being (lack of kendea) moregenerally.

De Boecke, wr iting about the Great Lakes region, shows how inap propr iate socialactivities can p rovoke famine to go around the land (1995). Periods of hun gerare thus ascribed social causes from within the community rather than beingconsidered from without.

Such possibilities should not be associated with an imagined isolated Africantrad itionalism, only to be overturned by modernity. For there is eviden ce ofsome vibrant persistence, albeit reworked, of such ideas in highly contemporary,worldly contexts, whether in associations drawn between the actions of Africanpoliticians and national droughts, or in contexts of conflict and war (e.g. Lan1985).

Nor are such ideas necessarily shared an d invoked u niformly. Rather, peoplemay m ake claims about their own or others links w ith spirits or the su pern atura lworld s in ways wh ich m ark and reprod uce social difference and p ower r elations.For example in many societies elderly women are particularly vulnerable towitchcraft accusat ions. Someon e gaining unexpected riches may be assum ed tohave privileged links with djinn sp irits (e.g. Jackson 1977). Susu eld ersmanufacture self-images as feared sorcerors to consolidate control over wivesand jun ior men , and their labour (Nyerges 1997).

Claims to control over, or pr ivileged relations w ith, land spirits have in a nu mberof societies been linked to broader power, wealth and ability to extort fromcommoners (Schoffeleers 1979). On the other hand , spirit possession is widely

7/31/2019 Booth Etal 1999

19/49

19

used as a mean s by which wom en and other dep ressed categories exert mysticalpressures upon their superiors in circumstances of deprivation and frustrationwhen few other sources are available to them (Lewis 1966: 318).

Moral econom ies: safety nets or exploitation?

With due regard to their diverse forms, many studies document moraleconomies and village redistributive systems of various kinds which in the pastserved to provide a minimum level of security and protection to the poor. OnKilimanjaro, for example, Howard describes how in the nineteenth century thedestitute were und er the care of the chiefs, who saw that they d id not starve or gonaked , and in return they ow ed labou r to the chiefs (1994: 242). The literatu redoes not paint a nostalgic picture of mer rie Africa. It is recognised that thesesystems were frequently pervaded by hierarchy and inequality, and excludedwh ole categories of people barren w omen, wid ows, twins, or those d efined asmad . Furthermore, different people may present such arrangements in verydifferent w ays; the image of moral village commu nity un der chiefly control m aybe an invented tradition of village elites (and the anthropologists who haveinteracted with them), overlooking the extortion, contest an d alternative survivalstrategies that the accounts of poor women and men might have stressed (e.g.Moore and Vaughan 1994).

The switch in representations of pre-colonial Rwanda in academic circlesfollowing the genocide is anoth er case in point. An exactive past, with the Tutsielite using forms of clientage which left extreme poverty among manycultivators, has been transformed into the checks and balances of the feudalsystem (Pottier 1995).

Both their complex and varied social relations, and the co-existence of differentinterpretations, should serve to nu ance and qualify the p icture of breakdow n oftrad itional arran gements sometimes evoked in broad -brush accounts of Africansocio-economic change. Nu merou s stud ies document transformations in ruralrelations of prod uction an d distribution und er the influence of colonial and post-independence policies, economic change, and the impact of world religions (e.g.Linar es 1991; Nyerges 1988; Hill 1977; McCann 1987, 1995). Whiletransformations in patterns of vulnerability are evident, the general breakdow nthesis does not hold.

Food security and famine

Anthrop ological literature on food security overlaps in man y respects with w orkon prod uction and coping more generally. Yet there are several key insightswhich anthropological work has generated, and discussions it has initiated,which serve to nuance and qualify the view that is taken of the relationship

between conjun ctural and structural issues in hun ger. They are worthsummar ising briefly here.

7/31/2019 Booth Etal 1999

20/49

20

De Waal (1989), in a provocative work, showed the cultural and historicalparticularity of the ways famine and hunger have come to be understood in theWest, and by developm ent and hum anitarian agencies. Dominant perceptionsderived from Malthusian n otions of crisis, he argu ed, misconstrued the p rocessesthrough which famine arose and the ways people in Darfur, Sudan (but alsoelsewhere in Africa, it was implied) defined and u nd erstood it. De Waalexplored peoples own definitions of hunger and famine, in which famine thatkills, as emph asised by hum anitar ianism, is but one category . The book argu edthat peop le defined and app roached destitution in ways wh ich placed priority onrecovering their livelihoods, hold ing onto seeds, cattle etc. And it argu ed thatpeople die less because of lack of food, than because of health problems anddisease, which are exacerbated by concentra tion into camps.

De Waals conclusions have been variously confirmed and debated in growinganth rop ological literatures arou nd each of these themes. Research has exploredemic (i.e. peoples own) definitions and categories of food problem and famine(see Shipton 1990), and the relationship between hunger and disease (e.g. Dias1981, Miller 1982). And a large literatu re examines coping resp onses to foodshortages and famine. This work generally emp hasises how ru ral peoplethrough out Africa both plan for contingencies and respond actively to them. Keyprevention strategies include diversification of livelihoods, consolidation ofsavings into illiquid, indivisible or incontestable forms, and social investments.In crisis, liquidation of savings, service labour, and movement, become key(Shipton 1990: 363).

Recent work, taking account of prevailing economic conditions in Africa, hasemphasised how coping strategies may become enduring livelihood adaptationswhich may reduce vulnerability but imply lower levels of well being in otherresp ects (e.g. Davies 1996, Mortimore 1989). Richard s (1993) points ou t that,given the level of hardship currently faced by Africas small-scale farmers, theability to maintain a livelihood at all is oftentimes a brilliantly innovativeachievemen t (1993: 70). Nev ertheless some argu e that too much attention tocoping is over-opt imistic, occlud ing the experiences of those wh o fail to cope.

A recurring theme in anthropological works on famine and food security is theway normal social relations may be transformed during food crises, sometimeswith end ur ing effects. Close atten tion to appar ent un ities such as descentgroup, compound, family or household reveals how relationships may shiftwithin them as food crises precipitate reallocations of responsibility by gender orgeneration (e.g. Pottier and Fairhead 1991, Vaughan 1987) or even force theabandonm ent of norm al kinship obligations. For examp le, Shu wa Arabs innorth-eastern Nigeria called the dearth of 1972-74 the era of refusing to acceptbrotherhood (in Iliffe 1987: 254).

7/31/2019 Booth Etal 1999

21/49

21

Anthropological work on recent food problems has addressed theirinterrelationsh ip with conflict and w ar in many par ts of Africa. In this, thestrategic creation of hunger, poverty and destitution as a part of conflict has beenemphasised (e.g. Keen forthcoming, Richards 1996), as has the ways that somepeople profit from famine/ destitution processes. In general, Keen emph asisesthat it is not always those w ho are most vu lnerable (economically) who su ffer themost in war-induced poverty and hun ger. Rather, those wh o are vulnerablepolitically, and often well-off finan cially, are frequen tly targets in war. Havingmore to be taken from th em, they lose relatively more, but can also become p artof a new category of conflict-induced poor .

Poverty and stru ctural chan ge: generalisin g abou t comp lexity

Many studies by anthropologists and others have been devoted to providingnuanced understandings of how large-scale structural changes have worked outin particular par ts of Africa. Often this has involved close attent ion to thequalitative dimensions of wh at they meant to people, and how th ey shaped an d,in turn, were shaped by social and cu ltura l relations and stru ggles. Trajectoriesof change, and peoples responses, took particular forms in particular places,with diverse implications for poverty and wealth, their form and their socialdistribution.5 One policy-relevant conclusion from this body of work as a wholeconcerns the range of factors that are of critical importance as interveningvariables between th e source of change and the impacts on p oor peop le.

That outcomes are invariably influenced by peoples responses, which areembed ded in locally-specific social, cultural and political relationships, is the keypoint here. This introd uces an element of variability that needs not to beun derestimated. Thu s a theme of man y studies has been the imp ossibility of, ordangers in, genera lising abou t the effects of colonial and capitalist transformation(cf. Moore 1988). Much of the relevan t work ad opts a far longer tim e-scale thanapplies to current policy interest in the impacts of economic liberalisation andinstitutional reform, but for this very reason it bears serious consideration. Thisforms our second th eme.

Land and rural produ ction systems

As anthropologists and social historians emphasise, rural resources are accessedthrough diverse, and frequently overlap ping, social institutions. Shiftingpatterns of resource access and control during the twentieth century, affectingpoverty and vulnerability, have both come about through, and in turn shaped,

5 More recent works have emphasised, also, how accounts of change, given by different

informants at different times, not only reflect the particular knowledge sources available tothem, but also particular social and political commitments in presenting change and itsoutcomes in certain ways (e.g. Moore and Vaughan 1994). This links up w ith the points abou t

positioned natu re of accoun ts and of discourse in Sections 2 and 4.

7/31/2019 Booth Etal 1999

22/49

22

social relations and id entities (Berry 1993). Some of the most impor tan t shiftshave involved changing intersections between local social relations andprocesses of state action and developm ent.

Changing relations to land are a key and well-documented d imension of patternsof poverty and w ealth du ring the twentieth century. While man y parts of thecontinent still have relatively plentiful land relative to labour, in many othersland scarcities have become significant, frequently linked to growing landmarkets. Sometim es scarcity has been manu factu red through exprop riation. InRwanda, for example, structural poverty was created in the nineteenth centurythrough the expropriation of land and labour by rulers, creating a dependentclass of day labourers (umucancaro) (Iliffe 1987: 62).

Land alienation to Europ eans in eastern and southern Africa had varying effectson the poor, depending on how the process interlocked with existing socialinstitutions. For examp le, among the Kikuyu, the mbari landholding families ofinegalitarian pre-colonial society contained many dependents ahoi, athami without full landhold ing rights, wh ile there w ere also many prop ertyless clients.Land expropriation to European settlers, while relatively mild in itself,intensified competition for land that was already increasing as a result ofpop ulation growth and commercial change. It was mainly clients, ahoi, athami,and women, who lost out alongside the expropriated families themselves(Kersh aw 1972).

More generally, tenurial reforms during and since the colonial period havefrequently marginalised p articular social group s, such as w omen of reprodu ctiveage (e.g. MacKenzie 1990, Davison 1996). By adding further layer s to thealready-multiple institutions through which people can legitimise land claims,they have p ushed people poor and wealthy alike to invest more r esources inup holding p osition in the institutions wh ich might safeguard their claims (Berry1993). Poor peop le may lose out in ensu ing struggles over resources less becauseof their lack of formal rights than because of their relatively weaker social powerto uph old them amid st other claiman ts. In this context it is evident that th e linksare between poverty an d resource control, rather than own ership as such.

Gen der an d the construction of rural livelihoods

A large literature documents transformations linked to the introduction of newcash and food crops. An imp ortant them e in this work is the imp ortance of socialrelations, and those of gender in particular, in mediating the eventualimp lications for p oor p eoples livelihoods.

The colonial introduction of cash crops frequently meant increasing rural wealth,but at some cost in terms of greater vulnerability to world markets and weath er,and w idening inequ alities. Positions in social relations gave some peop le thenecessary access to land, labour and capital, as well as suitable state contacts, to

7/31/2019 Booth Etal 1999

23/49

23

invest in cash cropping, while others were marginalised especially certainwomen an d stran gers (e.g. Leach 1994, Berry 1988, Guyer 1984). Loss of controlover labour an d income amon g wom en, and sometimes jun ior men, precipitatedheightened tensions and struggles within households.

Similar struggles characterised some post-independence development projects,such as those for irrigated rice in the Gambia (e.g. Carney and Watts 1991, Dey1988), and changes linked to crop commercialisation (e.g. Mackintosh 1989, onSenegal). As in the colonial per iod, interven tions predicated on assum pt ionsabout a male breadwinner and pooled household incomes sometimesconfronted very different arrangements for the allocation of labour andresponsibility. They includ ed, frequently, separate income/ expenditure streamsbetween women and men, and complex intra- and inter-household negotiationsover labour and prod uct rights.

The ensuing struggles have many implications for how poverty is understoodand experienced. First, individual p overty cannot be read off from householdexperiences, or pred icted from household data. Second, struggle can be linked toloss of economic au tonom y/ control for certain peop les, or to experiences ofdomestic violence, which negatively affect well being. Third, severa l au thorsdescribe the battle of the sexes as becoming a key idiom in which peopleencapsulate their experiences of poverty.

Schoepf and Schoepf (1984), for example, describe such a situation in the early1980s in Kivu, (ex) Zaire, where struggles turned particularly on cloth, as amarker of social status. Mens failure to meet their obligations to provide it fortheir wives, and w omens struggles to obtain it for themselves, provided a p otentidiom for reflections on poverty. Yet as Schoepf and Schoepf point ou t, sexismhere deflects attention from exploitation by plantations, traders and officials, aswell as serious p roblems of land shortage (1984: 118). Thus a fourth imp lication as Whitehead (1990) has emphasised is that a focus on intra-householdstruggles should n ot deflect attention from th e w ider m anifestations and causesof poverty.

Most of the anthropology of gender and rural production has emphasised theincreasing vulnerabilities faced by rural women, albeit sometimes stronglydifferentiated by age and other aspects of social status. How ever a small butnotable set of studies has begun to focus on African masculinities and socio-economic change, yielding a nascent area of work on male poverty. Forinstance, Pottier (1994), referring to north Kivu (ex Zaire), Zambia and Uganda,draws attention to m ens grow ing inabilities to m eet full bridewealth obligations.This is linked to red uced leverage in intra-household r elations and loss of statusand self-esteem, especially in economic settings where womens personalincomes from food cropping for the m arket are increasing.

7/31/2019 Booth Etal 1999

24/49

24

Other r ecent w ork d esigned to synthesise the results of anthrop ological stud ies ind ifferent socio-cultural settings in Kenya (Francis 1998) and more w idely (Francisforthcoming) contributes in another way to shifting the focus onto genderrelations, as distinct from the status of women . Initially concerned w ith tracingthe imp act on gend er relations of changing p atterns of ru ral livelihood, includ ingagricultural intensification and commercialisation, and the decline of migrantlabour , Francis overview d ocuments comp ellingly the reverse relationship. Thatis, a more or less common set of macro-economic conditions end up producingquite a varied set of livelihood outcomes, as their effects are mediated byd ifferent local forms of the partr iarchal bargain. Across seven stud ies of genderand rural livelihoods in different parts of Kenya, quite different householdprocesses emerged, dep ending not on ly on the potential economic reward s fromcooperation between men and women, but also on the pre-existing domesticau thor ity relations and ideologies of common or d ivided interest (1998).

As is well known, there is an on-going d ebate about how best to integrate gend erawareness into development policy and p lanning. In the wom en-in-developm ent(WID) era women were treated as a separate category, to be specifically targetedwith assistance in development initiatives. The critique of WID emp hasised tha twomen were being ghettoised through this process, and that there were severelimitations to what could be achieved in terms of improving their quality of life.Gender and development (GAD) has promoted the mainstreaming of genderawareness, so that gend er relations, rather than w omen, are the focus. Thisapp roach was to have been integrated at all levels into developm ent policy. Thelast few years have, however, seen gender sidelined once again, the specificneeds of women tending to be ignored within a more general concern with thepoor . Jackson (1996) has argued convincingly for the need to rescue gend er fromthe poverty trap, and ensure that gender relations maintain a central place inpolicy and planning.

The recent history of anthrop ology as a discipline has seen gend er integrated insuch a way that m ost anthrop ologists now take account of gend er relations evenwh ere their research is not primarily about gend er. This is probably because oneof the disciplines main concerns has always been kinship, and this formed afound ation for gender ana lysis in recent decades. The attention given to gend errelations as a basic explanatory variable in the stud y of the consequences of long-term macro change is another instance where, perhaps, anthropology hassomething more general to teach. The shifts in anthrop ological app roaches havehelped to demonstrate what mainstreaming gender might mean for practicaldevelopment thinking.

Resources, pop ulation change and ineq uality

Another key dimension of twentieth century change involves populationdyn amics. Here a num ber of long-term, multi-disciplinary studies involvinganthropology investigate trajectories of population change in relation to shifting

7/31/2019 Booth Etal 1999

25/49

25

social and resource-control relations. They show that in m any areas p opu lationincrease has been accommodated through agricultural intensification andlivelihood diversification, without growing general impoverishment butfrequen tly at the cost of grow ing inequality. For examp le, Haswells sequence ofstud ies of a Maninka village in the lower Casamance area of the Gam bia, in 1949-50, 1961-62 and 1973-74, showed that while village population increased from 483to 771 and land cultivated per capita fell, agricultural innovations such as cattle-keeping and wet rice cultivation allowed food p rodu ction to more than keep p aceand th e traditional hungry season largely to disappear. Yet the village wasincreasingly dominated by the 22 per cent of larger, longer-settled, householdswith access to labour an d good relations with the outside world . Others man ydrawn from ex-slave families lacked swampland and access to innovations, andbecame poorer the beginnings of a class of landless agricultural labourers(Haswell 1975).

A similar story emerges from Machakos in Kenya, wh ere the m uch-cited successof intensification between 1930 and 1990 resulted in increased food production,better average incomes and environmental improvement despite a five-foldincrease in pop ulation (Tiffen et al. 1993). Yet it has been convincingly shownthat inequality increased by class and gender (Murton 1997, Thomas-Slayter1992) and that certain women experienced new forms of struggle to survive(Rocheleau 1991).

For Rwand a, Andr and Platteau (1995) show how pop ulation increase has beenaccompanied by growing inequality in land distribution, and growing povertylinked to landlessness, as inheritance arrangements have intersected with theoperation of land m arkets. In Kanam a Commun e of Gisenyi they comp are 1988(672 peop le per km2) with 1993 (787 per km 2). The num ber of households owningless than 0.25 ha. of land increased from 36 per cent to 45 per cent, while thepercentage owning more than 1 ha. increased. Land sales had increased rap idlyas those with off-farm income bought up land from the poor (often as distresssales). Vulnerable people who had p reviously accessed land v ia customar yclaims found that such claims were no longer upheld, both as there was notenough land to go round and because many Rwandese claimed that they onlyapp lied to inherited, not purchased, land . Thus many return migrants, widow s,separated women, polygamous wives, handicapped people, orphans, andchildren of broken marriages, foun d themselves landless.

In Kivu, Zaire, similar processes were compounded by the confusion betweenmu ltiple polities and social institutions regu lating land access. Power fulcoalitions between state administrators, traditional chiefs, army and police, andlarge landholders, have made it possible for much land to be extorted frompoorer claimants, and for new claimants to extort heavy taxes of labour dutiesfrom those wh o lived on the land. This has contributed to growing structuralpoverty (Fairhead 1990, Pottier and Fairhead 1991).

7/31/2019 Booth Etal 1999

26/49

26

Migration

Changing patterns of migrant labour have also been linked to shifts in povertyand its distribution. Migrant labourers, frequently emp loyed on cash crop farmsduring (and since) the colonial period, became in some places a new category ofthe poor. For examp le, Mossi and other labourers from Burkina Faso constitu tedtwo-thirds of the labour force on Ivorian plantations in the 1960s and, excludedfrom minimum wage regu lations and forbidd en to form trad e unions, constitutedan exploited u nd erclass (Iliffe 1987). Migrant labou rers often already came fromthe p oorest strata of society.

Sometim es what began as stigma labouring for other s was seen locally as a signof poverty became more acceptable over time, as in Hausaland between the1930s and 50s. Farmers sometimes worked for others in particular years wh entheir own crops failed or they h ad economic or health problems in th e family, asin Yako village in sou thern N igeria du ring th e 1930s wh ere poor relations wereequated with those who needed to offer labour to others at harvest in exchangefor crops (Forde an d Scott 1946: 64). Yet migration cou ld also be seen p ositively,and v iewed as a means of escaping poverty and vulnerability. Among the Fulbeit was Fulbe themselves who tended to migrate rather than their cultivatingslaves, because the former had fewer means open to them to earn cashhonourably (Derman 1973: 157)

Iliffe draws attention to interaction between the old poverty of incapacitationand the new p overty of migrant labour. For example in Bugan da by the 1950smost cultivators grew cotton or coffee, often employing immigrant workers fromRwand a, and a few poor Bagand a, who earned p oor wages (only about one-thirdof rates paid by other employers). Many migrants hun g around w aiting for odd

jobs, an d a tenant p lot . A prop or tion of th ese were red uced to destitu tion bypersonal circumstances, including illness (for example, those who contractedleprosy feared to return home w here they wou ld be rejected) and alcoholism. In1961-64 10 per cent of all homesteads housed people living alone, a conditionstrong ly associated w ith poverty in local thou ght (Iliffe 1987: 153). Other stud ieshave examined th e impacts of migrant labour for w age emp loyment on those leftbehind (Murray 1981), or on gender dynamics in the production systems ofsend ing areas (e.g. Moore and Vaughan 1994).

Deb t and hered itary poverty

A further area in which anthropological work and long-term studies have madesome impact concerns the question of debt and its link to patterns of poverty. Ingeneral, debt and pawning have long been stock-in-trades of survival for mostpeop le. It has been part of norm al life, and the reverse side of saving, notnecessarily a sign of poverty. Indeed , in Hau saland from the 1950s through th e

1970s, Hill (1984) argu ed that it was th e inability to borrow (except from fools orstrangers) that was the acid test of poverty. How ever anthrop ologists have

7/31/2019 Booth Etal 1999

27/49

27

been somewhat divided in their views of rural indebtedness and relations withmoneylenders, perhaps reflecting different analytical lenses as much as realdifferences between areas (for example, see the debate between the Marxist-inspired position of Watts (1983) and the neo-populist position of Richards(1990) on Hausa and Mend e debt relations).

In savanna regions especially, it seems that the need to sell off harvest to meetdebts did and does sometimes hold farmers in a cycle of poverty an d h un ger.Yet long-term studies differ as to the extent to which poverty is inheritedbetween generations or is continuous among particular groups. A study of twosites among the Kikuyu in Kenya in the early 1970s did find that familiessup plying hired labour were descended from client group s of pre-colonial times(Collier and Lal 1980). Similar hered ity of poverty was foun d am ong plan tationworkers in Cameroon (Henn 1988). Watts (1983) argu ed that a hered itary class ofpoor farm ers was emerging in 1977-78 in Katsina, Northern Nigeria, among thosewho lacked inherited access to best land, were unable to practice risk-avertingstrategies, got into debt an d had to sell their labour pow er to survive.

Yet others do not suggest that hered itary classes are forming. Richard s (1986)argues th at am ong ru ral Mend e in central Sierra Leone, the d ynamics of dealingwith hazardous ecologies, uncertain social life and personal contingencies meanthat status is mu ch more dynam ic. Personal misfortun e and accident can plungeeven wealthy farmers into temp orary poverty. The fortun es of patrons andclients can reverse w ithin a few years, let alone a generation.

Hill (1977, 1984) makes a similar argum ent from her d etailed study of two Hau savillages. Poor people were those who lacked grain, manu red land or off-farmskills and were forced to sell their labour, collect bush products or practiceun rewarding crafts. While poor people found it hard to escape their condition,Hill argues that various factors militated against the formation of hereditaryclasses (for example: the dispersal of wealth among sons at a mans death; thefact that personal enterprise was still vital to success; the insecurity of thesavan na environ men t for farming fortunes; and the risks of ill-health w hich couldstrike anyone).

Coping and vuln erability in urb an areas

Taking a long historical perspective, Iliffe (1987) argues that the poor seen intowns in pr e-colonial times, and seen increasingly in m ultiplying an d expan dingtowns during the colonial period, were not bred in towns, products of urbandegeneration, but were largely rural poor seeking opportu nity. Yet towns gavepoverty new forms: crowdedness and slum squalor, unfriendliness,un emp loyment. New forms of poverty in town s sup plemented older onesdu ring the twentieth century: proletarianisation, unem ployment, delinqu ency. Inthe colonial period, most of the p oor in African tow ns w ere u nskilled labourers,wh ile the very poor were the ill, the very old, and children alone. Subsequent

7/31/2019 Booth Etal 1999

28/49

28

changes, including retrenchment of skilled workers linked to economic reforms,have qualified this situation.

Recently, key areas where anthropologists have contributed to understandingpeoples experiences of urban poverty include studies of urban housing on onehand , and w ork on coping and survival on the other. We consider these in turn.

Anthropological studies of the urban poor like studies in other disciplines generally find that overcrowded housing and poor health conditions are amongthe key problems people face. A significant literatu re has focused on hou singissues, examining how the personal circumstances and range of exchangeentitlements available interact to shape p eoples oppor tunities. In as mu ch asexperiences of poverty are linked to housing conditions, the emphasis is verymuch on the dynamic processes by which peoples housing fortunes change.With reference to Nairobi, for example, Amis (1987) shows the key role of stateurban housing policies, a role acknowledged by poor people themselves, indeclining opp ortu nities. Policies to restrict self-built hou sing mean thatindividuals must look elsewhere, principally to the market, where poor peoplefind their position weakened by low w ages and insecure employment.

In studies of economic coping and survival, the concept of the informal sectorcoined by anthrop ologist Keith Hart has framed man y studies. Debated andcritiqued as th e term h as been, it has nevertheless focused attention on th e man yways people get by; ways frequently invisible to official statistics andquan titative studies based on th em. Most towns in Africa are rich in informalmeans of survival, whether in small-scale industry, trade, artisanal activities,services or begging. Many insightful stud ies emp hasise how the activities andpractices of poor people, including their health-seeking behaviour, are shapednot only by economic factors but also by questions of social identity and culturalunderstandings (e.g. Wallman and associates 1996, King 1996, Nelson 1987).

Trade has been especially significant for women in many towns, especially inWest Africa where they dom inate markets (e.g. Clarke 1994). The mu ltiplicity ofpetty trad ers is itself a symptom of poverty, as consum ers can only afford to buyin small quan tities, and poor sellers comp ete for minute rew ard s (Iliffe 1987: 174;Hewitt d e Alcnta ra 1992: 8). The twentieth -centu ry history of informal-sectoractivities has involved processes of destruction and re-creation. Modern indu strydestroyed many occupations but created others; economic reforms and theactivities of the state both formal and informal have also had effects(MacGaffey 1987; Tripp 1997).

Nelsons recent re-study documents a number of such processes for Matharevalley, Nairobi (Nelson 1997). For examp le, beer brewing used to be a keyactivity throu gh wh ich peop le got by. But as the profitability of brewing wasrealised, small industries came in and undercut their market and moreoverlobbied successfully to outlaw artisanal brewing. This was not only an economic

7/31/2019 Booth Etal 1999

29/49

29

process but one wh ich involved the culture of consum ption, and changing tastesamong u rban p opulations.

Along with other disciplines, then, anthropology has joined the debate overwhether the informal sector is the last, secure bastion of survival for the urbanpoor, or vulnerable to take-over by the formal sector as soon as it gets profitable.However it examines such experiences as embedded in and shaped by broadershifts of cultural preference and urban social relations. Other studies havefocused on the culture and practices associated w ith small-scale trad e, and theirchanging ar ticulation w ith includ ing resistance to state policies. For example,Clarke (1994) provides a highly insightful and detailed study of the perceptions,life circumstances and actions of market women in Kumasi, Ghana, while Horn(1994) deals in a comp arable way w ith wom en trad ers in Zimbabwe.

Social netw orks an d associational life

Another strand of work examines the relationships, networks and forms ofassociation in wh ich people engage in towns, w hether for h ousing, or other socialor econom ic activities. Explicit references to perceptions and experiences ofpoverty in this literature are few, though the processes discussed are clearlyrelevant for un derstand ing poor peoples experiences.

In general, anthropological work shows that kinship ties were not dissolved orweakened in town s but remained vigorous, albeit in re-worked ways (e.g. diffusenetworks replacing corpora te groups). For example, Marris describes how tosupport aged parents was an absolute duty in Abidjan in the 1960s, and 49 percent of arr ivals went imm ediately to a relative (Marr is 1961). Yet stud ies alsoshow the cross-cultural and city-to-city variability in type of sup por t. Wheresup por t was lacking, it was a major cause of angu ish. As Iliffe reviews,Copperbelt families in controlled housing were less hospitable, while Ghanaianworkers in the 1970s said th ey were treated bad ly by their families.

Family structure made a difference: the large corporate households of the Luomad e it easier to accomm odate relatives in N airobi than it was am ong Amhara inAddis Ababa, characterised by shallow bilateral kinship and unstable marriage(only 121 of 600 hou sehold heads suppor ted extra relatives in 1960). Hencefamily structures helped to shape th e types of supp ort people could expect, withimp lications for who might face destitu tion. Again the natur e of local socialrelationships turns out to be critical to the implications for poverty of agenera lised p rocess of change.

Forms of association am ong n on-kin w hich p rovide forms of social sup port andsecurity include informal savings grou ps (e.g. Nelson 1996, Tierney forthcoming),urban-based cult associations and secret societies (e.g. Cohen 1969), andnu merou s informal networks an d mu tual help arrangem ents (e.g. Mitchell 1969,Ard ener and Burman 1995). A particularly interesting strand of work focuses on

7/31/2019 Booth Etal 1999

30/49

30

how such arran gements, and associated cultural practices, operate among streetchildren. Street children are widely assumed to be amongst the poorest andsoccially most deprived of urban inhabitants, yet they show some of the mostcreative means of gaining livelihoods and ensuring mutual support (e.g. PEA,1990).

A general theme to emerge from the anthropological literature is that many ofthose who face poverty display impressive personal qualities of ingenuity,innovation, resilience and perseverance, and also skills at working effectivelywith group s and networks to achieve what cannot be done on ones own . AsIliffe has put it, in th e past [the African p oor] relied for their survival chiefly ontheir own efforts [and this] is their inheritance amidst the harshness of thepresen t (1987: 8). The significance of the anthrop ological evid ence whichsupports this thesis is that it portrays the poor as being rich in the humanresources associated with entrepreneurship and teamwork, qualities which arewidely recognised as being of fundamental importance for the sustainability ofdevelopment initiatives.

We retur n to the implications of coping in Section 4. Here it remains tounderline some implications of the broad picture on responses to long-termchange.

Sum min g up comp lex change: diversity as a reality for policy

Anthropologists have still written relatively little that is focused specifically onpoor peoples responses to economic liberalisation and adjustment since the1980s. What exists generally combines this concern w ith an interest in processesthat , in th e per spective of history, are rather less episod ic, like commercialisationor urban growth. It makes sense therefore to add ress the literature on structuralchange and poverty as a whole and aim to extract some general principles thatmay be app licable to cur rent p olicy concerns.

On the basis of the above review, there is more than adequate evidence for theproposition that the imp acts of general processes of change on the livelihoods ofpoor people are never just that: a generalised set of outcomes reflecting in amechan ical way a common set of causes. People respondto change, and the waythey respond is constrained, shaped and given distinctiveness by the economic,social and political relations includ ing those that app ly at th e commu nity andhousehold level in which their lives are embedd ed. These structures remain,even today , highly diverse. The imp lication is that outcomes are often verydifferent in different parts of Africa, including within countries and regions ofcountries.

It is not that there are n o generalisations about large-scale change that stand upto scru tiny. For examp le, in Fran cis (1998) comparison of ru ral Kenyanexperience, it is clear that the existence of modern legislation on womens rights

7/31/2019 Booth Etal 1999

31/49

31

was consistently favourable to livelihood outcomes across an otherwise highlyvaried pan orama of change. Although more contested, the proposition that somemeasure of downsizing of state trading monopolies was necessary for ruraleconomic recovery in the last decades wou ld stand up well to scrutiny across therange of cases. As these examples suggest, however , the genera lisations that canbe mad e with confidence are limited and rather simple. No less imp ortant, fromthe p oint of view of designing m ore effective poverty-reduction app roaches arethe more complex propositions that can be advanced about the range ofvariability, and the kind of factors that seem to be relevant to the explanation ofdiversity.

Matching our