1 IntroductionWhile a great deal of the academic and policy dis-cussion about poverty has been conducted in ap-parently gender-neutral terms, closer scrutiny revealsthat it has often been premised on the concept of amale actor and of male-centred notions of well-being and agency, with obvious limitations foraddressing the gender dimensions of poverty Dif-ferent meanings have been given to poverty in thisdiscussion which favour different kinds of dataand generate different implications for policy Inthis paper, I want to explore what these differentapproaches can tell us about the gender dimensionsof poverty and how they can be used to build greatergender-awareness in poverty analysis.

A useful starting point for discussing different waysof looking at poverty is throughThe idea of depriva-tion. Poverty can then be seen in terms of an 'ends'perspective which focuses on the actual outcomesof deprivation or the extent to which basic needshave been met; or it can be seen from a 'means'perspective which entails a concern with the ad-equacy of resources at the disposal of the poor andtherefore the extent to which basic needs could havebeen met. A concern with ends is much more likelyto translate into interventions which seek to addressshortfalls in basic needs while a concern with meansis much more likely to translate into interventionswhich focus on expanding the resources at the dis-posal of the poor. In the next sections of this articlewe will be investigating the insights provided by a'means' as well as an 'ends' perspective into the gen-der dimensions of poverty In addition, however, athird approach would be to conceptualize povertyas an inseparable relationship between means andends; in this case, poverty encompasses both theneeds and priorities of poor people as well as thestrategies which most closely correspond to them.Such a conceptualization would entail a very differ-ent approach to the question of data and a very dif-ferent stance on policy and we will be exploring itin the concluding sections.

2 The 'Means' Perspective:Poverty and IncomeThe majority of economists have generally tendedto concern themselves with the 'means' aspect, be-lieving that this proxies the potential for exercisingchoice, while avoiding judgements about the actual

ii

choice of 'ends'. Moreover, they have also tendedto concern themselves with the measurable and themarketable. The equation of poverty with house-hold income, as in the poverty line approach,exemplifies both the measurement as well as theinstitutional biases of traditional economics. Themeasurement bias generally takes the form ofreducing the estimate of poverty to a single scalarwhile the focus on household income privilegesthe market as the main institutional mechanismthrough which basic needs are met. It also reflectsits gender bias since little attempt is generallymade to ascertain how equitably household incomeis distributed among household members and theextent to which gender inequalities in basic needsfulfilment are a feature of this distribution.

Recent shifts in the conceptualization of means from'income' to 'private consumption', which includedthe value of self-produced consumption goods, andexpansion of the concept to include common prop-erty resources and state provided goods all repre-sent attempts to overcome the market bias of tradi-tional measures. However, while most of these ex-panded versions are in principle measurable on theindividual, they have in practice focused on house-hold based measures. This partly reflects the com-plicated measurement problems associated withimputing the value of household consumption toindividual household members. It is instructive forinstance, that two recent studies, one in Bangladesh(Pitt et aL, 1990) and the other in the Philippines(Haddad et al., 1992), both attempting to imputehousehold consumption to different members whilemaking allowances for their differing energy require-ments, found in the Bangladesh case, that there was'some discrimination against males by the house-hold' (p 1155) andin the Philippines case, that calo-rie shortfalls were equally borne by all members ofthe family, irrespective of overall calorie adequacy'(p 13). While the conclusion for the Philippinesstudy might be considered plausible - evidence forgender biases in the intra-household distributionof basic resources is much weaker and less consist-ent in the Southeast Asian context - the findingsfor Bangladesh are less persuasive, given some ofthe evidence that we will be citing suggestingextreme gender inequalities in mortality as well asnutritional status. The Bangladesh finding suggestseither that the assumptions that economists areusing are flawed or else that life-threatening

12

gender asymmetries in well-being outcomes reflectbiases in the distribution of basic needs resourcesother than nutrition.

The concern with 'means' has become increasinglyassociated with a concern with the productivity ofpoor people's labour, because it is now widely rec-ognized that this is the primary income-earning assetat their disposal in much of the world. This is evi-dent in the World Bank's current poverty strategywhich prioritizes labour-intensive growth, generatedprimarily through the market, but supported bypublic investments in the human capital of the poor.Such an approach has the advantage of makingpoverty central to growth rather than an after-thought. Its success in addressing the genderdimensions of poverty, however, will depend onthe actual relationships between gender divisionsof labour, income and well-being which it encoun-ters as well as which it helps to promote.

3 The Ends Perspective: Povertyand Well-beingAn alternative approach to the measurement ofpoverty has focused on basic human indicators andwas prompted by a concern with the actual realiza-tion of basic needs rather than with the potentialvalue of basic entitlements or, to put it another way,with shortfalls in outcomes rather than in incomes.While this has sometimes prompted the combina-tion of incommensurable indicators of well-beinginto a scalar measure through the use of arbitraryweights - as in the Physical Quality of Life Index(Morris 1979) and the 1-luman Development Index(UNDP 1990) - this is not essential. The 'ends'perspective can also be operationalized as a vectorof key well-being indicators which permit policymakers to ascertain both value and trends in thedifferent dimensions of poverty, which while pos-sibly correlated, are not reducible to each other(Kabeer 1989). One of the major advantages ofwell-being indicators is that they are measureddirectly on the individual, the level at which pov-erty is actually experienced, and require far fewerassumptions in the interpretation of results thando household-based measures.

They also lend themselves more easily to genderdisaggregation in a way that household-based indi-cators do not. Another advantage, and one that may

make objective measures of well-being appropriate,at least in the first instance, for capturing the gen-der dimensions of poverty is spelt out by Sen(1990a): 'Especially in dealing with poor economies,there are advantages in concentrating ort suchparameters as nutrition, health and avoidance ofmorbidity, and educational achievements ratherthan focusing purely on subjective utility in theform of pleasure, satisfaction, desire fulfilment,which can be moulded by social conditioning and aresigned acceptance of misfortune' (p 133). Theproblem with using subjective measures to capturethe gender dimensions of poverty is that culturalrules, norms and values not only tend to devaluewomen's well-being in many societies but also tomilitate against recognition by women themselvesof 'the spectacular lack of equity in the rulingarrangements' (p 149).

The focus on 'human' indicators of well-being hasbeen associated in the development community withsuch agencies as the UNDP and is less tied to themarket as the key route to human-centred develop-ment. It sees the poor as deprived of the basic op-portunities to lead 'long, healthy and creative lives'as a result of the maldistribution of income, assetsand human capabilities and calls for a combinationof market opportunities and state support to coun-ter the exclusion of the poor from the fruits ofdevelopment. While the judicious use of the mar-ket is seen as creating an enabling environmentfor releasing people's creativity and talent, publicaction can play an important role in both counter-ing market imperfections and managing the fruitsof economic growth in the interests of all (see dis-cussion in Elson 1993). Such a policy approachrecognizes the exclusions and segmentations whichcharacterize 'real' markets and make them an in-hospitable environment for the disenfranchisedsections of society but its ability to extend thisinsight to recognition of the gender dimensions ofthis problem will depend on the analytical aware-ness of the agencies attempting to counter marketimperfections.

4 Gender Disjunctures in PovertyAnalysisAs we stated earlier, conventional conceptualizationsof poverty are implicitly male-centred. This allowsfor the assumption underpinning both income!

13

consumption as well as well-being measures thatshortfalls in either reflect poverty: shortfalls inincome/consumption translate into shortfalls inchoice and are manifested in shortfalls in well-being. A gender perspective introduces certaindisjunctures into this equation. The assumptionsunderpinning income/consumption as well aswell-being measures is that shortfalls in eitherreflect poverty: shortfalls in income/consumptiontranslate into shortfalls in choice and are manifestedin shortfalls in well-being. Gender introduces cer-tain disjunctures into this equation.

For instance, if household income/consumptionmeasures of poverty are used, evidence that onegroup of households has lower income than anothergroup can generally be taken as evidence that malemembers of those households enjoy lower purchas-ing power. Similarly evidence that one group ofmen is less well-nourished than another groupusually constitutes evidence that they come fromlower-income households. However, by and large,evidence that one group of households has lowerincome than another group tells us very little aboutthe levels of well-being enjoyed by its female mem-bers. Similarly evidence that one group of womenwas less well-nourished than another group ofwomen does not necessarily constitute evidencethat they came from lower-income households. Infact, it is possible in some contexts for femalemembers of households with higher income to beless well-nourished than male members from poorerhouseholds.

The basic conundrum that plagues attempts toconceptualize the gender dimensions of poverty -whether it is through the 'choice' concerns of eco-nomic measures or the 'well-being' concerns ofhuman indicators - stems from the analyticallydistinct but empirically seamless operation of thesocial forces that create scarcity on the one hand,and discrimination, on the other. In principle, itwould be possible to hold constant for householdincome and find evidence of gender inequalities inbasic well-being or to hold constant for gender,and find income-related inequalities of well-being;the gender dimension of poverty is concerned withthe interaction between the two. Other thingsbeing equal, it would be logical to assume that thejoint effect of gender discrimination and absolutepoverty would be the exacerbation of gender

inequalities among the poor. However, the forcesthat create inequalities of wealth and opulence in asociety embody quite different social norms andmaterial practices to those which create inequalitiesof gender and may prevent the automatic transla-tion of shortfalls in income into the widening ofexisting inequality It is only through a context-specific analysis of these forces that it becomespossible to determine whether gender inequalitiesin well-being and agency are offset, exacerbatedor unaffected by the problems of scarcity In thefollowing sections we will be considering some ex-amples of how gender disjunctures between income,effort and well-being work in different regions.

5 Disjunctures Between Incomeand Well-beingIncome/consumption resources are merely meansto the ultimate end of meeting the basic needs ofindividuals and therefore their validity as a measureof poverty ultimately rests on their ability to cap-ture the magnitude and distribution of shortfallsin individual well-being. The empirical data sug-gests that while such measures do appear to capturethe existence of shortfalls in well-being, they donot necessarily capture their distribution amongindividuals, specifically in this case their genderdistribution. The disjuncture between income andgender inequalities is discernible even at highlyaggregated levels of empirical analysis. Globally, itis still the case that the majority of the world's poorare concentrated in the Third World, in rural areasand that around half of them are to be found in SouthAsia. The head count index of poverty in 1985 inthe poorer regions of the world (World Bank 1990:29) ranked South Asia as having the highest per-centage of poor people below the poverty line (51per cent), followed by sub-Saharan Africa (47 percent), followed by the middle east and northernAfrica (31 per cent) and finally Latïn America andthe Caribbean (19 per cent).

These figures tell us something about the magni-tude of households with income shortfalls acrossthe world but very little about the gender distribu-tion of these shortfalls. The most direct estimatewould be the percentage of women and girls in thepopulation below the poverty line. However, unlessa significantly large enough number of householdsbelow the poverty line are populated entirely or

14

largely by females, there is unlikely to be a greatdeal of difference. One of the main reasons whyfemale-headed households has emerged as anindicator of the gender dimensions of poverty isprecisely because of their visibility in household-based measures of income-poverty. However,taking account of female-headed households, wewould still be none the wiser about women's accessto household income compared to men since wewould know nothing about the distribution ofincome or consumption in households below thepoverty line which were headed by men.

The use of well-being indicators offer a way for ex-ploring the connection between the incidence ofpoverty in a country and gender inequality. Variousmeasures of mortality (life expectancy, sex ratios,under-five mortality, maternal mortality and infantmortality) are often used as broad indicators of well-being because they summarize access to the keysources of physical survival across the population(Stewart 1985; Sen 1990). The overall length oflife expectancy appears to have a reasonable corre-lation with the wealth of a region. Among the poorercountries, overall life expectancy in ascending orderis sub-Saharan Africa (46-53 years); South Asia(55 years); North Africa (58-62 years); East/South-east Asia (60-64 years); West Asia (64-68 years);Latin America/Caribbean (64-72 years). Thedisaggregated life expectancies of women and menalso follow this order: in terms of absolute well-being, both sexes appear to be worst off in sub-Saharan Africa and South Asia which are also thepoorest regions in income terms.

However, while poverty goes a great deal of the wayin explaining differences in absolute well-being ofwomen and men across different regions of theworld, it is clear that even at this highly aggregatedlevel it does not fully account for relative differen-tials in their well-being. In terms of gender differ-entials in life expectancy, it is important to bear inmind that male infant mortality tends to be gener-ally higher than that of female (a pattern largelybelieved to have a biological rather than socialbasis) so that women tend to live longer than menin much of the world. Globally, the life expectancyat birth is 65 years for women and 62 for men(UNDP 1995) and this pattern of higher female lifeexpectancy prevails in most of the high-income aswell as poorer parts of the world. However, the

regional distribution of gender differentials in lifeexpectancy does not conform to the regional distri-bution of poverty or of overall levels of life expect-ancy As far as gender differentials in life expectancyare concerned, women are most disadvantaged inSouth Asia followed by North Africa and West Asïa.They are least disadvantaged in sub-Saharan Africafollowed by Southeast/East Asia.

The geographical distribution of sex ratios providefurther evidence that there are distinctive regionaland sub-regional patterns to gender inequalities inwell-being which have very little to do with pat-terns of poverty There is a belt of masculine sexratios (i.e. ratios of over 105 males per loo femalesin the 1980s) which stretches from areas of north-ern Africa across the middle east and the northernplains of the Indian sub-continent and includingChina and which encompasses quite different lev-els of national income, suggesting that factors otherthan economic resources are implicated (UnitedNations 1991; Momsen and Townsend 1987).Marked masculinity of sex ratios are reverse to thepattern to be expected on the basis of biological sexdifferentials in chances of survival. Yet so signifi-cant is the weight of social factors outweighing andreversing biological patterns in some regions that,despite biologically normal patterns in the rest ofthe world, we have the phenomenon of more than'100 million missing women' (Sen 1990b). In fact,the swathe of masculine sex ratios coincides withwhat Caldwell (1982) has called the 'patriarchal-patrilineal-patrilocal belt' suggesting that localgender relations or what we might call 'regionalpatriarchies' play a significant role in mediating thetranslation of economic resources into individualwell-being.

The evidence for gender differentials in basic well-being indicators is less well-documented, consist-ent or marked in other regions and indeed some-times occurs in the reverse direction. Studies fromSoutheast Asia found some milder forms of discrimi-nation within the household, usually against chil-dren rather than adults, female rather than maleand higher birth orders rather than lower (Deolalikar1992; Senauer et al. 1988 ). In Latin America andthe Caribbean there was also some evidence ofgender bias in favour of boys as far as nutrition is

For a discussion of excess female mortality in the gender power: a comparative study in SoutheasternIranian context see S. Razavi Agrarian chance and Iran, DPhil thesis, Oxford University, 1992

15

concerned (UNDP 1995). A review of studies froma number of SSA countries provide little evidenceof a bias in favour of boys; in fact there was evi-dence of both gender neutrality in nutritionaldistribution or a bias in favour of girls, particularlyin West Africa (Svedberg 1990). Where data doesexist for these areas, poverty appears to be associ-ated with greater intra-household inequality(Haddad and Kanbur 1990; Desai 1995).

What broad conclusions about the relationshipbetween gender, income and well-being can wedraw from this analysis? As we have noted, income!consumption measures tend to be collected at thehousehold level on the assumption that such income!consumption is distributed among household mem-bers in proportion to their need; extra informationon the intra-household distribution of well-being isonly necessary if such an assumption is violated inpractice. Lipton, for one, has suggested that theactual intra-household distribution of food andshelter in the world is 'surprisingly close' to the as-sumption of need-based distribution 'except inNorth India, Bangladesh and Pakistan' (1994).However, this should not be taken to suggest thatdeviations from the assumption of equitable distri-bution within the household are a minor regionalaberration since the areas referred to contain, aswe noted earlier, a significant proportion of theworld's poor. Moreover, recent data from the 1991census, suggests that inequalities in sex ratios, likelyto be a consequence of growing gender inequalitiesin mortality especially among younger children,have spread to the southern states of India andalso increased among poorer caste households (Drezeand Sen 1995). Furthermore Lipton's conclusionappears to be based primarily on the results ofliterature from South Asia (Harriss 1986) andsub-Saharan Africa (Svedberg 1990) and takes noaccount of the broader dïstribution of masculinesex ratios outside the Indian subcontinent. Whilethere were very few studies of intra-householdgender inequalities of the kind found in these twosub-continents, the persistence of masculine sexratios in this broader swathe of countries notedearlier suggests that the intra-household distribu-tion of welfare continues to be biased againstfemale members in these societies to the extent oftruncating their survival chances.'

However, it is also clear that measures of genderinequalities in well-being do not by themselves tellus a great deal about 'poverty' per se. The reluc-tance of most observers to draw the conclusion onthe basis of evidence on gender inequalities inwell-being that women are 'richer' than men in sub-Saharan contexts and 'poorer' in the South Asiancontext and that high-caste women in northernIndia are 'poorer' than low caste women suggeststhat few believe the gender differentials in well-being are purely a product of gender differentials inpoverty What our analysis also suggests thereforeis that local constructions of gender relations playan important intervening role between householdincome and weil-being so that 'culture' has pro-found material consequences, not only in shapingpriorities and perceptions, as noted by Baulch (thisvolume) but also in shaping allocative behaviourand practice.

6 Disjunctures Between Labourand IncomeAlthough measures of poverty were found to be poorpredictors of gender inequalities in deprivation andwell being in the cross-cultural context, the form ofwomen's economic activity appears to have greaterbearing on this question. In certain parts of theworld, particularly in the 'patriarchal-patrilocal-patrilineal' belt noted above, powerful norms aboutfemale propriety serve to constrain women's abilityto dispose of their own labour, channelling it intothose areas of the labour and commodity marketwhich are considered most compatible with theirseclusion within the precincts of the household.Where such norms do not prevail, women may befound in a larger range of economic activities andacross a wider set of locations. The higher rates ofeconomic activity recorded for women in sub-Sa-haran Africa compared to South Asia, and in south-ern states of the Indian sub-continent compared tothe northern states illustrates not only the greaterinvolvement of women in field-based stages ofagriculture in these areas but also in marketingproduce. This has implications for the distributionof gender inequalities in basic well-being. Sen(1990a) notes that gender differentials in labourforce participation rates are much more closelyaligned with gender differentials in life expectancyacross much of the low-income world. Within theIndian context, Dyson and Moore (1983) pointed

16

out that higher rates of female labour force partici-pation and greater public mobility in the southernstates of India than in the northern states were asso-ciated with both greater female autonomy and alsowith lower excess female mortality and with more'feminine' sex ratios. More recently, Dreze and Sen(1995) have shown that in the Indian context thenegative association between female labour forceparticipation and excess female mortality holds, evenwhen regional variables are held constant.

This suggests that attempts to generate labour-intensive growth may have positive implications forwomen's physical well-being but we need to dis-tinguish between improvements in women's well-being which are based on their increased economicvalue to their households and improvements whichreflect their increased agency The two need notnecessarily coincide and may have different impli-cations for gender-responsive poverty strategies. Thefact that women are economically active meansthat investments in their physical well-being has aproductivity payoff to the household economy butit does not necessarily imply that they exercise con-trol over their own labour. Even in regions with ahigh incidence of female own-account farming andenterprise, such as West Africa, gender asymmetriesare evident in the ability to command labour power,both of self arid others. Roberts (1988) notes howthis asymmetry occurs in the West African contextbecause husbands and senior males have non-re-ciprocal rights over the labour of wives and femalemembers. Women's obligations to work on hus-band's holdings or on compound holdings whichare controlled by the senior male restricts the amountof labour they have left for their own-account ac-tivities. Furthermore, their access to hired labour isconstrained by their lack of cash and of the socialpower necessary to sustain long-term contracts.There are possibilities for reciprocal labour arrange-ments between women, but these involve the inter-temporal redistribution of women's own labourrather than an enhancement of their labour entïtle-ments relative to men. Consequently, even in areaswhere women's own account activity is culturallysanctioned, they operate under greater labourconstraints than men from the same household.

In other parts of Africa and in much of South Asia,women's labour contributions to household produc-tion tend to he subsumed under male-controlled

processes. In areas of strict seclusïon, women eitherwork as unpaid family labour or in home-basedforms of economic activity where they relinquishcontrol over the production process to male house-hold members who initiate the production process,decide the scale of labour contributions required andare generally responsible for disposal of the proceeds.Even where the norms of female seclusion may notbe relevant men exercised control over the disposalof female labour; Sender and Smith (1990) foundfor instance in rural Tanzania that men, even thosefrom tIte poorest households, forbade their wives totake up waged labour.

One form of inflexibility in the disposal of ownlabour thus inheres in local norms and practiceswhich constrain women's access to and mobilitywithin the labour market while a second inheres inmen's non-reciprocal claim to female labour. A thirdand widely prevalent constraint that women face isthat while household labour almost everywhere isabout making a living and caring for the family itis largely women, and sometimes children, mainlyfemale children, who are responsible for unpaiddomestic chores and child care. Poverty is oftenassociated with forcing both women and childrento seek to make a living as well, so that poorerwomen are most likely to have longer working daysnot only than men in their households but alsocompared with women from better off households.In India, Sen and Sen(1985) note that poor womenare most likely to combine wage work, income-replacing activities and domestic chores whilespeciaiization in largely domestic chores was mainlythe feature of women from wealthier households.Illina Sen's (1988) study from rural Maclhya Pradeshfound that women worked longer hours overall thanmen and that women and men from poorer house-holds were also more likely to participate in workof greater arduousness and distance from the home.Lloyd and Brandons (1991) work in Ghana alsodraws attention to the leisure-work trade off: the)'found that male-headed and female-headed house-holds may have the same income levels but thatfemale heads have to work longer hours to achievethem.

Gender also differentiates individual ability to trans-late labour effort into income in the market place.The average wage for women is around 75 per centof the male wage outside agriculture for countries

17

where data is available; this is likely to be an under-estimate of the gender disparity because it excludesthe agricultural sector where most women workand disparities are likely to be high (UNDP 1995).The gender disparity in wages partly results fromthe various constraints on women's mobility notedabove and hence their confinement to the moremarginal segments of labour and commodity mar-kets. But it is compounded by other factors. Firstof all, investment in the productivity of femalelabour is generally lower. Women and girls aregenerally disadvantaged in the acquisition of for-mal education and specialized skills which couldimprove their access to a wider range of opportuni-ties, both in the market and state sector. Femaleenrolment ratios in higher education in lowerinconie countries are typically half the male ratio(UNDP 1990).

Second, the exchange rate for female labour in themarket place is lower. Studies of the gender dispar-ity in earnings suggest both direct and indirect wagediscriminatïon. Lipton (1983) cites data from In-dia showing that women received 70-80 per cent ofmale wages in agriculture which he attributes to ashorter hired working day for women, to loweraverage body weight and hence lower caloric costto employers and to the fact that women were chan-nelled into different and less weil remunerated tasksto men rather than to direct wage discriminationper se. However, it is worth noting that gender in-equalities in wages vary considerably across India;female wages are around a half of the male wage inMaharashtra, around two-thirds in most of SouthIndia and around three-quarters in West Bengal andparts of northern India (John and Lalitha 1995). ltis therefore possible that the causes of the disparityalso vary considerably across India and that directwage discrimination should not be ruled out.

Unfortunately, the kind of detailed empirical datathat would allow this to be explored is largely miss-ing. One exception is Kapadia's anthropologicalstudy (1993) which provides evidence of both wageand non-wage discrimination in one district inrural TamO Nadu. She documents the existence ofmale' and 'female' tasks fri agriculture as well asgender differentials in wage rates Two reasons weregiven for this disparity: that mens work was harderand that men would be humiliated if they were paidthe same as women. What this meant was that even

when women took up vacant 'male' jobs, they werepaid the female wage rate rather than the male one,bearing out the observation made by Elson andPearson (1981) that women enter the labour mar-ket as 'inferior bearers of labour' rather than as'bearers of inferior labour'.

Finally, gender disparities in returns to the sale oflabour reflects gender disparities in returns to the'human capital' embodied in labour; in other words,for any given level of 'human capital', women re-ceived lower returns. Lipton (1983) cites data fromIndia showing that women in the poorest ruralgroups found it harder than men to raise wage ratesthrough improved human capital endowments suchas health, education and nutrition. While there wasevidence to suggest that even among unskilled farmlabourers, extra education brought higher wagerates, Indian data showed that this effect was sub-stantial for men but insignificant for women. Simi-larly it was only among men that increases in wagerates were significantly associated with betternutritional status. Collier et al. (1986) notes that inrural Tanzania there was extreme skewed access tonon-farm wage employment. Access was deter-mined largely by education, age and gender. A 36year old man with secondary education had a threein four chance of such employment while a womanof the same education and age had only half of thatchance. With complete primary education, she hada quarter of the chance and with partial primaryonly a fifth.

The exclusionary and discriminatory practices whichgovern women's disposal over their own labour, theirability to access the labour of others and the returnsto their labour efforts suggests that while labour-intensive growth strategies may indeed generategreater demand for, and increase the returns to, thelabour of poor people, they are likely to have gen-der-differentiated effects. In the short term, it willbe men that will benefit first and where women dotake advantage of the new opportunities, it may beat the cost of intensified work loads for themselvesor for their daughters who are withdrawn from, ornot sent to school, in order that they can substitutefor their mothers in domestic labour. The with-drawal of daughters from school in the context ofmarket reform in Vietnam is opening up a gendergap in education which previously did not exist(Desai 1995). In the long run, some of the labour

18

market imperfections may begin to be eroded butthe next generation of young women will continueto be at an educational disadvantage in marketparticipation. Clearly public action now will helpto lower the greater immediate as well as longerterm costs of market participation for women andtheir daughters.

7 Listening for Change:Participatory Approaches toPoverty AssessmentIn an earlier section, 1 noted some of the problemsassociated with relying solely on subjective assess-ments of poverty as a way of identifying poor wom-en's priorities because of the power of social condi-tioning in shaping the 'choices' that women maketo the extent that they may be resigned to, andindeed actively promote, the distribution of re-sources which discriminate against themselves andtheir daughters. This is a valid point and is echoedin Jackson (1995) in her critique of the populistclaims made for FRA as 'giving voice' to the percep-tions of local people and the implicit assumptionthat their articulated perceptions are necessarilycomplete truths. The gender dimensions of thisproblem are spelt out by Jackson: 'Edwin Ardener'sanalysis of mutedness as a feature of the politics ofcommunication in research draws attention to theproblems of articulating perceptions (1975). Forwomen who are excluded from dominant worldviews and male vocabularies it is not wise toassume they can, or will, simply express theirpriorities as PRA assumes'.

Given these limitations, what role is there for par-ticipatory assessments of poverty in addressing itsgender dimensions? I would argue that participa-tory assessments, like any other methodologicaltool, is as gender-blind or as gender-aware as itspractitioner. At present, the gender biases of manyPPA practitioners is disguised by the populist rheto-ric of PRA discourse, a disguise not easily availableto researchers using more conventional quantitativetechniques. Yet to deny a role to participatory meth-ods of poverty assessment and policy design carriesthe danger of reinforcing the 'mutedness' of womenwithin their communities that Jackson is pointingout and of denying them 'voice' a second tïmearound. Participatory approaches which require 'us'to listen to 'them', and are informed by sensitivity

to the different forms that gender power and inequal-ity takes in different contexts, are critical in chal-lenging the assumptions, preconceptions and biaseswhich are part of all our cultural and disciplinarybaggage, whether the 'we' in question is the femi-nist researcher or the neo-classical economist. Theyallow us to analyse the 'choices' that women make,the meaning of these choices and the extent to whichthey are a product of agency or the denial of agencyto women within their households and communi-ties. When denial of agency is entailed, they allowus to explore the extent to which such denial isthe product of internalized ideologies or externalconstraints and hence what the priorities of policyintervention should be.

Allowing women to speak on their own behalf abouttheir own priorities and lives will not only help torescue them from the position of eternal and mutedvictims, but also has very practical implications. InGoetz and Sen Gupta's (1994) study of rural creditprogrammes for poor women in Bangladesh, it wasfound that around 60 per cent of loans given towomen were used by male members of the house-hold. If, as the authors assume, this was a matter ofmale predation, programme priority must be tostrengthen women's control over their loans. 1f, onthe other hand, this finding was a case of rationalchoice given the lack of access of women to pro-ductive sectors of local markets, then more atten-tion may have to be given to dismantling barriers towomen's participation in the market. As anotherexample (Kabeer and Murthy forthcoming), twoevaluations of the Tamil Nadu Women's Develop-ment Programme noted that men's savings wereoften used to meet the programme requirement of aminimum amount of savings before credit wouldbe given. However, different interpretations weregiven to this finding. One evaluation suggested thatwomen were usïng men's savings to speed up accessto programme credit, while the other suggested thatmen were using women to gain access to programmeloans. Again, programme responses are likely tobe very different if men were using women to gainaccess to loans (measures to ensure that womenbenefited from loan use?) or if women were usingmen to speed up their access to loans (dispensingwith the minimum savings requirement and usingother means to improve women's financial manage-ment skills?).

19

8 ConclusionThe gist of the discussion in this article has beenthat while conventional measures of income andwell-being are separately capable of capturing malepoverty because they are premised on the notion ofthe male actor and of the male experience of pov-erty, they are inconclusive about women's experi-ence of poverty The form in which women's pov-erty manifests itself depends on cultural context farmore than it does for men, suggesting that it cannotbe understood through the same conceptual lensas men. In order to get a preliminary handle on thegender dimensions of poverty, it. is necessary to un-derstand how discrimination looks in the contextof scarcity in different societies. But it also meansbeing aware of our own cultural biases in the inter-pretation of well-being and agency lt is here thatmore participatory assessments of poverty and thedesign of anti-poverty strategies may provide acomplementary or an alternative approach.

However, despite the complexities of collecting,interpreting and comparing data on the genderdimensions of poverty in different contexts, itsbroader causes are relatively clear cut. Women aregenerally poorer than men because they lack therange of endowments and exchange entitlementswhich male members of their households tend toenjoy They are less able than men to translatelabour into income, income into choice and choiceinto personal well-being. Policy makers may wishto pay attention to the gender dimensions of pov-erty because of the inefficient use of human resourcesthat it entails or because of the welfare consequencesfor women and children. However, a distinctivecontribution of a gender analysis of poverty andwell-being draws attention to the fact that holdingconstant for income/consumption does not removeall evidence of gender inequality Such inequalities- in life expectancy, physical well-being, educationand skills - reflect a different set of ideological andmaterial practices to those which create wealthand scarcity

In terms of policy therefore a gender analysis wouldsuggest a two-pronged approach. The first entails amore gender-sensitive interpretation of currentpoverty agendas which combine market incentivesand social investments. lt is clear from the empiri-cal evidence that time and energy constraints are

particularly binding for poorer women because ofthe conflicting demands of making a living andcaring for the family At the same time, labour poweris the single most important resource at the disposalof poor women - and men. Labour-intensive growthstrategies will only benefit poor women if effortsare made to address their disadvantages in the divi-sion of labour at home and in the market place.Making labour markets more accessible to womenis likely to have greater transformatory potential fortheir position within the family as well as for creat-ing sustainable livelihoods. Social expenditurespending can play a complementary role to labourintensive growth strategies in enhancing poorerwomen's access to, and ability to make choices about,labour market opportunities. Investment in thehuman capital of poor women will however nottranslate into greater access to market opportuni-ties unless the factors which hamper this access andchoice are also addressed. This means that content,design and delivery of social investments must begeared not only to enhance the human capital ofall poor people but also help to reduce some of

ReferencesCaldwell, J.c.' 1982, The Theory of Fertility

Decline, London: Academy Press

collier, ., Radhwan, S., and Wange, S., 1986, Laborand Poverty in Rural Tanzania, Oxford: OxfordUniversity Press

Desai, J., 1995, 'Vietnam through the lens of gender,Hanoi, Vietnam: UNDP

Deolalikar, A.B., 1992, 'Intrahousehold allocation ofhealth inputs and distribution of health outcomesamong Indonesian children', Tenth AnniversaryPublication, Economic Development Institute, WorldBank, Washington

Dreze, J., and Sen, AK, 1995, India: EconomicDevelopment and Social Opportunity Oxford.Oxford University Press

Dyson, T.,and Moore,M., 1983, 'On kinship structure,female autonomy and demographic behaviour inIndia', Population and Development Review, Vol9No 1: 35-60

BIson, D., 1992, 'Public action, poverty anddevelopment: a gender-aware analysis' paperpresented at the Seminar on Women in ExtremePoverty, UN Division for the Advancement ofWomen, Vienna

20

women's domestic labour overheads so that they arefree to participate in new and expanding markets,

However, what we have also highlighted in thearticle are the overarching structures of genderinequality which set the parameters within whichall women make - or do not make - choices regard-less of household income. In other words, not allwomen are poor and not all poor people are women,but all women suffer from discrimination. Thesecond prong of a gender equitable approach todevelopment therefore requires action whichaddresses the structural constraints that pervade andlimit all women's opportunity sets; action whichaddresses gender inequalities in all its various formsas a matter of basic human rights. In this the con-tribution of the international women's movementin campaigning to make women's rights recognizedas human rights should be seen, not as the concernsof a vociferous but sectional interest group, but asa critical and indispensable contribution to gooddevelopment and for some, a matter of life and death.

Elson, D., and Pearson, R., 1981, 'The subordinationof women and the internationalization of factoryproduction' in K.Young, c.wolkowitz and R.Mccullagh (eds), Of Marriage and the Market,London: CSE Books

Goetz, A., and Sen Gupta, R., 1994, 'Who takes thecredit: gender, power and control over loan use inrural credit programmes in Bangladesh', IDSWorking Paper No 8, Brighton: IDS

I-Iaddad,L.,andKanbur,R., 1990, 'Howserious is theneglect of intra-household inequality?', EconomicJournal Vol 100: 866-881

Haddad, L., Kanbur, R., and Bonis, H., 1992, 'Intra-household inequality and average household well-being: evidence on calorie intakes and energyexpenditures from the Philippines', paper presentedat the Institut National de la Recherche AgronomiqueConference on Household Economy and Modellingthe family farm in Agriculture, Montpelier, 1991

Harriss, B., 1986, 'The intra-family distribution ofhunger in South Asia' in j. Dreze and AK. Sen (eds),The Polítical Economy of Hunger in South Asia,Oxford: Clarendon Press

Jackson, C., 1995, 'Rescuing gender from the povertytrap', Gender Analysis in Development Series,University of East AngLa

John, M., and Lalitha, K,, 1995, 'Background reporton gender issues in India', Bridge report preparedby the Overseas Development Administration

Kabeer, N., 1989, 'Monitoring poverty as if gendermattered: a methodology for rural Bangladesh', IDSDiscussion Paper No 255, Brighton: lOS

Kabeer, N., and Murthy, RK., (forthcoming),'Compensating for institutional exclusion? Lessonsfrom Indian government and non-government creditprogrammes for the poor', IDS Discussion Paper,Brighton: lOS

Kapadia, K., 1992, 'Every blade of grass: landlesswomen labourers, production and reproductionin South India', Indian Journal of LabourEconomics, Vol 35 No 3: 266-276

Lipton, M., 1983, 'Labor and poverty', World BankStaff Working Paper No 616, Washington: WorldBank

-, 1994, 'Growing points in poverty research:labour issues', International Institute for LabourStudies Discussion Paper No 66, Geneva: IlLS

Lloyd, C., and Brandon, A.j., 1991, 'Women's role inmaintaining households: poverty and genderinequality in Ghana', Population CouncilWorking Paper No 25, New York: PopulationCouncil

Momsen and Townsend, 1987, Geography of Genderin the Third World, Albany: State University ofNew York P, London: Hutchinson

Morris, M.D., 1979, Measuring the Condition ofthe World's Poor: the Physical Quality of LifeIndex, Oxford: Pergamon Press

Pitt, MM., Rosenzweig, MR., and Hassan, MN,,1990, 'Productivity, health and inequality in theintra-household distribution of food in low-income countries', American Economic ReviewDecember: 1139-1157

21

Roberts, P, 1988, 'Rural women's access to labour inWest Africa' in S.Stichter and J. Parpart (eds),Patriarchy and Class: African Women in theHome and Workforce Boulder, Colorado:Westview Press

Sen, AK., 1990a, 'Gender and co-operative conflicts'in 1.Tinker (cd.), Persistent Inequalities, Oxford:Oxford University Press

1990b, 'More than 100 million women aremissing', The New York Review Dec 20th: 6 1-66

Sen, G., and Sen, C., 1985, 'Women's domestic workand economic activity', Economic and PoliticalWeekly Vol 20 No 17: 49-56

Sen, 1., 1988, 'Class and gender in work time allocation',Economic and Political Weekly Vol XXIII No 33:1702-1706

Senauer, B., Garcia, M., and Jacinto, E., 1988,'Determinants of the intra-household allocation offood in rural Philippines', American Journal ofAgricultural Economics Vol 68 No 4

Sender, J., and Smith, 5., 1990, Poverty, Class andGender in Rural Africa, London: Routledge

Stewart, F., 1985, Planning to Meet Basic Needs,London: Macmillan Press

Svedberg, P, 1990, 'Undernutrition in sub-SaharanAfrica: is there a gender bias?' Journal ofDevelopment Studies Vol. 26 No 3: 469-486

UNDO, 1990, Human Development Report,Oxford: Oxford University Press

1995, Human Development Report, Oxford:Oxford University Press

United Nations, 1991, The Worlth Women: Trendsand Statistics 1970-1990, New York: UN