challenge of poverty

FEATURE ARTICLES

FemaleHeadshipand Poverty

Gaps inMaternalMortality

Poverty Trapsin South Africa

IN REVIEW

Improving Water and SanitationSlums and the Urban Poor

International Poverty Centre

United Nations Development Programme May 2004

TheChallengeof Poverty

UNDP International Poverty Centre In Focus May 2004 2

According to an oft-quoted source, 1.2 billion people live in extreme poverty around the globe, surviving on less than US$1 a day. This figure speaks volumes about a

world that unabashedly spends trillions of dollars on conspicuous consumption anddefense, yet is challenged to spend a fraction of those resources to provide dignifiedliving conditions to so many of its impoverished citizens.

Ironically, the ‘1.2 billion’ figure may say a great deal about the state of the world, butit tells very little about poverty itself. Compressed into a single figure, the complex anddistinct circumstances experienced by those who find themselves in poverty — theirvaried trajectories into it, the dimensions in which they are deprived and the reasonsfor their predicament — essentially become concealed behind a common label.

We need to move beyond aggregation, snapshots and easy labels to improve ourunderstanding of poverty. This month, In Focus explores these issues by highlightingsome of the shortcomings of conventional analyses of poverty and the complexitiesinvolved in conceptualizing it in a more meaningful way.

We open with a provocative article that probes into the existence of an alleged closelink between household headship and the ‘feminization of poverty’. Female-headedhouseholds clearly are disadvantaged in many countries. But the inference that well-being is always enhanced in two-parent — that is, male-headed — families is wrong.It is largely informed by the narrow way in which well-being is defined and statisticsgathered in conventional poverty research, which also ignores the importantdimension of agency that allows women heads to offset gender discriminationas well as the foregone income from a male partner.

Next, we present a piece that sheds light on the extent to which risks for maternalmortality are unequally distributed among women from different socioeconomicbackgrounds. While not surprising, this finding is not what previous studies haveconclusively established. Done at a high level of aggregation, they typically show aweak relation between crude economic and mortality indicators for whole populations.But an examination of household level data reveals that poverty and maternalmortality are strongly associated, highlighting the need to move beyond countryaverages when assessing development performance.

Our third article goes deeper into the analysis of poverty, with post-apartheid SouthAfrica as the backdrop. It illustrates the limitations of static analyses in depictinghouseholds that are trapped in or prone to fall into poverty, calling instead for a greaterfocus on people’s ability to build and use assets to escape from it. The article goes onto argue that entrenched inequality and persistent poverty have a dampening effecton growth, thus providing a welcome contribution to contemporaneous discourseon the relationship between growth, poverty and inequality in developing countries.

We end by reviewing two recent publications by UN agencies. The first deals with theso-called urbanization of poverty and its most visible expression — the urban slumsthat punctuate the landscape of cities in developing countries. Solving the ‘problemof slums’ requires transcending the standard housing-and-services approach as wellas having plenty of foresight given predicted population trends in the decades ahead.The second publication makes a strong case for increasing investments in basic waterand sanitation to meet development targets adopted internationally. Such investmentswill yield large health, productivity and income gains to the poor families and resource-scarce economies of the developing world.

Together, the articles contained in this issue reflect a call for moving beyond simpledesignations and engaging in analysis that at once allows for a more dynamic,disaggregated and contextual understanding of poverty.

In Focus is an online bulletin of the UNDPInternational Poverty Centre (IPC). Its purposeis to present succinctly the results of recentresearch on poverty and inequality in thedeveloping world.

Based in Brazil, IPC joins the Oslo GovernanceCentre (Norway) and the Drylands DevelopmentCentre (Kenya) as the newest global thematicfacility of UNDP. IPC is designed to facilitate,promote, and disseminate knowledge andexperiences that may lead to tangibleimprovements in the lives of poor people indeveloping countries. A basic goal of IPC is toimprove global understanding about the causesof poverty and inequality – a first step fordevising policies and programs to tackle them.IPC’s activities are aligned with UNDP’s keyobjective of supporting the MillenniumDevelopment Goals, as unanimously adoptedduring the United Nations Millennium Summitin 2000.

IPC DirectorNanak Kakwani

EditorAlejandro Grinspun

Assistant EditorKandi Shejavali

DesignDesign GenuínoBernardo Lac, Maria de Oliveira

Front-page photoRocinha, Rio de Janeiro, 2001© Kita Pedroza

Special thanks to Astrid James (The Lancet),Joyce Boor and Ann E. Fitzmaurice (University ofAberdeen), and Vishnu Padayachee (Universityof KwaZulu-Natal).

United Nations Development ProgrammeInternational Poverty CentreSBS - Ed. BNDES, 10° Andar70076 900 Brasilia DFBrazil

www.undp.org/[email protected]

The content of this publication does notnecessarily reflect the official views of the UNDP.For comments and feedback, please [email protected]. Alejandro Grinspun

F R O M T H EE D I T O R



The term ‘feminization of poverty’has been much used, and arguablyabused, in the development lexicon inrecent years. It has become commonplaceto hear that poverty increasingly has a“woman’s face” given that femalespurportedly account for 70% of theworld’s 1.2 billion poor people.

Interestingly, the notion that womenrepresent a disproportionate andrising share of the world’s poor hasbeen attributed, in large measure, tothe growing incidence of femalehousehold headship. The alleged closelink between household headship andthe feminization of poverty is notsurprising. It is not only invoked by thegrowing numbers of women who headfamilies and are poor, but also by thewidely held belief that they are amongthe ‘poorest of the poor’.

This perception, in turn, rests uponthe notion that poverty is a majorcatalyst for the formation of female-headed households — and, moreparticularly, an almost invariableconsequence of it. Given that mostsocieties are characterized by genderinequalities in access to resources, it isnot unreasonable to suppose that thesedifferences are exacerbated when poorwomen become responsible forhousehold provisioning without malepartners. The resultant privation notonly affects these female heads butalso is allegedly perpetuated acrossgenerations.

Female headship is now so firmlyentrenched in the developmentdiscourse about gender and povertythat it has effectively become a proxyfor women’s poverty. This associationhas not been without its benefits insofaras concise slogans help secure resourcesto address the plight of poor women.

Development organizations may alsofind it appealing to have a female-onlytarget group for policies as it sparesthem the need to intervene in theconflictive ‘private’ domain of intra-household relations between two-parent families. Not surprisingly, suchexpediencies have flourished in aclimate in which targeting has becomean increasingly favored strategy inpoverty-reduction efforts.

However, branding single-parentfamilies headed by women as the‘poorest of the poor’ suggests thatpoverty is mainly determined byhousehold characteristics, rather thanthe socio-economic context in whichthey are situated. Such labeling not onlyscapegoats female headship as a causefor poverty but also deflects attentionfrom the wider structures of genderand socio-economic inequality thatcontribute to it. Placing excessiveemphasis on the economic disadvantageof female heads misrepresents anddevalues their enormous efforts toovercome gender obstacles. It alsoobliterates the personal significancethat headship has for women.

This stereotyping also suggests thatpoverty among women is confined tocases where females head their families.There is no denying that householdsheaded by women on low incomes standto benefit from institutional support forparenting and provisioning. But onemust acknowledge that women in male-headed households experience povertytoo — and may even be worse off. Thisfact is obscured when one assumes thatfemale-headed households are ofnecessity poorer than those controlledby men.

Prevailing methods of poverty researchhave undoubtedly played a role in

by Sylvia Chant,London School of Economicsand Political Science, UK

It is commonly believed thathouseholds headed bywomen are poorer than thosecontrolled by men.

But a closer look at intra-household relations, coupledwith a broader concept ofpoverty that focuses on socialdeprivation, reveals thatwomen — and indeed otherhousehold members — mayoften be better off without amale partner.

It will not be possible to devisepolicies that address thestructural bases of genderinequality without firstdebunking the myth of a closeassociation between femaleheadship and poverty.

Female Headship andthe ‘Feminizationof Poverty’


popularizing the notion of a‘feminization of poverty’ and its linkageto female household headship. Despitemounting rhetoric about the need forqualitative, participatory tools to assesspoverty, ‘money-metric’ approachescontinue to reign supreme. Theyprioritize questions of ‘physicaldeprivation’ to the detriment of ‘socialdeprivation’, which involves suchdimensions as lack of power, agencyor self-esteem. This concern has elevatedthe importance of the level of resourcesrather than focusing on who hascommand and control over them. Byoverlooking women’s power to accesshousehold resources, standard ‘money-metric’ approaches tend to over-emphasize the poverty of female headsrelative to women who reside in male-headed units.

Moreover, the household rather than theindividual persists as the dominant unitof measurement in poverty assessments.Consequently, households headed bywomen stand out in poverty statistics,even though they may not represent aslarge a proportion of poor people giventheir smaller size compared to dualparent households.

But even if one takes aggregatehousehold incomes as the welfaremeasure of choice, there is preciouslittle hard data that shows a systematiclink across space or time betweenhousehold headship and poverty.In Latin America, for example, theincidence of female headship in urbanareas — where women-headed familiesare usually more common — rose inevery single country for which data existfor 1990 and 1999. At the same time, theproportion of urban households inpoverty declined across the region from35% to 29.8%, while those below theextreme poverty line fell from 17.7%to 13.9%.

It appears, then, that the case for agreater share and depth of povertyamong female household heads restson rather tenuous grounds.

The lack of definitive links betweenfemale headship and poverty, assuggested by the ‘feminization ofpoverty’ thesis, results in part from the

heterogeneity of women-headed units.This heterogeneity can have importantmediating effects on poverty dependingon the social and cultural context.

It matters, for instance, whether awoman becomes head of a householdby ‘choice’ or involuntarily. Other sourcesof heterogeneity among female-headedunits can be linked to householdcomposition, the stage the woman isat in her lifecycle (e.g. age and relativedependency of offspring), thegeographic location of the household(rural versus urban), and whetherresources beyond the family unitare accessible.

It is not unusual for female headsto contend with discrimination, above-average work burdens and timeconstraints. But in order to offset thenegative effects of gender bias, theyoften organize their households soas to optimize the resources attheir disposal.

A common strategy is to invite co-residence by members of their extendedkin networks. Contributions from co-resident individuals as well as migrantfamily members can help compensatethe personal disadvantage femaleheads face because of their gender.Those contributions increase ahousehold’s access to resources forproductive and reproductive labor,thereby reducing its vulnerability whilealso bolstering its earning capacity.

Aside from intra-group diversity,two other factors help explain whyfemale headship does not automaticallytranslate into privation. The first relatesto intra-household power relations andthe attendant implications for resourceallocation, while the second pertains topoverty’s multi-dimensional and

subjective nature. Analysis of theseissues requires moving beyond thenarrow focus of conventional money-metric poverty assessments andincreasing scrutiny over what goeson inside households.

With respect to intra-household resourceallocation, one must eschew the idealizednotion that households are intrinsicallycohesive and internally undifferentiatedentities governed by ‘natural’ proclivitiesto benevolence, consensus and jointwelfare maximization. Indeed, a majorcontribution of feminist research hasbeen to debunk the myth of householdsas unitary entities operating on altruisticprinciples. Rejecting orthodox HouseholdEconomics thinking, many authors haveargued instead that households are ‘sites’of competing claims, rights, power,interests and resources, with domesticnegotiations frequently shaped by amember’s age, gender or position in thefamily hierarchy.

It is more difficult to shun idealizednotions of ‘female altruism’ and ‘maleegoism’. Findings from a remarkably largenumber of contexts confirm that womendevote the bulk of their earnings tohousehold expenditure, often withpositive effects on other members’nutritional intake, health care andeducation. Men, on the other hand, areprone to retain more of their earnings fordiscretionary personal expenditure. Insome instances, they even command alarger share of resources than theyactually bring home.

This implies that even if sufficientassets exist in a male-headed household,its female members (women and girls)may not be able to access them — atleast not in their own right. Alongwith reducing the resources availableto other members, irregular financialcontributions from male heads canlead to a situation of acute vulnerabilityand ‘secondary poverty’ among spousesand children. Moreover, women in male-headed families all too often sufferfrom other extreme forms ofdependence, including subjectionto authority and violence.

Accordingly, lower incomes amongfemale heads relative to their male

Stereotypes aboutwomen-headedhouseholds divertattention from the truesources of genderdisadvantage.


counterparts may be countered by theextent to which income and assets areconverted into consumption andinvestments for the benefit of the entirehousehold. Seen in this light, the absenceor loss of a male head may notnecessarily precipitate destitution andmay even enhance the economic securityand well-being of other householdmembers. Evidence from Mexico, CostaRica and the Philippines reveals thatmany low-income women feel moresecure financially without men, evenwhen their own earnings are low orprone to fluctuations. They also claimto be better able to cope with hardshipwhen they are not at the mercy of maledictat and thus freer to make decisions.

This evidence suggests that whilethe price of independence may behigh for poor women, benefits inother dimensions of their lives may beadjudged to outweigh the costs. Manywomen seem willing to trade a lowerincome for a position of greaterautonomy and self-reliance — a findingthat fits well with ‘social deprivation’thinking about poverty. Even if womenare poorer in terms of income as headsof their own household, they may feelbetter off and less vulnerable in theabsence of male control. It is difficultto grasp the importance of thesetransactions unless one incorporatespeople’s subjectivities into theconceptualization of poverty.

Indeed, viewing poverty as multi-dimensional helps explain why somelow-income women make choices that,at face value, could seem prejudicial totheir well-being. One such case is whenfemale heads refuse offers of financialsupport from absent fathers in order toevade ongoing contact or sexualencounters. Another instance is wherewomen forfeit assets such as houses andneighborhood networks in order to exitabusive relations. While financialpressures may force some women tosearch for new partners following aconjugal breakdown, it is significant thatothers choose to remain alone ratherthan return to ex-partners or form newrelationships. Those who live without amale partner often do so by choice,preferring to rely on sons or other familymembers rather than spouses.

In the end, the ‘feminization of poverty’thesis and its overriding preoccupationwith both income and women householdheads is dangerous for two reasons.First, it precludes an analyticalconsideration of the social dimensionsof gender and poverty. Second, suchpreoccupation tends to translate intosingle-issue, single-group policyinterventions with little power todestabilize the deeply embeddedstructures of gender inequality inthe home, the labor market andother institutions.

Instead, explicit policies are requiredto redress the processes that makewomen poorer than men. The mostimportant ones are gender-awarepoverty interventions that do not justtarget women in isolation or focusmainly on those who head theirown households.

We must recognize that, in most societies,women clearly lag behind men in termsof accessing the resources necessaryfor survival and self-determination.Yet in accepting the alleged ‘feminization’of privation, one also is implying thatthere is a counterpart process of‘masculinization’ of power, privilegeand material accumulation.

If this is the case, then future researchneeds to explore how that processsquares with the fact that, bothcollectively and individually, womenappear to be forging new spaces inmany countries. This is occurring notonly in politics, the law and the labormarket but even in their domesticenvironments, where women arearguably exerting more and moreinfluence over their own lives.

Sylvia Chant, New Contributions to theAnalysis of Poverty: Methodological andConceptual Challenges to UnderstandingPoverty from a Gender Perspective,ECLAC, 2003.

What matters to womenis who commands andcontrols householdresources.


Reducing the burden of mortalityand morbidity that affects poor people iswidely regarded as one of the foremostdevelopment challenges today. For manyinternational organizations and nationalgovernments, improving poor people’shealth has become a key policy goal.As a result, development assistance aswell as the strategies devised by manylow-income countries now commonlyprioritize interventions based on theiranticipated effect on poverty outcomes,including health. This focus stems froma growing recognition of the two-wayrelation between poverty and disease,and the importance to developmentof equity in health.

Debate continues, however, over theappropriate markers, measures anddata sources for monitoring progresstowards poverty-related goals. Maternalmortality has long been used as a markerof the health of populations. It is alsoseen as a litmus test of women’s statusand a barometer of the functioning ofa country’s health system. Its relevanceas a marker of health has receivedexplicit recognition in the UN MillenniumDeclaration, which set the target ofreducing maternal mortality ratesby 75% by 2015.

The problem is that tracking maternalhealth outcomes presents majoroperational challenges owing tomeasurement complexities and weakroutine health information systems inmost developing nations. While it isgenerally expected that maternalmortality will be highest among the poor,the majority of countries lack the data toconfirm this assumption beyond thecorrelation of crude indicators for wholepopulations. Yet existing analyses ofmaternal mortality ratios and economicmarkers such as per capita GNP reveal aweak correlation with significant outliers.

For this reason, assessments ofsocioeconomic differentials at thesub-national level often rely on proxyindicators such as uptake of antenatalcare or delivery with health professionals.These indicators typically show majorgaps between rich and poor women,with as high as 18-fold differences inuptake between the top 20% and bottom20% of a country’s population. Theexistence of such gaps highlights thedangers of monitoring developmenttargets only in terms of nationalaverages. Unfortunately, evidenceof the link between maternal mortalityand poverty is lacking at the level ofindividuals. But at the same time, surveydata already exist for many countriesthat make it possible to examine gapsin maternal mortality between a nation’srich and poor citizens.

Drawing from demographic and healthsurveys from ten developing countries(Burkina Faso, Chad, Ethiopia, Indonesia,Kenya, Mali, Nepal, Peru, the Philippines,and Tanzania), we have createdpercentage distributions of womenaccording to their poverty and survivalstatus. Each woman was assigned to oneof three survival categories reported inthe surveys: alive and aged 15-49 years,dying from non-maternal causes, anddying from maternal causes. The lattercategory includes all deaths to womenthat took place during pregnancy,delivery or in the six weeks after theend of pregnancy.

In order to estimate a woman’s povertystatus, we used proxy indicators suchas educational level, source of drinkingwater as well as the type of toiletfacilities and floor in their dwellings.Following a routine technique in surveyanalysis, we extrapolated responsesfrom some household members (theadult female respondents in the surveys)

by Wendy J. Graham,University of Aberdeen,

Scotland

With the use of householdsurveys, it is possible toexplore the link betweenpoverty and maternalmortality at the level ofindividuals rather thanwhole populations.

This type of analysis revealslarge disparities according towealth, with deaths frommaternal causes stronglyassociated with variouspoverty attributes across awide range of countries.

Health as well as otherdevelopment policies need tobe gauged based on theirpotential for reducingexisting gaps in outcomesbetween a nation’s rich andpoor citizens.

Exploring the LinksBetween MaternalDeath and Poverty


to others (their adult sisters), whosepoverty attributes were assumed to bethe same as those of their respondentsiblings. All of our surveys had a sampleof more than 10,000 adult sisters.Together, they represent a highly diverseset of countries, both in terms of GDP,HDI values and estimated levels ofmaternal mortality, ranging from 240to 1,800 maternal deaths per 100,000live births.

Despite the contrasts between thecountries selected, we found significantsimilarities in the associations betweenwomen’s survival and poverty statusacross countries. There is a gradientwithin and across survival categoriesso that the proportion of women dyingfrom maternal and non-maternal causestends to increase with poverty. Suchincreases are particularly consistentin the case of maternal deaths.

For example, Indonesian women whodied during pregnancy, delivery orpuerperium were more than twice aslikely to be unschooled than those whowere still alive at the time of the survey

(1997). They were also about 50% morelikely to rely on water unfit for drinking(open or unprotected water source), andhad a 40% greater probability of living indwellings with a dirt or bamboo floor ascompared with women dying from non-maternal causes or still alive.

Further analysis of Indonesian data for1994 and 1997 reveals large inequalitiesin survival status across households. Forboth years, around one-third of allmaternal deaths were in women from thepoorest 20% of the population, ascompared with fewer than 13% amongthe wealthiest 20%. In terms of crude riskratios, the probability of maternal deathwas three to four times greater in thepoorest than the richest group. Similarly,the proportion of pregnancy-relateddeaths among women aged 15-49 yearswas 25%-29% in the bottom quintileversus 10%-17% in the top.

The same poverty gradient was alsoapparent in the data sets for the othercountries in our study. As in Indonesia,data from Tanzania (1996) and thePhilippines (1998) show an increasing

Demographic andhealth surveys

Since 1984, when the first Demographic and Health Survey (DHS) was carried out, 174

surveys have been implemented in 71 developing or transitional countries. With

sample sizes ranging from about 3,000 to almost 100,000 households, the DHS use a

standardized questionnaire that collects information on a wide variety of

demographic, socioeconomic, health, nutrition, and health-service use variables

focusing on reproductive, maternal, and child health.

Typically, poverty status is captured in terms of assets or wealth, rather than income or

consumption. Questions are asked of the household head about ownership of

consumer items such as a radio or car, as well as characteristics of the dwelling such as

floor or roof type, toilet facilities, and water source. In addition, information on

relevant individual level characteristics (e.g. education and literacy) is collected for all

household members. For maternal mortality, data are gathered by interviewing adults

(usually, women of reproductive age) about survival of their sisters in order to yield a

retrospective estimate of the maternal mortality ratio as well as the lifetime risk by

establishing whether any sisters died during pregnancy, delivery, or the six weeks

after the end of pregnancy.

Since many DHS surveys include questions on both maternal mortality and poverty

status, they provide a unique opportunity to explore the relation between the two.

Moreover, several countries have had repeated surveys, thereby allowing time trends

to be assessed.

As long as developmentgoals are expressed aspopulation averages,disparities between richand poor may persisteven if targets are met.

risk of maternal death among womenfrom the poorest than the richestquintile. Poorer women in thesecountries were at least two to over threetimes more likely to die from pregnancy-related causes than were wealthierwomen. The proportion of deathsreported as maternal was also muchhigher among women from the bottomquintile in both countries: 39% versus22% in Tanzania, and 32% versus 12%in the Philippines.

The concentration of maternal deathsamong women from the pooresthouseholds is noteworthy. In general,women dying from maternal causeswere more likely to have had noformal education and worse dwellingconditions than those who had diedfrom non-maternal causes or werestill alive. While the strength of theassociations is especially striking foreducation, the risk of maternal deathalso is clearly related to the otherpoverty attributes reported inthe surveys.

Maternal mortality, moreover,seems to be a sensitive markerof disadvantage since non-maternaldeaths generally do not show suchextreme clustering in the poorestgroups. This finding is perhaps notsurprising in view of the unpredictablenature of obstetric complications andthe substantial costs of treatmentsto avert death, which are oftenunaffordable for poor families.The higher risks of maternal death inpoor women do not necessarily implythat they are more prone to pregnancy-related complications. Rather, thoserisks might be a reflection of thelower probability such women haveof receiving adequate treatmentbefore, during and after childbirth.


Our analysis relies on a numberof assumptions. First, the povertyattributes of our target group are inferredfrom the survey responses of theiradult female siblings, who are takento be representative of all women.By extrapolating data from a respondentto her siblings, we are assuming thatpoverty is common to families ratherthan being a personal attribute. Wefurther assume that the poverty statusof the adult respondents has not changedsince the reported deaths occurred.

Moreover, respondents are deemed ableto distinguish the time of death of theirsisters, specifically whether or not theyhappened while pregnant, during deliveryor in the six-week period followingchildbirth. This is particularly relevantgiven that demographic and healthsurveys have unrestricted referenceperiods, meaning that some women willhave died several years before theresponses captured through the survey.

Despite these caveats, there arestill definitive advantages to usingsurvey data to explore the synergybetween poverty and a health outcomesuch as maternal mortality. Mostimportantly, the data for both povertyand survival status derive from the samesource, thus bypassing the difficulties ofvarying quality, sampling frames andreference periods across surveys. Therequired data are already available formany low-income countries, and a similaranalysis could be extended to otheroutcomes (e.g. adult male mortality) aswell as process measures (e.g. use ofhealth services). Apart from being simple,the analysis is done at the individualrather than aggregate level, thusavoiding the ecological fallacy.

The discrepancy in the risk of maternaldeath between the poorest and richestcountries is often cited as a globalinjustice. Much less attention has beenpaid to differentials within countries.Yet, as long as health policy goalsremain expressed as populationaverages, disparities between poorand rich people could persist evenif the targets set by policy-makersare met.

Policies thus need to be assessed basednot just on their cost-effectiveness, butalso their contribution to povertyalleviation and equity enhancement.Better information is needed to trackgains to the poor — includingimprovements in health — bymaking more efficient use of existingdata while strengthening the arrayof methods and sources to monitordevelopment outcomes likematernal mortality.

Our findings have shown the magnitudeof the current gaps in maternal mortalitybetween a country’s rich and poorwomen. They should thus providea stimulus to setting and monitoringother development goals in accordancewith their potential for addressingsocio-economic inequalities.

Wendy J. Graham, Ann E. Fitzmaurice,Jacqueline S. Bell and John A. Cairns,“The familial technique for linking maternaldeath with poverty”, The Lancet (363),January 2004.


Despite its relative success in termsof macroeconomic stabilization, SouthAfrica’s post-apartheid economic policyhas mostly failed to promote sufficientgrowth, job creation and povertyreduction. While a strict monetary policyregime and high real interest rates havekept inflation in single figures, growthrates have not come near the estimated7%-8% per annum required to absorbnew entrants into the labor market andmake inroads into the large pool ofunemployed people. As a result,unemployment has kept rising, andboth poverty and inequality haveremained high and may have evenincreased during this period.

This disappointing record has takenplace against the backdrop ofapartheid’s dual legacy of raciallyembedded inequality and highlysegmented factor markets from whichpoor people are largely excluded. At theend of apartheid, almost half of SouthAfrica’s population was categorized aspoor using a national poverty line, andone-fifth was earning less than US$1 aday. Just 6% of the population capturedmore than 40% of national income andover 60% of Africans were poorcompared to only 1% of whites, makingSouth Africa one of the world’s mostunequal societies.

While there is much attention paid tothe impact of growth on poverty andinequality, we would like to turn thecausality around and argue that SouthAfrica’s sluggish growth may be rootedin its high and persistent inequality.Highly unequal income and wealthdistributions become economicallycostly and growth-reducing when largenumbers of a country’s citizens aremired in poverty over time, unable tosave, invest and progress. If this is thecase, then South Africa’s stunningly high

Accumulation Failuresand Poverty Traps inSouth Africa

South Africa’s weak economicrecord in the post-apartheidera may be rooted in its legacyof stunningly high, raciallyembedded inequality that hastrapped large numbers of itscitizens in poverty.

By lowering people’sexpectations of incomemobility and assetaccumulation, persistentpoverty can become a seriousimpediment for growth.

An agenda emphasizingmicro reforms is required toincrease the economicprospects of South Africa’spoor majority. The reformswill need to improve marketaccess and rewards for thepoor, while also lifting thefinancial and time constraintsthat prevent them fromaccumulating sufficient assetsto escape poverty.

by Julian May,University of KwaZulu-Natal,South Africa

inequality may partly explain its weakpost-apartheid economic record.

Indeed, panel data collected in KwaZulu-Natal shows evidence of widespreadpoverty traps in the province despite itsnot being among the country’s poorest.Expenditure-based headcount povertyrates jumped from 27% to 43% between1993 and 1998. In turn, the averageincome shortfall of the poor increasedfrom 27% to 33%, while the severity ofpoverty rose sharply.

While disturbing, these data do notnecessarily indicate a clear failure inaccumulation that traps people intopoverty. The reported trends could havebeen due to the presence of significantupward and downward mobility justabove and below the poverty thresholdso that initially poor and non-poorhouseholds simply swapped places inthe income distribution.

Clearly, a society in which some peopleare trapped in poverty over time is verydifferent from one where householdssuffer transitory spells from which theycan expect to escape periodically. Inorder to capture the differencesbetween these two scenarios, one needsto move beyond static depictions ofpoverty that often understate the extentof vulnerability to being poor at somepoint in time.

By looking at the households below thepoverty threshold in only one or bothwaves of our panel survey, we were ableto distinguish the transitory from thechronic poor. One can see now that only48% of sampled households were notpoor either in 1993 or 1998, as comparedwith 30% who were poor in at least oneyear and 22% who were poor in both.Despite the presence of considerablemovement into and out of poverty, it


Static depictionsof poverty tend tounderestimate people’svulnerability to beingpoor at some pointin time.

Poverty measures in KwaZulu-Natal,1993 and 1998

1993 1998

Poverty headcountPoor 26.8 42.5Indigent 3.3 10.0

Depth of poverty*Poor 27.1 33.0Indigent 30.4 4.9

Severity of poverty**Poor 0.03 0.063Indigent 0.005 0.0004

Gini coefficient 0.38 0.42

* Average income shortfall as % of poverty line.** Foster-Greer-Thorbecke-2.

meet the subsistence requirementsof incremental units of labor. Claimson other economic or social assets thusbecome necessary to lift a family abovethe poverty line. Secondly, financialconstraints limit the ability of poorpeople to utilize effectively theproductive assets and endowmentsthey do possess, such as land. Andthirdly, the burden of meeting basicneeds like water and fuel wood creates‘time poverty’, which further constrainsa household’s ability to employ theresources at its disposal. Indeed,for a given work capacity, householdreproduction time sharply diminishesits pool of uneducated labor powerand, therefore, its ability to generatean income.

It would seem that poverty not onlyreflects the lack of sufficient assetsto which the extant economy payssignificant returns, but also the presenceof tight constraints on the effective useof those assets. It is a case of multiplemarket failures, in which wageopportunities are weak and ancillaryfactor markets are not working well.

As a result, people remain poorbecause they cannot borrow againstfuture earnings to invest in inputsor accumulate assets for production,including education. They are unableor unwilling to engage in entrepreneurialactivities because the costs of failureare too high. They cannot insurethemselves against risks and lackinformation about market opportunities.And they are deprived of many publicgoods such as property rights, publicsafety and infrastructure, therebyincurring high direct and time costswhen trying to obtain them.

This may lead to a dilemma in whichpeople living in communities with

thus appears that a large number ofhouseholds were unable to escapepoverty at all during the time thatelapsed between surveys.

Further analysis reveals that a largeproportion of initially poor householdshas either held steady or fallen behind,while upwardly mobile households aremostly among those that had beenbetter off from the outset.

In order to explore the reasons behindthese movements, we reclassifiedhouseholds based on their possessionof assets and the returns they obtainfrom their use. This allows us to breakdown the categories of people belowthe poverty line to observe whethertheir situation improved or worsenedduring the five-year period betweenthe two surveys. The analysis seeksto distinguish random changes inwell-being from those that reflectsuccessful asset accumulation bya poor household.

What this exercise shows is that only4% of households were able to getahead during this period, but they stillremained poor. A much larger proportion(21%) fell behind, meaning their abilityto generate an income declined between1993 and 1998. Another 16% of thesample appears to be trapped in poverty.Unable to accumulate assets forenhancing their well-being, thesehouseholds are not improving theirsituation over time in order eventuallyto escape poverty. Finally, 11% of thehouseholds are found to be poordespite having the assets necessary toavoid poverty. These are the stochasticpoor, who may be said to have sufferedan entitlement shock from which theyshould be able to recover.

t is clear, then, that focusing on thelikelihood of being poor, rather thanon observed levels of poverty, permitsa richer understanding of changesin well-being.

More detailed scrutiny of our data setreveals three dimensions that correlatewith persistent poverty in South Africa.First, the marginal returns to uneducatedlabor are positive but so low that theyfail to generate an income sufficient to


equivalent current poverty levels butdifferent future expectations behavein dissimilar ways — politically, sociallyand economically.

Persistent poverty shapes people’sexpectations of income mobilityand could lead to risk-minimizingbehavior and potentially undesirableasset accumulation paths. Those whomay be able to pull ahead and escapethis poverty trap are only a small —and perhaps declining — portion ofthe population. If this trend persistsinto the future, it may lock SouthAfrica into a situation of increasedpoverty and sluggish aggregategrowth as large numbers of familiesfind themselves unable to make theinvestments needed to realize theirown and their children’s potential.

Indeed, if sluggish growth is partlyrooted in the numbers of peopletrapped with little prospects of incomemobility or asset accumulation, then afundamental rethinking of South Africa’seconomic strategy may be required.Particularly in the rural areas, fewpeople will be able to prosper aslong as their ability to utilize productiveassets and resources effectively remainshindered. To lift these constraints,policies will need to move beyondmacroeconomic stabilization into anagenda centered on “microeconomicreform” with support from an enablingand complementary macroeconomicpolicy framework.

Microeconomic reforms shouldseek to eliminate the conditionsthat create poverty traps, both throughinterventions targeting people with

Without a fundamentalrethinking of its presenteconomic strategy,South Africa maybecome locked intoa low growth-highpoverty trap.

limited or no assets and by improvingmarket access, information and rewardsfor the rural poor. Measures such as landreform, infrastructure development andfinancial services are examples ofpolicies that could help improve accessto productive assets and release thefinancial and time constraints faced bypoor households. In the absence of suchpolicies, widespread market failurereflected by missing, thin or distortedmarkets will keep on reproducing theossified socioeconomic structureinherited from apartheid.

The need for micro reforms has begunto enter the South African policy debate.But the development of these policies islimited by an inadequate knowledgebase. The linkages between macro-economic and microeconomic policieshave received little attention. At thesame time, there is not enoughinformation about the determinants ofthe behavior of poor producers andconsumers or the operation of marketsin which they participate.

These are areas where more research willbe essential to inform policy if SouthAfrica is to get ahead on a moresustained and equitable path.

Michelle Adato, Michael Carter, Julian Mayand Phakama Mhlongo, “Persistent povertyin South Africa: Severity, sources andsolutions”, 2003.


Slums have existed for many centuries, but their growth in recent decades hasbeen explosive. Today, almost one billion people live in overcrowded and un-servicedslums, often situated on marginal or dangerous land. As the world becomes moreurban, the challenge of ensuring adequate shelter and services for a growing numberof low-income households will become daunting in the absence of concerted action bymunicipal and national authorities, civil society and the international community.

In the first global assessment of its kind, UN Habitat calls for stepping up efforts toarrest the growth of informal settlements and improve the lives of slum dwellers inthe cities of the developing world. The report also lays the foundations for reachingthe ambitious goal of ‘cities without slums’.

Meeting this challenge will not be easy. Already, slum dwellers comprise nearly one-third of the world’s urban population, ranging from 6% in the industrialized countriesto 78% among the least developed. In no developing region did the pace of slumformation slow down during the 1990s, turning slums into a prominent feature of mostcities. Demographic changes will bring added pressure in the coming decades. As theworld’s rural population reaches its peak, urban settlements will absorb the bulk of arapidly expanding population at a rate that will far outstrip their capacity to providesufficient jobs, shelter and services. These developments will gradually shift the locusof global poverty to cities, where the number of slum dwellers may reach about twobillion people over the next 30 years.

Slums are a physical and spatial manifestation of urban poverty and intra-cityinequality. They result from rapid urbanization and the failure of housing and jobmarkets to accommodate the needs of a growing pool of low-income earners.Consequently, slum areas have high concentrations of socio-economic and physicaldeprivation, including substandard housing in hazardous locations, inadequate accessto basic amenities and infrastructure, and unhealthy living conditions.

Poor sanitation and indoor air quality, lack of waste disposal facilities and the presenceof vermin compromise the health of slum communities, leading to chronic illness andtruncated lives. Thus, urban mortality rates may generally be lower than in rural areasbut not among slum dwellers, who suffer disproportionately from AIDS, tuberculosisand water-borne diseases such as typhoid and cholera. To make matters worse, slumsare often recipients of the city’s nuisances, including industrial effluent and noxiouswaste, which are deposited in the only land the poor can afford because no one elsewants it.

Slum dwellers’ reliance on informal land and housing markets for shelter also resultsin a high degree of tenure insecurity and a constant threat of forced eviction. Insecuretenure not only undermines community cohesion but also limits access to urbanservices and the ability to build up assets to escape poverty. Not surprisingly,connections to infrastructure in informal settlements are, on average, only abouthalf the level in better-off parts of developing-country cities. Differences in serviceprovision may even widen under current strategies of fiscal decentralization andprivatization, since slum residents can seldom pay for services or mobilize effectivelyto attract resources to their neighborhoods.

The Challengeof Slums

I N R E V I E W

Combining several physicaland socioeconomicdimensions such as location,population density, quality ofhousing, tenure security andaccess to urban services andinfrastructure, a global reportby UN Habitat attempts a newdefinition of ‘slums’ and thenestimates the number of slumdwellers worldwide.

The estimates provide abaseline for monitoringprogress towards theMillennium DevelopmentGoals (MDGs), which call for asignificant improvement inthe lives of “at least 100million slum dwellers by theyear 2020”. They also are awake-up call to policymakersabout the urgent need toaddress the increasingurbanization of poverty.



Insecure as they may be, informal shelter arrangements are usually the only choicefor the urban poor. Due to stigmatization and geographic isolation, they are rarelyable to obtain formal sector jobs or access regular sources of finance to develop theirown businesses. Most earn their living from low-paying occupations in the informaleconomy, including a host of street- and home-based enterprises. A response to thefailure of formal job markets to absorb new entrants, the informal economy nowaccounts for well over one-third of the urban work force in the developing world andis expected to provide more than 90% of additional jobs in urban areas in the nextdecade. Despite being an important source of livelihoods for many people, informalemployment can easily lead to dead-end jobs in exploitative working environments.

At the same time, one must beware of stereotypes that equate slums with poverty. Notall slum dwellers are poor, and many of the urban poor live outside them. Slums mayactually perform numerous positive functions for their residents. Apart from providingcheap accommodation, slum housing can be used for profit, either as a source of rentalincome or a location for family-run businesses that deliver low-cost goods and servicesto urban consumers. Slums also serve as support networks for new migrants to the cityand, under certain conditions, a springboard for upward social and economic mobility.

But turning slums into places of opportunity requires an enabling policy andregulatory environment, and many past responses to the plight of the urban poorhave been either feeble or inadequate. They have ranged from negative policies suchas forced eviction, benign neglect and involuntary resettlement, to more positive onessuch as self-help and in situ upgrading.

Underlying many of these policy responses was the erroneous belief that solvingthe slum problem simply required their physical eradication or their upgradingthrough improved housing and services. Ill-conceived transport policies have oftencompounded the problem by relocating the urban poor to remote areas with limitedaccess to jobs. In the meantime, the poor have been left out of most housing efforts,including subsidized sites and services that have normally been more expensive thanpoor people can afford.

Past solutions essentially failed to address the underlying causes of slums. Futurestrategies need to go beyond the provision of housing and physical services andbecome part of a broader poverty reduction agenda that offers tenure security andemployment opportunities for low income-earning families. Along with urban finance,security of tenure helps unleash the potential of the urban poor for bettering theirliving and working environments. It allows them to improve their shelter conditions,access basic services, and raise credit for housing and livelihood activities.

More generally, urban policy should give more attention to the location of affordablehousing to maximize employment opportunities and reduce transport costs for low-income earners. Participatory upgrading programs that have spurred local economicdevelopment and socially cohesive communities need to be scaled up andaccompanied by investments in citywide infrastructure to ensure that networksof basic services reach the entire population. Still, upgrading slums will merelyaddress the backlog of urban neglect. Forestalling the growth of future slums willrequire substantial improvements in the provision of jobs, housing and servicesfor new urban residents over the next few decades.

The MDG target on slums reflects the growing global concern about urban poverty.Given the enormous scale of predicted growth in the number of slum dwellers, theMDG target must be considered the bare minimum that the international communityshould strive to achieve. Much more will be needed if ‘cities without slums’ are tobecome a reality.

UN Habitat, The Challenge of Slums: Global Report on Human Settlements, 2003.



For every US$1 invested, the world’s developing regions would gain between US$3and US$35 from introducing low-cost improvements aimed at halving the proportionof people without access to basic water and sanitation facilities by 2015, a MillenniumDevelopment Goal (MDG) established in the year 2000. The health benefits from thoseimprovements would be just as significant. Over the next 10-15 years, 546 millionepisodes of diarrhea could be prevented if the necessary investments are made.This amounts to a 10% average global reduction of predicted diarrhea cases and aproportionate drop in the number of deaths caused by water-borne and water-washed diseases.

These estimates are contained in a report released by the World Health Organizationlast month, which evaluates the costs and benefits of expanding access to safedrinking water and basic sanitation to the world’s poor citizens. Specifically, the studycalculates the full investment and operating costs of different types of water andsanitation improvements, as well as a number health and non-health benefits thatinclude the direct and indirect economic gains from avoiding diarrheal disease andassociated deaths.

Despite substantial investments during the 1980s and 1990s, the shortfalls in accessto improved water sources and sanitation services continue to be dismal. In regionslike Africa, 40% of the people do not have access to these basic necessities, whilemore than half of those living in Asia are deprived of adequate sanitation facilities.Not surprisingly, some 1.6 million deaths occur every year due to water-related causes.The burden of disease is concentrated among children under five, suggesting thatpriority interventions should target this group in an effort to achieve rapid andsustainable health gains.

Meeting the MDG target for water would bring improvements in water supply tonearly 700 million people around the globe. Another 1.5 billion people could benefitif the MDG target for sanitation is also reached by 2015. This would amount to 30%of the world’s population by that year, two-thirds of it in Asia and another 20% inSub-Saharan Africa.

Poor countries have a wide array of options for increasing access rates given that asizeable proportion of their population remains un-served. The options range fromlow-technology improvements that require simple maintenance to more costly andsophisticated ones, such as water disinfection at the point of use, regulated pipedwater supply, and household connection to the sewage system. Since total fundingrequirements will vary considerably depending on the technology used, a soundeconomic evaluation is crucial before deciding which type of intervention is bestsuited to a specific setting.

The health impacts from increased service provision would vary from one region toanother. Globally, WHO predicts that providing basic water and sanitation to the entireworld could help achieve a 17% annual reduction in the incidence of diarrheal disease.The present value of deaths avoided from these interventions would surpass US$5.5 billionannually, with the gains overwhelmingly concentrated in the poorest regions where thenumber of un-served people is higher and the burden of infectious diarrhea heavier.

Improving Water andSanitation Facilities

From now until 2015, morethan two billion people indeveloping countries wouldbenefit from actions gearedtowards meeting theMillennium DevelopmentGoals for water supplyand sanitation.

A critical input into economicdevelopment, increasedaccess to these services willhelp prevent the transmissionof water-related diseasesand bring large health andnon-health benefits to verypoor households aroundthe world.

Low-technologyimprovements

Water

• Stand post

• Borehole

• Collected rain water

• Protected spring or well

Sanitation

• Septic tank

• Simple pit latrine

• Ventilated improved pit-latrine

I N R E V I E W



Expanding access rates would confer many other benefits beyond reducing theburden of water-related diseases. For example, the relocation of a well or boreholeto a site closer to user communities, the installation of in-house piped water supplyand closer access to latrines could save people substantial time previously spent onfetching water or walking to an open-air pit-hole. Households would gain about20 billion additional working days per year from the time saved from havingthose facilities more conveniently located. Spread over the entire population,the value of these time savings represents an annual economic benefit ofUS$64 billion.

Similarly, improved health translates into higher school attendance and largeproductivity gains arising from lower morbidity and fewer deaths associated withdiarrhea. The WHO study estimates that the lower incidence of diarrhea would yielda global gain of over 270 million days of school attendance and 3.2 billion productivedays for the working population aged 15 to 59.

In addition to these direct benefits, meeting the MDG targets for water and sanitationcould deliver important cost savings. For the health sector, the resultant lower demandfor treatment could save as much as US$7 billion in annual health care costs. In turn,patients would avoid hefty treatment costs — including expenditures on drugs andtransport to a health facility — as well as the opportunity costs of time spent ill orseeking care. These savings could be very large for poor households, especially whenpatients have to travel long distances to a health facility or pay for its services.

When all the health and non-health benefits are combined, halving the proportionof people without access to basic water and sanitation is predicted to bring the totaleconomic gain to US$84 billion per year. In contrast, providing these services wouldcost less than US$2 billion annually for water alone and slightly over US$11 billionfor water and sanitation combined. While not insignificant, these expenditures dwarfin comparison with the potential benefits from such preventive measures. And whenthe costs are spread over the entire population, they range from US$0.2 to US$2.2per capita, depending on the region.

It is true that most expenses would have to be made upfront, while the benefits accrueover time. But even in the most pessimistic scenarios, a careful assessment of the totalbenefits and costs involved in water and sanitation projects tips the balance in favorof positive investment decisions. Whether simple or more sophisticated technologiesare used, WHO has found all water and sanitation improvements to be cost-beneficialin every one of the world’s regions. Choosing advanced technologies would yieldmassive overall health gains, but at a cost that very few countries can afford. Fordeveloping countries, therefore, low-cost options seem the best choice. They generatelarge health benefits without demanding heavy investments or maintenance.

Despite their low cost, it is unreasonable to expect the poor themselves to pay forthose investments. At the same time, the meager health budgets of developingcountries are unlikely to provide more than a fraction of the front-loaded expensesfor improving water and sanitation services. Given the large non-health sector outlaysand benefits of health care interventions, a variety of financing mechanisms spanningseveral sectors will have to be put in place to meet those costs in the coming years.

Ultimately, increasing resources for water and sanitation will require strong advocacyand political will at the national and international levels. The potential productivityand income effects of improved access are strong arguments for funneling resourcestowards those sectors. Even more compelling is the fact that greater access will leadto longer and healthier lives for the world’s poor.

World Health Organization (Guy Hutton and Laurence Haller), Evaluation of the Costs andBenefits of Water and Sanitation Improvements at the Global Level, 2004.

* Includes Algeria.

** Excludes Algeria but includes Middle Eastern countries,

Afghanistan, Iran and Pakistan.


International Poverty CentreSBS - Ed. BNDES, 10° Andar70076 900 Brasilia DFBrazil

[email protected]/povertycentreTelefax +55 61 2105 5000 M

ay 2

004

challenge of poverty

Documents