Report No. 36347-KE Kenya Inside Informality: Poverty, Jobs, Housing and Services in Nairobi's Slums May 31, 2006 Water and Urban Unit 1 Africa Region Document of the World Bank Table of contents Preface andAcknowledgements ..................................................................................................................... 5 Executive Summary ........................................................................................................................................... 7 2.Research methodology andthe data......................................................................................................... 12 1.Introduction................................................................................................................................................. 10 13 3.Poverty inNairobi's slums ......................................................................................................................... 2.1 Nairobi's populationand estimates regarding the number of slum dwellers .............................. 13 3.1 Disaggregating"poor" and "non poor" households....................................................................... 14 3.2 Non-monetary factors influencingpoverty, incomes and expenditures: Multivariate regression analyses ......................................................................................................................................................... 15 (a) Poor versus non-poorhouseholds ................................................................................................. 16 (b) Per capita income............................................................................................................................. 17 (c) Per capita expendture ...................................................................................................................... 18 4.Who lives inNairobi's slums? Demographcs, householdtypes and composition.......................... 20 4.1 The populationpyramid: Gender and age profiles of slum dwellers............................................ 20 4.2 Household size and composition....................................................................................................... 20 21 5. Economic base: Incomes, jobs and micro-enterprises inthe slums.................................................... 4.3 Female-headed households:Copingwith the gender handicap?................................................... 24 5.1 Household incomes and expenditures .............................................................................................. 24 5.3 Employmentinthe household........................................................................................................... 30 5.2 Indmidualsin the labor force .............................................................................................................. 28 5.4 Householdmicro-enterprises (HMEs).............................................................................................. 31 6. Housing, tenure and rents.......................................................................................................................... 34 5.5 Bankingand credit................................................................................................................................ 33 6.1 Tenure, length of stay and tenure security........................................................................................ 36 6.2 Housingsize and quality-crowding 37 6.3 Rental values inthe slums................................................................................................................... andconstruction materials................................................. 37 6.4 What drives rental values inNairobi's slums?.................................................................................. 38 6.5 Crime: Unsafe in their own neighborhood....................................................................................... 40 43 7. (Rural) Emigrants?:Previous residence, remittances and real estate assets outside.......................... 6.6 Stuck ina hgh-cost low-quality trap?................................................................................................ 43 7.1 Previous residence................................................................................................................................ 43 7.2 Ownership of real estate assets outside of Nairobi......................................................................... 44 7.3 Remittances ........................................................................................................................................... 44 7.4 Registeredvoters and participation inlast election-enhancing .................................... 45 8.Infrastructure access and service delivery................................................................................................ 49 "voice" 8.1 Water supply in the slums................................................................................................................... 50 (a) Primarywater sources ...................................................................................................................... 50 (b) Per capita use .................................................................................................................................... 50 (c) Unit cost of water ............................................................................................................................. 50 8.2 Electricity, other fuels, and street lighting ........................................................................................ 51 8.3 Sanitation and drainage........................................................................................................................ 52 (a) Access to toilets ................................................................................................................................ 52 (b) Excreta disposal and sewerage....................................................................................................... 52 (c) Grey water dsposal and drainage ................................................................................................... 52 3 8.4 Modes of transportation ...................................................................................................................... 54 (d) Solid waste disposal.......................................................................................................................... 53 9. Education ..................................................................................................................................................... 58 8.5 Infrastructure upgrading efforts thus far and results ...................................................................... 54 9.1 Enrollment rates among chilhen (5-14 years of age) ..................................................................... 9.2 Educational attainment (age 15 years or more)................................................................................ 59 59 9.3 Educational attainment at the householdlevel and its impact on poverty .................................. 60 10.9.4 Comparisonwith education statistics reportedinthe I<-DHS 2003 11.Developmentandpolicyimplications...................................................................................................... priorities ............................................................................................................................ 65 ............................................ 60 Conclusions 67 References ......................................................................................................................................................... 72 ANNEX 1:Univariate regression analyses.................................................................................................. 73 ANNEX 2: PopulationpyramidsinKenya (Rural, Urban,andNational)............................................. ANNEX 3: Male-headedandfemale-headed households........................................................................ 74 75 ANNEX 5: Householdmicro-enterprises, bankingandcredit................................................................ 77 ANNEX 6: Housingandpreviousresidence.............................................................................................. ANNEX 7: Highest level of education attained andprimaryactivity. by age andwelfare ..................81 82 ANNEX 8: Backgroundnote onmethodology.......................................................................................... 84 ANNEX 10:Householdquestionnaire -Nairobi...................................................................................... ANNEX 9: Note on defintion of slums usedby CBS. Kenya................................................................ 87 92 \ 4 Preface andAcknowledgements This report was prepared by a team comprised of Sumila Gulyani (Task Team Leader), Debabrata Talukdar and Cuz Potter, under the ditection ofJaime Biderman (Sector Manager, AFTUl), Colin Bruce (Country Director, CD5) and Geoffrey Bergen (Country Program Coordinator, CD5). This study buildsonwork startedunder theAfrica: Regional Urban UpgradingInitiative (2001-2004), financed inpart by a grant from the Nonvegan Trust Fund (NTF-ESSD),and managed jointly by a team comprised of Catherine Farvacque-Vitkovic, Sylvie Debomy and Sumila Gulyani. Itwas under t h i s initiative that the idea of a comparative study of the slums of Nairobi and Dakar was f i r s t proposed and financed. Specifically, NTF financing was used for the design and implementation of surveys of about 2000 households each inNairobi and Dakar. Descriptivereports containing tabulations of basic results were prepared by the consultants (COWI) in2004 and are available for both cities. Inthis study, we use data on the subset of 1755 slumhouseholds inNairobito generate a different and more analytical understanding of Nairobi's slums. A similar in-depth analysis of the data on Dakar's slums has been proposed and i s awaiting approval. This studywould not have beenpossible without support from several of our colleagues andwe wouldlike to express our sincere gratitude to: Catherine Farvacque-Vitkovic and Sylvie Debomy for their invaluable inputs into the design of the questionnaire and for their sustained help inco-management of the consultant contracts and coordination of field work inthe two cities. This collaboration between AFTUl andAFTU2 has not only helped ensure coordmated implementationof the studyin two very different parts of the African continent but also, and more importantly, enriched the intellectual content of this work. Makhtar Diop (previous Country Director for Kenya) for h i s enthusiastic support at the inception of h s study, for ensuring financing for a Nairobi-specific analyses, and for supportingthe idea of an addtional report comparingNairobiwith Dakar. Valerie Kozel, Judy Baker and Salman Zaheer for their advice on s a m p h g and study design, andJoseph Gadek for helping u s navigate procurement and contracts. Judy Baker andJesko Hentschelfor serving as peer reviewers and providingextremely thoughtfuland useful comments. Kathleen Beegle, Louise Fox, Luc Christiaensen, Michael Mills, Praveen Kumar, Tova Solo, Natasha Iskander, Genevieve Connors, Ellen Basset, K r i s t i n Little, Ian Gillson, B a j o r Mehta and Matthew Glasser for taking the time to review this tome and for their constructive comments. Not all of their excellent comments could be incorporated inthls version but wdl be infuture papers. Robert Buckley for helpingu s finance an additional qualitative study inthe slums and Ashna Mathema for delivering it inthe form of a report titled "A view from the inside." COWI consultants for managing the field work-entailing household surveys and focus group Iscussions-and data coding of the results, and for delivering a professional product. 5 Dr.James Mutero, as COWI's teamleader inNairobi, for h s hands-on leadership ofthe local survey teams and, subsequently, for his valuable inputs duringthe data analysis stage. Staff from the Central Bureau of Statistics for partneringincrucial aspects of this study-in particular, for their assistance inconstruction of the sample, for managing the field-based re- listingof households inselected EAs, and for generating summary tables on key indicators for Nairobi from the 1999 census data. Participants at the Urban Sector Brown Bag Lunch Presentation held on 28 November 2005 at the Bank in D C for a valuable discussion which has helped informsome of the analyses presented. Nicole Volavka for volunteering and delivering a thoroughly professional copy edit of an earlier draft. Perla SanJuan for her support ininnumerable ways over the course of this research project. 6 Executive Summary Africa i s the world's fastest urbanizing region and also its poorest continent. Incountries such as Kenya and Senegal, along with urbanization, the incidence of urban poverty i s also increasing. Informal or slum settlements are absorbing an increasing share of the expanding urban population and are home to the vast majority of the urban poor. Untilrecently, however, most poverty- oriented research has focused on rural areas. As a result, very little i s known about urbanpoverty in general, and about slums inparticular. In fact, inmost countries there are no reliable estimates even onbasic indicators-such as the number of people residing inthese slums and the proportion of them that are poor-let alone a good understanding of the livingconditions of s l u m dwellers, the nature of poverty that they face, and factors that may be helping slum households fight or escape poverty. Such ambiguity makes it l f f i c u l t to justify, design and implement appropriate programs for the poor livinginthese settlements and even harder to assess the impacts of policies and programs that do get implemented. This study attempts to fillgaps inour knowledge about slums inNairobiand to, hopefully, also create a precedent and basis for similar studies in other African cities. Drawing on data from a unique, population-weighted, stratified random sample survey of 1755 slum households inNairobi, t h s study attempts to shed light on housingand infrastructure conditions, economic opportunities, education, and poverty inNairobi's slum settlements. Analytically, it focuses on the following questions: How poor and inadequately served are slum dwellers inNairobi? What factors are correlated with poverty among slum households in this city? W e find that the incidence of economic poverty i s very highinNairobi's slums. About 73 percent of the slumdwellers are poor-that is, they fall below the poverty line and live on less than US$42 per adult equivalent per month, excludmg rent. The highrate of economic poverty i s accompanied by horrible living conditions and other forms of non-economic poverty. Slum dwellers' access to basic services such as water, sanitation, electricity, and transportation i s far worse than anticipated. For instance, only 22 percent of slum households have an electricity connection and barely 19 percent have access to a supply of piped water, in the form of either an in- house water connection or a yard tap. Such low connection rates stand insharp contrast to the relatively good coverage data reported for Nairobi as whole. Specifically, city-level data suggest that 71-72 percent of Nairobi's households have pipedwater (in-house connections or yard taps) and that 52 percent have electricity connections. Inother words, city-level averages mask a highlevel of inequality ininfrastructure access; Nairobi's slums lag city-wide averages by 50 percentage pointsin terms of connections to pipedwater and by 30 percentage points in terms of electricity connections. The housingunits are mostly Illegal, sub-standard inquality, and crowded. Yet the rents are high. Unlike inmany other cities of the world, an extraordinary 92 percent of the s l u m dwellers are rent- paying tenants (rather than "squatters" who own their units). Unit owners are mostly absentee landlords who seem to be operating a highly profitable business inprovidingshelter to the poor. In sharp contrast to the widely-held notion that slums provide low-quality, low-cost shelter to a population that cannot afford better standards, we find that Nairobi's slums provide low-quality but high-cost shelter. Slum dwellers have poor access to gainful employment. About 49 percent of adult s l u m dwellers have regular or casual jobs and 19 percent work ina household micro-enterprise, but at least 26 7 percent are unemployed. Unemployment rates are highest among youth (age 15-24) and women- 46 percent of the youth and 49 percent of the women report that they are unemployed. This i s problematic not least because the presence of an unemployed member ina householdi s strongly correlated with poverty. At the household level, micro-enterprises are helpingdiversify the income portfolio and appear to be assisting inthe struggle against poverty. About 30 percent of households report that they operate an enterprise and, encouragingly, ownership of an enterprise i s negatively correlatedwith poverty. Strikingly, as compared to male-headed households, female-headed households are more hkely to be operating an enterprise and usingthese to gainfully employ adults inthe household. Addtional research i s required to better understand the mechanisms throughwhich micro-enterprises can or do affect poverty. At the very least, the presence of these enterprises indicates that there i s significant and relatively successful entrepreneurial activity inthe slums; these enterprises appear to be worthy of some attention from public institutions and development agencies. The story on education i s very encouraging but deserves more attention. About 78 percent of adult s l u m dwellers report that they have completed primary school. Even more important, as many as 92 percent of school-age chlldren are actually enrolled inschool. These school enrollment rates are higher than the levels reported for Nairobi as a whole inthe 1999 census and inthe 2003 K-DHS; t h l s seems to be a positive outcome of the introduction of free primary education inJanuary 2003. This finding, albeitpreluninary, bringsinto question some of the negative assessments regarding the effects of the free education policy on net enrollment. Boththe highrate of school enrollment among children and the relatively hlghprimary-school- completion rate among adults are causes for optimism. Better still, and as we would hope, we find that higher education levels are positively correlated with income and negatively correlated with poverty among slum households. This i s not to argue that all i s well regarding education inthe slums.' Rather, it i s to suggest that it i s bothimportant and worthwhile to continue worlung on education challenges inurban slums. Specifically, more work i s required to enhance educational levels beyond primary schooling, reduce both the gender and welfare gap ineducation among s l u m dwellers, and address their concerns about the quality of primary school education. Finally, a systematic comparison between poor and non-poor households reveals five types of non- monetary factors that are positively correlated with household poverty inthe slums: (i) household demographcs-specifically, households that are large insize and have a highproportionof women; (ii) educationlevels;(ii)lackofownershipofamicro-enterprise;(iv)unemploymentinthe lower household; and (v) lack of access to infrastructure, inparticular, electricity and water supply. Given their strong correlation with poverty, these five factors can and should serve as a basis-a starting point-for the design of any poverty-alleviation efforts inthe slums. Policy andprogram implications Overall, living condtions inNairobi's slums are bleak and the incidence of poverty i s high. But there i s hope, not least because s l u m dwellers are educated, entrepreneurial, enfranchised, and seemingly able to enhance their economic welfare over time. Not only i s there need for developmental action inthese settlements but also the economic and social returns to well-chosen and well-designed programs are potentially very high. There i s also crude evidence that previous 8 s l u m upgrading efforts, despite having been extremely modest inscale and scope, have created some benefits. What shouldthe government prioritize? The slum dwellers themselves identify their top four development priorities as toilets, water, health, and electricity. Their priorities resonate strongly with the technical analyses. Infact, the technical analyses and residents' priorities have a clear area of overlap-infrastructure. Investments ininfrastructure-such as water, sanitation, paved paths and electricity-can help achieve improvements inseveral of the factors correlated with poverty as well as address some of the health concerns of slum dwellers. Inaddition to infrastructure, education deserves to be a highpriorityinthe slums. Although the "free primary education" program i s meeting the basic need of getting children enrolled in school, residents' concerns regardmg overcrowding and quality need to be reviewed and addressed. Equally important i s the need to reduce the welfare and gender gaps insecondary school completion rates. Area-wide programs or sector-specific ones? Ineducation, an independent sector-specific approach makes sense and can work. Interms of addressingvarious infrastructure deficiencies, we would argue that any serious and sustainable improvements will require a multi-sector and area-wide approach, given the base conditions inNairobi. Also, unlike inmany other cities, this i s a case where housingissues need to be dealt with alongside infrastructure. In fact, if w e were asked to identify just one entry point-that is, one sub-sector-into the problem of living conditions inthe slums, it would be the structure of the housingmarket. W e would argue that a key goal of any efforts inNairobi's slums shouldbe to break the low-quality, high-cost trap in slum housingand infrastructure, and that the only way to get there i s to start discussions with both landlords and tenants. 9 . 1.Introduction Indeveloping countries, an estimated 870 millionpeople were livinginurban slums in2001 (UN MdlenniumProject 2005). If current trends were to continue, the number of slum dwellers will grow to an estimated 1.43billion by 2020 (UNMillenniumProject 2005). World leaders and development agencies are again-after a significant hiatus-displaying their concern about the issue and slums appear to be back on the core development agenda. Indeed, at the United Nations MillenniumSummit in 2000 and subsequently at the Johannesburg Earth Summit in2002, world leaders agreed to a set of time-bound, measurable, and highly influential development targets- widely known as the Mdlennium Development Goals-which include a commitment to significantly improve the lives of 100 million slum dwellers by 2020 (UN2003). The commitment to improve the lives of slum dwellers i s well-intentioned and important, but the task of achieving &IS goal i s fraught with problems. First, there i s little informationand understanding of the scale and nature of urban poverty ingeneral, and the situation in slums in particular. Second, a whole generation of earlier efforts-starting inthe 1970s-to upgrade urban slums has, at best, been only partially successful.' Third, not only is the scale of the slum problem growingrapidly inmost cities of the developing world butit is also widely acknowledged to be increasingly complex-politically, institutionally, and, at times, technically-and therefore beyond the scope of simple and modest solutions. Overall, the urban s l u m problem appears to be a black box interms of its nature and dynamics, i s somewhat daunting in scale and scope, and often competes for policy attention and resources with the task of rural poverty alleviation. InSub-SaharanAfrica the slumproblemis particularly acute. Africa is theworld's fastest urbanizing region and its poorest continent. Incountries such as Kenya and Senegal, along with urbanization, the incidence of urban poverty i s also increasing. Informal or s l u m settlements are absorbing an increasing share of the expanding urban population and are home to the vast majority of the urban poor. These settlements are generally characterizedby highpopulation densities, h t e d or non- existent urban services, and low-quality housingstock. Here, even more so than in other regons of the world, the scale and nature of these settlements-even basic population and demographic indicators-remain a source of much contention and debate. Such ambiguity makes it difficult to justify, design and implement appropriate programs for the poor livingin these settlements and even harder to assess the impacts of policies and programs that do get implemented. A first task inmost cities, then, is to figure out what is inthe black box called slums and to agree uponpriorities for action inthat city. How many slums dwellers does the city have? Who are they and how poor? What aspects of their current quality of life need to be improved-should the prioritybe jobs or education or infrastructure or reduction ofviolence or some combination of such efforts? What are the factors that are currently helping slum dwellers intheir own quest for physical, economic and social upwardmobility? Ths studywas designed to fillgaps inour knowledge about slums intwo African cities-Nairobi and Dakar. Drawing on detailed surveys of households residmg in slums-1 755 and 1960 households inNairobi and Dakar, respectively-this study aims to develop a demographic, 1Untilrecentlybothdevelopment programs and research efforts have been focused on ruralpoverty. This is just starting to change with recent studies that show that poverty i s not entirely a rural phenomenon, even in a region such as Sub-Saharan Africa, and that in some countries, such as Kenya, urban poverty has been rising faster than rural poverty. 2See, for example, Gulyani and Basset (forthcoming) and Basset et al. (2003). 10 economic and infrastructure profile of slum settlements inthese two cities. Analytically, it focuses on the following questions: how poor and inadequately served are slumdwellers inNairobiand Dakar? What are the factors correlated with poverty among slum households in each city? Inthis paper we presentresults for Nairobi. The following findings and relatedarguments are worth highhghtingupfront. First, the incidence of economic poverty is very highinNairobi's slums and it is accompanied by horrible living conditions and other forms of non-economic poverty. The majority of slum dwellers fall below an expenditure-based absolute poverty line. At the same time, their access to basic services such as water, sanitation, electricity, and transportation i s far worse than anticipated-the conditions raise serious public health concerns and cannot but have a negative impact on overall productivity and well-being. Second, Nairobi's slums provide low-quality but high-cost shelter. This findmg directly challenges the widely-held notion that slums provide low-quality, low-cost shelter to a population that cannot afford better standards. The conventional understanding i s that, on the one hand, the quality of s l u m housingtends to be poor because of a combination of informality of tenure, use of low quality buildingmaterials and construction methods, and disregard of (legally-specified) minimum space/planning standards. On the other hand, these lower standards result inhousingthat i s cheaper and thereby affordable for lower-income people. While this may be the case inthe slums of some cities, it i s not true inNairobi-slum dwellers inNairobi, most of whom are very poor, are paying a lot for their sub-standard housing. Thud, somewhat encouragingly, there is heterogeneity among Nairobi's slums dwellers, their living conditions, and their economic welfare. The people living inthe slums are very poor but not universally so. Many are rural immigrants but a large proportionappears to have immigrated from other urban areas. Access to infrastructure services i s very poor but a small proportion of slum households have managed to gain access to services such as electricity and private piped water connections. Education levels vary signlficantly both among and within households, but at least they are not universally low. Although many households have at least one unemployed adult, their economic portfolios include some combination of a regular job, casual work, and/or household micro-enterprises. Each of these micro findings i s interesting initself and i s discussed indetail in the paper'to provide sector-specific insights. Taken together, these micro fmdmgs suggest that the situation i s not universally bleak and there are at least a few positive factors that can be built upon to foster economic and physical development inthese slums. Fourth, a systematic comparison between poor and non-poor households reveals five types of non- monetary factors that are strongly correlated with poverty inthe slums: (i) household demographics (size and gender and age composition); (ii) education; (iii) ownership of a micro-enterprise; (iv) unemployment in the household; and (v) infrastructure access, inparticular electricity and water supply. A slumhousehold is more likely to be poor, the larger its household size and the more the number of women in the household. Households who own a micro-enterprise are less likely to be poor, while those with even one unemployed adult are more likely to be poor. As the education level achleved by any member of the household rises, the likelihood of being poor falls. Finally, poor households are systematically and disproportionately less likely to have access to either electricity or a private piped water connection. Given their strongcorrelation with poverty, these 11 five factors can and should serves as a basis-a starting point-for the design of any poverty- alleviation efforts in the slums. Fifth, slum dwellers' own development priorities-a first proxy for "demand"-resonate strongly with the technical analyses. When asked to choose among competing investments, slum dwellers identified their top four development priorities as: toilets, water, health, and electricity. Their priorities, combined with the technical findings, provide a clear framework for action inthe city's slums. Sixth, although upgrading efforts in the slums have been piece-meal and modest thus far, they do appear to have created some benefits. For every 10 slum households who noted that a given sector- specific intervention had occurred in their neighborhood, nine said that it was working and that the situation was better than before. Addtional analyses, usingthe case of the water sector, support their general cominent-we find that indicators such as price, service access, and users' perceptions regarding price and quality of their water supply are all better in areas that have had a "water improvement" project as compared to those that have not. Although the degree of improvement in each water service indicator i s small, it i s nonetheless statistically significant and, therefore, encouraging. The paper i s structured as follows. Section two outlines the research methodology and the data. Section three estimates poverty incidence inthe slums and identifies factors correlated with poverty. Sections four through nine present both descriptive data and analyses on each of the following topics: demographcs, economic base, housing, previous residence of "emigrants," infrastructure, and education. Section 10 summarizes the development priorities of slum dwellers and Section 11 presents conclusions and policy implications. 2. Researchmethodologyand the data InFebruary/March 2004, a householdsurvey was administeredinNairobi's slum settlements. A total of 1755 households were surveyed in 88 Enumeration Areas (EAs). The sampling frame was constructed as follows. For census purposes, Kenya's Central Statistics Bureau (CBS) has divided Nairobiinto about 4700 EAs, of which 1263 as categorized as "EA5" or "informal settlements." EA5s are characterizedby poor-quality sub-standard housingand poorinfra~tructure.~The 88 EAs inour sample were selectedrandomly fromthe subset of 1263EA5sandweighted bypopulation. As the lists of households had not been updated for a few years, a complete re-listing was conducted ineach selected EA and the sample householdswere selected randomly from the new lists. CBS not only collaboratedindesigning the sampling frame of this study, but also took responsibility for the field-based re-listing of households in the 88 EAs. 3 CBS' methodology for creating the five categories (EA1-EA5) i s presented inAnnex 8; the definition for EA5 is excerpted here for ease of reference. While other categories (EA1-EA4) are largely formal planned settlements, the last category (EA5) i s largely composed of informal settlements. An EA5 area "has characteristics that distinguishit clearly from the rest of the categories. The structures are largely temporary, made o f mud-wall or timber-wall with cheap roofing materials, which may be iron sheets, makuti, grass or even d o n paper or cartons. The infrastructure inthese areas i s relatively poor as there i s n o proper sanitation, no clear roads for entry and even water i s not connected to the dwelling structures. The following types o f area fall inthis category: MkuruK w a Njenga, Korogocho, LainiSaba, Silanga, Soweto, Kamuthii, and Mathare Valley. .. .It is characteristic that, close to most of the high-income areas, there are informal settlements. However, our consideration i s what would be the mean in terms o f the facihties among all the residents o f the areas inthe categories. However, where a s l u m i s neighboring a class, which i s higher, the s l u m within that locality willbe identified and placed ini t s appropriate category." 12 Overall, this i s a population-weighted stratified random sample and it i s representative of the 1263 EAs categorized as "informal settlements" by CBS. Further, to complement the quantitative survey, qualitative studies-community questionnaires and focus group discussions-were conducted in 10 of the survey sites5 To the best of our knowledge, this is one of few large-sample, multi-sectoral, and representative surveys of s l u m households conducted, thus far, inthe city. O n e other comparable dataset i s APHRC's study of 4564 slumhouseholds, butit focuses almost entirely on healthissues (APHRC 2002). Indeed, the existence of the APHRC study-combined with the need to keep household interviews of reasonable length and complexity-is a key reason why our study examines several development sectors other than health. 2.1 Nairobi's population and estimates regarding the number of slum dwellers Inthe 1999 national census, Nairobi's populationwas foundto be 2.139 d o n and slums accounted for 0.6 d o n people or about 30 percent of the city's population.` By contrast, estimates inthe grey literature (e.g. consulting studies, reports by NGOs and aid agencies, etc.) are significantly higher. For instance, a study conducted in 1993 estimates that 55 percent of Nairobi's population lives in slums (Matrix Consultants/USAID 1993 cited inAdler 1995). There are at least two possible explanations for the divergence inestimates. First, it is highly likely that CBS and the other researchers use dfferent boundaries for Nairobi-that is, several of the studies prepare estimates for the entire Nairobi metropolitan area and include slum settlements that are on the periphery of the city's adrmnistrative boundaries; using CBS' categorization means that people residing in slums on the city's periphery but outside its administrative boundaries are excluded inthe estimate. Second, it i s possible that the CBS has underestimated the number of EAs that are s l u m EAs-that is, it may have mis-categorized some of the EAs either accidentally or by usingtoo narrow a definition. Clearly, additional research i s required to resolve this issue. Meanwhile, and for the purposes of thls study, the categorization used by CBS offers a more robust (but,possibly, conservative) starting point than the approaches and estimates used inother studes. W e see this number-0.6 million s l u m dwellers in 1999-as establishing a "floor" or minimumnumber of slum dwellers inthe city; it i s entirely possible that the actual number i s higher. The sampling and results of this study, therefore, pertain to the universe of 0.6 million slum dwellers. 3. Poverty inNairobi's slums Recent studies inKenya have noted that urban poverty has been increasing faster than rural poverty. In1992,29 percent of the people livinginurban areas fell below the povertyh e compared to 48 percent of those inrural areas (CBS 2000 cited inAPHRC 2002). By 1997, the poverty rate inurban 4 A similar household survey was conducted inDakar's slums to allowfor a comparative analysis withNairobiandto establish a base for comparative studies, inthe future, with other cities. The Dakar survey covered 1960 households selected randomly from a stratified random samp1,e o f 168 EAs, from a universe o f 2074 EAs in the city. The results fromDakar are not presentedinthis paper. 5 In addition, about 100 households were surveyed in nine EAs selected from known "sites and services'' (S&S) schemes, that is, areas that were developed under donor projects between the late 1970s and mid-1980s to provide affordable housing plots with basic services for low-income residents o f Nairobi. These data will allow for a separate (butrather preluninary) comparative analysis o fhouseholds livinginslums versus S&S schemes. 6 Between 1989 and 1999, Nairobi's population growth rate was about 4.8 percent per annum. 13 areas had increased to an extraordinary 49 percent, while that inrural areas increased modestly to 53 percent (CBS 2000 cited inAPHRC 2002). Urban areas now not only have a high poverty rate, but are also highly unequal-the country's first rigorous poverty mapping exercise reveals this very graphically (CBS 2003). For the case of Nairobi, the poverty mapping exercise estimated the poverty rate at 44 percent, with poverty headcount varying from below 20 percent in the richest &strict to over 70 percent inthe poorest districts of the city (CBS 2003). These numbers are calculated by usingproxyindicators (such as access to water and quality of housing)rather than actual income or expenditure data. This means, for instance, that residents of s l u m settlements- because they have poor quality housingand infrastructure-are almost by dehnition classified as poor. To move our understanding of urban poverty a step further, this study takes a closer look at both the level and nature of poverty within slums.1 For this, we use both monetary and non-monetary inhcators of poverty and analyze the linkages between them. Inthls sec.tion, we first &scuss monetary indicators of poverty and explain the measure selected to disaggregate slum households into two welfare categories-"poor" and "non-poor." W e then use multivariate analyses to examine w h c h non-monetary factors are correlated with poverty inthe slums. 3.1Disaggregating "poor" and "non poor" households For the purpose of this study, we used four differentrapid assessment techniques to ascertain poverty rates in the slums. As a first measure, a household-specific poverty line was calculated for each household-to adjust for household size and age composition-and respondents were asked whether their total monthly expenditure was above or below the computed amount.' Inaddition to this "discrete" (yes/no) measure of poverty, two "continuous" measures of householdwelfare were also computed-per capita income per monthand per capita expenditure per month. These were derived for each household from self-reported total monthly income and the total monthly expenditure that they incur ina typical month. Households were also asked to report actual spendmg on selected items-such as food, rent, utilities and transportation-in the previous month and these allowed for a cross-check on total expenditures reported. Finally, household assets were documented to allow for factor analyses and computation of a relative ranking of wealth. The discrete measure-an expenditure-based poverty line-was the one selected to disaggregate the sample into poor and non-poor households. This measure was selected because it i s based on and fully consistent with the Government of Kenya's (GoK) own methodology for assessing poverty in the country. Specifically, we take the 1999 poverty line as defined by the GoI< and adjust it for actual inflation to calculate the poverty threshold for 2004. * Usingthis expenditure-basedpoverty line-defined as an expenditure of Iin&, on4 about 20percentmore-than thepoor (Table 3 and Figure 3). Average monthly rent i s I