Intra-city differentials in urban poverty and slums in Nairobi, Kenya

Material Information

Intra-city differentials in urban poverty and slums in Nairobi, Kenya measurements, determinants, consequences and implications
Deng, Ning
Publication Date:
Physical Description:
xi, 193 leaves : ; 28 cm


Subjects / Keywords:
Urban poor -- Kenya -- Nairobi ( lcsh )
Slums -- Kenya -- Nairobi ( lcsh )
Poverty -- Measurement ( lcsh )
Poverty -- Measurement ( fast )
Slums ( fast )
Urban poor ( fast )
Kenya -- Nairobi ( fast )
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )


Includes bibliographical references (leaves 170-193).
General Note:
College of Architecture and Planning ; Design and Planning Program
Statement of Responsibility:
by Ning Deng.

Record Information

Source Institution:
University of Colorado Denver
Holding Location:
Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
181588736 ( OCLC )
LD1193.A735 2007d D46 ( lcc )

Full Text
Ning, Deng
B.S., Nanjing University, 1997
M.A., Nanjing University, 2000
A thesis submitted to the
University of Colorado at Denver/Health Sciences Center
in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
Planning and Design

by Ning, Deng
All rights reserved.

This thesis for the Doctor of Philosophy
degree by
Deng Ning
has been approved
Willem van Vliet


Ning, Deng (Ph.D., Planning and Design, College of Architecture and
Intra-city Differentials in Urban Poverty and Slums in Nairobi, Kenya:
Measurements, Determinants, Consequences and Implications
Thesis directed by Professor Thomas A. Clark
This dissertation research intends to document methods for measuring and
analyzing urban poverty in Third World Cities by using Nairobi as my primary
case. The main data source for this study is a 5% sample undertaken in the
Kenyan Population and Housing Census of 1999. The sample has more than
100,000 individual records with over 80 variables measuring basic living
conditions, socio-economic-demographic characteristics of urban residents,
and health indicators measuring, for example, child mortality. Moreover, a
supplemental survey dataset was collected by UN-HABITAT and the Kenya
government in 2002 to classify Nairobis 4700 Enumeration Areas (EAs) as
Through recoding and aggregation, I propose to construct a composite
material poverty index using principal components analysis. Next, I will build
a 2SLS (two stage least square regression) model to identify the most
significant variables determining the level of material poverty, given that there
is a recursive relationship between the material poverty in each sub-location
and its population density. Finally, a cross-sectional regression analysis will be
undertaken to examine the effect of living in a slum and residential
segregation of slums from non-slums on local child mortality rate.
The findings will have a three-fold implication for anti-poverty or anti-slum
practices. First, a composite material poverty index has a distinctively
empirical utility for the direct measurement of the living conditions of the
urban poor and slum dwellers. Second, lower population density significantly
decreases the material poverty level of a sub-location after controlling for
other factors, which lends some support to area-based anti-poverty or anti-
slum strategies that, for example, de-densify the slum areas through

redevelopment efforts. Lastly, the neighborhood effects of slums and
residential segregation on local child mortality rate are statistically significant
with negative correlations. This finding implies a new policy prescription for
the slum eradication: de-segregating slum dwellers from non-slum dwellers,
which is effective at least from a health perspective.
This abstract accurately represents the content of the candidates thesis. I
recommend its publication.

Many thanks go to my advisor, Thomas A. Clark, for his guidance,
contribution and support to my dissertation research. I also wish to thank all
the members of my committee for their valuable participation and thoughtful
insights. This dissertation was based on my consultancy work at the Global
Urban Observatory (GUO) program in UN-HABITAT in 2004.1 am grateful
to the comments made by staff members in GUO program in UN-HABITAT
on the earlier drafter of this study. Great thanks are extended to the
collaboration of the Central Bureau of Statistics of Kenya for making available
maps about 110 sub-locations in Nairobi. Last, I thank my wife for her
consistent and endless support to my dissertation work and my parents for
their warmly encouragement and spiritual support.

1. INTRODUCTION......................................................1
Research Questions.........................................9
Significance of the study.................................10
Structure of the dissertation.............................14
2. LITERATURE REVIEW................................................16
Area I: Concepts and Measurements of Urban Poverty........16
Area II: Poverty Determinants Analysis....................25
Area III: Residential Segregation, Urban Poverty,
and Public Health.................................31
Alternative Measures of Residential Segregation...31
Poverty, Residential Segregation and
Public Health.....................................37
3. STUDY SETTING: NAIROBI..........................................44
Brief History of Nairobi..................................44
Poverty and Slums in Nairobi..............................49
Residential segregation in Nairobi........................54
Socio-Political-Economic Structure and Process...........59
Taxonomy of Anti-Poverty and Anti-Slums Policy Prescriptions... 62

Top-down Approaches..............................64
Bottom-up Approaches.............................68
Public Health in Nairobi.................................71
4. MAJOR DATA SOURCES............................................75
Kenya Census, DHS and GIS................................75
Spatial Hierarchy and Unit of Analysis...................77
5. MATERIAL POVERTY INDEX........................................80
Issues of Variable Selection: Relevance and Significance.80
Data Processing: Recoding and Aggregatioa................87
Principal Component Analysis (PC A)......................91
Validating the Index Result..............................95
Poverty Mapping.........................................100
2SLS Material Poverty Determinants Analysis.............106
Density and Urbanism....................................109
Results and Interpretations.............................114
Summary Statistics.................................114
Correlation Test...................................124
Two Stage Least Squares (2SLS) Modeling Results....127
CHILD MORATLITY..............................................134
Measuring Segregation: Dissimilarity and Isolation Indexes.... 137
Modeling Child Mortality Rate...........................140

Segregation, Poverty and Public Health: Theoretic Angle.141
Results and Interpretations...........................146
8. CONCLUSIONS AND DISCUSSIONS..............................151
Summary of Major Results and Findings.................151
Limitation of the Study...............................156
Policy Implications: Discussions and Conclusions......159

AREA WITH ABOUT 300,000 INHABITANTS.............53
MEDIATION MODEL............................142

5.1 A description of recoding variables on living conditions...........88
5.2 Overall performance of the PCA model: total variance explained.....94
5.3 Variables coefficients for each component..........................95
5.4 Paired sample correlations of roof and floor material, access to water,
sewage and overcrowding in Census and KDHS data..................99
5.5 Paired sample test of variables in Census and KDHS.................100
6.1 Descriptive statistics on housing conditions and basic services in slums
and non-slums across 110 sub-locations in Nairobi................117
6.2 Descriptive statistics on the modeling variables...................121
6.3 Pearson correlation test among the modeling variables across 110 sub-
locations in Nairobi................................................126
6.4 Modeling results by using 2SLS -Two Stage Least Squares Regression
7.1 Modeling child mortality rate against slum incidence and residential
isolation of slums from non-slums

The urban population explosion underway in the less developed world means
that its urban growth will account for roughly 90 per cent of global population
growth in the first 25 years of the third millennium (United Nations, 1998). By
2025, more than half the population of the less developed world will reside in
urban areas (Brockerhoff & Brennan, 1998). Some of the highest urbanization
rates, often exceeding five per cent per annum, have beat witnessed in sub-
Saharan Africa over the last three to four decades (Todaro, 1989; Obudho,
1997). According to United Nations (1998) projections, the majority of the
population of sub-Saharan Africa will be urban residents by 2016. In addition
to the growing number of mega-villages with little or no health advantage
over rural areas, the number of cities with over one million inhabitants has
increased from 0 to 30 since 1950 (Brockerhoff & Brennan, 1998).
The world is becoming increasingly urban, which means that more poor
people will live in urban areas than ever before. Without any doubt, another
trend in parallel with the rapid urbanization process is the urbanization of
poverty, especially in developing countries and Sub-Saharan states. Urban
poverty as an enduring developmental problem has been intensively

investigated and debated not only among scholars, but also within political and
policy circles (Lipton, 1976; Sen, 1985, 1987, 1999; Satterthwaite,1997;
Wratten, 1995; UN-HABITAT, 2003a; World Bank, 2001, 2004). Part of the
reason is the wide incidence and huge magnitude of urban poverty worldwide.
According to the World Banks estimates (1999), there were 495 million urban
poor who lived below the $1 a day poverty line as of the year 2000, and
relatively speaking, 20 percent of poor people lived in urban areas. By 2020,
the proportion is projected to reach 40%, and by 2035 half of the worlds poor
people are projected to live in urban areas (http://www.peopleandplanet. net).
Further, as the continuing urbanization process has accelerated especially in
the developing world, this problem is becoming much more acute and is often
characterized by the newly added urban population being short of shelter,
employment, infrastructure facilities such as water and sanitation, health
services, and educational opportunities (UN-HABITAT, 2003a, b).
The poverty issue is one of the most urgent challenges to the African continent,
especially to the Sub-Saharan African countries. There weak and faltering
macroeconomic performance, inordinate debt burdens, civil strife,
environmental deterioration and global warming aggravate the condition and
impede intervention. There too fertility rates continue to outpace mortality
rates yielding high population growth rates. And various forms of regional

malaise provoke huge population displacements often flooding already over-
burdened urban destinations.
In most places conditions are worsening (Sahn and Stifel, 2003). Absent
intervention the long-run secular descent of urban Africa is likely to continue
unabated. If the reasons are debatable, the need for intervention is not. Yet
intervention must proceed from an understanding of essential conditions.
However, most literature and anti-poverty practices focus on the monetary
side of poverty, centered on the income and consumption aspects of peoples
welfare (Haddad & Ahmed, 2003; Meng, Gregory, & Wang, 2005; Woolard &
Klasen, 2005; Sackey, 2004; Grootaert, 1997). This does not represent the full
spectrum of deprivations from which individuals or households suffer because
poverty is a multidimensional phenomenon not limited to economic well-
being (Sen, 1985, 1987, 1999; Ravallion & Sen, 1986; Baharoglu & Kessides,
2001; Satterthwaite, 1997; Wratten, 1995). In this paper, one of the non-
monetary dimensions of poverty material poverty is addressed. It will
reflect the housing conditions and the quality of basic services provided to
urban residents in each sub-location in Nairobi, Kenya, including water,
sanitation, and energy. This kind of perspective is not novel in terms of
conceptual and ideological thinking. As early as the late 1970s, the basic
needs view on urban poverty was raised, elaborated, and applied in anti-

poverty practices (Streeton et al., 1981; Stewart, 1985; World Bank, 2004).
This material view on poverty became widely accepted partly because of the
enormous health benefits gained by the urban poor due to the improvement of
basic services including water, sanitation and sewage. Also, this perspective is
very relevant to the poverty situation in Nairobi as illustrated later in this
The most obvious manifestation of material poverty prevalent among the cities
in the developing world is the slum. According to a recent estimation by UN-
HABITAT, 31.6% of the worlds urban population (924 million people) in
2001 could be categorized as slum dwellers on the basis of an operational
definition of slums: durability of housing material, access to safe water and
sanitation facilities, sufficient living area, and security of tenure (UN-
HABITAT, 2003a).1 In the next 30 years, this figure is projected to double to
almost 2 billion unless drastic policy changes are put in place to alter this
projection (UN-HABITAT 2003a). More specifically, sub-Saharan Africa has
1 UN-HABITAT in close collaboration with the United Nations Statistic Division and
the Cities Alliance organized an expert group meeting in 2002 and the purpose was to
reach a consensus on an operational definition fa* slum dwellers that would be
applied to monitor the Millennium Development Goal 7, Target 11 on improving the
lives of at least 100 million slum dwellers by 2020. As a result of the meeting, a slum
household is defined as a group of individuals living under the same roof lacking one
or more of the following conditions: (1) access to improved water, (2) access to
improved sanitation facilities; (3) sufficient-living area, not overcrowded; (4)
structural quality/durability of dwellings; and (5) security of tenure (UN-HABITAT,

the largest proportion, around 71.95 percent of the urban population resident
in slums. Although in terms of sheer numbers, Asia hosts the largest number
of slum dwellers (554 million or 60% of the worlds total in 2001), Africa is
rapidly becoming a continent of slums; 166 million out of a total of 231
million urban residents in sub-Saharan Africa are classified as slum dwellers
(UN-HABITAT 2003a). In most cities in the developing world, poverty and
slums are symbiotic, given that the slum areas are desperate places with
concentrated poverty and deteriorated living conditions.
An important developmental problem associated with urban poverty and slums
is the challenge of public health for the poor. Historically in sub-Saharan
African most of the attention was paid to the health problems facing the poor
living in rural area rather than those in the city. Yet, the unprecedented rate of
urbanization and the accompanying disproportionate growth in the population
of poor city residents pose new challenges for health care in the region. This
problem has become particularly acute among poor people in urban
neighborhoods. For example, according to the WHO(2002), unsafe water,
poor sanitation, and hygiene especially in impoverished neighborhoods and
slum areas are the main cause of 4-8% of the overall burden of diseases in
developing countries and nine-tenths of diarrheal diseases. In addition,
findings from the China National Child Health Survey (Wang, Hughes, & Fan,

2002) show that in urban areas, children living in neighborhoods with poor
access to flush toilets have significantly higher mortality risks, when
controlling for household socio-environmental conditions. Similarly, in urban
Peru, children who live in neighborhoods with poor access to sanitation have
obviously lower nutritional status, measured by height for age (Alderman,
Hentschel, & Dabates, 2001).
In the literature it is well-recognized that the health outcomes of individuals
are heavily affected by the neighborhood environment in which they live
(Dannenberg, et al., 2003; Ellen & Mijanovich, 2001; Ellen & Turner, 1997;
Institute of Medicine, 2000, Pickett & Pearl, 2001; WHO, 2002). In reality, we
often observe enormous disparities in health outcomes between poor and rich
regions or countries. Statistics show the fact that maternal mortality rates in
middle-income countries are a fraction (one-seventh) of what they are in low-
income countries (Fay, et al., 2005). Moreover, there are considerable
variations in health status between and within cities. Child mortality is a
notable example. In cities well served by piped water, sanitation, drainage,
waste removal and good health care, child mortality rates are generally around
10 per 1,000 live births, and few child deaths result from water-related
diseases. In cities or neighborhoods with inadequate provision, mortality rates
are commonly 10 to 20 times higher. In a well-managed city, there is little

difference in mortality rates for low- and high-income areas; in a badly
managed city, they can vary by a factor of 10, 20 or more. Surveys in seven
settlements in Karachi found that infant mortality rates varied from 33 to 209
per 1,000 live births (Hasan, 1999).
Although child mortality in Africa as a whole has decreased markedly over the
last 40 years, the pace varies a great deal between areas. The overall situation
in sub-Saharan Africa has improved since the 1950s or 1960s, but progress is
slower than that of other regions in the world and appears reversible or
tenuous in the current context of economic crisis, poverty and the AIDS
pandemic (Tabutin & Schoumaker, 2004). For example, in some informal
settlements in Nairobi, where around half the citys population lives, under-
five mortality rates were more than double the average for Nairobi, as well as
significantly higher than the Kenyan rural average (APHRC, 2002).
Urban poverty, slums, and child mortality are the central topics in this
dissertation, with their transcendent importance that has attracted a lot of
attention from the international community. For example, since 1945, housing
policies in developing countries have aimed to address the slum in particular
and poverty in general by using the instruments of subsidized public housing
(1945-1960s), site-and-services programs (1972-1980s), and market enabling

strategies (1980s-present) (Harris & Giles, 2003). Recently, a series of global
commitments such as the Social Summit of 1995 in Copenhagen, HABITAT
II in Istanbul, Copenhagen +5 in Geneva, and the Millennium Summit 2000
in New York have been made to fight poverty world wide, with specific
targets for slum eradication and child mortality reduction by 2015.
The massive efforts that have been made to deal with these urgent challenges
can be justified by looking at how ineffective governmental interventions to
address the problems have been. Using child mortality as an example,
approximately 10.5 million children aged less than five years die each year in
less-developed countries, mostly from readily-preventable or treatable
conditions (Ahmad, Lopez, & Inoue, 2000). However, recent work has
demonstrated that health spending is extremely ineffective in reducing infant
or child mortality, which is instead mainly explained by a countrys income
per capita. Findings indicate that much health spending in developing
countries may be poorly targeted or otherwise ineffective (Hanmer, Lensink,
& White, 2003). Furthermore, despite the ability of some interventions to
reduce mortality, the coverage of such interventions is still too low and the
delivery of services is not sufficient (Steketee et al, 2003). As a result, poor
children are disadvantaged in terms of exposure to and resistance to disease,
coverage levels for preventive interventions, and use of health care when they

are sick. For many governments, the major issues are not only to reduce
mortality, but to ensure that interventions to reduce mortality also reach poor
Key Research Questions
In this dissertation, I intend to address three key research questions. First, how
do we measure material poverty, a non-monetary dimension of urban poverty
that focuses on housing conditions and basic infrastructural services provided
to the urban poor? Second, what are the most significant determinants that
affect material poverty formation in sub-city areas? Here we focus our
attention on the impact of environmental variables such as distance to city
center or density on determining local poverty level. Third, what are the health
consequences of living in a slum or segregated community? Here the child
mortality rate is used as a proxy to represent the health welfare of a place.
This study emphasizes the adoption of an intra-city perspective to investigate
poverty, slums, and health disparities. Traditional statistical reporting systems
often treat the city as a homogenous entity, which therefore ignores the
fundamental fact that the city comprises spatially heterogeneous units with
vast variations in social, economic, and heath conditions. This dilemma is due
in part to the lack of relevant datasets that are spatially disaggregated enough
to observe the immense differences within the city. For most developing

countries, this kind of spatially fine dataset is very rare; because those that do
exist are often confidential, most outside researchers do not have an access to
them. In the context of Nairobi, not only are there significant variations in
living conditions between sub-city areas, but there also appears to be a
significant dichotomy between slums and non-slums, and amongst slums
themselves. Intensive examination of these variations will form the
foundation for an appraisal of the broad classes of public policies now being
deployed to ameliorate poverty and to eradicate the most grievous slum
Significance of the Study
This study will illuminate three key implications for anti-poverty or anti-slum
First is about how to measure the poverty at a spatially disaggregated level.
Answering this question is very important in that it gives people a snapshot
about where the poor resides and how the poverty is distributed across the
space. Using this piece of information, the decision makers can therefore
tackle the spatial dimension of poverty through for example, the geographic
targeting program. Second, to better meet the practical needs of attacking
poverty, this poverty measurement should be comprehensive enough to
capture the full range of deprivations the poor suffer, and be empirically

sensitive to reflect local variations. As discussed later, the current anti-poverty
practice is mainly constrained by using narrowly defined monetary-based
poverty indicators and pre-determined yardsticks such as one dollar
consumption line per day to differentiate the poor from non-poor. Because of
this pre-judgment and narrow-minded thinking, the resulted measurements
provide misleading and biased information in terms of estimating the depth
and extend of poverty. As a result, the effectiveness of any anti-poverty
programs or policies is diminishing and the cost is increasing. Therefore, my
dissertation work has a pragmatic value in that it devises an analytical routine
that uses Census dataset, the most widely distributed and accessible dataset in
developing countries, to produce the material poverty index which is
methodologically reliable and empirically sensitive to local conditions. It also
fills a void in the literature since no precedent systematic effort has been made
on this made on this specific aspect.
The second implication is centered around the debate on the determinants of
material poverty. Basically I try to answer questions like what accounts for the
poors plight? Are cause and cure symmetrical or is the way out different from
the way in? (See Ravallion and Wodon, 1997). Actually there are volumes of
literature on poverty determinants; however most of them aimed to analyze the
determinants of income /consumption poverty by using household survey data.

In contrast, my focus is placed on the determinants of material poverty.
Traditional research has put much attention to die demographic, social, and
economic factors related to poverty formation. My work is different in that it
intends to isolate the effect of geographic variables such as density and
distance to the city center from others on local poverty level. Through this
research, I hope to discover some new evidence to support areal-based anti-
poverty or anti-slum practice. More specifically, the variable of central interest
is population density. If the modeling results support my hypothesis that
population density impacts local poverty level, then local authorities may
conceive a solution to alleviate the poverty situation by de-densifying slum
areas, where the most concentrated urban poverty resides, through
redevelopment efforts.
The third implication is about discerning the significant factors correlated with
public health status operationalized as the child mortality rate. The central
variables are material poverty index and residential segregation of slum
dwellers from non-slum dwellers. This research is aimed to explore the
impacts of the aforementioned variables on local child mortality rate by
controlling other social, economic, and demographic effects. Findings from
the analysis will help to, at least from a health perspective, assess the

effectiveness of using a new policy prescription for slum eradication: de-
segregating slum dwellers from non-slum dwellers.
Overall, any policy prescriptions aiming to create lasting improvement surely
emerges from empowerment of poor households. This is a result that flows
from both policies focused on household themselves, as well as those aimed at
the houses and the neighborhoods they occupy. Besides this policy
categorization, Hirschman (1958) mentions another dichotomy: SOC versus
DPA, which are acronyms for Social Overhead Capital versus Directly
Productive Activities respectively. Even though he was speaking about
economic development, there is still a lesson here. According to my view, the
primary challenge of urban poverty in Nairobi, according to my view, is to
build a strong economy, generate working opportunities and income, and then
channel the benefits of rising personal incomes and rising government tax
receipts, into useful investments in both SOC and DP A. Sometimes the
question is not about absolute growth in economy, but the justice of the social
redistribution system. In another word, we may inquire: does the growth really
benefit the poor or do the poor really enjoys the spillover effect of economic
development1 On most occasions it doesnt. To solve this problem it is
important to enable the market to work. Using housing as an example, the
housing market could play a constructive role in promoting housing supply

and meeting the housing demand of different socio-economic classes.
Meanwhile, the market is not omnipotent. Markets have their own weakness in,
for example, ensuring social fairness. Therefore, it is the responsibility of the
public sector, private sector and grass-root community organizations to
collaborate with each other to correct the market drawbacks and mitigate the
housing deficiency problem, especially among the urban poor.
Structure of the dissertation
The remaining part of the dissertation is organized as follows:
Chapter 2 reviews three groups of literature, including the concepts and
measurements of urban poverty, poverty determinants analysis, and research
on the neighborhood effects of residential segregation and urban poverty on
public health status, with an emphasis on local child mortality. Chapter 3 is a
description of the study area: Nairobi, Kenya. In this chapter I will briefly
document the urban history of Nairobi city since its inception, discuss its
poverty and slum problems, and summarize its public health status, especially
the child mortality incidence between slums and non-slums within the city.
Chapter 4 explains the main data sources that are used in this dissertation.
More specifically, I will elaborate on three major datasets: the Kenyan
population and housing census in 1999, the slum EAfenumeration areas)
survey, and the GIS database for 110 sub-locations in that city. In addition, the
spatial organization of the city and the unit of the analysis for this research

will also be discussed. Chapter 5 documents the methodological routine of
constructing a material poverty index to measure the physical deprivations at
the sub-location level in Nairobi city. Specific data processing steps include
recoding and spatial aggregation. With the aid of principal component analysis
(PCA), I construct the composite material poverty index, with the results
mapped and validated by the slum EA survey data Chapter 6 uses the 2 Stage
Least Square (2SLS) model to identify the most significant determinants of
local material poverty, in particular, the impact of population density on local
material poverty level. Chapter 7 uses a dissimilarity index to measure the
residential segregation by slum dwellers and non-slum dwellers. Also, it
explores how the poverty level, measured by the material poverty index, and
residential segregation by slum dwellers and non-slum dwellers impacts local
child mortality rates. Chapter 8 summarizes the major findings from the
previous analyses, and infers the policy implications of those findings for anti-
poverty and anti-slum strategy-making and local practice.

Area I: Concepts and Measurements of Urban Poverty
The literature on poverty is so vast that it is impossible to list. This is partly
due to the fact that there are various definitions of poverty, according to
income, consumption in general terms, food consumption, food ration, calories,
medical data, basic needs, among other criteria (Glewwe & Van der Gaag,
1990; Sen, 1999; Sen & Foster, 1997). After reviewing relevant literature, I
categorize the conceptual thinking on poverty into five groups: (1) the
opulence approach; (2) the utilities approach; (3) the capability
approach; (4) the entitlement approach; (5) and the developmental
The opulence approach to defining poverty essentially argues that peoples
command over commodities can be assessed by measuring income and
expenditures. If the income or expenditure level for a household or person is
below some predetermined threshold, this household or person is then defined
as poor. Examples of these definitions include the official poverty line set
by the U.S. government and the 1 U.S. dollar per capita per day consumption
line made by the World Bank, respectively.

Another relatively small number of scholars have considered poverty that
reflects the notion of utility, which a measure of satisfaction speaks for a
good thing. In short, these researchers define poverty as a certain welfare level
of utility and seek to determine how much income various family types need
to obtain that utility. This approach to poverty was of particular interest in the
1970s and 1980s (e.g., Colasanto, Kapteyn, & van der Gaag, 1984; Danziger,
van der Gagg, Taussig & Smolensky, 1984; Goedhart, Halberstadt, Kapteyn,
& van Praag, 1977; Hagenaars, 1986; van Praag, Split, & Van de Stadt, 1982).
A third general approach to thinking about poverty involves functioning and
capabilities, based on the work of Townsend (1979) and Sen (1981, 1985,
1995, 1999), and extended by Desai (1995). There poverty is identified as a
condition arising out of capability deprivation. Building on the basic needs
approach first outlined by Pigou (1920) and later by Streeton et al.(1981) and
Stewart (1985), Sen (1982,1985,1999) has argued that poverty is the result of
a lack of at least one of the basic conditions (or skills) that are prerequisites to
an effective life. In this sense, we are interpreting a lack of water to be
consistent with a lack of one of these basic prerequisites, but lack of water will
have many additional repercussions.

A fourth view on poverty is called the entitlement approach. This approach
explains deprivation not in terms of an overall lack of resources, but in terms
of households ability to command such resources by means of mechanisms
called entitlements (Amis & Rakodi, 1994). The contribution of the
entitlement approach to poverty is the recognition that poverty is more than
mere lack of resources, particularly income. It also contains important
elements such as exposure to and risks from sudden change (vulnerability).
Hence recent poverty analysis has focused on the concept of vulnerability, the
distinction between long-term trends and short-term shocks, and the
importance of assets.
The last category of poverty thinking, I termed here, is the developmental
approach, which is generally held by the developmental agencies such as the
World Bank or UN branches to serve the needs of practical developmental
activities. This type of thinking is not necessarily exclusionary to the previous
categories, although it has its own unique features. For example, the Asian
Development Bank (ADB) defines poverty as a deprivation of essential assets
and opportunities to which every human is entitled (ADB, 1999).
Essential assets and opportunities are further defined as
access to basic education and primary health services...the right to
sustain themselves by their labor [i.e., access to employment

opportunities], and ... having some protection from external shocks
[i.e., access to social protection], and, importantly, [a right to]
participate in making the decisions that shape their lives (ADB,
The ADB definition fits well with the conceptual framework on poverty and
human development established by the United Nations Development Program
(UNDP) over the past decade (UNDP, 1990, 2000). UNDP defined a human
development index combining measures of longevity, literacy, and infant
mortality, which is complementary with the income-oriented measures used by
the World Bank and other bilateral international agencies. More recently,
UNDP has added a stronger emphasis on improved governance and
participation by the poor as key factors in overcoming poverty (UNDP, 2000).
Moreover, we have observed a changing path in terms of poverty thinking by
the World Bank. Moving away from its traditional view on poverty that has
mainly dealt with monetary aspects, World Banks current position, based
partly on the results of extensive consultations with poor people around the
world, is to view poverty from the perspective of the poor themselves (World
Bank, 2001). It argues that the most fundamental failure in poverty analysis is
the lack of involvement of poor people in the definition of poverty and in
determining what should be done to reduce it (Satterthwaite, 1997). The
advantage of such perspective is to enable a better understanding of the needs

and priorities of poor people instead of arbitrarily making assumptions and
plans by the decision-makers themselves.
In line with the various ways of thinking about poverty, there exist alternative
measures to gauge the level of poverty. I divide them into monetary versus
non-monetary categories. As monetary indicators, income and consumption
are the most representative to measure peoples welfare level. The non-
monetary category includes measures to gauge the insufficient outcomes of
individuals or households with respect to health, nutrition, and literacy, and
with deficient social relations, insecurity, and low self-esteem and
powerlessness to meet basic needs. In practice, people sometimes use
composite indices to gauge the multiple dimensions of poverty by taking into
account income, health, assets, and education information.
Household income, especially per capita income, is commonly used as a
welfare indicator, yet there are criticisms of its use in defining poverty. The
theoretical objection is that in many developing countries, the incomes of a
large portion of the population may vary from one year to another. Thus, a
given year's income may not match a household's average level of welfare as
generated by the consumption of goods and services over time. Alternatively,
household consumption expenditures is a particularly attractive welfare

indicator because of its sound theoretical basis that is derived from welfare
economics to define poverty according to the individuals utility or welfare
functions. In practical use, theoretical considerations call for the use of
household equivalence scales to adjust income or consumption for household
size, although calculating such scales is controversial (Pollack & Wales, 1979;
Deaton & Muellbauer, 1986). This controversy is often avoided by using
either total consumption or per capita consumption.
Some economists (for example, Anand & Harris, 1985) propose per capita
food consumption as a measure of welfare. One advantage is that the data
requirements are fewer. Second, food consumption may be more accurately
reported in household surveys than non-food consumption. Third, it is often
easier to construct food price indices than to make price comparisons for non-
food items. Yet the major strength of this method, its focus on food alone, is
also a serious weakness: the consumption of non-food necessities such as
clothing and shelter is ignored. Moreover, the measure depends directly on a
household's propensity to consume food, which may vary across households.
One set of indicators that may have practical relevance comprises medical
indicators of health and nutritional status. These include anthropometric
measures to determine the incidence of stunting (low height for age) and
wasting (low weight for height), as well as medical tests (see examples cited in

Berg, 1987). These measures are particularly relevant when focus is on the
adverse effects of poverty on young children. While it is fairly easy to obtain
height and weight data in household surveys, other medical data may be
difficult to gather. In addition, using medical data to define poverty may be
misleading, because although health is correlated with household welfare, it is
not identical to it. If one finds that poverty, as measured by consumption data,
is highly correlated with health, it may be helpful to use medical data to
identify the poor (Berg, 1987).
Another approach to determining the economic well-being of a household is
the basic needs concept (Streeten, 1981; Stewart, 1985). Rather than
determining a households total consumption, or accepting a proxy measure
for this concept, households are defined as poor if their food, clothing, medical,
educational, and other needs are not met. Such needs are exogenously defined,
for example, by groups of experts on nutrition, health care, shelter, and so on.
There is usually no attempt to aggregate these various aspects of basic needs
into a single welfare indicator, which complicates the classification of
households as poor and nonpoor. A second problem with this approach is the
subjectivity involved in determining adequate levels of health care, housing,
education, cultural amenities, and so on.

After reviewing the literature presenting alternative definitions and
measurements on poverty, it is clear that poverty is a multidimensional
phenomenon that affects different aspects of peoples well-being. However,
most prior studies and practice have paid a lot of attention to the monetary side
of poverty using income or consumption data This mistreatment, as correctly
pointed out by Desai and Shah (1988), implies that poverty is a
unidimensional construct, when, in fact, it is multidimensional. Besides, this
understanding of poverty from a monetary perspective dominants the literature
and disconnects itself with the widely used anti-poverty instruments in the
field that are mostly in-kind benefits and services (e.g., food stamps,
Medicaid, work supports and services) rather than direct cash transfers.
To better meet the needs of my study which aims to depict and analyze the
physical dimension of poverty in Nairobi, I emphasize the notion of material
poverty (i.e., the inadequate and poor quality consumption of very basic
goods and services such as housing, water sources, sanitation, energy), a
relatively under-investigated issue in previous literature. Three are three
reasons for doing so: first, material poverty is empirically related to the
situation of the poor in Nairobi, which are reflected in the scene of the huge
magnitude and wide distribution of slums characterized as deprived housing
and living conditions. Therefore, studying material poverty provides an

alternative way and forthright to investigate poverty in that city. Further, I
assume the dynamics and determinants of material poverty are different from
those of income or consumption poverty, which is proved by the empirical
data analyses as elaborated later.
Second, study of material poverty is a good starting point for carrying out
responsive policy analysis on poverty and then deploying anti-poverty
activities on the ground. In fact, any anti-poverty and anti-slums programs in
Nairobi that have been deployed by the international donor agencies or local
governments are focusing on the in-kind support to the poor, which includes,
for example, slum upgrading or basic services provision. Therefore, my
analysis and findings will aid those anti-poverty activities with a better
targeting on deprived areas, a better understanding of what kind of
deprivations the area suffers from, and a better knowledge about the dynamics
and determinants of material poverty formation within in the city.
Third, in terms of the conceptual thinking on material poverty, it is more close
to the developmental approach of poverty, as elaborated above. The basic
needs approach that emerged late in the 1970s is also related to material
poverty (Streeton et al., 1981; Stewart, 1985). The case study of Nairobi can

enrich our understanding of this fundamental notion with more empirical
evidence derived from relevant data analyses and logic reasoning.
Area II: Poverty Determinants Analysis
In the literature, there are several strands of thinking trying to explain the
mechanisms and dynamics of poverty formation, especially in the developing
countries. In the 1960s, Lewis (1961) raised the culture of poverty thesis, in
which the poor were assumed to be marginal to urban development because of
innate and culturally determined personal characteristics and resulting deviant
behavior (Wratten, 1995). Later, most explanatory arguments on poverty
formation fall into two oppositional categories: poverty is either attributed to
the personal failings of the individuals concerned, which could be termed the
individualistic model; or urban poverty is viewed as the inevitable outcome of
an unfairly structured political and economic system which discriminates
against disadvantaged groups, called the structural or geographical model.
The latter argues that there exists a causal link between geography and the
level of well-being, which is supported by the empirical evidence that
geographic capital does indeed have a strong relationship with poverty
(Galster & Killen 1995; Jalan & Ravallion, 1997; Ravallion & Wodon, 1997).
However, literature on the general theoretic framework of poverty mechanism
may not be capable of capturing the essential forces that drive the local
poverty formation. Further, the synergy of the causal factors is very

complicated to estimate: they may be either one of these individual or
structural factors, or a combination of them (Crump, 1997). Therefore,
additional research on poverty determinants about specific cities or regions is
highly necessary and worthwhile, and it may shed light on local policy in
terms of eradicating poverty.
In the literature, there are a substantial number of articles devoted to poverty
determinants analysis targeting different areas in the world (Ahmed et al.,
1991; Khundker et al., 1994; Rahman et al., 1995; Ravallion & Sen, 1996;
Ravallion & Wodon, 1997; Wodon, 1997, 1998; Mills & Pemia, 1994;
Grootaert & Braithwaite, 1998). However, in the past, fewer studies have
investigated the determinants of poverty in Sub-Saharan African countries
(Fofack, 2002; House, 1991; Glewwe, 1991; Coulombe & McKay, 1996), and
even fewer focus on poverty in urban areas in these countries (Glewwe, 1991),
although many urban residents living in this area have suffered various kinds
of deprivations of housing, basic services, health care, education and so on.
Part of the reason for this dilemma is the scarcity of data sources available.
Due to confidentiality considerations, the relevant data sets may not be
accessible to most researchers out of the country. By taking advantage of the
Kenya Census dataset, I intend to remedy part of the deficiency in the

In addition, most previous works can be categorized as monetary-based
poverty analysis using either income or consumption-based indicators
(Haddad & Ahmed, 2003; Haddad, Ruel, & Garrett, 1999; Meng, Gregory, &
Wang, 2005; Wooldard & Klasen, 2005; Sackey, 2004; Mukheijee & Benson,
2003; Grootaert, 1997; Coulombe & Makay, 1996; Moller et al, 2003; Defirna
& Thanawala, 2004; Klasen, 1997; Quisumbing et al, 2001; Hardy &
Hazelrigg, 1993; de Janvry & Sadoulet, 2000; Murdock, Zhai & Saenz, 1999).
However, this type of approach is not comprehensive enough to capture the
full spectrum of deprivations from which the poor suffer. The income or
consumption-related variables that are mostly taken as proxies to measure
peoples poverty level artificially narrow the necessary scope of understanding
poverty; it is a multidimensional phenomenon not limited to economic well-
being (Baharoglu & Kessides, 2001; Satterthwaite, 1997; Wratten, 1995).
Besides the aforementioned monetary-based research, there exist a number of
studies focusing on other dimensions of poverty such as food poverty
(Kyereme & Thorbecke, 1991; Hassan & Babu, 1991) or shelter poverty
(Rutty, 2005; Grimm, Guenard, and Mesple-Somps, 2002). However, the
concept of shelter poverty has a different meaning than the concept of
material poverty that will be examined in this study. For example, shelter
poverty discussed in Kuttys (2005) paper means housing non-affordable to
the household, which is often measured in economic terms. Grimm et al.

(2002) propose the concept of existing living conditions poverty which is
close to the meaning of material poverty used in this context. However, there
exist some important distinctions. Living conditions poverty in Grimm et al.s
(2002) work refers to the overcrowding of the surveyed households plus no or
poor access to diverse basic commodities (e.g., water, electricity, toilets, fuel).
In my study, material poverty mainly deals with the inferior quality of housing
materials and other basic infrastructural services, and the overcrowding of the
households. Second, to calculate the living conditions poverty index, Grimm
et al. (2002) compute a score based on the feedbacks of the sampled
households, which is a subjective process. For my study, I will use the
coefficient values derived from the principal component analysis to calculate
the material poverty index, and this methodology is more empirically reliable
than Grimm et als (2002) approach.
The material perspective is not novel in terms of conceptual and ideological
thinking. As early as the late 1970s, the basic needs view on urban poverty
was raised, elaborated, and applied in anti-poverty practices (Streeton et al.,
1981; Stewart, 1985; World Bank, 2004). This material view on poverty
became widely accepted partly because of the enormous health benefits gained
by the urban poor due to the improvement of basic services including water,
sanitation and sewage.

Most datasets to study poverty determinants are collected from household
surveys, such as LSMS (Living Standard Measurement Study) data, a multi-
topic household survey from the World Bank applied in many developing
countries in the world, or a single-topic household survey like the welfare
monitoring survey deployed by the Kenyan government (CBS, 2003). The
strength of using a household survey is that it provides detailed information on
the total population by using a representative sample, and it covers different
domains of questions for a single person, thus deepening the understanding of
the scope and severity of deprivations experienced by urban poor. The 5
percent Census sample data is the major data source that I will use in the
Nairobi case. However, this survey may be questioned for their representative
and generalization power. It will not have these problems because the original
information has been collected from the total population and then sampled in a
demographically and geographically-balanced way.
From a methodological perspective, most prior research has focused on the
individual person or household as the basic unit of analysis. The behavior
patterns and dynamics of poor people are explored and analyzed to help the
effectiveness of human-centered poverty alleviation policies or programs
including safety net, training programs, food stamps, etc. In the field, there
exists another kind of aid practice which targets the impoverished places,

through, for example, slum upgrading or site-and-service provision. In contrast
with human-centered practices, this category of activities may be termed
area-based and characterized as geographic targeting. Previous research has
shown that geographic targeting is most effective when the geographic units
are quite small, such as a village or district (Baker and Grosh, 1994; Bigman
& Fofack, 2000). Therefore, exploring the causal factors that impact the
poverty formation of smaller geographic units is a worthwhile effort, the
causal mechanisms should be different from that in the human-centered
poverty analysis. After reviewing relevant literature, I found that fewer studies
on poverty determinants used the areal modeling technique (Benson,
Chamberlin & Rhinehart, 2005; Longley & Tobon, 2004; Rodriguez &
Arriagada, 2004; Enchautegui, 1997). For example, in Longley and Tobon
(2004)s work, an income measure is aggregated at the Enumeration District
(ED) level. Then with the aid of geographically weighted regression (GWR)
and an instrument approach, the spatial dependence, heterogeneity, and
possible endogeneity between dependent and some independent variables have
been accounted for in order to obtain more accurate coefficients of those
independent variables related to the dependent variable of poverty level of a
small geographic area My research is similar to Longley and Tobon (2004)s
work in that we both use an instrument approach to solve the endogeneity
problem between dependent and some independent variables. However, in

Longley and Tobon (2004)s case, the independent variables that have an
endogenous relationship with the dependent variables are social-economic
predictors. For my case, the endogenous predictor is the population density of
a sub-location, and through an instrument approach, I will capture the
recursive interaction between poverty level and density of a place. Other areal
modeling work done by Enchautegui(1997) focuses on the racial poverty of a
place, which is also different from my attention to material poverty in Nairobi.
Overall, my dissertation is innovative in that it takes the sub-location as the
basic unit of analysis, and examines the correlates to real poverty formation
especially geographic or environmental regressors, which I will use to inform
area-based anti-poverty policies or programs.
Area III: Residential Segregation, Urban Poverty, and Public Health
Alternative Measures of Residential Segregation
Residential segregation as a key topic in social science has attracted a lot of
attention in the field. More specifically, residential segregation by race in
metropolitan areas has been intensively documented and investigated (Glaeser,
Scheinkman & Shleifer, 1995; Massey & Denton 1993; Jargowsky, 1996,
Sethi & Somanathan, 2004). There are other important types of segregation as
well such as segregation by socio-economic status (Benabou, 1994),
occupational segregation (Benabou, 1993), educational segregation by schools

(Saporito, 2003), and housing segregation (Glaeser, Hanushek & Quigley,
In addition to the study of various segregation categories, there are heated
discussions on how to measure segregations especially among sociologists.
The debate has ebbed and flowed, and for a time the issue seemed settled. In
1955, Otis Dudley Duncan and Beverly Duncan published their seminal paper
demonstrating that there was little information contained in any of the then-
prevailing indices that was not already captured by the index of dissimilarity
(DI). For 20 years thereafter, this measure was employed as the standard index
of residential segregation.
The dissimilarity index aims to compare the distributions of two population
groups over discrete spatial units of an urban area, with the formula expressed
as follows:
DI = Ik=i "Tk | Pk-P | / [2T (1-P)].................eq. 2.1
where T is the total population size in city, Tk the total population size in unit
k, P the proportion of minority in city, and Pk the proportion of minority in
unit k.

This index expresses the number of relocations necessary to transform a
segregated distribution of complete segregation into a uniform distribution.
Based on this interpretation, the dissimilarity index is referred to as a "ratio of
efforts to achieve desegregation" (Jakubs, 1981; Morgan, 1983). The ratio
approaches 1 if the observed distribution represents a totally segregated
pattern. The ratio is close to 0 if the observed distribution resembles a uniform
distribution Typically in the urban context of the U.S., a dissimilarity index of
less than .3 is considered low, an index between .3 and .6 is considered
moderate, and an index above .6 is considered high (Massey and Denton,
The index of dissimilarity is attractive because it has minimal data
requirements, is easy to calculate, and has clearly defined upper and lower
boundaries. The Duncans (1955) suggested that the index of dissimilarity was
the most useful of these indices and it has, in fact, achieved preeminence in the
measurement of segregation. Later, Taeuber and Taueber (1965) have
presented what is probably the most detailed discussion on the measurement
of segregation to date. In practice, the dissimilarity index has been employed
extensively at the census tract and block level for the study of racial
segregation (Taeuber & Taeuber 1965; Farley & Taeuber 1968; Sorenson,
Taeuber, & Hollingsworth 1975; Van Valey, Roof, & Wilcox 1977) and at the

census tract level to describe occupational and industrial segregation in cities
(Duncan & Duncan 1955; Simkua 1978).
However, the literature has repeatedly alluded to several weaknesses of the
dissimilarity index such as its reliance on the uniform rather than Poisson
distribution to characterize the absence of segregation (Winship, 1977; Falk et
al., 1978; Kestenbaum, 1980), its reliance on proportions rather than absolute
population sizes (Peach, 1981; Lieberson, 1981), and its insensitivity to the
exchange principle (Winship, 1977).2 From a geographers' point of view, the
most disturbing shortcoming of the dissimilarity index is its insensitivity to the
arrangement of spatial units. A checker-board pattern of high and low minority
proportions (negative spatial autocorrelation) yields the same value for the
dissimilarity index as a clustered pattern (positive spatial autocorrelation).
Theil (1972) pointed out additional weaknesses of the index of dissimilarity: it
was not aggregation consistent, and it was limited to dichotomies.
But even if the dissimilarity index can be used to measure, for instance, the
proportion of blacks located in particular neighboihoods related to whites, and
In probability theory and statistics, the Poisson distribution is a discrete probability
distributioa It expresses the probability of a number of events occurring in a fixed
time if these events occur with a known average rate, and are independent of the time
since the last event.

we know that blacks are disproportionably distributed in some area, it is not
clear whether the blacks and whites have contact there (Blau, 1977). To
measure the isolation of blacks from whites, or to know how much interaction
there is between the races, a new indicator called the isolation index was then
created (Bell, 1954). The formula to calculate this index is therefore:
Index of Isolation = [£i=i (Black; /Black)* (Black/Total 0 (Black /Total)] /
[Min(Black/Totali, 1)- Black/Total]....................eq.2.2
where Total is the total population size in city, Black is the total black people
in city, Total i is the total population size in unit I, and Black is the total black
people in unit i.
If blacks and whites are equally distributed throughout the city, this measure
would be exactly the share of the city that is black. To separate out the size
effect of the black population from its distribution, we subtract the percent
black in the city as a whole. When there are low numbers of blacks in the city,
it will be impossible for blacks to be completely isolated from whites. As to
the denominator in this formula, White (1986) suggests always dividing by
one minus the black share in the citys population. While these two algorithms
will get the identical result for most modem cities, in a city with very few
blacks the min measure should be more accurate (Cutler, Glaeser & Vigdor,

Besides the standard dissimilarity and isolation indexes, there are other ways
to measure dimensions of residential segregation (Taeuber and Taeuber, 1965;
and Massey and Denton, 1988; Glaeser and Scheinkman, 1997). Douglas
Massey and Nancy Denton (1988) undertook a systematic analysis of some 19
segregation indices identified from a review of the extant literature. They
argued that segregation is not a unidimensional construct, but encompasses
five distinct dimensions of spatial variation. The five dimensions they
identified are: evenness, exposure, clustering, concentration, and centralization.
To verify this conceptualization, Massey and Denton carried out a factor
analysis of indices computed using 1980 Census data for U.S. metropolitan
areas. Their results showed that each index correlated with one of five factors
corresponding to the dimensions they postulated. On theoretical, empirical,
and practical grounds, they selected a single "best" indicator for each
dimension of segregation. The dimensional structure of segregation and
Massey and Dentons selection of indices have been reaffirmed using 1990
census data by Massey, White, and Phua (1996). Recently published work by
Wilkes and Iceland (2004) suggests this dimensional organization of
segregation remains valid for the year 2000. In addition, the empirical work
done by Cutler, Glaeser, and Vigdor (1999) suggested that dissimilarity,
isolation, and clustering are all highly correlated (correlation coefficients
over .72), and concentration and, particularly, centralization are less highly

correlated with the other measures when using the 1990 Census data for U.S.
metropolitan areas.
Poverty, Residential Segregation and Public Health
The second category of literature under this topic is about the neighborhood
effects of poverty and residential segregation on public health, particularly on
local child mortality rate. This issue has attracted a lot of attention in the
sociology and public health areas. It relates to a collection of literature
regarding neighborhood effects, which investigates whether the social and
physical environments have a positive or negative impact on individuals
outcomes and well-being overall.
Over 40 years ago, Kenneth Clark described inner-city ghetto communities,
isolated by poverty and racial segregation, as beset by a self-perpetuating
"tangle of pathology," which affected their residents' physical, emotional, and
mental health (Clark, 1965). Contemporary social theory holds that
concentrated-poverty neighborhoods have serious and lasting consequences for
their residents and that, all other things being equal, poor children who growup
in high-poverty neighborhoods will experience significantly worse outcomes
than poor children in more affluent communities (Brooks Gunn et al., 1993;
Jencks & Mayer, 1990; Sampson, Raudenbush, & Earls, 1997). Though they
come to different conclusions about the creation of concentrated-poverty

neighborhoods (see also Jargowsky 1996; Quillian 1999), both Wilson (1987)
and Massey and Denton (1993) agree that living in a neighborhood with
concentrated poverty has serious consequences above and beyond those of
growing up in a poor family because of the absence of role models, social
isolation from job networks, weakened social institutions, and other factors.
There have also been significant studies aiming at investigating the
relationship between residential segregation and public health (Bird &
Bauman 1998; Kennedy et al 1997; LaVeist 1992; LeClere, Rogers & Peters
1997, Polednak, 1996). Polednak (19%) used the index of dissimilarity that
measures the degree of black-white residential segregation based on census
block group data He argued that high rates of black infant mortality in highly
segregated areas might be related to the concentration of extreme poverty and
poorer neighborhood quality (e.g. inadequate high-density housing, crime,
noise, and psychosocial stresses). LeClere and colleagues (1997) found that
the mortality gap between blacks and whites was eliminated when the
concentration of minority population, community social-economic status (SES)
variables, and individual SES variables were included in the mode. Kennedy
and colleagues (1997) studied collective discrimination and mortality. They
defined collective discrimination as a lack of respect one group displays
toward another. A lack of respect was inversely correlated with state median

income level. They found that disrespect predicted age-adjusted black
mortality rates and white mortality rates.
There is also existed a lot of literature regarding the neighborhood effects of
poverty and residential segregation specifically on child mortality rate. Studies
show that those living in poor urban settlements face some of the most
difficult environmental conditions, and child mortality and morbidity rates in
these settlements equal or exceed those in rural areas. Research from five
communities in the Republic of Congo, for example, found the prevalence of
diarrhea was 3.5 times greater for urban than for rural children, and that the
rural-urban variable was more significant than socioeconomic, demographic
or behavioral factors (Mock et al., 1993). Studies comparing rural and urban
areas in Egypt, Zimbabwe and Malawi have also found a higher prevalence of
intestinal parasites and worms among urban children, especially those in poor
neighborhoods (Mason, Patterson & Loewenson, 1986; Curtale, et al., 1998;
Phiri, et al., 2000). High concentrations of people and wastes in urban areas
create more opportunities for exposure to pathogens, and a correspondingly
greater need for the levels of hygiene that adequate water and sanitation make

Moreover, literature on the determinants analysis of child mortality rate
suggests that there are some common factors that negatively affect child health:
(i) geographical location in rural or urban poor settlements; (ii) the low
standard of living of households in terms of assets; (iii) some community
elements, in particular morbidity, the insufficiency of vaccination and the
absence of childbirth assisted by qualified persons (Lauchaud, 2004). Also the
analysis on mortality determination shows that at the national level access to
electricity, incomes, vaccination in the first year of birth, and public health
expenditure significantly reduce child mortality (Wang, 2002). Overall, these
research findings reach the same conclusion that poverty and residential
segregation really have a negative impact on local child mortality rate, given
that residential segregation can be taken as a proxy factor to represent the
extent of accessing health care facilities and resources in developed or rich
However, there still exist a series of research issues in terms of exploring
neighborhood effects on health outcomes, despite numerous studies have been
done in this area First is how to define a neighborhood. As Yen and Syme
(1999) argued, much of the existing research done to describe the social
environment of a neighborhood has been indirect and suggestive. For example,
it is of interest to describe the environment by noting the percentage of people

living in crowded circumstances or in single family dwellings, but these
statistics are very crude measures of group life or community circumstances.
Although these statistics do describe the sort of people who live in an area and
their circumstances, they do not describe the richness of life in the
neighborhood or other important phenomena, such as crime occurrence, local
resources, or social cohesiveness.
Second, what are the mechanisms that relate the neighborhood or social
environment with health outcomes? The vast majority of studies have
examined the social environment in relationship with health results, but few
studies have considered the underlying mechanisms between them. One way
of approaching this issue is to examine health behaviors. Several studies have
now reported that people living in poor-quality social environments are more
likely to smoke and to have less physical activity and higher-fat diets (Blaxter,
1990, Diehr et al., 1993, Diez-Roux et al., 1997). It is important to pursue
these and other mechanisms so that we can better understand the social
environment and appropriate interventions.
Third, the difficulties in capturing neighborhood effects on individuals
outcomes using observational data are largely but not entirely the result of
selection bias: unmeasured factors that affect both neighborhood of residence

and individual outcomes that could potentially account for the association
between neighborhood poverty and segregation level and outcomes (Jencks &
Mayer, 1990). Thus, many results derived from this type of research have been
inconclusive and therefore social scientists have been unable to convincingly
demonstrate the neighborhood effect.
Exploring the connection between poverty, segregation, and health outcome
has important implications for prevention practices in reality. There are two
ways to think about prevention. An individual focus directs attention to the
fact that people need to change their behavior to lower their disease risk. Thus,
we develop smoking cessation, weight loss, and stress reduction programs. In
these programs, the onus is on individuals to change their way of life. For this
programmatic approach to have widespread impact, individuals must be
interested in the programs, and they must have the time and financial
resources to enable participation.
The second way to think about prevention is from the environmental
perspective. From this perspective, programs could be designed at several
different levels, including place-based or structural levels. A place-based
approach, for example, would include such activities as regulating billboard
advertising of cigarettes and alcoholic beverages, developing zoning laws to

include public recreation areas near residential areas, creating bike lanes, and
establishing community policing programs. Structural approaches to
prevention would include legislating speed limits, price-support policies for
food, alcohol, or tobacco, and cigarette taxes.
It has been verified in practice that the development of health-related
prevention programs that focus on places or structural dimensions can
influence the lives of more people and for longer periods of time than
individually-based interventions (Yen & Syme 1999). The effectiveness of
those programs, however, needs to be derived from a thorough understanding
of the effects of neighborhood elements on health outcomes. My research
focusing on the case of Nairobi will provide a robust and empirically reliable
foundation to support the making of those programs by addressing the
research issues mentioned above. For example, I define the neighborhood
environment by calculating the proportion of slum EAs out of total number of
EAs in a sub-location. I intend to examine not only the effect of living in a
slum area or living in a segregated area characterized by the isolation of slums
from non-slums on local child mortality rate.

Brief History of Nairobi
Nairobi in the local language Swahili means white highlands and it has
experienced significant growth since its inception in 1906, as elaborated below
(See Fig. 3.1) (Obudho & Aduwo,1992). The city sits between 5,000 to more
than 6,000 feet (1,500 to over 1,850 meters) above sea level, rising from the
Athi Plains in the southeast of the Kikuyu Plateau to the northwest. The lower
regions are fairly flat while the upper regions are characterized by a ridge and
valley landscape with steep slopes.

Established in 1896 as a caravan trading depot, Nairobi grew quickly after the
Mombasa Uganda railway reached it in 1899. During the founding period of
1899-1905, Nairobi became an official town when the colonial railway
authorities decided to make it as their headquarters. As time passed, it became
the capital of the British colonial government with 18 square kilometers of
previously ownerless land prior to that, the city was no more than a
grazing area for the pastoral Maasai. By 1905, Nairobi had a population of
over 10,000 people, and it has since remained Kenyas largest city, the
political and administrative capital, and the primary center of commerce,
finance, industry, education, and communications.
Just like other capitals in Africa for example, Harare in Zimbabwe and
Lusaka in Zambia Nairobi can be seen as initially a typical European city. It
primarily was a place for Europeans to live in and grew with little regard for
any preexisting pattern of housing and transportation network. At that time,
Africans were considered temporary workers who supposedly had permanent
homes in the rural areas. The Municipal Native Affairs Department,
established in 1927, stipulated that almost anybody not working for, and
therefore not housed by, the railways was considered to be unwanted urban
population. Thus, only those European colonists and Asians (mainly from
India and Pakistan) who initially arrived to help build and staff the railways

were the formal urban residents. Nevertheless, there was a small but steady
influx of Africans into Nairobi every year, and they soon began to open more
and more shops that catered to the growing administrative and settler
community in the city. Later, with the continued arrival of the wives and
children of the railway workers, as well as a number of single women, it was
not surprising that temporary dwellings began to spring up in the city because
the European-styled city characterized by narrow streets and limited number
of formal housing units could not meet the demand of the newly added
population members. As a result, the unwanted immigrants from the rural
areas plus the new arrivals formed the first generation of slum dwellers in
Coincidently, slum clearance as a policy strategy to address the increasing
number of slum dwellings in the city was soon adopted by the city government.
The official demolition of 120 shacks, as reported by the Medical offer of
Health in 1931, can be regarded as the first recorded abuse of the Public
Health Act, used by the colonialists and later by the independence Kenyan
authorities to justify slum clearance. Concurrently the implementation of the
Vagrancy Act of 1922 (the first had been enacted in 1902) meant not only that
unauthorized huts could be demolished, but also that any African found in
Nairobi without a job was liable to be identified as a potential criminal, who

ought to be repatriated to the so-called native reserves. By 1931 the African
population of the capital had risen to 27,000, of whom 3,375 were convicted
of vagrancy (Hake, 1977).
In 1963 Kenya won independence from British colonial government and
Kenyatta served as the first President of Kenya until his death in 1978. In the
Kenyatta era, the relaxation of the previously harsh measures taken against the
influx of Africans in Nairobi, in particular, and to other towns in general, helps
to explain why the urban population increased significantly in capital. By 1971
the population of Nairobi was estimated to have reached 500,000, one-third of
who were living in unauthorized housing (Hake, 1977). Many of those who
had recently arrived from the rural areas could not find accommodation in the
officially-recognized residential estates, such as Bahati and Kariokor, and so
they were allowed at least during the early years of Kenyattas era to put up
shacks in open spaces that were not too close to either the central business
district or the city square.
These soon became eyesores for the Kenyan leaders, who were trying hard to
prove that they could maintain law and order especially in the overcrowded
capital. So the honeymoon was soon ended, and the old colonial policy of
slum demolition was reinstated with much vigor and wrath, yet still officially

justified by the Public Health Act. However, there were relatively few slum
demolitions in the mid-1970s, partly because of the reduced rural-urban
migration during the so-called coffee boom.
After the death of Kenyatta in 1978, Daniel Moi took over the presidency and
publicly declared that he would follow the first Presidents footsteps
(Nyayo), but soon made clear that he wanted to be viewed as an
understanding, down-to-earth man, and a good listener. The political climate
in the late 1970s and early 1980s did not favor either the clearance of slums or
the harassment of those engaged in small economic enterprises. However, the
worst types of slum clearance were officially authorized in 1990, all too
reminiscent of the colonial period. Many Kenyans were both alarmed and
Statistics indicate that between 1971 and 1995, the number of informal
settlements and slums within Nairobi rose from 50 to 134, while the estimated
total population of these settlements increased from 167,000 to some
1,886,000 individuals (UN-HABITAT, 2003a). The size and densities of these
settlements vary from a few hundred people to hundreds of thousands of
people. Kibera, the largest of these settlements which is estimated to have a
population of around 300,000 is considered to be the largest slum in Africa,

if not the world. Today, both natural growth and rural-to-urban migration
continue to contribute to the growth of Nairobis slums and informal
settlements. These slums provide a large proportion of the formal and informal
labor force in the city and, it has been argued, play a useful role in providing
cheap housing for those who cannot or will not spend any more (Mumtaz,
2001). Cautiously, slum clearance happens from time to time, and raises a lot
of political controversial between slum dwellers and the local government who
often implement slums evictions (http://www.peopleandplanet.netA.
Poverty and Slums in Nairobi
Nairobi exemplifies the rapid urbanization amidst deteriorating economic
conditions that characterizes most big African cities. With an annual growth
rate of 7 percent over the last two decades, Nairobi is one of the fastest
growing cities in Africa (APHRC, 2002). Since the 1960s, Nairobis
population has increased six fold, from 350,000 in 1962 to about 2.1 million
people in 1999 (CBS, 2001). The city area has also been greatly expanded
from 25 km2 in 1920, to 82 km2 in 1940, and up to 684 km2 in 1995 up to the
present (UN, 1995). Today there are 194 urban centers in Kenya, with 45
percent of the urban population residing in Nairobi (GOK, 2001). With an
urban primary index of 2.6, Nairobi has continued to develop as a prime city
in Kenya, based on the Eleven-City Index of urban primacy (GOK, 2002).

A clear consequence of rapid urbanization amidst economic deceleration in
Kenya is that poverty is increasing among urban residents. The proportion of
people who cannot afford minimum food and non-food basic requirements
increased from 45 to 52 percent in Kenya between 1992 and 1997. While rural
poverty increased from 48 to 53 percent, the proportion of people living in
absolute poverty in urban areas increased from 29 to 49 percent (CBS, 2000).
Also for Nairobi, the levels of poverty from 1992 to 1997 are 26.5 percent and
50.2 percent respectively (CBS, 2000, 2001). The urban poor therefore
constitute the majority of the approximately 32 percent of Nairobis
population that is economically inactive (ROK, 2002). With the population of
the city expected almost to grow at high speed, the outlook for the city and its
residents is bleak. Similar trends are apparent in many African countries. Even
where rural poverty has been declining, the proportion of people below the
poverty line has been increasing in urban areas (World Bank, 2001).
Urbanization is largely characterized by deepening disparities between the
wealthy and the urban poor. The wealthy consume a disproportionate share of
resources while the unemployed and working poor are crowded into slums,
under worsening conditions. Nairobi is no different from many cities in this
respect as evidenced by the situation. According to estimates, a large
proportion of Nairobis urban population, around 40 percent of the city's

official total population of 2.1 million people, lives in slums and informal
settlements and about two-thirds are surviving on less than one dollar a day
(APHRC, 2002; Deng & Turkstra, 2004). With an annual growth rate of 5
percent, the municipality will host 5 million people by the year 2020, of whom
nearly 3 million will live in informal and often precarious settlements, if
current trends continue (UNCHS, 2001) (See Fig. 3.2). Slum dwellers have
limited access to health and educational services and to basic amenities.
Opportunities for steady employment are scarce, and very low wages are paid
to those fortunate enough to find work or engage in informal trading. Overall,
life in Nairobis slums is not easy by any standard. Although each of these
slums has its own socio-economic, political and ethnic characteristics, they
share a common theme: very high population density. From Fig 3.3, it can be
observed that there is a vast difference in terms of population density between
slums and non-slum areas (Syaggal et al., 2001). Sixty percent of the
population (including slum dwellers, low income households, and others) in
Nairobi occupy only 8.7 percent of all land used for residential purposes
(Deng & Turkstra, 2004). According to the 1999 census, the overall
population density for Nairobi was 3,079 people per km2 (CBS, 2001).
However, the wealthier parts of the city are much less dense, being inhabited
by about 300 persons per km2, while the slum areas are burdened with 75,000
persons per km2 (Alder, 1995). More strikingly, in the slum area as many as

1200 people live on one square hectare in the slum areas, most in shacks as
small as 10 by 10 feet.

The tremendous growth of Nairobs urban population, which has resulted from
both natural population growth and rural-urban migration, has led to an
increased demand for resources required to meet the consequent demand for
infrastructure services (Olima, 2001). However, the provision of basic urban
services has not kept pace with the rapid growth of the city. According to
estimates, only 45 percent of the citys residents had access to potable water
and only 63 percent had access to regular waste collection in 1993 (World
Bank, 1999). The city produces 1000 tons of solid wastes daily, but the

collection service is limited, and in only 20 percent of this solid waste is
collected and taken to designated dumpsites (Mitullah, 2003). Further, slum
residents are much worse off when compared to residents of other urban areas
in terms of their access to services and amenities. According to the Nairobi
Cross-sectional Slums Survey of 2000 and the Kenya Demographic Health
Study of 1998, the proportion of slum households in Nairobi having access to
pipe water is much less as compared to the whole of Nairobi (Wasao, 2002).
Consequently, most of the residents of informal settlements (75 percent)
purchase water for domestic use (Wasao & Bauni, 2001), which is confirmed
by other studies showing that slum residents tend to pay significantly higher
charges for water services, representing major fiscal burdens for the urban
poor (Aligula, 1999). In a word, Nairobi hosts some of the most dense,
deprived and insecure slums in the world.
Residential segregation in Nairobi
Accompanying the rapid growth of urban poverty and slum proliferation is the
extremely uneven distribution of land and resources allocated to different
socio-economic classes who often times live in an isolated and segregated
mode. Residential segregation is an original phenomenon in Nairobi as well as
other urban centers in Kenya Its origin in the Kenyan towns can be traced
back to the emergence of colonization. The major influence that set the frame
of the city was racial segregation, a practice that prevailed as late as the early

1960s (Desouza, 1988). With the aid of biased planning laws favoring certain
racial groups as well as exclusionary zoning regulations, the city was divided
into four distinct sectors along certain racial lines: North and East defined as
the Asian Sector (Parklands, Pangani and Eastleigh), East and South East
defined as the African Sector (Pumwani, Kariokor, and Donholm), South East
to South marked another small Asian enclave before it was bounded by the
Game Park (Nairobi South, and Nairobi West), and North and West marked
the European area Some scholars termed this phenomenon as racial
tripartition of the city made up of Europeans, Asians and Africans (Salau,
However, the spatial dimension of segregation went beyond the mere
separation of racial groups. The containment policy that complimented the
implementation of residential segregation by race resulted into gender
segregation where women and children were not allowed into the city and
occupational segregation where entry of the unemployed idlers was
restricted (Mabin, 2005). Apart from racial segregation, the colonial regime
also attempted to separate Africans into ethnic enclaves; resulting in ethnic
segregation (Christopher, 1994; Macoloo, 1998).

More recent experience indicates that economic capacity and income level
take over the traditional position of racial influence and play a more
significant role in shaping contemporary residential segregation. In Kenya,
especially after independence in 1963, affluence seems to be intensifying in
the former non-black areas upon abolition of state segregation, while poverty
and deprivation continues unabated in the irregular settlements (Macoloo,
1998). Consequently, the disproportionate consumption of urban space
continues to be in favor of the higher income groups just as it used to be
during the colonial period (Macoloo, 1998). This proposition is in agreement
with Kingoriah (1980)s observation that land ownership and choice of
residential areas in Nairobi by individuals were based largely upon the
economic ability to acquire land and housing. The lowest income group, who
after independence were migrating in large numbers into the city from the
countryside, became squatters. Europeans remained in the high-income
brackets and most of them live in the former European residential areas. The
high-income sector of the African population, mainly the better-educated and
more successful businessmen, joined them in upper Nairobi. The middle
income Africans joined the Asian and lived in Parklands, Eastleigh and
Nairobi South. The low-income groups were confined to the sprawling
Eastlands and the informal settlements. Virtually no European or Asian lived

in the Eastlands area. Thus the residential areas got differentiated mainly in
terms of income status.
Spatial segregation of poor people often occurs within slum areas. Most slums,
particularly those on public land, have extremely high densities resulting in
high congestion levels. Typically, there are 250 units per hectare in such
settlements as compared to 25 and 15 units per hectare in middle income areas
and high-income areas, respectively (Republic of Kenya, 1997). Sixty-six
percent of the poor population live in rental accommodation, often spending
40% of their disposable income on rents alone (KAkumn and Olima, 2007).
The substantial research in these areas report low level of nutrition, expensive
food, and high cost of cooking fuel including charcoal. The slums and squatter
areas also reveal the highest levels of disease, the lowest levels of
immunization and the highest levels of infant mortality.
Residential segregation has its various causes and consequences. KAkumn
and 01ima(2007) argue that it may occur for two main reasons: social
prejudice (state activated) or the malfunction of an economic system (market
activated). Social prejudice mainly occurs in racial or ethnic forms. According
to Valins (2003), segregation may also arise due to religious intolerance. A
malfunctioning economic system that does not promote social equity is also a

major factor. For example, spatial polarization of the first-order may lead to
spatial polarization of the second-order. Brueckner et al. (2002) have
demonstrated that separation of work places (say industrial land use) from
living place (residential land use) may transform into social problems when
workers with high commuting costs decide to setde in ghettos. On the other
hand, socioeconomic segregation may lead to the exclusion of the
disadvantaged from access to environmental or public goods (Sellers, 1999;
Schultz, Williams, Israel, & Lempert, 2002). The causes of residential
segregation are deeply interconnected such that causes arising from social
prejudice are quickly translated into economic ones. In the same vein, cause
and effect occur in a rapid and interconnected manner. Residential segregation
is therefore a dynamic urban phenomenon whose consequences may appear
completely look unrelated to the causes.
Residential segregation has exposed residents, particularly women and
children, to severe environmental health risks which critically affect their
ability to play a full economic role in the life of the city. Here the women
retain high levels of fertility (an average of 8+ per woman in squatter
settlements compared with decreasing average levels in Kenya). There are
poor levels of family planning and high incidences of teenage pregnancy.
Typically, there are also inadequate health care facilities (both primary health

care and hospitals), even though this typifies the city of Nairobi as a whole
(Lillis, 1992). Not surprisingly, educational enrolments are lowest in such
areas. Up to 60% of children in squatter areas do not attend pre-school;
nursery schools are available in middle and high-income areas but are
expensive and, even where the NCC operates nursery schools, for example, in
Mathare Valley, they are beyond the affordability (or comprehension) of
squatter families.
Socio-Political-Economic Structure and Process
The continued expansion and proliferation of slums in Nairobi is attributable
to the rapid growth rate of the citys population, the poverty of the inhabitants,
a deficient national housing policy framework, and the insubstantial,
inefficient and corrupt systems of urban governance.
Nairobi has been faced with the challenges of planning and development of its
urban neighborhoods to house the massive influx of low-income immigrants,
especially after its independence in 1963. According to Muwonge (1980),
during colonial rule the housing shortage was not as severe because the
government controlled the migration of Africans. However, after 1963,
Nairobis population rose sharply. Before independence, the population
growth rate in the city between 1948 and 1962 was 6%, and this rose to 9.3%

between 1962 and 1969. Africans migrated to the city in large numbers, thus
creating a housing crisis.
Notwithstanding the ever-increasing population pressures in urban areas over
the years, the government continued directly to plan a majority of the
neighborhoods in Nairobi. However, due to lack of funds, the inefficiency of
land utilization, the fragmentation of urban development patterns, and the
reliance on out-moded planning ideologies, housing strategies geared towards
the provision of complete housing units to the urban population have not
achieved the expected results. For example, the planning of Nairobi was based
on the neighborhood concept which was a strong planning tool adopted by the
colonial government. The driving force behind the use of the concept was the
recreation of a sense of belonging to a neighborhood. This attribute was
manipulated by the colonists to promote the stratification of residential
developments on a racial basis (Chana, 1974). The Africans were located to
the east of the city centre close to their working areas. Therefore this eastern
part of the city has become the area where most of Nairobis low-income
groups reside (Tuts et al., 1989). In stark contrast with African reserves in the
eastern and outskirts of the city, European and Asian neighborhoods occupy
the center and western part of the city (van Zwanenberg, 1972). Amis (1990, p.
86) reports that the effects of these racially segregated residential areas

continue to affect current population density and distribution in the city:
Nairobi has inherited an extremely unequal land distribution from colonial
racial residential planning such that in 1972 there were eight inhabitants per
acre in the ex-European zone, 32 in the Asian Zone, and 400 in the African
zone. By following this planning ideology, it creates an obvious
development paradox: suitable land for housing the poor is not available. Two
major reasons for this shortage are under-utilization of public residential land
and hoarding by both the rich and the local authority itself for speculative
Other significant factors that make the city fail to curb the housing shortage
and to meet the housing demand of the poor are attributed to rigid building
standards, the neglected needs of the poor, and non-affordable units for the
poor(Etherton, 1971; van Zwanenberg, 1972; Muwonge, 1980; Gome'z et al.,
1984; Kayongo-Male, 1988; PLD, 1996; El Partido, 2000). In addition, rapid
rural to urban migration, unemployment and shortage of low-cost housing
units have also contributed to the housing crisis (Werlin, 1974; Muwonge,
1980; Lee-Smith, 1984; Pons, 1994).
Due to the inability of the public sector in Kenya to meet the ever increasing
need for shelter provision, the 1980s witnessed the emergence of land buying

companies (LBCs), which have attempted to radically change Nairobis
residential landscape. This has been a completely new phenomenon of
residential neighborhood planning and development and involves the purchase
of land parcels, sub-division of plots and the carrying out of approved
residential developments on the site. Between 1980s and 1990s LBCs
acquired large portions of Nairobis urban land and planned it for private
residential estates and related purposes. The LBCs work closely with Nairobi
City Council (NCC) and Commissioner of Lands Office while undertaking
residential developments. Some LBCs plan and develop houses for sale while
others sell planned and serviced plots. Along with the rise of LBCs is a retreat
of the public sector. It has of late completely withdrawn from the provision of
housing. Instead its approach is infused with the precepts of the so-called
Global Shelter Strategy and Habitat II Agenda. These favor enabling
strategies in shelter provision. These seek to elevate the capacities of
individuals and households to secure appropriate housing. Housing, in this
perspective, resides at the end rather than the start of the causal chain, an end
achieved by other means. But of course, quality housing, in the larger causal
progressions will necessarily emerge once again as a means for the
attainment of higher living standards.
Taxonomy of Anti-Poverty and Anti-Slums Policy Prescriptions

Like other cities in less developed countries (LDCs), the housing shortages in
Nairobi have been on going for several decades and attracted a lot of attention
within political and academic circles. For example, towards the end of the
1990s, national governments have laid out National Action Plans to tackle the
shelter problem. These approaches stress recommendations made in previous
studies, as well as global housing policies advocated by the UN and the World
Bank. In this section, I will categorize and analyze the spectrum of policy
solutions proposed by government to curb housing crisis and alleviate urban
Broadly speaking, all the housing and poverty-related solutions implemented
in Nairobi could be grouped into two clusters: top-down vs. bottom-up
approach. However, this way of categorization is arbitrary since each can
address both the material expressions of povertythe built environment
itself-as well as the capacities of individual households. Nevertheless, the
difference between them is obvious. Generally speaking, the former are mostly
the results of governmental action, operating within a governance hierarchy,
with approaches stressing macro-level strategies in development projects and
include the demolition of illegal structures, the provision of sites and services,
and the upgrading of squatter settlements, as discussed in the following
paragraphs. However, the second approach, in stark contrast with the former

one, emphasizes micro-level strategies. Because it recognizes the contributions
of women and the poor in development projects, this bottom-up approach is
also referred to as an enabling approach. It aims to create a supportive
environment for the poor, to be proactive, and to promote self-sufficiency. A
distinction one needs to make here is that the top-down approach does not
necessarily imply a focus on the supply side of the housing market. It can, as
in the U S. context, also focus on the demand side through various
empowerment strategies.
Top-down Approaches
During an earlier period of time, after Nairobis independence in 1963, the
government undertook a series of efforts to ease the housing shortage, most of
which could be viewed as constituting a top-down approach. For instance, the
government declared in its 1964-1970 Development Plan that low-income
urban housing and slum clearance would be tackled by providing rental and
home ownership schemes.
The Plan also reported that since the cost of a minimal two-room house was
too expensive for most low-income earners, the government was going to
introduce sites and services schemes. Towards the end of the 1%0s, the
Nairobi City Council formulated new strategies to deal with the housing crisis.
The City Council declared it would support the sites and services programs,

lower the building code, and introduce squatter upgrading schemes. Both sites
and services schemes and squatter upgrading programs got a boost from the
World Bank when it began funding housing projects in the early 1970s.
Following the Banks lead, other national and international agencies also
began to fund and support these projects (GOK, 1964; Muwonge, 1980;
HRDU, 1977).
Even though the slum upgrading and site-and service schemes became
dominant strategies for housing urban poor in Nairobi during an earlier era,
the efficacy of these solutions has been seriously questioned. At issue is their
cost and replicability (World Bank, 1993; UNCHS, 1996). This approach is
further limited by the ambivalent attitude of government regarding the
appropriateness of irregular settlements: in the eyes of the political elite,
the administrator, and the professional, upgrading is not attractive for political
display (Otiso, 2002). The upgraded areas are too low a standard to be good
show pieces (Syagga and Kiamba, 1992). These strategies have failed because
of a reliance on inappropriate building bylaws and infrastructural standards,
modem designs, construction technology, and conventional building materials
that make housing unaffordable to the poor, even after subsidies. Although
there have been calls to government to adopt more suitable alternatives, few
changes have taken place, because policy makers view proposed modifications

as being too old-fashioned... and incongruent with their quest to modernize
the country (Otiso, 2002).
Another important factor that has contributed to the dysfunction of top-down
approach is the seeming inability of hierarchical governing bodies in the larger
cities of LDCs to manage low-income housing projects. Classic examples of
such failed schemes in Nairobi include the New Pumwani Urban Renewal
Project and the Dandora Phase II Community Development Housing Project,
which became too expensive for most households. Only a small number of the
original residents could afford the new units. In addition, some of the
beneficiaries sold the new units and moved out of the area to establish slum
and squatter settlements in other parts of the City (Mclnnes, 1995).
In addition to slum upgrading and site-and-service schemes, the top-down
based solution space was further enriched by the adoption of IMF-World
Bank-sponsored structural adjustment programs (SAPs) since the early 1980s.
Proponents of SAPs think that the provision of housing and basic services (e.g.,
potable water and sanitation) to Third World urban residents, especially the
poor, is difficult because of: (1) poor planning, (2) weak municipal
governments, and (3) rapid urban growth from high natural increase and
excessive rural-to-urban migration rates (Otiso, 2002). Therefore, SAPs seek

to reduce urban expenditures, forcing states to withdraw from the provision of
urban social services such as education, sanitation, and health, and by
encouraging privatization of these services to improve efficiency (Aina, 1997).
SAP proponents believe that privatization is the key to efficient housing and
urban services for urban residents, including the poor who are both willing and
able to pay for private sector services (e.g., those provided by the informal
sector), and are already doing so. While this is true, it is undeniable that SAPs
have worsened the plight of the urban poor (Aina, 1997). Consequently, the
widespread belief among SAP advocates that the market is the solution to poor
peoples housing and service needs is exaggerated (Potts, 1997). Moreover,
SAP supporters ignore important questions pertaining to poor peoples unjust
expenditures on basic necessities, e g., potable water from informal sector
vendors (Nigam, 1999).
Although the top down approaches is sometimes controversial, it has recently
gained new momentum. But unlike past practice, wherein demolition of
informal setdements played a central role, current practice employs a blend of
approaches. As a result it has gradually become more acceptable, especially
since the early 1980s. Nevertheless episodes of demolition and resettlement
do still occur when alternatives might have been more appropriate (for

example, in Nairobi Muoroto and Kibagare in 1990, Mitumba in 1993).
Moreover, the government introduced new and drastic measures to ease the
housing shortage in urban areas by launching the Urban Renewal Programme
in 1987. The programs was designed to redevelop low-income housing areas,
build high rise flats and phase out both the sites and services schemes and the
upgrading of squatter settlements. By the early 1990s, Nairobi and other urban
areas were already experiencing the effects of these new policies, In 1990,
no units were completed under the site and service scheme, which in the early
eighties accounted for more than half of the total units completed by the
[NHC], The shift to Tenant Purchase, Mortgage and Rental schemes has
contributed to the phasing out of site and service schemes (GOK, 1991).
Bottom-up approach
During the 1980s, the UN launched the bottom-up approach in recognizing
the obstacles encountered under the top-down approach: namely displacement,
affordability, cost-recovery, and replicability as previously mentioned.
Differing from the top-down approach, it assumed that people did not know
their needs and that communities were unable and unwilling to pay for
services (UN, 1976; UNCHS, 1989, 1991; World Bank, 1993; Pugh, 1994;
Aldrich and Sandhu, 1995; UNCHS and ILO, 1995; Pugh, 1997). Advocators
of the bottom-up approach argued that most of the poverty and housing
policies failed because they did not engage the community or the poor in the

planning, decision-making and implementation processes (Pugh, 1994;
UNCHS and ILO, 1995).
As argued before, the key word for bottom-up approach is enablement, which
is defined below: Enablement is defined as providing the legislative,
institutional, and financial framework that help all actors engaged in the
housing sector to be more efficient (Pugh, 1994). Unlike top-down approaches,
enabling strategies do not begin by estimating housing needs but by listing the
measures to be taken to enable communities to help themselves (UNCHS,
Recently the Kenyan government has adopted enabling policies in its
economic reforms for 1996-1998 and National Action Plans which include:
facilitating the efforts of all actors in providing shelter; the use of non-
conventional building materials; and the upgrading of low-income housing
(GOK, 1996). It appears that the bottom-up approach is more likely to be
successful in overcoming factors that inhibit the provision of low-income
housing as compared to the top-down approach.
One of the most important accomplishments of the so-called bottom-up
approach is that it recognizes the special contribution of non-government

organizations (NGOs), community-based organizations (CBOs) and the
private sector to reducing poverty and providing housing assistance. Take
Kenya as a vivid example. Traditionally the main actor in urban housing and
service provision in Kenyan cities (and in many other African cities) has been
the state. Due to inefficiency, fiscal difficulties, poverty, and market failure, it
has failed to meet poor peoples needs, thereby creating a vacuum that has
largely been filled by voluntary sector actors such as NGOs and CBOs (Tripp,
1992; Wekwete, 1997). There have also been increased donor flows through
the sector because of the perception that it avoids many of the weaknesses of
the public and private sectors. Thus, the voluntary sector is not as market-
oriented as the private sector, is less bureaucratic compared to the state, and is
closer to poor urban residents (Cemea, 1988; Otiso, 2000; Wekwete, 1997).
These qualities have led to the belief that the voluntary sector is best suited for
the role of spearheading efforts to meet the urban poors many needs (Hotz,
Despite the euphoria accompanying the growth of the voluntary sector since
the 1980s, it is evident that like the other actors, this sector cannot single-
handedly meet the needs of poor urban residents because it lacks the capacity
to do so (Fowler, 1995). Moreover, the sector needs the inputs of the other
participants, especially the state, to be effective (Aina, 1997). Researchers

argue that trisector partnerships involving the state, the private sector, and the
voluntary sector are a promising way forward (Otiso, 2003).
Public Health in Nairobi
On the African continent two marked developments suggest a need for urgent
attention to an emerging urban crisis: rapid urbanization amid poverty (United
Nations, 1998), and the urban character of the HIV/AIDS epidemic (DesGrees
du Lou, 1999). Research suggests a linkage between these two trends, with
deteriorating economic conditions in urban areas increasing the likelihood that
women, especially adolescents, will engage in risky sexual behavior that
encourages the spread of HIV/AIDS (Ulin, 1992; Carael & Allen, 1995).
There is growing evidence suggesting that substantial segments of urban
Africa are actually even more disadvantaged than rural areas in health and
education (Brockerhoff & Brennan, 1998). Given the high rates of
urbanization on this continent which has experienced consistent economic
decline over the last 30 years, impoverishment is prevalent among the
exploding marginalized urban population, who may easily become slum
dwellers. Research indicates that the extreme deprivation of slums traps
residents into engaging in risky sexual behavior for economic survival (Zulu et
al., 2003). The underlying causes are the unemployment and inadequate wages
among the employed, both of which restrict slum residents ability to meet
personal needs and familial obligations. As other economic options run out,

economic desperation pushes them, especially women, to rely on sexual
relations to obtain sufficient money for rent, childrens schooling, and other
basic necessities, and many of them maximize the number of sexual partners
they have in order to increase their economic security (see also Akuffo, 1987,
Bassett & Mhloyi, 1991, Orubuloye et al., 1994).
Furthermore, a recent study examining differences in sexual behavior between
slum residents and non-slum residents in Nairobi shows that slum residents
start sexual intercourse at earlier ages, have more sexual partners, and are less
likely than other city residents to know of or adopt preventive measures
against contracting HIV (Zulu et al, 2002). Other findings from the same data
suggest that three features of slum conditions help to socialize children into
very early (including preadolescent) sexual intercourse: economic stresses, a
social context that is accepting of prostitution, and residential arrangements
that do not afford privacy for sexual intercourse within households (Dodoo et
al, 2003). Given the urbanization rate and the continents bleak economic
future, an increase in multiple sexual partnerships and a declining age at first
sexual intercourse are to be expected. As a result, these trends will pose
serious health risks, especially in view of the very low levels of condom use.

In addition to the challenge presented by sexual diseases to the urban poor in
Nairobi, childrens health is another important issue that needs further
attention. Rapid urbanization and inequitable distribution of social services in
African cities significantly contribute to the current deterioration of child
health indicators in the region. Children of the urban poor are continually
exposed to higher risks of fatal diseases due to inequitable income distribution
and access to social services. For instance, the prevalence rate of diarrhea in
children under 3 years of age in the informal settlements (slums) was more
than double that of Nairobi as a whole. The infant mortality rate (IMR) in
Nairobi slums (91/1000) is higher than in any other parts of Kenya Nairobi as
a whole had an IMR of 39/1000 while the rural areas of Kenya had a rate of
76/1000 (APHRC, 2002).
Overall, poor living environment, lack of water and sanitation facilities, food
insecurity, and cultural malpractices in childcare are the hallmarks of life in
the slums. Unemployment as well as irregular and unskilled livelihood
opportunities limits financial capabilities to avail nutritious food for children
and opportunity for quality healthcare at the household level. Slum residents
are also heterogeneous in some socioeconomic features. Not all of the slum
residents are poor and uneducated migrants from rural communities, even
though they all share similar living environments. Differences in income,

migration status, education and ethnic background may influence health
outcomes of an individual especially children. Evidence from Demographic
and Health Surveys (DHS) indicates that the urban poor in sub-Saharan Africa
have less access to health services, and consequently exhibit higher child
mortality rates than residents from other population sub-groups including rural
residents (APHRC, 2002; Caldwell & Caldwell, 2002).
Therefore, selecting Nairobi as my case in this study has global implications.
It is a city embedded in a context typical of the developing world: rapid
urbanization, decelerated economy, constrained budget, and enormous
infrastructure stress on urban residents. In addition, exploring the causal
mechanism of material poverty endured by the urban poor is crucial to local
authorities, given that infrastructure and service deprivation are widespread
problems for urban residents. Understanding the causal mechanism is a
prerequisite for local governments to take any responsive action to address the
worsening plight of the urban poor, especially slum dwellers.

Kenya Census, DHS and GIS
The main dataset used in this study is called The Fourth Population and
Housing Census of Kenya, which was implemented by the Central Bureau of
Statistics (CBS) of Kenya in August 1999.3 From the Census, a 5% sample
was drawn for Nairobi at random in such a way that the geographic
distribution characteristics of the sample are the same as for the Census: each
division, location and sub-location have approximately the same proportion of
households in the sample and the census.
In addition to the 5% Census sample data, another dataset used in this research
is related to the number of households living in slum conditions for each of the
4700 enumeration areas (EAs) in Nairobi. This information was obtained
through a joint exercise between the Central Bureau of Statistics (CBS), the
3 The 1999 Population and Housing Census, hereinafter referred to as the 1999
Census, was the sixth census to be carried out since 1948 and the fourth since
independence. It was carried out between the night of the 24th and 31st of August
1999. The 1999 Census was undertaken at the time when diverse demographic and
socio-economic data were required to ascertain the achievements made since
independence and the way forward to address current challenges and those envisaged
during the 3rd millennium. This commitment necessitated the establishment of an
elaborate census organization and administration to guarantee the collection of
comprehensive, accurate and complete information.

Department of Housing, the Nairobi City Council and UN-HABITAT in 2002.
The team used three indicators to identify those among the enumeration areas
in Nairobi that can be classified as slum EAs: access to water, sanitation, and
electricity; permanency and durability of structure; and security of tenure
(Harvey, 2002).
The third source of data is the sub-location (n=110) map for Nairobi city
provided by the Central Bureau Statistics (CBS) of Kenya. For each sub-
location, information such as the name, population, area, perimeters are
contained in a corresponding GIS database, which makes the spatial
visualization and presentation of analytical results possible. The GIS database
could be matched with the 5% Census data because after some transformation,
the two data sources could be joined by a shared coding system for each sub-
The last data source that was used in this study is called the Kenya
Demographic and Health Survey (KDHS), implemented in 1998. 4 The
purpose of incorporating this dataset in the study is to verify that the Census
4 With the assistance of Mr. Gora Moup from the GUO section in UN-HABITAT, I
obtained the Kenya DHS dataset for comparison use in this study.

data have an acceptable quality by comparing the paired categories of
variables in the Census and KDHS.
The database used in this study thus has more than 80 variables. The variables
are related to focus such as demographics, education, employment, housing,
and basic services. Within the category of basic living conditions and services,
there are variables related to the building materials of the dwelling (roof, wall
and floor), water sources, sewage types, cooking fuel, lighting type, slum EA,
and overcrowding (the number of persons per habitable rooms).
Spatial Hierarchy and Unit of Analysis
Nairobi is divided into eight divisions (Dagoretti, Embakasi, Kasarani, Central,
Westlands, Kibera, Makadara and Pumwani). These divisions are in turn
divided into 46 locations, 110 sub-locations, and 4700 enumeration areas
(EAs). Kibera is the largest division and includes the Nairobi national park;
the second largest division is Embakasi, which includes the international
airport. In addition, some notorious slum sub-locations such as Kibera,
Mathare, Korogocho are shown in Fig. 4.1.

Selected Locations
0 3 6 12
^ i Klometers
In the sub-locations that are categorized as slum areas, the population density
is very high: for example, sub-location Silanga, located in the Kibera division,
has a population density of 906 persons/ha, while Huruma, in the Central
Nairobi division, the population density is 871 persons/ha. One the other hand
though, Karen, in the Kibera division, has a population density of only 4
persons per ha. Karen is generally taken as the high-income living area on the
outskirts of Nairobi. These density figures verify to some extent that reaching
down to the sub-location level has some discriminative power in terms of
making sense of the vast disparities within the city. Therefore, I chose the sub-
location as the basic unit of analysis in this research. The data availability is

another reason for doing so. Sub-location information is the minimum level
that can be accessed at the moment and smaller spatial units such as
enumeration areas have not been digitalized in a GIS format.

Issues of Variable Selection: Relevance and Significance
As described before, this study chooses eight Census variables to represent
different dimensions of material poverty and they are wall, roof and floor
materials, water quality, sanitation, cooking fuel, lighting sources and
overcrowding. Why are these variables important and being selected? Besides
the reason of data availability, one of the critical considerations is that these
variables are conceptually important for us to measure material poverty. As
defined before, material poverty in this study means the inferior quality of
basic living conditions and infrastructural services provided to the urban
residents. By following this line of thinking, material poverty has different
meanings for different sectors such as housing materials, water and sanitation,
energy sources, etc, all of which are under the umbrella of living conditions
and basic infrastructural services. Therefore, the inclusion of above variables
is extremely important to characterize different dimensions of material poverty
by gauging the impoverishment among those different sectors. In addition,
another key consideration here is more related to the empirical relevance of
including those variables in the analysis. More specifically, I think that those
indicators depict a comprehensive picture of material poverty in the city by
describing the actual experience of living in poverty. In doing so, it is feasible

to make sound policy prescriptions derived from empirical analysis in order to
improve the life of the urban poor. Next, I will start to explain one-by-one the
applicability of above points for different sectors.
The quality of housing materials is a very important indicator of the durability
and permanence of housing, and the adequacy of shelter that housing affords.
Durability is a function of construction materials employed, most importantly
for roofs, walls and floors as well as the manner of their assemblage. Solid
structures are a barrier against the vagaries of weather and climate, and
criminal intrusions. Sturdy and lasting structures are also less likely break
apart, injuring occupants. And they are more likely to ensure a safe and
cleanly environment, decreasing air pollution, increasing ventilation, and
reducing exposure to chemical and other contaminants such as those
associated with pesticides. Tighter structures, moreover, block the spread of
indoor contaminants such as mold-laden air or insulation fibers.
Properly built structures can also improve interior light quality, control noise,
and enhance a sense of privacy and well-being. This latter is all the more
essential in crowded slum environments where neighbors share walls,
strangers traverse close-by walkways, and the din associated with mixed use
seriously compromises individuality. Finally, quality materials, well

assembled reduce the cost of maintenance, extend the life of housing
structures, minimize the ill-effects arising from the procurement of scarce
building materials from over-populated places, and lessen the need for costly
reconstruction or demolition activities. Related to these are the savings
realized by households as a result of efficiencies achieved in heating and
cooling. Whether we gauge such transactions in terms of monetary
expenditure or personal labor, the reduction in cost can be substantial,
enabling such savings to be devoted to other important familial purposes. Of
course, quality materials and sufficient building practices entail costs as well,
and these occur at the front-end of occupancy, potentially burdening
households having few expendable resources.
Durability of building materials is to a very large extent subject to local
conditions as well as to local construction and maintenance traditions and
skills. Even though some houses may be built with materials classified as
durable, the dwellers may still not enjoy adequate protection due to the overall
state of a dwelling. Alternatively, a material may not look durable, in the
modem sense, but is, in the traditional sense, when combined with skills of
repair. Such cases are vernacular housing made of natural materials in villages,
maintained by its residents annually. The observation of the building material
has therefore to be supplemented by an observation of the state of repair of a

house. Adequate shelter is thus operationalized in terms of building material in
combination with state of repair.
Water is an important livelihood asset for petty trading, especially of prepared
food, drink, beer, ice and ice products. It is also a livelihood asset for services
such as laundry, car-washing, and even car-window washing at stoplights.
Because of the precarious nature of these livelihood activities, any disruptions
to the water supply directly affect incomes. The point could be verified by the
fact, as observed in slums around the world, that without connections to the
main water supply system, the urban poor and slum dwellers typically have to
buy water, often of dubious quality, from enterprising water vendors that
charge anywhere from 5 2500% more per liter of water than what a
consumer connected to the mains supply would pay. Often, this water will
come after hours of waiting, arrive at inconvenient hours, and at designated
places from which the household supply would need to be carried home. As a
result, income of the urban poor and its economic capacity is further
compressed because of this extra buying.
Water quality is another challenge facing urban poor. Due to the generally
poorer quality water, the urban poor suffer the debilitating effects of water-
borne diseases like diarrhea, gastro-enteritis, cholera, thus requiring them to

spend on medicines and medical treatment, causing their children to miss
school, or causing the adults to lose income through missed working days. The
alternative for this is for the urban poor household to clean their drinking
water by boiling it, thus spending more on fuel.
The testimonies that follow, from peri-urban dwellers in five case studies, help
to understand their perceptions and experiences of water poverty. The
insecurity associated with practices such as illegal connections is a recurrent
factor highlighted by most interviewees, as illustrated by a peri-urban resident
in Caracas, Venezuela: here is where the water problem is most visible,
on Terrace 11. We have the connection to the pipe furthest away, on the main
highway ... we connected an illegal tap, but it doesnt meet our needs. Water
doesnt reach my house at leas We have no responsibility, some people
waste a lot of water, there are broken pipes and they arent repaired. While in
some areas, like Cairo, the poor quality of water is a central concern, in other
places the main problem affecting peri-urban communities is related to its
irregular and inadequate supply. A woman in Caracas explains how this
affects her life: When they give me water every fourth day, I dont do any
other chores, I just get water The next day I do all my chores, because
water takes a lot of your time, fetching water, filling bottles, checking that
there are no leaks. Peri-urban poor residents develop different coping

strategies to deal with irregular and insufficient water supply. A woman from
Milpa Alta (Mexico) explains: They give us water every third day at the tap,
by hour. They give us three or four hours a day and we organize by number on
a list of families. There are 17 families in this area, around 80 people including
children. We get half an hour of water each When your turn comes, you
grab the hose and connect it to the barrel. (Adriana, Julio and Pascale, 2006).
The lack of access to adequate and affordable sanitation services is another
key factor that directly impacts the health of the urban poor, and thus, their
capability to earn livelihoods. In most poor communities and slums in
developing world, sanitation facilities for personal hygiene are often non-
existent. If they have access to either communal facilities or the facilities of
their landlords, use of these facilities adds to household expenses. Open
defecation brings with it particular risks to the personal security of women and
children and more general risks of diseases to the whole community.
One the other hand, it is quite often to observe that stagnant water from poor
drainage and drainage canals and streams strewn and blocked with uncollected
refuse and sewage become breeding grounds for mosquitoes & other pests that
are disease carriers. Refuse heaps and open sewage are regular sources of
contamination of food and drink. And because of the density of housing in

many slum communities, it is easy for diseases to spread rapidly throughout
the whole community, lowering incomes, and in some cases, claiming lives.
Acceptable quality of lighting material and cooking fuel are indispensable
resources for maintaining a decent urban livelihood, especially for urban poor
and slum dwellers who are particularly hard hit by the lack of access to
modem energy sources. The urban poor suffer direct physical harm from
indoor air pollution. The urban poor spend a much greater share of their
household income on energy than higher income groups. They have smaller
and less predictable incomes, and their appliances are less energy-efficient.
This situation applies particularly to urban poor households headed by women.
Global evidence shows that most expenditure on energy services by the poor is
on fuels for cooking, while the remainder is spent on fuels or batteries for light,
typically in an 80/20 percent ratio. In general, fuelwood provides heating and
cooking for the urban poor at a higher cost than LPG. Likewise, kerosene
provides lighting for the urban poor at a higher cost than electricity. Both the
direct cost and opportunity cost of acquiring energy for the urban poor is
increased by having to collect fuelwood and buy charcoal and kerosene in
small amounts. And this burden is disproportionately borne by women and

A case study carried out in Nairobi by Gitonga (1999) showed that the main
fuels the urban poor households use were, in order of importance, charcoal,
kerosene and fuelwood. A notable aspect of the survey is that all households
visited that had a charcoal stove also possessed a kerosene stove. The reason
was that kerosene is used mostly in the morning for breakfast while charcoal is
used for the hard foods that take long to cook. A similar pattern is found in
most countries of the region. Overall, kerosene is an important fuel and is
ranked second due to the ease of use, affordability and its use for both cooking
and lighting.
Last, lighting has very important security implication. The residents are also
putting up external security lighting at the gates of their premises and on the
external walls of their buildings to be able to reduce crime levels and improve
the security within the neighborhood. Effective lighting increases visibility
and exposes the potential offenders as it improves on the chance of
identification and detection. Successful lighting, in addition, is a valuable
amenity. The residents have come to the realization that the quality of
neighborhood is determined by the communal lighting within the blocks of
flats as well as the surrounding blocks and lighting on routes into and around
the estate.
Data Processing: Recoding and Spatial Aggregation

Most of the Census variables are qualitative in nature with values in a
multinomial format, which is not easily managed in statistical routines.
Therefore, through the recoding process, the values of the original variables
are transformed into a dichotomous (0 vs. 1) category, representing good
quality or bad quality of living conditions, or slum EAs versus non-slum EAs,
respectively (See Table 5.1). The recoding also facilitates using spatial
aggregation procedure to calculate percentages of people living in a sub-
location enjoying bad or good quality of basic services. The recoding
procedure adopted in this study takes into account the operational definition of
slum dwellers set by UN-HABITAT (UN-HABITAT, 2003a).
Table 5.1 A description of recoding variables on living conditions
Variable Name Values Label Recoded Values Label
Roof material 1 Corrugated Iron Sheet 0 Good quality
2 Tiles
3 Concrete
4 Asbestos Sheets 1 Bad quality
5 Grass (Thatch)
6 Makuti *
7 Tin
8 Other
Wall material 1 Stone 0 Good quality
2 Brick/Blocks
3 Mud with wood 1 Bad quality
4 Mud with cement
5 Wood only
6 Corrugated Iron Sheets
7 Grass/Reeds
8 Tin
9 Other
Floor material 1 Cement 0 Good Quality
2 Titles
3 Wood
4 Earth 1 Bad quality

Tab. 5.1 (Cont.)
Variable Name Values Label Recoded Values Label
Floor Material 5 Other 1 Bad quality
Water sources 7 Borehole 0 Good quality
8 Piped
9 Jabias/Cristem Tank
0 Other 1 Bad quality
1 Pond
2 Dam
3 Lake
4 Stream/River
5 Spring
6 Well
Sewage types 1 Main sewer 0 Good quality
2 Septic tank
4 Pit latrine 1 Bad quality
3 Cess Pool
5 Bucket latrine
6 Bush
7 Other
Cooking fuel 1 Electricity 0 Good quality
3 Gas
5 Charcoal
2 Paraffin 1 Bad quality
4 Firewood
6 Solar
7 Other
Lighting type 1 Electricity 0 Good quality
2 Pressure lamp
3 Lantern 1 Bad quality
4 Tin lamp
5 Fuel wood
6 Solar
7 Other
Overcrowding 1 Less than 3 persons per room 0 Not overcrowded
2 3 or more persons per room 1 Overcrowded
SlumEA 1 Slum EA 1 SlumEA
2 Non-slum EA 0 Non-slum EA
*Makuti: palm leaves used as traditional type of tile for roof material.
The strength of the recoding process is that it provides a pragmatic way to
simplify the complicated dataset into a more manageable one. However, the