POVERTY, FOOD SECURITY AND WOMEN

EMPLOYMENT IN URBAN ETHIOPIA

(A Case Study in Bahar Dar City)

By
Tenaw Gedefaw

In Partial Fulfillment of the Requirement for the Degree of

Bachelor of Arts (BA) in Economics

Advisor
Ato Kumsa Dandena (MSc)

Submitted to
THE DEPARTMENT OF ECONOMICS
ARBA MINCH UNIVERSITY

JUNE, 2008
Arba Minch

1
 Acknowledgment

I would have been much more at ease, if only, words could explain my indebtedness
to those who extended their hands whenever I am in need of one.

In particular, my appreciation to my advisor Ato Kumsa Dendena is beyond any

scale of measurement. And, to my brother Simeneh Gedefaw and all the rest of my
family deserve to be praised and credited for their unfailing and unconditioned
material, moral as well as financial support not only for this paper but for the whole
completion of Economics study. Finally, but most importantly, I would like to thank
all those friends who shared me their concerns and ideas about my achievement, on
one way or the other.

Above all I would like to thank the almighty God; it will not be possible without
Him.

2
 Table Of Contents

CONTENTS------------------------------------------------------- PAGE
Acknowledgment---------------------------------------------------------------------------------------iii
Table of contents----------------------------------------------------------------------------------------iv

List of tables---------------------------------------------------------------------------------------------
vii

Acronyms-----------------------------------------------------------------------------------------------viii

Abstract---------------------------------------------------------------------------------------------------
ix

Chapter One…………….……………………………………………...1
INTRODUCTION___________________________________________________________________1
1.1 Background of the Study_____________________________________________________________1
1.2 Statement of the Problem_____________________________________________________________3
1.3 Objective the Study__________________________________________________________________4
1.4 Research Hypotheses_________________________________________________________________5
1.5 Scope of the Study____________________________________________________________________5
1.6 Significance of the Study_____________________________________________________________5
1.7 Limitation of the Study_______________________________________________________________6
1.8 Organization of the Paper____________________________________________________________6
Chapter Two.
………………………………………………........7
LITERETURE REVIEW____________________________________________________________7
2.1 Theoretical Literature Review________________________________________________________7
2.1.1 Food security and poverty: Definitions and measurement issues 7
2.1.2 Food adequacy, food security and protein-energy malnutrition 9
2.1.3 Characteristics, Causes and Measurement of Urban poverty 12
i. Characteristics and Causes of Urban Poverty_____________________________________12
ii. Identifying Poverty Lines and Poverty Measures_________________________________14

3
 2.1.4 Gender and poverty 21
2.2 Empirical Literature Review________________________________________________________24
2.2.1 The poverty profile of urban Ethiopia 24
2.2.2 Empirical findings concerning characteristics of the poor 25

Chapter
Three………………………………………………….27
METHODOLOGY_________________________________________________________________27
3.1 Nature and Source of Data__________________________________________________________27
3.2 Method of Data Analysis____________________________________________________________28
3.2.1 Model specification 28

Chapter Four
…………………………………………………..35
RESULTS AND DISCUSSION______________________________________________________35
4.1 Descriptive Analysis_________________________________________________________________35
4.1.1 Poverty issues 35
4.1.2 Food security 38
4.1.3 Women employment 39
4.2 Econometrics Analysis______________________________________________________________41
4.2.1 Determinants of poverty 41
4.2.2 The effect of percentage share of female’s income in a family on the expenditure
of different food items 44

Chapter Five
…………………………………………………...48
CONCLUSION AND RECOMMENDATION________________________________________48
5.1 Conclusion__________________________________________________________________________48
5.2 Recommendations___________________________________________________________________49
Bibliography-------------------------------------------------------51
Annex---------------------------------------------------------------53

4
 List of Tables

TABLE NO
PAGE

Table 4.1 Poverty indexes of Bahar Dar City………………………………………..35

Table4.2 Poverty indexes for different demographic characteristics………………..37

Table 4.3 Poverty indexes for different educational levels and economic
activities……………………………………………………………………….38

Table 4.4 Meals eaten per day……………………………………………………….39

Table 4.5 Women’s main activity……………………………………………………40

Table 4.6 Logit maximum likelihood estimates……………………………………..42

Table 4.7 The effect of percentage share of female’s income in a household on food
expenditure ……………………………………………………………...45

Table 8 Logit maximum likelihood estimates……………………………………….53

Table 9 Test of multicollinerity using pair-wise correlation coefficient

………………....54

Table 10 Marginal effects of the explanatory variables on the probability of being

poor…………………………………………………………………………………..55

Table 11(a-e) Engel curve regression results………………………………..………56

Acronyms
DES- Dietary Energy Supplies

5
FAO- International Food and Agriculture Organization

FHH- Female Headed Household

GNP-Gross National Product

HHH- Household Head

IMF- International Monetary Fund

KMs-Kilometers

LDCs-Least Developing Countries

MOLSA-Ministry of Labor and Social Affairs

MEDAC- Ministry of Economic Development and Cooperation

NGOs-Non-Governmental Organizations

UN- United Nation

UNCHS-United Nation Centre for Human Settlements

Abstract
Data from a survey of one hundred dwellers of Bahir Dar city in the 2000 Ethiopian
Urban Household Survey are used to examine the extent and determinants of poverty

6
in the city, and the systematic relationship between women’s income and household
expenditure pattern. To this end, the study employed both descriptive and
econometrics analysis. With the help of one of the behavioral models, Logit Model,
the study tried to analyze and identify the determinants of poverty in the city. On
doing so, educational level, occupation and age of the household head and family size
are found to be the significant parameters in determining the probability of being
under poverty. On the top of that, by applying a simple microeconomics concept,
Engel curve, the study attempted to analyze the effect of percentage share of female’s
income in a household on the expenditure pattern of the household on different food
items. The result attested that percentage share of female’s income in a family has a
positive and significant effect on vegetables and fruits, and spices expenditures.

Chapter One
INTRODUCTION

7
1.1 Background of the Study

Poverty is a global concern. Chen and revallion(2008) had revised and updated the previous
studies of population living below poverty line. In their recent study, the population living
below 1$ a day poverty line is estimated to be 1.4 billion. It is one forth of the population of
the developing world some 25 years ago; there were 1.9 billion poor people in the world.
The country suffers spells of drought, with resulting famines and such conditions have a
strong influence on the living standards of the whole population, particularly in the north and
the dry southeast part of the country. Another major growth deterrent for many years in the
country had been internal conflicts. These major shocks have important implications for the
welfare of both urban and rural households. In urban areas, the impact of the shocks is felt
mainly through higher food prices and increased rural- urban migration, often contributing to
increased urban poverty. (Abbi Mamo Kedir)

Poverty is widespread and multi- faceted in Ethiopia. Measured mainly in terms of food
consumption, set at a minimum nutrition requirement of 2,200 calories per adult per day, and
also including non-food consumption requirements, an estimate of 1995/96 shows that 45.5
percent of the population were below the poverty line. Poverty was prevalent both in rural
and urban areas, with coverage of 47 and 33 percent of the respective populations (IMF,
2000). Urban areas account for only 15 percent of the total Ethiopian population, but also
have a high rate of incidence of poverty. Unlike the findings elsewhere in the developing
world, urban and rural poverty levels in Ethiopia are not dramatically different from each
other. Depending on the methodology adopted and the data analyzed, the estimated urban
overall poverty and food poverty range from 33 to 50 percent (Kedir, 2003; Bigsten et al
2003; MEDAC, 1999 Taddesse and Dercon, 1997).

The poverty experienced by many Ethiopians is reflected in a range of well-being measures

of the population. For example, the life expectancy at birth in the country is approximately
46 years, which is substantially lower than the average 77 and 67 years recorded for countries
with high and medium human development indices, respectively. Moreover, three quarters of
the population do not have access to an adequate water source, a figure that is amongst the

8
highest for countries experiencing a low measure of human development. The percentage of
population with access to suitable sanitation, which stands at 12 per cent, is significantly
lower than the 53 per cent average for the sub-Saharan Africa (UNDP 2003, pp. 237-257).
On the other hand, the adult illiteracy rate at around 60 percent is significantly higher than
the average for sub-Saharan Africa and other developing countries. (Tesfaye, 2006)

There is little evidence on poverty trends in urban areas with much of the discussion focusing
on cross-section evidence. Taddesse (1998) showed the trends in urban poverty between
1995, 1997 using subjective, and objective (consumption) poverty lines. His findings show
that poverty slightly increased according to the subjective poverty lines (SPL) and decreased
according to the consumption poverty lines. When we look at the disaggregated results, we
observe heterogeneous trends across cities. Poverty has decreased in Addis Ababa, Awassa
and Mekele while it increased in Bahar Dar, Dessie, Diredawa and Jimma; according to SPL
and it has decreased in Addis Ababa, Awassa, Bahar Dar, and Mekele, increased in Diredawa
and Jimma, but remained the same in Dessie according to the consumption poverty line.
Bigsten et al (2003) reported poverty trends (using consumption poverty lines based on
Ravallion and Bidani, 1994) for urban Ethiopian between 1994 and 1997. For all urban areas,
the study showed an increase in poverty from 1994 to 1995 and a decline in poverty from
1995 to 1997. Likewise, in the case of Taddesse (1998), the trends vary by city. Between
1994 and 1995, poverty declined in Addis Ababa, Awassa, Bahar Dar and Jimma while it
increased in Dessie, Diredawa and Mekele.

Based on real total household expenditure per month as a preferred welfare indicator, Kedir’s
results indicate that there is a high-level chronic poverty (25.9 %) more concentrated in
central and northern cities. Households that experience transitory poverty constitute 23.0% of
the total.

This paper contributes to the literature examining poverty in Ethiopia. In particular, updated
estimates are provided of the nature and extent of poverty experienced by Ethiopians residing
in Bahar Dar city. Further, the determinants of poverty are sorted out by using the
appropriate econometric model. Moreover, the association between women and poverty is

9
concerned. This will have a direct bearing on policymaking, as it will aid in prioritizing
women and different types of households in the effort to tailor resources to the needy.

1.2 Statement of the Problem

“The unfinished business of the 21st century is the eradication of poverty” Joan Somavia, UN
world summit for social development.

Despite significant improvements over the past half century, extreme poverty remains
widespread in the developing world. More than 1.2 billion people live on less than $1 per day
and more than 2.8 billion(almost half of the world’s population) live on less than $2 a day.
(Todaro, 2003) This worldwide chronic poverty is extensively manifested in our country in
general and in urban centers in particular.

In Ethiopia poverty is the general feature for the nation and causing many sufferings and
anguish to the largest proportion of the population. It is high agenda of the government,
donor agencies, NGOs and other actors. The government has been formulating and
implementing various policy interventions and programs that are in one way or another
related to poverty reduction.

Currently, though poverty is taken as the country’s rural phenomena there is a diffusion and
growth of urban poverty. Indeed, the number of urban poor is increasing at unprecedented
rate. This is due in part to the highest rural-urban exodus and alarming internal population
growth (Dessalegn and Aklilu, 2002). In effect, the urban economy has limited capacity to
accommodate the populous. In such a situation, employment in the formal sector is tough and
the probability of getting commendable job opportunities, in fact, could be daunting.

The impoverished people who live in urban centers of Ethiopia often suffers from under
nutrition and health problem, have little or no literacy, have little political voice, live in
environmentally degraded areas and attempts to earn a meager living on dilapidated urban
slums.

10
Women constitute a substantial majority of urban centers poor. Across Ethiopia, women and
children experience the harshest deprivation .They are malnourished, receive less medical
services, clean water and sanitation, lower earning capacity, less access to education, formal
sector employment, social security and government employment programmes .Because of
these factors financial resources of poor women are meager and unstable relative to men’s.

Women are often paid less for performing similar task. In urban areas, women are much less
likely to obtain formal employment in private companies or public agencies and are generally
limited to low productivity jobs.

Having the above afro mentioned points in mind, the researcher is eager to conduct a study
on this title i.e. what is the feature of poverty and food security in Bahar Dar city and what
would happen to the existing situation if women in urban areas in general and in the city in
particular were drawn into economic mainstream?

1.3 Objective the Study

The overall objective of the study under investigation is assessing the existing poverty, food
security and women employment in Bahar Dar city. Specifically, the study aims:

 To examine the extent of poverty in the city

 To identify the different parameters that determines
poverty in city

 To evaluate the extent of these determinants

 To show the experience of food shortage in the city

 To analyze women’s employment
 To examine the existence of systematic relationship
between household expenditure patterns and women
income

 Based on the findings, to draw some policy implication

1.4 Research Hypotheses

11
On the out set, the following hypotheses are forwarded and are going to be tested with
reference to the empirical data what is available on hand.

 Education and type of economic activity have a significant effect on the extent of
poverty on households
 Increasing the share of female’s income in a family will have a significant implication
in achieving not only food security but also nutrition security.

1.5 Scope of the Study

Since studying the overall features of poverty, food security and women employment at the
national level calls for a good deal of knowledge, skill, experience, finance, time and
organized data ,the researcher only attempts to cover the major urban centers of the country
for which the required data is available. Moreover, due to lack of the skill on how to process
a large sample data with the help of computer applications, the study delimited its scope only
to Bahar Dar city.

1.6 Significance of the Study

Characteristics of poverty in LDCs are multi-dimensional. So, poverty reduction effort

requires multi-faced interventions that emanate from deep knowledge and awareness
(Todaro, 1994). Therefore, reducing poverty, achieving food security and empowering
women, and thereby ensuring sustainable development through tackling the determinants and
causes of poverty is the main development strategy of Ethiopia. Hence, this paper believed to
be significant since it attempts to address the determinants of urban poverty and gender
issues in the city. In addition to this, it is presumed that the paper has a paramount significant
as a springboard for those researchers who conduct a study on the topic.

1.7 Limitation of the Study

In the process of carrying out the research, the researcher has encountered by a number of
hindering factors which added to the cost of doing the research. The primary shortcoming
which poses an adverse effect on performing the study was lack of know-how about
processing (aggregating) data using computer applications and the availability of computers

12
itself. In addition, lack of finance and scarcity of time also put a black shadow on the smooth
running of the study.

1.8 Organization of the Paper

This paper is framed in five major chapters. The first chapter tries to incorporate things like
background, objective, scope, significance and hypotheses of the study. Chapter two
discusses on the review of related literature that includes both empirical and theoretical
conceptual framework evidence on national and regional level. Third chapter holds the data
methodology and model specification part of the paper. Chapter 4 gives the data analysis and
discussion part of the paper. The last chapter provides the conclusion and policy implication
generated from the study.

Chapter Two
LITERATURE REVIEW
2.1 Theoretical Literature Review

2.1.1 Food security and poverty: Definitions and measurement issues

13
"Poverty may be defined as an insufficiency of material necessities of life"(Encyclopedia of
Britannica volume 17, 1966)

Poverty is generally considered as a situation in which the underprivileged do not have

adequate food and shelter, lack access to education and health services, are exposed to
violence, and find themselves in a state of unemployment, vulnerability and powerlessness.

Poverty is multi-dimensional and has to be looked at through a variety of indicators such as

levels of income and consumption, social indicators and indicators of vulnerability to risks
and socio-political access and participation. The most common approach to the measurement
of poverty is based on incomes or consumption levels. It is widely understood that an
individual is considered poor if consumption or income level falls below some minimum
level necessary to meet basic needs i.e. poverty line. The nature and level of basic need
satisfaction varies along with time and societies and the poverty line to be established should
be appropriate to the level of development, societal norms and values (World Bank Poverty
Net, as quoted by Asmamaw, 2004)

United Nation Development Program (UNDP) in its 1997 report described poverty as
follows; “Poverty has many faces .It is much more than low income. It also reflects poor
health & education, deprivation in knowledge &communication, inability to exercise human
& political rights, and absence of dignity, confidence & self-respect. There is also
environmental impoverishment & impoverishment of entire nations, where essentially every
one lives in poverty.”

The multi-dimensional character of poverty in Ethiopia is reflected in many respects, such as

destitution of assets, vulnerability and human development. The World Banks definition of
poverty indicates that poverty is “...a pronounced deprivation of well-being related to lack of
material income or consumption, low levels of education and health, vulnerability and
exposure to risk and voiceless ness and powerlessness (World Bank 2001a, as quoted by
Pradham et al., 2002).

This definition fairly describes the nature of poverty in the Ethiopian context. As the concept
of poverty reflects “socially perceived deprivation” of basic human needs, its understanding

14
also considers the minimum living standards of the people. Poverty alleviation and reduction
of economic inequality is the major socio-economic and political issue in the country. As
experience has shown, the existence of large number of poor people and the prevalence of
economic inequality may bring about social tensions, which would induce various criminal
acts if situations go beyond the limits of social tolerance. Poverty alleviation would,
therefore, enhance economic development and result in improved incomes and better well-
being of the people, which is a pre-requisite for peace and further development. However,
attempts to eradicate poverty would require strong commitment on the part of concerned
authorities in favor of economic development to induce the sustainable livelihood of millions
in urban and rural areas of Ethiopia. (Asmamaw, 2004)

Understanding poverty in the Ethiopian context also needs to consider its multidimensional
characteristics, which go beyond mere income and food provision. Such characteristics
include aspects of human capabilities, assets and activities necessary for sustainable
livelihoods. A sustainable livelihood is one that can “cope with and recover from stresses and
shocks and maintain or enhance its capabilities and assets both now and in the future, without
undermining the natural resource base” (Carney, 1998 as quoted by Asmamaw).

The fundamental bases of livelihood comprise natural (land, forests, water, pastures, and wild
life), physical (farm animals, tools/machinery, economic and social infrastructure), financial
capital (income and savings), social relations and human capital (health, education etc). The
Ethiopian situation clearly reflects the degree to which the bases for sustainable livelihood
are adversely affected by natural and fabricated calamities. The underprivileged poor have
limited access to most of the livelihood capital assets, which has widened income disparity
and undermined their bargaining power to establish sustainable livelihoods. This socio-
economic condition emphasizes the need for “political capital” as a means of ensuring better
participation in deciding on matters that affect the well-being of the poor and enhance
improved security of subsistence needs (Dubois, 2002 as cited in Asmamaw).

2.1.2 Food adequacy, food security and protein-energy malnutrition

Individual food adequacy (IFA) is shown, in the short term, by calorie intakes that are
sufficient for needs, varying with age, health, work and adult height; in the medium term, by

15
the absence of acute protein-energy malnutrition (PEM), for example low child weight for
age (WA) or low adult weight for height (WH)); and in the long term by the absence of
chronic PEM in children under five, of low height for age (HA). The proximate cause of
PEM appears to be energy deficiency relative to requirements and infection burdens.
Although protein deficiency causes widespread and serious damage, most nutritionists now
reject the earlier view that protein deficiency, let alone specific amino acid deficiency, is an
independent problem requiring special protein-rich foods, supplements or cereal varieties.
Only exceptionally is a protein problem not curable simply by providing "more calories".
(FAO, 2000)

Individual food security (IFS) refers to "access to adequate safe and nutritious food to
maintain a healthy life ... without undue risk of losing such access", i.e. IFA as well as the
confidence that it can be maintained (FAO. 1996. The Sixth World Food Survey.
Rome.).Without such confidence, people take hyperactive cautious decisions that forfeit their
chances to escape from chronic hunger. (Ibid)

A poor person usually obtains 70 to 80 percent of his or her calories (and most other
nutritional requirements) from one or two of the world's seven main food staples. These are
by far the cheapest sources of energy, and of most other nutrients. For the poor, access to
these staples is the key to achieving IFS. (Ibid)

Household food adequacy/security (HFA/HFS) is necessary for IFA/IFS, but not sufficient,
because food may be distributed among household members disproportionately to their
individual needs. (Ibid)

National food security (NFS) refers to a nation's capacity to ensure HFS/IFS without undue
departure from other policy goals. NFS in a given year is often measured by: dietary energy
supplies (DES) per person, allowing for the distribution of food and needs among individuals
and times; or the ratio of food imports to total exports, although food aid must be allowed
for; or staples stocks (publicly controlled or likely to be marketed if prices rise) as a share of
normal consumption. (Ibid)

16
The measurement of food security at the national level entails the calculation of the extent to
which the production of the staple food in a country can provide the population with the
nutritional minimum of 2,200 calories per person per day. The existence of national food
security, however, does not guarantee that all regions and all people, especially the poor, will
have access to the nutritional minimum because of existing regional, economic and social
inequalities. Furthermore, this classical definition of national food security has been
criticized because it does not take into account the need for a variety of foods (meat, fish,
oils, dairy products, vegetables and fruits) that provide protein, fat, micro-nutrients as well as
energy, that make up a nutritious diet (FAO 2000). In fact, food security, defined only in
terms of the staple food, does not provide a sufficient input for an adequate nutritional status
(Alderman and Garcia 1993).

Food security at the national level can also be ensured through exports from countries with
surplus food production. In fact, many countries, especially many less-developed countries,
import a part of the needed cereals for national food security at present. While improved
trade regulations have made such imports easier, the process of importing brings about a
degree of political dependence of the importing countries on the exporting countries. Self-
sufficiency, therefore, in terms of national food security, is often argued as desirable, if not
essential.

Although there may be food security at the national level, there may be food insecurity for
some rural and urban populations. Some urban populations may not have sufficient
purchasing power to ensure food security. There may be food insecurity for some rural
populations because they do not produce sufficient food and/or do not have sufficient
purchasing power to cover their food needs. Alternatively, they may produce enough staple
food and a variety of other foods, but they may market a part of the staple foods and all the
non-staple foods (vegetables, fruits, etc., that fetch a good price), because they must have
cash to cover other basic needs. It is only when their overall household income is sufficiently
high to afford non-staple foods that it can be said that they truly enjoy food security and an
adequate nutritional status (FAO 2000). It is, necessary, therefore, to measure food security
at the household level, since there is usually considerable social and economic inequality
between households. (Constantina, 2001)

17
What is the role of gender in reducing PEM and/or over nutrition? In some countries, tests
showed that stunting or low weight was much more prevalent than predicted from average
DES. This is partly because of adaptation, but also partly because the gender distribution of
DES and health care (and hence treatment of infections, which alters the efficiency of DES
use) is especially unequal in some countries. This contributes to the unfavorable stunting and
wasting status of India - North India, like Bangladesh and Pakistan, shows strong food
discrimination against girls aged two to five years, with known harmful outcomes. The same
is probably true of Mauritania. The nutritional ill-effects of gender discrimination may rise
even where nutrition is generally improving; in India, the incidence of adult body mass index
(BMI) below 16 (third degree chronic energy deficiency) was 11.4 percent of men in 1975-
79 but 8.8 percent in 1988-90, while for women the incidence only fell from 12.7 percent to
11.3 percent.

Girls aged two to four years suffer serious disadvantages compared with boys in DES
(relative to need) and/or access to health care - and as a result worse PEM - in many parts of
Asia, but not in sub-Saharan Africa or Latin America. Furthermore, girls almost everywhere
have fewer chances of education than boys, so that women have less access to skilled work.
Women also suffer discrimination in pay even for the same job or task and more severely in
access to land, legacies and credit. Although the evidence (especially in Asia) does not
suggest a much greater poverty risk for women, their control over income is certainly less.
(FAO, 2000)

Apart from directly damaging the food security of women, these gender disparities harm
children's food security, in two ways. First, income - including extra income - is more likely
to be spent on improving the nutritional status of children under five if it accrues to women.
Second, discrimination against small girls in feeding or health care not only cuts female life
expectancy to the levels of males - elsewhere it is three to six years more - but is transmitted
to later generations. Difficult pregnancies due to small womb size induce low birth weight,
imperiling infant life and development. This helps explain why, for example, several South
Asian regions suffer more frequent stunting and wasting than many African countries with
lower calorie intakes and similar inter household distribution.(Ibid)

18
2.1.3 Characteristics, Causes and Measurement of Urban poverty

i. Characteristics and Causes of Urban Poverty

Some authors doubt the distinction between urban and rural poverty because of the fear that
such distinction would remove one from considering the main determinants of poverty
(Ellen, 1995). There is, however, some distinguishing features of urban poverty that need to
be recognized and understood. The incidence, economics, demography and politics of
poverty differ between urban and rural areas. Therefore, the analysis, formulation and
implementation of policies should be differentiated although policy coordination is obviously
needed. (MEDAC, 2000)

Wage labor or the labor market is the main determinant of urban poverty (Haan, 1997) Most
of the urban poor earn income from the informal sector. While their earnings do not have as
large a seasonal component as those of the rural poor, their incomes are probably almost as
unstable because they have little protection from sickness and injury and the unpredictable
demand for their services. The poor possess little human capital and almost no physical
capital that can be sold or consumed at the time of a sudden dip in their earnings. The poor,
having no asset that can be used as collateral, also lack access to credit markets (Mills and
Perenia, 1994).

Other dimensions of urban poverty include poor environmental conditions, changes in prices
of basic goods, lack of social network, violence and insecure tenurial status (Haan, 1997;
Wratten, 1995). Wratten (1995) adds commodization of urban economy and the negative
effects of government actions and policies to the list of urban poverty dimensions. In
addition, absence of mechanisms to involve the poor in the decision, making process could
also be cited as another dimension of urban poverty. (MEDAC, 2000)

The environment and health risks faced by the urban poor result from the juxtaposition of
industrial and residential functions; competition for land; high living densities, overcrowded
housing, traffic congestion, pollution and the mismatch between urban growth and the
provision of clean water supply, sanitation, solid waste disposal etc. The quality of the urban
environment poses health threats including typhoid, diarrhea, cholera, intestinal worms.

19
These diseases come from contamination of water and food; poor garbage collection,
overcrowded housing and insufficient water for hygiene (Wratten, 1995).

The urban economy is more commercialized than the rural economies and such
commodization affects the lives of the urban poor by affecting their needs for subsistence,
housing, education etc. Increases in the prices of food, house rent and educational fees will
put pressure on the urban poor.

Urban areas are marked by social diversity and impersonal relationship. Urban people have
different ethnic, cultural and linguistics origins (Wratten, 1985). Such diversity leads to
fragmentation and crime. The poor are more often the victims of urban crime. The growing
phenomena of urban street children, prostitution and begging are associated with family
breakdown, poverty and absence of social networks.

The urban poor are also affected negatively by the state policy. For instance, the government
policy on security of land makes the poor to live in an increasing terror of losing their only
assets and personal possession. Insecure tenurial status forces the urban poor to live in self-
built, illegal housing, which is not, provided with government services such as schools,
health services etc.

There is a considerable debate concerning the causes of urban poverty. Some authors
emphasize the structural adjustment programs of different countries as the main cause of
poverty. The argument is that the structural adjustment programs have made numerous
people to suffer from increases in food price and service charges, restrictions on wage levels,
reduction in employment and declines in urban infrastructure expenditure (Amis and Rakodi,
1994). These people argue the emergence of the new poor (as opposed to the chronic poor)
because of loss of jobs due to structural adjustments (Vanderschueren et al, 1996).

The emphasis on such macro factors however neglects other factors, which explain the
impoverishment of individuals, families or social groups at micro level. Hence micro level
explanation dwells on the relation of poverty with a particular point in the life cycle of
families caused by sudden shocks, migration etc. Vanderschueren et al (1996) argues that the
loss of adult family income earner, the confiscation of street traders stock, the demolishing of

20
housing because of illegality, the high cost of illness in the family, civil war etc are factors
which result in considerable poverty particularly in some African countries. Similarly, rural
impoverishment initiates rural-urban migration, which fuels urban poverty in cities.

ii. Identifying Poverty Lines and Poverty Measures

A. Identifying Poverty Lines

Given an appropriate measure of welfare, the identification of the poor necessitates that a
poverty line be determined below which individuals or households are considered poor.
There are a number of ways that such a poverty line may be identified. The most common
approach is to estimate the cost of a consumption bundle for which basic consumption needs
will be met. This is known as the cost of basic needs approach. It proceeds by first estimating
the food expenditure necessary to attain some recommended food energy intake. This
expenditure level can be considered as the food poverty line. Next, an allowance is made for
non-food goods to arrive at the total poverty line (Ravallion 1994; Lipton and Ravallion
1995).

Deriving the poverty line using the cost of basic needs approach, however, presents some
difficulties. For example, setting the food energy requirement may be problematic as there
are significant variations among people in physical features and work habits. This renders the
task of setting a minimum energy requirement, even for a specific group in a specific region,
daunting. Even after a minimum requirement is set, there remains the problem of choosing a
food bundle that meets it (Sen 1999). A bundle that meets the requirement at minimum cost
(given prevailing prices) could be chosen, but such a bundle is of little relevance if it is not in
tune with the eating habit of the poor. As Sen (1999, p. 12) points out, ‘… the actual incomes
at which specified nutritional requirements are met will depend greatly on the consumption
habits of the people in question’. The second difficulty associated with deriving the basic
needs poverty line is in making an allowance for non-food goods. This stems from the fact
that there is nothing that can serve the same role as food energy requirements in anchoring
the non-food component of the poverty line (Ravallion 1994).

21
In practice, two methods have been commonly used to derive the poverty line; the ‘food
energy intake’ and ‘food share’ methods (Ravallion 1994). Both approaches are based on the
assumption that there is a minimum energy requirement for a typical person to keep up
normal activities, such as the 2,200 Kcal per day threshold stipulated by the World Health
Organization (1985). Thus, the ‘food energy intake’ method attempts to identify the total
consumption expenditure at which a person is expected to attain the minimum food energy
requirement. This is accomplished by regressing calorie intake on consumption expenditure
or income. The poverty line, then, becomes that level of total expenditure at which the
minimum energy requirement is met (Greer and Thorbecke 1986; Ravallion 1994). The
advantage of this method is that it automatically includes an allowance for non-food goods,
circumventing one of the difficulties mentioned above. However, it may lead to an
‘inconsistent poverty comparison across sub-groups or over time since people with the same
command over basic consumption needs will not in general be treated the same way’ (Lipton
and Ravallion 1995, p. 257).

In the ‘food share’ method, the cost of the food bundle that meets the minimum energy
requirement is estimated for each population sub-group. These food poverty lines are then
divided by the share of food in total expenditure of the poorest households, such as the
poorest deciles, in each sub-group to obtain the total poverty line. This method may also lead
to inconsistencies in poverty comparison since the share of food in total expenditure does not
remain constant across sub-groups (Ravallion 1994).

An alternative method of deriving the poverty line, which is a version of the cost of basic
needs approach, is suggested in Ravallion and Bidani (1994). In this method, a basket of
goods for which basic food requirements will be met is defined. The cost of this basket at
market prices becomes the food poverty line. An allowance for non-food goods is then added
on the food poverty line to obtain the total poverty line. This is done by estimating a food
Engle curve and determining the food share of the representative household whose total
consumption is exactly equal to the food poverty line (Ravallion 1994; Ravallion and Bidani
1994).

B. Poverty Measures (Measurement of Urban Poverty)

22
Measurement of urban poverty is a prerequisite for policy formulation. There are two main
approaches: one, which emphasizes the absolute nature of poverty, while the other focuses
on relative deprivation (Townsend cited in Amis and Rakodi, 1994). In the absolute
approach, a minimum or basic datum level is established. Individuals below this datum line
are designated to be in conditions of poverty. The datum level involves consumption norms,
usually with some nutritional criteria, which are translated into food requirements and then
into a required income (Amis and Rakodi, 1994). The line then serves as basis for head
counting and gives an idea of what proportion of the people are below the poverty line. The
relative deprivation approach defines poverty in relation to either average levels or societal
norms. The approach is developed in the context of developed country and it attempts to
relate the definition of poverty to its potential causes such as economic exploitation and
problems of social marginality (Townsend cited in Amis and Rakodi, 1994). It is assumed
that poverty measured with respect to some average is difficult to eradicate. (MEDAC, 2000)

The above two measures use income or consumption based approach to poverty. Income
measure of poverty, however, is criticized because it does not consider peoples’ access to
basic services such as education, health care, adequate quality housing with basic services
etc. Satterthwaite (1997) argues that there is uncritical transfer of the concept of poverty line
from the North to the South. He further argues that in the North, income based poverty line
concept is applied by governments where a smaller proportion of their population lived in
housing which lacked adequate provision of water supply and sanitation and few suffer from
lack of access to education, health care etc (Satterthwaite, 1997). The situation in the South
however is different. Inadequate provision of water, sanitation, drainage, poor quality and
overcrowded housing are common phenomena. Similarly, lack of access to education and
health is one of the fundamental influences on the capabilities of individuals to generate
income and lead decent life. Thus various indicators of welfare, notably health, education,
shelter, sanitation etc have substantial implications for poverty. (Ibid)

A different view on poverty is the entitlement approach. The approach explains deprivation
not in terms of an overall lack of resources, but in terms of household’s ability to command
such resources by means of mechanisms called entitlements (Amis and Rakodi, 1994).
Entitlements depend on income from wage labor, sale of assets, own production etc. The cost

23
of purchasing resources (food) and social security claims also influence entitlement (Sen
cited in Amis Rakodi, 1994). The contribution of entitlement approach to poverty is the
recognition that poverty is more than mere lack of resources particularly income. It contains
important elements such as exposure to and risks from sudden change (vulnerability). Hence
recent poverty analysis has focused on the concept of vulnerability, distinction between long-
term trends and short term shocks and the importance of assets. (Ibid)

The income, welfare or entitlement approaches give outsiders’ or professionals’ perspectives

on poverty. A different perspective of poverty is to view poverty from the perspective of the
poor themselves. It is stated that the most fundamental failure in poverty analysis is the lack
of involvement of the 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 that it
enables to understand the needs and priorities of the poor people instead of making
assumptions on external measures of income and expenditure. The poor may put high value
on welfare dimensions such as better health, better schools, secure housing tenure, home
ownership, improved provision of water and sanitation. Similarly, they may also place higher
values on qualitative aspects such as independence, security, self respect, identity and legal
and political rights (Satterthwaite, 1997). Thus, insiders view on poverty is better equipped to
prioritize and deal with poverty. (Ibid)

Identifying the poverty line facilitates identification of the poor. The next important issue is
the choice of an appropriate poverty measure to aggregate the information on individual’s
welfare. There is quite an extensive literature on poverty measures that is briefly discussed
here. The earliest and perhaps most famous measures of poverty are the head count ratio and
the poverty gap measure. The head count ratio gives the proportion of people who are poor.
That is:

24
Where H is the head count ratio, q is the number of individuals identified as poor and n is the
population size. The poverty gap measure provides an indication of the aggregate shortfall of
the poor from the poverty line. That is:

Where PG represents the poverty gap, z represents the poverty line and y i is the consumption
(income) of individual i, i= 1, 2…n. This index can be normalized to obtain the ‘income gap
ratio’, by expressing it as the percentage shortfall of the average income of the poor from the
poverty line. This is given by:

Sen (1976) argued that poverty measures should satisfy the following two desirable axioms:

Axiom 1: A reduction in the income of a poor person increases the poverty measure. This is
known as the ‘monotonicity’ axiom.

Axiom 2: A pure transfer of income from a poor person to anyone richer increases the
poverty measure. This is known as the ‘transfer’ axiom.

The head count ratio fails to meet both the ‘monotonicity’ and ‘transfer’ axioms. As it is
sensitive to the number of poor only, it does not take into account the extent of income
shortfall. In addition, the head count ratio does not rise if there is a transfer of income from a
poor person to a richer. It may, in fact, decline if a poor person closer to the poverty line
becomes non-poor because of the transfer (Sen 1999). The poverty gap measure and the
income gap ratio, on the other hand, meet the ‘monotonicity’ axiom but not the ‘transfer’
axiom. The other disadvantage of these measures is their insensitivity to the number or
proportion of poor people, unlike the head count index.

25
These limitations in the commonly used poverty indices prompted Sen (1976) to develop an
alternative poverty measure, popularly known as Sen’s index, given by:

Where G is the Gini coefficient of the distribution of income/consumption among the poor.
Sen’s index is effectively a weighted sum of the shortfall from the poverty line, where the
weight is the rank order of the poor in the ranking of the poor (Sen 1976). The index satisfies
both the ‘monotonicity’ and ‘transfer’ axioms. (Tesfaye, 2006)

Sen (1999) underscores the need to use a complex measure such as P to capture the different
concepts of poverty. The approach has not, however, escaped criticism. Streeten (1998, p.15)
argues that it tries to capture too many aspects in a single formula and ‘introduces an element
of relative deprivation into an absolute poverty measure’. Others note that the index fails to
meet other desirable axioms, other than the ones suggested by Sen (1976). For example,
Kakwani (1980) shows that the index fails to meet certain sensitivity properties identified in
axioms 3 and 4 below, while Foster and Shorrocks (1991) show that it does not satisfy the
subgroup consistency property given in axiom 5. The desirable properties stipulated in these
axioms are as follows:

Axiom 3: The increase in poverty measure be greater the poorer the transferor of income,
when the number of positions between the income transferor and recipient is fixed in the
ordering of poor people (Transfer-sensitivity I).

Axiom 4: The increase in the poverty measure be greater the poorer the transferor of income,
when the amount of income (and not the number of income positions) between a transferor
and recipient is fixed (Transfer-sensitivity II).

Axiom 5: The overall level of poverty to fall if poverty increases in a subgroup of the
population. This property is called subgroup consistency.

26
An alternative is the class of additively decomposable poverty measures developed in Foster,
Greer and Thorbecke (1984). This class of measures (commonly known as FGT measures) is
given by:

Where α is the poverty aversion parameter, g i is the income shortfall of the ith individual or
household, and z is the poverty line. The poverty aversion parameter (α) reflects the concern
attached to the proportionate shortfall from the poverty line. When α is equal to zero, the
FGT measure corresponds to the head count index in which no concern for the depth of the
shortfall is shown. When α is equal to unity, the FGT measure collapses to the poverty gap
ratio and is consistent with a uniform concern for the depth of the poverty shortfall at
different points of the distribution. When α is greater than unity, the poverty measure
becomes more sensitive to the poorest of the poor. The most commonly used value of α
(greater than 1) is two, which measures the severity of poverty. In this case, the FGT measure
is simply the weighted sum of poverty gaps, as α proportion of the poverty line, where the
weights are the poverty gaps themselves (Ravallion 1994).

The FGT measures satisfy axioms (1)-(5) for specific values of the poverty aversion
parameter. The monotonocity axiom is satisfied for α > 0, the transfer axiom for α >1, the
transfer sensitivity axiom for α > 2, and, the subgroup consistency property for all values of
the inequality aversion parameter. Another desirable feature of the FGT measures is that they
are additively decomposable with population share weights and thus can be expressed as:

Where yj is the income/consumption vector of subgroup j = 1, 2...m, and nj is the population

of subgroup j. The additive decomposability of the FGT measures is very useful in applied

27
work since it highlights the contribution of different subgroups to poverty. The percentage
contribution of subgroup j to poverty can also be calculated using the following expression:

2.1.4 Gender and poverty

Women constitute a substantial majority of world’s poor. Across the developing world,
women and children experience the harshest deprivation. They are malnourished; receive less
medical services, clean water, and sanitation, lower earning capacity, less access to
education, formal sector employment, social security and government employment
programmes. Because of these facts, financial resources of poor women are meager and
unstable relative to men’s.

Women are often paid less for performing similar task. In urban areas, women are much less
likely to obtain formal employment in private companies or public agencies and are generally
limited to low productivity jobs. Rural women have less access to resources to generate
stable incomes. In many regions of world, these exists a strong bias against females in areas
such as nutrition, medical care, education and inheritance. The low levels of living among
women and children are common where economic status of women remains low. Women
and their dependants remain the most economically vulnerable group in developing
countries.

In long run, the low status of women is likely to cause slower rates of economic growth. The
welfare of women and children are strongly influenced by the design of development policy.
To improve the living conditions for the poorest individuals, women must be drawn into
economic mainstream. This would entail increasing female participation rates in educational
and training programmes, formal sector employment and agricultural extension programmes.
It is also of primary importance that precautions be taken to ensure that women have equal
access to government resources provided through schooling services, employment and social

28
security programmes. The economic status of women would also be improved by legalizing
informal sector employment where majority of female labor force is employed.

Education and enhanced economic status for women are of crucial importance in long term
for developing human capital and speed up economic growth.

The proportion of female headed household (FHH) is high in Ethiopia, compared to

international standards. Ethiopia has the fifth highest percentage of female-headed
households among the 22 African countries for which such data is available (see Kodama
2006). Nationwide, 19 percent of the households is female-headed. The high prevalence of
female-led families is rooted in the historical and cultural roots of Ethiopia. For example,
Meehan (2004) finds that the high prevalence of FHH in Tigray (30 percent of the
population) is related to the loss of male combatants in both the civil war that ended in 1991
and the more recent Ethio-Eritrean conflict, a traditionally high age gap between wives and
husbands, leading to a high number of widows compared to widowers, and migration
patterns, resulting in de facto FHH even where legal marriage still exists.

Urban areas have a strikingly higher proportion of FHH than rural areas: 33 percent of the
urban families are female-headed with no significant variation across the urban spectrum,
against 17 percent of rural families. The percentage is high, compared to the 13 percent for
urban Ivory Coast in 1988 (Grootaert et al. 1997) and the 33 percent for Mauritania
(Coulombe and McKay 1996). Drawing on data from the Amhara region, Kodama (2006)
points to the local socio-economic conditions pushing female heads to migrate to urban areas
as one explanatory factor for the very high percentage of female heads in urban areas,
compared to rural areas. Once a woman residing in rural areas becomes the family head, she
is unlikely to be able to sustain herself unless she can get access to land and be allowed to put
it to productive use. The ability of a woman to access land is very low in the Ethiopian
context, where there is shortage of land and restrictions to both access and use of land by
women (e.g., inheritance laws).30 Those female heads who are not able to access land have
often no choice but to migrate to urban areas.

29
Women predominates the urban poor because the jobs women obtain are poorly paid, part
time or insecure. In addition, women do not always control their own income because of
inequities in resource distribution and decision making power within the household and in
equities in society in commanding resources or assets compared to men (Beall, 1996).
Women outnumber men in Addis Ababa. In 1994, the number of women was 1,089,285
while the number of men was 1,023,452. Female-headed households in Addis account for
about a quarter of the population. It is quite true that women in Addis Ababa especially those
who are less educated have less access to formal sector employment. The 1984 and 1994
census indicated that the unemployment rate of women was higher than men’s. Those
employed women largely work in the informal sector in petty commodity production,
domestic services, prostitution and fuel wood collecting. The vast majority of fuel wood
collectors (76%) are women and they travel long distances between 10 and 20 kms with a
fuel load of 20 kg on their backs (Dierig, 1999).

The educational status of women in Addis Ababa revealed that women are more illiterate
male illiteracy rate was 10.7 % and the women illiteracy rate was 23.48 %. Among the
literate population, the median grade completed by males was 7.9 while it was 6.1 for
females (Analytical report, 1999). The rates of enrollment as evidenced by gross enrollment
ratio (GER) shows that though the enrollment of girls in primary schools is almost similar to
that of boys, it is much lower for girls than boys in secondary schools. Hence, gender gap in
enrollment increases with rising educational levels. Similarly, the number of dropouts and
repeaters is higher for females than for males in the higher grades. This shows that females
do not go through high grades as compared to males. Prostitution entirely affects women and
in terms of street children, it is estimated that 25 % of the 40,000 street children are females.
These street girls working and living in the streets are faced by all kind of abuses. This group
of women is very likely to join the rank of street mothers and enter in the vicious circle of
poverty (Ministry of labor and social affairs, 1998).

The environmental pressures of insufficient water supply, water pollution, fuel wood
collection etc fall more on poor women than poor men in the city as women have to cater
these services for the household. Thus, the gender dimension of poverty in the city is
reflected in the economic, social and environmental pressures and hardships.

30
2.2 Empirical Literature Review

2.2.1 The poverty profile of urban Ethiopia

The mean consumption expenditure per adult equivalent has been computed for urban
Ethiopia and each of the urban centers to highlight the average standard of living enjoyed by
the urban society. The average consumption for urban Ethiopia was 151 Birr in 1994, but this
masks substantial variation across urban centers. The highest figure was recorded in the city
of Dire Dawa, followed by Bahar Dar, Awassa, Dessie, Addis Ababa, Jimma and Mekelle.
There was a 6-percentage decline in real mean consumption per adult equivalent between
1994 and 2000 for urban Ethiopia. Similarly, there was a decrease in all the cities with the
exception of Awassa and Mekelle during the period. Specifically, significant declines were
recorded in Dire Dawa, Dessie and Bahar Dar where the mean consumption per adult
equivalent fell by 26, 25 and 21 percentage points respectively. Conversely, there was an
increase in Mekelle and Awassa by 41 and 16 percent respectively. (Tesfaye, 2006)

The findings of Tesfaye confirm that the incidence of poverty in urban Ethiopia is indeed
high with a head count index of 41 percent in 1994 and 43 percent in 2000. The largest
proportion of poor in 1994 was found in the city of Mekelle. This is consistent with a priori
expectations because the economy and residents of Mekelle had to directly withstand the
worst of the civil war and its ensuing isolation from the rest of the country. The second
poorest city in 1994 was Awassa, followed by Addis Ababa, Jimma and Dessie. The high
poverty levels in Addis Ababa and Dessie may be associated with the influx of demobilized
soldiers and migrants into the two cities at the end of the civil war. Poverty was found to be
lowest in Dire Dawa in 1994 (a head count index of just 13 percent), which may in part be
attributed to the booming contraband trade in the city at the time. Bahar Dar also performed
relatively well in 1994. Being the main grain producing area of the country, Bahar Dar, may
have benefited from the liberalization in grain trade that had been put into effect as part of
the economic reform programs

Tesfaye’s analysis also shows that there were remarkable poverty declines in the cities of
Awassa and Mekelle between 1994 and 2000. In these two cities, the head count indices fell

31
by 46 percent and 30 percent respectively. Awassa had become a seat of regional
government, which resulted in more professionals moving into the city. This may have had a
positive contribution in reducing poverty in Awassa. In addition, the booming coffee trade in
the late 1990’s may have boosted the economy of Awassa, which grows mainly coffee. The
improvement in Mekelle, on the other hand, may be a result of the reconstruction efforts in
the city after the end of the civil war. There has been a steady flow of investment in the city
during the period. There was also a marginal decrease in the incidence of poverty in Addis
Ababa. Unfortunately, the performance of the remaining cities has not been as impressive.
Poverty increased dramatically in Dire Dawa, where the head count index more than tripled.
This result may in part be associated with the decline in contraband trade. The poverty
incidence had also increased significantly in Bahar Dar, Dessie and Jimma.

2.2.2 Empirical findings concerning characteristics of the poor

Based on panel data (1994, 1995&1997) Kedir reach a conclusion that says that “Chronic
poverty is often strongly associated with households having high dependency rates. While
these may be life cycle effects, such households are nonetheless often persistently poor over
many years, more than the time horizon of this data set. This is indeed the case in urban
Ethiopia, where chronically poor households are more likely to be large and likely to have
more children in them compared to households that are only sometimes poor. Similarly, the
households that were never poor over this period are more likely to be smaller and likely to
have fewer children than those that were sometimes poor. However, the never poor
households are also more likely not to have any household members aged 55years and above
compared to the other groups. The number of adults though tends not to vary very much
across these four groups of households, so indicating that poor households in general and the
chronically poor in particular typically have somewhat higher dependency rates. This of
course is potentially a very important determinant of persistent poverty.

There are also important variations across households according to the characteristics of their
head. A greater proportion of poor households are female-headed compared to the never
poor, though among the poor female headed households are not more likely to be chronically

32
poor. There are some variations by ethnicity, with the Gurage being more likely to be
chronically poor and the Tigre less so.

The marital status and religion of the head were not strongly associated with poverty status
(results not presented). However, the strongest association between poverty status and the
characteristics of the household head is with education. The heads of households that are
never poor are much less likely to have no schooling and much more likely to have
completed secondary education or above compared to the poor in general, but again
especially in comparison with the chronic poor. Low levels of education are clearly another
strong feature of chronic poverty.

Chapter Three
METHODOLOGY

33
3.1 Nature and Source of Data

This study is based on the 2000 household survey data obtained from the EUHS conducted
by the Department of Economics, with the collaboration of the Department of Economics,
Goteborg University.

The surveys were conducted over a period of four successive weeks during a month
considered to represent average conditions so that seasonal factors affecting the patterns of
household consumption and prices will be least operative. Seven cities and towns thought to
reflect the major socio-economic characteristics of the urban population in Ethiopia were
selected for the survey. These were the capital Addis Ababa, Awassa, Bahar Dar, Dessie,
Dire Dawa, Jimma, and Mekele.

A total sample size of 1,500 households was allotted in proportion to the size of the
population residing in the selected urban centers in 1992. Thus, 900 households were drawn
from Addis Ababa and smaller samples from each of the remaining cities (Tesfaye, 2006).
Proportional samples were then taken from all woredas (districts) in each of the urban centers
and half of the kebeles (the lowest administration units) selected randomly from each
woreda. Finally, using the registration of residential houses at the kebele administrative
offices as the sampling frame, systematic sampling was used to select households from each
of the kebeles. Using such a frame captures households living in own residences, government
and kebele houses and tenants in registered private housing. Such an approach, however, fails
to capture homeless individuals and family units. If poverty is particularly pronounced
among the homeless, as one might expect a prior, the level of poverty measured may be
underestimated.

In the survey, information was collected on a multitude of socioeconomic variables of

interest including the structure and composition of the household, educational and health
status, employment and income, consumption and expenditure, and credit.

3.2 Method of Data Analysis

34
To meet the very objectives of the study the analysis should be undertaken with a great care.
To this end, the researcher will employ different data analysis methods consisting of both
descriptive (includes: percentage, averages, indexes, etc.) and econometrics tools.

3.2.1 Model specification

As it is mentioned earlier, the study tries to apply econometrics tools in order to analyze

 The determinants of urban poverty

 The relationship that exist between expenditures on different food items and other
interest of variables (especially with percentage share of women’s income in a
household.)

A) Model 1

In order to analyze the correlates of urban poverty a Logistic regression model was employed
with the dependant variable being the dichotomous of whether the household is poor (1) or
not poor (0). The explanatory variables considered in the analysis are demographic
characteristics (sex, age, and family size), marital status, educational level, occupation,
religion and location.

Construction of poverty line

To begin with, the econometric analysis, first defining the poverty line for the period, which
the survey was conducted, is mandatory. The households are classified as poor and non-poor
by using the poverty line as a cut off point.

To do so, in this study a poverty line estimate previously determined for urban Ethiopia by
Tadesse (1999) which was 88 birr per month per adult equivalent is used. The food poverty
line that underline this estimate was defined using the cost of basic needs approach. Tadesse
(1999) constructed a food basket consisting of the average quantities of food consumed by a
standard consumer to acquire 2200 cal of energy per adult per day to determine the poverty
line. The bundle of urban household consumption items value at market price to obtain food

35
poverty line. In this analysis hence, if a household average consumption is below 88 birr per
month the household is classified as poor and if not below 88 birr, not poor.

The Logit Model

Logit model is appropriate when we assume the random components of response variables
follow binomial distribution & when most variables have categorical responses. Put
differently, it is suited when the dependent variable is dichotomous and of the type that have
a yes or no response. The form of the Logit model following Gujarati (2006) is:

Y    1 X 1   2 X 2   3 X 3  .....   k X k   ……………………… (1)

 i X 1 runfrom1 X 1t to k X K

Where,

Ý = Probability of a household being poor or non-poor

 = Intercept (constant) term

 k =Coefficients of the predictors estimated using the maximum likelihood method

Xi= Predictors (independent variables)

 = Random effect (error term)

Aggregating the value yields

k
   k X k  
Ý= k 1 ……………………………. (2)

In practice Y is unobserved, and  is symmetrically distributed with zero mean and has
cumulative distribution function (CDF) defined as F (  ). What we observe is a dummy
variable y, a realization of a binomial process defined by

1ify  o 
 
y= 0othewise  ………………………………………. (3)

From equation (2) leaving the constant term and rewriting the model yields

36
  k 
  k X k    0 
Prob(Y=1) = Prob  k 1 

 k

     k X k 
= Prob  k 1 

 k 
1 F   k X k 
=  k 1  …………………………….. (4)

The Logit model usually takes two forms. It may be expressed in terms of Logit or in terms
of event probability. When expressed in Logit form, the model is specified as

 P y  1  K
1  P y  1     K X K
Log   K 1 …………………………………… (5)

Using equation 4 and 5 can be transformed into a specification of the Logit model of event
probability by replacing the general CDF, F, with a specific CDF, L representing the Logistic
distribution

 k X k
k  k  e k 1
Pr ob y  1  1  L   k X k   L   k X k   k
 k 1   k 1   k X K
1  e k 1 …………………. (6)

The above equation represents the probability of an event occurring. For a non-event, the
probability is just 1 minus the event probability.

k
  k X k
 k  k  e k 1
Pr ob y  0      k X k   L   k X k   k
 k 1   k 1    k X k
1  e k 1 …………………. (7)

37
On the light of the above explanation, we can develop our urban poverty model as follows

UP=f (AGE, AGE SQUARE, SEX, EDUST, MS, OCC, FS, DLC, U)

The list of variables with their definition is given below

EDUST= Educational status of the house hold head

As the variable education is qualitative (ordinal) in nature, it is necessary to consider the

mutually exclusive levels of education separately. Accordingly, the household head can be
classified as illiterate, primary, secondary, and college and above. Since incorporating all
educational level in terms of dummies in to our model results in multicollinearity, we have to
omit one of them: for our case, we omit illiterate.

EDUST1= 1, if the hhh has attended primary school.

= 0, otherwise.
EDUST2= 1, if the hhh has attended secondary school.
= 0, otherwise.
EDUST3= 1, if the hhh has attended college and above
= 0, otherwise.
AGE= Age of the hhh, in year
AGE SQUARE= Square of the age of the hhh
SEX= Sex of the hhh
Sex= 1, if the hhh is female
= 0, otherwise
MS= Marital status of the respondent
Ms=1, if the respondent have married.
= 0, otherwise
OCC= Occupation of the hhh

38
 Occ1= 1, if the hhh is self-employed
= 0, otherwise
Occ2= 1, if the hhh is wage earner
= 0, otherwise.
Occ3= 1, if the hhh is casual worker
= 0, otherwise
Occ4= 1, if the hhh is pensioner
= 0, otherwise
Occ5= 1, if the hhh is unemployed
= 0, otherwise
FS= Family size in numbers
DLC= Duration of living in the city
Dlc= 1, if the house hold live less than 10 years
= 0, otherwise
U= Stochastic term, showing the impact of all other factors not included in the model.

Description of Explanatory Variables

1. Educational status of the household head

This variable is generally expected to have a negative sign. This is so because, under normal
circumstances a more educated hhh has a better employment opportunity and a good access
(know how) in income generating activities. Further, it is known that whenever one achieved
a higher and higher educational level the earning power of the individual will be increased.
2. Age of the household head

One’s experience is expected to increase with time and he/she will be equipped with how to
manage his /her family. In addition to this, it is expected that as age goes up, the assets of a
household is assumed to be increased. However, in some cases, as people gets older they are
unable to perform what is expected from them. As a result, there might be a decline in
income.

39
3. Sex of the household head

Since women in LDCs have not a better access to economic activities, a female-headed
household is expected to be more vulnerable to poverty.

4. Marital status of the household head

It is presumed that being married is well for higher income. Meaning there is a probability of
different income sources.

5. Occupation of the household head

In reality, it is explicitly known that wage earner and those who have their own business are
not that much expose to poverty. In the contrary, disables, casual workers and unemployed
people are usually highly vulnerable to poverty.

6. Family size of the household

Those households, which have large family size, are supposed to have higher probability of
being under poverty.

7. Duration of living in the city: It is expected that households living more than 10 years in
the city are less vulnerable to poverty than new comers.

B) Model 2

In order to analyze the relationship between household consumption pattern and women’s
income we use the simple microeconomic concept i.e. Engel curve. The analysis begins with
a standard Engel curve relationship between expenditure on specific groups of goods and
household income, introduced by Working (1934). This is a linear relationship between the
share of expenditure on each good and the natural logarithm of total expenditure. Household
demographics and other characteristics are also included, following Deaton (1997). The
components of household consumption expenditure are divided into food cereals and pulses,
spices, vegetable and fruits, and drinks & stimulants, to determine whether there happens a

40
systematic relationship between the percentage share of women’s income in the family and
the respective expenditures.

The functional form of the basic household level regression is

Wi = o + 1 (Y/n) + 2(n) + 4 ln(pswi) + 3edust + ui

Where Wi is the share of the budget devoted to good(food item) i, and the five categories of
food items considered in this preliminary study are food cereals & pulses, spices, vegetable
and fruits, milk & meat, and drinks & stimulants for home consumption (that is, excluding
meals bought outside the home). Y is total household income; n is the size of the household;
edust is the educational level of the household head (ordinal variable tacking 1, 2, 3, and 4
for illiterate, primary, secondary and college respectively); pswi is the percentage share of
women’s income in the household and ui is stochastic term, showing the impact of all other
factors not included in the model.

Note; the details of each food type is given below

Cereals and pulses; teff, maize, sorghum, lentils, wheat, barley, millet, split lentils, cow
peas, split cow peas, chick peas, horse beans, shiro (purchased)

Spices; berbere, salt, sugar, onion, garlic, jingible, cooking spices, and other spices

Milk and meat food items; milk/ergo, cheese, butter, beef, mutton, eggs, and fish

Vegetables and fruits; potatoes, tomatoes, carrot, gomen, keysir, karia, oranges, banana,
papaya, pineapple, avocado and others.

Drinks and stimulants; tella, tej, areki, soft drinks, beer, other liquor, coffee, tea, chat,
cigarettes.

In Working’s analysis, goods are necessities or luxuries and have the appropriate sign on the
coefficient, that is, when o < 0, the consumption declines are income rises and if o > 0, the
reverse is true

41
 Chapter Four
RESULTS AND DISCUSSION

In this section, the researcher is going to analyze the available data by employing different
data analysis methods to address the cited objectives. For the sake of understanding, the
descriptive part of the analysis presented first followed by the econometrics analysis. Here it
goes!

4.1 Descriptive Analysis

4.1.1 Poverty issues

Economists and development specialists agree on the perplexity of setting genuine poverty
lines. For instance, the minimum calorie intake requirements for households, which are
believed to play crucial roles for individual cases, in a specified period, though popularly
utilized, are still flawed with debates. This is because households are composed of family
members with different age and sex categories leading to differences in needs, consumption
habits, and preferences. It is also true that the same level of income cannot serve equally the
needs of households that are different in composition.

To minimize such problems scholars have been busy probing for a number of alternatives
among which the adult equivalent scale, which establishes on equivalence in the
consumption of an adult, a child, and extra, is used eminently. This study has adopted the
adult equivalent consumption method in general and the cost of basic needs approaches in
particular. The method of calculation for arriving this can be found in chapter two. Following
this approach, this part discusses major findings of the study.

Table 4.1 Poverty Indexes of the City for the Year 2000
Poverty Indexes
Head count Total poverty Average poverty Normalized Severity
Year index (P0) gap(TPG) gap (APG) poverty gap (P1) index(P2)
2000 0.45 992.4722 22.05494 0.25 0.092
Source: Authors compilation from EUHS 2000

42
As it is evident from the above table, the head count index of the city is 0.45 referring that
out of the total 100 surveyed households 45% are under absolute poverty. Put differently, 45
households are unable to meet the minimum requirement of 88 birr per month per adult for
the purchase of life sustaining food items. TPG in its part shows that 992 birr per month is
needed (required) for making those households get out of poverty. In average terms, 22 birr
per month per adult should be given as a transfer for the 45 households who are in absolute
poverty to boost the purchasing power (dollar vote) of those households and thereby meet the
minimum 88 birr per month per adult equivalent. The table also tells that the normalized
poverty gap is 0.25 showing the size of income shortfall in relation to the poverty line.

Poverty analysis for different socio-economic and demographic aspects

To take a deep look at on the possible cause and determinant of poverty levels of different
socio-economic characteristics, it needs the compilation of poverty indexes for each
individual determinant that we are interested in. The findings (specifically P 0) in the table
below suggest that, the incidence of poverty is high among female-headed households. The
reasonable argument for this situation is that, males have the highest access for income
generating activities than their female counterpart in our country in general and in the city in
particular. However, when we see P1 and P2, they show that even if out of the total female
headed households approximately half of them are under poverty, less amount of birr per
month per adult is require for those families to enable them get out of poverty than poor
male-headed households.

Though it can not be conclusive with respect to the age of the household head over the whole
sample, it seems that poverty declines with age up to a certain age, specifically up to 55 years
of age. Thereafter, the incidence of poverty increase with age. The decrease in poverty up to
a certain age can be interpreted as the negative impact of work experience of the head (which
is gained through age) on poverty.

On the other hand, poverty increases with household family size. This may reflect the
importance of family planning for reducing poverty. However, care should be taken in
interpreting the result because some studies argue that the hypothesis that poverty leads

43
households to have more children is equally plausible implying that the direction of causation
between poverty and family size require further investigation.

Table 4.2 Poverty Indexes for Different Demographic Characteristics

Determinants Poverty indexes
P0 P1 P2
Sex of the [Link] 0.4375 0.2239 0.1645
HHH [Link] 0.5054 0.2415 0.1474
Age of Up to 35 0.4667 0.2471 0.1632
the HHH 36-45 0.4285 0.2388 0.1423
46-55 0.4371 0.2365 0.1198
56-65 0.5023 0.2234 0.1068
>65 0.5123 0.2547 0.1406
Household [1-3] 0.4578 0.2243 0.1345
family [4-6] 0.4678 0.2298 0.1678
size [7-9] 0.5587 0.2034 0.1106
>9 0.5754 0.2456 0.1587
Source: Authors computation from EUHS 2000

The findings with respect to educational level of the head reveal that poverty indexes
decrease with an increase in educational attainment i.e. when we see illiterate to primary, the
poverty indexes decrease, also it decrease from primary to secondary and then to college
level of education. A reduction in the extent of poverty with level of education of the HHH
confirms that how education is significant in fighting poverty.

The poverty profiles by the type of activity that the HHH engaged demonstrates that those
households whose head are the unemployed and casual worker is the most vulnerable for
poverty. The reason is basically that these households do not have regular and sufficient
income (earning). In the other case, pensioners are also highly attacked by poverty while
wage earners and own account workers experience the least incidence of poverty in relative
terms.

Table4. 3 Poverty indexes for different educational levels and economic activities

44
 Determinant Poverty indexes
P0 P1 P2
Illiterate 0.5687 0.1978 0.1547

Educational level of the HHH Primary 0.4756 0.1647 0.0987

Secondary 0.3047 0.2057 0.1745
College 0.1978 0.1489 0.1147
Unemployed 0.6425 0.2745 0.1945
Casual worker 0.6234 0.2945 0.2024
Main activity of the HHH Pensioner 0.5027 0.2957 0.1297
Own account worker 0.4594 0.2345 0.1345
Wage earner 0.4376 0.1879 0.1247
Source: Authors compilation from EUHS 2000

4.1.2 Food security

The idea of food as a human right might be as old as human history, since food and nutrition
security is a primary concern in any society. In 1948, the United Nations incorporated for the
first time the freedom from hunger and malnutrition into the Universal Declaration on
Human Rights, Art. 25:

In addition, the “International Covenant on Economics, Social and Cultural Rights” (Art. 11)
as well as the “Convention of the Rights of the Child” (Article 24) included aspects of food
and nutrition security. Amongst those were adequate food, physical and mental health,
medical services, disease treatment, hygiene, sanitation, sustainable environment and care.

Nevertheless, although accepted nation-wide, the right to adequate food and to be free from
hunger has not yet been given sufficient attention in the context of operational development
concepts.

To have a crude look at on food security, let us take meals per day of the concerned
households

Table 4.4 Meals Eaten Per Day

Meals Households Total

45
 per day Female headed Male headed
In number In In number (2) In percentage In number In
(1) percentage (1+2) percentage
1 7 20 6 9.23 13 13
2 10 28.57 8 12.31 18 18
3 14 40 40 61.54 54 54
4 4 11.43 11 16.92 15 15
Total 35 100 65 100 100 100
Source: Authors computation from EUHS 2000

As we can observe from the above simple presentation, households, which constitute the
lion’s share of the total, ate 3 meals per day. On the other hand, those households taking 2
meals per day have relatively higher figure than the others. Having considered the sex of the
household head, we arrive at a deduction that female-headed households take a lesser amount
of meal per day than male-headed households do. However, meals per day do not suggest the
achievement of food security or not, it gives some rough clue about it i.e. those households
which can afford 3 or 4 meals per day for each members of the household have a higher
probability of being food secured. Since 31% of the total surveyed household ate only one or
two meals per day we suggest that there is a higher probability of incidence of food
insecurity in the city.

The mean consumption expenditure per adult equivalent on food consumption has also been
compiled for the city to highlight the average standard of living enjoyed by the dwellers of
the city. And its figure is 125.6 birr per month per adult equivalent, which is well above the
minimum requirement. However, it is not plausible to say that food security in the city is on
better status because there is a considerable variation among households i.e. the maximum
was 495.033birr per month per adult equivalent and the minimum was 25.7 birr.

4.1.3 Women employment

Women have lower participation rates and face higher level of unemployment compared to
their male counterparts. Empirical evidence indicates that the unemployment rate among
young women (20-24) was 38.7 per cent while it was only 23.2 per cent for young men in
that age category during the same year. These figures stand in contrast with the general trend
for the Sub-Saharan region, where the unemployment rates for young men (23.1 per cent)

46
exceed those for young women (18.4). The unemployment rates for the total labor force,
youth and urban youth, disaggregated by gender indicates that women experience higher
levels of unemployment than their male counterparts in Ethiopia. This is consistently true for
all working age groups (10 years and above), for all youth (both urban and rural youth) as
well as for urban youth across all reference periods.

Unemployment rates have increased more for women than for men over the five years prior
to 2004. In 1999, the youth unemployment rate among women was 17.3 per cent, compared
to 6.8 per cent among men. The urban female youth unemployment rate was 43.7 per cent
compared to 29.4 per cent for urban male youth. This concurs with the findings from
MOLSA data on registered job seekers. In 2003, female job seekers accounted for 56 per
cent of the total registered job seekers. This upward trend in unemployment is, in part, a
consequence of the increased participation of women in the labor force and in education.

To have a look at on the employment condition of women in the city, we should examine the
main activity of adult females whose age fall under the labor force (16-64). In our
examination of women’s employment, we exclude adult females at school and disables who
are not on the position to undertake any economic activity.

Table 4.5 Women’s Main Activity

Type of main activity of Respondents
adult females In number In percentage
House wife 55 34.48
Domestic worker 17 10.96
Casual worker 3 1.93
Unemployed 19 12.25
Pensioner 8 5.16
Household female business 15 9.67
Own account worker 6 3.87
Wage earner 38 20.64
Total 155 100
Source: Authors compilation from EUHS 2000

From the above table we observe that most of adult females engaged in non-economic
activities i.e. 34.48% of the total adult females are house wife implying that they do not have
their own income generating source. In other words, they are merely dependant on their

47
husbands’ income which makes them subordinate and ineffective in passing decisions
concerning the household. The number of female unemployed out of the total adult females
is also higher with respect to the other non-economic activities. Surprisingly, only 20.64% of
adult female are wage earner and 3.87% has run their own formal business. In general,
almost 78% of adult female are engaged in non-economic activity in the city.

4.2 Econometrics Analysis

4.2.1 Determinants of poverty

The descriptive analysis in the previous section has already clearly identified some
determinants of the probability of being under poverty. However, to investigate this more
carefully calls for a multivariate analysis, considering many factors together. This is
considered here by estimating the factors influencing the likelihood of a household being
under poverty, by means of a logit model. The explanatory variables used in this model are
summarized in chapter three; these include characteristics such as household demographics;
main economic activity of the head; education of the head; gender, and residence before ten
years. As before, these are the values of these variables in the survey year (2000). While
many of these were considered individually in the previous section, the regression model
enables the simultaneous effects of these different factors to be considered and so gives a
more robust assessment of their importance. For the estimation of the model, stata-9
application software was used.

Before analyzing determinants of urban poverty, all variables, which were hypothesized to
depict incidence of poverty in the study area, were checked for multicollinearity using pair-
wise correlation coefficient. The result of the test confirmed that there is a moderate
collinearity between college education and wage employment (0.47) whereas the rest of the
variables did not show significant collinearity. Wage employment and college education are,
therefore, taken as detrimental variables for their collinearity is less than 0.7.

Besides, the overall fitness of the model is found to be very good because the calculated
value of LR chi2 is much greater than its tabulated value. In other words, its p-value is 0.

48
Basing the estimates and the above tests, now we are going to interpret and analyze the
significance of each explanatory variable in determining the probability of being poor. In
addition, the type of relationship that exists between the explanatory variables and dependant
variable was analyzed.

Table 4.6 Logit maximum likelihood estimates

Explanatory variables Coefficient Z P>|Z|
sex of hhh 2.4826 1.99 0.043**
primary -0.1132247 -0.12 0.907
secondary -3.652882 -1.97 0.049**
college -3.117586 -2.02 0.034**
age of hhh -0.1563392 -2.9 0.004***
age square 0.0005966 1.35 0.178
marital status -3.499883 -3.07 0.002***
wage earner -3.665469 -1.78 0.097*
own account worker -3.42582 -1.58 0.139#
pensioner 0.6524294 0.22 0.827
casual worker 0.0387654 0.01 0.989
unemployed 0.1909243 0.06 0.95
family size 0.9636165 2.91 0.004***
residence before 10 year 0.3028024 0.26 0.798
Source: Authors computation from EUHS 2000
***, **,*, # significant at 1, 5,10and 15 percent level respectively

As we can clearly saw from Table 4.6 of this page and Table 8 of the annex, gender of the
household head has a positive and significant relationship with the probability of the
household to be poor i.e. those households whose head are female are highly affected by
poverty than male-headed households. The possible and reasonable argument for this
situation could be male-headed households had good opportunities for employment and
investment, and had wider probability to escape from the prevail poverty than females-
headed households.

The econometrics results from the logit model suggest that, the probability of being poor has
a significant and negative correlation with age. Meaning, the higher the age of the household
head the lesser probability that the household being under poverty. This may be explained by
the benefit in terms of the increase in earning power associated with work experience of the

49
head as his age increases. However, from the positive age square result we can draw an
inference that the probability of being poor is higher at very low and very high age levels.

As far as the marital status of the household head and the probability of being under poverty
are concerned, a negative and significant relationship does exist between them. This is to
mean that the probability of being poor for households’ whose head is at marriage is less than
those households headed by persons with out marriage (divorced, widowed and single). This
can be elucidated by the argument that those households at marriage have the chance of two
different income sources.

The econometric results also reveal that, household family size has a positive and significant
effect on the probability of being poor. This attests that households with large number of
family size are more likely to be poorer than households with small (few) members. This
may suggest that the disadvantage from free riding in consumption by some household
members is by far larger than the benefit that the household obtains as a result of returns to
scale in consumption as household size increase in the city.

On the other hand, the impact of educational level of the household head is found to be a
significant one. This is to mean that all the dummies for completion of primary, secondary
and college are negative associated with probability that the household is poor though their
significance is not same i.e. secondary and college educational levels are found to be
significant while primary is not. This discloses that those households with formally educated
heads are less likely to be under poverty than those with no formal education. In other words,
the probability of being poor increases as the household does not have formal education. This
is not surprising since it is consistent with human capital theory where education is assumed
to be increase production and hence earnings. This is also intuitively appearing especially in
the urban areas where formal education is taken as a requirement in getting a job at least for
skilled laborers. On the top of the significance, one point that should be get stress is that the
coefficients of the schooling dummies are not the same for the different levels of education.
This may suggests that different levels of education affect poverty in different amount
compared to those no formal education at all.

50
Looking on the econometrics results, we can also identify the relationship that exists between
type of economic activity of the household head and the probability of being poor. As we
have seen from the table, wage earner’s and owns account workers are less likely to be
exposed to poverty while those families whose head are casual worker, unemployed, and
pensioner are in risk of poverty.

Finally, residence before ten years has a positive but insignificant relationship with the
probability being under poverty i.e. those households which lived more than ten years in the
city have relatively low probability being under poverty.

Marginal effect of the explanatory variables on the probability of being poor

Table 10 in the annex displayed the marginal effect of each individual explanatory variable
on the probability of being under poverty. From this table we understand that the marginal
effects of sex, marriage and college are found to be much higher. And the marginal effects of
secondary, wage, and own account worker also appeared significant.

4.2.2 The effect of percentage share of female’s income in a family on the

expenditure of different food items

Based on the already specified Engel curve model in chapter three, the respective
expenditures on each food items was regressed on per capita income, household family size,
educational attainment of the household head, and natural logarithm of the percentage share
of females income in a household and we get the following results.

51
Table4.7 The effect of percentage share of female’s income in a household on
food expenditure
Explanatory Dependant variable: Expenditure on different food items
variables Cereals and Spices Milk and meat Vegetables Drinks and
pulses and fruits stimulants
Female income 1.525 0.201 0.596 0.315 -0.108
share (1.763) (1.988)* (1.370)  (2.192)** (0.922)
Per capita 1.360 0.418 0.293 0.168 0.115
income (8.904)*** (7.399)*** (3.857)***  (4.464)*** (3.548 )***
Family size 35.923 10.496  9.747 3.551 2.201
(7.976)*** (6.299)*** (4.352)** (3.195)***  (2.299)**
Educational -4.603(-0.386)  11.857  5.97(1.006) 7.688 4.081
level of HHH (2.685)*** (2.610)** (1.609)
R-SQUARE 0.649 0.549 0.358 0.374 0.214
Source: Author’s computation from EUHS 2000

Note: the numbers in the bracket are the t-values of each parameter.
: ***, **, *, Significant at 1, 5 and 10 percent level respectively

As we can observe from the above Engel curve regression results, both income per capita and
family size have a positive and significant relationship with expenditure on cereals and
pulses at 1% significance level. This is to mean that as income per capita or family size
increases, the expenditure on cereals and pulses will increase as well, keeping the other
explanatory variables unchanged. On the other hand, the effect of percentage share of
female’s income in a household is positive but statistically insignificant at 5% significance
level. On the other extreme, the effect of education in this particular food item expenditure is
negative and insignificant. The implication is that as the household head completed higher
and higher education, his expenditure on cereals and pulses decrease, citrus paribus, though
by an insignificant amount.

Having diverted our concern from cereals and pulses to spices, we get similar but slightly
different results from the former finding. The similarity is that both per capita income and
family size retain their positive and significant effect here also. However, the difference
arises from the relationship that exists between education and expenditure on spices i.e. the
Engel curve regression result suggests that there is a positive and significant relation ship
between education and expenditure on spices. When we take a look at on the effect of
percentage share of female’s income in a family on the expenditure of spices, we can say that

52
it has a similar effect as that of expenditure on cereals and pulses. Even if this deduction is
right, however, the effect of this explanatory variable on the two expenditures differs on the
ground of its significance. Meaning at significance level of 5% the effect on both
expenditures is not worthwhile. However, if we take 10% significance level the result will be
different i.e. the extent of the effect of percentage share of female’s income in a family is
appeared to be greater on spices expenditure.

As far as the effects of income per capita and family size on the expenditure of milk and meat
food items are concerned, the Engel curve regression result confirms that the two explanatory
variables still keep their positive and significant effect. Where as both education and
percentage share of female’s income in a family has a positive but insignificant effect on the
expenditure of milk and meat at both 5 and 10 percent significance level. Since our business
here is to examine the effect of percentage share of female’s income in a family on different
types of food expenditures, we have to compare the degree (significance) of its effect on each
food item. By doing so, we can say that the effectiveness of percentage share of female’s
income in a family in influencing the expenditure pattern of households on milk and meat is
found to be lesser than the above two.

As it is evident from the Engel curve regression result, income per capita and family size are
found to be positively as well as significantly related with expenditure on vegetable and fruits
once again. Moreover, education is also has a positive and significant effect on household’s
expenditure on vegetables and fruits at 5% significance level. In other words, as the level of
education of the households head increases the expenditure on vegetables and fruits
increases, citrus paribus. The regression result also attested that the percentage share of
female’s income in a family has a positive and significant effect on vegetables and fruits
expenditure at 5% significance level. This could be interpreted as the more the share of
females’ income in the family, the higher expenditure on vegetables and fruits will be.

Finally, observing the Engel curve regression results once again bears out that, income per
capita and family size till keep hold of their positive and significant effect on the expenditure
on drinks and stimulants. What is strange here is that, the impact of percentage share of
female’s income in a family has a negative but insignificant effect on the expenditure of

53
drinks and stimulants. The result also told us that educational level and expenditure on drinks
and stimulants are positively related.

In conclusion, from the overall findings of this sub section we perceive that percentage share
of female’s income in a family has a positive and significant effect on the expenditure of
spices, and vegetables and fruits. In addition, it is found that there is a positive but
insignificant relationship between percentage shares of female’s income in a family and
expenditure on cereals and pulses, and milk and meat food items while its effect on drinks
stimulants is negative and insignificant.

What is the implication of the results? The implication of the results is that if female’s
income shares in a family increases, it is possible to achieve not only food security but also
nutrition security. For this to happen, it is obligatory to empower women in socio-economic
and political affairs.

54
 Chapter Five
CONCLUSION AND RECOMMENDATION

5.1 Conclusion

The objectives of the study were: to assess determinants and their quantitative relationships
up on urban poverty, to examine the effect of percentage share of female’s income in a
family on food expenditure, and the condition of women employment in Bahar Dar city. To
this end, primary data sources that were obtained from EUHS were used to carry out the
study. A total of 100 households were selected from the dwellers of the city during the 2000
EUHS.

The research used the cost of basic needs approach in the identification of the poor from the
non-poor. Based on this approach out of the 100 surveyed households 45(45%) of them were
found below the poverty line. That incidence of poverty is rampant and hardcore among the
surveyed: 0. 45 the head count, 0.25 normalized poverty gap, and 0.092 as the severity index
in the town respectively calls for urgent interventions aimed at curbing the fate of the poor.
One way of doing this is through identifying factors that account for urban poverty. This,
however, requires analytical rigor, as the factors are complicated and important as well in
fighting against poverty. Although a number of specific conclusions could be drawn from the
estimation results, summaries and sum ups of this research, among other things, include the
following:

The study found out that female-headed households are more likely to be poorer than
households of which the head is men.
Household size was positively and significantly correlated with poverty in Bahar Dar
city as the study depicted. This has a clear implication for the residents of the city in
that households with large size will fall into the hardcore sections of poverty easily
than those who have not.

55
 Educational status of the head was appeared a significant factor in determining the
probability of being poor. Meaning, those households whose head has attended more
schooling is less likely to be in absolute poverty.

Age of the household head also has a negative and significant effect on the
probability of being poor. This is to mean that as age of the hhh increases the
probability of being under poverty go down. However, this did not hold for all age
levels because as we can observe from the age square result, the very high the age of
the hhh is, the higher the probability of the household being under poverty.

The economic activity of the hhh also has found a significant effect on the status of
poverty of a household i.e. households whose head are wage earner and own account
worker have less probability of being under poverty.

Most of the surveyed households i.e. 54% could afford three meals per day for their
members. On the contrary, a significant amount of households were unable to eat
three meals per day.

As to the employment of women the study found that 78% of adult females were
engaged in non-economic and less paid jobs and activities.

Finally but most importantly, the study found that the percentage share of female’s
income in a household has a positive and significant effect on expenditures of a
household on vegetables and fruits, and spices.

5.2 Recommendations

Based on the findings reached up on, the study come up with the following recommendations

 As it was evident from the head count poverty index, nearly half of the surveyed
households were under poverty i.e. they were unable to meet the minimum
expenditure on food items for survival. From this result, therefore, the study suggests
that urban poverty should be get due emphasis from the concerned body.
 The findings revealed that most female-headed households were under poverty than
their male counterparts. In addition, the lion share of adult females was engaged in

56
 non-economic and less remunerative jobs and activities. Thus, it entails that both
intensive and extensive projects and programs must be done to empower
women in socio-economic and political affairs. Here, it is worth mentioning to put
the assertion of UNDP in its 1997 annual report i.e.” if development is not
engendered, it is endangered. And if poverty reduction strategies fail to empower
women they will fail to empower society.”

 As the educational attainment of the head of the household is found to be the most
important factor associated with urban poverty, it clearly suggests ways of focusing
on the value of education. Adequate education is central in addressing incidence
of poverty. Specifically, college education is found to be of paramount importance
in reducing poverty in Bahar Dar city.

 The other most important recommendation that the study come up with is that
creating a good employment opportunities for women has a paramount of
significance not only in the alleviation of poverty and achievement of food security
but also for realization of nutrition security

 Since most of the households which are headed by old age persons are under
absolute poverty, there should be a good social security system as far as what is
possible.

57
 Bibliography
Abbi [Link] and Andrew McKay (2003), Chronic Poverty in Urban Ethiopia: Panel Data
Evidence, Paper prepared for International Conference on ‘Staying Poor: Chronic
Poverty and Development Policy’, hosted by Institute for Development Policy and
Management, University of Manchester, UK, 7 – 9 April 2003

Abebe Shimeles and Arne Bigsten (2008), The persistence of urban poverty in
Ethiopia: A tale of two measurements, Working Papers in Economics No 283,
Goteborg University

Asmamaw Enquobahrie (2004), Understanding Poverty: The Ethiopian Context, A Paper

presented at The Gambia AAPAM Roundtable Conference, Banjul, The Gambia, April
19 - 23, 2004

Constantina Safiliou-Rothschild (2001), Food Security and Poverty: Definitions and

Measurement Issues, Dialogue Working Paper 4

Dejene Aredo (2005), Migrant Remittances, Shocks and Poverty in Urban Ethiopia: an
Analysis of Micro Level Panel Data, Addis Ababa University

Elisa Muzzini (2008), Urban Poverty in Ethiopia: A Multi-faceted and Spatial

Perspective, The World Bank Group, Washington D.C.

Esubalew Alehegn (2006), Determinants of Urban Poverty: The Case of Debre Markos,
MA Thesis in Regional and Local Development Studies

G.S. Maddala (1992), Introduction to Econometrics, second edition, Macmillan publishing

company, New York

Gujarati, Damodar N. (2004), Basic Econometrics, 4th edition, Tata McGraw-Hill

Company, New York

Ines Reinhard and K.B.S Wijiyaratne (2000), The use of Stunting and Wasting as
indicators of Food Insecurity and Poverty, Working Paper 27

John Simister and Jenifer Piesse (2002), Household Consumption Decisions and Nutrition
in South Africa, Centre for the Study of African Economies Conference,St Catherine.s
College, Oxford

58
Martin Ravallion (2005), Evaluating Anti-Poverty Programs, Development Research
Group, World Bank

MEDAC and UNCHS (Habitat) (2000), An Overview of Urban Poverty in Addis Ababa,
Addis Ababa.

Michael P. Todaro and Stephen C. Smith (2003), Economic Development, 8th edition,
Dorling Kindersley Pvt. Ltd, India

Milu Muyanga (2006), Household Vulnerability to Transient and Chronic Poverty:

Evidence from Rural Kenya, Egerton University (Kenya)

Quisumbing, Agnes R. and Bonnie McClafferty (2006), Using Gender Research in

Development, International Food Policy Research Institute

Raj Nallari (2004), Poverty Measurement, World Bank Institute

Sarah Gammage (2006), A Menu of Options for Intra-Household Poverty Assessment,

United States Agency for International Development

Tesfaye Alemayehu Gebremedhin (2006), The Analysis of Urban Poverty in Ethiopia, The
University of Sydney

Woodldridge, Jeffrey M., Introductory Econometrics, A Modern Approach, second edition

[Link]

59
 Annex
Table 8 Logit maximum likelihood estimates
Logistic regression Number of obs = 99
LR chi2(14) = 92.02
Prob > chi2 = 0.0000
Log likelihood = -22.362667 Pseudo R2 = 0.6729

------------------------------------------------------------------------------
povstatus | Coef. Std. Err. z P>|z| [95% Conf. Interval]
------------------------------------------------------------------------
sexhhh | 2.4826 1.383928 1.99 0.043** -.2298484 5.195049
primary | -.1132247 .9673393 -0.12 0.907 -2.009175 1.782726
secondary | -3.652882 1.915089 -1.97 0.049** -7.406387 .1006235
college | -3.117586 1.724896 -2.02 0.034** -6.263147 .1000754
age | -.1563392 .0539907 -2.90 0.004*** -.262159 -.0505194
agesquare | .0005966 .0004427 1.35 0.178 -.000271 .0014643
maritalstu~s | -3.499883 1.140234 -3.07 0.002*** -5.734701 -1.265066
wage | -3.665469 2.271472 -1.78 0.097* -9.489446 2.158509
#
ownaccount~r | -3.42582 2.906577 -1.58 0.139 -9.122606 2.270966
pensioner | .6524294 2.990513 0.22 0.827 -5.208869 6.513727
casual | .0387654 2.840184 0.01 0.989 .000423 .527892
unemployed| .1909243 3.05345 0.06 0.950 -5.793728 6.175577
familysize | .9636165 .3313162 2.91 0.004*** .3142487 1.612984
residenceb~s | .3028024 1.186032 0.26 0.798 -2.021778 2.627383
_cons | .908611 3.396427 0.27 0.789 -5.748263 7.565485

Source, Authors computation from EUHS

***, **, *, # Significant at 1, 5, 10 and 15 percent level respectively

60
 Table 9 Test of multicollinerity using pair-wise correlation coefficient for model one

Pwcorr

| povsta~s sexhhh primary second~y college age agesqu~e

-------------+---------------------------------------------------------------
povstatus | 1.0000
sexhhh | 0.3277 1.0000
primary | 0.1608 0.0859 1.0000
secondary |-0.4797 -0.1628 -0.4561 1.0000
college |-0.1823 -0.2200 -0.3135 0.1988 1.0000
age |-0.3459 -0.0969 -0.0500 0.1250 0.0421 1.0000
agesquare | 0.1014 -0.0986 0.0173 -0.2432 -0.2449 0.2514 1.0000
maritalstu~s |-0.3295 -0.4557 -0.1803 0.4427 0.2541 0.2611 0.0595
wage |-0.3771 -0.1610 -0.2010 0.3610 0.4726 0.0533 0.2618
ownaccount~r |-0.1862 -0.2054 0.0368 0.0524 -0.1286 -0.0735 0.0265
pension | 0.1384 -0.2017 0.1559 -0.1444 -0.1833 0.0300 0.2215
casual | 0.2974 0.2023 -0.0074 -0.2199 -0.0626 -0.0105 0.0459
unemployed| 0.1962 0.3903 0.0953 -0.1203 -0.1707 0.0193 0.2031
familysize | 0.2726 -0.1884 0.1071 -0.1634 0.0391 0.0764 0.0049
residenceb~s | 0.0288 -0.0047 -0.0899 0.0410 0.0997 -0.0519 0.0105

| marita~s wage ownacc~r pension casual unempl~d family~e

-------------+---------------------------------------------------------------
maritalstu~s | 1.0000
wage | 0.3176 1.0000
ownaccount~r | 0.1823 -0.3340 1.0000
pension |-0.0527 -0.3034 -0.2446 1.0000
casual |-0.2876 -0.2502 -0.2017 -0.1833 1.0000
unemployed|-0.2270 -0.2826 -0.1607 -0.2070 -0.1707 1.0000
familysize |-0.0483 0.0598 0.0339 0.0998 -0.0235 -0.1460 1.0000
residenceb~s | 0.0454 0.1210 -0.0688 0.0342 -0.0720 -0.1782 0.0063

| reside~s
-------------+---------
residenceb~s | 1.0000

61
Table 10 Marginal effects of the explanatory variables on the probability of being poor

Marginal effects after logit

y = Pr(povstatus) (predict)

= .29785064
Variable dy/dx Std. Err. z P>z [ 95% C.I. ]
sexhhh* .5072762 .23345 2.17 0.030 .049713 .964839
primary* -.0235608 .19964 -0.12 0.906 -.414843 .367722
second~y* -.4810376 .12132 -3.97 0.000 -.718818 -.243257
college* - .6444211 .22401 -2.88 0.004 -.805361 .48348
age -.0326961 .01309 -2.50 0.013 -.058355 -.007038
agesqu~e .0001248 .0001 1.23 0.217 -.000073 .000323
marita~s* -.6105133 .14135 -4.32 0.000 -.887551 -.333475
wage* -.5230431 .30152 -1.73 0.083 -1.11401 .067924
ownacc~r* -.4395887 .24553 -1.79 0.073 -.920826 .041649
pensio~r* .1461117 .69775 0.21 0.834 -1.22146 1.51368
casual* -.00806 .58715 -0.01 0.989 -1.15886 1.14274
unempl~d* .0409318 .6698 0.06 0.951 -1.27185 1.35372
family~e .2015265 .06444 3.13 0.002 .075224 .327829
reside~s* .0659258 .2678 0.25 0.806 -.458953 .590805
(*) dy/dx is for discrete change of dummy variable from 0 to 1

62
Table 11(a-e) Engel curve regression results

Coefficientsa

Standardi
zed
Unstandardized Coefficien
Coefficients ts
Model B Std. Error Beta t Sig.
1 (Constant) 74.421 40.269 -4.331 .000
FEMSHARE 1.525 .552 .185 1.763 .211
PERINC 1.360 .153 .566 8.904 .000
FAMSIZE 35.923 4.504 .538 7.976 .000
EDUST -4.603 11.935 -.024 -.386 .701
a. Dependent Variable: CEREALS

Coefficientsa

Standardi
zed
Unstandardized Coefficien
Coefficients ts
Model B Std. Error Beta t Sig.
1 (Constant) 55.841 14.898 -3.748 .000
FEMSHARE .201 .204 .075 1.988 .098
PERINC .418 .057 .533 7.399 .000
FAMSIZE 10.496 1.666 .482 6.299 .000
EDUST 11.857 4.415 .188 2.685 .009
a. Dependent Variable: SPICES

63
 Coefficientsa

Standardi
zed
Unstandardized Coefficien
Coefficients ts
Model B Std. Error Beta t Sig.
1 (Constant) 34.527 20.026 -3.722 .000
FEMSHARE .596 .275 .197 1.370 .233
PERINC .293 .076 .331 3.857 .000
FAMSIZE 9.747 2.240 .397 4.352 .000
EDUST 5.971 5.935 .084 1.006 .317
a. Dependent Variable: MILKMEAT

Coefficientsa

Standardi
zed
Unstandardized Coefficien
Coefficients ts
Model B Std. Error Beta t Sig.
1 (Constant) 6.207 9.937 -4.650 .000
FEMSHARE .315 .136 .208 2.192 .023
PERINC .168 .038 .379 4.464 .000
FAMSIZE 3.551 1.111 .288 3.195 .002
EDUST 7.688 2.945 .215 2.610 .011
a. Dependent Variable: VEGFRUIT

Coefficientsa

Standardi
zed
Unstandardized Coefficien
Coefficients ts
Model B Std. Error Beta t Sig.
1 (Constant) 11.961 8.558 -1.398 .166
FEMSHARE -.108 .117 .093 .922 .359
PERINC .115 .032 .338 3.548 .001
FAMSIZE 2.201 .957 .232 2.299 .024
EDUST 4.081 2.536 .148 1.609 .111
a. Dependent Variable: DRINKS

64
65