Determinants of Rural Poverty in Banja District of Awi Zone, Amhara National Regional State, Ethiopia

DETERMINANTS OF RURAL POVERTY IN BANJA DISTRICT OF AWI ZONE, AMHARA NATIONAL REGIONAL STATE, ETHIOPIA

MSC THESIS

DESALEGN TESHALE WOLDIE

FEBRUARY 2019

HARAMAYA UNIVERSITY, HARAMAYA

DETERMINANTS OF RURAL POVERTY IN BANJA DISTRICT OF AWI ZONE, AMHARA NATIONAL REGIONAL STATE, ETHIOPIA

A Thesis Submitted to College of Agriculture and Environmental Sciences, School of Agricultural Economics and Agri-business Management

HARAMAYA UNIVERSITY

In Partial Fulfillment of the Requirements for the degree of Master of Sciences in Agriculture (Agricultural and applied Economics)

By: Desalegn Teshale Woldie

February 2019

Haramaya university, Haramaya

HARAMAYA UNIVERSITY

POSTGRADUATE PROGRAM DIRECTORATE

As thesis research advisors, we hereby certify that we have read and evaluated the thesis prepared, under our guidance and direction, by Desalegn Teshale Woldie, entitled “Determinants of Rural Poverty in Banja District of Awi zone, Amhara National Regional State, Ethiopia”. We recommend that the thesis be submitted as it fulfills the requirements.

Jema Haji (Professor) ______Major Advisor Signature Date

Abule Mehare (PHD) ______Co-advisor Signature Date

As members of the Examining Board of the MSc Thesis Open Defense, we certify that we have read and evaluated the thesis prepared by Desalegn Teshale Woldie and examined the candidate of Collaborative Master of Sciences in Agricultural and Applied Economics.

Mohammed Amman (Mr.) ______Chair Person Signature Date Mengistu Ketema ( PHD) ______Internal Examiner Signature Date Bezabeh Emanna (PHD) ______External Examiner Signature Date Final approval and accepted of the Thesis is contingent upon the submission of its final copy to the council of graduate studies (PGDP) through the candidate’s department or school graduate committee (DGC or PGDP).

STATEMENT OF THE AUTHOR

By my signature below, I declare and affirm that this thesis is my genuine work and that all sources of materials used for this thesis have been properly acknowledged. This thesis has been submitted in partial fulfillment of the requirements for collaborative M.Sc. degree at Haramaya University in Agricultural and applied Economics and is deposited at the University Library to be made available to users or borrowers under the rules of the Library. I confidentially declare that this thesis has not been submitted to any other institution anywhere for the award of any academic degree, diploma, or certificate.

Brief quotations from this thesis are allowable without special permission, provided that accurate acknowledgement of source is made. Requests for permission for extended quotation from or reproduction of this manuscript in whole or in part may be granted by the head of the major department or the Dean of the School of Graduate Studies when in his or her judgment the proposed use of the material is in the interests of scholarship. In all other instances, however, permission must be obtained from the author of the thesis.

Name: Desalegn Teshale Woldie Signature: Place: Haramaya University Date of Submission:

ACCRONYEMS AND ABBREVATIONS

AE Adult Equivalent CSA Central Statistical Agency CBN Cost of Basic Needs DAO District Agricultural Office ENHRI Ethiopian Health and Nutrition Research Institute EFY Ethiopian Fiscal Year ET Ethiopia ETB Ethiopian Birr FPL Food Poverty Line FGT Foster Greer and Thorbeck FAO Food and Agricultural Organization GDP Gross Domestic Product GTP Growth and Transformation Plan HDI Human Development Index HICE Household Income and Consumption Expenditure IFAD International Fund for Agricultural Development IMF International Monetary Fund MDG Millennium Development Goal MoFED Ministry of Finance and Economic Development NFPL Non-food Poverty Line PPT Purchasing Power Parity RGDP Real Gross Domestic Product TPL Total Poverty Line TLU Tropical Livestock Unit UNDP United Nations Development Programm WB World Bank WBI World Bank Institute WM Welfare Monitoring

BIBLOGRAPHY

The author was born in Injibara town of Awi zone, Amhara Regional State on September 11, 1986. He attended his primary and secondary schools in Injibra. Following completion of his Secondary school in 2002, he joined Debub University (the then Hawassa) at Wondogenet college of Forestry and Graduated with Diploma in Forestry in 2004. There after he employed in Ministry of Agriculture in Maichew and Woreta TVET colleges as an as assistant instructor. Soon after he joined Haramaya university in 2008 and graduated with BSc degree in Agricultural Economics in 2012. Since then he was employed in Ethiopian Institute of Agricultural Research in 2013 as a Junior Researcher. He served the institute for three years until he joined Haramaya University in 2017 to pursue his MSc degree in Agricultural and Applied Economics.

ACKNOWLDEGMENTS

First and for most, I indebted to Almighty God, who allowed me to begin and concluded this research thesis.

I would like to express my depest gratitude to my major advisor Professor Jema Haji and co- advisor Dr. Abule Mehare for their valuable guidence and suggestions, for their insightfull thoughts and kindness in providing me valuable and constructive advises. Their generous time devotion from the early design of the research propozal to the final write up of the thesis is key to finalize the thesis timely.

Thirdly, my deepest apprection goes to Ethiopian Institute of Agricultural Research for giving me the chance to join the Masters program. Besides, I would like to express my indebted thanks to Pawe Agricultural Research centre collegeous and more particularly to the centre Dirctors, Birhanu Ayalew (the former) and Yalew Mazngia (the current). It is my pleasure to apperciate African Econmoic Researh Consortium (AERC) for their financial support to accomplish my MSc study at Haramaya University and University of Pretoria.

Finally, I would like to extend my deepest gratutude to my family who are helping and nurturing me through out my life.

. .

vii

TABLE OF CONTENTS

STATEMENT OF THE AUTHOR iv

ACCRONYEMS AND ABBREVATIONS v

BIBLOGRAPHY vi

ACKNOWLDEGMENTS vii

TABLE OF CONTENTS viii

LIST OF TABLES xii

LIST OF FIGURES xiii

LIST OF TABLES IN THE APPENDIXES xiv

ABSTRACT xv

1. INTRODDUCTION 1

1.1. Background of the Study 1

1.2. Statement of the Problem 3

1.3. Research Questions 6

1.4. Objectives of the Study 6

1.5. Significance of the Study 7

1.6. Scope and Limitations of the Study 7

1.7. Organization of the Thesis 7

2. LITRATURE REVIEW 9

2.1. Concepts and Definition of Poverty 9

2.2. Theories of Poverty 10 2.2.1. Individualistic Theory of Poverty 10

viii

…Continued 2.2.2. Cultural Theory of Poverty 10 2.2.3. Geographic Theory of Poverty 11 2.2.4. Structural Theory of Poverty 11

2.3. Theoretical Framework 12

2.4. Determination and Types of Poverty Line 14 2.4.1. Poverty Lines and Types 14 2.4.1.1. Absolute poverty line 15 2.4.1.2. Relative poverty line 15 2.4.2. Setting Poverty Lines 16 2.4.2.1. Cost of Basic Needs Approach (CBN) 16 2.4.2.2. Direct Calorie Intake (DCI) 18 2.4.2.3. Food Energy Intake Approach (FEI) 18

2.5. Measurement and Indices of Poverty Analysis 19 2.5.1. Head Count Index 20 2.5.2. Poverty Gap Index 20 2.5.3. Poverty Severity Index 21

2.6. Time needed to Exit Poverty 22

2.7. Review of Potential Analytical Models 24

2.7.1. Binary Choice Models 24 2.7.2. Ordered Models 25 2.7.3. Multivariate Regression Models 26 2.7.4. Censored Tobit Regression Model 26

2.8. Empirical Review of Poverty in Ethiopia 27 2.8.1. Status of Regional Poverty 29 2.8.2. Empirical Studies on the Determinants of Rural Poverty in Ethiopia 29

2.9. Conceptual Framework of the Study 32

…Continued

3. RESEARCH METHODOLOGY 33

3.1. Description of the Study Area 33

3.3. Sampling Technique and Sample Size 35

3.2. Data types, Sources and Methods of Data Collection 35 3.3.1. Sampling Technique 35 3.3.2. Sample Size Determination 36

3.4. Methods of Data Analysis 37 3.4.1. Descriptive Analysis 37 3.4.2. Poverty Measures and Procedures 38 3.4.3. Econometric Models for Determinants of Household Poverty Status 38

3.5. Definitions of Variables and Working Hypothesis 42 3.5.1. Dependent Variable 42 3.5.2. Independent Variables 42

4. RESULTS AND DISCUSSION 49

4.1. Poverty Line Determination 49

4.2. Poverty Measures and its Status 51

4.3. Consumption Expenditure 53

4.4. Association of Livelihood Capitals of Households with Poverty 55 4.4.1. Human Capital 55 4.4.2. Physical Capital 57 4.4.3. Financial Capitals 59 4.4.4. Natural Capital 60 4.4.5. Other Institutional Capital or Characteristics 61

4.5. Time Needed to Exit Poverty 62

…Continued

4.6. Econometric Model Results 63

5. SUMMARY, CONCLUSSION AND RECOMMENDATIONS 69

5.1. Summary 69

5.2. Conclusion and Policy Recommendations 69

6. REFERENCES 72

7. APPENDEXES 81

LIST OF TABLES

Table Page 1: Trends of national poverty (1995/96—2015/16) 28 2: Trends in regional poverty headcount indices 29 3: List of selected kebeles with their respective no of sample households 36 4: Summary of the variables included in the Tobit model 48 5: Food consumption, food prices and food poverty line (N=190) 50 6: Poverty measure of sample households 51 7: Distribution of Poor and non-poor sampled households by Peasant association 53 8: Total expenditure and food expenditure of sampled households per year in Birr 55 9: Human capitals with poverty level 56 10: Physical capital of sample households with poverty status. 59 11: Financial capitals with poverty status (ETB per annum) 59 12: Natural capitals with poverty status 61 13: Average time needed to exit poverty at 7.7 % growth rate of RGDP of the country 63 14: Tobit model regression estimates and marginal effects of poverty determinants 68

xii

LIST OF FIGURES

Figure Page 1: Poverty and livelihood concetptual framework 32 2: Location of the study area 34

xiii

LIST OF TABLES IN THE APPENDIXES

Appendix Page 1: Conversion factor for kilocalories per gram of different food items 81

2: Conversion factor for tropical livestock unit (TLU) 81 3: Conversion factor for adult equivalent (AE) 82 4: National real gross domestic product in million Birr EFY 2005/6-2009/10 82 5: Tobit regression coefficient and marginal effects after Tobit 83 6: Survey questioner 86

xiv

DETERMINANTS OF RURAL POVERTY IN BANJA DISTRICT OF AWI ZONE, AMHARA NATIONAL REGIONAL STATE, ETHIOPIA ABSTRACT

Poverty is one of the most serious problems of human deprivation and a complex phenomenon. Ethiopian government have been implementing different poverty reduction programs and strategies to fight extreme hunger and poverty. The struggle to reduce rural poverty at household level is a continuing challenge. This study is conducted in Banja district of Awi zone which is known for its low agricultural production and high population. The specific objective of this study was to estimate the rural poverty status and average exit time of poor households and identify factors determining rural poverty. In order to achieve these objectives, cross sectional data on human capital, physical capital, financial capital, natural capital and other institutional charachetrstics were collected from 190 households drawn from randomly selected five kebeles using structured household questioner. Descriptive and inferential statistics, and econometric model were used to analyze data on poverty status and poverty level, respectively. Hence, setting poverty line, identifying poor and non-poor rural households, measuring the incidence, depth and severity, and mean comparison between the groups were made. Accordingly, using Cost of Basic Needs approach, the estimated poverty line was Birr 4301 per adult equivalent per year. The Foster Greer and Thorbeck measure of poverty found that 44 percent of sample households were found below poverty line and the poverty gap and poverty severity were 9 percent and 2 percent, respectively. The average exit time of the poor households based on the five-year average per capita Gross Domestic Product growth rate was estimated. Accordingly, the average exit time poor households from poverty was estimated to be 3.35 years. Tobit model result showed that household size was significantly and positively influence poverty whereas number of livestock and oxen ownership, educational level of the household head, input utilization, asset ownership and credit utilization negatively influenced poverty in the study area. The result suggests that improving adult education, provision of input for smallholder farmers, improving access and availability of credit, improving the livestock sector will be important policy interventions.

Keywords: Rural Poverty, Banja, Determinant, Tobit

1. INTRODDUCTION

This chapter presents the background, the problem statement, the objectives, the significance of the study, the scope, limitations and organizational structure of the thesis work.

1.1. Background of the Study

Poverty is one of the most serious problems of human deprivation and a complex phenomenon. It is a multi-dimensional concept which encompasses inadequate income and short of the necessities such as education, health services, clean water and hygiene which are crucial elements for human dignity and survival. Therefore, dealing with poverty is a priority development concern in many developing countries in general and sub-Saharan countries in particular (World Bank, 2007).

Poverty is a threat to the world, especially in developing countries. The governments, national and international development institutions have tried to understand the nature of poverty and mechanisms of reducing it. Development economists argued that the fight against poverty is a necessary condition for any economic growth thereby achieve the wellbeing of citizens.

Although the proportion of households living in poverty and extreme poverty in developing countries have been declining over the past three decades, the numbers remain high, with almost one billion people considered to be extremely poor and another one billion are poor. Notably, Extreme poverty has fallen substantially in East Asia and the Pacific as well as in South Asia. However, in sub-Saharan Africa, little progress has been made and almost half the population is extremely poor (FAO, 2015). Hence, the highest regional poverty rate is in Sub-Saharan Africa, where 42.7 percent of the population is estimated to be below the global poverty line, followed by South Asia (18.8 percent) and East Asia (7.2 percent) (World Bank, 2015). That's why any country in these regions consider poverty reduction programs as the prime objective of their national development plan. This was witnessed by world leaders by setting the Millennium Development Goal on Eradicating extreme poverty and hunger by 2015. Besides, it set a foundation for sustainable development goals for 2030. Thus, poverty is an issue of continuing global agenda.

Nearly, three-fourth of the poor in the developing world lives in rural areas, and rural poverty remains high and persistent. Majority of sub-Saharan African population, 63 percent, lives in rural areas where their daily income is less than US$ 1.90 a day. Overall, the rural population is hit much harder by poverty than people living in urban areas (WFP, 2016). Even though rapid economic growth and quick poverty reduction are witnessed in East Asia, sluggish and deteriorating growth in Sub Saharan countries is not quick enough to reduce extreme poverty (IFAD,2011).

Majority of people in Ethiopia, by any standard, are among the poorest in the world (Teshome and Sharma, 2014). It is a common and widespread phenomenon in which a larger proportion of its population lives below one US dollar a day. Based on the Human Development Index (HDI), Ethiopia is classified as a low human development implying that many of its citizens are seriously deprived of basic needs like food, shelter, education and health. The HDI for 2015 was 0.448 which is below the average of sub-Saharan African countries (0.523) and put a country on a rank of 174 out of 187 countries. The figure clearly indicates that Ethiopia is one of the poorest nations in the world (UNDP, 2016).

The two-major poverty/welfare related nationwide government leads major survey studies in Ethiopia are the Welfare Monitoring (WM) and Household Income and Consumption Expenditure (HICE). The Household Income and Consumption Expenditure Survey is the most extensive survey available on the status and extent of poverty in the country which was conducted every five years since 1995/96. Based on this survey result, the incidence of poverty has declined substantially between 1995/96 and 2015/16. The decline was reflected both in rural and urban areas. The percentage changes of decline in rural poverty indices were smaller compared to the urban poverty. This might be due to lack of identifying the poor rural farmers and develop appropriate and area-specific poverty reduction strategies that actually address their problems.

Realizing the worst situation of poverty in the country, the government of Ethiopia has implementing different poverty reduction strategies in order to make the country among one of the middle-income country by 2025. Notably, Sustainable Development and Poverty reduction program, Plan for Accelerated and Sustained Development to End Poverty, Growth and

Transformational Plan I and most recently phase two of GTP have been implemented to reduce the extreme poverty. Following these major national programs and strategies, the country has registered a promising economic growth that helps a significant proportion of people move out of the poverty trap. According to UNDP (2016) report, the Ethiopian GDP per capita for the year 2015 was $1,530 US dollar (at 2011 PPP). The real Gross Domestic Product growth rate for the same year was 8.7 percent.

Despite positive economic growth registered in the past years, a substantial number of people are still suffering from poverty. Poverty seems to persist in large sections of the country both in rural as well as urban, with little hope for a substantial improvement of the living conditions of the rural poor as well as urban poor in the near future. In order to combat such a devastating situation of the human being given limited and scarce resources, it has to be well understood about the magnitude, the extent, and determinants of rural poverty. Once these issues are identified, it may not be a difficult task to provide policy recommendations particularly for those who are suffering from poverty.

This study was conducted in Banja District of Awi zone, where mixed farming system is the main livelihood of the rural farmers. The rural poverty scenario in the study district requires depth analysis in that the situation of rural poverty might be more chronic than other areas in the region. Besides, those who are assumed non-poor now, may inevitably be vulnerable to poverty in the coming years. Therefore, identifying and analyzing the poverty measures and determinant factors of the poverty level in the study area is crucial. Besides, it could help to close the information and knowledge gaps that are hindering in addressing the poor households through implementation of area specific development interventions and strategies. Hence, this study could have paramount importance in the study area in particular and the region in general.

1.2. Statement of the Problem

In much of sub-Saharan Africa countries, agriculture is the mainstay of economic growth, overcoming poverty, and enhancing food security. Of the total population of SSA in 2003, 66 percent lived in rural areas and more than 90 percent of people in these regions depended on agriculture for their livelihoods (Husmann, 2016). Nearly, 70 percent of the population in the

4 world who are below the poverty line are located in Sub Saharan African countries (Ahmed, 2013). Hence, improving rural areas means improving the lives of the most chronically poor people. Thus, the agricultural sector will have paramount importance to lift millions of people out of poverty.

Ethiopia, like the other Sub-Saharan African country, is still suffering from widespread and severe poverty. Since 80 percent of the Ethiopian economy is based on the agricultural sector, the country's national policies and strategies are targeted to reduce poverty through increasing the productivity of this sector. The programs provided credit, agricultural input, access to better extension packages, expansion of rural infrastructural services mainly irrigation canals, rural road construction, health services, telecommunication services, and primary education.

In mid-1990’s, Ethiopia was one of the highest poverty rates in the world, with 55.3 percent of the population living on less than the global poverty line (US$1.25 per day) and 45.5 percent of its population live below the national poverty line. By 2011, 33.5 and 29.6 percent of the populations lived on less than the global and national poverty line respectively (World Bank, 2016). Thus, a country had achieved substantial progress in reducing extreme poverty in the last twenty years.

Even though various poverty reduction policies and strategies have been implemented and, hence rapid economic growth in the past decade was registered, a significant proportion of the population still live in absolute poverty situation. For instance, in the years 2014/15, the Ethiopian economy registered the real gross domestic product (GDP) of 8.7 percent which was better than sub-Saharan countries (NBE, 2015). But this economic growth was not enough to eradicate extreme poverty in the country particularly rural areas.

In rural Ethiopia, poverty is a common phenomenon. Compared to urban-rural poverty rate is high. This captures the attention of researchers and policymakers to develop poverty reduction strategies that actually address the situation of rural poor farmers. Household Income and consumption expenditure survey (HICE) taken at 1995/96, 1999/2000, 2004/05, 2010/11 and 2015/16 showed that the incidence of poverty was 45.5, 44.2, 38.7, 29.6 and 23.5 percent respectively. At the same time, rural poverty was 47.5, 45.4, 39.3, 30.4 and 26.1 percent. Even though one can see the declining trend of rural poverty, the figures are still very high compared

5 to many sub-Saharan countries. Major changes in the political environment, and migration caused by civil conflict, and the increased frequency and severity of drought and erratic climatic shocks have all taken their toll on the country's poor rural households and continue to affect them. The majority of the rural population lives far below the internationally recognized absolute poverty threshold of less than the US $1.90 a day at 2011 PPPs, and most of these people are chronically, or at least periodically, food insecure. As copping mechanism, most households resort to seasonal or permanent migration to urban areas in search of wage employment to feed their families (IFAD, 2011).

Following different reforms and poverty reduction strategies that have been implemented in the country, there could have been a reduction of poverty both at the national and regional levels. For instance, in 2015/16 the national poverty level was 23.5 percent. Across regions, the analysis of HICE survey result of 2015/16, Amhara region experienced headcount index of 26.1 percent which is higher than the national average. Moreover, rural and urban poverty headcount index in the region stood at 28.8 percent and 11.6 percent, respectively in which the former is above the national rural headcount index of 25.6 percent. This implies that rural poverty is a widespread problem in the region in particular and the country as a whole.

Banja district is one of the nine administrative districts in Awi Zone. According to agricultural offices of the district, it is characterized by high population, low agricultural productivity, the high number of migrants to urban centers, poor rural infrastructure and weak institutional support service. Accordingly, the district is identified as one of the poorest districts in Awi Zone. Following this, efforts have been made to improve the livelihood of farmers. By increasing the speed of the growth of the agricultural sector, the living standard of the rural farmers can be improved. However, this is not the reality in Banja district. Based on the information obtained from the agricultural office of Banja district, substantial number of poor farmers are selling or renting their natural asset like land and move out of their village in search of food to his/her family to urban areas at least temporarily. Besides, family dissolving is getting common phenomenon in the study area. Moreover, production and productivity of farm households are getting worse due to the acidic nature of the soil. Thus, rural farmers in Banja district are hardly breaking the vicious circle of poverty.

Therefore, to fight against poverty in rural areas of the study district and reverse the situation at the minimum possible level, it requires depth understanding and need to design pro-poor and location or district-specific poverty reduction policies and strategies. Hence, critical assessment of rural poverty in the district helps to identify who the poor are and why they are being poor and getting poor, what contributes them to be poor and what poverty reduction intervention or strategy might be appropriate.

Moreover, poverty analysis carried out elsewhere might not be applicable in the other areas due to differences in socio-cultural, economic, geographic zones and differences in livelihood strategies. Thus, critical analysis of rural poverty at the district level is important.

To the best of the researcher, studies with special focus on poverty analysis and factors affecting rural poverty are not yet studied in Banja district. Hence, this study is assumed to fill the existing knowledge gap concerning the magnitude, extent and scenario of poverty and to suggest appropriate policy intervention options aimed at reducing and eradicating rural poverty.

1.3. Research Questions

The underlining research questions of this study were: 1. What is the magnitude, extent and status of poverty in rural part of Banja district? 2. How long would it take for the poor rural household to exit poverty if GDP per capita of the country grows at a positive rate per year? 3. What are the major determinant factors of rural poverty in the study area?

1.4. Objectives of the Study

The general objective of the study was to assess rural poverty situation in Banja District of Amhara National Regional State. The study had the following specific objectives;

1. To examine the extent, depth and severity of household’s poverty in rural areas of Banja district; 2. To estimate the average time required to exit the poverty of rural poor households, and 3. To analyze the determinants of rural household poverty in the study area

1.5. Significance of the Study

Studies with special focus on the magnitude of poverty, the average time required to exit poverty and factors affecting rural poverty are not yet studied in the rural area of Banja district. Hence, this study is deemed to fill the existing knowledge gap concerning the extent of rural poverty, the average exit time of poverty and its determinants in the district in particular. In addition, the outcome of this study helps to provide possible poverty reduction interventions for policymakers. Besides, it may be used as an important input for researchers and development practitioners as well as a good startup for further study in the district. Therefore, the outcome of this study would contribute towards the extent of poverty problems in the society and underlying features so as to help as base information in reducing poverty in the study area as well as areas with similar characteristics.

1.6. Scope and Limitations of the Study

The study was conducted in only one district, Banja, Amhara National Regional State. From the twenty-five, rural kebeles found in the district five rural kebeles total sample of 190 households heads were randomly selected. The study was limited to the use of cross-sectional data that was conducted for the study purpose. The unit of analysis in this study was household head. It is unquestionable that such study does not capture the dynamic and multidimensional nature of rural poverty. Hence, this study is limited only to those basic necessities of life which are being partially met economic issues by targeting rural households. Moreover, from such one-shot study cross-sectional data, one can hardly generalize the findings to other districts without considering the socioeconomic, demographic and agro-ecological characteristics. Besides, estimating the average exit time of the poor households based on national GDP per capita growth rate due to unavailability of regional GDP per capita that might not explain a specific district is one of the serious limitations of this study.

1.7. Organization of the Thesis

This thesis contains five chapters. Chapter two presents the concepts, definitions, and theories of poverty. Besides, empirical, theoretical and analytical poverty measures were briefly

8 discussed. Chapter three deals with brief description of the study area and the research methodology for sampling, data collection, and analysis. Chapter four presents the results and discussion of the the research. The last chapter but not the least presents the summery, and conclusion and policy recommendations based on the research finding.

2. LITRATURE REVIEW

This chapter presents the different concepts, theories and perspectives of poverty. Besides, theoretical, conceptual and analytical framework of poverty were presented. In each section detailed review were made.

2.1. Concepts and Definition of Poverty

The commonly referred and comprehensive definition of poverty by the world Bank (2001) characterizes it as: “…a pronounced deprivation of wellbeing related to lack of material income or consumption, low levels of education and health, vulnerability and exposure to risk, voicelessness and powerlessness”. From this definition, one can understand that poverty is a multi-dimensional phenomenon. One of the dimensions is the material deprivation, lack of access to goods and services, which is measured in terms of income or consumption as indicators. The second dimension refers to low capabilities as manifested by low level of educational achievement and poor nutritional and health conditions. Vulnerability and exposure to risk, and voicelessness, and powerlessness are considered, respectively as the third and fourth dimensions of being poor. These, the four, dimensions of poverty are interrelated and reinforce each other.

Poverty can be defined as lack of household income or consumption that results from many interlinking factors found in the poor people’s experience. It encompasses not only low monetary income and consumption but also low human development, such as education, health and nutrition. More generally, poverty means the inability to meet basic needs, including food, shelter, clothing, water and sanitation, education, and healthcare. In this sense, poverty generally reflects a combination of income poverty at the household level and poverty at the community level in the provision of basic infrastructure and public services (SDSN, 2012).

The concept of poverty, in simple terms, describes a situation of whether or not individuals or households possess enough resources or capabilities to meet their current needs for a living. The poor are, hence, underprivileged segments of society who do not have adequate food, shelter and access to education, health and other services (Dawit et al., 2011). It is thus a household or a group of society lacking to achieve reasonably minimum standard of living.

Poverty is also a social phenomenon which goes further than economic spheres and encompasses inability of individuals to participate in social life and political environment. One way of defining poverty is by letting the poor to explain their own poverty. It is allowing individuals or groups who are practically facing poverty to define what represents their basic requirements in life (Ahmed, 2013).

2.2. Theories of Poverty

Once we understood the definition of poverty, the next step is to look deep into different theories of poverty which give us a comprehensive explanation of why people are poor. Since poverty is a multidimensional concept, understanding the root causes of it is important. The following poverty theories summarizes how and in what context poverty exist and arise. The major ones are individualistic, cultural, geographical and structural. Hence, the following sub-sections give a comprehensive review in this regard.

2.2.1. Individualistic Theory of Poverty

Individualistic theory of poverty perceives the poor as if they are born with it (i.e. born being disabled like crippled, blind, or deformed) and for that reason they cannot do anything to change the situation in which they are living (Rainwater, 1970). Furthermore, the individualistic theory perceives that poverty is resulted due to acquired personality traits like character and actions of individuals. The idea here is that some individuals who are born being lazy do not voluntarily participate in tasks that have meaningful effect in their life. Put differently it is an intrinsic to an individual character and it consists of personal ability and intelligence of the person.

2.2.2. Cultural Theory of Poverty

This theory suggests that poverty is created by the transmission over generations of a set of beliefs, values, and skills that are socially generated but individually held. Individuals are not necessarily to blame because they are victims of their dysfunctional subculture or culture. Technically, the culture of poverty is a subculture of poor people in ghettos, poor regions, or social contexts where they develop a shared set of beliefs, values and norms for behavior that are separate from but embedded in the culture of the main society resources (Bradshaw, 2007).

The culture of poverty is a syndrome that develops in some specific situation. It occurs in an economic setting with low wages, high rate of unemployment, and people with low skills. In the absence of deliberate support from the government, the low-income population have a tendency to build up the culture of poverty against the prevailing ideology of expanding the middle class. The poor, who consider themselves negligible in a society, create survival strategy by developing their own subculture and institutions, and finally come to embody a common pattern of behavior, norms and values. The subculture developed by the poor is characterized by pervasive feelings of dependency, helplessness, marginality, and powerlessness (Lewis, 1959 as cited by Ryan, 1976).

2.2.3. Geographic Theory of Poverty

Geographic theory of poverty corresponds to spatial characterization of poverty. Rural poverty, ghetto poverty, urban disinvestment, Southern poverty, third-world poverty, and other forms of the problem represent a spatial characterization of poverty that exists separate from other theories. While these geographically based theories of poverty build on the other theories, this theory calls attention to the fact that people, institutions, and cultures in certain areas lack the objective resources needed to generate well-being and income, and that they lack the power to claim redistribution (Weber and Jansen, 2004).

The geographic theory of poverty traces the flows of capital as well as knowledge. For example, rural areas are most of the time the last stop of technologies, and less competitive pricing and low wages dominate production (Ahemed, 2013). Limited infrastructural development inhibits the rural poor economic performance and let them behind the largest competition.

2.2.4. Structural Theory of Poverty

Finally, the structural theory is a progressive social theory. This theory advocates that elimination of structural barriers and implementing a wide range of socioeconomic policies generates substantial numbers of successes in reducing poverty. The range of socioeconomic policies that can be adjusted to realize poverty reduction include raising wages, providing jobs, assuring effective access to medical care, expanding the safety net, and coordinating social insurance programs (Leonared, 1988).

2.3. Theoretical Framework

Based on the theories, it is possible to disscuss the various factors that affect the likelihood of households to experience poverty. Poverty may arise due to household or individual level characteristics. It may also arise due to factors that are external to the household, i.e. due to community level, and/or regional level characteristics. It is possible to disaggregate household level characteristics in to two broad categories as demographic and socioeconomic characteristics. Indicators of demographic characteristics that may be associated with poverty are household size and structure, dependency ratio, the age of the household head and the gender of household head. The most familiar socioeconomic characteristics that explain poverty are household asset and household employment. Typically, in rural areas, the cropping system of the household can affect the income obtained from farming activities. Cash crop farmers may generate higher income and, therefore, be less poor than food crop farmers irrespective of the amount of inputs and the size of the cultivated land.

Moreover, community level characteristics might be related with poverty for certain neighborhood. Infrastructure is the core determinant at this level. It includes access to electricity, proximity to paved roads, access to market, access to schools and health care service centers. In addition, inadequate social service provision, social exclusion and discrimination are associated with poverty (Grant and Marcus, 2009). Besides, poverty might be associated with several features like remoteness. Generally, poverty is higher in areas characterized by low resource base, geographical isolation, rainfall deficit, and other harsh climatic conditions (World Bank, 2005).

The three, major, school of thought in literature concerning the definition and measurement of poverty are; the welfare school, the basic needs school and the capability school (Garza, 2000). Even if each of the thoughts perceive poverty differently, there are areas in which they share some common meaning and all of them judge a person to be poor whenever he/she is lacking a reasonable minimum standard of living.

The Welfare Approach: It defines poverty to the economic well-being of the society. It assumes that when; societies are not able to attain a level of economic well-being deemed to constitute a minimum by the standard of that society, then the person/society faces poverty. It

13 sees income as a determining factor for the presence of poverty. It bases composition of wellbeing solely on individual utilities, which are based on social preferences (Ravallion, 1992 cited in Esubalew, 2006). Put differently, economic actors are rational and that they behave in ways to maximize their utility.

The Capability Approach: The basic building blocks of the capability approach defined as a broad normative framework for the evaluation and assessment of individual well-being and social arrangements, the design of policies, and proposals about social change in society. It is used in a wide range of fields, most prominently in development studies, welfare economics (Alkire, 2002). The core characteristic of the capability approach is its focus on what people are effectively able to do and to be; that is, on their capabilities. This school emphasizes on neither the economic well-being nor the basic needs deemed to satisfy the minimum standard by the society; instead on human abilities or capabilities to achieve a set of functioning. This is an alternative criterion for the definition and measurement of well-being which tells the extent to which people have capabilities to do things of intrinsic worth. Such approach to the definition and /or measurement of poverty suggests a broader set of criteria for assessing poverty than just income and/or consumption. The capability approach includes publicly provided but non- marketed services; like road, sanitation, health care, education and life expectancy (Sallila and Hilamo, 2004).

The Basic Needs Approach: The basic needs approach defines poverty when one lacks basic needs either goods or services to sustain his life. It concentrates on the degree of fulfillment of basic human needs in terms of nutrition/ food, health, shelter, education, transport and so on. A person is said to be poor if he or she is unable to meet his or her basic needs. This definition of poverty can apply to a given society or district when it is unable to meet the basic requirements deemed by them.

Garza (2000) argued that the definition and measurement of poverty are the major areas of limitations of the basic needs approach as the set of basic goods and services are different for different individuals and society depending on social, cultural, political and economic status of an individual. Determining the set of basic needs for an individual is challenging task. Moreover, there is even a high disagreement among professionals on the determination of set the basic needs.

Generally, there is no universal consensus among scholars in defining poverty as it has various interpretations within economic, social, political, institutional, environmental, and cultural contexts. The sociologist conceives poverty as a reflection of social inequality. It is explained as the absence of access to enjoy fundamental human rights. This could include lack of access in social participations such as social class and social group. To environmentalists, poverty is conceived as a situation in which one has access to environmentally fragile natural resources that reduce income and utility. Biologists conceptualizes poverty as the lack of entities for survival. They postulate that poverty exists when the necessary minimum requirements for physical efficiency are not fulfilled (Sen, 1981).

Since the focus of this is to exmine the status and extent of poverty in Banja district, it is more appropriate to use the basic needs approach. This approach usually considers on the minimum requirments of basic needs of a household. Hence, rural households in Banja district are struggling to fullfill their basic needs they are more concerned for the present needs. Thus, the the threshold line that demarcates the poor from the non-poor households were determined by using the basic needs approach.

2.4. Determination and Types of Poverty Line

Poverty line indicates deprivation in absolute sense, and it refers to the minimum level of income or consumption expenditure assumed to fulfill the minimum requirements of life or wellbeing. Hence, those households income or expenditure is below the poverty line are considered as poor and those housholds income or expenditure above the poverty line are considered as non-poor. However, setting an arbitrary poverty line in order to distinguish the households or individuals in to poor and non-poor categories is central question in any poverty analysis. In the following sections poverty line, types and methods of poverty line determination were discussed.

2.4.1. Poverty Lines and Types

The term “Poverty line” is probably familiar to most economists, and indeed much of the population, however it may not necessarily be fully understood. It is the value of income or consumption expenditure necessary for fullfiling the minimum requirments to sustain life (Sian, 2014). It can

15 also be as defined, based upon a minimum level of consumption, where the cost of a bundle of goods (both food and non-food) required to assure that basic consumption needs are met and below which survival is threatened. Put differently, poverty line for a household can be defined as the minimum spending or consumption (income, or other measure) needed to achieve the minimum level given the level of prices and the demographic characteristics of the household. In its simplest term, poverty line is a demarcation line that separates the poor from their counter non-poor groups. Thus, construction of poverty line is inevitably a precondition to go further in poverty analysis.

2.4.1.1. Absolute poverty line

Absolute poverty is the inability to secure the minimum basic needs for human survival. It is the worst kind of poverty in which people do not have access to basic necessities to fulfill their basic physical needs, and therefore are undernourished, weak and susceptible to diseases (Hussain, 2003). The concept of absolute poverty is preferable for doing comparative analysis in any poverty research. While employing an absolute measure of poverty, the first step is to establish an absolute poverty line, where a poverty line is used to differentiate the poor from the non-poor groups. The poverty line should represent the same standard of living across groups or regions within the country and should be adjusted for spatial differences in the cost of living. This means, the poverty line is set so that it represents the same purchasing power year after year, but this fixed line may differ from country to country or region to region.

2.4.1.2. Relative poverty line

Relative poverty in loose terms is a state of having less than others. It defines "poverty" as being below some relative poverty threshold point (Sallila et al., 2004 and Morduch, 1998). Relative poverty measures define the segment of the population that is poor in comparison with the consumption expenditure or income of the general population. It defines how income and inequality is distributed in a society. Put differently, it observes poverty as a function of relative deprivation in terms of commodities, hence defining poor households who are unable to attain a given commodity that are assumed to be normal for their society. The statement itself is self- intuitive in that this poverty is defined by the position of an individual compared to other members of a given society. The relatively poor, therefore, are those whose incomes are lower

16 compared to that of the rest of the community even if they are in a position to secure an adequate level for survival.

Relative poverty is, therefore, the share of people whose income or consumption expenditure falls below a poverty line. In practice, the most popular choice to set poverty line in this method is done by taking certain percentage of mean or median incomes of the population. It specifies the poverty threshold as a “cut-off” point in the distribution of income or expenditure and hence it can be updated automatically for changes in living standards (Foster, 1998 and Muller, 2006). Consequently, relative poverty line varies across countries usually between 50 to 60 percent of the national mean income.

2.4.2. Setting Poverty Lines

The poor are those whose expenditure or income falls below a poverty line. The starting point is, therefore, the identification of the poor from the non-poor. To deal with this, poverty line plays a vital role in quantifying the various indicators of wellbeing into a single index (Ravallion, 1992). Thus, constructing relevant and appropriate poverty line has a paramount importance. However, setting poverty line is a difficult task in measuring poverty. The common argument used to set the line is a minimum level of consumption of goods and services below which it is difficult to sustain life (World Bank, 2000). The most popular and widely used methods in constructing poverty line are the Cost of Basic Needs method, Food Energy Intake method and Direct Calorie Intake method.

2.4.2.1. Cost of Basic Needs Approach (CBN)

The definition of basic needs is believed to be a socially determined minimum requirement of income or consumption expenditure for which basic needs will be met. Thus, the cost of basic needs approach is the most widely used to set the poverty line. It proceeds by first estimating food expenditure necessary to attain some recommended food energy intake (2,200kcl) which reflects the minimum nutritional threshold needed for a healthy life (EHNRI, 2000). This expenditure level can be considered as food poverty line. Next adjustments are then made for non-food expenses like housing, clothing and social values and applicable to arrive at total poverty line (Ravallion, 1994).

In order to construct poverty line using this approach, first estimates the cost of acquiring enough food for adequate nutrition usually 2,200 kilo calories per person per day and secondly add the cost of other essentials such as clothing, shelter, transport, medical services, social and religious expenses, etc. In other words, first determining the food consumption bundle “Food basket” just adequate to meet the required food energy requirements; and second, adding an allowance for non-food basic needs to this cost. The food consumed is then valued at the prevailing local or regional prices if the objective is to arrive at a consistent food poverty line across regions and groups (Ravallion and Bidani, 1994). The food poverty line obtained is translated and incorporate to the expenditure required to attain non-food basic needs

However, it is not free of problems. The first problem is the minimum required nutrition level varies across regions or ethnic groups. In other words, estimating food components is influenced by different socioeconomic and agro-climatic regions. The second problem is it is hard to know the utility level of non-food components. Thus, it is difficult to estimate the non- food basic components. Despite the limitations, the cost of basic need approach shall be used as it is the most widely used method of poverty line determination provided that the threshold line ensures the basic requirements’ of a household (WBI, 2005). In practice, two methods have been commonly used to derive the total poverty line, the ‘food energy intake’ and ‘food share’ methods (Ravillion, 1994). Both approaches are based on the minimum energy requirement for a typical person to keep or sustain normal daily activities. Alternatively, World Bank developed a simple linear regression to obtain non-food poverty line. Accordingly, the non-food poverty line can be estimated as the share of the food expenditure to total expenditure of each household in adult equivalent on a constant and the log of the ratio of total expenditure to food poverty line (Ravallion, 1992). 푇퐸 푆 = 훼 + 훽푙표푔 ( ) + 푒 (1) 푖 퐹푃퐿 푖

퐹퐸 Where 푆 = , Share of per adult food expenditure to total expenditure 푖 푇퐸

푇퐸 , Total expenditure

퐹푃퐿 , Food poverty line

퐹퐸, Food expenditure

훼 푎푛푑 훽 , Food share and slope respectively

𝑖 , sampled respondent households

퐹푃퐿 퐹푃퐿(1−훼) Therefore, and and the total poverty line and non-food poverty line respectively 훼 훼 Therefore, the total poverty line can be obtained by the arithmetic sum of food poverty line and non-food poverty lines.

Mathematically,

푃퐿 = 푃퐹퐿 + 푃푁퐿 (2)

Where; PL is the poverty line, 푃퐹퐿 is food poverty line and 푃푁퐿 is non-food poverty line

2.4.2.2. Direct Calorie Intake (DCI)

Direct calorie intake method defined poverty line as the minimum calorie requirements for the individuals to survive, and those who consume below a predetermined minimum level of calorie intake are considered to be poor. Inother words, direct calorie intake method simply measures poverty with malnutrition. However, this method does not take into account the non-food basic need requirements that are essential for survival and it does not give costs of acquiring the minimum calorie requirement. Therefore, if the interest is to measure poverty as lack of command on basic needs, it is unlikely to reveal the extent of deprivation for a given society (Tassew et al., 2008).

2.4.2.3. Food Energy Intake Approach (FEI)

This approach locates the poverty line as the income or consumption expenditure level at which a person’s typical food energy intake is just sufficient or adequate to meet a predetermined food energy requirement for healthy life and normal activities (Ravallion and Bidani: 1994). It is an improvement over the direct calorie intake regarding to the representativeness of the poverty line since it provides the monetary value rather than a purely nutritional concept of poverty. It is simply determined by regressing the per capita consumption expenditure on calorie intake or income and the expressed value of the per capita consumption expenditure at the predetermined

19 calorie intake level is taken as the poverty line. The poverty line becomes that level of total expenditure at which the minimum energy requirement is met (Greer and Thorbecke, 1986).

However, in the case of FEI approach, the necessary point is to know that either an individual meets the minimum level of calorie intake or not whatever the types of food baskets are consumed. Hence, it is affected by individual’s preference, regional variations, activity level, age, sex, consumption habit and relative price. Consequently, applying food energy intake method to construct poverty line in different regions and over the periods even within the country yields inconsistent threshold (poverty line) and it does not provide robust poverty line.

Accordingly employing the cost of basic needs approach for this particular study is appropriate and easier to measure to food and non-food poverty line. This is to mean that consumption is a better indicator of a household. It actually reflects the ability of a particular household to meet his or her basic needs. Besides, expenditure on basic needs can easily be recalled by a household to measure his or her consumption. In general, cost of basic needs approach in measuring poverty is advantageous in that it ensures consistency (treating individuals with the same living standards equally) and adjustments for spatial and inter-temporal variations could be made to set the poverty line most commonly used method. Moreover, price level at the district or regional level can be easily available.

2.5. Measurement and Indices of Poverty Analysis

Poverty reports in developing countries use all the three poverty indices; namely headcount poverty index, the poverty gap index, and the severity of poverty index. Generally speaking, poverty analysis follows the three common procedures; choosing the appropriate welfare measure, constructing poverty line, and computing poverty indices using available information to report for the whole population. There is no single measure of poverty and all the choices have their own pros and cons. The choice of which index to use in poverty analysis depends on the purpose of measurement and even the availability of data (WBI, 2005). The three poverty indices generally called the Foster Greer and Thorbeck (FGT) class of poverty measures developed by Foster et al. (1984).

2.5.1. Head Count Index

Head count index is the most widely used measure of poverty analysis; defined as the proportion of the population whose measured standard of living (income or consumption) is below the poverty line. Hence, it tells us the proportion of poor people in a given population.

Formally,

n H = (3) N Where, H represents head count ratio, n represents the number of people considered as poor, and N is the total number of the sample population.

It can also be rewritten as follows:

n 1 푃 = ∑ 퐼(푥 < z) (4) 0 N 𝑖 i=1

Where, 푃0 represents measure of head count ratio, I is an indicator function that takes a value

th of 1 if the expression 푥𝑖 < 푧 holds true, 0 otherwise. 푥i , is the i poor household’s consumption expenditure or income and 푧 is the poverty line.

Despite its simplicity to construct and interpret the values, poverty head count ratio failed to address some of the important poverty measures like how intense the poverty is and also failed to indicate how the poor are poorer. Thus, it has less value for policy implication.

2.5.2. Poverty Gap Index

The poverty gap index indicates the depth of poverty which is the difference between the poverty line and the mean income or consumption of the poor households expressed as a percentage of the poverty line. Put differently, the poverty gap measure captures the mean aggregate income or consumption shortfall relative to the poverty line across the whole population. It is obtained by adding up all the shortfalls of the poor (assuming that the non-poor households have a shortfall of zero) and dividing the total by the population. Since this index

21 is based on the aggregate poverty deficit of the poor relative to the poverty line, it is by far better than the head count index and is known as modest measure of poverty analysis.

More specifically, poverty gap (Gi ) is the poverty line (z) less actual consumption expenditure

(푥푖) for poor individuals; the gap is considered to be zero for everyone else (non-poor). Using the index, we have

퐺푖 = (푧 − 푥푖). 퐼(푥푖 < 푧)

More formally, the poverty gap index (푃1) can be computed as:

푛 1 퐺𝑖 푃 = ∑ ( ) (5) 1 푁 푧 𝑖=1

Even though poverty gap index can be used as an indicator of potential for eliminating poverty by targeting the poor households, it is insensitive to the distribution of income among the poor (Kimalu et al., 2002).This calls the third measure of poverty.

2.5.3. Poverty Severity Index

Poverty severity index measures not only the distance separating the poor from the poverty line (the poverty gap), but also captures the inequality among the poor. Higher weight is placed on those households further away from the poverty line. This index measures the severity of poverty by squaring and averaging the gap between the income or consumption expenditure of the poor households and the poverty line. Unlike the poverty gap index, this measure reflects the severity of poverty in that it is sensitive to inequality among the poor (Fitsum, 2002, WBI, 2005, and Tassew et al., 2008).

Poverty severity (푃2) is can be computed using the following formula: 푛 1 퐺 2 푃 = ∑ ( 푖) (6) 2 푁 푧 푖=1

Where, Gi = z − 푥푖 (with Gi = 0 when 푥푖 ≥ z), 푥푖is consumption expenditure in the 𝑖푡ℎ poor household, z is the poverty line, n is the number of poor households, and N is the size of the sample population.

Generally, the measures of depth and severity of poverty are important complements of the incidence of poverty. It might be the case that some groups have a high poverty incidence but low poverty gap (when numerous members are just below the poverty line), while other groups have a low poverty incidence but a high poverty gap for those who are poor (when relatively few members are below the poverty line but with extremely low levels of consumption or income).

2.6. Time needed to Exit Poverty

This concept was first developed by Kanbur (1987). He noticed that knowing the average time of the poor households to exit poverty is useful in designing appropriate poverty reduction policies and strategies. Thus, the ultimate goal of any developing country is to reduce and exit poverty within a given period of time. Accordingly, it is very useful to estimate the poor households poverty exit time given positive economic growth rate is assumed and enjoyed by the poor. Morduch (1998) employed this measure of poverty analysis in Bangladesh and Bolivia and concluded that this measure is distributionally sensitive, additively decomposable, and cardinally meaningful and satisfies standard of poverty axioms. Therefore, the statistics is decomposable by population sub-groups and is also sensitive to how consumption expenditure or income is distributed among the poor. For the ith person below the poverty line, the expected time to exit poverty (i.e. to reach poverty line), if consumption per capita grows at positive rate 푔 per year;

ln(푍)−ln (푥푗) 푊 푡푖 ≈ = (7) 𝑔 𝑔 𝑔

Where; W is watts index 푔 is consumption per capita growth rate

푥푖 is the expenditure of the person below poverty line 푡푖 is the expected poverty exit time for the ith poor individual

In other words, the time taken to exit poverty is the same as the Watts index divided by the expected growth rate of consumption per capita growth rate.

Watts index: the first distribution sensitive poverty measure was proposed by Watts in 1968 (Zheng,1993). It can be put in its discrete version in the form of: 풏 ퟏ 푾 = ∑[풍풏(풛) − 풍풏(풚풊)] (8) 푵 풊=ퟏ

Where N individuals in the population are indexed in ascending order of income (consumption expenditure) and the sum is taken over in individuals whose income or consumption expenditure falls below the poverty line. Ravallion and Chen (2011) argue that three axioms are essential to any good measure of poverty. Under the focus axiom the measure should not vary if the income of the non-poor varies. Under the monotonicity axiom, any income gain for the poor should reduce poverty; and under the transfer axiom, inequality reducing transfers among the poor should reduce poverty. Hence, the Watts index satisfies these thee axioms, but the poverty head count and severity measures do not.

Research conducted in rural Bangladesh using Household expenditure survey show that if economic growth rate is 3 percent, it takes 4.9 years for the average poor to exit poverty provided the economic growth rate is assumed continuous (Morduch, 1998). A study conducted by Ali et al. (2010) to estimate regional trends of average time to exit poverty in Pakistan by comparing inter temporal and inter provincial comparisons using six-year GDP growth on average by 4 percent showed that the average poor households to exit poverty requires 6.25 years.

2.7. Review of Potential Analytical Models

Poverty determinants can be analyzed in a number of econometric models. However, in the literature binary choice models, ordered models, multivariate regression models and censored regression models are most widely used methods.

2.7.1. Binary Choice Models

In these model it is assumed that the actual per-capita income or consumption of households is not observed. We act as if we only know whether the household is poor or not, which takes the value of one if the household is poor and zero if the household is non-poor. Gujarati (1995) and Maddala (1993) indicated that the logit and probit models are the only binary dependent models used to establish a relationship between household demographic and livelihood capitals with a binary response variable. Even though logit and probit models produce similar parameter estimates, a binary logit regression model is the appropriate and preferred probability model recommended mostly from mathematical point of view (Feder et al., 1985).

Majority of poverty studies in rural Ethiopia used Binary regression models to determine its determinant factors. For example, Ayalneh et al. (2005) conducted a study on the determinants of poverty in three selected rural districts of Ethiopia. They employed a binary logistic regression model and found that land holding, oxen ownership, per capita income and educational level significantly affect poverty.

Indris (2012) conducted study on assessment of food insecurity, its determinants and coping mechanisms among pastoral households in Chifra district of Afar Regional state, Ethiopia. Logit regression model was applied and found that family size, dependency ration, age of the household head affect food insecurity positively where as non-farm income affects food insecurity negatively.

Nega (2015) employed logistic regression model to find out determinants factors of poverty in rural households of Gulomekeda district of Tigray region. Accordingly, total family size and dependency ratio had positive associated with poverty level of household whereas farm size, total livestock unit, value asset, educational status, credit and access to off farm income negatively affect poverty status of the household.

However, the binary regression model is not free of critiques. The main problem is that the binary estimates are sensitive to specification error, the parameters will be biased if the underlying distribution is not normal. Besides, it does not make use of all information available because it collapses income or expenditure in to binary variable (WB, 2002). In other words, the logit/probit model involve unnecessary loss of information in transforming household expenditure in to binary variable (being poor or non-poor). It does not capture how far the observation is below or above the poverty line or cut-off point. Hence, construction of artificial dependent variable in which information about the actual relationship between the level of consumption and the dependent variable is lost (Simler et al., 2004 and Fageras et al., 2007).

2.7.2. Ordered Models

If the objective is identifying different population groups in several stages, order model can be more applicable. In the first stage identifying the population in to poor and non-poor groups. In the second stage, examining the probability of being in extreme poverty condition (identified as extremely poor groups). Following the process of identifying poor and non-poor, the focus will be on those who are extremely poor versus moderately poor and non-poor.

This approach is justifiable, because one can make the ordering of the population sub-samples, using extreme, absolute and moderate poverty lines as cut-off or threshold points in a cumulative distribution of expenditure (Maddala, 1993 and Sen, 1985). The model is built around a latent regression (Greene, 2003).

P∗ = β풳 + e (9)

Where 푃∗ is unobserved. The observed one as follows: ∗ 푃푖 = 0 𝑖푓 P ≤ 0 ∗ 푃푖 = 1 𝑖푓1 < P ≤ µ1 ∗ 푃푖 = 2 𝑖푓 µ1 < P ≤ µ2

Where 0,1,2 are ordered extremely poor, poor and non-poor respectively, whereas µ is known to be estimated with β.

However, ordered model is not free of shortcomings. This model is highly dependent on poverty lines. Hence, its dependability on exact level of poverty lines derivation will make sensitive results of poverty analysis when poverty line changes (Appleton, 2004).

2.7.3. Multivariate Regression Models

Linear regression model is also an alternative econometric model used to analysis the determinants of poverty. The poverty determinants can be assessed by regressing per adult equivalence consumption expenditure (as a proxy for households’ welfare indicator) against a series of independent variables, i.e., variables that affect household consumption expenditure exogenously (Baker, 2000).

Mathematically,

푙표푔 퐶푗 = 훽풳푗 + 푒푗 (10)

th Where 퐶푗 is per capita consumption of the i individual, 풳푖 is a set of explanatory variables.

However, information obtained from the regression of consumption expenditure on the explanatory variables might yield misguided policy recommendation if the poor and non-poor groups of the population have different behavioral patterns. Besides, some of the hypothesized determinants of welfare might have different returns for the poor and non-poor groups.

2.7.4. Censored Tobit Regression Model

Censored Tobit regression model has been extensively used for poverty analysis to measure the effect of changes in the explanatory variable on the probability of being poor and the depth or intensity of poverty. The Tobit model was originated from the study of James Tobin in 1958. The poor households are represented by poverty depth and non-poor households have zero value as the dependent variable. The Tobit model presumes that the dependent variable in the model is censored (has upper or lower limit) and the variables takes on these limiting values for a good number of the respondents, with the other respondents having a wide range of values beyond (below) this boundary. The Tobit model measures not only the probability of an individual is poor but also the intensity of poverty (Tobin, 1958).

The model expressed based on the Tobin (1958):

∗ 푃𝑖 = 훽풳𝑖 + 푒𝑖 (11)

∗ 푃푖 = 0 𝑖푓 푃푖 ≤ 0

∗ ∗ 푃푖 = 푃푖 𝑖푓 푃푖 > 0, 𝑖 = 1, 2, 3, … 푛 ∗ Where 푃푖 is the limited dependent (latent) variable. It is observable when households are poor and unobservable when they are non-poor.

푍−푌 푃 is the poverty depth defined as 푖 for 푌 < 푍 and 푃 = 0 zero for 푌 ≥ 푍. 푖 푍 푖 푖 푖

th 푍 is the poverty line and 푌푖 is the mean annual consumption expenditure of the i household.

풳푖 is explanatory variables, β is a vector coefficient and 푒 is error term.

Wooldridge (2002) recommended both the marginal effects on the latent dependent variable ∗ (푃푖 ) and the expected value for uncensored observations have to be reported. In the first case, the reported Tobit coefficients indicate how a one unit change in an independent variable alters the latent dependent variable. In the second case, the reported Tobit coefficients indicate how a one unit change in an independent variable affects uncensored observation.

2.8. Empirical Review of Poverty in Ethiopia

In mid-1990’s the country experienced one of the highest-level of extreme poverty recorded internationally. In 2000 the rate of the extreme poverty was 55 percent. Since then poverty has declined to 33 percent in 2011 at international standard level. At the same time the national poverty rate in 1996 was 45.5 percent compared to 23.6 percent in 2015. In both cases level of poverty has been declined substantially (Xinshen, 2010, World Bank, 2016).

Based on the 2015/16 HICE survey result, the poverty head count index, which measures the proportion of population below the national poverty line was estimated to be 23.5 percent with marked differences between urban (14.8 percent) and rural (25.6 percent) areas of the country. The poverty gap index that measures the average poverty gap in the population as a proportion of the poverty line is also estimated to be 6.7 percent. At the same time the rural poverty gap was 7.4 percent which was more than twice the urban poverty gap index of 3.6 percent. Moreover, the national poverty severity index is found to be 2.8 percent with rural poverty

28 severity index of 3.1 percent being considerably higher than that of urban areas 1.4 percent (National Planning Commission, 2017).

The headcount poverty rate fell in both rural and urban areas in the past decades. Rural poverty has been declined from 47.5 percent in 1995/96 to 25.6 percent in 2015/16. Over the same period, urban poverty has also declined from 33.2 percent to 14.8 percent. The figures showed that significant improvement in rural poverty has registered than urban poverty in the past ten years.

Although the declining trend of poverty exists, it is still predominantly rural phenomenon compared to urban areas. According to national interim report of national planning commission (2017), the poverty head count index of rural poverty is nearly two times higher than urban poverty in 2015/16. Besides, the poverty gap between rural and urban areas had been narrowing until 2004/05, this gap started diverging after 2004/05 and widened in 2015/16, with 3.7 percent for urban versus 7.4 percent for rural areas.

Table 1: Trends of national poverty (1995/96—2015/16)

Poverty indices 1995/96 1999/00 2004/05 2010/11 2015/16 2015/16 over 2010/11 National Head count 45.5 44.2 38.7 29.6 23.5 -20.5 Poverty gap 12.9 11.9 8.3 7.8 6.7 -13.9 Poverty severity 5.1 4.5 2.7 3.1 2.8 -10.8 Rural Head count 47.5 45.4 39.3 30.4 25.6 -15.9 Poverty gap 13.4 12.2 8.5 8 7.4 -7.1 Poverty severity 5.3 4.6 2.7 3.2 3.1 -3.4 Urban Head count 33.2 36.9 35.1 25.7 14.8 -42.3 Poverty gap 9.9 10.1 7.7 6.9 3.7 -46.6 Poverty severity 4.1 3.9 2.6 2.7 1.4 -48.4 Source: National planning commission (2017)

2.8.1. Status of Regional Poverty

Based on the analysis the results of the 2015/16 HICE survey data regions like Tigray, Benshangul Gumuz and Amhara regions experienced highest poverty incidence with 27 percent, 26.5 percent and 26.1 percent respectively while Harari region, Dire Dawa and Addis Ababa city administration registered lowest poverty incidence of 7 percent, 15.4 percent and 16.8 percent respectively. Tabel 2 below shows regional poverty status from 1999/00 up to 2015/16 disaggregated with rural and urban poverty head count ratio.

Table 2: Trends in regional poverty headcount indices

1999/00 2004/05 2010/11 2015/16 Total Region Rural Urban Rural Urban Rural Urban Rural Urban Tigray 61.6 60.7 36.7 36.5 31.1 14.2 26.8 14.2 27.9 Afar 68 26.8 42.9 27.9 41.1 23.7 26.5 10.6 23.6 Amhara 42.9 31.1 40.4 37.8 30.7 29.2 28.8 11.3 26.1 Oromia 40.4 35.9 37.2 34.6 29.3 24.8 25.7 15.3 23.9 Somale 44.1 26.1 45.2 35.3 35.1 23.1 22.3 22.9 22.4 Benishangul 55.8 28.9 45.8 34.5 30.1 21.3 28.7 17.7 26.5 SNNP 51.7 40.2 38.2 38.3 30 25.8 21.9 14.4 20.7 Gambela 54.6 38.4 NA NA 32.5 30.7 26.4 16.6 23 Harari 14.9 35 20.6 32.6 10.5 11.7 8.5 6 7.1 Addis Ababa 27.1 36.2 29.9 32.6 NA 28.1 NA 16.8 16.8 Dire Dawa 33.2 33.1 39.8 32.9 14.2 34.9 23.3 11.1 15.4 Total 45.4 36.9 39.3 35.1 30.4 25.7 25.6 14.8 23.5 Source: National Planning Commission interim report (2017), NA stands for not available

2.8.2. Empirical Studies on the Determinants of Rural Poverty in Ethiopia

Dawit et al. (2011) examined extent and determinants of income poverty in selected rural villages located in different parts of the country using maximum likelihood estimates of the probit model. They revealed that family size, ownership of livelihood assets (land and livestock), diversification in crop production, engagement in non-farm activities, access to roads and utilization of microfinance services were found to be the significant factors affecting household poverty. Increased ownership of the basic livelihood assets, farm land and livestock, showed positive effect on the poverty status of the rural households.

Ayalneh et al. (2005) and Derjie and Eric (2015) conducted a study on the determinants of poverty in rural districts of Ethiopia using binary logistic regression model using both the food calorie intake and the cost of basic needs approaches of setting poverty line. The findings showed that the factors behind the persistence of rural poverty is strongly linked to entitlement failures understood as lack of household resource endowments to crucial assets such as land, human capital and oxen. Age of household head, age at first marriage of household head and livestock ownership, family size and income from sale of livestock and livestock products, and remittances were found to be the determining factors of poverty.

Abebe (2011) assessed the dimensions and determinants of rural household poverty in Itang special district of Gambella Region of Western Ethiopia using logistic regression model. The model output showed that the total households’ income was significantly influenced rural poverty. Household size and household head age were found to positively and significantly influence rural poverty.

Degye (2013) investigated the dynamics, determinants and vulnerability of rural poverty in Ethiopia using panel data taken in two rounds (2004 and 2009). The findings showed that the depth and severity of poverty were reduced from but the incidence of poverty was increased in these periods. The Probit model result showed that household size, livestock holding, farming occupation, life status, social network and other exogenous variables significantly influenced poverty in rural Ethiopia. Besides, he pointed out that the likelihood of households to be poor was about 45.5 percent.

Tsegaye (2014) carried out a household level analysis of rural poverty in Gozamen district of Amhara region using binary logit model. Education, livestock ownership, cultivated land holding, oxen holding, off-farm/non-farm income, credit utilization and frequency of extension contact were found to be statistically significant and have a strong negative association with the poverty status of rural households whereas family size alone was found to have a positive association with poverty status of rural households.

Tassew and Tekie (2002) conducted a study on national poverty profile of Ethiopia and found that poverty incidence, depth and severity being higher for those engaged in farming than those engaged in non-farming activities; poverty incidence, depth and severity were higher for the

31 illiterate than for the literate in both rural areas; the consumption poverty incidence, depth, and severity sharply declines in accordance with the households’ level of educational attainment and the incidence, depth and severity of poverty increases as family size increases.

2.8.3. Empirical Studies on the Time Needed to Exit Poverty

Reducing poverty is a key developmental goal. Clear understanding of what factors determine the poverty status of individuals and households is crucial to achieving the goal. Equally importantly understanding and estimating how long the poor households would take to exit from poverty draws attention of policy makers or researchers. If positive economic growth is registered, then one can estimate the average time needed to exit poverty. It means economic growth has a significant effect on poverty reduction. For every one percent GDP growth rate, there will be 0.15 percent reduction in poverty (World Bank, 2014). Tsegaye (2014) estimated the average time needed to exit poverty in Gozamen District of Amhara National Regional State. He found that, the estimated poverty exit of the poor rural households was 4.4 years provided that the 6.4 percent GDP per capita growth rate per year continues to grow.

Even though growth is unlikely to be constant or uniform across households, the average exit time provides the potential for poverty reduction strategies. Therefore, achieving broad-based economic growth helps to target the poor households and prepare analytical tools that consider poverty and growth jointly. It is in mind that the time under consideration is estimated average exit time not the exact duration.

2.9. Conceptual Framework of the Study

In this section the conceptual framework of the study is sketched. It shows how the national poverty reduction policy and strategies links with the different livelihood capitals. Hence, these livelihood capitals that were assumed to influenced poverty at household level were shown by the arrows.

National Poverty Reduction Policy and Strategies

Human Capital: Sex, Age, Family size, Educational attainment, Dependency ratio, Extension contact

Physical capitals: livestock, input Financial capitals: credit utilization, proximity to market and utilization, on/off income, non- other infrastructures farm income, asset ownership ownership Livelihood capitals Institutional Natural Capitals: land holding, irrigation use, plantation Characteristics: human Health, veterinary and cooperatives services

Determine rural poverty status

Poor

or non-poor

Figure 1: Poverty Framework: Adapted from Scoones (1998)

3. RESEARCH METHODOLOGY

This chapter presents socioeconomic characteristics of the study area, data type and method of data collection, sampling technique and sample size, method of data analysis, and definitions of dependent variable and basic independent variables with hypothesized sign.

3.1. Description of the Study Area

This study was conducted in Banja district of Awi zone in Amhara National Regional State of Ethiopia. The district is bordered on the south by Ankesha, on the West by Guangua, on the North by Fageta Lekoma and in the East by the West Gojam zone. The district is consists of 26 kebeles of which 25 are rural kebeles and one urban kebele. Injibara town is both the capital of Awi zone and Banja district. The town is located about 450 kms North West of Addis Ababa and 118 Km South of Bahir Dar.

Total population of the district was 111, 975 out of this 56,364 (50.3 percent) were female and 55,6115 (49.6 percent) were male (CSA, 2007). The number of farm households were 16,239 out of this 13,684 were male headed households and 2,555 were female headed households. The district has the total area of 47,915.82 ha. Land use pattern of the district is 12,277 ha cultivated land, 21,141.57 ha grazing/pasture land, 12,346 ha covered by forest and the rest 2,151.24 ha for other uses. The district comprised of Dega (80 percent) and Winadega (20 percent) agro ecologies and the altitude ranges from 1900- 2750 meter above sea level. The annual temperature is 26°C at the maximum and 16°C at minimum. It has unimodal rainfall distribution pattern. The rainy season for the area starts in May and extends to the end of October. The average annual rainfall reaches 2300 mm. Crop production, livestock farming and forestry are the main sources of livelihood of farmers in the district (DAO, 2018).

Crop production: The farming system of the district is mainly characterized by mixed farming where crop and livestock production are undertaken as it is the major livelihood strategies of the rural poor in the district. The rural farmers in the Banja district had not been able to produce sufficient amount of output to feed its population throughout the year. Hence, limited amount of crop yield is being produced which merely used for home consumption. Major crops produced in the district are, potato, teff, maize, wheat, barley, finger millet and other crops like bean and

34 onion The major crops area coverage produced in the district in 2017/18 cropping calendar was potato, teff, wheat and maize which covers 3200, 2800, 2000 and 1350 hectares, respectively. Some farm households use irrigation, particularly for potato and onion production which is the main crops produced and traded in the district (DAO, 2018).

Livestock production: Livestock plays a significant role in the study area as it is an immediate source of income to purchase food and non-food items. Livestock in the study area includes cattle, shoats, equine and poultry. The population of the livestock sector in 2017/18 production year were 74,379 TLU. The livestock sector contributes to crop production as draught power, source of cash income, manure, transport services and cultural services. By selling livestock and livestock products rural communities in the district tried to subsidize their consumption. Horses are the main source of draught power in the study area.

Source: Ethiopian Map Agency (2018) Figure 2: Location of the study area

3.3. Sampling Technique and Sample Size

Sampling is the process or technique of selecting a suitable sample for the purpose of determining the characteristics of the whole population. The population from which the samples drawn were rural households of Banja District who are practicing mixed farming system as their major source of livelihood.

3.2. Data types, Sources and Methods of Data Collection

Both primary and secondary data sources were used to collect quantitative and qualitative data. Primary data on demographic characteristics, socioeconomic and institutional factors and other relevant data assumed to meet the objective of the study were collected from randomly selected farm households in the study district. The Primary data collection process were made in May 2018 using structured questioner which was administered by trained enumerators with the supervision of the researcher. The questionnaire was designed and pre-tested in the field for its validity and content, and to make the overall improvement of the study and in line with the objectives of the study. While secondary data were collected from different published and unpublished sources, such as national planning commission, regional office of agriculture, district economic and finance development offices, district agricultural offices and kebele development agents were consulted to generate relevant data for the study. Besides, average local food price list was collected from trade and industry offices of the study district that show monthly basic food items.

3.3.1. Sampling Technique

Three stage sampling procedure was employed to select respondent farmers. The first stage was stratification of the District consisting of 25 rural kebeles (excluding urban kebele) in to two agro-ecological zones of which twenty rural kebeles were found in Dega and five kebeles were found in Woyena dega. In order to be a good representative for the District, the researcher believed and took 20 percent of rural kebeles were subjected to sampling. In the second stage five rural kebeles proportional to the agroecological zones were randomly selected in both stratum. Implying that four kebeles in Dega and one kebele in Woyena dega were selected. Finally, after identifying the sampling frame which contains the complete fresh list of all

36 households within each selected kebele with the help of kebele leaders and DA’s, respondent rural household heads were selected randomly in proportion to their total number of households in each kebele.

3.3.2. Sample Size Determination

Sample size determination is an important element in any survey. It is related to survey precision, survey budget and data quality. Besides, it is a complex decision taking many factors like survey precision, reference indicator or important variable, data quality etc. have to be considered. It varies with the nature and types of research designs. Despite its importance determination of sample size is a difficult task. Its wide applicability and from the perspective of this research objectives, the total number of sample size was determined by simple formula of Yamane (1967).

Hence, Yamane formula is given as follows: 푁 푛 = (12) 1 + 푁(푒)2 where, n is the sample rural household, N is the total population of rural households in the selected kebeles of the district, and the level of precision (e) set at 7 percent. As more and more of the population group is considered for sampling, then its sample variance decreases there by increase the inference level of the sample to the true population. Considering precision level and financial resource the researcher decided the precision level to be at 7 percent. Accordingly, the total population in the selected five rural kebeles 2833 household heads. Therefore, the total sample size of this study was 190 household heads.

Table 3: List of selected kebeles with their respective no of sample households

Agro ecology Total Selected kebeles Total household Selected kebeles head household head Dega 20 Arsa-na-gembiha 280 18 Chaba-Gisa 633 42 Dangia 560 38 Aseramariyam 580 40 Woyinadega 5 Kidamaja 780 52 Total 25 5 2833 190 Source: DAO and own computation (2018)

3.4. Methods of Data Analysis

In order to answer the specified objectives of the study, descriptive analysis, Foster-Greer- Thorbecke (FGT) poverty measures and econometric methods of analysis were employed.

3.4.1. Descriptive Analysis

Descriptive statistics like percentages, ratios, mean values, frequencies, and others tests were used to assess the status of poverty in the study area. Independent sample t-test (for continuous variables) and Pearson chi-square analysis (for discrete variables) were applied to test whether significant mean difference was observed between poor and non-poor counter groups of respondent households.

In order to set poverty lines, the cost of basic needs approach (CBN) was used. It comprises a food bundle items that would provide minimum of 2,200 Kcal per person per day, which is the minimum calories required for an adult to maintain an average physical life under normal conditions (ENHRI, 2000). Therefore, a household is considered to be living in poverty provided that the per capita daily household consumption expenditure was unable to attain 2,200 kcal. Besides, data on household’s annual expenditure on non-food basic needs were included. But expenditure on durable goods were not included in this study. Hence, the sum of expenditure on food consumption and the expenditure on non-food basic needs results the total annual expenditure of the household. Accordingly, the total poverty line was determined.

Once we know the total poverty line using consumption expenditure on food and non-food needs, the three well known dimension of poverty measures were computed (Foster et al.,1984). Foster-Greer-Thorbecke (FGT) measure of poverty is highly dominant and preferred to other poverty measures due to its ethical flexibility (captured by the parameter α), decomposability across subgroups, sub- group consistency, ability to capture the most desirable properties of a poverty indices and understandability. Hence, Foster-Greer-Thorbecke (FGT) index, measure of poverty can provide us the incidence of poverty (measured by the headcount ratio α = 0), the depth of poverty (measured by poverty gap index α=1), and the severity of poverty (measured by the squared poverty gap index α=2).

Mathematically, the FGT index can be computed; 푛 1 푧 − 푥𝑖 훼 푃 = ∑ ( ) (13) 훼 푁 푧 푖=1

Where, α = 0, 1, 2

푃훼 , poverty measure Z, poverty line Xi, consumption expenditure of the household per adult equivalent N, number of sample households n, number of poor households α, measure of sensitivity of the index to poverty

The second objective of this study was analyzed using poverty statistic in which the property of decomposability by population subgroups and sensitivity to expenditure/income distribution among the poor (Morduch,1998). Thus, for the ith household below poverty line, the expected time needed to exit poverty if consumption per capita grows at a positive rate 푔 per year is:

ln(푧) − ln (푦푖) 푤 푡푖 = (14) 𝑔 ≅ 푔 푔

th Where, 푦i = per capita annual consumption expenditure in the i poor household, 푔 is consumption per capita growth rate, 푧 is poverty line and W is Watts index.

If one can estimate the individual poor household exit time, it is not hard to estimate the average exit time of the poor households. It considers the per capita consumption expenditure of the poor households per year given consumption per capita grows at positive rate per year is;

ln(푧) − ln(µ푝) 푤 푡푎푣 ≈ = (15) 𝑔 푔 푔

Where; µ푝 the average per capita consumption expenditure of the poor households (those who are below the poverty line).

3.4.2. Poverty Measures and Procedures

As discussed in chapter two, the study used cost of basic needs approach that demarcates the poor household from their counter non-poor groups by obtaining the predetermined minimum

39 daily caloric requirement of 2,200 per adult equivalent per day. Different household surveys use different types of recall periods even for the same goods. For this study thirty days’ recall period was used. In order to get the predetermined minimum kilocalorie, household were asked to recall the quantity (Kg or Lt) and the type of the food (crop) item consumed in the last thirty days of non-fasting season. Once it is known, it can easily be converted to kilocalories based on the conversion factor (EHNRI, 2000). The total kilocalorie obtained from basket food items of the family was divided by the adult equivalent to get the amount of average kilocalorie a particular household obtained per adult equivalent per day. By taking the average local market prices of each food items purchased and attaching the monetary value of own produces and multiplying by the value of kcal/AE/day, we obtained the amount of money needed to get the basket of food items for an individual required per day. This can be converted to yearly amounts of food poverty line. Similarly, expenditures on non-food of basic goods including clothing, schooling, transportation, fuel and miscellaneous expenses were asked as how much was spent during the last month/year depending on the frequency of purchase or payment. Once we obtain food poverty line, non-food poverty line was determined by using a simple linear regression (Equation 1).

Finally, the consumption expenditure for food and expenditure on non-food basic items are added up and resulted in an adjusted “cut-off” point as a poverty line which could enable us identify respondent households. If households annual mean consumption expenditure is below the poverty line or the “cut-off” point, then they are as poor and non-poor if their annual mean consumption expenditure above the poverty line.

3.4.3. Econometric Models for Determinants of Household Poverty Status

Poverty determinants can be analyzed in a number of econometric models as discussed in chapter two. For the aforementioned reasons, censored Tobit regression model was employed to analyze the determinant of rural poverty as discussed below.

Tobit Regression model: It was first developed by Tobin in 1958 and has been widely used by economists for measuring effect of changing explanatory variables on probability of being poor. The Tobit model is a member of censored regression models, which has a latent (dependent) variable that is not observed, whereas the explanatory variable is observable.

Tobit model is an extension of Probit model and it is one of the approaches dealing with the problem of censored data (Johnston and Dandiro, 1997). It is a hybrid of the discrete and continuous dependent variables. When the data are censoring problem, the distribution that applies to the sample data is a mixture of discrete and continuous variable. The use of Tobit model is conceptually preferable to conventional linear regression models because parameter estimates from the former overcome most weaknesses of linear probability models namely: providing estimates which are asymptotically consistent and efficient (Mcdonald and Moffit, 1980). In Censored regression model independent variables are known for all observation in the sample but data of dependent variable is observable only in limited boundary (in this case those households who are below poverty line). In addition, Tobit model measures not only the probability of a household being poor but also the intensity or depth of poverty (Tobin, 1958).

Therefore, for this particular study Tobit regression model was conceptualized to determine rural households’ poverty determinants. The model is chosen because it has an added advantage over other discrete models (logit or probit) in that it measures not only the probability of a household being poor but also the intensity of poverty level. Therefore, the third objective of this study was estimated by Tobit regression model.

The Tobit will be specified as the follows;

∗ 푃푖 = 훽풳푖 + 푒푖 (16) ∗ 푃푖 = 0 𝑖푓 푃푖 ≤ 0 (𝑖. 푒 푍 < 퐼) Hence, we have censoring at zero, ∗ ∗ 푃푖 = 푃푖 𝑖푓 푃푖 > 0 (𝑖. 푒 푍 > 퐼) Therefore, the observed model is ∗ ∗ 푃푖 = 훽풳푖 + 푒푖 𝑖푓 푃𝑖 > 0 푎푛푑 푃𝑖 = 0 표푡ℎ푒푟푤𝑖푠푒

Where 푃푖= Limited dependent variable. The dependent variable is poverty level of the respondents. The poor households are represented by poverty depth, while the non-poor households have zero as their dependent variable. ∗ 푃푖 is latent variable that may or may not be directly observable. ∗ 푃푖 is observable for poor and unobservable for non-poor.

풳푖 = explanatory variables, those were supposed to determine poverty (age of household head, household size, sex of household head, TLU, land holding, oxen ownership, non/off-farm income, credit utilization, input utilization, asset ownership, irrigation utilization, cooperative membership etc.) and 푒푖 is disturbance error.

훽푖 = vector of estimable coefficient parameters. I= the mean annual consumption expenditure on food and non-food basic items Z = Poverty line і = 1, 2, 3,…..n, the number of observations (sample).

Accordingly, the loglikelihood condition of the Tobit model has two sets or the entire sample consists of two different sets of observations. The first set contains the observations for which the value of 푃푖 is zero and for these observations only the explanatory variables were known ∗ and the fact that 푃푖 less than or equal to zero whereas the second set consists of all observations for which the values of both explanatory variable and the dependent (푃푖) are known and the fact ∗ that 푃푖 is positive.

1 1 (푃 − 훽 풳 )2 훽 풳 퐿 = ∑ − [log (2휋휎2)] − ∑푛 푖 푖 푖 + ∑ 푙표푔 [1 − 훷( 푖 푖)] (17) 푃푖 2 2 푖=1 휎 푃푖=0 휎

Where the first two parts constitute the first part of the likelihood and the third part constitute the second part of the likelihood estimation (Raghbendra et al., 2006).

Unlike the case of ordinary least square (OLS) coefficients, it is difficult to interpret the estimated coefficients of the Tobit model as a marginal effect because there are three main conditional expectations of interest. These are; the conditional expectation of the underlying latent variable (y*), the conditional expectations of the uncensored observed dependent variable (y|y>0) and the conditional expectation of the observed dependent variable (y). Following Greene (1997), Johnston and Dinardo (1997), McDonald and Moffitt (1980), the marginal effects of these conditional expectations of the Tobit model, respectively are given as follows;

∂퐸(푦∗⁄푥) = 훽 (18) ∂푥

∂Pr (y > 0/풳 풳훽 훽 = 훷 ( ) (19) ∂풳 휎 휎

∂E(푦⁄푥) 풳훽 = 훽훷 ( ) (20) ∂푥 휎

The choice of marginal effect depends on the point of interest based on the objective of the study. For instance, if the interest is to make statements about the conditional mean function in the population despite the censoring, Equation 18 is used for the censored data. If the researcher is interested on average values of the dependent variable for those who have already participated in a program or those who are considered as poor, equation 19 is used. Finally, if one wants to know the average of value of the whole population, and how those values vary with the independent variables equation 20 is used.

In this study the researcher interested to interpret the three marginal effects. Because it clearly shows the probability of a household being poor, the intensity or level of the poor households and intensity of the poverty status among the whole population.

3.5. Definitions of Variables and Working Hypothesis

Once the analytical procedure and its requirements are known, it is necessary to clearly state and identify the dependent and potential explanatory variables that are expected to affect poverty and describe their unit of measurements.

3.5.1. Dependent Variable

Poverty level (POVLV): The dependent variable is poverty level of the respondents. The poor households are represented by poverty depth, while the non-poor households have zero as their dependent variable.

3.5.2. Independent Variables

The independent variables that are expected (hypothesized) to have influence on household poverty level in the study area were included for Tobit regression model. The independent variables were selected based on economic theories, past research findings, and experts and author’s knowledge of the poverty status and situation of the study area. Therefore, by taking the poverty depth as dependent variable, the following variables were the potential determinants of

43 household poverty in the study area. Hence, any exogenous variable having negative coefficient is expected or hypothesized to reduce poverty whereas, any exogenous variable positively related to the poverty would deteriorate the wellbeing of the household by increasing the poverty level.

Household size (HHSZ): It is a continuous variable. It is the number of household members who share resources or meals under the same household measured in adult equivalent (AE). According to the findings of Tassew and Tekie (2002) household size has positive association with poverty level of a household. They justified their findings that as the household size increase the number of individuals sharing the resources available in the household increases; in which this might leave the household under risks of short fall of basic needs. Hence, it is hypothesized that household size and poverty level are expected to be positively related.

Age of the household head (HAGE): This a continuous variable which refers to the age of the household head measured in years. The higher the age of the household head, the more stable the economy of the farm household would be. Besides, the older the household head, the more experience of the social capital and physical environments such as weather forecasting and risk minimizations through crop diversification and endowed with more capitals than the younger heads (Datt and Jolliffe, 2005). However, there are cases where poverty and age go together. Since agricultural operation by its nature requires physical wok, the ability of a farmer in farm operations become weaker and weaker as his age increases beyond a certain limit. Reduction in farming operation leads to a subsequent low production and income (Etim et al., 2009). Therefore, age of the household head and poverty level is hypothesized as indeterminate which may negatively or positively related.

Sex of household head (HSEX): This is a dummy variable indicating male or female of the household head. It takes a value of 1 if the household head is male, 0 otherwise. Akerele and Adewuyi (2011) found that male headed households are in a better position to pull more labor force together than the female headed ones. Hence, a family headed by male has a better chance to generate income than female do. Moreover, with regard to farming experience, males are in a better position than the female headed ones. Therefore, it is hypothesized that male headed households have less likely to be poor.

Dependency ratio (DPRTIO): In a household where adults or productive age groups are higher than the non-productive age groups, the probability of the household to be in poverty would be less. It is the ratio of children under age of 15 and old age of above 65 to active labor force (age between 15-64) expressed in terms of AE. The existence of large number of children under age of 15 and old age of 65 and above in the family could worsen the poverty level of a household. Household with large number of dependent members tends to be poorer than those households with small dependent groups (Denano, 2018). Thus, it is hypothesized that a household with relatively large number of dependent members has a positive relation with household poverty status.

Education level of the household head (EDULVL): This is a continuous variable which stands for education level of the rural household head measured in years of schooling completed. Households who have household heads with relatively better education are more likely to be non-poor than those households headed by uneducated (illiterate) ones. Education equips individuals with the necessary knowledge of how to make living and influence in decision making process. Hence, literate individuals are very keen to get and use information (Ejigayhu, 2012). It increases the farmers’ awareness about the possible advantages of modern agricultural technologies and method of production; this helps them to diversifying household income sources. Therefore, the educational status of the head of the household is expected to be negatively related with poverty.

Livestock ownership (LVSTOWN): This is the total number of livestock holding of the household excluding oxen measured in Tropical Livestock Units (TLU). The livestock ownership has great importance for rural farmers. A Household with large livestock size are less likely to be poor. Dawit et al. (2011) justify that possession of livestock serves as a hedge against food insecurity, source of cash income, principal form of saving and investment. Hence, it is expected that households with high number of livestock in TLU would likely to be less poor than those households who have few or no livestock. Thus, it is expected to be negatively related with poverty.

Oxen ownership (OXOWN): There is strong relationship between crop production and oxen ownership in mixed farming system. Oxen provide manure and source of draught power for

45 crop production (Ayalneh et al., 2005). Therefore, oxen ownership and poverty status of households is expected to be negatively related in the study area.

Off-farm and non-farm income (OFINCM): It is a continuous variable measured in monetary values (in ETB). It represents the amount of total annual income (log normalized) per household obtained from various off-farm/non-farm activities in 2017/18 production year. Various income sources of off farm and non-farm income generating activities and opportunities are common practices of most rural households. Households who are engaged in various activities or receiving incomes from remittances, renting of pack animals and other informal businesses were better endowed with additional income to meet their food and non-food requirements (Dereje, 2008). Hence, such off-farm or non-farm income sources determine the poverty status of the household. As a result, it is expected to have a negative impact on poverty.

Distance from the nearest market center (DSMRKT): It is a continuous variable measured in kilometer (KM). Households who have proximity to market center have better chance to improve their income (Semere, 2008). Access to market and other public infrastructures may create opportunities for more income by providing non-farm employment and easy access to transportation. Besides, proximity to market centers creates access to additional income by providing opportunities of engaging in non-farm employment as well as selling different agricultural products; hence, better chance to reduce household’s poverty. Therefore, as a household is far from the market center, he is likely to be poor. Thus distance to the nearest market center is expected to be positively related to poverty status.

Land size (LNDHLD): This refers to the total area of land owned and cultivated by the household measured in hactar. Cultivated land is one of the livelihood capitals available for food production thereby ensuring household entitlement to food (Alemayehu et al., 2006, Adugna and Sileshi, 2013). Land being an important asset and factor of production in the rural households, a household with larger land holdings have better opportunity to reduce households’ poverty, and hypothesized to have a negative impact on the household poverty level.

Credit utilization (CRTUTZ): This is a continuous variable measured in ETB. It is the amount of credit received in 2017/18 production year either for the purchase of agricultural inputs or

46 for the purchase of food and non-food items. Credit utilization is theoretically expected to reduce poverty through cash investment in different productive activities and to generate better income. The possible explanation is that, in the study area, those households who participate in credit scheme to earn higher amount and become capable to improve their income position by performing different activities. Besides, households who get credit at times of food shortage coped the risk by using the credit directly for food consumption.

Alemayehu et al. (2006) found that credit influenced poverty are negatively. It is to mean that those households who used credit were less likely to be poor than those who did not received credit. Hence, it was hypothesized that credit would have a negative impact on poverty level of households in the study area.

Frequency of extension contact (FREXTN): This is a continuous variable measured by the number of times that the household head visited extension agents per month. Access to extension services enhances his/her understanding of modern agricultural inputs and technologies which will enhance the way to adopt new agricultural technologies. Studies conducted in southern Ethiopia and rural Nigeria proved that, households’ who have access to extension services help them to undertake modern farming techniques and farm management principles that improves their agricultural productivity which ultimately lead them to escape poverty (Apata et al., 2010; Adugna and Sileshi, 2013). Hence, it is expected that the greater the frequency of extension contact, the less likely that the household is in poverty.

Utilization of irrigation (UTLIRG): This is a dummy variable which assumes a value of 1 for irrigation users and 0 otherwise. Irrigation as one of the technology options available to enable the farmers to diversify their production, practice multiple cropping and supplement their land when rain shortage happens. Hence, the farm household can be capable to produce more than once a year. Accordingly, it helps the farmers to increase production and income. Besides, it can benefit the poor through raising yields and production, lowering the risk of crop failure, and generating income. Through time it can enable smallholders to adopt more diversified cropping patterns, and to switch from low-value subsistence production to high-value market- oriented production (Agerie, 2016). Therefore, access to irrigation is expected to be negatively related with poverty.

Utilization of agricultural inputs (UTLINP): This is a dummy variable. It takes a value of 1 if a household have access and used the input and 0 otherwise. Access and use of agricultural input included in the study were fertilizer, improved seeds, , modern agricultural implements, small generators, use of pesticides and weedicide. Use of appropriate agricultural input is supposed to increase agricultural productivity (Tesfaye, 2013). Accordingly, it has a positive effect on household’s welfare status. Therefore, in this study it is hypothesized that, households who used at least one of the improved agricultural input in the production year, would less likely to be poor than those who did not.

Asset ownership (ASTOWN): It is a continuous variable valued in monetary terms (ETB). It includes farm implements like small generators, carts, sickle, modern bee hive and a number of other asset available at household level that used for agricultural operation. The assumption is that a household with sufficient asset possession is supposed to have a better opportunity to produce more and generate income for his family. It is to mean that a household with valuable asset is expected to use it to improve his or her welfare by increasing work efficiency (Nega, 2015). Therefore, the current value of asset holding is expected to be negatively related with poverty level of the household.

Cooperative membership: This is a dummy variable that takes a value of 1 if a household is member of cooperatives, 0 otherwise. Cooperative is an institution that meets most of the dimensions of poverty, providing opportunities, facilitating empowerment, and enhancing food security (Kwapong and Hanisch, 2013). Hence, cooperative is crucial to improve living standard of the rural poor household and reduce extreme poverty. Therefore, in the study area cooperative membership is hypothesized to be negatively related with poverty.

Table 4: Summary of the variables included in the model

Variable code Variable type Variable definition and measurement Expected sign Dependent POVLV Limited Assumes a value of 0 for non-poor, and Dependent Poverty depth for poor households Independent HAGE Continuous Age of the household head in years -/+ HHSZ Continuous Household size in adult equvivalent (AE) + HSEX Dummy Sex of the head: 1 if male; 0 otherwise - DPRTIO Continuous Dependency ratio in adult equivalent (AE) + EDULVL Continuous Education level of the head in years of - schooling LVSTOWN Continuous Number of livestock ownership in TLU - OXOWN Continuous Number of oxen ownership in number - OFINCM Continuous Off-farm/non-farm income earned in - 2017/18 in ETB DSMRKT Continuous Distance to the nearest market center in + Km LNDHD Continuous Land holding of a household in hectares - CRTUTZ Continuous Amount of credit recived in 2017/18 in - ETB FREXTN Continuous Number of of extension contact in a month - UTLIRG Dummy Utilize irrigation: 1 if utilize; 0 otherwise - UTLINP Dummy Utilize agricultural inputs: 1 if used inputs; - 0 otherwise ASTOWN Continuous Household current asset ownership in ETB - COPMEM Dummy Cooperative membership: 1 if member, 0 - otherwise Source: Own definition (2018)

4. RESULTS AND DISCUSSION

This chapter presents the way to distinguish the poor and non-poor household groups, and presents the overall findings from the descriptive, FGT and econometric analysis. The descriptive analysis include mean, percentages, standard deviation and frequency distribution. Inferential statistics were also employed to compare the mean difference between poor and non- poor household groups with respect to different livelihood capitals. FGT indices of poverty analysis results were also discussed. Besides, econometric analysis (Tobit model) outputs of poverty determinants and its marginal effects were interpreted accordingly.

4.1. Poverty Line Determination

In order to achieve the objectives of this study constructing poverty line using cost of basic needs (CBN) approach was used. Based on the cost of basic needs approach, first estimate the cost of acquiring enough food for adequate nutrition usually 2,200 kilocalric requirement per person per day, and secondly add the cost of other non-food basic essentials such as clothing, shelter, transport, medical services, schooling, grinding, social and religious expenses, etc. In other words, first determining the food consumption bundle “Food basket” just adequate to meet the required food energy requirements at prevailing market price; and second, adding an allowance for non-food basic needs to food expenditure.

Accordingly, the “Food basket” which constitute seventeen food items either from their stock and or purchase were identified. Then these food items were valued at the annual average local prices of the study district in 2017/18. These consumed food items were converted to kilocalorie and then divided to households in adult equivalent. Therefore, on the nationally predetermined minimum caloric requirement (2,200 kc) for daily activities, the food poverty line in the study area was determined. Subsequently, the food poverty line that gives the minimum daily food caloric requirement was Birr 3232.42 per adult per year as shown in Table 5 below. Hence, the food poverty line that demarcates the poor from the non-poor households was birr 8.85 per day per adult equivalent. Put differently, an adult requires 8.85 Birr per day to attain the minimum 2,200Kcaloric requirement for his/her daily life.

From the total sample respondents, 83 households were found to be unable to meet the minimum subsistence food requirement whereas 107 households were found to meet their energy requirement. The food poverty line of the study district was below the national average of Birr 3772 in 2015/16. The reason might be low food prices in the study district compared to other parts of the country. Besides, the majority of farmers in the district were producing non- marktable crops, which further leads them unable to purchase the required food items.

Table 5: Food consumption, food prices and food poverty line (N=190)

Food *Mean Gm per Mean Mean price Value of food Expendit type Kcl per day/ Kcl per per kg/Lt poverty line per ure share gram/Lt adult Gm per (ETB) year (ETB) (%) day Teff 3.589 75.3 270.14 14 542.66 0.16 Maize 3.571 134.3 479.55 8.5 587.83 0.18 Barely 3.723 22.8 84.76 12.5 146.55 0.04 Millet 3.41 47.2 160.63 12 291.10 0.09 Wheat 3.623 30.9 111.91 8 127.25 0.05 Potato 0.87 331.9 288.81 4 683.84 0.21 Onion 0.713 10.3 7.37 7 37.26 0.01 Tomato 0.373 1.329 0.48 8 5.33 0.01 Pepper 0.933 2.9 2.79 10 15.43 0.005 Cabage 0.37 14.8 5.49 3 22.93 0.007 Beans 3.514 10.9 38.37 11.5 64.67 0.02 Milk 0.737 15.5 11.43 8 63.92 0.02 Egg 0.68 1.3 0.9 31 21.20 0.006 Oil 8.964 5.2 46.36 45 119.84 0.04 Butter 7.364 3.5 25.66 90 161.53 0.05 Coffe 1.103 8.9 9.92 70 324.09 0.10 Salt 1.783 8.3 14.71 4 16.99 0.001 Total 1559.3 3232.42 0.999 Source: Own computation (2018), *EHNRI (2000)

Once the food poverty line is determined it is easy to compute the total poverty line. The food poverty line obtained is translated and incorporated the expenditure required to attain basic non- food needs. Hence, the non-food poverty line can be estimated as the share of the food expenditure to toal expenditure of each household in adult equivalent on a constant and the log of the ratio of total expenditure to food poverty line (Ravallion, 1992). Accordingly, the food share of households that had failed to attain the food poverty line was found to be 75.14 percent

51 whereas the share of non-food was 24.86 percent. Subsequently, the non-food expenditure in the study area was Birr 1069.45 per adult per year and, therefore, gives the total poverty line of Birr 4301.85 per adult per year. In other words, the absolute poverty line (the sum of food poverty line and non-food poverty line) that demarcates the poor households from the non-poor was found to be Birr 4301.85. Those households whose mean annual consumption expenditure falls below Birr 4301.85 per adult were counted as poor and those households whose mean annual consumption expenditure above this “cut-off” point were counted as non-poor. In other words, an adult requires a subsistence consumption expenditure of Birr 359 per month at 2017/18 crop prices in the study area.

According to the National Planning Commission of Ethiopia (2016), the annual consumption expenditure poverty line of the country was Birr 7184. Compared to this figure, the study area annual consumption expenditure is low. The probable justification might be, the ability of farmers, in the study area, to spend on non-food items is weak as they are short of cash and usually hardly fulfill their non-food items.

4.2. Poverty Measures and its Status

The three well known group of poverty measures namely head count ratio (incidence of poverty), poverty gap (extent of poverty) and poverty gap squared (severity of poverty) were analyzed using Foster-Greer-Thorbeke equation (13). Based on the poverty line determined above, the FGT class of poverty indices were found to be 0.44, 0.09 and 0.02 for head count, poverty gap and poverty severity respectively.

Table 6: Poverty measure of sample households

Poverty index Values Poverty head count index 0.44 Poverty gap index 0.09 Poverty severity index 0.02 Source: own survey computation (2018)

Poverty head count index: The headcount index measures the proportion of the sample population that is counted as being poor. Table 6 above revealed that the absolute head count

52 index of the sampled population showed that 44 percent of households were considered as poor or 44 percent of sampled populations were unable to fulfill the predetermined minimum consumption requirement. Put differently these proportion of sampled population households were unable to meet the minimum amount of consumption expenditure of Birr 4301per adult equivalent per year and hence found under absolute poverty. This figure is considerably higher than the national and regional figure reported by the government of Ethiopia (NPC, 2017). Hence, the study district is one of the highest poverty incidence registered in the region. Poverty head count does not capture how the poor are and does not change if people below poverty line become poorer.

Poverty gap index: This measures the average proportionate poverty gap of consumption expenditure of the population where the non-poor households have zero poverty gaps. It tells us the extent to which an individual is found below the poverty line. Accordingly, the mean aggregate consumption shortfall relative to the poverty line across the whole population is found to be 0.09. In other words, the poor households require an additional 9 percent of the present consumption expenditure to attain their minimum basic needs. Although the poverty gap index is the mean of the gaps between the welfare of the poor and poverty line, it does not capture the consumption inequality among the poor households.

Poverty severity: This index measures not only the distance separating the poor households from the poverty line but also considering the inequality among the poor. Thus higher weight is placed on those households further away from the poverty line. The severity index for this particular study was 0.02 implying that there exist 2 percent consumption inequality among sampled poor household in the study area. The figure is consistent with the Amhara National Regional State severity of 2.4 percent in 2015/16.

According to 2015/16 interim report of Ethiopia, the Amhara National Regional State the absolute head count index, poverty gap and poverty severity was 28.8, 6.8 and 2.4 percent respectively. There is discrepancy between incidence of poverty in the study district and the regional. The reason might be the number of sampled respondents used for this particular district was low as it might reduce its representativeness of the whole population. Besides, local prices would not reflect the true values of food items at regional level. Moreover, in the study

53 district natural shock particularly frost and heavy snow frequently damaged most of their annual and perineal crops.

Even though the incidence of poverty in the study district vary from the national and regional poverty incidence, previous findings support the result. Study conducted by (Degye, 2013) in four major regions (Tigray. Amhara, Oromiya and SNNP) showed that incidence of rural poverty was increased from 37.5 percent in 2004 to 52.9 percent in 2009 while the the national incidence of poverty estimated by the government was 29.6 percent. Similar finding by Denano (2018) suggest that the incidence of poverty in rural Soro district was 65.7 where the national poverty incidence estimated by government was 23.5 percent. Besides, study conducted by Zegye (2017) in Damote Gale Distrcit of Southern region the incedince of poverty was 56.17 percent which was higher than the national incidence estimated by the government.

Table 7 below describes the sample population by the poverty status with their respective kebeles. The highest number of poor households were found in Dangia and Asrammarm while the lowest number of poor households were found in Arsan-gembeha keble. Besides the highest number of non-poor households was found in Kidamaja kebele.

Table 7: Distribution of Poor and non-poor sampled households by Peasant association

Category Sample kebeles by poverty status (N=190) Arsan- Chaba Kidmaja Asramram Dangia Total gembeh -Gisa Poor Count 15 6 19 22 22 84 Percentage 17.8 0.07 22.6 26.2 26.2 100 Non poor Count 37 8 23 19 19 106 Percentage 34.9 0.07 21.7 17.9 17.9 100 Total Count 52 14 42 41 41 190 Percentage 27.3 7.4 22.1 21.6 21.6 100 Source: Own survey result (2018)

4.3. Consumption Expenditure

In the study area the majority of farmers consume mainly cereal crops like teff, barly, maize, wheat and vegetable crop like potato. Annual income from sale of agricultural products and charcoal making were the major source of annual cash in the district.

The survey result, table 8 below, indicates that the overall annual mean consumption expenditure per adult equivalent for the sample population was Birr 4624.2 per year with standard deviation of 1512.8. The mean consumption expenditure of poor and non-poor groups of households were Birr 3355.1 and 5629.8 with standard deviation 1285.4 and 446.8 respectively. Significant mean difference was observed between poor and non-poor sample population at 1 percent significance level. Besides, annual mean consumption expenditure on food and non-food sample population was Birr 3810.8 and 2092.9 with standard deviation of 1463.7 and 443.6. There was statistically significant mean difference between the poor and non- poor sample households in terms of food and non-food expenditure per annum per adult equivalent at 1 percent significance level.

Specific to food items, significant mean difference was observed between poor and non-poor households in their expenditure on maize and finger millet at 5 percent and 10 percent significance level. Besides, annual mean expenditure on beans, oil, egg and onion observed significant difference between poor and non-poor households at 1 percent probability level. Moreover, specific to non-food annual expenditure on kerosene, transport and religious expense showed significant mean difference between the poor and non-poor Household at 10 percent probability level whereas housing and schooling expenses at 5 percent probability level.

Table 8: Total expenditure and food expenditure of sampled households per year in Birr

Expenditure Poor (N=84) Non-poor (N=106) t-value Mean St. Dev Mean St. Dev Total 3355.1 446.7 5629.8 1285.4 15.48*** Food 2579.4 543.9 4786.6 1203.4 15.58*** Non-food 1781 293.3 2339.5 383.9 11.01*** Teff 193.39 177.9 206.76 217.4 0.45 Maize 203.53 127.8 156.45 131.64 -2.47** Barely 58.18 110.06 41.77 78.66 -1.19 Millet 77 156.5 121.6 183.6 1.77* Wheat 52.85 75.25 45.39 71.29 -0.67 Potato 333.03 263.49 293.23 245.96 -1.07 Onion 15.45 14.32 27.07 14.8 4.12*** Tomato 1.01 3.82 2.88 10.55 1.55 Beans 28.61 27.59 46.96 53.07 2.87*** Milk 46.6 19.55 46.99 21.32 -0.02 Egg 3.04 1.48 5.04 3.3 2.79*** Oil 40.98 40.22 87.96 72.49 5.32*** Butter 21.23 44.90 25.13 67.16 0.46 Coffee 144.31 71.8 79.41 62.92 -0.98 Source: Own survey computation (2018) ***, ** and * significant at 1%, 5% and 10% significance level

4.4. Association of Livelihood Capitals of Households with Poverty

In this section different livelihood capital which are assumed to influence the rural farmers were analysed. It includes human capitals, physical capitals, finical capitals, natural capitals and other institutional aspects were discussed accordingly.

4.4.1. Human Capital

Human capital represent once skill, knowledge, ability, education, good health which helps an individual to increase his/her productivity and to pursue different livelihood strategies. In this study human capital included sex, education, family size, extension contact, age and dependency ratio. Accordingly, the total size of sample household members was 1186 out of this sample household members 629 (53 percent) were male and 557 (47 percent) were female. Of the total sample household members 16 percent were household heads.

Based on the survey result showed in table 9 below, from the total sample households 87 percent and 13 percent were male and female headed households respectively. The average household size in AE was 5.1 per household with standard deviation of 1.7. The figure exceeds the national average of 4.2 persons per household. Compared to lowland areas, the number of people living in the highland area per square kilometer is high. This could be the probable explanation for higher number of family size in the study area. Educational attainment of sampled households or literacy rate was 1.6 years of schooling on average with in the rage of illiterate and eleventh grade maximum. Besides, 57 percent of sampled households were unable to read and write. The mean age of sample household head was 47.8 years with standard deviation of 10.23. The youngest household age was twenty and the oldest was eighty years. The average dependency ratio in the study area was 0.76 with minimum and maximum of zero and 3.5 respectively. Extension visit made by household head or contact on average was 3.3 days per month with standard deviation of 1.7.

Analyzing human capital based on poverty status can give us more sound information, educational attainment of household head showed statistically significant mean difference between the poor and non-poor sampled households. The average educational attainment of the head of the poor households was 0.7 years with standard deviation of 1.7 whereas the non-poor household heads were 2.4 years with standard deviation of 3 with statistical mean difference at 1 percent significant level. The negative value of t-test indicted that education attainment of the household head and the poverty status was negatively correlated. The average household size of poor households in AE was 5.3 per household with standard deviation of 1.6 and that of non- poor households was 4.9 with standard deviation of 1.7. There was no significant mean difference in terms of family size between poor and non-poor sampled households.

The average age of poor households was 48.5 years with standard deviation of 9.02 whereas the non-poor households was 47.30 years with standard deviation of 11.12. There was no significant mean difference between the two groups in terms of the age of the head of the household. The dependency ratio which indicates ratio of non-active members to that of active household members between the poor and non-poor was 0.8 and 0.73 with no statistical mean difference. Extension visit or contact made by a household between the poor and non-poor groups was 3.3 and 3.4 days per month respectively with no significant mean difference.

Table 9: Human capitals with poverty level

Human capital Poor (N=84) Non-poor (N= 106) Total (N=190) t-value Mean St. Dev Mean St. Dev Mean St.Dev Age 48.5 9.1 47.3 11.1 47.8 10.23 - 0.8 Household size 5.3 1.6 4.9 1.7 5.1 1.71 - 1.08 Education level .7 1.7 2.4 3.0 1.3 2.63 - 4.56*** Dependency ratio 0.8 0.6 0.7 0.6 0.7 0.63 - 0.78 Extension contact 3.3 1.7 3.4 1.8 3.4 1.77 - 0.06 Source: Own survey result (2018) *** significant at 1% significance level

Sex of the head of the household is one of human capital in which it influences poverty. From the total sampled household population 87.9 percent of the respondents were male headed whereas 12.1 percent were female headed households. The highest number of male headed households was found in kidamaja and smallest in Arsana-gembeha rural kebeles. Moreover, 89 percent of poor households were male headed and 11 percent were female headed. The non- poor male headed households were 80 percent and that of female headed were 20 percent. Out of the total female headed respondents the majority (67 percent) were poor. Correlation test using Person chi-square does not show any significant mean difference between the poor and non-poor groups of the respondents in terms of sex of the household head.

4.4.2. Physical Capital

In this study housing, livestock ownership, oxen ownership, access to farm inputs, distance to nearest market place in kilometer and input utilization were considered as physical capitals. The housing type of sample households were either corrugated iron sheet, grass roofed or both types. Among the sampled respondent households 68.9 percent, 5.8 percent and 25 percent were possessed corrugated iron roof, grass roofed and both types respectively. Majority of farmers in both poor and non-poor groups live in corrugated iron sheet homes with 65.3 percent and 80 percent, respectively.

Livestock ownership in the rural areas can be taken as one of the basic livelihood strategy and source of income either by selling the animal or its products. Hence, large number of livestock unit help the rural farmers to fulfill their food and non-food requirements. As shown in table 10

58 below, the average number of livestock ownership measured in TLU and oxen size per household were 3.3 and 0.9 respectively with standard deviation of 1.83 and 0.84. Moreover, maximum and minimum livestock and oxen holding were 10.1 and 3.3 respectively.

Proximity to nearest market had significant influence on poverty status of the household. In the study area the average distance that sampled households reach the nearest market is 7.2km with the longest distance of 24km and the smallest of 0.5km with standard deviation of 4.6km. Input utilization is also one important factor in rural farmers in which it helps to increase production and productivity. Input includes improved seed, organic and inorganic fertilizer, chemicals like pesticides and herbicides, and different agronomic practices. Following this, 62 percent of the sampled respondents were utilized at least one agricultural input notably inorganic fertilizer in 2017/18 production season.

Physical capital might have different compositions in poor and non-poor sampled household groups. As indicated in table 10 below, the average number of oxen owned by the poor and non-poor groups of sampled households were 0.5 and 1.3 respectively. There is statistically significant mean difference in oxen ownership between poor and non-poor sampled households at 1 percent probability level. Here, nearly 43 percent of the respondents in the study area did not have any oxen in the survey year. The average livestock ownership in TLU in the study area was 2.7 units for the poor and 3.8 units for non-poor households with standard deviation of 1.4 and 2.1 respectively. There was highly statistically significant mean difference observed between poor and non-poor households in terms of TLU at 1 percent probability level. Regarding to proximity to the nearest market center, the average distance of the poor farmers was 8.04km with standard deviation 4.59 whereas their non-poor counter households average distance was 6.51km with standard deviation of 4.38. The result showed that poor households are living far away from the market place than the non-poor households. Besides, significant mean difference was observed between poor and non-poor groups in terms of nearest market center at 5 percent probability level. Moreover, among input users 38 percent and 81 percent of poor and non-poor sample households utilized at least one of the agricultural input. Input utilization showed significant mean difference between the poor and non-poor households at 5 percent significance level.

Table 10: Physical capital of sample households with poverty status.

Physical capital Poor (N=84) Non-poor Total(N=190) t-value, (N=106) chi2 Mean St.Dev Mean St.Dev Mean St.Dev Number of oxen 0.5 0.7 1.3 0.96 0.9 0.84 6.54*** Livestock size (TLU) 2.7 1.36 3.8 3.03 3.3 1.83 4.21*** Nearest market (Km) 8.1 4.72 6.5 4.38 7.2 4.60 -2.28** Input utilization(1=yes)% 0.38 0.8 0.6 36.91** Source: Own survey result (2018) ***, ** and * significant at 1%, 5% and 10% significance level respectively

4.4.3. Financial Capitals

Financial capital comprises cash and non-cash availability and regular and non–regular inflows of money enabling people to adopt different livelihood strategies and hence achieve their livelihood strategies (Scoones, 1998). For this particular study the financial capitals include credit utilization, asset ownership, on farm income and off/non-farm income received in 2017/18 production year. As shown in table 13 below, the annual average credit received by respondent farmers were Birr 4408 with minimum. The maximum credit taken by respondent households was Birr 20,500. The main and most accessible source of credit is from Amhara Credit and Saving Institute (ACSI). Besides, from the whole sample population 63 percent were used credit either for input purchase or household food consumption. The average annual income obtained from on-farm income and off/non-farm income was Birr 6294.8 and Birr 6896. Asset possession also help farmers increase production and productivity. It includes different farm implements like sickle, saw, axe, small irrigation generator, hand cart and others. These assets were valued in monetary terms. Hence, on average, sample population households’ current asset value was Birr 1172 with standard deviation of 1110.

In order to better understand the financial utilization of the sample population the mean comparison was made to check its significance difference exists between poor and non-poor households. Accordingly, annual mean credit received in a year by the poor and non-poor households was found to be Birr 3148 and 5407 with standard deviation of 2886 and 5056 respectively. There is highly statistically significant mean difference between the two groups were observed at 1 percent probability level. The mean annual current asset value of poor rural

60 farmers was Birr 750 with standard deviation of 612 and their counter non-poor was Birr 1500 with standard deviation of 1304. Besides, there is highly statistically significant mean difference between poor and non-poor sample households at 1 percent probability level. Similarly, annual earnings obtained from sale of agricultural goods either crop yield or livestock and livestock products was Birr 3,783 while their counter non poor groups was Birr 6225 with highly statistically significant mean difference at 1 percent probability level.

Table 11: Financial capitals with poverty status (ETB per annum)

Financial capitals Poor (N= 84) Non-poor (N= 106) Total (N=190) t-value Mean St. Dev Mean St. Dev Mean St.Dev Credit Recived 3,148 2886 5407 5056 4408 4370 3.64*** Asset Value 750 612 1,500 1,304 1172 1116 4.88*** On-farm income 3,783 5,386 6,225 6,762 6294 5146 2.67*** Off/on-farm income 5,281 4,450 8,176 8,596 6896 7199 2.8*** Source: Own survey result (2018) , ***significant at 1% significance level respectively

4.4.4. Natural Capital

Natural capital is the quality and quantity of natural and environmental stock from which resources flows and services are useful for livelihood of rural farmers (Scoones, 1998). This study includes irrigation use and land holding as its natural capital. Accordingly, the average land holding in the study area was 1.04ha with a maximum area of 4ha, and 4.7 percent of sampled households did not have any type of land in the survey season. The other important natural capital included in this study was irrigation use. In the study area only 22 percent of the respondents have access and use of irrigation water implying that the majority were entirely depend on rain fed agriculture. Unavailability of irrigation water and shortage of land was identified as the major constraint of the irrigation sector in the study area.

Regarding to the poverty status of sampled households the average land holding of the poor household was 1.03 ha with standard deviation of 0.55 where as its counter non-poor was 1.04ha with standard deviation of 0.62. Poverty status and land holding did not show any significant differences between poor and non-poor households. Besides, nearly 14 percent of poor farmers had access and irrigated their land while 28 percent of non-poor farmers were also irrigated their land with statistical mean difference at 5 percent probability level.

Table 12: Natural capitals with poverty status

Natural capitals Poor (N = 84) Non-poor (N = 106) t-value, chi2 Mean St. Dev Mean St. Dev Own land holding 1.03 0.55 1.04 0.62 0.11 Irrigation use (1=yes)% 0.14 - 0.28 - 5.34** Source: Own survey result (2018), ** at 5% significance level

4.4.5. Other Institutional Capital or Characteristics

Institutional capitals can play significant impact in any society. It is to mean that institutions guide or arrange an individual or community to make life easier. The main motive of any government is to create conducive institutional arrangements that help economic agents can operate at their lowest transaction costs. These institutions might be governmental or non- governmental bodies. Besides, Institution may be formal or informal laws, norms or culture of any given society. In this study institutional capitals or services included were human health services, veterinary services and cooperative access. Following this, the majority of (79 percent) of sample population had access to veterinary services.

Regarding to the association, the was no significant mean difference between the poor and non- poor sample households in terms of Veterinary service. At the same time human health coverage in the study area was 72.6 percent. The human health service delivers mostly for women and babies on disease prevention and free delivery of drugs usually (contraceptives, antibiotics of TB, etc). There was significant mean difference observed in terms of human health service between poor and non-poor groups. Cooperative availability and membership can be considered as an important factor for the rural poor in providing agricultural inputs at the lowest transaction costs. Hence, in the study area 61.5 percent of respondent farmers had access to it. There was no significant association between the poverty status and cooperative membership in the study area. Usually most primary cooperatives in the study area provide inorganic fertilizer up on credit or cash.

4.5. Time Needed to Exit Poverty

Any developing country’s ultimate target or goal is eradicating poverty and enhancing the wellbeing of its citizens. To do so, achieving sustained economic growth could be considered as great importance. Despite the importance of economic growth, it will generally take more than just growth to rapidly improve the lives of the poor. Estimating expected time to exit poverty for those who are below the poverty line given positive economic growth rate is increasingly becoming popular these days for policy issues and poverty reduction interventions (Morduch, 1998).

Thus, targeted programs are needed to deliver benefits to the poor for instance in the form of improvements in their human and physical assets or through interventions that improve the returns they get from assets (WBI, 2005). Therefore, the concept of average time needed to exit poverty is central to lift majority of the poor households from poverty.

For policy makers, the average exit time for the poor households might sound more than the average exit time of the whole sample households. If the poverty exit time for the whole sample population is estimated, the conclusion might lead the policy makers neglecting to remember that many people are already poor. Hence, based on the national bank of Ethiopia, the average national real gross domestic product (RGDP) from 2005/06 up to 2009/10 (EFY) was used to estimate poverty exit time. The computed average RGDP for the five year was 7.7 percent.

Therefore, the estimated average time needed for poor sample households to exit poverty would be 3.35 years as shown in table 13 below. The is to mean that it requires 3.35 years so that the poor households to move out of poverty or at least bring them to the pre-determined poverty line given average per capita consumption expenditure of the poor households was minimum of Birr 3355 per annum and GDP per capita continue to register at least 7.7 percent per year (Appendixes 4). Besides, the average time required to bring the average poor household to the poverty line was 3.22 year. The additional exit time across the whole poor sample households and average poor sample household is 0.13 years. This is due to low consumption inequality amongst the poor households the additional exit time (0.13 year) looks insignificant. If the analysis was made for the whole sample population, the poor households in the study area were already non-poor before three months (back to the survey year). It is therefore, important

63 concept for identifying economic opportunities and challenges that poor people are benefiting and suffering. Besides, respective authorities can make use of necessary policy or strategy adjustments based on the estimated exit time. Previous findings showed that sustained economic growth can have a potential to reduce poverty. Research conducted by Tsegaye (2014) in Gozamen district of Amhara National Regional State found that the average exit time for the average poor household was four years.

Even though growth is unlikely to be uniform or constant for years across households or regions, the average exit time provides a simple and quick way for poverty alleviation through growth. Therefore, achieving broad based economic growth is important policy agenda, and developing analytical tools to consider poverty and economic growth jointly.

Table 13: Average time needed to exit poverty at 7.7 % growth rate of RGDP of the country

Poverty measure Estimated Value Average exit time of poverty for the average poor sample household 3.22 years Average exit time of poverty for poor sample households 3.35 year Additional year due to inequality amongst the poor 0.13 year Average GDP growth rate of the country* 7.7% Poverty line for the sampled households Birr 4301.85 Average per capita consumption expenditure of the poor sampled households Birr 3355 Source: Own survey result (2018), *Average five year RGDP (EFY2005/06—2009/10)

4.6. Econometric Model Results

To ascertain the effects of the explanatory variables related to poverty level of the households, Tobit regression model was employed. The dependent variable is the poverty level of the households, where the poor households were represented by poverty depth whereas non-poor households have zero as their dependent variable. The Tobit model assumes the dependent variable (poverty depth for poor and zero for non-poor) is censored and the variable takes on the these limiting values (zero and one) for a substantial number of respondents. Tobit model measures not only the probability of a household being poor but also the intensity of poverty level.

Following the results of Tobit model, the possible explanations for each significant explanatory variable are discussed below. Moreover, the three marginal effects after Tobit model namely conditional expectations on the latent variable (the probability of being poor), conditional expectation of the uncensored dependent variable (the intensity of poverty level) and the unconditional expected value of the dependent variable for the whole sample observation were interpreted.

Accordingly, sixteen explanatory variables were included for Tobit regression model that were assumed to influence the poverty level in the study area. Seven variables were found to be statistically significantly influenced poverty level of the households in the study area. These were number of livestock ownership in TLU, number of oxen ownership, total household size in AE, educational level of the household head in completed years, input utilization, asset ownership and credit utilization. Except household size, all other variables significantly and negatively influenced poverty in the study area. Household size was positively and statistically influenced poverty at 10 percent significance level. While, oxen ownership, credit utilization and use of improved agricultural input were found to be negatively and highly statsticaly influenced poverty at less than 1 percent significance level. Educational attainment of the household head in completed years was negatively and statistically influenced poverty level at less than 5 percent significance level. Value of asset and livestock ownership in TLU were also negatively and statistically influenced poverty at 10 percent probability level. In light of these summaries of Tobit model output, the possible explanations for each significant explanatory variable are briefly discussed below.

Educational level of household head (EDUCLV): Education is considered as one of the basic human capital that help individuals move out of poverty. According to this study, educational attainment of the household head found to have negative and significant influence on poverty at 5 percent significance level. The result implies that households who have household heads with relatively better education are less likely to be poor than those headed by uneducated (illiterate) household heads. The marginal effects, keeping other variables constant, showed that as the household head education level increases by one grade the probability of a household being poor would decrease by 4.7 percent while it decreases the intensity or expected value of poverty by 0.7 percent and 0.9 percent for the poor households and for the whole observation,

65 respectively. This might be the fact that educated household heads have in a better position to adopt improved and best bet agricultural technologies than less educated or uneducated ones. This further raises the productivity, efficiency and income of the educated heads with subsequent improvement of their living condition. The result is similar with the previous findings of Adane (2002) and Tasew and Teke (2002).

Oxen Ownership (OXOWN): As expected, the number of oxen ownership was found to negatively and significantly influenced poverty level at less than 1 percent significance level. The negative sign indicates that households with large number of oxen have less likely to be poor compared to households with less number of oxen holding. The negative sign indicates that households with higher number of oxen are less likely to be poor than households with few or no oxen. The marginal effects of oxen ownership revealed that the probability of a household being poor tends to decrease by 19.2 percent for every addition of ox possessed by the household, whereas it decreases the intensity of poverty by 2.7 percent and 3.6 percent for the poor households and the whole sample observation, respectively. The possible reason could be; rural households are largely dependent on oxen for draught purpose there by plowing either their land or may enter for sharecropping arrangements with other households who have cultivable land holding but no oxen. Accordingly, those who have oxen can produce more food and income for his family. The result is consistent with Ayalneh et al. (2005).

Number of Livestock Ownership (TLU): Number of livestock ownership in TLU in rural areas is considered as one of the basic livelihood assets. As hypothesized, number of livestock owned measured in TLU was found to negatively and significantly contribute to the level of poverty at 10 percent significance level. Keeping other variables constant, the marginal effect of TLU indicates that increasing the number of livestock by one TLU the probability of a household being poor decreases by 4.4 percent while it decreases the intensity of poverty level by 0.7 and 0.9 percent for the poor households and for the whole observation, respectively. The result revealed that livestock do have an important asset for the majority of rural households. It served as an immediate source of income by selling the lives and/or its product to fulfill food and non-food requirements of the household. The finding is similar with that of Degye (2013).

Household size (HHSZ): Household size is a demographic variable measured in adult equivalent. As expected, household size was found to significant and positively related with

66 poverty level at 10 percent significance level. The positive relationship shows that as household size increases the probability of a household being poor would increase. Considering the marginal effect, as the member of household increases by one adult equivalent the probability of a household being poor would increase by 3.8 percent while it increases the intensity of poverty by 0.5 and 0.7 percent for the poor households and for the whole observation of the study, respectively. The possible reasons might be when most members of the households are dependents due to existing high rate of unemployment and less job opportunities in rural areas, an additional household member shares the limited resource that lead the household to become poor. This result has been supported with the findings of Derje and Eric (2013.

Credit Utilization (CRUTL): As expected, credit utilization of a household was found to significant and negatively influenced poverty at less than 1 percent significance level. The negative relationship shows that the probability of a household being poor decreases as a household receives credit. The marginal effect showed that an increase in the amount of credit by one thousand ET birr, the probability of a household being poor decreases by 3.2 percent while it decreases the intensity of poverty by 0.4 percent and 0.6 percent for the poor ones and for the whole sample observation of the study, respectively. The possible explanation is that credit plays vital role when cash constraint happened either to finance farm input and/or purchase other immediate food and non-food basic requirements. Besides, credit helps rural farmers to involve in long term income generating activities that ultimately help them move out of poverty trap. The result is consistent the findings of Ayalneh and Shimles (2009).

Asset Ownership (ASTOWN): As expected, value of asset ownership available in the household was significantly and negatively related with poverty level at 10 percent significant level. Considering the marginal effects, as the value of household’s current asset increases by one thousand Birr the probability of a household being poor would decrease by 10.1 percent, while it decreases the intensity of poverty decreases by 1.4 percent and 1.9 for the poor households and the whole sample observation of the study, respectively. The possible could be households having larger asset have the capacity to withstand economic shocks and income shortfalls. Hence, they can finance their household needs be it farm or household goods that ultimately help farmers increase their production and productivity. Accordingly, ownership of household assets helps farmers as a fallback strategy against shocks because some of the assets

67 could be sold to support households cash requirement. The finding is similar with Girma (2012) and Asogwa et al. (2012).

Input Utilization (UTLINP): The study result suggests that households who used improved agricultural input was found to significant and negatively influenced poverty at less than 1 percent significant level. The negative relationship indicates, a household who used improved agricultural input are less likely to be poor than those who did not. The marginal effects showed that those households who used improved agricultural input, the probability of being poor would decrease by 32.1 percent, while it decreases the intensity of poverty by 4.5 percent and 6 percent for the poor ones and the whole observation of the study, respectively. The result agrees with the previous findings of Eshetu (2011).

Table 14: Tobit model regression estimates and marginal effects of poverty determinants

Independent Variables Coefficient Std. Err Marginal Effects P * / x  Pry  0/ x EP / x

x x x Number of livestock owned (TLU) -0.020* 0.011 -0.047 -0.007 -0.009 Number of oxen own -0.081*** 0.021 -0.192 -0.027 -0.036 Household size(Adult Equivalent) 0.016* 0.008 0.038 0.005 0.007 Age of the household head 0.003 0.002 0.007 0.001 0.001 Sex of household head 0.053 0.056 0.127 0.018 0.024 Education level of household head -0.019** 0.008 -0.047 -0.07 -0.009 Distance from the nearest market 0.002 0.004 0.004 0.001 0.001 Land size owned (ha) 0.006 0.032 0.014 0.002 0.003 Frequency of extension contact 0.003 0.009 0.007 0.001 0.001 Utilization of irrigation 0.021 0.044 0.049 0.007 0.009 Utilization of input -0.014*** 0.034 -0.321 -0.045 -0.059 Value of asset owned (ETB) -0.043* 0.025 -0.101 -0.014 -0.060 Dependency ratio 0.032 0.028 0.077 0.011 0.014 Non/off-farm income (ETB) -0.004 0.005 -0.009 -0.001 -0.002 Credit utilization (ETB) -0.014*** 0.005 -0.032 -0.004 -0.006 Cooperative memebrship 0.009 0.034 0.023 0.003 0.004 Constant 0.012 0.118 _se 0.167 0.014 Observation summary Number of observation = 190 Log chi 2 (16) = -101.24 Left censored observations =106 Prob > chi = 0.000 Uncesnored observations = 84 Pseudo R2 = 0.0000 Log likelihood = -142.04363

Source: OwnSurvey result (2018) ***, ** and * Significant at 1%, 5% and 10% significance level respectively

5. SUMMARY, CONCLUSSION AND RECOMMENDATIONS

This chapter presents the summary, conclusions and policy implications emanating from the present study.

5.1. Summary

This study was designed to investigates the household level poverty situation and determinant factors of poverty in rural areas of Banja district Amhara National Regional State, Ethiopia. In order to investigate poverty situation of the study area, the researcher employed the cost of basic needs approach. Accordingly, the total poverty line that demarcates poor households from non- poor households was Birr 4301 per adult per year. The food poverty line and non-food poverty line was Birr 3232 and 1069 respectively per adult per year. FGT class of poverty measure were 0.44, 0.09 and 0.02 for poverty head count, poverty gap and poverty severity, respectively.

Based on the survey result, an attempt was made to describe whether there exists significant mean difference was observed between the poor and non-poor sample respondents regarding to annual mean consumption expenditures. Accordingly, there is significant mean difference between poor and non-poor sample households in terms of annual total consumption expenditure, annual food expenditure and non-food expenditure per adult per year.

Significant mean difference was observed between poor and non-poor sample households in terms of educational attainment of the household head, oxen ownership, TLU, input utilization, asset ownership, on/off farm income, distance to the nearest market and credit utilization. Poor households have larger family size in AE than counter non-poor households. Besides, poor household head had low level of educational attainment than their counter non-poor household heads.

Estimating the average time needed to exit the poor households from poverty in the study area was estimated to be 3.35 years. In other words, it takes 3.35 years to push the poor households in to poverty line given GDP per capita continues to grow at minimum rate of 7.7 percent per annum. Besides, the average exit time for the average poor farmer was 3.23 years.

Econometric results of Tobit regression model employed sixteen explanatory variables. Accordingly, seven variables were found to significantly influence rural poverty in the study area. These were livestock ownership in TLU, number of oxen ownership, total family size in AE, educational attainment of the household head, input utilization, value of asset ownership and credit utilization significantly influenced poverty.

5.2. Conclusion and Policy Recommendations

Based on the findings of this study the following conclusions were made. The proportion of people living below poverty line still remains higher compared to the national and regional rural head count index of 25.6 percent and 28.8 percent, respectively. The result revealed that the average poverty exit time by assuming sustained positive economic growth can lift the poor farmers in nearly three and half years. To do so pro-poor and area specific policy intervention given the consumption pattern of the poor households help them move out of poverty trap. Therefore, sustainable economic growth at the national and regional level is needed so that considerable number of households will be non-poor in the estimated time period. Educational attainment of the household head, livestock ownership, family size, input utilization, asset ownership and credit utilization was found to be an important poverty determinant.

Household size which is an important component of demographic character showed positive contribution for rural poverty. This has direct implication on the income and resources available in the household which worsen the poverty situation of the household. Therefore, family planning through awareness creation and integrated health programs, creating job opportunities for the rural households may lead to acceptable number of children. Livestock have great importance for the majority of households in rural areas. Hence, the livestock sector should be strengthened through the provision of veterinary services and feed supply. Besides, intervention projects that enhance the livestock sector like dairy cow credit, sheep credit and fattening credit need to be supported with the necessary husbandry skill and knowledge.

Educational attainment of the household head was highly related with rural poverty in the study district. The relevant authorities should develop educational program through promoting and expanding of schools in the nearby rural villages as well as strengthen adult education would improve the living standard of the rural poor households.

Utilization of improved agricultural input remarkably improve production and productivity there by increase household income. Therefore, access and provision of improved agricultural input, best bet farming technology and management practices would help the farmers to increase production and productivity. Another important poverty determinant in the study area was asset ownership. Accordingly, relevant government authorities or development actors have to design and implement rural household asset building program that enhance the production and productivity of farmers and there by improve their lives.

The last but not the least important poverty determinant in the study area was credit service. Credit can create an opportunity to be involved in economic activity that generates revenue for the rural households. It also supports farm households in solving cash constraints to either to buy food for family or to buy farm input. Thus government bodies, non-governmental organizations and other relevant authorities shall focus on enhancing and expanding rural credit services for poor rural households at minimum possible transaction and/or collateral cost.

6. REFERENCES

Abebe Zewdu. 2011. Dimensions and Determinants of Poverty among Rural Household. The Case of Itang Special District of Gambela Region. M.A Thesis, Addis Ababa University, Addis Ababa, Ethiopia.

Adane Nabso. 2002. Magnitude and Determinants of Rural Household Poverty in Central Ethiopia: The case of Berehe Aleltu District. M.Sc Thesis, Haramya. University, Hramaya, Ethiopia.

Adugna Eneyew and Sileshi Mengistu. 2013. Determinants of Poverty in Agro-pastoral Societies of Southern Ethiopia. Livestock Research for Rural Development, 25(20).

Agerie Nega. 2016. Determinats of Smallholder Rural Farm Households’Participation in Small Scale Irrigation and Its Effect on Income in North Gonder Zone of Amhara Region, Ethiopia. International Journal of Innovative Research and Development, 5(6): 592- 601.

Ahmed Mohamed. 2013. Determinants of Poverty in Rural Ethiopia: A Household Level Analysis. MSC Thesis, Lund University, Lund, Sweden.

Akerele, D. and Adewuyi, S. 2011. Analysis of Poverty Profiles and Socioeconomic Determinants of Welfare among Urban Households of Ekiti State, Nigeria. Current Journal of Social Science, 3(1): 1-7.

Alemayehu Geda, Abebe Shimles and Daniel Zerfu. 2006. Finance and Poverty in Ethiopia: A household Level Analysis. United Nations University, World Institute for Development Economics Research, Research paper No.2006/51.

Ali, I., Abdul, S., Sarfraz, A. and Mustafa, 2010. Identifying the Pathways Out of Poverty: Evidence of Exit Time Poverty Estimations in Pakistan; PMAS (Pri Mehr Ali Shah) Arid Agriculture University Rawalpindi, Pakistan. Pakistan Journal of Life and Social Sciences, 8(1): 24-29.

Alkire, S. 2002. Valuing Freedoms: Sen’s Capability Approach and Poverty R eduction. Oxford University Press. New York.

Amemiya, T. 1975. Qualitative Methods: Annals of Economic and Social Measures, 1(2): 363- 372.

Apata, T.G., Apata, O.M. and Igbalajobi, O.A. 2010. Determinants of Rural Poverty in Nigeria: Evidence from Smallholder Farmers in South-western Nigeria. Journal of Science and Technology Education Research, 1(4): 85- 91.

Appleton, S. 2004. The Rich are Just like us, only Riches: Poverty Function or Consumption Function. Centre for studies of African Economies. Oxford University Press, Oxford. Journal of African Economies, 4 (10): 433- 469.

Asogwa, B., Ummeh, J. and Okwoche, V. 2012. Estimating the Determinants of Poverty Depth among the Peri-Urban Farmers in Nigeria. Current Research Journal of Social Sciences, 4(3): 201-206.

Ayalneh Bogale and Abebaw Shemles . 2009. Household Level Determinants of Food Insecurity in Rural Areas of Dire Dawa, Eastern Ethiopia. African Journal of Food, Agriculture, Nutrition and Development, 9(9): 1914-1926.

Ayalneh Bogale, Konard, H. and Benedikt, K. 2005. Determinants of Poverty in Rural Ethiopia. Quarterly Journal of International Agriculture, 44(44): 101-120.

Ayalneh Bogale. 2002. Land Degradation, Improvement and Livelihood Strategies of Rural Households in Ethiopia: Farmers Perception and Policy Implications. Doctoral Dissertation, Humboldt Universty, Berlin, Germany.

Baker, J. 2000. Evaluating the Impact of Development Projects on Poverty: A Hand book for practitioner, World Bank Press, 225.

Bradshaw, T. K. 2007. Theories of Poverty and Anti-Poverty Programs in Community Development. Journal of the Community Development Society, 38(1): 7-25.

CSA (Central Statistics Authority). 2007. Intercensal Population Survey, Addis Ababa, Ethiopia.

Damisa, M., Sanni, S., Abdoulaye, T., Kamara,A and Ayanwale, A. 2011. Household Typology Based Analysis of Livelihood Strategies and Poverty Status in the Sudan Savannah of Nigeria. Journal of Agriculture and Environment Studies, 2(1): 146 – 160.

Datt, G. and Jolliffe, D. 2005. Poverty in Egypt: Modeling and Policy Simulations. Economic Development and Cultural Change, 53(2): 327-346.

Dawit Alemu, Woldeamlak Bewket, Gete Zeleke, Yemisrach Assefa. and Peter, T. 2011. Extent and Determinants of Household Poverty in Rural Ethiopia: A Study of Six

Villages. Eastern Africa Social Sciences Research, 27(2): 21-49.

Degye Goshu. 2013. The Dynamics of Poverty and Vulnerability in Rural Ethiopia. Ethiopian Journal of Economics, 22(2):1-20

Denano Tesfaye and Tekle Getachew. 2018. Determinants of Rural Households’ Poverty with Selected Link Functions: The case of Soro District, Hadiya Zone, Sothern Nations Nationalities and People Regional States, Ethiopia. International Journal of Economics and Managmeent Science, 7(1): 1-6

Dercon, S. and Krishnan, P. 2000. Vulnerability, Seasonality and Poverty in Ethiopia. Journal of Development Studies, 36(6): 82-100.

Dercon, S. and Krishnan, T. 1998. Changes in Poverty in Rural Ethiopia from 1989-1995: Measurement, Robustness Tests and Decomposition. Centre for the Study of African Economies, Oxford University. Working Paper Series, 98:7.

Dereje Haile and Eric, N. 2015. Determinants of Poverty in Gebi-Resu Pastoralists Area of Afar Region, Ethiopia. Journal of Poverty, Investment and Development, 17: 56-72.

EHNRI (Ethiopian Health and Nutrition Research Institute), 2000. Food Composition Table for Use in Ethiopia. Addis Ababa. Ethiopia.

Ejigayehu Sisay and Abdi Kehali. 2013. Determinants of Food Insecurity in Addis Ababa City, Ethiopia. Fourth International Conference of Afican Association of Agriccultral Economists, 22-25, 2013. Hammamet, Tunisia.

Eshetu Sied and Adam Bekele. 2011. Determinants of Farm Household Poverty: The Case of Lay Gaint District, South Gonder Zone, Amhara Regional State. M.Sc Thesis, Haramaya University, Haramaya, Ethiopia.

Esubalew Alehegn. 2006. Determinants of Urban Poverty in Debre Markos, Ethiopia: Household Level Analysis. M.A Thesis, Addis Ababa University, Addis Ababa, Ethiopia.

Etim, N., Edet, G. and Esu, B. 2009. Determinants of Poverty Among Peri-urban Telfera Occidental Farmers in Uyo, Nigeria. Journal of Agricultural and Social Sciences, 5(12): 49-51. Fageras, S. and Lindsay, W. (2007). Determinants of Poverty in Sierra Leone. Economic and Statistics Analysis Unit. Overseas Development Institute. Working Paper 19, London, UK.

FAO (Food and Agricultural Organization). 2015. The State of Food and Agriculture: Social Protection and Agriculture; Breaking the cycle of Rural poverty. Rome, Italy.

Feder, G., Just R.R. and Zilberman, D. 1985. Adoption of Agricultural Innovation in Developing Countries. World Bank Staff Working Paper 542, Washington D.C, USA.

Fields, G.S. 1994. Data for Measuring Poverty and Inequality Changes In the Developing Countries. Journal of Development Economics, 44(1): 87-102.

Fitsum, T. 2002. Poverty in Addis Ababa: A Comparison of Female and Male Headed Households. M.A Thesis, Addis Ababa University, Addis Ababa, Ethiopia.

Foster, J. 1998. Absolute versus Relative Poverty: American Economic Review, Papers and Proceedings, 88: 335-341.

Garza, J. R., 2000. The Determinants of Poverty in Mexico. Doctoral Dissertation,

Girma Gezimu. 2012. Determinants of Food Insecurity Among Households in Addis Ababa City, Ethiopia. Journal of Interdisciplinary Description of Complex Systems, 10(2): 159-173.

Grant, U. and Marcus, R. 2009. Chronic Poverty and PRSP: Desk Study, Background Paper for the Chronic Poverty Report.Chronic Poverty Research Centre.

Greene, W. 1997. Econometric Analysis, 3rd ed. New York: Macmillan. Greene, W. 2002. Econometric Analysis, Fifth Edition. New York University, NewYork USA.

Greer, J. and Thorbecke, E. 1986. A Methodology for Measuring Food Poverty Applied to Kenya. Journal of Development Economics, 24(1): 59- 74.

Gujarati, D. 1995. Basic Econometrics: McGraw-Hill, Inc. 3rd edition. New York.

Husmann, C. 2016. Marginality as a Root Cause of Poverty: Identifuying Marginality Hotspots in Ethiopia.World Development, 78 (C): 420-435.

Hussain, I. (2003). Alleviating Rural Poverty in Pakistan. A Dawn Group of newspapers.

IFAD (International Fund for Agricultural Development). 2011. Rural Poverty Report: New Realities, New Challenges, New Opportunities for Tomorrow’s Generation.

IMF (International Monetary Fund). 2006. Th Federal Democratic Republic of Ethiopia: Poverty Reduction Strategy Paper: Annual progress Report.

Indris Siraje. 2012. Assessment of Food Insecurity, its Determinants and Coping Mechanisms Among Pastoral Households. The Case of Zone One Chifra District, Afar National Regional State, Ethiopia. M.Sc Thesis, Haramaya University, Haramaya, Ethiopia.

Johnston, J. and Dandiro, J. 1997. Econometrics Methods, fourth Edition, New York: McGraw Hill Companies, Inc.

Kanbur, S. 1987. Measurment and Allevation of Povertywith an Application to the Effects of Macroeconomic Adjusmeents. IMF Staff paper, 34(1): 60- 85.

Kawani, N. 1993. Measuring Poverty. Definitions and Significance Tests with Application to Cote d’ivore.

Kwapong, N. and Hanisch, M. 2013. Cooperatives and Poverty Reduction: A Literature Review. Journal of Rural Cooperation, 41(2): 114- 146. Leonard, B. 1988. Individual and Structural Explanation of Poverty. Population Research and Policy Review, 7(3): 201-222.

Maddala, G. 1993. Limited Dependent and Qualitative Variables in Econometrics. Cambridge University Press. Cambridge.

McDonald, J. and Moffitt, R. 1980. The Use of Tobit Analysis. The Review of Economic and Statistics, 62(2): 318 -321.

MoFED (Ministry of Finance and Economic Development). 2012. Assessing Progress towards the Millennium Development Goals. MDGs Report, Addis Ababa, Ethiopia.

MoFED (Ministry of Finance and Economic Development). 2013. Development and Poverty in Ethiopia 1995/96-2010/11. Addis Ababa, Ethiopia.

Morduch, J. 1998. Poverty, Economic Growth, and Average Exit Time. Economics Letters, 59(3): 385-390.

Muller, C. 2006. Defining Poverty Lines as a Fraction of Central Tendency. Southern Economic Journal, 72(3): 720-729.

NBE (National Bank of Ethiopia). 2017. Annual Report, Addis Ababa, Ethiopia.

Nega Afera and Sudhakara, B. 2015. Determinants of Poverty in Rural Tigray, Ethiopia: Evidence from Rural Households of Gulomekeda Wereda. International Journal of Science and Research, ISSN: 2319-7064, 4(3): 822-828.

NPC ( National Planning Commision). 2017. Ethiopias Progress Towards Eradicting Poverty: An intrim Report on 2015/16 Poverty Analysis Study. Addis Ababa, Ethiopia.

Raghbendra J., Gaiha, R. and Sharma, A. 2009. Calorie and Micronutrient Deprivation and Poverty Nutrition Traps in Rural India. World Development, 37(5): 982-991.

Rainwater, L.1970. Neutralizing the Disinherited: Some Psychological Aspects of Understanding the Poor. Psychological Factors in Poverty, 9-28.

Ravallion, M. 1992. Poverty Comparisons: A Guide to Concepts and Methods. Living Standard Measurement Study. Working Paper, No, 88. Washington, D.C, USA.

Ravallion, M. and Bidani, B. 1994. How Robust is a Poverty Profile? World Bank Economic Review, 8(1): 75-102.

Ryan, W. 1976. Blaming the Victim. New York: Vintage.

Sallila, S. and Hilamo, H. 2004. Rethinking Relative Measures of Poverty. Maxwell School of Citizenship and Public Affairs. Working paper No3 68. Syracuse University of New York.

Scoones, I. 1998. Sustainable Rural Livelihoods: A Framework for Analysis. International Development Studies Working Paper.

SDSN (Sustainable Development Solutions Network). 2012. Global Profile of Extreme Poverty. London, Britain.

Semere Gebretsadik. 2008. Dimensions and Determinants of Rural Household Poverty: The Case of Eastern Zone, Tigray Regional State, Ethiopia. M.Sc Thesis, Haramaya University, Haramaya, Ethiopia.

Sen, A. 1981. Poverty and Famines: An Essay on Entitlement and Deprivation. Oxford, Clarendon Press.

Sen, A. 1983. Poor, Relatively Speaking. Oxford Economic Papers, 35(2): 153-169.

Sian, M. 2014. Essay on Poverty and Wellbeing. Doctoral Disertation, Stirling University, Stirling, United Kingdom.

Simler, K., Mukherjee, S., Dava, G. and Datt, G. 2004. Rebuilding After the War: Micro-level Determinants of Poverty Reduction in Mozambique. International Food Policy Research Institute, Research Report, Washington D.C, USA.

Storck, H., Berhanu Adnew, Bezabeh Emana and Begander R. 1991. Management Strategies for Farming Systems in an Uncertain Environment and Approaches for Their Improvement. Farming Systems and Resource Economics in Tropics, Vol 27, Hannover, Germany.

Tassew Woldehana, Hoddinott, J. and Dercon S. 2008. Poverty and Income Inequality in Ethiopia: 1995/96 – 2004/05, Addis Ababa, Ethiopia.

Tassew Woldehana. 2004. Pro-poor Budgeting and the Role of Parliament in the Implementation of Poverty Reduction Strategy in Ethiopia. Addis Ababa, Ethiopia.

Tassew Woldehanna and Tekie Alemu. 2002. Profiles and Determinants of Poverty. Paper Presented at an International Seminar of Development Strategies for Less Favored Areas. Wageningen, Netherlands.

Tesfaye Gedefaw, Fredu Nega and Manasebo Gebru. 2017. Rural Household’s Poverty and Vulnerability in Amhara Region, Ethiopia: Case Study in Gubalafeto Woreda. Journal of Poverty, Investment and Development, 39: 22-34.

Teshome Kebede and Sharma, M.K. 2014. Determinants of Poverty in Ethiopia. Ethiopia Journal of Economics, 13(1): 114 -130.

Tobin, J. 1958. Estimation of Relationships for Limited Dependent Variables, Econometrcia 26: 24-36.

Tsegaye Molla, 2014. Analysis of Rural Poverty and Exit Time: The Case of Gozamen District of East Gojiam Zone, Ethiopia. Journal of Economics and Devlopment, 5(27): 310-320.

UNDP (United Nations Development Program). 2016. Human Development Report. Human Development for everyone, New York, USA.

WB (World Bank). 2000. Panama Poverty Assessment: Priorities and Strategies for Poverty Reduction. Working Paper 237 , Washington, USA.

WB (World Bank). 2001. World Development Report 2000/01: Attacking Poverty. New York, Oxford University Press.

WB (World Bank). 2007. World Development Report, World Bank, Washington D.C, USA.

WB (World Bank). 2014. Ethiopia Poverty Assessment: Poverty Global Practice Africa Region.

WB (World Bank). 2015. Policy Research Working Paper, Washington D.C. USA.

WB (World Bank). 2016. Priorities for Ending Extreme Poverty and Promoting Shared Facility: Systemic Country Diagnostic.World Bank, Washington D.C, USA.

WBI (World Bank Institute). 2005. Introduction to Poverty Analysis: Poverty Manual, All, JH Revision.

Weber, B. and Jensen, L. 2004. Poverty and Place: A Critical Review of Rural Poverty Literature. Oregon State University. Rural Poverty Research Center, Working Paper 04- 03.

Wooldridge, J. 2002. Econometric Analysis of Cross Section and Panel Data. Cambridge: Massachusetts Institute of Technology Press.

Xinshen, D. 2010. Economic importance of Agriculture for Sustainable Development and Poverty Reduction and Poverty Reduction: The Case of Ethiopia. Global Forum on Agriculture, Paris, France.

Yamane, T.1967. Statistics, An Introductory Analysis, 2nd Edition. New York.

Yonas Tesfamariam. 2013. Determinants of Poverty of Zoba maekel of Eriteria: A Household Level Analysis. International Journal of Food and Agricultural Economics, 1(2):73-84.

Zegeye Paulos. 2017. Determinants of Poverty in Rural households: The Case of Damote Gale District, Wolita Zone, Ethiopia. A household Level Analysis. International Journal of African and Asian Studies, 29: 68-75.

Zheng, B. 1993. An Axiomatic Characterization of the Watts Poverty Index. Economics Letters, 42(1): 81-86.

7. APPENDEXES

Appendix 1: Conversion Factor for kilocalories per Gram of different food items

Food types Unit Kilocalorie Teff Gram 3.589 Wheat Gram 3.623 Sorghum Gram 3.805 Maize Gram 3.751 Haricot bean Gram 3.451 Potato Gram 0.87 Sweet potato Gram 1.36 Onion Gram 0.713 Meat Gram 1.148 Milk Gram 0.737 Egg Gram 0.061 Butter Gram 7.364 Oil Gram 8.964 Cabgeges Gram 0.37 Source: Ethiopian Health and Nutrition Research Institute (2000)

Appendix 2: Conversion Factor for Tropical Livestock Unit (TLU) Livestock type Conversion factor Oxen 1.1 Cow 1.0 Heifer 0.5 Bull 0.6 calves 0.2 Shoat (Adult) 0.13 Shoat(young) 0.06 Donkey 0.5 Horse 1.1 Mule 0.7 Poultry 0.013 Source: Storck et al. (1998)

Appendix 3: Conversion Factor for Adult Equivalent (AE)

Age category in years Male Female <10 0.6 0.60 10 - 13 0.9 0.80 14 - 16 1.0 0.75 17 - 50 1.0 0.75 >50 1.0 0.75

Appendix 4: National Real Gross Domestic Product in million Birr EFY 2005/6-2009/10 Ethiopian Fiscal 2005/06 2006/07 2007/08 2008/09 2009/10 Average Year (EFY) Real GDP 93,474 104,499 116,190 127,844 141,187 116,639 Growth in RGDP 11.5 11.8 11.2 10 10.4 10.98 RGDP per capita 1441 1553 1664 1764 1953 1675 Growth in RGDP per capita 8 7.8 7.1 6 9.6 7.7 Source: National Bank of Ethiopia (EFY 2010)

Appendix 5: Tobit regression coefficient and Marginal effects after Tobit dtobit2 depth LIVSTOWN OXOWN HHSZ HAGE HSEX EDULVL DSMinKM LNDHLD FREXTN UTLIRG UTLINP ASSET depratio lnOff CRTUTZ Coop_member, ll(0)

Tobit estimates Number of obs = 190

LR chi2(16) = 116.80

Prob > chi2 = 0.0000

Log likelihood = -15.147885 Pseudo R2 = 0.7940

------

depth | Coef. Std. Err. t P>|t| [95% Conf. Interval]

------+------

LIVSTOWN | -.0200816 .0106502 -1.89 0.061 -.0411018 .0009386

OXOWN | -.081182 .0209198 -3.88 0.000 -.1224712 -.0398928

HHSZ | .0161259 .0083257 1.94 0.054 -.0003064 .0325582

HAGE | .0029383 .0020177 1.46 0.147 -.001044 .0069207

HSEX | .0536616 .0555814 0.97 0.336 -.056039 .1633622

EDULVL | -.0198399 .0080521 -2.46 0.015 -.0357323 -.0039474

DSMinKM | .0018485 .0038509 0.48 0.632 -.005752 .0094491

LNDHLD | .0060675 .0317601 0.19 0.849 -.056617 .0687521

FREXTN | .0030244 .0086698 0.35 0.728 -.0140871 .020136

UTLIRG | .0207214 .0438248 0.47 0.637 -.0657752 .1072179

UTLINP | -.1356112 .0344791 -3.93 0.000 -.2036624 -.06756

ASSET | -.0426515 .0247487 -1.72 0.087 -.0914978 .0061948

depratio | .0326609 .0281326 1.16 0.247 -.0228642 .0881859

lnOff | -.004104 .0049944 -0.82 0.412 -.0139614 .0057535

CRTUTZ | -.0136091 .0049716 -2.74 0.007 -.0234214 -.0037967

Coop_member | .0099718 .0337815 0.30 0.768 -.0567024 .076646

_cons | .0119266 .1178369 0.10 0.919 -.2206471 .2445004

------+------_se | .1669135 .0141616 (Ancillary parameter) ------Obs. summary: 106 left-censored observations at depth<=0 84 uncensored observations

| Marginal Effects at Observed Censoring Rate

|------

| Unconditional Conditional on Probability

Name | Expected Value being Uncensored Uncensored

------+------

LIVSTOWN | -.00887819 -.0066841 -.04749108

OXOWN | -.035891 -.02702117 -.19198752

HHSZ | .00712935 .00536746 .03813622

HAGE | .00129906 .00097802 .00694889

HSEX | .02372409 .0178611 .12690448

EDULVL | -.0087713 -.00660363 -.0469193

DSMinKM | .00081725 .00061528 .0043716

LNDHLD | .00268248 .00201955 .01434909

FREXTN | .00133712 .00100667 .00715249

UTLIRG | .00916102 .00689703 .04900396

UTLINP | -.05995441 -.04513774 -.32070712

ASSET | -.01885647 -.01419643 -.10086672 depratio | .01443955 .01087107 .07723978

lnOff | -.00181438 -.00136599 -.00970545

CRTUTZ | -.00601663 -.00452973 -.03218407

Coop_member| .00440858 .00331908 .0235823

------+------

Appendixe 6: Survey Questioner

Introduction This questionnaire is prepared to undertake a study on the Determinants of Rural Poverty in Banja District of Awi Zone Amhara National Regional State, Ethiopia. The purpose of the questionnaire is to gather information on household’s socio-economic, agricultural and non- agricultural activities, access for services and other important information. Dear respondents, the result of this study will help different stakeholders and policy makers to make appropriate measures on Poverty reduction and welfare of the society. The information you provide remain confidential and will only be used for research purpose. Therefore, you are kindly requested to provide genuine responses. Thank you for your time and cooperation!

ID Information 1. Woreda/Distric Banja Kebele/PA ______3. Enumerator’s name______

5. Date of Interview ______6. Signature______

PART I: Household Capital

1.1.Basic Household Information

No Name of Household Relationship Age Sex Marital Educational Reason for Drop members to the Status Status out if any household Grade head 00 A B C D E F 1 2 3 4 5 6 7 8 9 Code A. Head =01, wife/husband =02, Son/daughter =03, Mother/father = 04, Brother/sister=05 , Grandparents = 06, Relative = 07, not relative = 08 Code C: Male= 1, Female =0 Code D: Married = 01, Single =02, Divorced =03, Widowed = 04 Code E: Illiterate = 0, Read and write = 1, Write the grade of formal education

Code F: No access to school= 01, Lack of money = 02, Do not want/ no interest= 03. To help family through herding animals = 04, Old age = 05. Lack of awareness =6 1.2.Total Family Size ______Number of family members below age of 14 ______M______F______Number of family members between 15 & 64______M______F______Number of family members above 64______M______F______Part 2. Land Ownership

Here, the interviewer would like to ask about the land Owned by a household member, attention is given to own land. For further use, sharecropped in or rented in from any other will be included for data collection

2.1 Do you have your own land? 1= Yes, No =2 2.2 If yes, Land size in timad ______

2.3 If question 2.1 is Yes, did you cultivate it in 2016/17 production season? 1= Yes, No

3. Please identify your production 1. Crop production plot Type of crop Plot size in Land Type How much was the Have you Have you If you sold any amount ; no grown last Timad (Code B) harvest given any part Sold any year (Code A) of the harvest amount of to others as your harvest payment

Code (A) Quantity Unit Quant Unit Yes=1, 2= Quantity Unit Amount in ity No Birr 1 2 3 4 5 6 7 8 9 10 Code A. 6. Beans Code B. 1. Teff 7. Peas 1.Own land 2. Barly 8. Noug9.Oil 2. Sharecropped out 3. Wheat seed 3.Rented in 4. Fingermillet 10.Uncultivated 4. Sharecropped in 5. Maize land

2. Fruit, Vegetables and Wood lot production Crope type Type of Plot size in Land Type How much the Have you given Have you If you sold any amount crop timad (Code A) harvest any part of the Sold any grown harvest to others amount of last year as payment your harvest Vegetable Quantity Unit Quantity Unit Yes=1, 2= Quantity Unit( Amount in (Kg) No Kg) Birr Potato Onion Cabage Pepper Selata, Kosta Others Fruit Apple coffe Lemon Orange Banana Avocado Wood lots Eucalptus Gesho Khat Others Code A 1. Own land 2. Sharecropped out 3. Rented in 4. Sharecropped in

3. Did you rent out your land in the last main production season? Yes= 1, No=2, if yes, land size______, amount of money you received? ______Birr. Reason for rent out was the reason ______4. If you cultivated through rented in, land size ______timad, Amount of money / you paid in the last season? ______Birr 5. Have you sharecropped out your land in the last main production season? Yes=1, No=2, 6. If Q5 is yes, how much income you get from sharecropped out? ______Birr 7. Do you utilize all your family labor for farm activities? 1= Yes, 2 =No 8. If no, where do surplus labors go? Go to neighbor for labor selling= 1 Go to town for labor selling = 2, Stay at village for nothing= 3 Move out of their village in search of job = 5 , Others (specify)______How many days do any of your family stay out of his village ?______

Part 3. Access and Utilization of Irrigation, Agricultural inputs, Credit service and Extension service 3.1. Irrigation Utilization 1. Is there any irrigation project or irrigable water sources in your community? Yes=1, No =2. 2. If yes do you or your household member has irrigable land? Yes= 1, No=2 3. If Q2 yes, what is the size of irrigable land in timad? ______4. How many times do you produce per year using irrigation? ______5. Where is the irrigable water source available (accessible)? Highland=1 (upper stream), middle land (middle stream) =2, lowland (lower stream) =3 6. If your answer in Q2 is no, what are the main reasons? 1= Lack of water source, 2= Lack of capital, 3 =Lack of interests, 4=Lack of technical skill, 5= Steeply slope of plots, Others (specify) ______3.2. Utilization of Agricultural inputs 1. Did you use agricultural inputs for your farm? Yes =1, No =2 2. If yes, which agricultural inputs you used? 1. Fertilizer 2. Improved seed Herbicide/Pesticide 4. Others (specify)______3.3. Credit services

1. Is there credit support access in your locality? Yes = 1, No = 2 2. If Q1 yes, did you have utilized credit for the production of the Commodities? Yes= 1, No= 2 3. How much you received from your credit customer? ______Birr 4. What was the source of your Credit? 1. Banks, 2. Microfinances (ACSI ), 3. Rural saving cops 4. Agri union 5. Friends /Relatives, 6. Traders

6. If question number is no 1, what is the reason for not taking credit? Fear of repayment= 1, High interest rate on credit =2, No interest for credit = 3, Fear of defaulters = 4 in group collateral =5, Enough money= 6, Other specify ______3.4. Extension Service

1. Did you contact/visit any extension agent during the last cropping season? Yes = 1, No =2 2. If yes, for how many days per month you frequently visited extension agents for receiving agricultural advice from them? ______3. Did you gain any knowledge from the extension agents that could help you to do things differently on the specific commodities? Yes=1, No= 2, If no, specify your reason______3.5. Cooperative membership 1. Is there farmers’ union or cooperative in your locality? Yes =1, No = 2 2. If yes, are you member of it? Yes= 1, No= 2 3. If Q2 is yes, did you get credit during the last cropping season? Yes = 1, No= 2 4. If Q3 is yes, how much did you get? ______NB. If it is in kind please convert it in monetary value. 5. Are you member of equb at present? Yes =1, No = 2 6. If Q4 is yes, how much contribute per month ?______Birr. 7. Are you a member of Eddir at present? Yes=1, No=2 8. If yes, how much did you contribute per month?______Birr

Part 4. Housing

1. What is the type of house you own and live in? 1= Corrugated Iron Sheet, 2=Thatch/ grass roofed, 3= both 2. Does your house have partitioned rooms for humans and livestock? Yes = 1, No = 2 3. Does your house have toilet for the family? Yes =1, No= 2 4. Do you own a house in nearby town/village? Yes =1, No = 2 5. If yes, how much if it is sold? ______Part 5. farm implements and household durable assets Item Quantity Value at current Remark market price in Birr Agricultural Implements Ploughing set Sickle, Axe, Megaz etc Generator Tridal pump Bee hive (local & modern) Cart Weaving equipment Furniture & household durables Radio, Tap recorder Televsion Bicycle Motor Vehicle Mobile phone Solar set Refrigerator Sofa set Improved stoves

Part 6. Livestock and Oxen ownership 1. Did you have own livestock in previous year ? Yes = 1, No =2, If yes answer the following table . Livestock Numb Value in Source of Ownership Did you sell any animal in last year How many type er Birr died if any Own Gift/par Purchase No sold Total sale in Reason for sale ent d Birr (Code A) Oxen Cow Bull Heifer Calf Sheep Goat Horse Mule Donkey Poultry Code A 4. To buy other livestock 1.To buy food 5. To buy housing materials 2. To pay credit 6. For educational expense 3. to buy input 7. Health expense 8. Other specify ____

Part 7. Access to Public infrastructure

1. Where did you usually obtain food and non-food items you want to purchase? 1. Nearby central/district market 2. Village level market 3. Other 2. Did you sell farm and non-farm products during the last cropping season? Yes= 1, No =2 3. If yes, did you receive reasonable prices for your products sold? Yes = 1, No =2 4. How far is your house from the nearest market ______(Walking minutes) 5. Does the market distance from your house encourage farmers producing marketable products? Yes =1, No =2 6. Distance to all weather road ______walking minutes 7. Do you have the access for market information from your village? Yes =1, No = 2 8. If yes, what is/are your basic source of market price information? 1= Radio, 2=Traders, 3=Development Agents 4=Neighbors/fellow farmers 5. Mobile/phone 9. Did you get health services in your locality? Yes= 1, No =2 10. If yes, how much Birr paid for health service in last year? ______Birr. 11. Did you access veterinary services in your locality? Yes =1, No =2 12. How much you paid for your sick animals including purchase of drugs if any in the previous year ? ______Part 8. Off arm and or non-farm income. 1. Please mention the amount of income obtained accordingly in the last production period. No Source of income from on farm Obtained If yes annual income Remark and offarm/ non-farm Yes =1, No =2 (Birr) 1 Sale of crop produce 2 Land rent out 3 Sale of livestock & products 4 Khat, vegetable, fruit 5 Handicrafts 6 Sale of labor in neighbor

7 Sale of labor in a town 8 Sale of Charcoal/Wood 9 Sale of seedlings 9 Others specify

2. Remittance and other transfers receipts 1. Has the household received any other income (such as remittance, gifts or other transfers) in the last 12 months? Yes= 1, No =2

Type of recipent How many Amount For what times you recivied value purpose used received last in birr if it was (mainly) year? in kind Code A Remittance in cash Remittance in kind Food aid/nonfood aid Gift/inheritance Code A 1. To buy food for the house hold 2. To buy nonfood items 3. Payment for debit 4. To agric. Inputs 5. To household assets

3. Household food consumption expenditure

1. What are your staple food crops? ______2. Where did you get these food stuffs? own produce = 1, purchase= 2, From Relatives = 3, others specify = 4.

3. Please list the type and amount of each food consumed in 30 days and valued at that time market price (bought and own production for consumption purpose)

Food item Unit Quantity Kcal Market value Remark Teff Kg Maize >> Barely >> Finger millet >> Wheat >> Potato >> Cabage/kosta Tomato Onion Pepper Oil Litter Milk Litter Meat Kg Beans/Peans Linseed/Noug Salt Kg Butter Kg Egg No Honey Tea, coffe Areqi, Lit Tella, Tejj Lit Beer Soft drink Other (specify)

4. What is the monthly produce from animals that used for food? Food type Unit Produced Consumed Market Kcal Remark value Milk Litter Butter Kg Meat Kg Poultry No egg

5. Did you produce enough food for your household consumption? 1. Yes= 1, No= 2 6. If no, how do you cope up it? Purchase food =1, Sale of livestock=2, Borrow from neighbors/relatives= 3, Others specify______

7. Which of the month/s in a year are good in terms of food availability (more than one answer is possible). Sept Oct Nov Jan Feb Mar May Jun Jul Aug

8. Which of the months in a year are worst in terms of food availability Sept Oct Nov Jan Feb Mar May Jun Jul Aug

9. Which months were food shortage happened in your household? list the month/s______10. How many times did your household eat per day where the availability of food is average in non-fasting season? Once=1, Twice= 2, Trice=3 More than 3 times = 4. As obtained =5 3. Non-food expenditure of households Which of the following nonfood types were major expenditures? Nonfood type Amount in Birr/ specify (Total Remark expenses) Tax Fertilizer & improved seed School fee Health fee Clothing & shoes House utensils (including durable household items ) Cleaning & personal care items( Soap, Omo, Cosmetics, hair salon Veterinary services Transportation Kerosen or any Grinding mill Social & other contributions (Edir, association: women, youth, ANDM, Contribution to churches, school, clinics expenses Ceremonial expenses(Weeding, teskar, senbete, mahiber, Kerstna, etc

Part 9. Experience of Shocks 1. Did you face any natural and man-made shocks in your farm like frost, thunder, heavy rain, disease outbreak? Yes= 1, No= 2.

2. If Q1 is yes had this household been affected by the shock? Type of shocks Yes=1, No Shock resulted How widespread Remark =2 in (Code A) this shock (Code B) Too much rain/flood Frost/hailstrom Crop pest/disease Storage loss due to pest Livestock disease Death of husband/wife Illness of husband/wife Death of any household member Illness of any household member Other specify

Code A. Code B. 1. Loss of Productive asset 1. Only affected my household 2. Loss of household income 2. Affected some households in this 3. Reduction in household consumption village 4. Other specify 3. Affected all in the village 4. Affected this village & nearby village NB. More than one answer is possible

3. How did you cope up these most important shocks?

1. Own fund, savings 9. Land rent out 2. Sell livestock 10. Share livestock with relatives 3. Sale productive assets 11. Loans with interest from relatives, 4. Sale nonproductive assets village level lenders 5. Reduction in consumption 12. Loans without interest from community 6. Rent out land organizations 7. Send children to work 13. Emergency Food aid 8. Migration or family dissolution 14. Assistance from government 15. Assistance from NGO 16. Loans from credit institutions ( ACSI, Banks, )

Thank you!