IMPACT OF SMALL-SCALE IRRIGATION ON HOUSEHOLD FOOD

SECURITY: THE CASE OF BISHOFTU TOWN

MSc THESIS

BACHA TAYE GURMU

APRIL 29

OROMIA STATE UNIVERSITY

Impact of Small-Scale Irrigation on Household Food Security: The Case

BISHOFTU TOWN

A Thesis Submitted to the College of DEVELOPMENT ECONOMICS

OROMIA STATE UNIVERSITY

In Partial Fulfillment of the Requirements for the degree of

MASTER OF SCIENCES IN DEVELOPMENT ECONOMICS

Bacha Taye Gurmu

April 2024

OROMIA STATE UNIVERSITY

 APPROVAL SHEET
POSTGRADUATE PROGRAM DIRECTORATE
OROMIA STATE UNIVERSITY

As thesis research advisors, we hereby certify that we have read and evaluated this thesis
prepared under our guidance and direction, by Bacha Taye Gurmu, entitled, “Impact of Small-
Scale Irrigation on Household Food Security: The Case of Bishoftu Town and we
recommend that the thesis be submitted as it fulfills the thesis requirements for the degree of
Masters of science in Development Economics.

Mengistu Ketema (Professor)

Major Advisor Signature Date

Beyan Ahmed (Ass.Prof.)

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 Bacha Taye Gurmu and examined the candidate.
We recommend that the thesis be accepted as fulfilling the thesis requirements for the degree of
Master in Collaborative Master of Sciences in Development Economics.

Million Sileshi (PhD)

Chair Person Signature Date
Ketema Bekele (PhD)
Internal Examiner Signature Date
Sisay Debebe (PhD)
External Examiner Signature Date

Final approval and acceptance 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).

ii
 STATEMENT OF THE AUTHOR

By my signature below, I declare and affirm that this thesis is my own work. I have followed
all ethical and technical principles of scholarship in data collection, data analysis and
compilation of this Thesis. All sources of materials used for this thesis have been properly
acknowledged.

This thesis is submitted in partial fulfillment of the requirements for collaborative Masters of
Science degree at Oromia State University in Development Economics and is deposited at the
University Library and available for users and 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: Bacha Taye Gurmu

Signature:
Place: Oromia State
University
Date of Submission: May 04/2024

iii
 BIOGRAPHICAL SKETCH

The author was born in August 27, 1998 in Oromia National Regional State, West Showa
Zone, Bako Tibe District, from his father Mr. Taye Gurmu and his mother Mrs. Tajitu Abebe.
He attended elementary, high school and preparatory school at Bako Tibe elementary school
and Bako Tibe secondary and preparatory school. He then joined Rift Valley University in
2017 and graduated with Bachelor of Arts Degree in Economics In 2019. After his graduation,
the author was employed in Cooperative Bank of Oromia in 2020 as a Junior customer service
officer. He served the institute for Three years until he joined Oromia State University in 2021
to pursue his MSc degree in Development Economics.

iv
 ACKNOWLDEGMENTS

First and foremost, I thank the Almighty God for being with me from the start to the end of
my works and for giving me the opportunity and the capacity to realize my aspiration.

I would like to express my deepest gratitude to my major advisor Dr Baro Beyan and my co-
advisor Mr. Beyan Ahmed, whose contribution is immense, and without their involvements,
the accomplishment of this research would have been difficult. Besides, their kind and gentle
guidance and advisor-ship from the early design of the research proposal to the final write-up
of the thesis by adding valuable, constructive and ever-teaching comments highly improved the
contents of the thesis.

Thirdly, my heartfelt thanks go to the Ethiopian Institute of Agricultural Research for giving
me the chance to join the MSc program. Besides, I would like to thank National Agricultural
Biotechnology Research Center and Holeta Agricultural Research Center’s staff members,
more particularly Mr. Obsi Desalegn, Dr. G/rufael Girmay, Mr. Adisu Getahun, Mr. Takele
Mebratu, Mrs. Yewubdar Tadese and Mr. Shumet Embiale
Finally, I would like to extend my deepest gratitude to my beloved daughter Situnaf and her
mother Hawi who withstand loneliness when I was apart from them and gave me endless love
and courage during my thesis work.

v
 ACCRONYMS AND ABBREVIATIONS

AE Adult equivalent
AERC African Economic Research Consortium
ATT Average Treatment on Treated
CIA Central Intelligence Agency of America
CSA Central Statistics Agency
EHNRI Ethiopian Health and Nutrition Research Institute
EIAR Ethiopian Institute of Agricultural Research
FAO Food and Agricultural Organization
FCS Food Consumption Score
GDP Gross Domestic Product
GTP Growth and Transformation Plan
HDDS Household Dietary Diversity Score
HH Household
HHH Household Head
IFC International Finance Corporation
IMF International Monetary Found
IPC Integrated Food Security Phase Classification
Kcal Kilo calorie
PSM Prosperity Score Marching
SSI Small-scale Irrigation
TLU Tropical Livestock Unit
UNDP United Nation Development program
UNICEF United Nations Children's Fund
WFP World Food Program

vi
 TABLE OF CONTENTS

STATEMENT OF THE AUTHOR iii

BIOGRAPHICAL SKETCH iv
ACKNOWLDEGMENTS v
ACCRONYMS AND ABBREVIATIONS vi
TABLE OF CONTENTS vii
LIST OF TABLES xi
LIST OF FIGURES xii
LIST OF TABLES IN APPENDICES xiii
ABSTRACT xv
1. INTRODUCTION 2
1.1. Background of the Study 2

1.2. Statement of the Problem 3

1.3. Research Questions 5

1.4. Objectives of the Study 5

1.5. Significance of the Study 5

1.6. Scope and Limitations of the Study 6

1.7. Organization of the Thesis 6

2. LITERATURE REVIEW 7
2.1 Definitions of Basic Concepts 7

2.2 Food Security Status in Ethiopia 9

2.3 Theoretical Framework of the Study 10

2.4 Measuring Household Level Food Security 13

vii
 TABLE OF CONTENTS
2.4.1. Food consumption score (FCS) 13
2.4.2. Household consumption and expenditure survey (HCES) 14
2.4.3. Household dietary diversity score (HDDS) 14
2.4.4. Coping strategy index (CSI) 15
2.4.5. Household hunger scale (HHS) 15
2.4.6. Individual’s daily calorie intake (DCI) 16

2.5. Irrigation Schemes Development in Ethiopia 16

2.6. Impact Assessment Methods 17

2.6.1. Propensity score matching (PSM) 17
2.6.2. Difference-in-difference (DID) 18
2.6.3. Randomized selection methods (RSM) 19
2.6.4. Regression discontinuity (RD) 19

2.7. Empirical Literature Review 20

2.7.1. Determinants of household participation in small-scale irrigation 20
2.7.2. Impacts of small-scale irrigation on household food security 21
2.7.3. Determinants of household food security 22

2.8. Conceptual Framework of the Study 23

3. RESEARCH METHODOLOGY 26
3.1. Description of the Study Area 26
3.1.1. Demography 26
3.1.2. Climate 27
3.1.3. Agriculture 27
3.1.4. Description of irrigation schemes in the district 27

3.2. Data Type, Sources and Method of Collection 28

3.3. Sampling Technique and Sample Size Determination 29

viii
 TABLE OF CONTENTS
3.4. Methods of Data Analysis 30
3.4.1. Descriptive statistical analysis 30
3.4.2. Econometric analysis 31
3.4.2.1. Factors affecting household participation in small-scale irrigation 31
3.4.2.2. Factors affecting household food security 31
3.4.2.3. Impact of small-scale irrigation on household food security 32

3.5. Variables Definitions and Hypothesis 35

3.5.1. Dependent variable 35
3.5.2. The outcome variables 36
3.5.3. Independent variables 36

4. RESULTS AND DISCUSSION 42

4.1. Descriptive Results for Demographic and Socio-economic Variables 42
4.1.1. Descriptive results of dummy variables 42
4.1.1.1. Participation in irrigation 42
4.1.1.2. Household food security status 43
4.1.2. Descriptive results of continuous variables 45
4.1.2.1 Participation in irrigation 45
4.1.2.2 Household food security status 47
4.1.3. Summary of food consumption score by participation in irrigation 49
4.1.4. Summary of daily calorie intake by participation in irrigation 50

4.2. Econometric Results 52

4.2.1. Factors affecting household participation in small scale irrigation 52
4.2.2. Factors affecting household food security 56
4.2.3. Impact of small-scale irrigation on household food security 61
4.2.3.1. Estimation of propensity score 61
4.2.3.2. Restricting the common support region 63

ix
 TABLE OF CONTENTS
4.2.3.3. Choosing the matching algorithm 66
4.2.3.4. Assessing the matching quality 67
4.2.3.4.1.Balancing test 67
4.2.3.4.2.Estimation of the average treatment effect on treated (ATT) 69
4.2.3.5. Sensitivity analysis 70

5 SUMMARY, CONCLUSIONS AND RECOMMENDATIONS 73

5.1 Summary and Conclusions 73

5.2 Recommendations 75

6 REFERENCES 78
7 APPENDICES 90
7.1. List of Tables in Appendices 90

7.2. Lists of Figures in Appendices 92

7.3. Questionnaires Used for Data Collection 93

x
 LIST OF TABLES

Table Page
1. Modern small-scale schemes in Bishoftu Town 28
2. Total sample distribution over the selected kebeles 30
3. Summary and hypothesis of explanatory variables 41
4. Descriptive results for dummy variables (Participation in irrigation) 43
5. Descriptive results for dummy variables (Household food security status) 44
6. Summary of continuous explanatory variables (Participation in irrigation) 47
7. Summary of continuous explanatory variables (Household food security) 49
8. Food consumption score category by participation in small scale irrigation 50
9. Daily calorie intake by participation in irrigation 51
10. Variance inflation factor test (for 13 explanatory variables) 52
11. Logit estimation of factors affecting participation in small scale irrigation 55
12. Variance inflation factor test (for 14 explanatory variables) 56
13. Factors affecting household food security estimated using daily calorie intake 59
14. Factors affecting household food security estimated using food consumption score 60
15. Logistic regression result for estimating propensity scores 62
16. Summary of estimated propensity score to restrict common support region 64
17. Comparisons of different matching methods 66
18. Balancing test for covariates 68
19. Estimation of ATT based on daily calorie intake and food consumption score 69
20. Rosenbaum sensitivity analysis for daily calorie intake method 71
21. Rosenbaum sensitivity analysis for food consumption score method 72

xi
 LIST OF FIGURES

Figure Page
1. Conceptual Framework of the Study 25
2. Location map of Bishoftu Town 26
3. Kernel density for participants, non-participants and both, before matching 63
4. Kernel density of propensity score for non-participants in common support 65
5. Kernel density of propensity score for participants in common support 65

xii
 LIST OF TABLES IN APPENDICES

Appendix Table Page

1.
Conversion factor for kilocalories per gram for different food items 90
2.
Conversion factors for tropical livestock units (TLU) 91
3.
Conversion factor for adult equivalents (AE) 91

xiii
 LIST OF FIGURES IN APPENDICES

Appendix Figure Page

1. Histogram of propensity score by participation in irrigation 92
2. Graphical presentation of propensity score 92
3. Graph of propensity score for unmatched and matched covariates 93

xiv
 Impact of Small-Scale Irrigation on Household Food Security: The Case of Bishoftu
Town

ABSTRACT

Small-scale irrigation is among the feasible way in which agricultural production and
productivity can be boosted to meet the ever-growing food demand of least developed countries
like Ethiopia. But, unlike the irrigation potential, the utilization level and its impact on food
security were not empirically analyzed in specific areas of the country. This study was
conducted to identify factors affecting household participation in small-scale irrigation, factors
affecting household food security and assessing the impact of small-scale irrigation on
household food security in Bishoftu Town. Both primary and secondary data were used.
Primary data were collected from 220 irrigation users and non-users. Descriptive, inferential
and econometric data analysis were executed. The logistic regression applied to estimate
factors affecting participation in small scale irrigation revealed that age, livestock holding,
sex, family size, land owned, occurrence of crop pests and diseases, distance from irrigation
site and access to credit services were the variables that significantly affected. Similarly, the
logistic regression applied to estimate factors affecting household food security revealed that
sex of the head, family size, dependency ratio, livestock holding, land holding, access to
extension contact, access to irrigation services and access to credit services were the variables
that significantly affected. To analyze the impact of small-scale irrigation on households’ food
security, PSM method was applied. Radius matching with band width of 0.1 was the matching
algorithm used. The quality of covariate balancing was checked using pseudo R2, mean bias
and t-test. Finally, ATT was estimated and the result revealed that family members of irrigation
user households on average consumed more calories of 529 kcal than irrigation non-users, and
this result is statistically significant. Sensitivity analysis was done and the estimated ATT was
insensitive to unobserved bias up to 200%. Therefore, policy interventions giving priority to the
variables mentioned above to increase participation in small-scale irrigation and also improve
household food security status are recommended.

Keywords: Impact, Small-scale irrigation, Food security, Bishoftu, Calorie, PSM

xv
 1. INTRODUCTION

1.1. Background of the Study

Food insecurity and hunger are alarmingly increasing in the world. The number of people
under acute food insecurity escalated from 80 million during 2015 to 108 million in 2016 and
reached 124 million during 2017 (FAO et al., 2018). Moreover, the figure of undernourished
people in the world has also increased to more than 820 million from 804 million during the
same years, indicating one person out of nine is undernourished. These figures are clearly
showing that the achievements of Sustainable Development Goal, specifically that of SDG-2
(Hunger Eradication) is at risk (FAO et al., 2018; FSIN, 2019).

Conflicts, climate variabilities and extremities (severe drought, floods, storms) are the major
driving factors behind global hunger and severe food crises. These problems are even worse in
agrarian countries like those in Africa, in which the larger proportion of the population
depends on agriculture (World Bank, 2018; World Vision, 2018). In Africa, about 20 percent
of the population, (257 million) was hungry, out of which 237 million were from sub-Saharan
Africa during 2017. For example, the prevalence of undernourished people in the region was
escalated from 181 million in 2010 to 222 million in 2016, and then raised to more than 236
million during 2017 (FAO et al., 2018; FAO and ECA, 2018).
Ethiopia is the second most populated country in Africa following Nigeria, having 105 million
people estimated for 2017 (World Bank, 2019). Displacement induced by intercommunal
clashes (3 million), coupled with drought, worsen the food and nutritional insecurity situation
in the country, and more than 8.1 million people are in need of food assistance during 2019
(FAO, 2019). However poverty reduction and human development efforts have contributed
much to lift out human lives out of poverty, about 22 million people are still below the
national poverty line during 2017. Even though the human development index (HDI) of the
country is showing improvements, Ethiopia is yet among the poorest countries in the world,
holding the rank of 173rd out of 189 countries (UNDP, 2018a).

To withstand the problems of drought, and to sustain the food security of the highly increasing
population, agricultural technologies that can improve production and productivity with the
 2

existing resources, such as irrigation are promoted for their contributions of improving
livelihood, increasing income (Hirko et al., 2018; Tulu, 2014), employment opportunity, food
security, poverty reduction, production improvements, diversification of crops grown,
improved household income, improved health status, source of balanced diet, and easier
access to medications, good sources of feeds for animals and good sources of asset ownership
(Asayehegn, 2012).

Ethiopia has immense water resources from rivers, lakes and ground which is 124.4, 70 and 30
billion cubic meters, respectively (Berhanu et al., 2014). Out of these, the irrigation potential
of major rivers and rift valley lakes is estimated to reach around 3,800,733 hectares of land
(Ayalew, 2018). But, 97 percent of cereal production is by rainfed, while irrigation is serving
only 3 percent (FAO and IFC, 2015). Ethiopia is also endowed with 362,590 square kilometers
agricultural land and 15,119,000 hectares of arable land (CIA World Fact Book, 2019).
Recently, the total irrigated land area in the country is increasing and reached 2.4 million ha in
2015 from 885,000 hectares during 2011, and the expected plan is to increase it to 4 million
hectares by 2020 (ATA, 2016).

In Oromia region, the trend of crop production is not rising with the pace by which the
population growth is rising due to different reasons like recurrent drought and environmental
degradation. Therefore, using small-scale irrigation is the most feasible remedy to alleviate
such problems and improving production and productivity of small-holders and also for
improving their food security status. According to OIDA (2014), Oromia region is endowed
with ample amount of irrigation water and irrigable land resources. Out of the total of 1.7
million hectares of irrigable land, only 800,000 hectares (47.05 percent) were irrigated yet
both in traditional and modern schemes.

According to the information from Bishoftu agriculture and rural development bureau (2020),
the district has high potential of irrigation water and irrigable land. Out of 23 rural and 8 urban
kebeles in the district, irrigation access exists in 20 rural and 4 urban kebeles, but in limited
area coverage in 3 rural and 1 urban kebeles. The estimated irrigation potential of Bishoftu Tis
more than 9,055 hectares.
 3

According to Mengistie and Kidane (2016), small-scale irrigation has direct and indirect
impact in enhancing farmers’ livelihoods through diversification of crops grown, increasing
agricultural production, and increasing household income. But, the impact of small-scale
irrigation on household food security, factors affecting farmers’ participation in small-scale
irrigation and factors affecting household food security in the study area is not empirically
analyzed and this study is proposed with this background.

Therefore, in this study, the impact of small-scale irrigation on household food security,
factors affecting farmers’ participation in small-scale irrigation and factors affecting
household food security in Bishoftu Town will be analyzed.

1.2. Statement of the Problem

Small-scale irrigation has direct and indirect impacts in enhancing the livelihood of farm
households through diversification of crops grown, increased agricultural production,
increased household income, and increased employment opportunities (Mengistie and Kidane,
2016). This is why irrigation development is being viewed as promising approach to ensure
food security and improved livelihood under climate variability and increasing population
growth (Passarelli et al., 2018). Irrigation is also promoted for its contribution for seasonal
food security of households, dietary diversity, health, and resilience to drought and climate
changes (Domènech, 2015; Lefore et al., 2019). It has the potential of increasing agricultural
yields by more than 50%, in which the majority of increased income helping smallholder
farmers (Xie et al., 2014).
The issue of food insecurity is the major policy issue for poor countries like Ethiopia as it is
the problem of millions. Therefore, empirically analyzing such issues is very important for
policy intervention. Recurrent drought, lower average land holding, higher average family size
and low soil fertility are among the factors contributing to food insecurity status in the
country. The average land holding is about 0.9 hectare (FAO, 2015), the average family size is
about 5 persons per household and the average cereal productivity is 2.45 tons per hectare
(CSA, 2017). According to Sileshi (2016), households of Bishoftu Town are reported to be
food insecure and used to run through a hand to mouth style of living.
 4

Production and productivity improving agricultural technologies like small-scale irrigation

play inevitable role to sustain household food security, and limited information on such
technologies, pest management, seeds, markets and modern equipment prevent farmers from
making informed decisions, and it also prevent private investment from entering into the
market, which again make irrigation sector contributes less to food security, wealth creation
and resilience (Lefore et al., 2019).

In Ethiopia, there is wider gap between irrigation potential and the current level of
implementation in terms of irrigated land and the number of participating farmers. This is
because of technical, physical and economic challenges. Unlike the potentials, about 97
percent of cereal production in the country is being produced using rainfed and irrigation is
contributing only 3 percent for cereal production (FAO, 2015). According to ATA (2016), the
challenges hindering farmers from fully utilizing the existing irrigation potential were not
exhaustively identified in specific areas of the country. But, recently, efforts are being made to
transform agricultural sector from traditional and rain-fed to technology intensive and
mechanized, irrigated and market-oriented agriculture, with packages of post-harvest
technologies (FAO and IFC, 2015).

Similar to the country, utilization of irrigation potential for Bishoftu Town is also below the
potential (Sileshi, 2016). But, there was no studies conducted in the district on why farmers
are not fully utilizing the existing irrigation potential; what factors are affecting food security
status of the households and the impact of small-scale irrigation on household food security in
Bishoftu Town.

As confirmed by different scholars like Asayehegn, 2012, Hirko et al., 2018 and Tulu, 2014,
small-scale irrigation improves production, productivity, income, consumption, welfare and
general wellbeing of the farm households. But, according to Sileshi (2016) these opportunities
of irrigation are not well utilized in the study area and the farmers is experiencing hand to
mouth mode of live. But the research conducted by this scholar did not specify factors
affecting farmers’ participation in small scale irrigation, factors affecting household food
security and also the impact of small-scale irrigation on household food security.
 5

Therefore, this research will assess the impacts of small-scale irrigation on household food
security status, factors affecting households’ participation decision in small scale irrigation
and also factors affecting food security status of the households in the study area and come up
with relevant information for policy intervention, extension actions, and for further researches.

1.3. Research Questions

This study has tried to answer the following questions:
1. What are the factors that affect households’ participation in irrigation in the study area?
2. What are the factors that affect households’ food security in the study area?
3. What is the impact of small-scale irrigation on household food security in the study area?

1.4. Objectives of the Study

The general objective of the study is to analyze the impacts of small-scale irrigation on food
security status of the households in the study area.

The specific objectives of the study are:

1. To identify factors affecting household participation in small-scale irrigation in the study area
2. To identify factors affecting household food security in the study area.
3. To analyze the impact of small-scale irrigation on household food security.

1.5. Significance of the Study

To sustainably supply food for the ever-increasing population of Ethiopia, promotion of

affordable and profitable agricultural technologies such as small-scale irrigation should get
priority area as it contributes many dimensionally in combatting food insecurity,
unemployment, malnutrition and other related problems. It increases agricultural production
and productivity and income and improve the livelihood of the rural households.

Therefore, realizing the impact of irrigation on food security, identifying factors determining
household participation in irrigation and also identifying factors affecting household food
 6

security in the study area will avail information for farmers, policy makers, researchers and
other stakeholders for decision making and formulation of intervention policies and strategies.

1.6. Scope and Limitations of the Study

The study focused on the assessment of the impact of small-scale irrigation on households’
food security and it was limited to only Bishoftu Town because of resources and time
limitations. Therefore, the findings of this study have some limitations. First, since the study focused
only on one district, generalization of the result at regional level may not be possible. Additionally,
since cross-sectional data of one year was used, it is relatively shorter period of time to
understand the impact of small-scale irrigation on household food security. Moreover, since
there was no baseline data, propensity score matching, which require more variables and also
more sample size is used. Finally, since the study will be using cross sectional data of one
season, generalization of calorie intake that may differ from season to season may lead to bias.
1.7. Organization of the Thesis
This thesis is organized in five chapters. The first chapter contains background of the study,
statement of the problem, objectives of the study, significance of the study and scope and
limitations of the study. Under chapter 2, review of literatures regarding basic concepts of
small-scale irrigation, methods of measuring food security, impact assessment methods and
empirical literature reviews on the impact of irrigation on household food security are
discussed. The third chapter presents the description of the study area, types, sources and
methods of data collection, and methods of data analysis. In chapter 4, the findings on the
socio-cultural and demographic characteristics of sample households, factors affecting
household participation in small scale irrigation, factors affecting household food security and
impacts of irrigation on household food security were presented. Finally, chapter 5 presented
summary, conclusion and recommendations.
 7

2. LITERATURE REVIEW

2.1 Definitions of Basic Concepts

The concept of food security is broad and flexible. It was first launched during 1940s and
improved onward. During 1970s, national and global food security gets concern and then in
1980s, the focus was shifted to household and individual level access to food (Maxwell and
Wiebe, 1998). It is a broad concept about the issues related to the nature, access, quality and
security of foods. The concepts had undergone several redefinitions from Hot Springs
Conference of Food and Agriculture onward. These are food surplus disposal (1940-
1950), food for development (1960s), food assurance (1970s), broadened food security
(1980s) and freedom from hunger and malnutrition (1990). The food security issue got
prominences during 1970s, particularly in 1974 during the World Food Conference in Rome.
Since 1974, the food security concept evolved, developed and diversified at global, national,
household and individual level (Maxwell, 1996).
There are many definitions of food security given by many scholars and organizations.
However, the recent and widely accepted definition of food security is the definition including
the quantity, quality, safety, socio cultural acceptance of food, which can be mentioned as
follows:
Food Security: Is defined and said to exists when all people, at all times, have physical and
economic access to sufficient, safe and nutritious food to meet their dietary needs and food
preferences for an active and healthy life (WFS, 1996).
Food Availability: is achieved when sufficient quantities of food are consistently available to
all individual households. Such food can be supplied through household production, other
domestic outputs, commercial imports or food assistance. It addresses whether food is actually
or potentially physically present for purchase or acquisition for consumption (IPC, 2019)

Food Access: Is ensured when households and all individuals within them have adequate
resources to obtain appropriate food for a nutritional diet. Access depends upon income
available to the household, on the distribution of income within the household and on the price
of food. It focusses on different sources food like; own production, purchases, gifts, aid,
gathering, etc. and whether the households are able to acquire enough food to cover their
nutritional needs from the sources available.
 8
The ability to access enough food will directly depend on physical access like own production,
distance to markets), financial access (e.g. purchasing power, access to credit) and social
access (e.g. ability to secure food through social networks, based on extended family,
ethnicity, religion or political affiliation) (IPC, 2019).
Food Utilization: It is the proper biological use of food, providing sufficient energy and
essential nutrients, potable water, and adequate sanitation. Utilization is the question that
comes after food availability and access. It is the question whether households are maximizing
the consumption of adequate nutrition and energy, which is usually a factor of food
preferences, preparation, storage and access to adequate quantity and quality of water (IPC,
2019).
Food Stability: It refers to the ability of the household to obtain food over time. It is the last
pillar of food security that focusses on whether the whole system is stable and ensuring that
the household will be food-secure in the future (IPC, 2019).
Household: Household is defined as a unit of people living together headed by a household
head and derives food from a common resource, obtained mainly from farming activities. The
head is regularly a man and maybe a woman occasionally when there is no man. Grandparents
and adolescents may also a household head where both parents have dead (Callens and
Seiffert, 2003).
Farm households are individuals or a group of people sharing one home and living together
and deriving food from a common resource, headed by a household head and obtain their
income mainly from farming activities (Ellis, 1993).

Understanding the difference between national food security and household food security is
basic since measures of improving household food security are quite different from that of
improving national food security. National level food security is mainly focus on macro-level
production, distribution, marketing and ensuring food security of the population generally (De
Haen, 2003).

In this study, following the arguments of Frank Ellis (1993) and Callens and Seiffert (2003),
the farm household is taken as individual or a group of people sharing one home, headed by
one household head, deriving food from a common resource and obtaining its income mainly
from farming activities (more than 50 % of its income).

Irrigation: is defined as the science of artificial application of water to the soil to increase the
moisture content of the soil which is vital for plant root growth and plant development and to
prevent stresses that may cause yield as well as quality reduction of the crops (Tesfaw, 2018).
 9

Small Scale Irrigation: is defined as the irrigation started, owned, managed and financed by
farmers themselves, individually or in a small group in which farmers produce high value
horticultural crops and staple crops (Lefore et al., 2019; Otoo et al., 2018).
2.2 Food Security Status in Ethiopia
Ethiopia is one of the fastest growing economies in the world, with an average annual gross
domestic product (GDP) growth rate of 10 percent (Paul et al., 2016). But the economy of the
country is prone to several challenges that affecting its development status and also affecting
food security status of the country. Among these factors, extreme poverty, chronic
malnutrition, civil and political conflicts, displacements of people from their homes,
unemployed young urban populations, recurrent droughts, crop-threatening pests, and etc. are
the major factors affecting rural livelihoods, income and food security (USAID, 2018; Garnett
et al., 2013).
Poverty in Ethiopia is a persistent problem and chronic and acute food insecurity is prevailing
in the country. About 10 percent of the citizens are chronically food insecure, which can also
increase to more than 15 percent in the years of drought. The major contributing factors for
food insecurity of the country are: population growth, drought, poor technologies use, poor
extension services, and post-harvest crop loss (Endalew et al., 2015).
According to World Bank (2016), 55.3 percent of the citizens of Ethiopia were living under
international poverty line, and 44.2 percent of the population was living below the national
poverty line during 2000. By showing moderate reduction, the percentage of the population
living below international and national poverty line reached 33.5percent and 29.6 percent
respectively during 2011 (World Bank, 2016).
Food insecurity has been critical challenge to Ethiopia for decades. Efforts made during the
past 30 years to improve food security status in the country have resulted in improvements
both in food security and health. But, the problems of food security and nutrition and health
status is continued to be a challenge yet. When we see the health status of children under five
year and women of reproductive age, 57 percent of children under five years were anemic, 38
percent were stunted, 24 percent were underweighted, 13 percent were born with low birth
weight, 10 percent were wasted, and 1 percent was overweight and 24 percent of women of
reproductive age are anemic during 2016 (USAID, 2018). Similarly, despite the attempts made
to improve food security situation, the actual number of people exposed to food shortages in
Ethiopia has remained significantly high. A rapidly increasing population, slow productivity
growth and climate-related disasters such as droughts increase food insecurity. Thus, the
agricultural sector of the country has been unable to produce sufficient food to feed the ever-
 10
growing population of the country. In recent years, commercial food import and food aid have
been accounting for a significant proportion of the total food supply in the country (Cochrane,
2011; Lefort, 2011).
2.3 Theoretical Framework of the Study

Utility maximization theory was used. It refers to the concept in which individuals and firms
seek to get the highest satisfaction from their economic decision makings. In this study, it is
assumed that from their experience, farmers know major irrigation related benefits and
problems, and they can state their preferences. They are expected to rationally reveal their
preference in line with the objective of improving their welfare. This preference can be
represented by a utility function and the decision problem can be modeled as a utility
maximization problem.
Suppose that the farmer drives utility from using small scale irrigation and his resource
endowment. Let farmer’s participation in small scale irrigation be represented by k, where k =
1 if the farmer decides to participate in small scale irrigation to maximize his utility and k = 0,
otherwise. Resource endowment of the farm household is represented by r, and ‘x’ represents
other observable attributes of the farm household that might potentially affect participation in
small scale irrigation.
If a farmer participates in small-scale irrigation for crop production, the farmer's utility is
given by the function U1 = U (1, r, x) and, if the farmer does not participate in small scale
irrigation, the farmer’s utility is given by U0 = U (0, r, x). Thus, based on this economic
theory, the farmer chooses the best alternative that offering best value, subject to his/her
constraints. According to Bekele (2004) and as it is most common in the specification of a
utility function, we assume additively separable utility function in the deterministic and
stochastic components where the deterministic component is assumed to be linear in the
explanatory variables. Functionally, it can be expressed as:

U1= U (1, r, x) = t (1, r, x) + ɛ1..........................................................................................................................................................

and
Uo = U (0, r, x) = t (0, r, x) + ɛ0....................................................................................................................................................2

Where, Uk (.) is the utility from the use of small-scale irrigation water (yield and income) and
tk(.) is the deterministic part of the utility, and ɛk is the stochastic component representing the
component of utility known to the farmers but are unobservable to the economic investigator.
It is obvious that farmers are assumed to know their resource endowment, r, and the implicit
cost of using irrigation water or practicing irrigation in terms of engagement of their resources
 11
and can make a decision whether to use it or not. Assume that farmer’s implicit cost of
deciding and using small scale irrigation is represented by ‘C’.

Therefore, a farmer will decide to participate in small-scale irrigation if the utility from
participation exceeds the utility from not participating. Functionally:
 12

U1(.) > U0(.)

t (1, r - C; x) + ɛ1 > t (0, r; x) + ɛ0................................................................................................................................................3

The presence of the random component permits us to make probabilistic statements about
farmers’ decision behavior. If the farmers participate in irrigation, the probability distribution
is given by:

P = Pr (t (1, r - C; x) + ɛ1 > t (0, r; x) + ɛ0..............................................................................................................................4

and it will be given by the function:
P = Pr (t (0, r; x) + ɛ0 > t (1, r - C; x) + ɛ1..............................................................................................................................5
if the farmer did not participate in irrigation.

By the assumption that the deterministic component of the utility function is linear in the
explanatory variables, the utility functions of small-scale irrigation participation and non-
participation can be expressed as:

U1 = β′1xi + ɛ1 for participation and

U0 = β′0xi + ɛ0 for non-participation,

Where: β1' and β0' are the vectors of the response coefficients and
ɛ1 and ɛ0 are random disturbances

Similarly, the probabilities of participation and non-participation in small scale irrigation can
be given as:
P = Pr (U1(.) > U0(.))
P = Pr (β′1xi + ɛ1 > β′0xi + ɛ0)
P = Pr (β′1xi - β′0xi > ɛ0 - ɛ1)
P = Pr (xi (β′1 - β′0) > ɛ0 - ɛ1)
P = Pr (xi α > vi)
P = Pr (xi α)..................................................................................................................................6
Where: P = is the probability function,
vi = (ɛ0 - ɛ1) is a random disturbance term
α = (β′1 - β′0) is a Qx1 vector of parameters to be estimated
 13

xi = is n x Q matrix of explanatory variables

P (xi α) = is the cumulative distribution function for vi evaluated at xiαi

The probability that a farmer will use the irrigation technology is then a function of a vector of
the explanatory variables, the unknown parameters and the disturbance term.

2.4 Measuring Household Level Food Security

Food security can be measured at household level using different methods. Among these
methods; Food consumption score, Food expenditure approach, Dietary diversity index,
copying strategy, Household hunger scale and Individual dietary intake are the most
commonly used methods.

2.4.1. Food consumption score (FCS)

The Food Consumption Score is the index recommended by World Food Program and called
the FAO method. FCS is calculated using consumption of different food groups consumed by
the household during the past 7 days. It is the combination of dietary diversity and food
frequency. Dietary diversity focuses on the number of different foods eaten within the specific
days and food frequency focuses on the number of days that a specific food item is eaten
within consecutive seven days. A food group is a food items with a similar caloric and nutrient
content.

To calculate Food Consumption Score, we go about six-steps. First, the seven-day food
consumption frequency is collected. Second, the collected data is grouped into nine specific
food groups. On the third step, the repetition of consumption for each food groups are added
up. The value of each food group which are above seven are recorded as seven. On the fourth
step, the consumption frequency obtained is multiplied by the weights given to each food
group (As mentioned in Appendix 3 question no. 10) to get new weighted scores. The weight
increases as energy, quality protein and micronutrients contents increase. The fifth step is
calculating Food Consumption Score using the following formula:

FCS=2*Xstaple+3*Xpulse+1*Xvegetable+1*Xfruit+4*Xmeat&fish+4*Xmilk+.5*Xsugar+.5*Xoil.......................7
 14

Finally, the Food Consumption Score of the household is categorized based on the following
thresholds. Accordingly, the household is considered as ‘poor food security profile’ if the
value of FCS is from 0 to 21, the household is said to be on the borderline threshold if the
value of FCS is between 21.5 and 35, and the FCS is acceptable if the value is greater than 35.
The value of the borderline can be increased to be between 28 and 42 for the populations
frequently consuming sugar and oil. During interpretation, the failure of FCS to capture
seasonal changes of food consumption should be considered (WFP, 2008; Sassi et al., 2018).

2.4.2. Household consumption and expenditure survey (HCES)

This method estimates the proportion expenditure on food out of the total income of the
household within 30 days. It also includes the values of foods produced and consumed at
home. If the household is poor, the portion of income he/she spend on food is high and he/she
is more vulnerable to food insecurity. Based on the value of this computation, households are
classified as low vulnerable if the share of their expenditure on food is less than 50 percent,
medium vulnerable if the value of their expenditure on food is between 50-65 percent, highly
vulnerable if the share of their expenditure on food is between 65-75 percent, and are very
highly vulnerable if their expenditure on food is greater than 75 percent (Kennedy et al.,
2011).

2.4.3. Household dietary diversity score (HDDS)

Dietary Diversity is a commonly used measure as data collection on food group consumption
are easy and dietary diversity has positive associations with both the nutrient quality of diets
and child anthropometry. It also has an association with various measures that are commonly
considered as proxy indicators of household food security (Andrew et al., 2013).

Dietary diversity score is a qualitative measure of food security which measures the
individual’s access to diversities of food groups. It measures the types of food groups
consumed by an individual or the household during specific period of time. HDDS is similar
with FCS, but HDDS preferably uses the 24-hour recall to minimize food consumption
memory errors, but 3 to 7 days recall is also possible (Kennedy et al., 2011; Andrew et al.,
2013).
 15

2.4.4. Coping strategy index (CSI)

This Strategy is the corrective measures that the households adopted to withstand the short-
term food insufficiency conditions. This is the coping strategies that the HH adopted during
the past 7 days when they do not have enough food. Generally, two Coping Strategies are
commonly used, namely: the contextual and the reducing. The context-specific is group or
location specific behaviors (Maxwell and Caldwell, 2008). To analyze coping strategy of
different income levels and different livelihood levels, reduced strategy is preferable. There
are five generally accepted coping strategies. These are: Borrowing from others, changing
preference to the less preferred ones, limiting adult intake, limiting portions and reducing the
number of meals per day (Maxwell and Caldwell, 2008; Sassi et al., 2018).

The severity of reduced coping strategy is calculated by multiplying the frequency of each
coping strategy by the standard universal severity weights. The standard universal weights are
1 for eating less-preferred foods, 2 for borrowing food or money, 1 for limiting portions at
mealtime, 3 for limiting adult intake and 1for reducing the number of meals per day. The
largest value of the index is expected to be 56 assuming that the household implemented all
coping strategies in all 7 days of the week and the higher the value of the index, the severe the
level of coping strategy or the more the food insecurity level. This strategy shows the image of
current food insecurity extent and it is commonly used in measuring the impacts of food
assistance program and indicator of early warning food crisis (Maxwell and Caldwell, 2008).

2.4.5. Household hunger scale (HHS)

If the population under investigation is expected to be in higher level of hunger, this index is
recommended to measure the severity of household food security problems. Three questions
about the perceptions of the household about the varying degrees of hunger. The perceptions
of the households about the degrees of hunger is measured from the number of times
experienced hunger during the past 30 days before the survey day. This scale is used in
monitoring the progress of the implementation of programs and policies and for warning about
food and nutrition security (Ballard et al. 2011).
 16

2.4.6. Individual’s daily calorie intake (DCI)

This method is measured by using different methods. These are: 24-hour recall method,
individual food records method and using questionnaires of food frequency. The drawback of
the 24 hour and the food frequency questionnaire methods is that they depend on the memory
of respondents and the recording of foods method requires individual’s ability of recording
consumed food.

Compared to other methods, individual dietary intake method has some exclusive and
common advantages. Among these it enables us to measure direct food consumption in
addition to food availability. It also enables us to measure both dietary quality and caloric
intakes at the individual level. It allows us to map food insecurity at individual and national
level and help us understand the patterns of intra household food consumptions (Pérez and
Segall, 2008).

This method is selected and used in this thesis research to measure and compare food security
level of irrigation participants and non-participant households in the study area.

2.5. Irrigation Schemes Development in Ethiopia

Access to irrigation water is crucial in the improvement of rural smallholder farmers. It is one
of the promising options of increasing agricultural production and productivity, diversify
agricultural products, create opportunity of employment and improve the income of
smallholders (Deribe, 2008).

During and prior to the Dergue regime, the focus on irrigation sub sector was on construction
of large-scale irrigations, while the currently ruling government mainly focuses on the
development of SSI schemes in which several small-scale irrigations has been developed and
the existing facilities are rehabilitated all over the country. But, irrigation development history
in Ethiopia is blamed of its prominence on technical and engineering aspects while it gives
less concern to policy, institutional and socio-economic factors (Deribe, 2008).

Unlike the abundance of water resources, Ethiopia is one of the countries whose agricultural
production is mainly dependent on rainfed and prone to climate variabilities. Ethiopia’s
 17

irrigation potential is about 5.3 million hectares of land out of which only 7.5 percent in
average is irrigated and this make Ethiopia among the least irrigation user in Africa (Mosissa
and Bezabih, 2017).

Effective investments in agricultural water management water will contributes to the

improvement of food security status by increasing production, product diversification,
increasing income and stabilizing prices (FAO and ECA, 2018). Improving agricultural water
management on small farms in developing countries is critical forcmd increasing and ensuring
food security and improved nutrition, particularly given expected growing food demands and
climate variability (Otoo et al., 2018). But, if not in controlled manner, irrigation causes
undesirable social and environmental damages such as degrading soil fertility by erosion and
creating conflicts among users over the use of the resource (Theis et al., 2018; Lefore et al.,
2019).

2.6. Impact Assessment Methods

To undertake impact evaluation, there are different methods that we can use. These methods
are: Propensity Score Matching PSM), Difference in Difference (DID), Randomized Selection
Methods (RSM) and Regression Discontinuity Design (RDD) (Abadie et al., 2004).

2.6.1. Propensity score matching (PSM)

Propensity score matching is a statistical matching technique that estimate the effects of a
treatment given the covariates. It allows finding a control group from a sample of non-
participants closest to the treatment group in terms of observable characteristics so that both
groups are matched on the basis of the propensity score. Propensity score is a predicted
probability of participation given observed characteristics (Ravallion, 2008). The propensity
score is estimated using statistical models, logit or probit, and the average treatment effect
(ATE) of the outcome of the two groups in absence of baseline data is calculated (Abadie et
al., 2004).

It is used when it is possible to create a comparison group from a sample of non-participants

closest to the treated group using observable variables. Both groups are matched on the basis
of
 18

propensity scores, predicted probabilities of participation given some observed variables.

Propensity score matching consist of four phases most commonly: estimating the probability
of participation, that is, the propensity score, for each unit in the sample; selecting a matching
algorithm that is used to match beneficiaries with non-beneficiaries in order to construct a
comparison group; checking for balance in the characteristics of the treatment and comparison
groups, along with estimating the program effect and doing sensitivity analysis (Caliendo and
Kopeinig, 2008). Propensity score matching (PSM) has two key underlying assumptions.
These are conditional independence (CI) and the existence of common support region (Baum,
2013).

Conditional Independence: It states that there exists a set of ‘X’ observable covariates such that
after controlling for these covariates, the potential outcomes are independent of treatment
status.

The Common Support: It states that for each value of ‘X’, there is a positive probability of
being both treated and untreated. It is used when creating a comparison group is possible from
a sample of non-participants closest to the treated group using observable variables.

2.6.2. Difference-in-difference (DID)

The difference in difference (DID) is designed for empirical analysis of causal effects, and has
a long history in and outside econometrics and is one of the most heavily used empirical
research designs to estimate the effects of policy changes or interventions in empirical
microeconomics nowadays (Lechner, 2011).

Difference in difference is a method in which we compare treatment and control group before
project (first difference) and after project (second difference). Comparators should be dropped
when propensity scores are used and if they have scores outside the range observed for the
treatment group. In this case potential participants are identified and data are collected from
them. However, only a random sub-sample of these individuals is actually allowed to
participate in the project. The identified participants who do not actually participate in the
project form the counterfactual (Jalan and Ravallion, 1999; Baker, 2000).

Panel based difference in difference estimator has both advantages and disadvantages. Among
its advantages, its intuitive appeal and simplicity can be mentioned. Additionally, estimates of
 19

DID are known to be less subject to selection bias because they remove the effect of any
unobserved time invariant differences between the treatment and comparison groups. It also
has disadvantages. First, panel-based estimation of DID can be expensive, time consuming,
and logistically challenging particularly because we need to collect baseline and follow-up
data that straddle the implementation of a program. Second, the design assumes that the
potential selection bias (i.e., due to administrative targeting or volunteering) is linear and time
invariant such that it can be subtracted off in the first differencing (Jalan and Ravallion, 1999).

However, these assumptions might be violated if the time period between two panel data sets
is long enough so that the unobservable variables of subjects are altered. In addition, the
unobservable variables can be changed as the subjects participate in the program which leads
the estimate to be biased (Tolemariam, 2010).

Difference in difference has the advantage that the idea is very intuitive and easy to
understand for an audience with limited econometric education. If compared with othr
matching methods, it has further advantage that there is no need to control all confounding
variables, since we have double difference. In many applications, time is an important variable
to distinguish the treated and control groups in difference in difference (Roberts and Lemmon,
2007).

2.6.3. Randomized selection methods (RSM)

Randomized selection methods (RSM) is the process of randomly selecting both group,
treatment and control, from clearly defined population to evaluate the outcome of the
intervention. Based on this, the control group is similar with the treatment group, and the only
difference is the participation in the required program. This method can do before and after or
pre and post matching and this leads to matching of variables that change due to participation.
Furthermore, randomization also does not require the untestable assumption of conditional
independence on observables (Abadie et al., 2004).

2.6.4. Regression discontinuity (RD)

Regression discontinuity (RD) method is one of the rigorous nonexperimental impact

evaluation approach that can be used to estimate program impacts in situations in which
candidates are
 20

selected for treatment based on whether their value for a numeric rating exceeds a designated
threshold or cut-point (Jacob et al., 2012). It allows us to account for observed and unobserved
heterogeneity. It initially assigns scores for the intervention unit and then compares the
outcome of individuals above the cut-off point with a group of individuals below it.
Individuals around the cut-off point are similar (Buddelmeyer and Skoufias, 2004).

Regression discontinuity is based on the cut-off point in observable characteristic, often called
the rating variable. RD techniques are considered to have the highest internal validity (the
ability to identify causal relationships), but their external validity (ability to generalize
findings to similar contexts) may be less impressive, as the estimated treatment effect is local
to the discontinuity (Baum, 2013).

The treatment is not randomized, but there is some process that deterministically dictates
whether a unit is treated or not, cut-off point. In this design, units receive treatment based on
whether their value of an observed covariate is above or below a known cut-off (Calonico et
al., 2013). But when using instrumental variable for causal inference, one must assume the
instrument is exogenously generated as if by a coin-flip (Lee and Lemieux, 2010).

For this study purpose, Propensity Score Matching (PSM) is selected and used for analyzing
the impact of small-scale irrigation on household food security for several reasons. Firstly,
there is no baseline data on participants and non-participants to compare before and after as it
is common in many researches that are conducted on impact evaluation. Second, the
participants in small scale irrigation are self-selected to participate on the program and this
leads to bias. Furthermore, the data to be used is cross-sectional survey data. Finally, it is
possible to identify some features, in these case socio-economic, institutional and physical
characteristics, to match the participants and non-participants.

2.7. Empirical Literature Review

2.7.1. Determinants of household participation in small-scale irrigation

Hirko et al. (2018), using probit regression of double hurdle model revealed that; market
information and access to credit are factors that positively affect participation in small scale
 21

irrigation and conversely; distance from market and distance from irrigation water source are
the factors that negatively affect participation in small scale irrigation.

Astatike (2016), used Heckman two-staged model and the probit model indicated that;
ownership of irrigation land, pumping motor and dissatisfaction with the existing irrigation
schemes are among the most determinants factors that influence irrigation participation.
Regassa (2015), using logistic regression model revealed that; age, sex, income, input use and
participation in cooperative organization are among the factors that significantly and positively
affecting irrigation participation. On the other hand, farm experience, distance to the district
market and total livestock holding significantly and negatively affected households’ decision
to participate in irrigation. Agerie (2013) using probit model found that distance from the
nearest market, education level of household head, total livestock holding, distance to water
sources, access to information, family labor, access to credit and gender of household head
were significantly affected participating in small scale irrigation scheme. The result of the
research conducted by Kuwornu and Owusu (2012) revealed that factors such as land size,
household size, market access and extension services are significant predictors of irrigation
participation

Haji et al. (2013) using logit model result found that dependency ratio, family size, size of
land holding and education of household head were significantly affected household
participation decision in small scale irrigation scheme.

The result of the research conducted by Kuwornu and Owusu (2012) revealed that years of
schooling, dependency ratio, size of livestock holding, frequency of extension contact, income
per capita, land covered by dam and distance from homestead to irrigation scheme
significantly determine the decision to participate in irrigation. Owusu et al. (2011) also
reported that farm land size and age of the household are significantly and positively
correlated with the decision to participate in irrigation.

2.7.2. Impacts of small-scale irrigation on household food security

According to Haileyesus (2019), Tobit model revealed that age of the household head and
crop disease are reported to affect food security negatively and conversely, access to
extension
 22

services, land holding size, livestock ownership and market access are among the factors
positively affected food security status of households.

Mengistie and Kidane (2016) found that irrigation has positive impact on well-being, asset
ownership and food security of rural smallholder farm households. Dawit and Balta (2015)
using a logistic regression model found small scale irrigation investment has appositive
impacts on household food security by reducing rainfall risk, farm household’s decision on
irrigation enabled households to diversify production to new types of marketable crops like
fruits, cash crops and vegetables, increased crops yield and marketing, reduced
unemployment, reduced inequality and poverty. Similarly, a study conducted by Hagos, et al.
(2012) also indicated that irrigation increased yields per hectare, income, consumption and
food security of the households.

Norton et al. (2014) indicated that investment in irrigation can increase farmers’ independence
on rainfall, increase irrigated farm land, generate employment and promotes farmers to
produce two or three times in a year, which is the major power for agricultural growth and
reduction of food insecurity.

According to Abera (2015) binary logistic regression model result showed that distance of the
irrigation scheme, number of livestock holding, total annual income, access to market
information, dependency ratio, age and sex of household head were found to be the
determinants of small-scale irrigation scheme use and the development of small-scale
irrigation positively impacted food security of beneficiary households.

2.7.3. Determinants of household food security

According to the studies conducted by Regassa (2011) and Bedeke (2012), ownership of
livestock, farmland size, family labor, off farm income, market access, use of improved
technology, education, health status, amount of rainfall and distribution, crop diseases, number
of livestock, and family size are identified as major determinants of household food security.

The study conducted by Torell and Ward (2010) revealed that participation in small-scale
irrigation improved farm households’ incomes, diet, health and food security. Sinyolo et al.
 23

(2014) using propensity score matching model indicated that irrigation access for smallholder
farmers was identified to be significantly and positively improving household welfare and
poverty reduction.

Abdissa et al. (2012) using multivariate logistic regression analysis indicated that dependency
ratio, household family size and market accessibility have showed significant and negative
effect on food security; whereas cultivable land size, access to irrigation, number of livestock
showed positive role for food security.

Fanadzo (2012) also reported that irrigation schemes in South Africa have increased
employment opportunities, and stabilized and increased rural wage rates; and increased family
consumption of food through enhancing food availability, reducing levels of consumption
shortfall, increasing of irrigation incomes and reducing food prices thereby ensure food
security.

The result of the study conducted by Beyene and Muche (2010) using binary logit model
showed that age of the household head, size of land cultivated, livestock ownership, soil and
water conservation practice and oxen ownership have positive and significant
relationship with household food security; whereas, education of household head, household
size and off- farm/non-farm income have negative and significant influence on household food
security.

Similarly, a study conducted by Bogale and Shimelis (2009) using binary model reveals that
age of household head, cultivated land size, livestock ownership, total income of the
household, irrigation and amount of credit receive have negative and significant effect on
household food security. Tesfay (2008) using PSM also showed that about 70% of the
irrigation users were food secure while only 20% of the non-users were food secure.

2.8. Conceptual Framework of the Study

Irrigation contributes to agriculture by improving production, generating income, poverty

reduction, employment creation and more. The income generation potential of irrigation is
affected by the crop grown using irrigation, market access to the products and supporting
institutional frameworks. Irrigation increase yields and lowers the risks of crop failures, which
 24

also help the farmer get income throughout the year. It also helps smallholders adopt diverse
cropping patterns and focus on high market value products (Akuriba et al., 2017).

Households having access to irrigation are able to increase crop production through harvesting
twice and more per year and also increase cash from increased production. Since water will be
available for crop year-round, the risk of crop failure is reduced. Since irrigation reduces risk
of crop failure, farmers get better income and they prefer to use high yielding varieties,
sufficient fertilizers and apply other relevant chemicals. Higher production and productivity of
high value crops lead farmers to fetch high income provided favorable market prices and
institutional arrangements are in place (Hagos et al., 2012).

Increased production and income of farmers enable them have increased access to food for
family. Furthermore, farmers who use irrigation also able to increase their asset base and
enable them to invest on other household assets and this farther contributes to the
improvement of households’ livelihood and make them safe in case of food shortage.
Moreover, other factors such as household resource endowments (land and livestock holding),
institutional services (agricultural inputs, credit and extension service), labor availability and
education level of household head can affect the overall success (Akuriba et al., 2017; Dittoh
et al., 2013).

In general, the conceptual framework showing how farm households can improve their
production through high value crops cultivation using irrigation, how they improve their
income, food security and livelihood of their family is shown by the following diagram:
 25

Figure 1 Conceptual Framework of the

Study Source: Own sketch.
 26

3. RESEARCH METHODOLOGY

Under this chapter, description of the study area, data type, source and method of collection, sampling
technique, method of data analysis, variable definition and hypothesis and others were covered.

3.1. Description of the Study Area

3.1.1. Demography

The study was conducted in Bishoftu Town. Bishoftu is among the eight districts of Finfinnee

surrounding Oromia special zones. The district is located at 47.9km away from Southeast of
Addis Ababa. The geographical location of the district is in between 8°50'-9°15' N and 38°25'-
38°45' E. It is bordered by Dukam district on the west, by Gimbichu district on the north west,
by Liban Chukala on the east and by Mojo town on the north. According to the census report
of CSA (2007), the population of Walmara district were 83,823 (42,115 male and 41,708
females). Similarly, according to the population projection report of CSA (2019), the
population of Walmara district were 197,557 out of which 93,631 male and the rest 103,926
were females, and the area of the district is 65,605 hectares.

Figure 2. Location map of Walmara district

Source: Adopted from Tiruneh and Geta (2016).
 27

3.1.2. Climate

The agro-ecology of the study area is classified into highland and midland. Majority parts (61
percent) of the district is highland, and the rest 39 percent is midland with an altitude ranging
from 2060 to 3380 meters above sea level. The mean altitude of the district is 2400 meters
above seas level. The annual rainfall of the district varies from 795mm to 1300mm, and its
average annual rainfall is 1144 mm. The annual temperature of the district also ranges from 6
ºc to 24ºc, and the average annual temperature is 14 ºc.

3.1.3. Agriculture

There are 7 Kebele administrations and two towns excluding the capital town of the district.
The farming system of the district is characterized by crop and livestock production similar to
other central highlands of the country. The major crops grown in the main season in the
district were wheat, barley, tef, pulses, oilseeds and potatoes respectively. These crops are the
major staple food crops according in the district. Moreover, the major vegetable crops grown
in the district using irrigation were potatoes, cabbages, tomatoes, carrots and onions
respectively.

3.1.4. Description of irrigation schemes in the district

Bishoftu Town has different streams and rivers that are suitable for irrigation activities. The
district has long history in both traditional and modern irrigation schemes. The total irrigable
land potential in the district is more than 9,055 hectares, out of which 7580 hectares were
cultivated yet under different types of irrigation. Out of the total irrigated land in the district,
828 hectares were cultivated using modern irrigation, 4890 hectares were cultivated using
traditional irrigation, 1788 hectares were cultivated using motor pump irrigation and 74
hectares were cultivated using wells, which in general benefitting about 2,163 households. The
irrigation user households mainly produce vegetables like potatoes, cabbages, tomatoes,
carrots, onions and others.

Since Bishoftu town is among the surplus producing districts and also among the high
irrigation potential districts of Oromia region, it was selected to be one of the agricultural
 28

growths and transformation program two (AGP II) districts and construction of modern
irrigation schemes in different kebeles were done by this program as shown in Table 1. This
study was conducted on three of these modern small-scale irrigation schemes found in
Talacoo, Barfata Tokkoffaa and Bakakkaa & Qoree Oddoo, and on one traditional irrigation
that is found in Waajituu Harbuu kebele.

The irrigation schemes of Barfata Tokkoffaa, Talacoo and Bakakkaa & Qoree Oddoo has the
capacity to irrigate more than 65, 94 and 65 hectares of land respectively. These schemes were
constructed by AGP II during 2008, 2005 and 2004 respectively and currently being managed
and administered by water user associations formed by the user members.

Table 1 Modern small-scale schemes in Walmara district

Kebeles Irrigable Constructed Programs Number of Beneficiaries

land (hec.) year (E.C) Constructing Male Female Total
Biftu 65 2008 AGP II 125 27 152
Hora 94 2005 AGP I 75 23 98
Daka Bora 30 2007 AGP II 339 68 407
Dirre 65 2004 AGP I 107 14 121
Erar 106 2005 AGP I & II 352 106 458
&2011
Kilole 30 2007 AGP II 103 29 132
ARSEDE 130 2011 AGP II 297 95 392
158 2011 AGP II 120 33 153
150 2011 AGP II 199 51 250

Source: Bishoftu Town office of agriculture

3.2. Data Type, Sources and Method of Collection

For this study, both primary and secondary data were collected. The primary data were
collected from selected sample households. Pre-tested and amended questionnaires were used
to collect the data. Selected researchers from National Agricultural Biotechnology Research
Center were trained and participated in data collection under close supervision of the
principal researcher.
 29

Secondary data were collected from different records of district’s office of agriculture, Kebele
administration offices, journals, websites and other relevant sources.

3.3. Sampling Technique and Sample Size Determination

The sample households were selected by using multi-stage sampling procedure. Bishoftu
Town was purposively selected. First, kebeles in the district were classified into highland and
mid highland based on their agro-ecology. There are 7 kebeles in the highland part and 12
kebeles in the mid-highland part. Second, kebeles in each agro-ecology were stratified into two
based on irrigation access (kebeles with irrigation access and without irrigation accesss).
Third, two kebeles from each agro-ecology, a total of four kebeles were randomly selected
from those kebeles having irrigation access. On the fourth stage, a total of 220
reprentative sample households (96 from highland agro-ecology and 124 from mid highland
part) were selected using systematic random sampling technique. Out of the total 220 sample
households, 121 were non-irrigators, while 99 were irrigators.

Irrigators are the households living in the selected kebeles who used any irrigation system
during 2018/19, and non-irrigators are those households living in the sample kebeles and who
did not participated in any irrigation activities within the same period. Irrigation non-users
were used as a control group to measure the impacts of small-scale irrigation on household
food security in this research.

Sample size were determined using the rule of thumb, which states that the minimum number
of requred sample size for a given research should be 10 times the number of important
independent variables included (10X, where X = number of important independent variables
included in the model) . Based on this rule, the required sample size is a function of relevant
independent variables included in the model, and is preferably ten times number of covariates
or more. Therefore, since the total number of independent variables included in the model
were 14, the minimum requred number of sample size will be 140. But, to minimize error and
increase the chance of getting counterfactual, the number of sample size for this research was
decided to be 220 including contingency for non respndents. Out of the total sample size, 121
households (55%) were irrigation non-users and 99 households (45%) were irrigation users.
 30

Samples from each kebele were selected based on proportional sampling and the sample
selected from each selected kebeles were proportional to the population in each kebele and
was determined using the following formula:

𝑛𝑖 =
(𝑁𝑖)
(𝑛) ……………………..……………………………………………………8
Ʃ𝑁𝑖

Where ni - the sample to be selected from ith kebele

Ni - the total population living in ith kebele.
ƩNi - the summation of population living in selected four kebeles
n - total sample size for the district

Table 2 Total sample distribution over the selected kebeles

Total Sample Non

Name of Kebeles Participant Proportion
household selected participant
Hora 544 60 27 33 27.3%
Biftu 777 86 39 47 38.9%
arsede 325 36 16 20 16.3%
kilole 349 38 17 21 17.5%
Total 1995 220 99 121 100%

Source: District office of agriculture and own computation result

3.4. Methods of Data Analysis

Both descriptive and econometric data analysis methods were used.

3.4.1. Descriptive statistical analysis

The data analysis was done using descriptive and inferential statistics. Means, percentages,
standard deviation, maximum, minimum, and variance were used for descriptive statistics. T-
test and chi-square tests were performed to see the statistical significance of the continuous
and dummy variables, respectively. STATA V15.3 was the software used to analyze the
collected data.
 31

3.4.2. Econometric analysis

3.4.2.1. Factors affecting household participation in small-scale irrigation

Farmer’s participation decision in small-scale irrigation practice is dependent variable. It is a

dichotomous (binary) which takes a value of one if the household is irrigator and zero if the
household is non-irrigator. To see the participation decision of the households, logit or probit
models can be used as the result of the two models are similar (Green, 2003). So, logistic
regression model was chosen in this research to assess factors affecting farmers’ participation
decision in small-scale irrigation. The logit model used to estimate factors affecting
households’ participation decision is expressed using the following equation:

……………..…… 9

Pi
Where: 1−𝑃 = is the odds ratio in favor of participation in irrigation and household food
𝑖
security. The ratio of irrigators to non irrigators.
Pi = is the probability of being food secure, ranging from 0 to 1
Zi - is a function of “n” explanatory variables
βo - is constant.
βi = coefficients to be estimated
Xi = explanatory variables deter
i = observations
ui = error term.

3.4.2.2. Factors affecting household food security

To assess factors affecting household food security, logit model was used again to model the
relationship between the dichotomous dependent variable and set of independent variables that
are hypothesized to affect the outcome (Wooldridge, 2010). The result of this estimation
shows the effect of each explanatory variable on its log of odds ratio. Positive coefficients are
expressed as “the odds ratio is increased as the explanatory variables are increased and the
odds ratio will
 32

decrease as explanatory variables are decreased. The coefficients of logistic regression are
estimated using the maximum likelihood estimation method. Food security is a dummy
variable in its nature and different factors like demographic, socio-economic and other factors
are hypothesized to influence it as shown in the table presenting the summary of variables.

3.4.2.3. Impact of small-scale irrigation on household food security

Propensity Score Matching (PSM) is used to compare irrigation user households (treatment
group) and non-irrigation user households (control group) lying in the common support
region. Propensity Score Matching (PSM) is a statistical matching technique that estimates the
effect of treatment or intervention given covariates. It allows a comparison group from a
sample of non- participants closest to the treatment group in terms of observable
characteristics so that both groups are matched on the basis of the propensity score, which is a
predicted probability of participation given observed characteristics (Ravallion, 200).
Propensity value is estimated using logit or probit, and used to estimate the average treatment
effect of the outcome in absence of baseline data using observable variables (Abadie et al.,
2004).

Propensity score matching (PSM) is used in this study for different reasons i.e., there is no
baseline data to see the difference between before and after, there may be self-selection bias as
small-scale irrigation participant households may be self-selected to participate, the cross-
sectional survey data is used for matching the participant and non-participant groups, and etc.
PSM controls for self-selection bias by creating a statistical comparison group by matching
every individual observation of the treatment group with individual observations from the
control group with similar observable characteristics.

These groups are matched based on the propensity scores, which is the predicted probability of
participation given some observed variables. Even though PSM has many advantages, it has
limitations such as requiring large samples, lack of common support region and hidden bias
since matching controls only for the observables.

The Propensity score is defined as the probability of receiving treatment based on measured
covariates.
E(x) = P (D=1 | X)....................................................................................................................10
 33

where E(x) = propensity score,

P = probability, D=1 a treatment indicator with values 0 for control and 1 for treatment
"|" = is a symbol stands for conditional on (predicted), and
X = is a set of observed covariates.

As stated in Caliendo and Kopeinig (2008), there are five steps of performing Propensity
Score Matching. These are: Estimating PS, choosing matching algorism, restricting common
support region, balancing test and sensitivity analyzing. Each five steps are described as
follows:

Step 1: Estimation of Propensity Scores

A Logit model is used to estimate PS for both participants and non-participants. Using logit
model has an advantage since the probabilities are limited between zero and one. The
dependent variable takes the values of one if an individual is participating in small scale
irrigation and zero otherwise.

Step 2: Choosing the Matching Algorism

The second step is choosing the matching algorithm that match the estimated propensity score
to the best level from the Nearest Neighbor Matching, Caliper and Radius Matching,
Stratification and Interval Matching, Kernel and Local Linear Matching (Caliendo and
Kopeinig, 2008). Choosing matching method comprises a tradeoff between matching quality
and its variance. The matching algorithm which has high matching sample size, low R 2 value
and high matching balance is selected among the methods (Dehejia and Wahba, 2002).

The Nearest Neighbor Matching Method: For each control unit, the nearest neighbor
matching method assigns a weight of one for the nearest comparison unit and zero to the other
comparison observations. So, a single comparison unit can be used as a matching for more
than one control unit.

The Radius Matching Method: The nearest neighbor matching method may face the risks of
bad match when the closer neighbor is far away. This problem can be corrected by adjusting
the tolerance level to the maximum propensity score distance (caliper). By using this method,
the
 34

bad match will be adjusted and the matching quality rises. This is another way of imposing the
common support region.

Kernel Matching Method: In all matching algorithms discussed yet, few observations from
the comparison group is used to construct the counterfactual outcome (unobservable outcome)
of the treated individual. But, Kernel matching (KM) is non-parametric matching method
which uses weighted averages of all individuals in the control group to construct the
counterfactual outcome. The choice of matching estimator depends on available data set, and
there is no winner for all cases and matching estimator choice mainly depends on the situation
at hand. It is widely accepted that a good matching estimator is the matching which gives the
lower pseudo R2, statistically insignificant likelihood ratio test after matching and the
matching with larger observations (Caliendo and Kopeinig, 2008; Leuven and Sianesi, 2006).

The choice of the matching algorithms will be based on the most important tests to reduce the
bias and inefficiency simultaneously. These tests include mean bias, number of matched
samples, value of pseudo-R square, and number of the balanced covariates. When considering
the mean bias, the one with lowest mean bias is better matching algorithm. Based on number
of samples matched, the one with the highest matched number of observations is the best and
selected. When coming to the value of the pseudo-R square after matching, the matching
algorithm with the lowest pseudo-R square is the best matching algorithm. On the other hand,
the matching algorithm with the highest number of balanced covariates is more appropriate
(Dehejia and Wahba, 2002).

Step 3: Restricting Common Support Region

Identifying the common region is a critical step. In fixing the common support region, two
guidelines might be helpful to do it more precisely; the first is comparing the maxima and
minima of the p-score in both groups and the second is estimating density distribution in both
groups. The overlap condition is the area which contains the maximum and minimum
propensity scores of control and treatment group. It ensures that any combination of
characteristics observed in control group also observed in the treatment group (Caliendo and
Kopeinig, 2008).
 35

Step 4: Testing Matching Quality or Balancing Test

When using PSM method, balancing test is very important. The quality of matching depends
on the ability of the procedure to balance the relevant covariates. Rosenbaum and Rubin
(1985) proposed a standardized bias which is commonly used method to quantify the bias
between control and treated groups. The comparison of the pseudo R2 before and after
matching, in which the value of pseudo R2 after matching should be lower because of the
matching use those households that have similar characteristics which mean that no significant
difference of covariate of treated and the control group is also proposed. In other word, the t-
test value of all covariate after matching is insignificant (Leuven and Sianesi, 2006).

It helps us to evaluate the impact of treatment on the treated groups. It is the difference
between the outcomes of treated and the outcomes of treated observations had they not been
treated (counterfactual) computed as:

ATT = E (Yi1 - Yi0 | D = 1)........................................................................................................11

E (Yi1 | D = 1) - E (Yi0 | D = 1)..................................................................................................12

Step 5: Sensitivity Analysis

The main question needs to be answered in sensitivity analysis is how strongly an unmeasured
variable influences the selection process in order to undermine the implication of matching
analysis (Caliendo and Kopeinig, 2008). Hence, sensitivity analysis was undertaken to detect
the identification of CIA (conditional independency assumption) is satisfactory or affected by
the dummy confounder.

3.5. Variables Definitions and Hypothesis

3.5.1. Dependent variable

Participation in Small Scale Irrigation (partirrig): It is a dummy variable taking a value

one for irrigation users and zero otherwise. Irrigation user is a household owning land, rented
in, shared in, shared out or obtained land through gift for irrigation during 2018/19 and non
user otherwise.
 36

Household food security (calsec): This is also dummy variable of one if the HH is food
secure and zero otherwise. It represents daily calorie intake of the household per adult
equivalent.
3.5.2. The outcome variables
Daily Calorie Intake per Adult Equivalent (calorieAE): It is a continuous variable
measured in kilocalorie. To measure it, the types and amounts of food items consumed by
each household during the last 24-hour were collected first. Second, the calorie contents of
consumed food items were calculated using calorie conversion factor (WFP, 2017). Thirdly,
the total calorie consumption was divided by family size of adult equivalent (AE) and
compared with internationally accepted minimum energy requirement of 2200 kilocalories per
person per day (Sassi, 2018). Similarly, according to FDRE (2010), 2200 kilocalorie is the
minimum nationally recommended energy requirement per adult per day, and is used as a cut-
off point for deciding that the household is food secure or not. Therefore, the households
above this threshold are taken as food secure and insecure otherwise.
Food Consumption Score (fcs): It is a continuous variable measuring the types of food
consumed by the household and also the frequencies of the food consumed within seven days.
In measuring it, first the type and frequency of consumed food were collected. Second, the
consumed foods were grouped into specific food groups. Third, the consumed food under
groups was added up. Fourth, the value under each group was multiplied by the weight given
to each food group. Fifth, all are added together and the food security profile of the household
was decided. Accordingly, the households with score value less than 28 were considered as
food insecure, those whose food consumption score value between 28 and 42 were considered
as borderline, and food secure if their score values are greater than 42 (WFP, 2008).
3.5.3. Independent variables
Independent variables are the variables affecting the dependent variables, participation in
irrigation and household food security, particularly in this research. This effect is a combined
effect of various factors such as demographic, socio-economic and institutional factors. The
independent variables that are proposed to affect the dependent variable are defined and
hypothesized as follows:
 37

Sex of the Household Head (Sexhead): Sex is a dummy variable with values of one if the
household head is male and zero otherwise. As identified by scholars, Haji et al. (2013) and
Hadush (2014), male headed households are more likely irrigation participants compared to
female headed households. According to the findings of Sami and Kemaw (2019), male headed
households are more food secure than female headed households. Therefore, this study
hypothesized that male household heads are more likely to participate in irrigation and to be
more food secure than female headed households.
Age of the Household Head (Agehead): Age is a continuous variable measured in years. In
Ethiopia, the household is the decision maker for farm activities. According to Haileyesus
(2019), keeping other things constant, household age was negatively related with food
security. According to Daniel (2008), age of the household head is negatively related to
participation in irrigation. Thus, age of the household head is hypothesized to negatively affect
participation in irrigation and food security.
Educational Status of the Household Head (Educhead): This variable refers to the formal
schooling grades that the household head completed. Human capital is assumed to be a key
source of income, growth and an important block for building wellbeing. Education is
assumed to improve the attitudes and awareness of farmers towards the adoption of new
agricultural technologies, which is irrigation in the context of this work. Ayele et al. (2013),
Etwire et al. (2013) and Kinfe et al. (2011) confirmed that education level of irrigation
participants is higher than that of non-participants. According to Habtewold (2018),
educational status of the household head is positively related with food security. Thus,
education status of the household head is hypothesized to positively affect participation in
irrigation and household food security.
Dependency Ratio (Dependratio): It is a continuous variable which refers to the ratio of
economically inactive labor force, child and elders (below 15 and greater than 65 years), to the
labor force that is active (from 15 to 65 years) existing in the household. As the economically
inactive individuals are dependent on the active family members to fulfil their food demand, it
has negative relation with food security of the household. According to Kuwornu and Owusu
(2012), dependency ratio is identified as the factor that negatively affects participation in
irrigation. According to Abdissa et al. (2017), dependency ratio is reported to negatively affect
 38

household food security. Therefore, this variable is hypothesized to negatively affect irrigation
participation and food security status of the households.
Livestock Holding (LivstokTLU): It is a continuous variable representing the number of
livestock measured in Tropical Livestock Unit (TLU) owned by the HH. Livestock is an
important source of income, food and draft power and represents an asset, which indicates the
wealth and social status of the household and a source of finance. Hadush (2014), and
Kuwornu and Owusu (2012) reported that more livestock holding households are more likely
to participate in small-scale irrigation. Similarly, according to Abdissa et al. (2017), livestock
holding was reported to positively affect household food security. Therefore, livestock holding
is hypothesized to positively affect participation in irrigation and food security of the
households.
Household Size (famsizeAE): It is a continuous variable measured in adult equivalent.
Households having more adult equivalent household size are expected to adopt agricultural
technologies, irrigation in this context, than those having less household size since improved
agricultural technologies require more labor. The findings of the research conducted by
Tadesse et al. (2013) indicated that family size is positively related to participation in small-
scale irrigation. According to the findings reported by Abdissa et al. (2017), family size is
negatively related to household food security status. Thus, family size is hypothesized to
positively related with irrigation participation and negatively related with household food
security status.
Non-farm Income Source (Offarincom): This variable is a continuous variable which takes a
value of one if the household participates in non-farm activities and zero otherwise. It
represents income out of agriculture like from trade, salary, wage, remittance, etc. As reported
by Molla (2017) and Hadush (2014), non-farm income earners are more likely to participate in
small scale irrigation than non-earners. According to Habtewold (2018), access to off-farm
income is positively related to household food security status. Thus, non-farm income is
hypothesized to positively affect participation in irrigation and household food security.
Extension Contacts (Extncontact): This variable is a discrete variable measured by whether
or not the household has made contacts with extension agent per month. The contact
frequency between the farmers and the extension agents will potentially accelerate the
dissemination of
 39

information to the farmers. As reported by Sinyolo et al. (2014) and Abdissa et al. (2017),
extension service is positively related with irrigation participation. Moreover, according to
Habtewold (2018), access to extension service is positively related to household food security
status. Hence, this variable is hypothesized to positively affect the farmers’ decision of
participation in irrigation and food security status.
Distance from the Nearest Market (Mrktdist): This is a continuous variable measured in
kilometers. It affects the probability of participation in irrigation. Asayehegn (2012) reported
that the probability of participation in irrigation for the households with good market
information is more than two-fold of the participation probability of those households that are
far from information. Moreover, when farmers are far away from the market, the income that
they get from their farm product is lower due to lack of market opportunity and transportation
facilities and the farmers’ participation in irrigation will be less (Muez, 2014). According to
Abdissa et al. (2017), market distance was reported to have negative relation with household
food security. Thus, market distance is hypothesized to negatively affect participation in
irrigation and household food security.

Distance from Irrigation Water Source (Distirrig): This variable is a continuous variable
measured in kilometers. Sinyolo et al. (2014) reported that distance of irrigation water source
affects households’ participation in irrigation negatively. The nearer the household to the
source of irrigation water, the higher the probability the farmer will participate in irrigation.
This is due to the fact that the operation cost and time lost in travelling will be reduced.
According to Abdissa et al. (2017), households living nearer to irrigation water sources are
more food secure compared to those farmers that are far from irrigation water source. Hence,
distance to irrigation water source is hypothesized to negatively affect participation in
irrigation and food security.

Access to Irrigation (Irrigacces): Access to irrigation is a dummy variable of value one if a

household used irrigation and zero otherwise. Participation in irrigation would enable farmers
to produce twice or more and reduce rain dependency. So, farmers participating in irrigation
are likely to increase production, income and consumption than non-participant farmers.
Fanadzo (2012) and Abdissa et al. (2017) reported that irrigation user households are more
food secure
 40

than non users. Hence, access to irrigation is hypothesized to positively affect food security of
the households.

Access to Credit Service (Creditacces): It is a dummy variable which is a vital source of

investment. Households having credit access are more likely to use inputs like fertilizer,
improved seed, agricultural chemicals and adopt agricultural technologies. According to
Regassa (2015), households with access to credit and members in cooperative unions are more
food secure. According to the research finding reported by Wakene (2018), households with
access to credit services are more likely to participate in irrigation. Therefore, access to credit
is hypothesized to positively affect households’ decision of participation in irrigation and their
food security status.

Size of Farm Land (Landown): is a continuous variable that determine household food
security. It is a total size of land on which food and cash crops are cultivated by households,
measured in hectares. Abraham et al. (2015) identified that size of cultivated land has a
positive effect on household participation in small scale irrigation. Similarly, Abdissa et al.
(2017) reported that households having more land are more food secure than those having less
land. Thus, this variable is hypothesized to have positive impact on household food security
and participation in small-scale irrigation.

Pests and Diseases (Cropdiseas): These variables are dummy variables of value one if pests
and diseases were occurred during the main season of the year under investigation and zero
otherwise. Plant pests and diseases potentially affects production and productivity of the
farmers. According to FAO and IPPC (2017), plant pests and diseases are responsible for
reduction of 20 to 40 percent global food production. Therefore, households whose
agricultural production was affected by pest and disease are more likely to be food insecure
than those households that are not affected by pests and diseases. According to Abdissa et al.
(2017), infestations of crop pests and diseases negatively affects household food security. So,
pests and diseases are hypothesized to negatively affect household participation in irrigation
and food security status.
 41

Table 3 Summary and hypothesis of explanatory variables

Nominated Type of Unit of Effect on Effect on

Variable Definition
Symbol Variable measure participation Food S.
Sex of the household head Sexhead Dummy 1 and 0 + +
Age of the household head Agehead Continuous Years - -
Education status of HHH Educhead Continuous Years + +
Dependency ratio Dependratio Continuous AE - -
Livestock holding LivstockTLU Continuous TLU + +
Family Size FamilsizeAE Continuous AE + +
Non-farm income earned Offarmincom Continuous Birr + +
Extension contacts Extncontact Dummy 1 and 0 + +
Distance to nearest market Mrktdist Continuous Kilometer - -
Distance of irrigation site Distirrig Continuous Kilometer - +/-
Access to irrigation Irrigacces Dummy 1 and 0 +
Access to credit service Creditaccess Dummy 1 and 0 + +
Size of Farm Land Landown Continuous Hectares + +
Pests and Diseases cropdiseas Dummy 1 and 0 - -
 42

4. RESULTS AND DISCUSSION

Under this chapter, analysis of the collected data and interpretation of the results were
addressed. Descriptive statistics and econometric results of different factors were discussed in
this chapter.

4.1. Descriptive Results for Demographic and Socio-economic Variables

Under this topic, descriptive results of dummy and continuous variables were addressed for
both participations in irrigation and household food security.

4.1.1. Descriptive results of dummy variables

4.1.1.1 Participation in irrigation

Generally, out of 220 sample households, 99 (45 percent) were irrigators and 121 (55 percent)
were non-irrigators. As presented in Table 4, sex, access to credit services and occurrence of
crop disease were the variables that significantly related with participation in irrigation.

Sex of the household head: The result showed that 19.8 percent of non-irrigators and 11.11
percent of irrigators were female headed, while 80.2 percent of non-irrigators and 88.89
percent of irrigators were male headed households. This indicates that male headed households
are more irrigators compared to female headed households. The Chi-square test result showed
that the mean difference was statistically significant at 10 percent probability level.

Access to credit services: The result in Table 4 ris showing that 95.96 percent of households
having access to credit services and 85.96 percent of the households that not having access to
credit services are irrigation users. This indicates that households getting access to credit
services are more irrigators than households not having access to credit services. The chi-
square test also showed that the mean difference was significant at 5 percent.

Pests and diseases: According to the survey result, 57.6 percent of households whose crop
field affected by pests and diseases and 43.8 percent of the households whose crop field did
not affected were irrigators. This indicates that farmers whose crop affected by diseases during
the main season are more irrigation users. The chi- square test also indicated that the mean
difference is statistically significant at 5 percent.
 43

Access to extension services: From the result in Table 4, access to extension services was
almost similar for both irrigators and non-irrigators. 86.86 percent of irrigators and 85.95
percent of non-irrigators get access to extension service. The chi- square test also indicated
that the mean difference was not significant.

Table 4 Descriptive results for dummy variables (Participation in irrigation)

Dummy variables
Dependent
Sex of the head Credit access Crop disease Extension contact
Irrigation use Fem Male Total No Yes Total No Yes Total No Yes Total
No 24 97 121 17 104 121 68 53 121 17 104 121

Yes 11 88 99 4 95 99 42 57 99 13 86 99
Total 35 184 220 21 199 220 110 110 220 30 190 220
Percent 16 84 100 10 90 100 50 50 100 13.6 86.4 100
chi2(1) 3.097 6.3178 4.132 0.039
Pr. 0.078* 0.012** 0.042** 0.843

Note: * and ** shows that the variables are significant at 10% and 5% significance levels
Source: Computed from own survey data of 2020.

4.1.1.2 Household food security status

As indicated in Table 5, all the variables; sex, access to credit services, occurrence of crop
diseases and access to extension contact were significantly related with household food
security.

Sex of the household head: The result of the survey revealed that 68.6 percent of female
headed households and 92.4 percent male headed households were food secure. This indicates
that male headed households are more food secure compared to female headed households.
The Chi- square test result showed that the mean difference was statistically significant at 1
percent probability level.

Access to credit services: The result in Table 4 indicated that 895.9 percent of households
getting access to credit services and 76.1 percent of the households that not having access to
credit services are food secure. This indicates that households getting access to credit services
 44

are more food secure than households not having access to credit services. Chi square test also
showed that the mean difference was significant at 10 percent significance level.

Crop diseases and pests: According to the survey result, 82.7 percent of household heads
whose crop field affected by pests and diseases and 94.5 percent of the households whose crop
field did not affected by pests and diseases were food secure. This indicates that farmers
whose crop fields were affected by diseases and pests are less food secure than households
whose crop fields did not affected. The chi- square test also indicated that the mean difference
is statistically significant at 1 percent probability level.

Access to extension services: The result revealed that, 90.5 percent of household heads
having access to extension contact and 76.7 percent of the household heads that not having
extension contact was food secure. This indicates that households having access to extension
contacts are more food secure than that of the households not having access to extension
contacts. The chi- square test also showed that the mean difference was statistically significant
at 5 percent probability level.

Table 5 Descriptive results for dummy variables (Household food security status)

Dummy variables
Dependent
Sex of the head Credit access Crop disease Extension contact
Calorie security Fem Male Total No Yes Total No Yes Total No Yes Total
No 11 14 25 5 20 25 6 19 25 7 18 121

Yes 24 171 195 16 179 195 104 91 195 23 172 99

Total 35 185 220 21 199 220 110 110 220 30 190 220
Percent 16 84 100 9.5 90.5 100 50 50 100 13.6 86.4 100
chi2(1) 16.636 3.57 7.626 4.941
Pr. 0.000*** 0.059* - 0.006*** 0.026**

Note: *, ** and *** shows the variables are significant at 10% 5% and 1% significance
levels Source: Computed from own survey data of 2020.
 45

4.1.2. Descriptive results of continuous variables

4.1.2.1.Participation in irrigation

According to the results presented in Table 6, participation in irrigation was significantly

affected by age of the household head, livestock ownership, family size, distance from
irrigation site and the size of land owned.

Age of the household head: The result in Table 6 showed that the average age for irrigators
and non-irrigators were 42.3 and 45.4 years with standard deviations of 10.4 and 9.2
respectively. This is to mean that younger households are more irrigation users compared to
older household heads. The t-test result also showed that the mean difference is statistically
significant at 5 percent probability level.

Educational status of the household head: According to the survey result, the mean
educational level for irrigation participants and non-participants were 4.5 and 3.7 years of
schooling with standard deviations of 4.0 and 3.7 respectively. This is to mean that more
educated household heads are more irrigation users compared to less educated households. But
the t-test showed that the mean difference was not statistically significant.

Dependency ratio: Dependency ratio is a proportion of family members whose age is less
than 15 years and those whose age is above 65 years. As shown in Table 6, the average
dependency ratio of the sample households was about 1 person with standard deviation of 0.7.
This means, on average, about one dependent family member exists per household in the study
area. The mean dependency ratio for both participant and non-participant households were 0.9
(about one person), with standard deviations of 0.6 and 0.8 respectively. The t-test result
revealed that there was no significant mean difference between irrigation user and non-users.

Livestock holding (TLU): The result of both focus group discussion and questionnaire
revealed that livestock is the most important asset in the study area. Livestock is mainly
serving as the main source of draft power, means of transportation for farmers themselves and
for their farm products, source of income to cover different expenses, sources of different
products and bye- products for home consumption and sale, and etc. Especially, livestock
serves as the main
 46

coping mechanism for majority of the farmers during food shortage periods. Households
purchase food by selling livestock and livestock products during the time they face food
shortage. The result showed that the mean livestock holding for irrigators and non-irrigators
were 6.3 and 7.5 with standard deviation 3.7 and 4.1. According to this result, households
having a smaller number of livestock are more irrigation users than those having larger
livestock. The t-test result showed that the mean difference was significant at 5 percent
probability level.

Family size AE: According to the survey result the average adult equivalent family size for
irrigation users was 4.9 (about five persons per household) with standard deviation of 2
persons, while it was 4.4 (about four persons per household) with standard deviation of 1.8
(about 2 persons) for irrigation non-users. On average, the family member for irrigation
participants were larger than that of irrigation non-participants. The t-test result also showed
that the mean difference was statistically significant at 5 percent probability level.

Off farm income: The main types of off-farm income in the study area are petty trade,
temporary employment (wage) selling firewood or charcoal, selling local beverages, animal
driven cart services, permanent employment, hand craft services, and etc. The result in Table 6
revealed that the mean off-farm income for irrigators and non-irrigators were 7,164 Birr and
35,900 Birr with standard deviation of 9,671 and 10,383 respectively. From the result, average
off-farm income of non-irrigators was larger than that of irrigators. But the t-test result
revealed that the mean difference was not statistically significant.

Market distance: The mean market distance for irrigation participant and non-participant
households were 5.2 and 5.5 kilometers, with standard deviations of 2.3 and 2.1 respectively.
From the result, the average market distance for irrigation non-users was longer than that of
irrigation users. But the t-test result revealed that the difference was not statistically
significant.

Distance from irrigation site: This variable is one of the variables whose mean for irrigation
participants and non-participants showed significant differences. The mean distance of
irrigation site for irrigation users and non-users were 1.9 and 2.3 with standard deviation of 0.8
and 0.7 respectively. The average distance of irrigation water source was longer for irrigation
non-users than irrigation users, and the t-test result in Table 6 also showed the difference was
statistically significant at 1 percent probability level.
 47

Land owned: The results of the data collected through both focus group discusion and
questionnaire revealed that land is the most important asset that highly affecting both crop and
livestock production in the study area. The size of land owned by the sample households were
varied from 0 to 4.5 hectars. As presented in Table 6, the mean land holding of the sample
households was 1.5 hectars, with standard deviation of 1.2. Similarly, the average land holding
for irrigation participnats and non-participants were 1.7 and 1.4, with standard deviation of 0.2
for both. In general, irrigation users hold more land compared to non-users on average, and the
t-test also showed that the mean difference was statistically significant at 5 percent probability
level.

Table 6 Summary of continuous variables (Participation in irrigation)

Participants Non-participants Combined

Variables T-
value mean St.dev mean St.dev mean St.dev
Age of head (years) 42.3 10.4 45.4 9.2 43.9 9.9 -2.47**
Education of head (years) 4.5 4.0 3.7 3.7 4.1 3.8 1.38
Dependency 0.9 .62 0.9 0.8 0.91 .71 -0.31
Livestock (TLU) 6.3 3.7 7.5 4.1 7.0 3.9 -2.23**
Family size (AE) 4.9 2.0 4.4 1.8 4.6 1.9 2.65**
Off-farm income (Birr) 7164 9671 8090 10383 7673 10057 0.67
Market distance (KM) 5.2 2.3 5.5 2.1 5.4 2.2 -1.30
Irrigation distance (KM) 1.9 0.8 2.3 .7 2.2 .8 -3.31***
Land owned (hectares) 1.7 1.2 1.4 1.2 1.5 1.2 2.25**

Note: ** and *** is showing the significance level at 5% and 1% significance levels
Source: Computed from own survey data of 2020.

4.1.2.2. Household food security status

As indicated in Table 7, age, educational level, dependency ratio, livestock holding, family
size, market distance, distance from irrigation site and size of land owned were the variables
that showed significant mean difference for food secure and non-secure households.
 48

Age of the household head: The result showed that the average age for food secure and non-
secure households were 43.3 and 48.9 years with standard deviations of 9.75 and 9.7
respectively. From the result, the mean age of food insecure households was higher than that
of food secure. This is to mean that age is negatively related to household food security status.
The t-test also showed that the mean difference was statistically significant at 5 percent.

Educational status of the household head: From the result, the average education level for
food secure and non-secure households were 4.3 and 2.3 years with standard deviations of 3.9
and 2.9 respectively. From the result, the mean age of food secure households was higher than
that of food insecure. This is to mean that education is positively related to household food
security. The t-test result showed the mean difference was statistically significant at 5 percent.

Dependency ratio: The mean dependency ratio for both food secure and food insecure
households were 0.8 and 1.5 with standard deviations of 0.68 and 0.7 respectively. From the
result, the mean dependency ratio for food secure households was lower than that of food
insecure. This is to mean that dependency ratio is negatively related to household food
security. The t-test result also showed the mean difference was statistically significant at 1
percent.

Livestock holding (TLU): The result showed that the mean livestock holding for food secure
and insecure households were 7.2 and 5.5 with standard deviations of 4.1 and 2.1 respectively.
This is to mean that livestock holding is positively related to household food security. The t-
test result also showed the mean difference was statistically significant at 5 percent.

Family size AE: The mean family size for food secure and insecure households were 4.5 and
5.9 with standard deviations of 1.8 and 2.5 respectively. From the result, the mean family size
for food secure households was lower than that of food insecure. This is to mean that family
size is negatively related to household food security. The t-test result also showed the mean
difference was statistically significant at 1 percent.

Market distance: The mean market distance for food secure and insecure households were
5.3 and 6.3 respectively. The mean market distance for food secure was shorter than that of
insecure. This is to mean that market distance is negatively related to food security status.
Similarly, the t-test result also showed that the mean difference was statistically significant at
5 percent.
 49

Distance from irrigation site: The mean irrigation distance for food secure and insecure
households were 2.1 and 2.6 respectively. The mean irrigation distance for food secure was
shorter than that of insecure. This is to mean that irrigation distance is negatively related to
food security status. Similarly, the t-test result also showed that the mean difference was
statistically significant at 5 percent.

Land owned: The result showed that the mean land holding for food secure and insecure
households were 1.5 and 0.6 with standard deviations of 1.2 and 0.6 respectively. This is to
mean that land holding is positively related to household food security. The t-test result also
showed the mean difference was statistically significant at 1 percent.

Table 7 Summary of continuous explanatory variables (Household food security)

Food secure Food insecure Combined
Variables T-value
mean St.dev mean St.dev mean St.dev
Age of head (years) 43.3 9.75 48.9 9.7 43.9 9.9 -2.7**
Education of head (years) 4.3 3.9 2.36 2.9 4.1 3.8 2.37**
Dependency 0.82 .68 1.5 0.7 0.9 .7 -4.6***
Livestock (TLU) 7.2 4.1 5.5 2.1 7.0 4.0 2.04**
Family size (AE) 4.5 1.8 5.9 2.5 4.6 1.9 -3.7***
Off-farm income (Birr) 7508 10160 8856 9298 7673 10057 -0.67
Market distance (KM) 5.3 2.2 6.3 2.2 5.4 2.2 -2.16**
Irrigation distance (KM) 2.1 0.8 2.6 .8 2.2 .8 -2.61**
Land owned (hectares) 1.5 1.2 0.6 0.6 1.5 1.2 3.9***

Note: ** and *** is showing the significance levels at 5% and

1% Source: Computed from own survey data of 2020.

4.1.3. Summary of food consumption score by participation in irrigation

Food consumption score is one of the outcome variables used to measure food security status
of sample households in this study. Table 8 shows that sample households that are irrigation
users and rest on acceptable food consumption score category were 38.18 percent while
those who
 50

practice irrigation and rests on borderline food consumption score category were 6.82 percent.
Moreover, households who are not irrigation users and rest on acceptable, borderline and poor
food security profile were 31.82%, 21.82% and 1.36% respectively. These figures show us
that food security status of irrigation users is better than that of irrigation non-users. The p-
value is 0.000, showing that there is significant relation between food consumption score
categories and irrigation participation since the p value is significant at 1 percent significance
level.

Generally, the food security status of the sample households is questionable based on food
consumption score method in which only 70.00, 28.64 and 1.36 percent of the sample
households were in the range of acceptable, borderline and poor food security profile ranges
respectively. These figures show us that 70 percent of the sample households in the study area
were food secure, and the rest (about 30 percent) were food insecure. Therefore, from this
descriptive result, the researcher believes that the food security status of the sample
households was not good since the data were collected during peak season of food availability.

Table 8. Food consumption score category by participation in small scale irrigation

Food Consumption Irrigation Status

Total
Score Category Non-users Users
Acceptable 70 (31.82%) 84 (38.18%) 154 (70.00%)
Borderline 48 (21.82%) 15 (6.82%) 63 (28.64 %)
Poor 3 (1.36%) 0 (0.00%) 3 (1.36%)
Total 121 (55.0%) 99 (45.00%) 220 (100 %)

Pearson chi2(2) = 19.5540 Pr. = 0.000

Source: Computed from own survey data of 2020.
4.1.4. Summary of daily calorie intake by participation in irrigation

Daily calorie intake is one and the most preferred method of measuring food security at
individual, household and national level. It enables us measure direct calorie intake at
individual
 51

level, food availability and dietary quality. It also helps us understand the patterns of intra-
household food consumption, e.g., how it can be influenced by gender.

As shown in Table 9 below, the proportion of sample households that are irrigation
participants and food secure are 42.73 percent and those who are food insecure are 2.27
percent. Moreover, the sample households that are irrigation non-participants that are food
secure and insecure are
45.91 and 9.09 percent respectively. Based on daily calorie intake measure of food security,
majority of the sample households (88.64 percent) are food secure, while only 11.36 percent
are food insecure, provided that the data was collected during the peak season of food
availability. The chi-square result showed that there is significant difference between mean
food security of irrigators and non-irrigators since the p-value is 0.008, which is significant at
1 percent.

Generally, when we compare food security status of the sample households measured by food
consumption score and daily calorie intake method, the result obtained by the later method
showed a better food security status of the sample households. Based on the descriptive result
obtained using daily calorie intake method, the researcher believes that the food security status
of the sample households was in a good state for the season at which the data was collected.

Table 9 Daily calorie intake by participation in irrigation

Food security status Irrigation Status

measured in Daily Total
calorie intake Non-users Users

Food secure 101 (45.91%) 94 (42.73%) 195 (88.64%)

Food insecure 20 (9.09%) 5 (2.27%) 25 (11.36 %)
Total 121 (55.0%) 99 (45.00%) 220 (100 %)

Pearson chi2(1) = 7.1225 Pr. = 0.008

Source: Computed from own survey data of 2020
 52

4.2. Econometric Results

4.2.1. Factors affecting household participation in small scale irrigation

To identify factors affecting participation in small scale irrigation, logit model was used to
predict the probability that the sample households are participating in small scale irrigation
using the hypothesized independent variables, and the results were presented in Table 11.

Before running the model, multi-collinearity was tested using variance inflation factors (VIF).
The problem of multi-collinearity exists when the value of VIF is greater than or equal to ten.
According to the result in Table 10, there was no problem of multi-collinearity in this case
since the mean VIF is 1.20. So, all the variables were included in the model to estimate factors
affecting households’ decision of participation in small-scale irrigation in the study area.

Table 10 Variance inflation factor test (for 13 explanatory variables)

Variables VIF 1/VIF

Family size (AE) 1.49 0.672
Age of the head 1.41 0.710
Land owned 1.27 0.788
Crop disease 1.25 0.802
Dependency ratio 1.24 0.804
Livestock (TLU) 1.23 0.815
Education of the head 1.17 0.854
Distance from irrigation 1.14 0.879
Sex of the head 1.12 0.892
Credit access 1.12 0.896
Market distance 1.09 0.919
Extension contact 1.09 0.920
Off-farm income 1.03 0.968
Mean VIF 1.20

Source: Computed from own survey data of 2020

 53

Therefore, to proceed with the estimation of factors affecting households’ participation in

small scale irrigation, logistic regression was executed, taking participation in irrigation as
dummy dependent variable, and 13 variables (four dummy and nine continuous) as
independent variables. The explanatory variables were selected based on theoretical
explanations and results of different empirical studies.
Accordingly, the result of logistic regression presented in Table 11 revealed that, out of
thirteen independent variables included in the model, eight independent variables (sex of the
household head, age of the household head, family size, livestock holding, land owned,
distance of irrigation source, crop diseases and access to credit services) were the variables
that significantly affected participation in small scale irrigation in the study area.
Sex of the household head: As expected, sex of the household head positively and
significantly affected household participation in small scale irrigation at 1 percent probability
level. The possible reasons might be the physical, socio•cultural and time constraints that
females are facing hindered females from participating in small scale irrigation in the study
area. This result is consistent with the findings of Wakene (2018), Kinfe et al. (2012) and in
contrary to the result reported by Jirane (2015)

Age of the household head: Age of the household head was hypothesized to negatively affect
households’ participation in irrigation. Accordingly, it affected household participation
decision negatively and significantly at 1 percent probability level. The possible reason is that
the older farmers are less likely to adopt modern technologies compared to the younger and
educated farmers. This result is consistent with the findings reported by Molla (2017) and
Jirane (2015).

Family size AE: Adult equivalent family size positively and significantly affected household
participation in small scale irrigation at 1 percent probability level. The result of focus group
discussion showed that family labor is one of the most frequently used inputs for production
under irrigation in the study area; most commonly for protecting the field, hoeing and other
agronomic practices. Therefore, irrigation participation demands more labor force and the
households with large labor force are more likely to participate in small-scale irrigation than
the households with smaller number of labor forces. This result is consistent with the findings
of Wakene (2018) and Kalkidan et al. (2016).
 54

Livestock holding: Against the hypothesis, livestock holding negatively and significantly
affected irrigation participation decision of the households at 1 percent probability level. The
possible reason might be the additional income that the households drive from selling animal
byproducts like selling milk, butter, cheese and others as their additional income sources to
support their lives, and hence may not look for irrigation use. This result is consistent with the
findings reported by Hamda (2016).

Land owned: As expected, size of land owned positively and significantly affected
participation in small scale irrigation at 1 percent probability level. The result of focus group
discussion highlighted that land is the most important determinant of crop and livestock
production. Households having larger land size easily perform crop production and livestock
rearing compared to the households owning smaller land size or not owning land at all. This is
because of high land rental price for crop production and for grazing. This result is consistent
with the findings reported by Agena (2017), Molla (2017), Wakene (2018), and Temesgen
(2019).

Distance from irrigation water source: As expected, distance from irrigation water source
negatively and significantly affected household participation in small scale irrigation at 5
percent probability level. Households those living closer to the irrigation sites are more likely
to use irrigation. The reason may the advantage of performing agronomic practice, suitability
to guard the plots during day and night, lesser walking time required and etc. This result is
consistent with the findings reported by Molla (2017), Wakene (2018), and Temesgen (2019).
Occurrence of crops pests and diseases: Contrary to the hypothesis, occurrence of crop pests
and diseases positively and significantly affected household participation in small scale
irrigation at 5 percent probability level. According to the survey result, crop pests and diseases
are commonly occurring on crop fields of the farmers mainly in the main season than off-
season (irrigation) and impose severe yield loss that may reach almost total yield loss. To
withstand these problems, affected farmers try all the best to participate in small scale
irrigation during the off-season, even by paying expensive prices for land rent according to
responses of the focus group discussion participants. This result is consistent with the results
reported by Birilie (2017).

Access to credit services: As expected, credit service affected participation in small scale
irrigation positively and significantly at 10 percent probability level. This is to mean that
 55

households getting the access of credit services were more likely participating in small scale
irrigation than the households without access to credit services. This is because of those
farmers having access to credit services are more capable to purchase irrigation inputs like
seeds, fertilizers, labor, motor pumps, irrigable land, and etc. This result is consistent with the
results reported by Birilie (2017) and Temesgen (2019).

Generally, from the value of likelihood ratio, which is 40.93, the overall fitness of the model is
good as it is significant at 1 percent probability level.

Table 11 Logit estimation of factors affecting participation in small scale irrigation

Participation in irrigation Coef. St. Err. t-value

Sex of the head 1.286 0.473 2.72***
Age of the head -0.069 0.022 -3.18***
Education of the head 0.052 0.044 1.19
Family size (AE) 0.425 0.111 3.84***
Dependence ratio -0.041 0.238 -0.17
Livestock holding (TLU) -0.150 0.047 -3.18***
Land owned (hec.) 0.466 0.145 3.22***
Distance of irrigation site (KM) -0.493 0.239 -2.06**
Off-farm income (Birr) 0.000 0.000 -0.13
Crop diseases 0.979 0.384 2.55**
Extension contact 0.666 0.539 1.24
Market distance (KM) -0.056 0.070 -0.80
Credit access 1.315 0.699 1.88*
Constant -1.025 1.733 -0.59
Mean dependent var. 0.450 SD dependent var. 0.499
Pseudo r-squared 0.219 Number of obs. 220.0
Chi-square 40.939 Prob. > chi2 0.000
Akaike crit. (AIC) 264.427 Bayesian crit. (BIC) 311.9

Note: *** p<0.01, ** p<0.05 and p<0.05* (significance levels at 1%, 5% & 10%)

Source: Computed from own survey data of 2020

 56

4.2.2. Factors affecting household food security

To estimate factors affecting households’ food security status, food security measured in Daily
Calorie Intake (DCI) method was converted to dummy variable of food secure and food
insecure, and used as dummy dependent variable in the logit model. As independent variables,
fourteen (five dummies and nine continuous) variables were used in the model. These
explanatory variables were selected based on theoretical explanations and the results of
different empirical studies as explained in variable definition and hypothesis part.

Before running the model, variance inflation factor (VIF) was tested again because of one
additional dummy independent variable (access to irrigation) was included. Tables 12 is
showing there were no problems of multi-collinearity since mean VIF is 1.25. Therefore, all
the variables were used in the model to estimate factors affecting household food security.

Table 12 Variance inflation factor test (for 14 explanatory variables)

Variables VIF 1/VIF

Family size (AE) 1.60 0.625
Age of the head 1.48 0.675
Irrigation access 1.35 0.742
Land owned 1.33 0.752
Crop disease 1.30 0.767
Dependency ratio 1.29 0.776
Livestock (TLU) 1.24 0.804
Education of the head 1.18 0.848
Distance from irrigation 1.16 0.860
Sex of the head 1.15 0.868
Credit access 1.14 0.875
Market distance 1.10 0.912
Extension contact 1.09 0.917
Off-farm income 1.03 0.625
Mean VIF 1.25 .

Source: Computed from own survey data of 2020

 57

Finally, estimation of factors affecting household food security was done using Daily Calorie
Intake (DCI) as dummy dependent variable and 14 variables as independent variable.
Households with daily calorie intake exceeding 2200 kilocalorie were taken as food secure
and the rest were taken as food insecure. The result in Table 13 showed that eight variables
(sex of the household head, family size, dependency ratio, livestock holding, land holding,
access to extension contact, access to irrigation services and access to credit services) were the
variables that statistically and significantly affected household food security in the study area.

Sex of the household head: As expected, sex of the household head affected household food
security positively and significantly at 5 percent probability level. This means male headed
households are more likely to be food secure than female headed households. The possible
reasons might be due to physical, socio•cultural and time constraints that females are facing as
compared to males, and these contributed to food insecurity status of female headed
households. This result is consistent with the findings of Ahmed (2015), Mustapha et al.
(2018) and Sani and Kemaw (2019).

Family size AE: In contrary to the hypothesis, adult equivalent family size negatively and
significantly affected household food security at 1 percent probability level. This result is
indicating that the households with larger adult equivalent family size are less likely to be food
secure compared to the households with smaller family size. This might be because of adding
additional family member to limited resources may cause scarcity and lead the family to food
insecurity status. This result is consistent with the findings of Mesele et al. (2018), Mustapha
et al. (2018) and Temesgen (2019).

Dependency ratio: As expected, dependency ratio affected household food security

negatively and significantly at 5 percent significance level. Households with high dependency
ratio were less likely to be food secure compare to the households with less dependence ratio.
The reason for this is that as dependent family members increase, it will impose pressure on
household resources. Additionally, as dependents increase per family, there will be higher
burden on the active family members and this in turn affects households’ food security status.
This result is consistent with the results reported by Goshu (2016), Dawit and Zeray (2017),
Mustapha et al. (2018) and Akukwe (2020).
 58

Livestock holding in TLU: As hypothesized, livestock holding affected household food

security positively and significantly at 5 percent significance level. From this result, the
households having more livestock are more likely to be food secure compared to those
households holding lesser numbers of livestock. The reason for this was the multi-dimensional
contributions of livestock products and byproducts in combating household food insecurity.
Livestock serves as source of food as well as source of income to purchase food during the
time of food shortage. This result is consistent with the findings reported by Haileyesus
(2019), Mesele et al. (2018), Habtewold (2018)

Land owned: Land owned also affected household food security positively and significantly
at 5 percent significance level. This result is consistent with the hypothesis and the households
having more land are more likely to be food secure than those household owning lesser land.
This is because households with wider area of land would have the possibility to produce more
or diverse agricultural products which would diversify consumption either through the product
that they produce or though the additional income that they would get and hence contribute to
improve food security. This result is also consistent with the findings reported by Mustapha et
al. (2018), Mesele et al. (2018), Habtewold (2018) and Haileyesus (2019).

Access to extension services: This variable positively and significantly affected household
food security at 5 percent significance level. This is to mean that households getting extension
services are more likely to be food secure compared to households without the services. This
is because of agricultural extension play significant role in improving production,
productivity, food security and rural livelihood. This result is also consistent with the findings
reported by Mustapha et al. (2018) and Haileyesus (2019).

Access to credit services: As expected, this variable also affected household food security
positively and significantly at 10 percent significance level. The households having access to
credit services are more likely to be food secure than household without the services. This is
because of those households having access to credit services can easily buy agricultural inputs
like improved seed, fertilizer, labor, rent farm land, and etc. and improve their production and
productivity which improves their food security status. This result is also consistent with the
findings reported by Mustapha et al. (2018), Habtewold (2018) and Sami and Kemaw (2019).
 59

Participation in irrigation: As expected, participation in irrigation positively and

significantly affected household food security at 1 percent significance level. From this result,
irrigation user households are more likely to be food secure than household not using
irrigation. The reason for this is because of participation in irrigation enables the households to
efficiently use the available agricultural inputs like land, labor and other resources during the
off (dry) season and increase their production. This result is also consistent with the results
reported by Hamda (2016), Agena (2017), Ngema et al. (2018), Mesele et al. (2018) and Sami
and Kemaw (2019).

Table 13 Factors affecting household food security estimated using daily calorie

intake Food security measured in

calorie intake (DCI) Coef. St. Err. t-value
Sex of the household head 1.693 0.721 2.35**
Age of the head (years) -0.020 0.037 -0.55
Education of the head (years) 0.077 0.108 0.71
Family size (AE) -0.638 0.206 -3.10***
Dependence ratio -1.272 0.497 -2.56**
Livestock holding (TLU) 0.306 0.122 2.51**
Land owned (hec.) 1.243 0.482 2.58**
Distance of irrigation site (KM) -0.694 0.448 -1.55
Off-farm income (Birr) 0.000 0.000 -0.83
Occurrence of crop diseases -0.620 0.712 -0.87
Access to extension contact 1.644 0.785 2.09**
Market distance (KM) -0.145 0.155 -0.94
Access to irrigation services 2.508 0.923 2.72***
Access to credit service es 1.730 0.936 1.85*
Constant 3.139 2.826 1.11
Mean dependent var. 0.886 SD dependent var. 0.318
Pseudo r-squared 0.516 Number of obs. 220.0
Chi-square 80.333 Prob. > chi2 0.000
Akaike crit. (AIC) 105.449 Bayesian crit. (BIC) 156.4

Source: Computed from own survey data of 2020

 60
*** p<0.01, ** p<0.05 and * p<0.1 (significance levels at 1%, 5% and 10%)

Source: Computed from own survey data of 2020

 61

Similarly, factors affecting household food security also estimated using Food Consumption
Score method by converting FCS into dummy variable of one if the value of the score is equal
to and exceeding 42.5 and zero otherwise. This is because of the household is said to be food
secure if his/her food consumption score value is exceeding 42 and insecure otherwise.
Therefore, all households in both border line and poor food consumption score profile were
considered to be food insecure. Table 14 is presenting the logistic regression result.

Table 14 Factors affecting household food security estimated using food consumption score

Food security measured in

Coef. St. Err. t-value
Food consumption score (FCS)
Sex of the household head 1.239 0.475 2.61***
Age of the head (years) -0.010 0.024 -0.41
Education of the head (years) 0.062 0.055 1.13
Family size (AE) -0.256 0.125 -2.04**
Dependence ratio -0.338 0.283 -1.20
Livestock holding (TLU) 0.153 0.059 2.61***
Land owned (hec.) 0.495 0.199 2.48**
Distance of irrigation site (KM) -0.201 0.239 -0.84
Off-farm income (Birr) 0.000 0.000 -0.39
Occurrence of crop diseases -0.173 0.417 -0.41
Access to extension contact 0.417 0.515 0.81
Market distance (KM) -0.055 0.088 -0.62
Access to irrigation services 1.674 0.469 3.57***
Access to credit service es 0.804 0.593 1.36
Constant -0.826 1.669 -0.50
Mean dependent var. 0.700 SD dependent var. 0.459
Pseudo r-squared 0.303 Number of obs. 220.0
Chi-square 81.48 Prob. > chi2 0.000
Akaike crit. (AIC) 217.3 Bayesian crit. (BIC) 268.2

Note: *** p<0.01 and ** p<0.05 (significance levels at 1% and 5%)

Source: Computed from own survey data of 2020

 62

From Table 14, five variables (sex of the household head, family size, livestock holding, land
holding and participation in irrigation) were found to significantly affect household food
security in the study area. Sex of the household head, livestock holding, land owned and
participation in irrigation were the variables that positively and significantly affected food
security status of the household in the study area, while family size was the variable that
negatively and significantly affected food security status of the sample households in the study
area. Moreover, sex of the household head, livestock holding and participation in irrigation
were the variables significantly affected household food security at 1 percent probability level,
while family size and land holding were the variables that significantly affected the household
food security at 5 percent probability level.

4.2.3. Impact of small-scale irrigation on household food security

Under this sub section, impact of small-scale irrigation on household food security was
assessed using propensity score matching method of impact evaluation since there was no
baseline data. Assessment of impact evaluation using this method follows five basic steps.
These steps are: estimation of the propensity score, restricting common support region,
choosing matching algorithm, checking for balance, and sensitivity analysis.

4.2.3.1. Estimation of propensity score

Logistic regression was used to estimate propensity score. Participation in irrigation was
dummy dependent variable and 13 independent (four dummies and nine continuous) variables
were included to estimate propensity scores for both groups. As stated in Bernand et al (2007),
conditional on the ability of propensity score to overcome potential sources of bias, irrigation
user and non-user households become comparable. So, the propensity score was estimated as
shown in Table 15.

The logistic regression result presented in Table 15 showed that the model performed well
since the likelihood ratio chi-square value is 66.35 and the overall fitness of the model was
found to be significant at 1 percent. Moreover, the value of pseudo-R 2 is also smaller (0.21),
indicating that irrigation user households do not have much distinct characteristics over non-
users. This
 63

similarity between both groups make easier to get good match between irrigation users and
non- users (Caliendo and Kopeinig, 2008).

Table 15 Logistic regression result for estimating propensity scores

Participation in irrigation Coef. St. Err. z

Sex of the household head 1.286 0.497 2.580
Age of the head (years) -0.069 0.021 -3.240
Education of the head (years) 0.052 0.044 1.190
Family size (AE) 0.425 0.112 3.800
Dependence ratio -0.041 0.252 -0.160
Livestock holding (TLU) -0.150 0.047 -3.230
Land owned (hec.) 0.466 0.153 3.050
Distance of irrigation site (KM) -0.493 0.222 -2.220
Off-farm income (Birr) -0.000 0.000 -0.120
Occurrence of crop diseases 0.979 0.365 2.680
Access to extension contact 0.666 0.507 1.320
Market distance (KM) -0.056 0.074 -0.750
Access to credit services 1.315 0.647 2.030
Constant -1.025 1.541 -0.670

Logistic regression Number of obs. = 220

Log likelihood = -118.2 Prob. > chi2 = 0.0000

LR chi2 (13) = 66.35 Pseudo R2 = 0.21

Source: Computed from own survey data of 2020

Graphically, the estimated propensity scores for both irrigation user and non-user groups were
presented in Figure 3 using kernel density. From the figure, the normal line (the middle line)
represents the total sample household; the short dash line (left below and right upper line)
representing the propensity score of irrigation participants, while the long dash (left upper and
right below line) representing the propensity score of irrigation non-participant group.
 64

Figure 3 Kernel density for participants, non-participants and both, before matching

4.2.3.2. Restricting the common support region

Estimation of ATT and ATE are possible only in the region of common support. As presented
in Table 16, the estimated values of propensity score for the sample households were ranging
from 0. 0080573 to 0. 9653866, with mean and standard deviation of 0.45 and 0.26
respectively. The propensity scores for irrigation participant group ranges from 0.0263124 to
0.9653866, with mean and standard deviation of 0.6 and 0.2.

Similarly, the propensity score for non-participant group ranges from 0.0080573 to 0.8350227
with mean and standard deviation of 0.33 and 0.22. Based on these values of propensity score,
the common support region was restricted. Accordingly, the common support region is the
values of the propensity scores found between the larger value from the two minimums (the
larger value from minimum for participants and minimum for non-participants) and the
smaller
 65

value from the two maximums (the smaller value from the maximum of irrigation participants
and the maximum of irrigation non-participants).

From Table 16, the larger value from the two minimums was 0. 0263124 and the smaller value
from the two maximums was 0. 8350227. Therefore, the common support region was
restricted to the values of the propensity scores that are greater than 0.0263124 and smaller
than 0.8350227. Based on the restriction of the common support, 19 observations (16
observations from irrigation users and 3 observations from non-users) were found to be out of
the common support region and excluded from farther analysis (the impact of small-scale
irrigation on household food security).

Table 16 Summary of estimated propensity score to restrict common support region

Irrigation status No. obs. Mean St.dev. Min Max

Non-participants 121 . 3257946 . 2173374 . 0080573 . 8350227
Participants 99 . 6018065 . 2224784 . 0263124 . 9653866
Total 220 .4500000 . 2587913 . 0080573 . 9653866

Source: Computed from own survey data of 2020

Graphically, the region of common support can also be expressed using kernel density and
histogram. Therefore, the kernel density was used to visually present the common support
region for both participants and non-participants consequently as shown in figures 4 and 5.
Figure 4 is presenting the kernel density of the propensity score for irrigation non-participant
households and Figure 5 is presenting the kernel density of the propensity score for irrigation
participant households. Moreover, the histogram visual presentation of the estimated
propensity score is also presented as expressed in appendix 7.2.
 66

Figure 4. Kernel density of propensity scores for non-participants in common support

Similarly, the visual presentation (the kernel density) of the propensity score that rested in the
region of common support for irrigation participants was presented in figure 5.

Figure 5. Kernel density of propensity score for participants in common support

 67

4.2.3.3. Choosing the matching algorithm

After estimation of the propensity score and restriction of common support region, choosing
the appropriate matching technique was followed. Accordingly, basic matching techniques
such as the nearest neighbor matching, caliper matching, radius matching and kernel
matching with different bandwidth were tested and the result was presented in Table 17.

Table 17 Comparisons of different matching methods

Matching Methods Matched sample size Balancing test Pseudo R2

The Nearest Neighbor Matching
Nearest Neighbor (1) 201 9 0.077
Nearest Neighbor (2) 201 12 0.043
Nearest Neighbor (3) 201 13 0.031
Nearest Neighbor (4) 201 13 0.034
Caliper Matching
Caliper (0.01) 88 13 0.124
Caliper (0.10) 115 13 0.052
Caliper (0.25) 138 13 0.092
Caliper (0.50) 155 13 0.147
Radius Matching
Radius caliper (0.01) 151 12 0.063
Radius caliper (0.10) 201 13 0.009
Radius caliper (0.25) 201 13 0.029
Radius caliper (0.50) 201 12 0.129
Kernel Matching
kernel bandwidth (0.01) 151 11 0.071
kernel bandwidth (0.10) 201 13 0.010
kernel bandwidth (0.25) 201 13 0.017
kernel bandwidth (0.50) 201 13 0.075

Source: Computed from own survey data of 2020

 68

Based on Table 17, the choice of appropriate matching method was done by looking at their
balancing qualities. According to Dehejia and Wahba (2002), balancing qualities of matching
methods are measured by different standards like the number of matched samples, the value
of Pseudo R2 and the number of balanced covariates. The matching method with larger
matched sample size, lower pseudo R2 and the one balancing all the covariates is the most
preferred matching method. Therefore, from the matching methods presented in Table 17,
radius matching with band width of 0.1 was the best matching method since the largest
sample size were matched, the lowest value of pseudo R2 was registered and all the covariates
were matched.

4.2.3.4. Assessing the matching quality

4.2.3.4.1. Balancing test

As it conditioned on propensity score, rather than on covariates, we should have to check

wheather or not the matching procedure balances the distribution of relevant variable in both
irrigation participants and non participants. The main idea is to compare situations before and
after matching and check wheather or not any difference exist after matching conditioning on
propensity score. Balancing test is a test conducted to check whether there is significant
difference in the mean values of covariates for participants and non-participants. Based on
Table 17, radius matching with band width of 0.1 matched larger sample size, balanced all the
covariates and bear the minimum Pseudo R2 value compared to other methods.

Therefore, the radius matching method with bandwidth of 0.1 was tested for balancing of the
covariates. According to the survey result presented in Table 18, the t-value was showing that
eight covariates were statistically significant before matching, but all the covariates become
statistically insignificant after matching. Moreover, the standard bias for all the covariates
before matching were larger and in the range of 2.7% to 44.7 % in absolute value, and the
mean bias of the covariates before matching was 23.8% as presented in the bottom rows of
Table 18. But, after matching, the standard bias for all the covariates become smaller and
ranged from 1.9% to 10.2% with the mean bias after matching of 4.7% as presented in the
bottom rows of the same table. These values guarantee us that the matching method has high
matching quality.
 69

Table 18 Balancing test for covariates

Unmatched Mean %reduct T -test V(T)/

Variables bias
Matched Treated Control |bias| t P > |t| V(C)
Sex Unmatched .88889 .80165 24.2 1.76 0.079 .
Matched .87952 .8915 -3.3 86.3 -0.24 0.810 .
Age Unmatched 42.131 45.405 -33.3 -2.47 .014 1.29
Matched 43.084 43.267 -1.9 94.4 -0.12 0.903 1.22
Education Unmatched 4.4545 3.7355 18.7 1.39 0.167 1.15
Matched 4.4458 4.5932 -3.8 79.5 -0.23 0.816 0.95
Family size Unmatched 4.9419 4.361 30.6 2.27 0.024 1.15
Matched 4.7609 4.7097 2.7 91.2 0.17 0.868 0.7
Dependency Unmatched .87879 .90909 -4.3 -0.31 0.754 0.66*
Matched .86747 .88688 -2.8 35.9 -0.17 0.862 0.72
Livestock Unmatched 6.3584 7.5436 -30.4 -2.23 0.027 0.82
Matched 6.4448 6.6333 -4.8 84.1 -0.32 0.748 0.99
Land Unmatched 1.7278 1.3562 30.6 2.25 0.025 0.99
Matched 1.5819 1.6847 -8.5 72.3 -0.56 0.578 0.74
Irrigation di Unmatched 1.9798 2.3306 -44.7 -3.32 0.001 1.21
Matched 2.012 2.0289 -2.2 95.2 -0.14 0.889 1.50
Off-farm Unmatched 7163.6 8090 -9.2 -0.68 0.498 0.87
Matched 6979.5 6613 3.7 60.4 0.25 0.804 1.02
Crop diseas Unmatched .57576 .43802 27.7 2.04 0.042 .
Matched .49398 .44334 10.2 63.2 0.65 0.516 .
Extension Unmatched .86869 .8595 2.7 0.20 0.844 .
Matched .87952 .90195 -6.5 -144.3 -0.46 0.646 .
Market dist. Unmatched 5.1717 5.562 -17.6 -1.30 0.195 1.18
Matched 5.241 5.0476 8.7 50.5 0.57 0.567 1.22
Credit acces Unmatched .9596 .8595 35.3 2.54 0.012 .
Matched .95181 .95946 -2.7 92.4 -0.24 0.812 .

* if variance ratio outside [0.67; 1.49] for U and [0.66; 1.52] for M

Sample Pseudo R2 LR chi2 p>Chi2 Mean Bias Med bias B R %var

Unmatched 0.216 65.27 0.000 23.8 27.7 117.7* 0.94 11
Matched 0.006 1.45 1.000 4.7 3.7 18.6 1.06 0

* if B>25%, R outside [0.5; 2]

Source: Computed from own survey data of 2020

 70

Similarly, the joint significance test presented in the bottom rows of Table 18 revealed that the
value of pseudo R2 was very small (0.006) and the t-test was not significant. These also gives
as guarantee that the matching process created good balance between participants and non-
participants based on the included covariates. Therefore, estimation of the Average Treatment
Effect on the Treated (ATT) was proceeded.

4.2.3.4.2. Estimation of the average treatment effect on treated (ATT)

According to Caliendo and Kopeinig (2008), the main goal of performing propensity score
matching is to ensure that both participants and non-participants are in the same condition to
estimate average treatment effect on the treated (ATT). For this research, ATT measuring the
average difference of individual’s Daily Calorie Intake (DCI) between irrigation participants
and non-participants, and also average difference of household Food Consumption Score
(FCS) between irrigation participants and non-participants. Therefore, Average Treatment
Effect on the Treated (ATT) was estimated by taking daily calorie intake and food
consumption score as outcome indicators as presented in Table 19.

Table 19 Estimation of ATT based on daily calorie intake and food consumption score

Outcome Variable Parameter Treated Controls Difference S. E (B.st) T-stat

Daily calorie intake (AE) ATT 5199.61 4670.56 529.05 252.81 2.09**
Food cons. Score (FCS) ATT 49.12 45.43 3.69 1.83 2.02**

Note: ** shows that the variable is significant at 5percent probability level

Source: Computed from own survey data of 2020

The results in Table 19 revealed that concerning individual daily calorie intake (DCI) method
of measuring food security status, on average, the family members of irrigation participant
households consume more energy of 529 kilocalories than the family members of irrigation
non- participant households, and this difference was significant at 5 percent probability level.
Similarly, based on Food Consumption Score (FCS) method of measuring household food
security status, on average, irrigation user households got more food consumption score values
of 3.69 and have better food consumption score profile than irrigation non-user households,
and
 71

this difference was significant at 5 percent probability level. The bootstrapped standard errors
were obtained after replications of 100 since the standard error that we get from the model
does not take into account that the propensity score was estimated.

Therefore, based on the information presented in Table 19 and other relevant information
listed in the above steps, controlling for pre-intervention differences using Propensity Score
Matching (PSM), participation in irrigation has impacted (improved) household food security
measured in both daily calorie intake (DCI) and household food consumption score (FCS).

4.2.3.5. Sensitivity analysis

Sensitivity analysis was done to assess the effects of unobservable bias, and it also tests if the
estimated Average Treatment Effect on the Treated (ATT) is effective or not. It is used for
significant effects, and using for insignificant effects is meaningless. It helps to test the
robustness of the results (estimated ATT), and it also help for checking if the unobserved
confounders have effects on the estimated ATT or not. In non-experimental data, measuring
the extent of bias is impossible and this problem is solved using sensitivity analysis.

Accordingly, the sensitivity analysis was done to check whether the average treatment effect
that was estimated using individual’s daily calorie intake (DCI) method was sensitive to
unobserved bias using Rosenbaum bound of gamma values between 1 and 3, by adding 0.25
on 1 and continuing up to 3 and the result was presented in Table 20.

The result of sensitivity analysis presented in Table 20 indicated that all-important covariates
that affecting household participation in small scale irrigation and household food security
were included, and the estimated average treatment effect on the treated (ATT) reported in
Table 19 was insensitive to hidden (unobserved) bias up to 200% (gamma value of 3).

Therefore, the estimated average treatment effect on the treated (ATT), which is about 529
kcal is the pure effect of small-scale irrigation, and this effect is statistically significant at 5 %
probability level.
 72

Table 20 Rosenbaum sensitivity analysis for daily calorie intake method

Gamma sig+ sig- t-hat+ t-hat- CI+ CI-

1 .000315 .000315 603.237 603.237 276.569 928.044
1.25 .005305 7.6e-06 461.38 751.882 112.344 1068.05
1.5 .030473 1.7e-07 349.395 864.264 -11.5488 1178.97
1.75 .095069 3.4e-09 239.754 959.054 -134.119 1271.94
2 .203758 6.6e-11 150.199 1040.48 -248.818 1356.49
2.25 .342246 1.3e-12 75.7718 1102.73 -338.455 1427.66
2.5 .487974 2.4e-14 4.75855 1164.6 -437.687 1492.77
2.75 .621741 4.4e-16 -56.1573 1211.7 -520.442 1555.94
3 .732778 0 -122.487 1262.97 -591.248 1609.84

* gamma - log odds of differential assignment due to unobserved

factors sig+ - upper bound significance level
sig- - lower bound significance level
t-hat+ - upper bound Hodges-Lehmann point estimate
t-hat- - lower bound Hodges-Lehmann point estimate
CI+ - upper bound confidence interval (a= .95)
CI- - lower bound confidence interval (a= .95)

Source: Computed from own survey data of 2020

Similar to the food security measured in individual’s daily calorie intake (DCI), sensitivity
analysis was done for the household food security measured in food consumption score (FCS)
method, using Rosenbaum bound of gamma values between 1 and 3, by adding 0.25 on 1 and
continuing up to 3 as the previous method and the result was presented in Table 21.

The result of sensitivity analysis presented in Table 21 also indicated that all-important
covariates that are affecting both household participation in small scale irrigation and
household food security were included, and the estimated average treatment effect on the
treated (ATT) was insensitive to hidden (unobserved) bias up to 200% (gamma value of 3).
Therefore, the result of ATT represented in Table 19, which is 3.69 is the pure effect of small-
scale irrigation, and this effect is statistically significant at 5% probability level.
 73

Table 21 Rosenbaum sensitivity analysis for food consumption score method

Gamma sig+ sig- t-hat+ t-hat- CI+ CI-

1 .002108 .002108 1.78765 1.78765 .512468 3.37058
1.25 .025396 .000071 1.18908 2.48609 -.007507 4.30301
1.5 .109418 2.0e-06 .720006 3.1274 -.482486 5.18073
1.75 .26564 5.1e-08 .33954 3.68253 -.947998 6.02752
2 .457653 1.2e-09 .054237 4.17616 -1.20976 7.02715
2.25 .637429 2.8e-11 -.246687 4.63444 -1.48287 7.82881
2.5 .777037 6.4e-13 -.467814 5.13393 -1.79916 8.59754
2.75 .871912 1.4e-14 -.75919 5.66321 -2.04339 9.32921
3 .930357 3.3e-16 -.946214 6.02475 -2.29168 9.98932

* gamma - log odds of differential assignment due to unobserved factors

sig+ - upper bound significance level
sig- - lower bound significance level
t-hat+ - upper bound Hodges-Lehmann point estimate
t-hat- - lower bound Hodges-Lehmann point estimate
CI+ - upper bound confidence interval (a= .95)
CI- - lower bound confidence interval (a= .95)

Source: Computed from own survey data of 2020

 74

5. SUMMARY, CONCLUSIONS AND RECOMMENDATIONS

5.1. Summary and Conclusions

The study was conducted with the main objective of assessing the impact of small-scale
irrigation on household food security in Bishoftu Town Oromia.
The specific objectives of the study were identifying factors affecting participation in small
scale irrigation, assessing factors affecting household food security and assessing the impact of
small-scale irrigation on household food security in the study area.

Both primary and secondary data were used. Multi stage sampling technique was employed
and the primary data were collected from four randomly selected kebeles from a total of 220
households (121 irrigation non-users and 99 irrigation users). Secondary data were collected
from different sources. Descriptive and econometric data analysis were performed. Propensity
score matching was the method used in analyzing the impact of small-scale irrigation on
household food security since there was no baseline data.

The descriptive result showed that, sex of the household head, access to credit services,
occurrence pf crop pests and diseases, age of the household head, livestock holding, family
size, irrigation distance and land owned were the variables that showed significant relation
with irrigation participation. Similarly, sex, credit access, crop disease, extension contact, age,
education, dependency, livestock holding, family size, market distance, irrigation distance and
land owned were the variables that significantly related to household food security. Moreover,
mean individual’s daily calorie intake and food consumption score values were also found to
have significant difference between irrigation users and non-users.

The logistic regression result revealed that participation in small scale irrigation was
significantly affected by eight explanatory variables, namely, sex of the head, age of the head,
family size, livestock holding, land owned, distance from irrigation site, occurrence of crop
diseases, and access to credit services. Age of the head, distance of irrigation site and livestock
holding negatively affected participation in irrigation, while sex of the head, family size, land
holding, occurrence of crop diseases and access to credit services were the variables that
positively and significantly affected participation in irrigation.
 75

Similarly, the logistic regression result revealed that eight variables significantly affected food
security. These variables were sex of the head, family size, dependence ratio, livestock
holding, land owned, access to extension contacts, access to irrigation services and access to
credit services. Among these variables, sex of the household head, livestock holding, land
owned, access to extension contacts, access to irrigation services and access to credit services
affected food security positively, while family size and dependency ratio affected household
foods security negatively.

To assess the impact of small-scale irrigation on household food security, propensity score
was estimated and the common support region was restricted. Based on this common support,
118 irrigation non-users were matched with 83 irrigation users using radius matching of 0.1
bandwidth by discarding 19 observations that are out of the common support. Matching
qualities like pseudo R2, matched sample size and the number of balanced covariates were
checked, and accordingly, the pseudo R2 value was 0.006, the matched sample size was 201
and the number of matched covariates were 13. Furthermore, the standard error was
bootstrapped to capture all sources of errors and sensitivity analysis was done and the
estimated ATT was insensitive, showing its robustness.

Finally, the impact estimation result revealed that both methods of measuring household food
security (individual’s daily calorie intake and food consumption score) showed significant
mean difference for irrigation users and non-users. On average, family member of irrigation
user households consumes more calorie of 529 kcal than the family member of irrigation non-
users, which showed significant difference. Similarly, irrigation user households have better
food consumption score profile than irrigation non-users, and on average, irrigation user
households got 3.69 more value of food consumption score, which also showed significant
difference.

Therefore, participation in small-scale irrigation was significantly affected by sex of the head,
age of the head, family size, livestock holding, land owned, distance of irrigation source,
occurrence of crop diseases and pests and access to credit services, while household food
security was significantly affected by sex of the head, family size, dependency ratio, livestock
holding, land holding, extension contact, irrigation use and credit services. Finally,
participation in small-scale irrigation positively significantly affected household food security
status.
 76

5.2. Recommendations

Based on the empirical findings of this research, the following recommendations were
forwarded:

The results of this study revealed that households using small scale irrigation were more food
secure than the households not using small scale irrigation both in daily calorie intake (DCI)
and food consumption score (FCS) methods of measuring households’ food security.
Therefore, government policies and strategies (agricultural policies and strategies) focusing on
promoting small scale irrigation should be implemented to take out the lives of millions of
small-holders from the state of food insecurity and poverty, especially in erratic rain fall and
drought prone areas of the country.

According to the result of this study, sex of the household head has positive and significant
effect on both household participations in small scale irrigation and household food security.
This means, female headed households were less participants in small scale irrigation and less
food secure compared to male headed households. Therefore, the district bureau of agriculture
and rural development and the development agents working in each kebeles should take
initiative and give awareness to female headed households to make them participate in small
scale irrigation and improve their food security status through trainings and experience
sharings.

Age of the household head affected participation in small scale irrigation negatively and
significantly. This result is showing that aged households are less likely to participate in small
scale irrigation compared to younger household heads. Therefore, the concerned stakeholders
including development agents need to improve the awareness of the older household heads
using trainings and experience sharing to make them participate in small scale irrigation and
also improve their food security status.

The finding of this study also indicated that adult equivalent family size positively and
significantly affected participation in small scale irrigation. But negatively and significantly
affected household food security at 1 percent probability level. Therefore, government policies
and strategies promoting family planning should be implemented and health extension
workers should improve the awareness of the households towards family planning.
 77

Livestock holding negatively affected participation in small scale irrigation. But, positively
affected food security status. Therefore, governmental and non-governmental institutions and
other stakeholders should promote mixed farming (both livestock and crop production) to
improve household food security level in the study area.

Land ownership affected both participation in small scale irrigation and household food
security positively and significantly. This is to mean that households owning larger land size
are more likely to participate in small scale irrigation and also, they are more food secure than
those households owning smaller land size. Therefore, any development and agricultural
policy supporting the expansion of cultivated land is recommended. For example, urbanization
and industrialization, which draw larger population from rural areas to the urban areas that
expand cultivable land for the those remaining in the rural areas, and family planning for rural
dwellers.

The result of this study also revealed that access to extension services positively and
significantly affected household food security status. This means households getting extension
services are more likely food secure compared to the households not getting extension
services. Therefore, the district bureau of agriculture and rural development with development
agents in each kebeles should give trainings and experience sharing to farmers to enhance their
awareness on how to improve agricultural production, productivity and their food security
status.

According to the result of this thesis, distance of irrigation site affected participation in small
scale irrigation negatively and significantly. This means households living closer to irrigation
water source are more likely to participate in small scale irrigation compared to those
households living far away from irrigation water sources. Therefore, governmental and non-
governmental institutions and other stakeholders planning to construct small scale irrigation
should consider the distance of the schemes from the residence of the user households.

Finally, access to credit services positively and significantly affected household participation
in small scale irrigation and also household food security status. This is to mean that
households getting access to credit services are more likely to participate in small scale
irrigation than households who do not get access to credit services and similarly, households
getting access to credit services were more likely to be food secure than those households not
getting access to credit services. Moreover, most of the households who did not received
 78
credit services
 79

responded that the reason why they did not take credit was mainly high interest rate and
bureaucratic credit giving systems of microfinances in the study area. Therefore,
microfinances giving credit services to small-holder farmers should check their interest rate,
credit system and the amount they are giving.
 80

6. REFERENCES

Abadie, A., Drukker, D., Herr, J.L. and Imbens, G.W., 2004. Implementing matching
estimators for average treatment effects in Stata. The Stata journal, 4(3), pp.290-311.
Abdissa, F., 2012. Impact of Small-Scale Irrigation Schemes on Household Food Security in
Western Oromiya, Ethiopia: The Case of Sibu Sire District of Eastern Wollega
Zone (Doctoral dissertation, Haramaya University).
Agena M.A., 2017. Impact of small-Scale Irrigation on Farmers’ Cropping System Choice,
Productivity and Food Security in Meskan District of Southern Ethiopia (Doctoral
dissertation, Haramaya University).
Agerie, N.W., 2013. Determinants of smallholder rural farm households’ participation in
small scale irrigation and its effect on income in North Gondar zone: a cross-sectional
approach (evidence from Dembia Woreda) (Doctoral dissertation, Mekelle University).
Ahmed, M. 2015. Determinants of Household Food Security and Coping Strategies: The Case
of Bule-Hora District, Borana Zone, Oromia, Ethiopia. European Journal of Food
Science and Technology, 3, 30-44.
Akukwe T.I., 2020. Household food security and its determinants in agrarian communities of
southeastern Nigeria. Agro-Science, 19 (1), 54-60. DOI:
https://dx.doi.org/10.4314/as.v19i1.9
Akuriba, M.A., Haagsma, R., Heerink, N. and Dittoh, S., Irrigation governance and
performance: the case of smallholder irrigated agriculture in northern Ghana.
Andrew D. Jones, Francis M. Ngure, Gretel Pelto, and Sera L. Young, 2013. What Are We
Assessing When We Measure Food Security? A Compendium and Review of Current
Metrics. American Society for Nutrition. Adv. Nutr. 4: 481–505, 2013
Asayehegn, K., 2012. Negative impact of small-scale irrigation schemes: A case study of
Central Tigray regional state, Ethiopia. Agric Res Rev, 1(3), pp.80-85.
Astatike, A.A., 2016. Economic Valuation of Improved Irrigation Water in Bahir Dar Zuria
Woreda, Ethiopia. Economics, 5(3), pp.46-55.
ATA, 2016. Ethiopian Agricultural Transformation Agency Annual Report. Addis Ababa,
Ethiopia.
 81

Ayalew, D.W., 2018. Theoretical and Empirical Review of Ethiopian Water Resource
Potentials, Challenges and Future Development Opportunities. Int J Waste Resour 8:
353. doi:10.4172/2252-5211.1000353
Ayele, G.K., Nicholson, C., Collick, A.S., Tilahun, S.A. and Steenhuis, T.S., 2013. Impact of
small-scale irrigation schemes on household income and the likelihood of poverty in
the Lake Tana basin of Ethiopia. ILRI.
Baker, J.L., 2000. Evaluating the impact of development projects on poverty: A handbook for
practitioners. The World Bank.
Ballard, T., Coates, J., Swindale, A. and Deitchler, M., 2011. Household hunger scale:
indicator definition and measurement guide. Food and Nutrition Technical Assistance
(FANTA), 2.
Baum, C.F., 2013. Propensity Score Matching Regression Discontinuity Limited Dependent
Variables. http://fmwww. bc. edu/ECC S, 2013.
Bedeke, S.B., 2012. Food insecurity and copping strategies: a perspective from Kersa district,
East Hararghe Ethiopia. Food Science and Quality Management, 5, pp.19-26.
Bekele, W., 2006. Analysis of farmers' preferences for development intervention programs: a
case study of subsistence farmers from East Ethiopian highlands. African development
review, 18(2), pp.183-204.
Berhanu, B., Seleshi, Y. and Melesse, A.M., 2014. Surface water and groundwater resources
of Ethiopia: potentials and challenges of water resources development. In Nile River
Basin (pp. 97-117). Springer, Cham.
Bernard, T., Taffesse, A.S. and Gabre‐Madhin, E., 2008. Impact of cooperatives on
smallholders' commercialization behavior: evidence from Ethiopia. Agricultural
Economics, 39(2), pp.147-161.
Beyene, F. and Muche, M., 2010. Determinants of food security among rural households of
Central Ethiopia: An empirical analysis. Quarterly Journal of International
Agriculture, 49(892-2016-65219), pp.299-318.
Birilie, A.B., 2017. Impact of Koga Large Scale Irrigation Scheme on Poverty Status of Rural
Households: The Case of Mecha District in West Gojjam Zone of Amhara Regional
State, Ethiopia (MSc Thesis, Haramaya University)
 82

Bjornlund, V. and Bjornlund, H., 2010. Sustainable irrigation: a historical perspective (Vol.
37). WIT Press.
Bogale, A. and Shimelis, A., 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).
Buddelmeyer, H. and Skoufias, E., 2004. An evaluation of the performance of regression
discontinuity design on PROGRESA. The World Bank.
Caliendo, M. and Kopeinig, S., 2008. Some practical guidance for the implementation of
propensity score matching. Journal of economic surveys, 22(1), pp.31-72.
Callens, K. and Seiffert, B., 2003. Methodological guide: Participatory appraisal of nutrition
and household food security situations and planning of interventions from a
livelihood’s perspective. Food and Agriculture Organization of the United Nations,
Rome.
Calonico, S., Cattaneo, M.D. and Titiunik, R., 2014. Robust data-driven inference in the
regression-discontinuity design. The Stata Journal, 14(4), pp.909-946. causal studies.
The Review of Economics and Statistics, 84(1): 151–161
CIA World Fact Book, 2019. Ethiopia- United States Central Intelligence Agency.
https://www.cia.gov/library/publications/the-world-factbook/geos/et.html
Cochrane, L., 2011. Food security or food sovereignty: The case of land grabs. The Journal of
Humanitarian Assistance, 5.
CSA, 2017. Central Statistical Agency. The Federal Democratic Republic of Ethiopia, Central
Statistical Agency, Agricultural Sample Survey 2016/7 (2009 E.C.), Volume I, Report
on Area and Production of Major Crops (Private Peasant Holdings, Meher Season),
Statistical Bulletin 532, May 2017, Addis Ababa, Ethiopia.
Daniel, T., 2008. Access and Utilization of Agricultural Information by Resettler Farming
Households: The Case of Metema Wereda, North Gondar, Ethiopia.
Dawit, D. and Balta, A., 2015. Review on Impact of Small-Scale Irrigations in Household
Food Security in Ethiopia. Journal of Economics and Sustainable Development, 6,
p.21.
Dawit, D. and Zeray, N., 2017. Determinants of rural household food security in Wolaita
Zone: the case of Humbo Woreda. Journal of Poverty, Investment & Development 32,
65-82
De Haen, H., 2003. The state of food insecurity in the world 2003: monitoring progress
towards the world food summit and millennium development goals. FAO, Roma
 83
(Italia).
 84

Dehejia, R. and Wahba, S. 2002. Propensity score matching methods for non-experimental
Deribe, R., 2008. Institutional analysis of water management on communal irrigation systems:
the case of Atsbi Wemberta district in Tigray region and Ada'a district in Oromiya
region, Ethiopia (pp. 3-7). Addis Ababa University.
Dittoh, S., Bhattarai, M. and Akuriba, M.A., 2013. Micro Irrigation-Based Vegetable Farming
for Income, Employment and Food Security in West Africa.
Domènech, L., 2015. Improving irrigation access to combat food insecurity and
undernutrition: A review. Global Food Security, 6, pp.24-33.
EHNRI, (Ethiopian Health and Nutrition Research Institute), 2000. Food Composition Table
for Use in Ethiopia. Addis Ababa. Ethiopia.
Ellis, F., 1993. Peasant economics: Farm households in agrarian development (Vol. 23).
Cambridge University Press.
Endalew, B., Muche, M. and Tadesse, S., 2015. Assessment of food security situation in
Ethiopia: A Review. Asian Journal of Agricultural Research, 9(2), pp.55-68.
Etwire P., Dogbe. W., Wiredu1. A., Martey. E., Etwire. E., Robert. K., and Wahaga, E. 2013.
Factors Influencing Farmer’s Participation in Agricultural Projects: The case of the
Agricultural Value Chain Mentorship Project in the Northern Region of Ghana.
Journal of Economics and Sustainable Development, 4(10): 1•9
Fanadzo, M., 2012. Revitalisation of smallholder irrigation schemes for poverty alleviation
and household food security in South Africa: A review. African journal of agricultural
research.
Fanzo, J., Hawkes, C., Udomkesmalee, E., Afshin, A., Allemandi, L., Assery, O., Baker, P.,
Battersby, J., Bhutta, Z., Chen, K. and Corvalan, C., 2018. 2018 Global Nutrition
Report: Shining a light to spur action on nutrition.
FAO and ECA, 2018. Regional Overview of Food Security and Nutrition. Addressing the
threat from climate variability and extremes for food security and nutrition. Accra. 116
pp
FAO and IFC, 2015. Irrigation Market Brief of Ethiopia. Food and Agriculture Organization
of the United Nations. Rome, Italy.
FAO and IPPC, 2017. Plant health and food security: International Plant Protection
Convention (IPPC) Viale delle Terme di Caracalla, 00153 Rome, Italy
 85

FAO, 2019. Early Warning Early Action report on food security and agriculture (April–June
2019). Rome. 62 pp. License: CC BY-NC-SA 3.0 IGO
FAO, IFAD, UNICEF, WFP and WHO. 2018. The State of Food Security and Nutrition in the
World 2018. Building climate resilience for food security and nutrition. Rome, FAO.
License: CC BY-NC-SA 3.0 IGO.
FDRE, 2010. Federal Democratic Republic of Ethiopia, Growth and Transformation Plan 20
10/ 11 -20 14/ 15 Volume I: Ma in Text
FSIN, 2019. Food Security Information Network, Global Report on Food Crises: Joint
Analysis for Better Decisions.
Garnett, T., Appleby, M.C., Balmford, A., Bateman, I.J., Benton, T.G., Bloomer, P.,
Burlingame, B., Dawkins, M., Dolan, L., Fraser, D. and Herrero, M., 2013. Sustainable
intensification in agriculture: premises and policies. Science, 341(6141), pp.33-34.
Green, W. 2003. Econometric Analysis. 5th ed. Prentice Hall. Inc, New York.
Goshu, M.T., 2016. Determinants of Rural Food Security and Child Nutrition: The Case Study
of Gubalafto District of Amhara Regional State, Ethiopia (Unpublished PhD thesis).
Department of Economics, Aligarh Muslim University Aligarh, India
Guo, S., Barth, R.P. and Gibbons, C., 2006. Propensity score matching strategies for
evaluating substance abuse services for child welfare clients. Children and Youth
Services Review, 28(4), pp.357-383.
Habtewold, T.M., 2018. Determinants of food security in the Oromiya Region of Ethiopia.
In Economic Growth and Development in Ethiopia (pp. 39-65). Springer, Singapore.
Hadush, H., 2014. Adoption and Impact of micro Irrigation on Household Income: The case of
Eastern Tigray (Doctoral dissertation, Mekelle University).
Hagos, F., Jayasinghe, G., Awulachew, S.B., Loulseged, M. and Yilma, A.D., 2012.
Agricultural water management and poverty in Ethiopia. Agricultural Economics, 43,
pp.99-111.
Haile, T., 2008. Impact of irrigation on poverty reduction in Northern Ethiopia. A PhD
dissertation presented to the Department of Food Business and Development National
University of Ireland, Cork, p.200.
 86

Haileyesus, M., 2019. The Contribution of Irrigation to Food Security in Farta Woreda, South
Gondar Zone, Amhara National Regional State, Ethiopia (MSc Thesis, Addis Ababa
University).
Haji, J., Aman, M. and Hailu, T., 2013. Impact analysis of Mede Telila small scale irrigation
scheme on house poverty alleviation: Case of Gorogutu District in Eastern Hararghe
Oromia National Regional State Ethiopia. International Journal of Development and
Economic Sustainability, 1(1), pp.15-30.
Hamda, A.K., 2016. Impact of Small-Scale Irrigation on Household’s Food Security in
Fentale District East Showa Zone of Oromia National Regional State, Ethiopia (MSc
Thesis, Haramaya University).
Hirko, T., Ketema, M. and Beyene, F., 2018. Evaluating the impact of small-scale irrigation
practice on household income in Abay Chomen District of Oromia National Regional
State, Ethiopia. Journal of Development and Agricultural Economics, p.384.
Index, C.C.S., 2008. The coping strategies index: field methods manual, 2nd edition, CARE.
IPC Global Partners, 2019. Integrated Food Security Phase Classification: Technical Manual
Version 3.0: Evidence and Standards for Better Food Security Decisions. Food and
Agriculture Organization of the United Nations. Rome.
Jacob, R., Zhu, P., Somers, M.A. and Bloom, H., 2012. A Practical Guide to Regression
Discontinuity. MDRC.
Jalan, J. and Ravallion, M., 1999. Income gains to the poor from workfare: estimates for
Argentina's Trabajar Program (Vol. 2149). World Bank Publications.
Jirane, A.I., 2015. Determinants of Use of Small-Scale Irrigation and its Effects on Household
Food Security: The Case of Bako Tibe District, West Shoa, Ethiopia (Doctoral
dissertation, Haramaya University).
Kalkidan, Fikirie, Yemiru, T. and Ephrem, T. 2017.Challenges in women•managed small scale
irrigation practices: the case of Lume district, central rift valley of Ethiopia. World
Research Journal of Agricultural Sciences, 3(1): 54•59.
Kinfe, A., Chilot, Y., & Rajan, S. 2012. Effect of small•scale irrigation on the income of rural
farm households: the case of Laelay Maichew district, central Tigray, Ethiopia. Journal
of Agricultural Science, 7(1): 1.
 87

Kennedy, G., Ballard, T. and Dop, M.C., 2011. Guidelines for measuring household and
individual dietary diversity. Food and Agriculture Organization of the United Nations.
Kuwornu, J.K. and Owusu, E.S., 2012. Irrigation access and per capita consumption
expenditure in farm households: Evidence from Ghana. Journal of Development and
Agricultural Economics, 4(3), pp.78-92.
Lechner, M., 2011. The estimation of causal effects by difference-in-difference
methods. Foundations and Trends® in Econometrics, 4(3), pp.165-224.
Lee, D.S. and Lemieux, T., 2010. Regression discontinuity designs in economics. Journal of
economic literature, 48(2), pp.281-355.
Lefore, N., Giordano, M., Ringler, C. and Barron, J., 2019. Viewpoint–Sustainable and
Equitable Growth in Farmer-led Irrigation in Sub-Saharan Africa: What Will it
Take? Water Alternatives, 12(1), pp.156-168.
Lefort, R., 2011. The great Ethiopian land-grab: feudalism, Leninism, neo-liberalism... plus
change. Open Democracy, 31.
Leuven, E. and Sianesi, B., 2018. PSMATCH2: Stata module to perform full Mahalanobis and
propensity score matching, common support graphing, and covariate imbalance testing.
Maxwell, D. and Caldwell, R., 2008. The coping strategies index: field methods manual.
CARE. Maxwell, D. and Wiebe, K.D., 1998. Land tenure and food security: A review of
concepts,
evidence, and methods. Land Tenure Center, University of Wisconsin-Madison.
Maxwell, J.A., 2012. Qualitative research design: An interactive approach (Vol. 41). Sage
publications.
Meier, J., Zabel, F. and Mauser, W., 2018. A global approach to estimate irrigated areas–a
comparison between different data and statistics. Hydrology and Earth System
Sciences, 22(2), pp.1119-1133.
Mengistie, D. and Kidane, D., 2016. Assessment of the impact of small-scale irrigation on
household livelihood improvement at Gubalafto District, North Wollo,
Ethiopia. Agriculture, 6(3), p.27.
Mesele, K.A., Suneetha, P. and Tigga, A., 2018. Determinants of household food security in
southern Tigray, northern Ethiopia.
 88

Molla, A., 2017. Impact of Small-Scale Irrigation on Farm Households’ Food Security: The
Case of Girana Irrigation Scheme in Kobo District, North Wollo Zone of Amhara
Region, Ethiopia (MSc Thesis, Haramaya University)
Mosissa, T. and Bezabih, B., 2017. Review on participatory small-scale irrigation schemes and
small-scale rainwater harvesting technology development and its contribution to
household food security in Ethiopia. International Journal of Water Resources and
Environmental Engineering, 9(3), pp.54-63.
Muez, H.A., 2014. The Impact of Small-Scale Irrigation on Rural household Food Security:
The case of Emba Alaje woreda (Doctoral dissertation, Mekelle University).
Mustapha, M., Kamaruddin, R.B. and Dewi, S., 2018. Factors affecting rural farming
household’s food security status in Kano, Nigeria. International Journal of
Management Research and Reviews, 8(9), p.1.
Ngema, P.Z., Sibanda, M. and Musemwa, L., 2018. Household food security status and its
determinants in Maphumulo local municipality, South Africa. Sustainability, 10(9),
p.3307.
Norton, G.W., Alwang, J. and Masters, W.A., 2014. Economics of agricultural development:
world food systems and resource use. Routledge.\
OIDA, 2014. Oromia Irrigation Development Authority (OIDA). Guidance for Irrigation
Development Project. May 2014.
Otoo, M., Lefore, N., Schmitter, P., Barron, J. and Gebregziabher, G., 2018. Business model
scenarios and suitability: smallholder solar pump-based irrigation in Ethiopia.
Agricultural Water Management–Making a Business Case for Smallholders (Vol. 172).
Owusu, E.S., Namara, R.E. and Kuwornu, J.K., 2011. The welfare-enhancing role of irrigation
in farm households in northern Ghana. Journal of International Diversity, 1, pp.61-87.
Passarelli, S., Mekonnen, D., Bryan, E. and Ringler, C., 2018. Evaluating the pathways from
small-scale irrigation to dietary diversity: evidence from Ethiopia and Tanzania. Food
Security, 10(4), pp.981-997.
Paul, C.J., Weinthal, E.S., Bellemare, M.F. and Jeuland, M.A., 2016. Social capital, trust, and
adaptation to climate change: Evidence from rural Ethiopia. Global Environmental
Change, 36, pp.124-138.
 89

Pérez-Escamilla, R. and Segall-Corrêa, A.M., 2008. Food insecurity measurement and

indicators. Revista de Nutricao, 21, pp.15s-26s.
Ravallion, M., 2008. Evaluation in the Practice of Development. The World Bank.
Regassa, A.E., 2015. Determinants of agro pastoralists participation in irrigation scheme: the
case of fentalle agro pastoral district, Oromia regional state, Ethiopia. International
Journal of Agricultural Research, Innovation and Technology, 5(2), pp.44-50.
Regassa, N., 2011. Small holder farmers coping strategies to household food insecurity and
hunger in Southern Ethiopia. Ethiopian Journal of Environmental Studies and
Management, 4(1).
Roberts, M.R. and Lemmon, M.L., 2007. The response of corporate financing and investment
to changes in the supply of credit. HKUST Business School Research Paper, (07-29).
Rosenbaum, P.R. and Rubin, D.B., 1985. Constructing a control group using multivariate
matched sampling methods that incorporate the propensity score. The American
Statistician, 39(1), pp.33-38.
Sani, S. and Kemaw, B. 2019. Analysis of Rural Households Food Security in Western Ethiopia
Food and Nutrition Sciences, 10, 249-265. https://doi.org/10.4236/fns.2019.103019
Sassi, M., Sassi and Acocella, 2018. Understanding Food Insecurity. Springer.
SIF, 2015. Status and Potential of Groundwater use in Ethiopian: Flood Plain-Spate Irrigation
Foundation (SIF).
Sileshi, A., 2016. An Evaluation on the Effectiveness of Household Irrigation Technologies on
the Lives of Farmers: The Case of Dhuffa, Berfete Tokoffa And Kore Odo Kebeles in
Welmera Wereda, Western Shewa of Oromia, Ethiopia (Master’s Thesis, St. Mary's
University).
Sinyolo, S., Mudhara, M. and Wale, E., 2014. Water security and rural household food
security: Empirical evidence from the Mzinyathi district in South Africa. Food
Security, 6(4), pp.483-499.
Storck, H. and Doppler, W., 1991. Farming systems and farm management practices of
smallholders in the Hararghe Highlands. Wissenschaftsverlag Vauk Kiel.
Tadesse, A., Bosona, T. and Gebresenbet, G., 2013. Rural water supply management and
sustainability: the case of Adama Area, Ethiopia. Journal of Water Resource and
Protection, 5(02), p.208.
 90

Tesfaw, M., 2018. Small Scale Irrigation Development. Irrigation Drainage System
Engineering, 7(206), p.2.
Tesfay, H., 2008. Impact of irrigation development on poverty reduction in Northern Ethiopia.
A PhD thesis. Department of Food Business and Development. National University of
Ireland, Cork. Ireland.
Tesfaye, A., Bogale, A., Namara, R.E. and Bacha, D., 2008. The impact of small-scale
irrigation on household food security: The case of Filtino and Godino irrigation
schemes in Ethiopia. Irrigation and Drainage Systems, 22(2), pp.145-158.
Theis, S., Lefore, N., Meinzen-Dick, R. and Bryan, E., 2018. What happens after technology
adoption? Gendered aspects of small-scale irrigation technologies in Ethiopia, Ghana,
and Tanzania. Agriculture and human values, 35(3), pp.671-684.
Temesgen, B.T., 2019. Determinants of Small-Scale Irrigation Use and Its Implication on
Poverty Reduction: Empirical Evidences from Bogena River Catchment in Awabel
District, East Gojjam, Ethiopia (MSc Thesis, Addis Ababa University).
Tiruneh, W.G. and Geta, E., 2016. Technical efficiency of smallholder wheat farmers: The
case of Welmera district, Central Oromia, Ethiopia. Journal of Development and
Agricultural Economics, 8(2), pp.39-51.
Tolemariam, A., 2010. Impact assessment of input and output market development
interventions by IPMS Project: The case of Gomma Woreda (Doctoral dissertation,
Haramaya University).
Torell, G.L. and Ward, F.A., 2010. Improved water institutions for food security and rural
livelihoods in Afghanistan's Balkh river basin. International Journal of Water
Resources Development, 26(4), pp.613-637.
Tulu, H., 2014. The Effects of Small-Scale Irrigation on Rural Households ‘income: The Case
of Adami Tulu Jido Kombolcha District, Oromia National Regional State.
UNDP, 2018a. Ethiopia’s Progress Towards Eradicating Poverty: Implementation of the third
United Nations decade for Eradication of Poverty (2018 – 2027).
UNDP, 2018b. Human Development Indices and Indicators: Briefing notes on 2018 Statistical
Updates of Ethiopia.
 91

Wakene, B.U., 2018. Determinants of Smallholder Farmers’ Participation in Small Scale

Irrigation: The Case of Deder District of Eastern Hararghe Zone, Ethiopia (MSc
Thesis, Haramaya University)
USAID, 2018. Global Food Security Strategy (GFSS) Ethiopia Country Plan, 22. The U.S
Government’s Global Hunger and Food Security Initiative.
WFS, 1996. Rome declaration on world food security and world food summit plan of action.
FAO, Rome.
Woodhouse, P., Veldwisch, G.J., Venot, J.P., Brockington, D., Komakech, H. and Manjichi,
Â., 2017. African farmer-led irrigation development: re-framing agricultural policy and
investment? The Journal of Peasant Studies, 44(1), pp.213-233.
Wooldridge, J.M., 2010. Econometric analysis of cross section and panel data. MIT press.
World Bank, 2016. Enabling the business of agriculture 2016: Comparing regulatory good
practices.
World Bank, 2018. Ethiopia’s Economic Update. The Inescapable Manufacturing-Services
Nexus: Exploring the potential of distribution services.
World Bank, 2019. An Overview of Ethiopian Economy. World Bank’s goals of promoting
rapid economic growth and improving service delivery.
World Food Program 2017. Conducting Food Security Assessments using Household
Expenditure Surveys (HES): Vam Guidance Paper
World Food Program, 2008. Vulnerability Analysis and Mapping Unit. Lao PDR:
Comprehensive Food Security and Vulnerability Analysis (CFSVA). World Food
Program, Vulnerability Analysis and Mapping Branch (ODAV).
Xie, H., You, L., Wielgosz, B. and Ringler, C., 2014. Estimating the potential for expanding
smallholder irrigation in Sub-Saharan Africa. Agricultural Water Management, 131,
pp.183-193.
Yihdego, A.G., Gebru, A.A. and Gelaye, M.T., 2015. The impact of small-scale irrigation on
income of rural farm households: Evidence from Ahferom Woreda in Tigray,
Ethiopia. International Journal of Business and Economics Research, 4(4), pp.217-228.
Yihdego, M.T.G. and Ghosal, S., 2016. The Impact of Small-Scale Irrigation on Household
Income in Bambasi Woreda, Benishangul-Gumuz Region, Ethiopia.
 92

Zhang, W., 2008. A forecast analysis on world population and urbanization

process. Environment, Development and Sustainability, 10(6), pp.717-730.
 93

7. APPENDICES

7.1. List of Tables in Appendices

Table in Appendix 1Conversion Factor for kilocalories per gram for different food items

Food items Unit Kilocalorie/kg

Maize Kg 3360
Sorghum Kg 3805
Millet Kg 3260
Tef Kg 3589
Wheat Kg 3574
Potato Kg 1037
Sweet potato Kg 1360
Lentil Kg 3523
Broad bean Kg 3514
Cheek peas Kg 3630
Field peas Kg 3553
Onion Kg 713
Pepper Kg 933
Beef Kg 1148
Milk Litter 860
Egg No 61
Sugar Kg 3850
Butter Kg 7363
Meat Kg 1148
Honey Kg 3605
Edible oil Kg 8960
Vetch Kg 3470
Garlic Kg 118
Vegetables Kg 370
Spices Kg 2970
Salt Kg 1780
Coffee Kg 1103

Source: EHNRI, (2000)

 94

Table in Appendix 2 Conversion factors for Tropical Livestock Units (TLU)

Livestock Category TLU Livestock Category TLU

Oxen and cow 1.00 Horse and mule 1.1
Bull/heifer 0.75 Sheep and goat (adult) 0.13
Calf 0.25 Sheep and goat (young) 0.06
Donkey (adult) 0.70 Chicken 0.013
Donkey (young) 0.35 Camel 1.25

Source: Storck, et al., 1991

Table in Appendix 3 Conversion factor for Adult Equivalents (AE)

Age group Male Female

0-1 0.33 0.33
1-2 0.46 0.46
2-3 0.54 0.54
3-5 0.62 0.62
5-7 0.74 0.7
7-10 0.84 0.72
10-12 0.88 0.78
12-14 0.96 0.84
14-16 1.06 0.86
16-18 1.14 0.86
18-30 1.04 0.8
30-60 1 0.82
60+ 0.84 0.74

Source: Storck, et al., 1991

 95

7.2. Lists of Figures in Appendices

Figures in Appendix 1 Histogram of propensity score by participation in irrigation

Figures in Appendix 2 Graphical presentation of propensity score

 96

Figures in Appendix 3 Graph of propensity score for unmatched and matched covariates

7.3. Questionnaires Used for Data Collection

Dear Respondent,

First of all, I would like to thank you for your willingness to participate in this study. The
questionnaire is designed to collect the necessary information to undertake a research on the
impact of small-scale irrigation on household food security in Bishoftu town The study is
conducted for the partial fulfillment of the requirements for the degree of Masters in
Development Economics.

The main objective of this research is to investigate the impact of small-scale irrigation on
household food security and all your answers will be used for this purpose. So, please answer
each question genuinely since your genuine responses are quite vital for the success of this
study. Finally, I would like to confirm you that all the information you provide in this
questionnaire will be strictly confidential and will exclusively use for this research purpose
only.

Thanks in advance!
 97

Survey Questionnaire on the Impact of Small-Scale Irrigation on Household Food Security: The
Case of Bishoftu Town.

1. Household Characteristics
1.1. Kebele Interviewer Date
1.2. Household name code Tele
1.3. Respondent Name (if different from the household head)
1.4. Ethnicity: 1. Oromo 2. Amhara 3. Gurage 4. Other
1.5. Religion: 1. Orthodox 2. Muslim 3. Protestant 4. Waaqeffataa 5. Other
1.6. Sex of the household 0. Female 1. Male
1.7. Age of the household head
1.8. Education level of the household head: Years of formal schooling
1.9. Marital status: 1. single 2. Married 3. Divorced 4. Widowed
1.10. Family Information: Use the following table

Table 1 Family member information

S. N Age Classification Male Female

1 0-1
2 1-2
3 2-3
4 3-5
5 5-7
6 7-10
7 10-12
8 12-14
9 14-16
10 16-18
11 18-30
12 30-60
13 60+
 98

2. Livestock holding during 2016 E.C.

2.1. Do you have livestock? 0. No 1. Yes
2.2. If yes for Q#2.1, indicate the number of livestock in the following

table Table 2 Summary of Livestock holding

Types of Livestock Number Owned Number Sold Income Obtained

Oxen
Cows
Bulls
Heifers
Calves
Horses (Adult)
Horses (Young)
Donkeys (Adult)
Donkeys (Young)
Goat (Adult)
Goat (Young)
Sheep (Adult)
Sheep (Young)
Poultry
Bee colony

3. Land holding and usage

3.1. Do you have your own land? 0. No 1. Yes
3.2. If yes, total land size ha; irrigable ha, and non-irrigable ha.
3.3. Do you think that your piece of land is enough to support your family? 0. No, 1. Yes
3.4. If no to Q#3.3, what are the reasons? 1. Low soil fertility, 2. Small land size,
3. Insufficient application of inputs, 4. Large family size, 5. others
3.5. What is the total land you cultivated in 2016E.C? total ha: irrigated rainfed
3.6. Did you shared-in/leased-in land during 2016 E.C? 0. No 1. Yes
3.7. If yes to Q#3.6, total land size ha; irrigated ha, and rainfed ha.
 99

3.8. How do you evaluate your level of production during the last three years? 1. excess of
family consumption, 2. sufficient for household consumption, 3. insufficient for family
consumption, 4. Others
4. Farming experience, irrigation utilization and income from farming
4.1.
How many years since you started farming (Farming experience) years.
4.2.
Did you produced crop under rain fed during 2016E.C? 0. No 1. Yes
4.3.
If yes to Q#4.2, what were the crops produced? Use the following table.

Table 3 Crop Production under Rainfed (Qt)

Crop Own
Total yield (Qt) Sold Given to others
consumed
Type Amount Value Amount amount Income Amount

4.4.
Did you produced crop using irrigation during 2016? 0. No 1. Yes
4.5.
If yes for Q#4.4, when did you started irrigation farming? ___________________
4.6.
If yes for Q#4.4, what are the crops produced in 2016 E.C? use the following table

Table 4 Crop Production under irrigation

Crop Own
Total yield (Qt) Sold Given to others
consumed
Type Amount Value Amount amount Income Amount
 10
0
4.7.
If yes for Q#4.4, what is the source of water for your irrigation? 1. Gravity River
diversion 2. Pumping rivers, 3. Ground water, 3. Ponds, 4. Others
4.8.
If yes for Q#4.4, how many times do you produce per year using irrigation?
4.9.
Have you ever faced food shortage? 0. No 1. Yes
4.10.
If yes for Q#4.9, during which months? according to severity
4.11.
How did you cope during food shortage? 1. Sale of livestock 2. Reduce the number of
meals 3. Sale of Animal products 4. Wage employment 5. Other
5. Distance from irrigation water sources km, or minutes of walk.
6. Non/ off-farm income in 2016 E.C
6.1.
Have you got off/non–farm income during 2016 E.C? 0. No 1. Yes
6.2.
If yes, indicate the amount and sources of income earned from non/off-farm:

Table 5 Types and magnitude of off/non-farm income

No Income Sources (use code) Annual income

1
2
3
4
5
6
7
8

Codes: 1=Pity trade, 2=Selling firewood /charcoal, 3= Renting land, 4=Remittance, 5=local
beverages, 6= Handcraft, 7=Cart, 8. Car/Bajaj, 9=Food aid, 10=others ______

7. Physical, Institutional and environmental factors

7.1.
Did diseases and pests affect your crop production in 2016E.C? 0. No, 1. Yes
7.2.
If yes for Q#7.1, what is the estimated loss occurred? birr.
7.3.
If yes for Q#7.1, what percent of the total production is damaged? percent.
7.4.
If yes for Q#7.1, how did you treat it? 1.spraying chemicals, 2. Cultural treatments,
 10
1

3. No any treatment, 4. others

7.5.
Is there farmers training center (FTC) in your kebele? 0. No 1. Yes
7.6.
How far is the FTC from your home? km.
7.7.
Do you have contacts with DA? 0. No 1. Yes
7.8.
If yes, how many times do you contact with them per month?
7.9.
Do you get different advices/ supports from DAs? 0. No 1. Yes
7.10.
If yes to Q7.9, do you practically use their advices accordingly? 0. No, 1. Yes
7.11.
If no to Q7.10, why? 1. is irrelevant 2. not timely, 3. not affordable, 4. Other
8. Distance to the nearest market
8.1.
How far is your home from the nearest market? km.
8.2.
Where do you sell m a j o r i t y o f your farm products? 1. On farm, 2. a t
Cooperative union shops, 3. at local markets, 4. Others
8.3.
Do you get market information about price of inputs and outputs? 0 = No 1 = Yes
8.4.
Did you get good price for your produce in 2016 E.C? 0. No 1. Yes
8.5.
If no, what are the reasons? 1. low demand for the products, 2. higher supply of the
products, 3. Lak of information, 4. Lack of road access, 5. Others
8.6.
How do you transport your produce to the market? Using: 1. Vehicle 2. Animal power, 3.
Cart, 4. Human power 5. Others
9. Access to credit services
9.1.
Do you have access to credit for your agricultural activities? 0. No 1. Yes
9.2.
If yes for Q#9.1, did you used credit services during 2016 E.C? 0. No, 1. Yes
9.3.
If yes for Q#9.2, the source? 1. Commercial banks, 2. Cooperative unions,
3. Neighbors and relatives, 4. Micro finance institutions 5. Other
9.4.
If yes for Q#9.2, what was the amount of credit that you have received? Birr.
9.5.
If yes for Q#9.2, for what purpose? 1. Purchase of seeds 2. Purchase of fertilizer,
3. Purchase of oxen, 4. For family consumption, 5. Others
9.6.
If no to Q#9.2, what are the reasons? 1. Collateral problem 2. No need of credit,
3. High interest rate, 4. Inadequate supply, 5. No access, 6. Others
 10
2

10. Food Consumption Score Data Collection Question.

Table 6 Types and frequencies of Food items consumed during the last seven days.

S. N Food Items No of days the weight Food consumption

food item eaten score Value
Main staples (Maize,
Rice, Sorghum,
Pasta, Bread/Cake,
1 2
Potato, Yam,
Cassava, Sweet
Potato, Taro, etc.)
Pulses (Beans, Peas,
2 3
Groundnuts and etc.)
Vegetables (Carrot,
Red Pepper, Sweet
Potato, Onion,
3 1
Tomato, Green
Beans, Peas, Lettuce,
etc.)
Fruit (Mango,
Papaya, Banana,
4 1
Apple, Lemon,
Orange)
Meat and fish (Beef,
5 Goat, Poultry, Pork, 4
Egg and Fish)
Milk (Milk, Yoghurt,
6 4
Cheese, etc.)
Sugar (Sugar, Honey,
7 0.5
etc.)
Oil (Vegetable Oil,
8 Palm Oil, Butter, 0.5
etc.)
Condiments (Spices,
9 Tea, Coffee, Salt, 0
Fish Power, etc.)
 103

11. Daily Calorie Intake Data Collection Question

Table 7 Types and Amounts of Food consumed during the past 24 hours

Food Type Amounts consumed in the last 24 hours in Kg.

Tef
Maize
Wheat
Barley
Sorghum
Rice
Beans
Peas
Chickpea
Meat
Fish
Chicken
Egg
Milk
Butter
Cheese
Potatoes
Sweet potato
Tomatoes
Onion
Pepper
Garlic
Cabbage
Sugar
Honey
Salt
Oil
Coffee
Others
 104

12. General opinion of respondents

12.1.
Please list all problems associated with irrigation development activities in your area.
a.
b.
12.2.
What interventions must be made for better implementation of small-scale irrigations?
a.
b.

Thank you for your cooperation and confidential responses!!!