Mahmud - Determinants of Agricultural Productivity and Off Farm Household Income in Rural Ethiopia, MSC Thesis
Mahmud - Determinants of Agricultural Productivity and Off Farm Household Income in Rural Ethiopia, MSC Thesis
Mahmud - Determinants of Agricultural Productivity and Off Farm Household Income in Rural Ethiopia, MSC Thesis
MSC THESIS
MAHMUD ABDUROHMAN
May, 2018
Mahmud Abdurohman
May, 2018
As thesis research advisor, I hereby certify that I have read and evaluated this thesis prepared
under my guidance by Mahmud Abdurohman entitled Determinants of Agricultural
Productivity and Off-farm Household Income in Rural Ethiopia. I recommend that it be
submitted as fulfilling the thesis requirement.
As members of the Board of Examiners of the M.Sc. thesis open defense examination, we
certify that we have read and evaluated the thesis prepared by Mahmud Abdurohman and
examined the candidate. We recommend that the thesis be accepted as fulfilling the thesis
requirements for the degree of Masters of Science in Economics (Development Policy
Analysis).
I dedicate this thesis manuscript to all Ethiopian rural poor people who are in the way of
struggle on livelihood improvement.
STATEMENT OF THE AUTHOR
By my signature below, I declare and affirm that this thesis is my own work. I have followed
all ethical principles of scholarship in the preparation, data collection, data analysis and
completion of this thesis. All scholarly matter that is included in the thesis has been given
recognition through citation. I affirm that I have cited and referenced all sources used in this
document. Every serious effort has been made to avoid any plagiarism in the preparation of
this thesis.
This thesis is submitted in partial fulfillment of the requirement for a degree from the School
of Graduate Studies at Wolaita Sodo University. The thesis is deposited in Wolaita Sodo
University Library and is made available to borrowers under the rules of the library. I
solemnly 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 may be used without special permission provided that
accurate and complete acknowledgement of the source is made. Requests for permission for
extended quotations from, or reproduction of, this thesis in whole or in part may be granted by
the head of the School or 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 interest of scholarship. In all
other instances, however, permission must be obtained from the author of the thesis.
Department: Economics
ABBREVIATIONS
EA Enumeration Area
FE Fixed Effect
HH Household
IV Instrumental Variable
RE Random Effect
i
SNNP Southern Nations Nationality and Peoples
ii
BIOGRAPHICAL SKETCH
The author was born in November 4, 1993 in Amhara Regional State, South Wollo Zone of
Kalu District from his mother Fatimah Said and his father Abdurohman Seid in a village
called Agerdaru. He attended his elementary education at Gerba elementary school; secondary
and preparatory education at three different schools namely Bulbula Secondary School in
Addis Ababa, Degan Secondary and Preparatory School in Degan and Hawas Preparatory
School in Adama. Then he joined Addis Ababa University in September 2013 and graduated
with BA degree in economics in June 2016. After his graduation he employed by Ministry of
Education as an assistant lecturer in September 2017. Flowing employment, the author
immediately joined the school of graduate studies of Wolaita Sodo University to follow his
M.Sc study in Economics with specialization of Development Policy Analysis.
iii
ACKNOWLEDGMENT
First of all I deliver my whole hearted gratitude to the Almighty ALLAH that endowed me
with an innate ability to recognize and acknowledge His existence. Without His grace and
wish, I would not be hear where I am now and obviously this work cannot be accomplished
successfully. May ALLAH send His praises upon the holly and beloved prophet, Mohammed
bin Abdullah, who brought the message of peace and happiness to all creatures.
After that I would like to extend my deepest appreciation and gratitude to my thesis Advisor
Tadele Tafese (Assist. Prof.). By his scholarly constructive criticism, guidance, countless
fruitful discussions and ever-teaching comments, he used to bring me back to the track,
recharge and stimulate my enthusiasm for pressing on further. Undoubtedly, his innovative
comments greatly improved the content of this thesis.
Dr. Atlaw Alemu from Addis Ababa University, Dr. Bidyut Ghosh from Symbiosis
International University of India and Mr. Ayana Anteneh, head of economics department at
Wolayta Sodo University are highly indebted to my heart full thanks for their technical and
moral support throughout the study period. My special thanks once again goes to CSA and the
World Bank International Standard Living Measurement (ISLM) team for providing the data I
used in this study, free of charge and without bureaucratic complications.
I owe more than I can say to my family for their unreserved help, care and encouragement
throughout my study. Really their contribution in my life cannot be undermined. In this
special moment I would like to thank my best friend Ali Indris. He is on the side of me in all
respects. Finally, all of my lovely post graduate colleagues; I am delighted with the
commitment we had made during our two years study. I would like to say let we please with
our dedication.
Mahmud Abdurohman
May, 2018
iv
TABLE OF CONTENT PAGE
ABBREVIATIONS ................................................................................................................... I
BIOGRAPHICAL SKETCH ................................................................................................ III
ACKNOWLEDGMENT ....................................................................................................... IV
LIST OF TABLES .............................................................................................................. VIII
LIST OF FIGURES ............................................................................................................... IX
LIST OF TABLES IN THE APPENDIX .............................................................................. X
ABSTRACT ............................................................................................................................ XI
1. INTRODUCTION .........................................................................................1
1.1. BACKGROUND OF THE STUDY....................................................................................... 1
1.2. STATEMENT OF THE PROBLEM ...................................................................................... 3
1.3. OBJECTIVES OF THE STUDY .......................................................................................... 6
1.3.1. General objective ............................................................................................................. 6
1.3.2. Specific objectives ........................................................................................................... 6
1.4. RESEARCH QUESTIONS ................................................................................................. 6
1.5. SIGNIFICANCE OF THE STUDY ....................................................................................... 7
1.6. SCOPE OF THE STUDY ................................................................................................... 7
1.7. LIMITATIONS OF THE STUDY......................................................................................... 7
1.8. ORGANIZATION OF THE THESIS..................................................................................... 8
2. LITERATURE REVIEW ...............................................................................9
2.1. THEORETICAL REVIEW ..................................................................................................... 9
2.1.1. Basic concepts and definitions ......................................................................................... 9
2.1.2. The relationship between agricultural productivity and poverty alleviation ................. 10
2.1.3. Linkages between off-farm income and agricultural productivity................................. 14
2.1.4. Agricultural production technologies and productivity ................................................. 16
2.1.5. Impact of macroeconomic variables on agricultural productivity ................................. 18
2.1.6. Household characteristics nexus agricultural productivity ............................................ 19
2.1.7. Demographic and environmental factors vis-a-vis agricultural productivity ................ 21
2.1.8. Factors affecting off farm rural household income ........................................................ 22
2.1.9. Theoretical model for analyzing the determinants of agricultural productivity and its
measurement .................................................................................................................. 24
v
Table of content (continued)
2.2. EMPIRICAL REVIEW ........................................................................................................ 25
2.2.1. Agricultural productivity and off-farm rural household income in developing
Countries........................................................................................................................ 25
2.2.2. Agricultural productivity and off farm rural household income in Ethiopia ................. 29
3. RESEARCH METHODOLOGY .................................................................32
3.1. DATA TYPE AND SOURCES ............................................................................................. 32
3.2. SAMPLE DESIGN ............................................................................................................. 32
3.3. METHODS OF DATA ANALYSIS ....................................................................................... 33
3.4. ECONOMETRIC MODEL SPECIFICATIONS ......................................................................... 33
3.4.1. Empirical agricultural productivity model specification................................................ 33
3.4.2. Estimation technique ...................................................................................................... 36
3.4.3. Variables definition and working hypothesis................................................................. 38
3.4.4. Empirical model specification for off-farm rural household income............................. 42
3.4.5. Estimation technique ...................................................................................................... 43
3.4.6. Hypothesis and definition of variables for off-farm rural household income model ..... 43
4. RESULTS AND DISCUSSION ....................................................................46
4.1. DESCRIPTIVE ANALYSIS OF THE SAMPLE DATA.............................................................. 46
4.1.1. Summary of agricultural productivity and significance test for its variation across years
and regions of the country ............................................................................................. 46
4.1.2. Summary of rural off-farm income and significance test for its variation across years
and regions of the country ............................................................................................. 50
4.2. DESCRIPTIVE STATISTICS OF VARIABLES USED IN THE ECONOMETRIC MODEL .............. 53
4.2.1. Descriptive statistics of variables used in the agricultural productivity model ............. 54
4.2.2. Descriptive statistics of variables used in the off-farm household income model ......... 55
4.3. ECONOMETRIC RESULTS AND DISCUSSION ..................................................................... 56
4.3.1. Econometric results for agricultural productivity model ............................................. 57
4.3.2. Econometric results for rural off-farm household income model .................................. 63
5. SUMMARY, CONCLUSIONS AND RECOMMENDATIONS ...............69
5.1. SUMMARY ...................................................................................................................... 69
5.2. CONCLUSIONS AND RECOMMENDATIONS ....................................................................... 70
vi
Table of content (continued)
REFERENCES ...................................................................................................73
APPENDIXES ....................................................................................................87
APPENDIX I............................................................................................................................ 87
APPENDIX II .......................................................................................................................... 89
APPENDIX III ......................................................................................................................... 90
APPENDIX IV ......................................................................................................................... 96
vii
LIST OF TABLES
Table page
1. Summary of agricultural productivity across regions for each year and its test for variation
significances across regions................................................................................................. 48
2. Summary of agricultural productivity across years and its test for variation significance . 50
3. Summary of off-farm household income across regions for each year and its test for
variation significances across regions ................................................................................. 52
4. Summary of total off-farm household income across years and its test for variation
significance .......................................................................................................................... 53
5. Summary statistics of variables used in agricultural productivity model ........................... 54
6. Summary statistics of variables used in off-farm rural household income model .............. 56
7. Random effect’s IV regression output for the determinants of agricultural productivity in
Ethiopia, 2012-2016 ........................................................................................................... 61
8. Random effect’s IV regression output for the determinants of off-farm household income in
rural Ethiopia, 2012-2016 .................................................................................................... 67
viii
LIST OF FIGURES
ix
LIST OF TABLES IN THE APPENDIX
1. Correlation between the variables used in the regression for agricultural productivity
model after off farm income variable is instrumented ........................................................ 87
2. Correlation between the variables used in the regression for rural off-farm household
income model after productivity variable is instrumented .................................................. 88
3. Fixed effect IV regression output for agricultural productivity model ............................... 90
4. Random effect IV regression output for agricultural productivity model........................... 91
5. Hausman test result for agricultural productivity model..................................................... 92
6. Fixed effect IV regression output for off-farm household income model .......................... 93
7. Random effect IV regression output for off-farm household income model .................... 94
8. Hausman test result for off- farm household income model ............................................... 95
9. The regional distribution of EAs, sampled households and actual households included in
the study ............................................................................................................................. 96
x
DETERMINANTS OF AGRICULTURAL PRODUCTIVITY AND OFF-FARM
HOUSEHOLD INCOME IN RURAL ETHIOPIA
ABSTRACT
In Ethiopia agriculture plays an important role and has profound implications in ensuring
quality of life and sustainable development of the rural areas. Even though the country has
highly productive resources, at present the sector is dominated by low-productivity. The
largest population in rural areas of the country is unable to get the expected farm output. The
main objective of this study was to investigate the factors that determine agricultural
productivity and off-farm household income in rural Ethiopia. The study employed both
descriptive (One Way ANOVA and Bonferroni tests) and random effect instrumental
regression over 1,915 sampled rural households of the Ethiopian Socioeconomic Survey
panel data to hit its objective. Descriptive analysis of the study indicated the existence of
significant variation in the level of agricultural output both across regions and time but with
significant variation of off-farm income across regions only. The random effect instrumental
variable regression output showed that land holding size, ownership of oxen, agricultural
labour input, educational level of the household head, gender of the household head, use of
chemical fertilizer, drought, use of improved seeds, use of extension services and annual off-
farm household income were the main determinants of agricultural productivity in Ethiopia.
On the other hand age of household head, education level of the household head, annual
agricultural output, number of dependents in the household and access for electricity are
found as the main factors that determine off-farm rural household income. Based on these
findings, the researcher recommended that emphasis should be given towards strengthening
different educational opportunities, asset building programs, modern agricultural
technologies, drought mitigation strategies, and increasing off-farm income through the
application of rural electrification and agricultural input diversification.
xi
1. INTRODUCTION
Scholars have long pointed to agricultural development as the most important contribution
toward national economic development. Great Britain, France, and the United States,
according to most economists, have followed this trajectory to industrialization and economic
expansion. Agriculture is a proven path to prosperity. No region of the world has developed a
diverse, modern economy without first establishing a successful foundation in agriculture.
This is going to be critically true for Africa where, today, close to 70% of the population is
involved in agriculture a smallholder farmers working on parcels of land that are, on average,
less than 2 hectares(AGRA, 2017). As such, agriculture remains Africa’s surest bet for
growing inclusive economies and creating decent jobs mainly for the youth.
The major objective of agricultural and rural development in developing nations is the
progressive improvement in rural levels of living achieved primarily through increases in
small farm incomes, output, and productivity, along with genuine food security as a route out
of poverty and toward genuine rural development. Enhancement of cereal productivity
1
represents only a small part of the agricultural opportunities (Todaro and Smith, 2012). The
New Economics of Labor Migration postulates that increased non-farm income may relax
rural households’ financial constraints and increase investment in new farming technologies.
It is important to identify the principal factors for agricultural progress with in the sector and
the basic off farm income conditions essential to its achievement (Stark and Bloom, 1985).
Ethiopia is one of the largest countries in Africa both in terms of land area, 1.13 million km2
of which 51.4 million hectares is arable land and population of 94.4 million of which rural
population accounts about 75.2 million whose livelihood depends on agriculture (CSA,
2017).The sector contributes 24.5 % for GDP growth in 2015 (UNDP, 2017). In spite of huge
agricultural potential, the growth in agricultural production has not been able to keep pace
with that of the demand (CSA, 2017). About one third of rural household’s farm is less than
0.5 hectares which, under rain fed characteristics of Ethiopian agriculture, cannot produce
enough food to meet their requirements (MOA, 2010).As a result, yields are low and farmers
can be trapped in a circle of poverty and food insecurity for decades (UNECA, 2009). In order
to came out from this trap; to be self-ensured from the enate variability of farm income and to
stabilize their total household income, the Ethiopian rural society uses both farm and nonfarm
activities.
In Ethiopian economy, agriculture plays an important role and has profound implications in
ensuring quality of life and sustainable development of the rural area. It provides for 85 % of
the labor force employment and accounts over 43 % of the GDP and about 70 % of export
revenue in 2013. Due to its importance, the Ethiopian government gives high priority to the
agriculture Sector by planning and implementing different strategies and allocating at least
10% of its total public expenditure on the sector (MOA, 2010). Agricultural Development Led
Industrialization (ADLI) is the central pillar of the economic policy of the country. The
Sustainable Development and Poverty Reduction Program (SDPRP), a Plan for Accelerated
and Sustained Development to End Poverty (PASDEP) and the recent two Growth and
Transformation Plans (GTP I and GTP II ) are some of the development strategies of the
government. Despite the focus of the government on the agricultural sector through different
2
strategies, the rural communities are unable to produce enough food to feed them throughout
the year (Bihon, 2015).
Many efforts have been made to lay agricultural foundation for industrial development in
Ethiopia. National development plans, notably the Growth and Transformation Plans (GTP) I
(2010/11-2014/15) and II (2015/16-2019/20) have placed emphasis on promoting
industrialization. In particular, GTP I identified the agricultural sector as a foundation for
industrialization and prioritized structural transformation from agriculture to manufacturing.
For achieving the plans government has made many efforts such as improving productivity of
smallholder farmers by disseminating effective technologies through the scaling up strategy
and conserve natural resources and improving irrigation facilities. Despite efforts made, the
share of the industrial sector in GDP remains low, averaging 12.2% between 20011/12 and
2015/16 (UNDP, 2017). Access to technology and capital were among the major bottlenecks
in the process of agricultural transformation (GTP II, 2015/16).
Agricultural development needs timely and adequate supplies of essential farm inputs.
Farmers in Ethiopia use their off farm income as a principal means for achieving their goal of
productivity. However, the investment capacity of majority of Ethiopian farmers is low, the
poor farm household cannot afford to meet increased demand for the purchase of improved
seeds, recommended quantity of fertilizer, buying or hiring of farm machinery etc; so lack of
finance is one of the main reason for low productivity in our agriculture (Tessema, 2015). The
decline of productivity in agricultural sector and unimproved level off- farm household
income are interrelated factors which are highly responsible for the Ethiopian rural population
to live in poverty trap now a day.
Poverty reduction has long been the foremost development priority in Ethiopia. Yet despite
progress toward eliminating extreme poverty, Ethiopia remains one of the poorest countries in
the world, due to both low agricultural productivity and low starting base. Ethiopia has
recorded an economic performance of 10.2 % and agricultural sector grew by 6.4 % in
3
2014/15. However, Growth in the rain dependent agricultural sector slowed due to the
induced EL Nino drought in 2015/16, decreasing from 6.4% in 2014/15 to 2.3% in 2015/16
(UNDP, 2017). Its share in GDP also slightly declined from 40.1 % in 2013/14 to 38.8 % in
2014/15 (NBE, 2015) and to 36.2 % in 2016/17 (IMF, 2017). As a developing country, this
decline in the share of agriculture in total production and employment poses different
challenges across regions (FAO, 2017). Whereas, its share in GDP has been declining steadily
in the past decade, Agriculture continues to be the backbone of the Ethiopian economy (NBE,
2015).
Based on the past experiences gained from agricultural sector, the Ethiopian government have
committed to rapid growth of agriculture as a means of accelerating the economic
transformation. In the transformation plan, vertical and horizontal linkages between
agriculture and industrial sector have strongly advocated. While progress in industrialization
specially in expanding the manufacturing sector is not satisfactory. At present the economy is
dominated by low-productivity agriculture on potentially highly productive land and labour
resources; the bulk of the people live in rural areas and many are isolated from the requisites
for a high level of agricultural productivity.
This low level of productivity is occurred due to many factors. Firstly, in Ethiopia, farmers
have not enough capacity to invest in their agricultural activity. Secondly, the sector itself has
not shown dynamic change due to a number of internal and external challenges. Therefore, to
identify directions of development that will increase the productivity and contribution of this
sector, it is significantly essential to investigate factors which have great impact on
agricultural productivity and off-farm household income in rural Ethiopia.
So far, in Ethiopia, different income and productivity focused researches have been carried
out. Anne (2011), on his survey of agricultural productivity and nutritional status of rural
South Wollo shows agricultural input and livestock ownership has an impact on household
wellbeing. A study in Tigray region by Bihon (2015) using cross sectional data indicates that,
agricultural production in Tigray region is determined by household characteristics, physical
environment, agricultural technologies, institutional factors, off-farm participation and
4
agricultural marketing issues. There are also other researchers who concern on identifying the
determinants of farm income and diversification of off farm activities (Mohammed (2013),
Tesfaye (2014), Prowse and Yisihak (2015), Beyan (2016)). However, all of these studies are
geographically limited. As Ethiopia has diverse climatic zones, results obtained in a specific
region or area cannot be generalized for the whole country. Therefore, there is a need of
further investigation on the determinants of agricultural productivity and off-farm household
income at the national level.
Yishak (2017) uses cross sectional data to analyze rural farm household’s income
diversification in Wolayta zone. His result revealed that factors such as sex, farm size,
livestock ownership, oxen ownership, education, leadership, annual cash income and market
distance were key determinants of farmer’s participation in income diversification. Similarly
Birhanu (2014) uses OLS and probit regression models to determine the factors which reduce
farmers off farm participation in Tigray region by using cross sectional data. His result
indicates gender, age, education, family size, draft animals, location and amount of credit
taken are significantly determine the probability of farmers participation in off farm activity.
Beyond their geographical limitation of these studies, using cross sectional data for analysis
has many problems. Since, it is a snapshot data; it is difficult to study dynamics of changes on
the variables. In addition to this, in cross sectional data due to small sample size, it is difficult
to obtain precise estimates. To overcome all these weaknesses of cross sectional data, we have
used panel data for our analysis which gives more variability, less colinearity among variables
and more degree of freedom.
Only limited inconclusive researches have done at the national level to show the determinants
of Ethiopian agricultural productivity and rural household income. Therefore, conducting this
study was found to be highly significant to narrow the existed knowledge gap and provide up-
to-date information. As far as my knowledge no one of previous research in Ethiopia tries to
show the empirical relationship between the household’s farm and off-farm livelihood in rural
Ethiopia which is part of my research objective.
5
Hence, with the above mentioned gaps, this study have intended primarily on identifying,
analyzing, and documenting the factors that affect agricultural productivity and off- farm
household income in rural Ethiopia. It has also tried to investigate the empirical relation
between agricultural productivity and off-farm household income in Ethiopia.
The prime objective of this study is to investigate the main determinants of agricultural
productivity and off-farm household income in rural Ethiopia. In order to address this main
objective we have focused on the following specific objectives.
The paper takes advantages of the available recently collected nationally representative
surveys and seeks to answer the following questions:
1. Does agricultural productivity and off- farm household income vary across years and
different regions of the country? If yes, to what extent?
2. What are the major factors which affect agricultural productivity in Ethiopia?
6
3. Which determinants of off-farm income are the most potent for the change in rural
household income of the country?
4. Is there any empirical relation between off-farm rural household income and
agricultural productivity in Ethiopia?
Ethiopia has taken agricultural sector as a base for economic transformation. Therefore, it is
important for policy makers to know the critical factors that could accelerate or reduce the
agricultural productivity and household income in the country’s rural area. As my research is
targeted on identifying factors which have significant impact on agricultural productivity and
household off-farm income, the findings of this study have significant benefit for the
stakeholders to drive and implement appropriate policies for enhancing agricultural
productivity and hence to transform the economy. It also farther open up the new research
areas which have not been covered by this study and serve as important addition to the
existing literature on the household income and productivity determinants.
Despite the possibility of including the existed small towns and urban agriculture of the
country in our analysis, in this study we have excluded them and completely focused on
agriculture of rural areas. In Ethiopia agricultural production comprises crop, fruit and
vegetable and livestock subsectors. However, our study does not concern on the livestock
subsector. In addition to this, due to the nature of the study, we have used only quantitative
type of analysis to address all of our objectives.
This study is limited only with secondary data sources, as it is difficult to collect panel data in
a fixed time span. Furthermore, it has depended on the current three years of panel data for the
analysis due to lack of continuous data for many years in our country.
7
1.8. Organization of the Thesis
The thesis is constructed with five chapters. The first chapter is concerned with the
introductory part, which comprises the background, statement of the problem, objectives of
the study, significance of the study and scope and limitation of the study. The second chapter
deals with the review of theoretical and empirical literature pertinent to the concern of the
study. The third chapter focused on research methodology that included a brief description of
how and who collected the secondary data, procedures, analytical model and techniques of
estimation methods. The chapter also briefly discussed procedures followed in data collection,
estimation, model used and hypothesis settings. Results and discussions of the study are
presented in the fourth chapter. Finally we have presented summary, conclusions and
recommendations of the study in the last chapter.
8
2. LITERATURE REVIEW
The term productivity has been used in different meanings and has aroused many conflicting
interpretations. Pandit (1965) has expressed the connotation of productivity as “Productivity
is defined in economics as the output per unit of input… the art of securing an increase in
output from the some input or of getting the same output from a smaller input”. From this
agricultural productivity can be defined a measure of efficiency with which an agricultural
production system employs land, labour, capital and other resources. Agricultural productivity
is the major determinant of the income gap that separates LDCs from other developing
countries and developed countries. Low productivity in agriculture is a major reason for the
prevalence and persistence of poverty in most LDCs.
There are two measures of agricultural productivity, partial factor productivity (PFP) and
total factor productivity (TFP). Partial factor productivity is a ratio of output to a subset of the
inputs, usually one input, described as single-factor productivity. Two commonly used
measures of PFP are land productivity (defined as the ratio of output to total harvested area)
and labor productivity. Total factor productivity, conceptually also a measure of output to
inputs, is commonly measured as an index of the ratio of total agricultural outputs to total
agricultural inputs. As such, TFP analysis can be seen as an extension of PFP analysis, since
the variables used in measuring PFP are included in the variables used in measuring TFP
(IFPRI, 2016). Regardless of which measure is used, empirical studies support the idea that
improvements in agricultural productivity are important for poverty reduction (Mellor, 1999).
9
General concepts on off-farm income
Off-farm income and non-farm income are used interchangeably in several studies. These
studies on off farm income are overwhelmed by vague definitions. The difference between
the two is that off-farm income is much broader than non-farm income and it is made up of
agricultural wage income plus non-farm income. Some authors adopt non-farm income, which
exclude income from agricultural employment on other people’s farm. According to
Haggblade et al. (2007) off-farm income means off the owner’s own farm that includes wage
income in agriculture earned on other people’s farms along with nonfarm earnings from the
owner’s nonfarm enterprises or from nonfarm wage earnings. Thus, off-farm income is the
sum of rural nonfarm income and wage earnings in agriculture.
Off-farm household income is generated when a farmer, spouse or other family member
works off of the farm, thereby generating extra income for the family. In this moment it is
essential to know what does household means in our context. The meaning of household has
two dimensions. A household may be either a one person household, that is a person who
makes provisions for his own living without combining with any other person to form part of
a multi- person household or it may be a multi-person household, that is, a group of two or
more persons who live together and make common provisions for food and other essentials of
living (CSA, 2017). The persons in the group may pool their incomes and have a common
budget to a greater or lesser extent. They may be related or unrelated persons or a
combination of both. These persons are taken as members of the household and get their
income either in cash or often in kind.
Around the world, agriculture will continue to be a major building block in the achievement
of the Millennium Development Goals (MDGs). The contribution of agriculture sector to
poverty reduction depends on its own growth performance, its indirect impact on growth in
other sectors, the extent to which poor people participate in the sector, and the size of the
10
sector in the overall economy (Abdulhafidh, 2013). In this section we will see the various
channels in which agricultural productivity influence poverty reduction.
History shows that different rates of poverty reduction over the past 40 years have been
closely related to differences in agricultural performance particularly the rate of growth of
agricultural growth (Abare, 2001). In simple terms, this means that these are the countries that
have managed to increase their agricultural productivity that have managed to reduce their
poverty rates. According to that, agriculture remains the economic heart of most developing
and developed countries. Moreover agricultural growth is a catalyst for broad-based economic
growth and development in most low income countries (Pinstrup et al., 1995).The
development economics literature (Xavier et al. 2004; Schneider and Gugerty, 2011), shows
that an increase in agricultural productivity reduces poverty by raising farm and off farm
incomes, improving health and nutrition, lowering food prices and broadening wage
employment opportunities. For more understanding we have discussed each of these
productivity effects on poverty as follow.
Agricultural productivity growth can drive rural growth and catalyze a pro-poor development
process (Thirtle et al. 2001). In theory, increasing agricultural production (output) increases
incomes for poor farmers who then increase demand for the goods and services produced by
the non-farming rural poor. Higher agricultural output thus stimulates employment in the rural
and urban non-farm sectors through both forward and backward linkages. This in turn
decreases urban poverty by slowing migration to urban areas and lowering food prices
(Mellor, 1999). Thus, agricultural growth benefits poor farmers and landless laborers by
increasing both production and employment, benefitting both the urban and rural poor through
growth in the rural non-farm economy (Thirtle, Lin, and Piesse 2003). The full general
equilibrium effects of this growth thus take place through the farm, rural non-farm and
national economies.
11
Price effects
Agricultural productivity determines the price of food, which then determines wage costs and the
competitiveness of tradable goods leading to a confluence of effects that determine the real
income effects of increased output for farming households (World Bank, 2007). Increased
agricultural output can change the relative prices of agricultural outputs in relation to substitute or
complimentary products, as well as the costs of inputs to production. If increased output drives
down product prices or the costs of production rise due to increased demand, than increased
agricultural output may not translate into higher real farm income (Irz et al. 2001). Output growth
may not increase farm household incomes if the price effects counteract the production gain;
however food price effects depend upon the tradability of the food.
On-farm employment
Growth in agricultural productivity can increase real wage rates, which both directly and
indirectly contributes to poverty alleviation (Datt and Ravallion, 1998). Increased agricultural
production is likely to increase the demand for farm labor through increases in area cultivated,
intensity of cultivation and frequency of cropping. The impact of farm labor opportunities on
poverty reduction depends on the extent to which the rural poor depend on farm laboring for
their livelihood. Technology also influences the scale of the change in labor demand. Some
technologies increase labor productivity and decrease input requirements, while others allow
for the expansion of cultivated area or multiple cropping per season.
Nutrition linkages
Timmer (1995) suggests that increased food production and farming incomes may allow for
better nutrition among rural labourers. Declines in food prices translate to an increase in real
income for net food buyer households and increase the resources available for consumption.
These resources may be used to increase consumption of staple foods or more diverse,
nutritionally rich foods (Hazell and Haddad, 2001). Enhanced nutritional status can contribute
to further increases in labor productivity in both current and future generations.
12
Multipliers in the rural non-farm economy
Increased agricultural production creates demand for products and services both upstream
(inputs, services for agriculture) and downstream (processing, storage, transport). It also
generates consumption links as farmers and farm laborers spend increased incomes on goods
and services. The degree of these multiplier effects depend on several factors including the
extent of rural infrastructure, population density, the extent of immediate processing needs for
agricultural products, the nature of technological change in farming and the tradability of
goods and services both produced and demanded by agricultural communities (Irz et al.,
2001). The overall impact of increasing agricultural productivity on poverty reduction is
shown in the chart below.
Increase agricultural
productivity
Increase farm
household real income
Increase output
Decrease
Improve
nutritional and Increase non-
health status Increase farming household
Increase
of the society hours of real income
real wage
work
13
At the macro-economic level, growth in agriculture has been consistently shown to be more
beneficial to the poor than growth in other sectors. Furthermore, analysis reveals that
increasing agricultural productivity has probably been the single most important factor in
determining the speed and extent of poverty reduction during the past 40 years (Anne, 2004).
All of these theories beg the question: how can we augment agricultural productivity?
Investigating its main determinant is one way for answering this question, because knowing
their effect on agricultural productivity is strictly important for policy formulation and
implementation.
Over the last two decades, it has become widely accepted in both academic and policy
researches that rural off-farm income make up a significant component of rural livelihoods in
developing countries (Cornilius, 2004). Farming as a primary source of income has failed to
guarantee sufficient livelihood for most farming households. In the context of low
agricultural productivity and missing credit and insurance markets, which characterizes most
rural economies of developing countries, diversifying sources of income into off-farm
activities is one of the ways by which households may overcome some of their credit and
insurance market constraints (Oseni and Winter, 2009). The share of off-farm income is
expected to increase substantially in the coming years, especially in sub-Saharan Africa where
increasing population growth and limited agricultural resources are threatening the growth of
the agricultural sector (Haggblade et al. 2007). In sub-Saharan Africa, a range of 30–50 %
reliance on non-farm income sources is common (Frank, 1999).
Theoretically, the effect of off-farm income on agricultural production within the same
household could be positive, negative or nil, depending on the household’s degree of
integration with factor or product markets (Lopez-Feldman et al., 2007). For instance, it could
be positive when off-farm incomes are spent on financing farm investment so that the positive
investment effect outweighs the negative effect of removing family labour from agriculture. A
positive effect of off-farm income on agricultural production could imply that stimulating the
rural off-farm sector will enhance growth of the farm sector. On the other hand, it could be
14
negative, if the income earned off the farm is not spent on agricultural production, but rather,
on increasing consumption, financing investment in non-agricultural activities, or migration
out of agriculture entirely (Pfeiffer et al., 2009). The effect could also be nil when the positive
effect through agricultural financing just equals the negative effect of family labour loss from
agriculture.
The New Economics of Labor Migration postulates that increased non-farm income may relax
rural households’ financial constraints and increase investment in new farming technologies
(Stark and Bloom 1985). Remittances from family members working in cities allow rural
residents to purchase high-quality seeds and fertilizers to boost crop yields; agricultural
productivity also increases if migrant remittances are invested in technology and
infrastructure. However, migrant remittances do not necessarily induce productive
investment, especially in the absence of profitable investment opportunities.
The agricultural investment effect of off-farm income is particularly important for poor farm
households. This is because lack of liquidity and poor access to credit are the most pressing
constraints to improved agricultural productivity among farm households in developing
countries (Deininger et al., 2007; Haggblade et al., 2007). Apart from providing flows of cash
income that can be used to purchase farm inputs and hire labour for agricultural production,
evidence of a steady off-farm income has been used as collateral for agricultural loans, given
the inadequacy of land, in certain settings (Collier and Lal, 1986; Hoffman and Heidhues,
1993, Hert, 2009).
The existing literature alludes to two main potential direct effects of income from off-farm
income (Babatunde and Qaim, 2015): the liquidity-relaxing effect, which supposes a potential
increase in farm expenditure/investment, and the lost-labour effect, which supposes a
potential allocation of labour away from the farm. Authors such as Ruben and van den Berg
(2001), Ellis and Freeman (2004), and Stampini and Davis (2009) point to the liquidity-
relaxing effect of income from off-farm work in different studies. These studies highlight the
positive spillover effects of engaging in off-farm work on farm input among producers.
15
Ellis and Freeman (2004) identified positive effects of off-farm income on land productivity,
the hiring in of labour, and the acquisition of farm inputs. Furthermore, Oseni and Winters
(2009) observed a greater use of hired labour and inorganic fertilizers among Nigerian
farmers engaged in off-farm work. Similarly, Anriquez and Daidone (2010) found off-farm
work to enhance investment in farm inputs among farmers in rural Ghana, while Maertens
(2009) found off-farm employment to increase fertilizer use and cultivated areas in a study of
Senegalese farmers. Pfeiffer et al. (2009), on the other hand, found non-farm income to have a
positive impact on the demand for farm inputs but a negative impact on output and use of
household labour in production. An inverse relationship between the diversification of
household income through working off-farm and farm investment is also reported by Ahituv
and Kimhi (2002) and Davis et al. (2009).
Despite the complex nature of the off-farm – farm linkages that make it very difficult to have
a priori expectation of the net effect of off-farm income on agricultural outcomes, it is
hypothesized that off-farm income contributes positively to better agricultural production in
terms of larger farm output, crop input expenses and technical efficiency. Changes in
agricultural production and input use by farm households that receive off-farm income are
likely to represent an important part of the overall effect of off-farm income in economies
where agriculture remains an important source of livelihood for many households. Therefore
due to its great impact on boosting agricultural productivity, to increase the amount of off-
farm rural household income is compulsory. But in Ethiopia off-farm rural income itself is
restricted by many factors.
16
It is believed that chemical fertilizer, if soil organic matter is not depleted, is an ingredient
that feed roots with sufficient nutrients in the nutrient poor soils; and enables to adopt high
yield verity and thereby increase agricultural production. In a certain country, chemical
fertilizer adoption can be determined by economic, social, physical and technical aspects of
farming (Abay and Assefa, 2004); and these aspects influence the type of crops to be grown
and the production method to be used (Sassenrath et al., 2012).
There are also views that the dependency on chemical fertilizers only for agricultural
production might not be sustainable as it results in the depletion of organic soil contents
thereby reducing the potential benefit of fertilizer utilization (Ghosh, 2004). Most of the time,
the application of chemical fertilizers is not based on soil tests which leads to the utilization of
fertilizer either above or below their requirements (Ogoke et al. 2009). Excessive use of
fertilizer and intensive production system could result in fertilizer contamination; and thereby
environment and production potential of the land and soil would further be harmed.
In combination with chemical fertilizers, improved varieties of seeds are critical agricultural
inputs that help farmers to obtain improved agricultural yields. However, Formal sector
supplied improved seeds should fulfill certain quality standards set by the national regulations
(Bishaw et al. 2012). Seeds that fulfill the quality requirements have a positive impact on the
productivity of land.
Irrigation and irrigation technologies are important ingredients for accelerating agricultural
production. It can be taken as an input to stabilize crop yield and patterns; it is an asset that
can be exploited when rainfall is insufficient; it also enhances cropping intensity by letting to
produce twice or more per year; and paves the way to use high yield variety (Datar and Del
Carpio, 2009). Access to good irrigation allows the poor to increase production, gives them
opportunities to diversify their income base and reduce their vulnerability to the seasonality of
agricultural production and external shocks (Hussain, and Hanjra, 2004). Though it is
pertinent for production, its adoption and expansion can be influenced by extension services,
education, water price, cost of irrigation equipments and farmland size (Genius et al., 2013).
However, irrigation can also trigger socio-economic upheavals when it causes problems such
17
as disease, land degradation, water pollution and destruction of living beings and natural
ecosystems (Hussain, &Hanjra, 2004).
Food aid
There are two major arguments in the debate over food aid impacts on agricultural
productivity, those made by Schultz in the 1960s and Barrett in recent years. Schultz
maintained that food aid discourages domestic production in recipient countries, whereas
Barrett claimed that, rather than affecting production, food aid displaces imports to recipient
countries (Pane, 2006). At the surface level the relationship between food aid and agricultural
production and household supply to agriculture is assumed to be negative. As per capita food
production has declined in conjunction with increasing amounts of food aid deliveries, the
assumption suggests a causal relationship between food aid and the decline in production.
Sofia, (2010) sum up this hypothesis as “The logic behind this assertion is that food aid
deliveries increase supply faster than they stimulate demand, depressing the food prices
received by recipient country producers and traders, and thereby creating disincentives for
producers to invest in improved technologies or for marking intermediaries to invest in
storage and transport capacity”.
For the greater part of the last three decades, food aid has helped to sustain large segments of
the Ethiopian population. Although food aid is intended to serve as a temporary mechanism to
provide populations in need of sustenance, the nature and necessity of food aid has shifted
from transitory to chronic in countries like Ethiopia, which currently receives more food aid
than any other country in the world (Sofia, 2010). Households receiving food aid were
associated with a small decrease in the number of days supplied to agricultural preparation, as
well as a decrease in the value of production from the “Belg” season.
18
Agricultural credit
The household characteristics consist of many variables that affect the agricultural production
of farm operators. Some of these variables are: education level, gender, age, family size,
landholding size and possession of oxen.
Gender refers to socially constructed roles and relationships of women and men in a given
culture or location (Adeoti et al. 2012). In enhancing agricultural production and income, the
full participation of men and women is very important. Gender is one of the significant
determinants of agricultural production since male-headed and female-headed households
(HH) could not have the same capability and endurance in enhancing agricultural production;
19
where the former are stronger (Nyanga et al., 2012). In Kenya, Ekbom et al. (2012) found that
female-headed households are inefficient and unproductive compared with their counter parts.
Historically, for thousands of years, oxen have been recognized as the first draft animals to
serve human beings, to cultivate land and pull heavy loads (Bryant, 2010).The possession of
oxen determines the farming ability of farm operators because if farmers do not have oxen
they would be obliged to rent out their land to other farmers (Holden et al., 2004). In this
case, farmers would enter into sharecropping. This further diminishes the production and
income of the household as the yield is shared with oxen owners. There are advantages
associated with owning oxen. Oxen owners can cultivate and sow their land at the right time.
This has a positive impact on the productivity of land. In addition, oxen could also be rented
out on a daily payment basis to till the land for other households. Therefore, they may serve as
a source of additional income for the owners.
20
2.1.7. Demographic and environmental factors vis-a-vis agricultural productivity
Population density
Boserup’s (1965)hypothesize that increasing population density leads to increased input use
per unit of land, and increased production per unit of land as farmers move successively from
long fallow to short fallow, to annual cropping, and finally to multiple cropping cycles per
year. Contrary to this, Binswanger and McIntire (1987), argued that increases in rural
population density should induce a number of challenges on tropical agricultural farming
systems, including declining labor productivity, decreased fallows, increased landlessness, the
development of land, labor and informal financial markets, and declining livestock tenancy.
As rural communities become more heavily populated, farmers move from shifting cultivation
to annual cropping of the same plots. Fallows are reduced and more labor time is devoted to
each unit of land produced. We would also expect that agricultural wage rates will decline in
areas of high population density, as the number of workers increases relative to the amount of
land, as predicted by the induced innovation hypothesis.
Soil erosion
Climate can also affect agriculture in a variety of ways. Temperature, radiation, rainfall, soil
moisture and carbon dioxide (CO2) concentration are all important variables to determine
agricultural productivity, and their relationships are not simply linear. Climate change
21
manifests through increasing average temperature and changing rainfall regime and it has
many negative effects, both globally and regionally. Drought is the most important climatic
factor influencing crop growth and yield. It is a major factor in agricultural productivity
especially in developing regions. Recent events in East Africa are testament to this.
In computing the income of the household from the different types of off-farm income, the
determinants of off-farm participation variables are applied. Because the variables that
influence the probability of participation in off-farm employment are determining the income
from that employment (Babatunde, & Qaim, 2009). As many literature explains age,
education, family size, land holding size, information and distance to the town, land
ownership security, land fertility, value of farm output, access to formal credit and access to
electricity affect rural household off farm participation and thus off-farm income.
As the age of the household increases the farmer acquires more knowledge and experiences
with possible positive impact on off-farm income. At a younger age the probability of
working off farm will increase. At older ages the overall labor hours will diminish and the
demand for leisure will increase. As a net result, a humped-shaped life cycle profile will
appear.
The education level of household’s head is also of the determinant factors of off-farm income.
If the supply of labour in the labour market is in excess of demand, an educated individual,
specifically at university level, has a better chance of getting higher wage in off-farm
activities compared to an individual without a university education (Bhaumik et al., 2011). In
demand constrained labour markets, education is used as a screening device and is the most
determining factor for high-paying and skilled employment.
22
Among the determinants of income portfolio choices, gender is recognized to be crucial.
Evidence shows that incentives for economic diversification may vary between men and
women both in the same and different households (Abdulai, 2001). Not surprisingly, in
households with a higher proportion of women the returns to off-farm activities are lower than
in households with a higher share of adult men. Female sole parents, for example, are likely to
have the lowest family income and have also experienced a decline in employment
(Johnstone, nd). This disparity may be due to constraints in access to productive assets,
biases in the labor markets, differences in preferences and attitudes towards risk, or social
norms governing gender roles in the household production. Women and men are distinct
comparative advantages and inequalities in the labor market, with women usually facing
barriers to entry into certain sectors and lower wages (Sara, 2009).
What is clear is that family size is an important factor when looking at family income. In
terms of income levels, some families are more disadvantaged than others and are therefore
more vulnerable to social exclusion. Families with higher dependent members have
constraints, typically discouraging the mother's labour force participation and hence inhibiting
family income-generation ((Johnstone, nd).It is these families with poor employment patterns
who also tend to have the largest number of children. However families with large member of
young children are expected to have higher off farm income level.
Some studies have established the relationships between the households’ landholding size and
their participation in off-farm activities. Hazell & Haggblade (1993) reported that the share of
non-farm income to total household income related to the corresponding land holding of farm
operators. They added that landless and near landless households in all the countries
considered in the study depended on non-farm income. Therefore, regions with small land
sizes have more people in off-farm employment because the farm income is not sufficient to
support a household (Chaplin et al., 2003). In addition to the size of the landholding, the
quality of the land is a determinant factor for farmers to participate in off-farm activities. A
study conducted in China by Guoqiang & Wenting (2013) found that the off-farm
23
participation of farmers residing in areas endowed with better natural resources was lower
than farmers in areas with poor natural resources.
In developing countries, lack of money is the most critical problem for farmers (Kwon et al.
2006). Off-farm income makes farmers more able to fulfill the monetary requirements for
new technology. In the absence of credit arrangements, farmers are forced to use a large
proportion of their income to purchase inputs such as fertilizers and pesticides (Pfeiffer et al.,
2009).
2.1.9. Theoretical model for analyzing the determinants of agricultural productivity and
its measurement
Empirical research has generally used the Cobb-Douglas production function to measure the
relationship between inputs and output, marginal products, and production elasticities.
Measuring agriculture productivity can clearly shows the level of incomes of the rural
household, those who are engaged in agricultural activity. Agricultural productivity most
commonly estimated using parametric and the non-parametric approach. In the parametric
approach, the coefficients of the production function are estimated statistically using
econometric approach whereas, in the non-parametric approach by using the mathematical
programming approach. It is the parametric approach commonly used in the estimation of
24
production functions while the non-parametric approach used in efficiency analysis (Coelli et
al., 1998).
This study has considered the parametric approach to estimate the agricultural productivity
function. Because the econometric approach has the advantage of being statistical, hence
permitting hypothesis testing and calculation of confidence intervals to test the reliability of
the model estimated. This approach explicitly measures the marginal contribution of each
category of inputs to aggregate agricultural output.
For our analysis of agricultural productivity, we will use the general neoclassical production
function, Cobb-Douglas production function, found to be theoretically and empirically
reasonable, since it is easy to estimate and mathematically manipulate and possible to test the
significance of the estimated elasticities using standard test statistics such as t–ratios and f-
ratios. Cobb-Douglas production function is given by the equation:
Where Y is the output level, Xs are the inputs; A, α & β are positive constants; K & L are
capital and labour input respectively. A is the total factor productivity, α & β are capital and
labour elasticities respectively.
Many empirical researches have done in relation to agricultural productivity, off-farm income
and their linkage with poverty reduction in developing countries across the world.
Abdelhafidh (2013), on his simultaneous equation model uses a panel data of 32 Sub Saharan
Africa (SSA) countries, have shown the interrelationship between agricultural productivity,
poverty and technological innovation. His finding indicates that agricultural productivity
25
growth would lead to a 32% decrease in poverty in SSA. This implies that for SSA countries
accelerating growth in agriculture is fundamental to reduce poverty and allow countries to
achieve economic transformation. This passes through the ability of agriculture to generate
employment, to stimulate the economy through linkages, and to reduce the real cost of food
accounts.
A number of studies have sought to explain the factors leading to increase or decrease in
agricultural productivity and off-farm rural households, particularly in developing countries.
Reuben (2014) using generalized method of moments (GMM) method, identifies the impact
of foreign agricultural aid on agricultural GDP and productivity in sub-Saharan Africa by
applying secondary data from 47 SSA countries spanning from 2002-2010. The finding
reveals that foreign agricultural aid has a positive and significant impact on agricultural GDP
and agricultural productivity at 10% significance. The study also reveals that bilateral foreign
agricultural aid influences agricultural productivity more than multilateral foreign agricultural
aid and that multilateral foreign agricultural aid influences agricultural GDP more than
bilateral foreign agricultural aid.
A study by George (2017) in Kenya, by employing Cobb Douglas production function and
Ordinary List Square estimation technique as the method of analysis and using secondary
26
data from the period 1980 – 2013; found that increase of one percent in labour force caused
an increase in agricultural productivity by 0.1984402%. From this result we can understand
labour force is directly related with agricultural productivity.
Hannan (2008), tries to investigate the relationship between farm size and agricultural
productivity for Ukraine farmers. This study used a parametric (econometric) and operational
approach to estimate the farm productive efficiency. The study was conducted using the data
for 1170 Ukraine farmers during the five year (2001-2005). The finding of the study implies
the relationship between land size and farm productivity is non linear. This means that up to
some point productivity increases (for farms with total land size up to 120 ha) and then
decreases. A similar study by Fidele (1995) also shows a similar inverse relationship between
farm size and productivity in Uganda.
A research conducted by Ighodaro et al. (2013) analyses the impact of soil erosion on
agricultural potential and performance in the eastern cape of South Africa by using cross
sectional data. The findings revealed that there exists a high negative correlation between soil
erosion and agricultural productivity in the study area. It recommended that soil control
mechanism should be pushed to the farmers by extension services to curb the negative impact
of soil erosion on productivity.
27
investment will depend, mutually, on the ability of rural wage employment activities to
generate surpluses and the extent to which there is unemployed labour in the household.
Another researcher, Babatunde (2015), by using 2SLS analyzed the effect of off-farm income
on agricultural production, using farm households survey data collected from 40 villages in
Kwara State of Nigeria. It was found that off-farm income is important for the vast majority
of the households: almost 90% of the sampled households have at least some off-farm income
and on average it accounts for about 50% of total household income. His result of the
Instrumental Variable estimation suggest that off-farm income contributes to higher farm
production and larger expenses on purchased inputs, while it decreases the use of family
labour.
Using a panel of village data from China, Wang et al. (2011), have examined the extent to
which non-farm income influence agricultural productivity in rural China. Their finding
shows that nonfarm revenue has a significant positive effect on agricultural land productivity.
Although non-farm activities do withdraw labor out of agriculture and therefore dampen land
productivity, that negative effect is negligible in comparison with the land productivity
improvement brought by nonfarm revenue-financed infrastructure capital investment.
According to their study the growth in non-farm income has induced investment in
infrastructure capital through collective provision, but has not increased private capital such
as machinery and livestock power. Although their study is based on a panel of Chinese
villages, the results apply to other developing countries.
Lhing et al. (2013), by using Cobb-Douglas functional form and Logistic regression model
found the most common influencing factor on household off-farm income in Myanmar.
Among these factors educational level, age of the head of household, gender of the head of
household, educational level of the head of household, land holding size, number of crops,
and established new enterprise have a significant influence on household income. Another
study by Egyei and Adzovor (2013), on their analysis of household non farm income in
Ghana, found that availability of telecommunication infrastructure, availability of banks,
28
availability of motorable roads, forest and savannah zones are significant factors in explaining
non-farm income.
2.2.2. Agricultural productivity and off farm rural household income in Ethiopia
By using cross sectional data, a study conducted in northern Tigray for assessing the factors
affecting agricultural production, shows that, agricultural production in the study area is
determined by household characteristics, physical environment, agricultural technologies,
institutional factors, off-farm participation and agricultural marketing issues (Birhanu, 2014).
A researcher used a Probit regression analysis to arrive in this conclusion. He also revealed
off- farm participation is determined by gender, age, education, family size and tropical
livestock unit, presence of pack animals, location dummies and amount of credit taken. From
these all, age, tropical livestock unit and location dummies carried a negative coefficient
indicating their negative implication on farm household’s probability of participation and the
resultant effect on agricultural production while farm income is significantly influenced by
age, family size, land size, plot distance, plot slope, fertilizer use, row spacing, access to
credit, membership to a certain association and Tropical livestock unit.
In the same region Bihon (2015), by using farm income model, landholding size , possession
of oxen, amount of fertilizer, improved seeds, irrigation, soil quality, village distance to the
district market, average distance of plots from the homestead and crop rotation were
determinant variables for agricultural productivity. As the researcher revealed, the
determinant variables of off-farm participation were: irrigation, age, amount of money
29
borrowed, village distance to the “woreda” market, fear of land confiscation and access to
electricity.
Josephson and Anna (2013), uses household-level panel data to estimate how population
density impacts agricultural intensification and farm income in Ethiopia. They found that
increases in population density lead to lower farm sizes, which have major implications for
agricultural intensification and household well-being; cause farmers to purchase more
inorganic fertilizer per hectare. This is due to population density’s direct effects on market
access. Population density is a significant factor that influences input demand, but not output
supply is the main conclusion of the study.
A study by Elias et al. (2015), on their analysis of farmer’s satisfaction with agricultural
extension services and its influencing factors, conducted in North West Ethiopia based on
ordered logit model. They found that perceived economic return, regular extension contact,
family size and off-farm income were driving factors for farmers’ satisfaction. On the other
hand, limited technology choices, high price of inputs, inconvenient loan system and
undefined boundary between the extension services and the local politics were among the
reasons for farmer’s dissatisfaction.
Another research done by Tesfaye et al. (2015), investigates how off-farm income affects
crop output market participation decision and marketed surplus of smallholder farmers in
Ethiopia. They used Double-Hurdle model as estimation technique and use three waves of
panel data from Ethiopian Rural Household Survey (ERHS) for their study. Results showed
that off-farm income has no significant influence on household output market participation.
But conditional on positive market participation, each additional earning from off-farm work
has negative and statistically significant effect on marketed surplus. This indicates farmers use
earnings from off-farm source rather for consumption purpose than as a source of liquidity to
invest in agricultural production and increase marketable surplus.
Research in the agricultural sector has and continues to be carried out. This can be attributed
to the significant role agricultural sector plays in the economy, especially in the developing
30
economies. Since agricultural sector keep on a very significant sector of the Ethiopian
economy, there is need for vigorous and extensive research so as to provide updated data to
enable the relevant authorities to formulate policies and programmes which are up to date and
relevant to the current trends. This study is; therefore, serve the purpose of expanding the
body of literature available to enable policy makers to formulate relevant policies.
31
3. RESEARCH METHODOLOGY
For achieving our objectives, in this study we have used quantitative secondary panel data.
Three wave Ethiopian Rural Socioeconomic Survey (ERSS) panel data, which covers the
period of 2011/12, 2013/14 and 2015/16 have used. The data is collected by the CSA of
Ethiopia in collaboration with World Bank Living Standards Measurement Study (LSMS)
team as part of the integrated surveys on agriculture program.
The survey is conducted in 2011/12 for the first time in full sample coverage at National level
and after that data collection has been conducting in the gap of each two years. This data
compiles a set of basic statistics such as household’s demographic characteristics, educational
and health status of the household, household assets and housing characteristics, data on
agricultural production, nonfarm enterprises, other income and assistances, households
Consumption, Expenditure, Food Security, Shocks and coping mechanisms.
For collecting the data CSA uses the list of Enumeration Areas (EA) of rural Ethiopia and its
respective agricultural households obtained from the 2007 Population and Housing Census.
On the Survey the data collector employed a stratified, two-stage probability sampling
techniques to select its sample where the regions of the country serve as strata. The first stage
of sampling entailed, selecting enumeration areas using simple random sampling (SRS). From
the rural area a sample of 290 EAs were selected.
The second stage of sampling was the selection of households to be interviewed in each EA.
From 290 rural sampled EAs, a total of 10 households are randomly selected in each EA.
Based on this procedure, at the rural country level the totals of 2,900 sample households are
selected. However, the number of actual interviewed households is reduced from year to year
across the three panel periods. As a consequence, we have used a balanced panel data of 1,915
32
sampled households for our analysis. The regional distribution of these households is
presented in appendix IV.
We have employed quantitative methods to analyze the data. In order to better understand the
variation of agricultural production and off-farm rural household income portfolio across
years and different areas of the country a descriptive analysis have performed. The study used
one way ANOVA and Bonferroni tests to show the variations of productivity and off- farm
rural incomes across time and regions of the country. The univariate approach, which is very
popular in econometric analysis of panel data (Carlos, 2014) have used for explaining the
determinants of agricultural productivity and off-farm rural household income independently.
For analyzing the determinants of agricultural productivity and off-farm rural household
income, fixed effect instrumental variable and random effect instrumental variable regression
models are used and the best fitted model is chosen by using Hausman identification test. To
execute all the descriptive and econometric models in our study, we have used Stata software
version 14.
The econometric models applied in this research are based on the scientific requirements of
the variables (dependent and explanatory) that are considered and models used by other
researchers with similar topics.
The study have used the general Neoclassical Cobb Douglas production function model to
show the joint impact of independent variables which has explained in the literature review
part to agricultural productivity. The logical basis for choosing Cobb-Douglas production
function is based on the fact that it is relatively simple and convenient to specify and interpret.
Moreover, application of Cobb- Douglas production function has been found applicable in
33
similar studies to this one (Thabit 2015, Tessema 2015, George 2017). Most of the studies
using the Cobb Douglas production function approach stated that the functional form of the
Cobb-Douglas production model is assume homogeneity, unitary elasticity of substitution
between input and output. And also it is among the best well known production function
utilized in applied production and productivity analysis.
By considering all this, the model we employee have specified and start with output supply
equation. The output supply equation includes inputs, farm characteristics and household
characteristics. The current analysis have considered all the factors of production such as use
of quality inputs, chemical fertilizer, cultivated area of land, number of oxen etc. as proxy for
capital inputs. Therefore, the econometric model of production function is constructed as:
Yit= ( β1 β2
it Kit ) euit…………………………….……………………………… (1)
Where: Yit= is the amount of all farm outputs produced by the ith household’s in
To eliminate the bias in Cobb-Douglas production function, reduce the influence of outliers
and be able to interpret coefficients as elasticities, we transform equation 1 by taking the
logarithms of both sides. Comparing the transcendental logarithmic function (trans-log) and
Cobb-Douglas production function, the former is relatively more flexible, thus it is more
appropriate especially when estimating a production relationship which is not well
understood. So, it gives us:
34
Therefore, in the case of our several independent variables the log-linear model would be:
Where,
lnYit = the log of total annual farm output produced by ith household during period t
lnLHSit = the log of land area holding size of ith household during period t
lnOPOit = the log of number of oxen owned by ith household in each period
EDUit = Educational level of the ith household head during the period t
FAIDit = total amount of food aid received by ith household during period t
IRGit = shows whether the ith household uses irrigation system or not during period t
DRit = if drought was happened to the ith household farms during the period t
SOEit = if the land of the ith household is exposed to Soil erosion at period t
UESit = if the household participate in extension service program during the period t
35
αoit is a constant term; α and β are parameters to be estimated.
uit represents the time-varying unobserved factors (idiosyncratic errors) that affect
Agricultural productivity.
Following, Verbeek (2000), Greene (2003), Baltagi (2005), Gujarati and Porter (2009) and
Wooldridge (2010), we have presented the panel data regression model as:
Where,
Yit = is a scalar dependent variable
Xit= is a K × 1 vector of independent variables
it= is the unobserved individual heterogeneity or the individual fixed effect
β= are the parameters to be estimated
uit= is a scalar disturbance term, i indexes individual, firm or country in a cross section, while
t indexes time.
36
By expecting the existence of endogeneity problem between the variable agricultural
productivity and off-farm household income in our model, we have used fixed effects
instrumental variable and random effects instrumental variable estimation techniques and
applied Hausman identification test to choose the best fitted model among them. Accordingly,
Hausman identification test selected random effect instrumental variable estimation
techniques in favor of fixed effect. Due to this random effect instrumental variable estimation
technique is used for our agricultural productivity model.
Where, E (μit/zi, ci) = 0, ci is unobserved effect and μit are the idiosyncratic errors.
In order to allow for correlation between the regressors and the idiosyncratic errors, we
assume the existence of instruments, zit, which are strictly exogenous conditional on ci. This
permits for unspecified correlation between zit and ci, but requires zit to be uncorrelated with
uit. In Fixed effects instrumental variable estimator we deviate variables from time averages to
remove ci and apply instrumental variable as it is given in equation (6). Since the fixed effect
estimator involves time-demeaning, we assume that all variables in xit and zit are time-
varying.
žit = zit – źi
The key condition for fixed effect IV to be consistent is that the instruments (zit) are strictly
exogenous with respect to μit. With T ≥ 3 time periods, this is easily tested as in the usual
fixed effect case. On the other hand, random effects instrumental variable assumes ci is
uncorrelated with zi, and normally imposes serial independence on μit.
37
The fixed effect model has the advantage of yielding unbiased estimates of β, but the
estimates can be subject to high variability. The random effect model on the other hand
usually introduces bias in estimates of β, but can significantly reduce the variance of those
estimates (Gelman& Hill 2007). Hence, there is a trade-off between bias and variance while
choosing between the two models (Wooldridge 2010; Gelman& Hill 2007). Hausman (1978)
proposed a method for testing this assumption based on the difference between the random-
effect and fixed-effect estimates (Wooldridge 2010). We have used this test to decide whether
FE-IV or RE-IV is the preferred specification for our data. The FE-IV is inefficient under the
null but consistent under the alternative; while, the RE-IV being efficient under the null, but
inconsistent under the alternative. Rejection by the Hausman test suggests that αi is correlated
with the covariates and thus estimation of β should be based on the FE-IV estimator.
Dependent variable
The dependent variable in our model is the total annual agricultural output. It is the amount of
the ith rural household total farm output measured in kilogram. All quantities of the output are
converted to a standard measurement, kilogram, on the basis of conversion factors calculated
by the Ethiopian CSA for different traditional measurement units used in different areas of the
country.
Conventional inputs for agricultural production which are listed in the literature review are
used as the determinants of agricultural productivity (explanatory variables) in our model.
Among the number of factors, which have been related to productivity, in this study,
demographic and socioeconomic factors are hypothesized to explain the dependent variable.
1. Off-farm income: It is the sum of rural nonfarm income and wage earnings in agriculture.
This is measured by aggregating all the incomes obtained from nonfarm activity and
38
employment wages gained by each household from agriculture in Ethiopian Birr. Off-farm
income directly affects agricultural productivity through increasing the potential of farm
expenditure, relaxing rural households’ financial constraints and increase investment in new
farming technologies (Babatunde and Qaim, 2015). As a result we expected off farm income
has a positive effect on productivity.
2. Land holding size: this implies the sum of owned cultivated land, rented-in land and land
secured through sharecropping arrangements during the survey period by the household. It is
a continuous variable measured in hectare. Those who have large farm size can expand
production by exploiting economies of scale; higher input usage and tend to reject the
traditional broadcasting method by adopting row-planting method which is pertinent for
increasing productivity (Endrias et al. 2013). Therefore, it is hypothesized that the larger the
land holding size of the farm household, the higher the volume of productivity will be.
4. Agricultural labour input: It refers to the total number of family members of the
household who have directly involved on the agricultural production process. The more the
labor force utilized for the farm production process the more farm land preparation will be
made. Therefore, we hypothesize that, agricultural labour have a positive impact on
agricultural productivity.
5. Land dependency ratio: it is a measure of the intensity of land use, expressed as the ratio
of total area of land holding size per family size in a given household. This variable is taken
as a proxy for showing the impact of population density on the amount of agricultural output
at the household level. The larger the land dependency ratio induce a number of challenges on
tropical agricultural farming systems, including declining labor productivity, decreased
39
fallows, increased landlessness, and decline in livestock tenancy. Therefore we have expected
its negative impact on agricultural productivity.
7. Gender of household head: It is a dummy variable, “1” if gender of the household head is
male and “0” otherwise. Male-headed households are physically strong and capable than
female headed households and then the former would have better opportunities for enhancing
their farm productivity. In addition to existing biological gender differences; male headed
households have mobility, participate in different meetings and have more exposure to
information about better farm inputs and practice. Therefore it is hypothesized that male
headed households have more productive to produce farm output.
8. Food aid: refers to the provision of food or cash to purchase food either by foreign
government or by charitable organizations for households in need. It is a dummy variable
which takes “1” if the household receive food aid and “0” otherwise. Households receiving
food aid were associated with a small decrease in the number of days supplied to agricultural
preparation, as well as a decrease in the value of production from the “Belg” season in
Ethiopia (Sofia, 2010). Therefore, the relationship between food aid and agricultural
productivity and household supply to agriculture is expected to be negative.
9. Technologies: The widely used agricultural technologies by farm operators in Ethiopia are
irrigation, fertilizer and improved seeds, which serve as a proxy variable for measuring the
impact of technology in agricultural productivity in our model. A dummy variable, irrigation,
takes “1” if the household used irrigation system during the survey period and “0” for those
households who do not use irrigation at all. The variable fertilizer is also a dummy which
takes “1” for households who have used fertilizer and “0” for non fertilizer user. Similarly;
40
improved seeds will take a dummy variable which is“1” for households using improved seed
and “0” otherwise during all the survey periods. It is hypothesized that agricultural
productivity is positively influenced by the application of each of these technologies.
10. Drought: it refers to a period of abnormally low rainfall which causes an extensive
damage on crops and prevents them from giving the expected yield. It takes dummy variable
“1”if there is the households face a drought during the last 12 months prior to the survey
period and “0” otherwise. We proposed a negative relationship between drought and
agricultural productivity.
11. Soil erosion: refers to the removal of topsoil faster than the soil forming processes can
replace it, due to natural, animal and human activities. It is a dummy variable which takes “1”
if the agricultural land of the household is exposed to soil erosion and “0” for those
households whose crop land is not exposed to erosion. The rates of soil erosion that exceed
the generation of new top soil are a dynamic process which leads to decline in the soil
productivity, low agricultural yield and income (Pani and Kumar, 2013). As a result we have
expected soil erosion exerts a negative impact on agricultural productivity.
12. Access to credit: Access to credit is measured in terms of whether at least one member of
the household has received a credit or not during the last 12 months prior to the survey period.
Financial loan from government, banks, and other financial institutions for primary
production, processing and distribution of agricultural products and for the production and
distribution of inputs is important for the development of agriculture (Aggelopoulos et al.
2011). Hence, we expected the positive relations between credit access and agricultural
productivity.
13. Use of extension services: Extension service is a dummy variable representing “1” if
households are visited by extension worker and “0” otherwise. Those households who have
access for extension service during their farm production process would expect to increase the
probability of farm productivity. Therefore, it is hypothesized that the use of extension
services positively impacts agricultural productivity.
41
3.4.4. Empirical model specification for off-farm rural household income
The second model we have employed for examining the determinants of off-farm rural
household income is also a panel data model. If OFIit is rural off-farm income for household i
at time t, we can define OFIit as a function of explanatory variables, identified in the literature
as the major determinants of off-farm income. Thus, the model that deals with the
determinants of off-farm income can be written as:
In panel data models, the unobserved heterogeneity (αi) is called a random effect, if it is
treated as a random variable and a fixed effect if it is treated as a parameter to be estimated for
each individual observation i (Wooldridge 2010). In the fixed-effects (FE) model, the αi in
equation (7) is permitted to be correlated with the regressors Xit. In the random-effects (RE)
model, αi is assumed to be purely random, with zero correlation between the observed
explanatory variables and the unobserved effect that is Cov (Xit, αi) = 0, t = 1, 2……..T. This,
according to Wooldridge (2010) is a relatively stringent assumption and allows for time-
invariant variables to play a role as explanatory variables.
42
3.4.5. Estimation technique
We have applied Instrumental Variable (IV) estimation approach to solve the endogenity
problems in the explanatory variables. According to Cameron & Trivedi (2010), the
individual fixed-effects model gives consistent estimates of the coefficients of the time-
varying parameters under a limited endogenity of the regressors Xit. These regressors may be
correlated with the fixed-effects extraneous variables which are captured by uit in equation (7).
This is why we have used the IV approach. The IV regression provides an improved way of
allowing for Xit to be correlated with εit, under the assumption that there exist variables or
instruments Zit that is correlated with Xit but not with εit.
3.4.6. Hypothesis and definition of variables for off-farm rural household income model
Dependent variable
The dependent variable, Off-farm rural household income, refers to the portion of farm
household’s income which constitutes the sum of rural nonfarm income and wage earnings in
agriculture. It is measured by aggregating all the incomes obtained from nonfarm activity and
employment wages gained by each household from agriculture in Ethiopian currency.
Explanatory variables
In computing the income of the household from the different types of off-farm income, the
determinants of off-farm participation variables are applied. This approach is in line with
similar studies on the participation of off-farm employment. The explanatory variables are
applied again for the off-farm employment income. Because the variables that influence the
probability of participation in off-farm employment are determining the income from that
employment (Babatunde, &Qaim, 2009, Birhanu, 2014, Bihon, 2015).
43
produced higher output from their land tend to participate in off-farm activities as well
because of availability of finance for starting up new off-farm business. Due to this we
hypothesized that agricultural output is positively relate with off-farm income.
2. Land holding size: This is a continuous variable measured in hectare. Those with larger
farm lands participate less frequently in nonfarm activities due to the productive potential of
their large farm size that could restrain off-farm participation and income from it, since they
have an opportunity to diversify their farm activities (Abebe, 2008; Alaba and Kayode, 2011).
Due to this we hypothesized a negative relationship between land holding size and off-farm
income.
5. Number of dependents in the household: This variable refers to those members of farm
households whose age is 5 years or younger and 65 years or above. It is a continuous variable
measured in number. A farm household with a majority of dependants is less likely to
participate in off-farm work and hence less income from off farm activities. Due to this the
expected sign is negative.
44
6. Number of adults in the household: It is a continuous variable measured in numbers and
whenever households have larger number of child aged in between 6-15, may tend to
participate in off-farm activities and their off farm income will increase (Birhanu, 2014). The
expectation is that the more existence of young members in the household the higher the
probability off-farm income the household will get.
7. Tropical livestock unit (TLU): It is the total number of livestock holding of the household
measured in livestock unit. Those who possess a flock of TLU were expected to participate in
off-farm activities much better than those who possess less and their income will be higher
(Shumet, 2011). TLU is expected to serve as a source of a startup capital for off-farm
employment and thereby raise the amount of income from off the farm. Therefore, we
hypothesized livestock holding has a positive impact on off farm income generation.
8. Access to electricity: It is a dummy variable which takes “1” if the household has access
for electricity and “0” otherwise. Access for electricity in the area or a nearby area may
encourage small businesses and other non-agricultural activities in that area (Bihon, 2015). As
a result, it is expected to positively stimulate off-farm participation and increase off-farm
income.
9. Access to credit: This is a dummy variable that takes the value “1” when the household
takes loan and “0” otherwise. Credit helps farm operators to improve the productivity of their
land. It gives them access to farm inputs to benefit more from their land. Farmers engage in
off-farm activities to fulfill the cash requirement of the household when there is no credit
access. Therefore, we hypothesized; this variable is negatively related to off-farm income.
45
4. RESULTS AND DISCUSSION
This chapter have presented and discussed the results of empirical analysis based on the
model specified in chapter Three. The researcher organized the analysis of the study in to two
parts: descriptive analysis part and econometrics analysis part. Both methods of analysis used
the Ethiopian Rural Socioeconomic Survey data of the year 2011/12, 2013/14 and 2015/16 for
the sample of 1,915 households from all regions of Ethiopia.
In this section results for the summary of agricultural production and rural off-farm household
income across years and regions of the country have discussed by using descriptive statistics
of mean, standard deviation, frequency appearance, minimum and maximum values.
Furthermore, we have used one-way ANOVA test to show the significance of variation for
agricultural productivity and rural off-farm household income across years and regions of the
country. We have also employed Bonferroni test to specifically know between which year and
regions of the country does the significant variation occurs.
4.1.1. Summary of agricultural productivity and significance test for its variation across
years and regions of the country
Even though the result of our descriptive analysis revealed the variation in the volume of
agricultural output across regions of the country, About 75.2 million of the rural population
livelihoods depends on agriculture (CSA, 2017). The statistical summary given in table 1
showed that for the year of 2011/12 farmers in Gambella region produces the highest
agricultural output in which each individual farmer produces 793 kg on average followed by
The SNNP farmers with mean value of 773 kg. Contrary to this in the same year the lowest
level of agricultural production takes place in Afar region with mean value of 100 kg and
standard deviation 50kg. There is much variation in the level of annual agricultural output
between regions in 2011/12 ranging from a minimum of 0.4 kg to a maximum of 17,999 kg.
46
To know whether this variation in agricultural output across regions is statistically significant
or not, we have used ANOVA test with a null hypothesis (mean value of agricultural
production in all regions are the same) and alternative hypothesis which states that at least
there is significant mean value difference between two regions in the country. By doing so,
we have obtained P-value of 0.0370 and this enables us to reject our null hypothesis and
concluded that there is statistically significant variation on the mean value of agricultural
production among regions of the country in 2011/12.
For the year 2013/14 farmers in Oromia region produces the highest mean value of output
(2,565 kg) with standard deviation of 5,674 kg. Similar with the year of 2011/12, the lowest
production in this year is undergone in Afar region in which on average each farmer produces
5.56 kg. As it is shown in the table, there is a statistically significant variation in the level of
production in this year at 1% of significance (p-value = 0.0000); output ranging from a
minimum value of 1.5 kg to a maximum of 54,401 kg.
A Farmer in Benshangul regional state produces the highest mean value of agricultural output
for the year 2015/16. In which, on average each farmer produces 2,600 kg with standard
deviation of 4,560 kg. In opposite to this the lowest level of production for this year is done
by the farmers of dire dawa with mean value of 333 kg. By seeing the value of F =4.52 and P
=0.0000 we have concluded that there is a statistically significant variation in the level of
production among regions of the country with 1% level of significance in 2015/16.
For all years of the survey periods, The ANOVA test results have strengthened our findings
by showing the difference in mean value agricultural production is statistically significant.
This is not, however, unexpected; as each region in Ethiopia has different climatic zones and
land fertility, in turn this has strong impact for the variation in agricultural productivity
throughout the country.
To understand which pair of regions in the country has shown a statistically significant
variation in mean values of their agricultural output we have applied Bonferroni test for each
47
year. In 2011/12 the variation in output for each regional comparison is not significant even
though there is statistical significance at the aggregate level.
Table 1: Summary of agricultural productivity across regions for each year and its test for
variation significances across regions
48
Benshangul 2207.7333 4983.8032 75 103 41364
2013/14 SNNP 1286.8629 2146.4586 631 6 40522 5.18 0.0000***
Gambella 1054.4444 1010.7868 36 3 3505
Harari 874.75281 742.7104 89 3 4189
Dire dawa 822.64286 712.51526 70 9 3740
According to our finding, In Ethiopia, The average agricultural output level for an individual
farmer is not similar across the three survey periods. As it is revealed in table 2 a farmer
produces the highest mean value of output (1,649 kg) with standard deviation of 3,481 kg in
the year 2013/14, where as in 2011/12 on average he/she produces the lowest level of output
(670 kg) on average with standard deviation 1,211 kg. Agricultural output ranges from a
minimum of nothing production to a maximum of 128,464 kg for the survey periods in a
given agricultural household. There is statistically significant difference in the annual mean
value of output at 1% probability level. The largest F- value in the table implies high level of
variation in agricultural output across years for an individual farmer.
The level of production increased as we compare the year 2011/12 with 2013/14 and again
reduced in the year 2015/16. This year-wise diffence in the mean amount of output can be
attributed to the 2015/16 El Nino drought which occurs in the country. From the Bonferroni
49
comparison test we have obtained a significant variation in the average level of output
between the year 2011/12 and 2013/14 with mean difference of 978 kg; between the year
2011/12 and 2015/16 with mean diffence of 824 kg. There is no statistical significant
variation in the level of production for a farmer between the year 2013/14 and 2015/16.
Table 2: Summary of agricultural productivity across years and its test for variation
significance
4.1.2. Summary of rural off-farm income and significance test for its variation across
years and regions of the country
Under rain fed characteristics of Ethiopian agriculture, most rural farmers in Ethiopia
produces and harvest once in a year and this entails the harvest must generate income to last
the entire year. Consequently, many rural households turn to different off-farm activities
outside their farm during the hungry season to cope with short-term financial needs as income
earned from agriculture alone cannot satisfy their needs. It can be visually observed from
table 3 that farmer in all regions of the country obtained a certain amount of off-farm income
in all of the study periods. In 2011/12 the highest amount of rural off farm income is obtained
by a farm household in Oromia region with a mean value of 1,003 Birr. Sampled rural
households in Afar region get the lowest amount of off farm income in this year. Contrary to
50
this, on average, farmers in Afar region got the largest amount of off-farm income (4,200
Birr) with standard deviation 7,274 Birr and the lowest off-farm income is obtained by
farmers in Harari region with mean value of 331 Birr in 2013/14. In a similar manner in
2015/16, farmers in Afar region takes the first rank in the amount of off-farm income with a
mean value of 9,358 Birr and standard deviation of 14,936 Birr while Oromia region obtained
the lowest mean value of 301 Birr with standard deviation 1,166 birr for this year.
Rural off-farm income gained by each regions of the household ranges from a minimum of
zero Birr to 48,000 Birr, 50,250 Birr, and 300,000 Birr for the year 2011/12, 2013/14 and
2015/16 respectively. We have made statistical significance test for the variation of rural off-
farm income mean values for each year across regions of the country. Accordingly, in the year
2011/12 and 2013/14, there is statistical significance in mean value variations across regions
at 1% level with P-value of 0.0084 and 0.0005 respectively. Despite of variations in rural off-
farm income for the year 2015/16, this variation is not statistically significant as P value >
0.05.
The Bonferroni test for regional comparison on the variations of rural off- farm income
disclosed that, there is statistical significant variation between regions of Oromia and Amhara
with mean difference of 600 Birr plus SNNP and Oromia with mean difference of 579 Birr in
2011/12. Regional differences in off-farm income in 2013/14 were also found to be
statistically significant at 5% level. This significant variation is between regions of Dire Dawa
and Oromia with mean difference 1,312 Birr plus Oromia and Amhara with mean difference
of 693 Birr.
Our analyses further cleared out the existence of increment in the mean value of rural off-farm
household income from year to year in the survey periods. As table 4 shows in page 53, on
average the highest rural off farm household income in the country is obtained in the year
2015/16 with mean value of 919 Birr and the lowest off-farm income is in the year 2011/12
with mean value 514 Birr and standard deviation of 2,449 Birr.
51
Table 3: Summary of off-farm household income across regions for each year and its test for
variation significances across regions
Year Household’s Summary of total off farm income ANOVA test results
residence in Ethio. Birr
region
Mean Std. Dev. Freq Min. Max F- P- value
value
52
During 2011/12, 2013/14 and 2015/16 physical years, annual rural off- farm household
income ranged from a minimum of nothing to a maximum of 300, 000 Birr. This suggests
that, there are variations in the level of rural off-farm household income across years.
However, this variation is rejected by an F test with P value 0.0667 and leads us to conclude
no statistical significance variation exist in rural off farm household income across the survey
years. Farmers in Ethiopia greatly rely on the income obtained from agriculture for their
livelihood and that’s why insignificant relative variability of off-farm income between
households existed.
Table 4: Summary of total off-farm household income across years and its test for variation
significance
This study used a balanced panel data for investigating the determinants of agricultural
productivity and off farm rural household income in Ethiopia. The panel descriptive statistics
of all variables used in both econometric models are given as follow.
53
4.2.1. Descriptive statistics of variables used in the agricultural productivity model
The summary of the data on a variables used in the agricultural productivity model is given in
Table 5. It shows that, the rural farm household’s agricultural output level was used as a
dependent variable in the productivity model. As a result the mean value of agricultural
productivity is 6.33 kg with its standard deviation of 1.41. The mean value of Household’s
off-farm income is 700 birr with its standard deviation of 5459. The mean value of farm
household those who are uses fertilizer input is about 55 percent out of the total sampled for
the survey with its standard deviation of 0.49.
The mean of the household that that affected by drought and soil erosion are 16% and 25% of
the total sampled household with their standard deviation of 0.37 and 0.43 respectively.
However, the mean value of land holding size for a particular farm household is about 0.23
hectare and its standard deviation is 1.17. The mean value of Extension service access to
credit and agricultural labour input are 0.41, 0.23 and 0.13.29 with their standard deviation of
0.49, 0.42 and 12.71 respectively.
54
Gender of household head 5745 0.82 0.37 0 1
4.2.2. Descriptive statistics of variables used in the off-farm household income model
Here the mean value of dependant variable, which is annual off-farm income, i.e. the total
income earned from household’s farm income and non farm income including remittance is
1.62 birr with its standard deviation of 2.99 as shown in table 6 below. Similarly the mean
value of the independent variables, land holding size and age of household head 0.23 and
46.36 with their standard deviation of 1.17 and 14.7 respectively. However, the mean values
of the number of dependents in the household, number of adults in the household and total
livestock ownership are 1.18, 1.82 and 12.25 with their standard deviation of 0.93, 1.43 and
16.56 respectively.
Those who get an income from off-farm activity and the number of household those who have
access for electricity are very small in magnitude. This implies that, only few rural household
have earns off-farm income and have only few gates electricity access.
55
Table 6: Summary statistics of variables used in off-farm rural household income model
The results of econometric analysis in this section looked over the major factors that
determine agricultural productivity and off-farm household income in rural Ethiopia. The first
part of this section is about the determinants of agricultural productivity, while the second part
focuses on the determinants of rural off-farm household income. We have used random effect
instrumental variable (RE-IV) regression results for our discussion in both models in favour
of fixed effect instrumental variable (FE-IV) as it is rejected by Hausman identification test.
56
4.3.1. Econometric results for agricultural productivity model
There are various socio-economic and demographic factors that determine the level of
farmer’s agricultural production output in rural Ethiopia. Among these factors it is
hypothesized that, off-farm rural household income have a positive impact on the output of
agricultural production in Ethiopia. However, we have suspected the presence of endogeneity
problem in our model between these two variables. Due to this, it is mandatory to run an
endogeneity test in the agricultural productivity model. As the test result shown there is a
serious endogeneity problem between agricultural production output and rural off-farm
household income with P- value = 0.0001 (see appendix III (a)). The Durbin and Wu –
Hausman F test forced us to reject our null hypothesis, variables are exogenous, and we
accept the presence of endogeneity problem in our model.
To come out from the trap of endogeneity problem and generate unbiased estimates in the
model, we have applied fixed effect Instrumental Variable and random effect Instrumental
Variable estimation techniques. Off- farm income is instrumented by a variable, household’s
access to electricity, which has strong relation with off-farm income but have not any relation
with our error term (ui) in the model. The identification criteria for the instrument are also
fulfilled.
After controlling for the endogeneity problem for off-farm income, the next task is the choice
for best fitted model between the fixed effects IV and random effects IV estimation
techniques. The study has employed the Hausman test to select the best model estimator to
avoid the ambiguity of selecting the correct model. The Hausman test (appendix 3(a)) clearly
shows that the P-value is 0.8934 which is far greater than 0.05 leads us to accept our null
hypothesis; random effect IV regression is best fitted estimation technique, in our agricultural
productivity model. Therefore, in this study we have presented and discussed only the
regression results obtained from random effects IV regression output.
The model was also tested for the possible appearances of heteroscedasticity and
multicollinearity problems. The Heteroskedasticity problem is adjusted by regressing the
57
model with robust standard and the multicollinearity problem is also checked and tested using
the observed information matrix (OIM) during the estimation of the variance -covariance
matrix. As a result shown we do not find any multicollinearity (see appendix I (1)) problem
during the estimation for the determinants of agricultural productivity.
After all the appropriate tests, we have run random effects IV regression on fifteen
independent variables that were hypothesized to have influence on the level of agricultural
output in Ethiopia. As it is displayed in table 7, out of these fifteen variables, ten of them
have shown statistically significant impact in determining agricultural productivity during the
study periods. Of the ten significant variables; land holding size, ownership of oxen,
agricultural labour input, educational level of the household head, gender of the household
head, use of chemical fertilizer and drought were highly significant in determining
agricultural productivity at 1% level. The variables use of improved seeds and use of
extension services were also significant at 5% level and off-farm income at 10% level.
However, land dependency ratio, food aid, irrigation use, soil erosion and access to credit
were found to be insignificant factors in determining agricultural productivity in this study.
Furthermore, all the significant independent variables were with the expected signs. Having
this background, the effects of each significant explanatory variable on the level of
agricultural output is discussed briefly in the subsequent paragraphs below.
Land is an important fixed asset of rural farm households in Ethiopia. This asset is a
prerequisite in the production process of agricultural sector. As the land size increases, the
household is able to increase and diversify the quantity and type of crop produced from that
land. The findings of this study also revealed this fact, land holding size positively associated
with agricultural production at 1% level of significance. The result disclosed that other things
remains constant at their mean value, a 10% increases in land holding size leads to an increase
in agricultural output by 1.7%. . This result is consistent with the findings of other researcher
such as Hanan (2008), in Ukraine, Birhanu (2014) and Bihon (2015) in Tigray region of
Ethiopia.
58
Most farmers in Ethiopia predominantly apply plough-based farming system in their
agricultural production process. As it was hypothesized, for a smooth management and
timeous cultivation of land, a household needs a pair of oxen. Hence, the finding indicates
that an additional ox brings with a 0.015% increase in agricultural output with 1% level of
significance. This result is in conformity with the study of Bihon (2015) in Tigray region. If
farmers have at least a pair of oxen, they will be able to cultivate and sow their land at the
appropriate time and increases their land productivity.
The quantity of agricultural output is also directly determined by the variable, agricultural
labour input. Agricultural production in Ethiopia is extremely labour-intensive and employs
little physical capital. As the model result in table 7 revealed, a unit increase in agricultural
labour input for a particular household generates 2.8% increases in agricultural output and this
result is statistically significant at 1% level. This positive association implies the fact that like
all other LDCs, in Ethiopia labor is one of the most extensively used input in the agricultural
sector.
The knowledge and skills of the farmers towards technology and technological changes are
also critical to an inclusive and sustainable growth of agricultural sector in Ethiopia. Educated
peoples can manage their fields properly and then this activity results have pushes to get good
production and productivity of the land. The study result indicates that the level of education
acquired by head of the household is one of the key determinants of agricultural productivity
and is highly significant at 1% level. This is due to the fact that education of the household
heads can raise their information acquisition and adjustment abilities thereby-providing
awareness regarding opportunities for productive employment and rational expectation for
decision making. The coefficient of the variable shows that educated household head
produces 27% more than none educated households.
Similarly, the level of agricultural productivity shows variation due to the case of sex
difference of household heads. Females in Ethiopia have production, reproductive and
childcare burdens which make them less productive than their male counterparts in the
agricultural or other sectors. Also they have less access to information about the technology
59
due to the above mentioned burden than male headed households. Moreover, Male-headed
households are physically strong and capable than female headed households and then the
former would have better opportunities for enhancing their farm productivity. The result of
this study also confirmed this fact. All other things being constant, the output level for male
headed households is 23.7% higher than female headed households and is statistically
significant at 1% level of significance.
Use of chemical fertilizer is another variable which have shown a significant impact on the
volume of agricultural output at 1% level of significance. A research result reported by Khalil
(2015) from Pakistan is also approves this finding. The use of fertilizer increases the
productivity of a given land and hence increases the output level. The study result indicates
that the movement of an agricultural household from non fertilizer user to fertilizer user
increases the agricultural output by 23%. However, in Ethiopia the cost of fertilizer is so high
and farmers with low income level are unable to move simply in this way. Therefore, there
should be a good environment and opportunities for low income farmers by implementing
laws and regulations that facilitate fertilizer access, promote fertilizer use and increase
agricultural productivity.
The history of climate extremes, especially drought, is not a new phenomenon in Ethiopia.
Even though there is a long history of droughts in Ethiopia, studies show that the frequency of
droughts has increased over the past few decades, especially in the lowlands. Since 1957
Ethiopia has experienced serious droughts culminating in the international reaction following
extensive media coverage in 1984. With 75% of people dependent on rain dependent
agricultural activities and another 10% living entirely on livestock, the country remains
vulnerable to drought. The finding of this study also exposed the highly significant negative
impact of drought on agricultural productivity at 1% level. The coefficient result shows when
drought occurs in a particular production year, the level of farm output shrink by 35% as
compared to the normal years output. Therefore, to cope from the drought impact and able to
increase productivity in the agricultural sector government should adopt risk mitigation
strategies.
60
Table 7: Random effect’s IV regression output for the determinants of agricultural
productivity in Ethiopia, 2012-2016
___________________________________________________________________________
Explanatory variables RE + IV
___________________________________________________________________________
Annual off-farm income in Ethiopian Birr 0.0002168*
(0.0001287)
Log of land holding size in hectare 0.1746763***
(0.0234838)
Log of no. of oxen owned by the household 0.0157027***
(0.0040899)
Agricultural labour input 0.027594***
(0.0020413)
Population density in the household (land dependency ratio) - 0.0059576
(0.0036204)
Educational level of the household head 0.2681975***
(0.0395945)
Gender of the household head (male = 1) 0.2372341***
(0.0679847)
Food aid in the given production year (received = 1) - 0.1397301
(0.1098008)
Irrigation (use= 1) - 0.0389242
(0.0786542)
Use of chemical fertilizer (use = 1) 0.2323298***
(0.0594002)
Use of improved seeds (use= 1) 0.1444578**
(0.0572635)
Drought in each production year (yes = 1) - 0.351821***
61
(0.0754646)
Soil erosion (land exposed = 1) 0.0444283
(0.0487319)
Access to credit (yes = 1) 0.0170979
(0.0480014)
Use of extension service (use = 1) 0.1422318**
(0.058843)
_cons 5.338789***
(0.1371782)
___________________________________________________________________________
No. of observations 5,745
No. of groups 1,915
R2 (overall) 0.0478
Wald chi2 (15) 829.13
Prob > chi2 0.0000
Source: calculated from Ethiopian Socioeconomic Survey data, 2012-2016
Note: In parenthesis Robust and clustered standard errors adjusted from hetroskedasticity problem.
***, ** and * represents the coefficients are significant at 1, 5 and 10 percent level respectively.
Use of Improved seeds is also another significant variable of the regression model at 5%
level. The result indicates that farmers who utilized improved seeds in their farm activity got
14.4% more output compared to farmers who did not use them. The result of this study is
consistent with the finding of Bihon (2015) in Tigray region. Therefore, for boosting
agricultural output, the quality of seed is crucial. Regulations should establish mechanisms
that guarantee access for improved seeds for farmers.
It is widely accepted that agricultural extension services play a pivotal role for productivity
improvement. They are crucial for facilitating the dissemination of new agricultural
62
technologies, their learning and adoption by farmers. This research also shows the significant
impact of agricultural extension services in the volume of output at 5% level. The coefficient
results discloses farmers who have access for extension services produced 14.2% more output
compared to farmers who have not any access for extension program. This result is in line
with the findings of Elias et al (2015), a study in north western part of Ethiopia.
Lastly, although only statistically significant at 10% level with p-value 0.092 (see appendix
III (a)), we have observed that off farm income is positively correlated with agricultural
output in Ethiopia. The result of this finding is consistent with other LDCs research work
outputs such as a research done by Cornilius (2005) in Zimbabwe; Wang et al. (2011) in
China and Babutunde (2015) in Nigeria. The coefficient of the regression result shows, in
Ethiopia, when the household’s off-farm income increases by 100 Birr agricultural output
raised by 2.1%. From this result it is possible to say, farmers in Ethiopia uses a large
proportion of their off- farm income for consumption purpose. However, from the total of
1,915 sampled households only 25% of them got income from off the farm. With this
background and the results explained before, we cannot conclude a household does not invest
income from off the farm for his/her farm activities if he/she obtained more. Therefore,
implementing policies and strategies for the increment of rural household’s off-farm income
is important for raising productivity in the agricultural sector.
63
To make our model free from the problem of endogeneity and able to generate unbiased
estimates we have followed a similar procedure with the previous agricultural productivity
model. The log of annual agricultural output is instrumented by two variables, agricultural
labour input and drought. These instruments have strong relationship with agricultural output
but have not any relation with our error term (ui) in the rural off-farm household income
model. The appropriateness of the instruments is also confirmed by the instrumental variable
identification test criteria.
Fixed effect Instrumental Variable (IV) and random effect Instrumental Variable (IV)
estimation techniques are applied and then the best fitted model is chosen by using Hausman
test. As it is shown in the Hausman test (appendix III (b)), the P-value is 0.0555 which is a
little bit greater than 0.05. This result implied that, random effect IV estimation technique
explains our model better than fixed effect IV estimation. Therefore, all of our discussion for
the determinants of rural off- farm household income is based on the regression results
obtained from random effect IV regression output.
In addition to the above diagnostic tests we have checked for the presence of
heteroscedasticity and multicollinearity problems in our model. Due to the existence of
heteroscedasticity problem we have regress the model with robust standard. The
multicollinearity problem is also checked and tested using the observed information matrix
(OIM) during the estimation of the variance -covariance matrix. As a result shown, there is no
any multicollinearity (see appendix I (b)) problem between variables during the estimation for
the determinants of rural off-farm household income model.
After all these tests, we have executed and present the random effect IV regression over nine
independent variables which are hypothesized as the main determinants of off-farm rural
household income in the methodology part. Out of nine variables used in the model, five were
found to be highly and statistically significant factors at 1% level in determining off-farm
household income in rural Ethiopia. These variables are annual agricultural output, age of
household head, education level of the household head, number of dependents in the
household and access to electricity (see table 8). However, the rest four explanatory variables
64
were found to have no significant influence. The coefficients of all significant variables have
expected signs as before except annual agricultural output and age of the household head. The
effect of each significant explanatory variable on Ethiopian off-farm rural household income
is discussed bellow.
Priory, we have hypothesized for the existence of positive relations between agricultural
output and off- farm rural household income in Ethiopia relying on the theory, Farmers who
have produced higher output from their land tend to participate in off-farm activities due to
availability of finance for starting up new off-farm business. Unexpectedly, the actual
estimation result in table 8 revealed the negative relation between these two variables. The
interesting insight here is not only the change in sign but also the level of significance. All
other things being constant, a 10% increase in annual agricultural output will reduce the
amount of off-farm rural household income by 3.1% and is statistically significant at 1%
level. This may be due to the fact that off farm activities compete with farm activities in terms
of the limited household resources. In Ethiopia, agricultural output is strictly labour intensive
but this labour is also required for off-farm income generation. For simultaneous increment in
off-farm income and agricultural output and hence for overall improvement of rural livelihood
the application of physical capital inputs in the agricultural sector is recommended.
It was found that age of household head influences off-farm rural household income
positively and significantly at 1% level. Other things held constant, a one year increase in the
age of household head will increase rural off-farm income by 2.6%. Here also the sign of the
coefficient is in contradiction with our hypothesis. The possible justification is that, as the age
of the household head increases the farmer acquires more knowledge and experiences with in
farm and off-farm activities and able to generate more income outside of farm activities. The
result of this study confirms the earlier findings of Lhing et al. (2013) in Myanmar and
Birhanu (2014) in northern part of Ethiopia.
As the model result indicated, the variable education had positively and significantly
influenced off-farm rural household’s income at 1% level of significance. This finding is in
line with theoretical literature because those farmers with high educational level are more
65
likely to diversify livelihood strategies into non-farming or off-farming activities than those
who less educated. The coefficient of the variable also indicates educated household head
gains 22.6% more income outside of his/her farm than none educated households. This is due
to most probably educated person gain better skill, experience, knowledge and this again help
them to engage in diversified livelihood strategies. Literate individuals are very ambitious to
get information and use it. Therefore, in Ethiopia strict attention should be given for the
development of human capital.
Moreover, the amount of off-farm income in rural Ethiopia is negatively and significantly
determined by the number of dependants in the family at 1% level of significance. The
coefficient of the variable implied all other factors remained at their mean values; an addition
of one dependent member in the household has reduced the household’s off-farm income
level roughly by 11%. This is because a farm household with a majority of dependants is less
likely to participate in off-farm work and hence less income from off farm activities.
Access for electricity is an additional variable which have a positive and highly significant
impact on the amount of rural household’s off-farm income with 1% level of significance.
Surprisingly, the regression parameter indicates other things being constant at their mean
values, households who have access for electricity obtained 63.5% more income from off the
farm as compared to the households who did not have any access for electricity during the
study period. The result collaborates with a study of Bihon (2015) in Tigray region. This
implies how electricity plays a significant role for generating off-farm income in rural areas of
Ethiopia.
66
Table 8: Random effect’s IV regression output for the determinants of off-farm household
income in rural Ethiopia, 2012-2016
__________________________________________________________________________
Explanatory variables RE + IV
___________________________________________________________________________
67
Wald chi2 (9) 85.54
Prob > chi2 0.0000
Note: standard errors in parenthesis and are adjusted from hetroskedasticity problem.
Dependent variable is log of annual off-farm rural household income measured in Ethiopian Birr.
We can fetch out another important finding of this research from the preceding agricultural
productivity and off-farm rural household income regression models. The random effect IV
regression result in table 7 shows, off-farm rural household income positively affects
agricultural productivity and its impact is statistically significant. However, the regression
output in table 8 indicated the increase in agricultural output will reduce the generation of off-
farm income. This implies the impact of off-farm income on productivity reduces when the
level of production increases. The main reason for this non linear relationship may be due to
the substitutability of household’s limited labour input between farm and off-farm activities in
the country. To sum up, the finding indicates the existence of a positive non linear empirical
relationship between off farm income and agricultural productivity in Ethiopia. However, it is
possible to reverse this non linear relation by applying physical capital inputs in the
agricultural sector.
68
5. SUMMARY, CONCLUSIONS AND RECOMMENDATIONS
5.1. Summary
Ethiopia’s rural population livelihood relies on rain dependence agriculture and households
are subject to extreme poverty. Small farmers of the country who produces low level of output
and obtain small amount of off-farm income, encompasses the largest portion of the
population. This low level of productivity and income is one of the reasons for rural
households to live in the vicious circle of poverty for a long period. Given that increasing
agricultural productivity and income matters for reducing poverty level in the country;
identifying the factors that determine productivity and off-farm income is vital for future
policy design.
Accordingly, this paper has examined the determinants of agricultural productivity and off-
farm household income in rural Ethiopia. For identifying these factors, a study employed the
Ethiopian rural socioeconomic survey (ERSS) data. Three waves balanced panel data for a
total of 1, 915 rural sampled households are used for the study. A random effect instrumental
variable analysis method is applied to identify the determinants of both agricultural
productivity and off-farm household income at the country level.
The descriptive result of the study showed the existence of significant variation in the level of
agricultural output across regions of the country. The regions Dire Dawa and benishangul,
Dire Dawa and Oromia, SNNP and Benishangul, SNNP and Oromia plus SNNP and Amhara
showed a significant variation in their agricultural output. The average agricultural output
level for an individual farmer is also not similar across the three survey periods. There is
significant variation in the average level of output between the year 2011/12 and 2013/14.
In view of off-farm income our descriptive analysis indicated the presence of a significant
variation in the amount of off-farm income between regions of Oromia and Amhara, SNNP
and Oromia, plus Dire Dawa and Oromia. Even though there is an increment in the level of
69
off-farm incomes from year to year this study revealed the absence of significant variation in
the amount of rural off farm household income across the survey periods.
The random effect instrumental variable regression model for the determinants of agricultural
productivity showed that among fifteen explanatory variables, which were included in the
model, land holding size, ownership of oxen, agricultural labour input, educational level of
the household head, gender of the household head, use of chemical fertilizer and drought were
found to be highly significant in determining agricultural productivity at 1% level in Ethiopia.
Use of improved seeds and use of extension services were also significant at 5% level and off-
farm income at 10% level in determining agricultural productivity. This study also found a
positive non-linear empirical relationship between farm off-farm livelihoods of the rural
population.
The analysis of the study further indicated that the amount of off-farm rural household income
in Ethiopia is significantly and positively determined by age of household head, education
level of the household head, and access for electricity at 1% level. However, annual
agricultural output and number of dependents in the household are found to have negative
significant impact on off-farm household income in rural Ethiopia.
This study has tried to identify the determinants of agricultural productivity and off-farm
household incomes in rural Ethiopia. Given the above findings of the study we have suggested
the following general recommendations for further consideration and improvement for the
livelihood of rural population.
The empirical results of this study showed that asset holdings such as oxen and size of
cultivated land are positively associated with agricultural productivity. Therefore, this
is an insight that rural household asset building program should be implemented to
enhance the level of production in the agricultural sector. However, expanding any
more land holding today is increasingly rare because of increasing demographic
70
pressure on land and degradation of the existing land resource. This calls for
introducing the issue of increasing productivity per unit of land area and
intensification. This would be fruitful provided that it is accompanied with policy
interventions of investment in water harvesting and irrigation technologies than
depending merely on erratic rainfall thereby relieving livelihood risks emanating from
shortage of rainfall.
This study revealed a high significant negative impact of drought in the agricultural
production process. In order to cope up from the drought impact, risk mitigation
strategies should be adopted. Risk mitigation strategies at household levels include:
crop diversification, mixed crops and livestock production. Public level risk mitigation
strategies include: water harvesting, resources conservation and management,
irrigation, voluntary resettlement programs, household extension packages, agro-
ecological packages, productive safety net programs, and the newly pilot studies that
is weather-indexed drought insurance and commodity exchange programs.
Human capital namely education level of the household have a significant impact on
both agricultural productivity and off-farm household income. The positive
contribution of this human capital for rural livelihood improvement calls for integrated
intervention of rural people-centered education especially it is preferable to have
farmer teaching centers aimed to promote their educational level in all issues linked
with rural livelihood.
71
Furthermore, the finding shows off farm income involvement in the process of
agricultural production significantly improved the level of productivity. However,
only limited households obtained off-farm income in rural Ethiopia in the study
periods. Therefore, government should give more attention over the significant
determinants of off-farm income which are found in this study to increase the level of
rural off-farm income and to extend its impact for increasing agricultural productivity.
72
REFERENCES
Abbas, A., Chandio, I., Yuansheng, J., Abdul Rehman, & Luan, J. (2016).Impact of
Government Expenditure on Agricultural Sector and Economic Growth in Pakistan.
American-Eurasian Journal on Agriculture and environmental science, 16 (8).
Abay, A., & Assefa, A. (2004). The role of education on the adoption of chemical fertilizer
under different socioeconomic environments in Ethiopia. Agricultural Economics, 30
(2004).
Abdulhafidh, D. (2013). Agricultural productivity and poverty alleviation: what role for
technological innovation. University of Sousse, Sousse, Tunisia.
Adeoti, A.I., Cofie, O., & Oladele, O.I. (2012).Gender analysis of the contribution of urban
agriculture to sustainable livelihoods in Accra, Ghana. Journal of Sustainable
Agriculture, 36, 236-248.
Adomi, E.E., Ogbomo, M.O., &Inoni, O.E. (2003). Gender factor in crop farmers’ access to
agricultural information in rural areas of Delta State, Nigeria. Library Review, 52.
AGRA (2017).Africa agriculture status report: the business of smallholder agriculture in Sub-
Saharan Africa (Issue 5). Nairobi, Kenya: Alliance for a Green Revolution in Africa,
Issue No. 5.
73
Ahituv, A., & A. Kimhi (2002). Off-farm work and capital accumulation decisions of farmers
over the life-cycle: the role of heterogeneity and state dependence. Journal of
Development Economics, 68: 329–53.
Anne, M. (2011).A survey of agricultural productivity and nutritional status in rural South
Wollo, Ethiopia. Anthropology Department Theses and Dissertations 15.
Retrived from http://digitalcommons.unl.edu/anthrotheses/15
Anne, T. (2004). Agriculture growth and poverty reduction. Agriculture and Natural
Resources Team of the UK Department for International Development, Oxford Policy
Management, Oxford.
Anriquez, G. & Daidone, S. (2010). Linkages between the farm and nonfarm sectors at the
household level in rural Ghana: a consistent stochastic distance function approach.
agricultural economics, 41: 51–66.
Assefa, A., (1989). Some factors influencing agricultural credit, among peasant farmers in
Ethiopia: A case study of two districts. Ethiopian Journal of Development Research,
11(1): 10.
Asfaw, A., & Admassie, A. (2004).The role of education on the adoption of chemical
fertilizer under different socio-economic environments in Ethiopia. Agricultural
Economics, 30, 215-228.
Babatunde, R.O., &Qaim, M. (2009). The role of off-farm income diversification in rural
Nigeria: Driving forces and household access. Paper presented at the Centre for the
Study of African Economies (CSAE) Conference, Economic Development in Africa,
March 22-24, University of Oxford, United Kingdom.
Babatunde, R., &Qaim, M. (2015).Impact of off-farm income on food security and nutrition
in Nigeria. University of Ilorin, PMB 1515 Ilorin, Nigeria.
74
Baltagi, B. H. (2005). Econometric analysis of panel data. 3rd edition, John Wiley & Sons Ltd,
The Atrium, Southern Gate , England.
Beyan, A. (2015).What factors contribute to the small holder farmers farm income
differential? Evidence from east Hararghe, Oromia, Ethiopia. Haramaya University,
Ethiopia.
Bhaumik, S.K., Dimova, R., & Nugent, J.B. (2011). Off-farm labour supply and labour
markets in rapidly changing circumstances: Bulgaria during transition. Economic
System, 35, 378-389.
Bishaw, Z., Struik, P.C., & Van Gastel, A.J.G. (2012). Farmers’ seed sources and seed
quality: physical and physiological quality. Journal of Crop Improvement, 26, 655-
692.
75
Bryant, S.J. (2010). Draft animal power. Journal of Agricultural and Food Information, 11,
360-366.
Boserup, E., (1965). The conditions of agricultural growth: the economics of agrarian change
under population pressure. Allen &Unwin, London.
Cameron, A. C., &Trivedi, P. K. (2010).Micro econometrics using Stat. college station, stata
press, USA.
Central Statistical Agency. (2017). Agricultural sample survey 2016/17: volume 1. Report on
area and production of major crops. Addis Ababa, Ethiopia.
Chaplin, H., Davidova, S., & Gorton, M. (2003).Agricultural adjustment and the
diversification of farm households and corporate farms in Central Europe. Journal of
Rural Studies, 20, 61-77.
Coelli, T.D.S., Prasada, R., & Battese, G.E. (1998).An introduction to efficiency and
productivity analysis. kluwer academic publisher, pakistan research repository.
Collier, P. &Lal, D. (1986).labour and poverty in Kenya from 1900-1980. Oxford University
Press, New York.
76
Cornelius, C. (2004). Rural off-farm employment and farm investment: an analytical
framework and evidence from Zimbabwe. Discussion paper 2004/03.
Datar, G. & Carpio, X.V. (2009). Are irrigation rehabilitation projects good for poor farmers
in Peru? A spatial regression discontinuity approach. Policy Research Working Paper
5154, Impact Evaluation Series No. 42.
Datt, G, & Ravallion, M. (1998a). Farm productivity and rural poverty in India. Journal of
Development Studies, 34(4): 62-85.
Datt, G. and M. Ravallion (1998b). How important to India’s poor is the sectoral composition
of economic growth? The World Bank economic review, 10: 1-25.
Davis, B., Winters, P., Reardon, T., & Stamoulis K. (2009). ‘Rural Nonfarm Employment and
Farming: Household-Level Linkages’. Agricultural Economics, 40: 119–23.
Egyei, R., & Adzovo, r.P. (2013).Household non-farm income: any influence on agricultural
productivity in rural Ghana?. University of Ghana, Legon, Ghana.Vol.3, No.9, 2013.
Ekbom, A., Yonas, A. & Sterner, T. (2012).Integrating soil science into agricultural
production frontiers. Working Papers in Economics No 550.
Elias A. ,Nohmi M., Yasunobu k., & Ishida A. (2015). Farmer’s Satisfaction with Agricultural
Extension Service and Its Influencing Factors: A Case Study in North West Ethiopia.
Vol. 18: 39-53, May 2015.
77
Ellis, F.(2000. Rural livelihood and diversity in developing countries. New York; Oxford
University Press.
Ellis, F. & Freeman, H.A. (2004).Rural livelihoods and poverty reduction strategies in four
African countries. journal of Development Studies, 40(4): 1-30.
Endrias, G., Ayalneh, B., Belay, K. & Eyasu, E.(2013). Productivity and efficiency analysis
of smallholder maize producers in Southern Ethiopia. J Hum Ecol, 41(1): 67-75 (2013).
FAO (2017).The future of food and agriculture – Trends and challenges. Rome.
Frank, E. (1999). Rural livelihood diversity in developing countries: evidence and policy
implication. Overseas development institute, Number 40, April 1999.
Gelman, A., & Hill, J. (2007).Data analysis using regression and multilevel/ hierarchical
models. Cambridge: Cambridge University Press.
Ghosh, N. (2004). Reducing dependency on chemical fertilizers and its financial implications
for farmers in India. Ecological Economics, 45, 149-162.
Gollin, D. (2015). Agriculture as an Engine of Growth and Poverty Reduction: Lessons for
Africa. Current and Emerging Issue, Oxford University Press, Oxford.
Greene, W. (2003), Econometric Analysis. Fifth edition. New York : Prentice Hall.
Gujarati, D. N., & Dawn C.(2009). Basic Econometrics, McGraw-Hill International Edition,
5th Edd., Boston.
78
Guoqiang, R., &Wenting, D. (2013).Analysis on the evolution of the village’s off-farm
employment considering the resource endowment. Journal of Applied Sciences
13,4072-4076.
Haduse, H. (2014).Adoption and impact of micro irrigation on household income: the case of
Eastern Tigray, Ethiopia. M.Sc thesis, Mekele University, Ethiopia.
Haggblade, S., Peter B.R., &Thomas R. (2007). Transforming the rural nonfarm economy:
opportunities and threats in the developing world. The Johns Hopkins University Press:
Baltimore.
Hazell, P. & Haggblade, S. (1993). Farm-nonfarm growth linkages and the welfare of the
poor, In including the Poor. The World Bank, Washington, D.C.
Hoffman, B. & Heidhues, F. (1993).Credit access in the informal sector: a case study of two
villages in the province queme in Benin. Wageningen, Netherlands, 271–277.
Holden, S., Shiferaw, B., & Pender, J. (2004).Non-farm income, household welfare, and
sustainable land management in a less-favoured area in the Ethiopian highlands. Food
Policy, 29, 369-392.
79
Hussain, I., &Hanjra, M.A. (2004). Irrigation and poverty alleviation: Review of the empirical
evidence. Irrig.and Drain., 53(2004), 1-15.
Ighodaro, I., Lategan, F., Yusuf, S.F. (2013).The impact of soil erosion on agricultural
potential and performance of sheshegu community farmers in the eastern cape of South
Africa. Journal of agricultural science, vol 5, No. 5: 2013.
Irz, X., & Roe, T. (2000). Can the world feed itself? Some insights from growth theory.
Agrekon, 39(3), 513-528.
Irz, X., Lin, T., & Wiggins, S.( 2001). Agricultural Productivity Growth and Poverty
Alleviation. Development Policy Review, 19(4), 449-466.
Jones, B. (2007). Soil erosion.Paper presented at the thematic strategy for soil protection,
common criteria for risk area identification in the soil framework directive.25th April,
2007, BGR, Hannover.
Johnston, K. (nd). New Zealand families: size income and labour force participation.
University of Waikato population studies center, New Zealand.
Josephson P., & Anna L. (2013). How population density influences agricultural
intensification and productivity: evidence from Ethiopia. Open Access Theses, Paper
31.
80
Kanagawa, M., & Nakata, T. (2008).Assessment of access to electricity and the socio-
economic impacts in rural areas of developing countries. Energy Policy, 36, 2016-
2029.
Janvry, A., & Sadoulet, E. (2002). World poverty and the role of agriculture technology:
Direct and indirect effects. Journal of Development Studies, Volume 38(4), 1-26.
Kooijman, V., &Dijk, A.L. (2011).The role of energy in creating opportunities for income
generation in the Indian Himalayas. Energy Policy, 41, 529-536.
Kwon, C.W., Orazem, P.F., & Otto, D.M. (2006).Off-farm labour supply responses to
permanent and transitory farm income. Agricultural Economics, 34, 59-67.
Leigh, A. (2011). Yield gap and productivity potential in Ethiopian agriculture: Staple
grains and pulses. Discussion paper.
Lhing, N., Nanseki, T., & Takeuchi F. (2013).An analysis of factors influencing household
income: a case study of pact microfinance in kyaukpadaung township of Myanmar.
American journal of human ecology Vol. 2, No. 2, 2013, 94-102.
Lopez, F., Pfeiffer, L., & Taylor, J. (2007).Is off-farm income reforming the farm? Evidence
from México. Department of economics draft paper, University of California, USA.
81
Mainali, B., &Silveira, S. (2012). Renewable energy markets in rural electrification: Country
case Nepal. Energy for Sustainable Development, 16, 168-178.
Mellor, J. (1999). Faster, More Equitable Growth – The Relation between Growth in
Agriculture and Poverty Reduction. Agricultural Policy Development Project ,Research
Report No. 4, Washington D.C.
Retrieved from http://www.abtassociates.com/reports/19998977199472.pdf
Nepal Rastra Bank. (2014). Agricultural credit and its impact on farm productivity: a case
study of Kailalidistrict Nepal. Rastra Bank Banking Development and Research Unit
Dhangadhi Office, November,2014.
Nyanga, P.H., & Umar, B.B. (2012). A Polycentric approach to rural poverty reduction: Case
of Zambian smallholder farmers Progress. A conference paper for IFPRI, Norwegian
University of Life Sciences.
Ogoke, I.J., Ibeawuchi, I.I., Ngwuta, A.A., Tom, C.T., & Onweremadu, E.U. (2009).Legumes
in the Cropping Systems of South-eastern Nigeria. Journal of Sustainable Agriculture,
33823-834.
Oseni, G., &Winters, P. (2009). Rural nonfarm activities and agricultural crop production in
Nigeria. Agricultural Economics, 40(2), 189-201.
82
Pandit, A.D. (1965), Application of productivity concept to Indian agriculture and
productivity. Special issue on agricultural productivity 6, 1965, P.187.
Pane, B. (2006).The impact of food aid in Ethiopia. Research report on poverty action
network. Addis Ababa, Ethiopia. August 2006.
Pfeiffer, L., Lopez-Feldman, A. & Taylor, J.E. (2009). Is off-farm income reforming the
farm? evidence from Mexico. Agricultural economics, 40(2), 125-138.
Pinstrup, E. (1995). Agricultural growth is the key to poverty alleviation in low income
developing countries. International food policy research 2020 brief 15, 1995.
Rahman, S.U., Hussain, A., &Taqi, M. (2014). Impact of agricultural credit on agricultural
productivity in Pakistan: an empirical analysis. International journal of advanced
research in management and social science.vol.3.no.4.
Ruben, A. (2014). Impact of agricultural foreign aid on agricultural growth in sub Saharan
Africa: a dynamic specification. A grouped working paper, V. 0006, July 2014.
Ruben, R., & Van, D. M. (2001).Non-farm employment and poverty alleviation of rural farm
households in Honduras. In world development 29 (3): 549-560.
83
Sassenrath,G.F., Schneider, J.M., Gaj, R., Grzebisz, W. & Halloran, J.M. (2012). Nitrogen
balance as an indicator of environmental impact: Toward sustainable agricultural
production. Renewable agriculture and food systems.
Sara, R. (2009). Gender and occupational choices in Africa: the role of time poverty and
associated risks. University of Florida, USA.
Schneider, K., & Mary K. G. (2011). Agricultural Productivity and Poverty Reduction:
Linkages and Pathways, The Evans Review, 1(1): 55-74.
Sofia, M. (2010). How does food aid impact agricultural production and household supply to
agriculture in Ethiopia? Master thesis, Washington. D.C. April 30, 2010.
Stampini, M., & Davis, B. (2009). Does non-agricultural labour relax farmer’s credit
constraints? Evidence from longitudinal data for Vietnam. Agricultural Economics,
40(2), 177–188.
Stark, O., & Bloom D. (1985).The new economics of labor migration. American Economic
Review, 5(1985):173-178.
Tefera, D. (2004). Determinants of smallholder farmer’s demand for non-formal credit: The
case of Farta Woreda. Un published M.Sc thesis, Alemaya University, Ethiopia.
Tesfaye, B., Thomas, H., & Surry, Y. (2015).Effect of off-farm income on smallholder
commercialization: panel evidence from rural households in Ethiopia. Institute for
Food and Resource Economics, University of Bonn, Germany.
84
Tessema, U. (2015). The determinants of agricultural productivity and rural household
income in Ethiopia. M.Sc thesis, Addis Ababa University, Ethiopia.
Tesfaye, Y. (2015). Analyzing the determinants and the effects of income diversification in
rural Ethiopia. Hague, the Netherlands. December 2014.
Thabit, A. ( 2015). Economic analysis of factors affecting the farmer income under
traditional farming system in South Darfur state – Sudan. Journal of Agricultural
Science and Engineering, Vol. 1, No. 3, 2015, 114-119.
Retrieved from http://www.aiscience.org/journal/jase
Thirtle, C., Lin, L., & Piesse, J. (2003).The impact of research-led agricultural productivity
growth on poverty reduction in Africa, Asia and Latin America. World Development,
31(12), 1959-1975.
Thirtle, C., Irz, X., Lin, L., Mckenzie V., & Wiggins, S. (2001). Relationship between
changes in agricultural productivity and the incidence of poverty in developing
countries. Report No. 7946, London.
Timmer, P. (1995). Getting agriculture moving: Do markets provide the right signals? Food
Policy notes toward a research program. 20(5): 455–472.
Retrieved from http://www.econ.yale.edu/~cru2//pdf/agdevmombasa2009paperc.pdf
Todaro, M., & Smith, S. (2012). Economic development.11th edition, United States of
America.
UNECA (2009). Economic Report on Africa 2009: Developing African agriculture through
regional value chains. Economic Commission for Africa, Addis Ababa, Ethiopia.
85
United Nations World Institute for Development Economics Research (2017).Effect of non-
farm work on agricultural productivity: Empirical evidence from northern Ghana.
Working Paper, 2017/38.
Verbeek, M. (2000).A Guide to modern econometrics 2nd edition. John Wiley & Sons press,
England.
Wooldridge, J. (2010). Econometric analysis of cross section and panel data. Cambridge, MA:
The MIT Press.
World Bank. (2007a). Enabling the business of agriculture. World balk group discussion
paper, 2007.
Ye Wang, Y., Wang, c., & Pan, S. (2011). The impact off nonfarm activities on agricultural
productivity in rural China. Department of agricultural and applied economics, Texas Tech
University.
86
APPENDIXES
Table a. 1: Correlation between the variables used in the regression for agricultural
productivity model after off farm income variable is instrumented
OFI lnLHS lnOPO ALI LDR EDU GHH FAID IRG UCF UIS
OFI 1.0000
IRG -0.2680 0.1258 0.1175 -0.2240 0.0376 0.2180 -0.2013 -0.1836 1.0000
UCF 0.0352 0.2338 -0.1369 -0.0448 -0.0734 -0.0836 0.0187 0.0809 -0.1091 1.0000
UIS -0.4527 -0.0290 0.3849 -0.1750 0.0758 0.2064 -0.2496 -0.0174 0.1752 -0.1456 1.0000
DR -0.4653 -0.1001 0.3032 -0.1702 0.0640 0.2027 -0.1580 -0.2470 0.2451 -0.2917 0.2315
SOE 0.1455 0.0191 0.1433 0.1493 -0.0217 -0.1993 -0.0696 0.1221 0.0792 0.0357 -0.0241
ACR 0.1095 -0.1071 -0.0251 0.1660 -0.0175 -0.1153 -0.0501 0.0045 -0.0238 -0.1688 0.0427
UES 0.1597 -0.1014 -0.2555 0.0265 -0.0120 0.0106 0.1846 0.0200 -0.1492 -0.4552 -0.4101
87
DR SOE ACR UES
DR 1.0000
Table a. 2: Correlation between the variables used in the regression for rural off-farm
household income model after productivity variable is instrumented
lnAGO 1.0000
ACR -0.1863 0.0595 0.0644 0.0924 0.0346 -0.0074 0.0126 0.0889 1.0000
88
Appendix II: test of endogeneity for agricultural productivity and off-farm household
income model
89
Appendix III: Instrumental Variable Regression Outputs and Hausman Test Results
90
Table a. 4: Random effect IV regression output for agricultural productivity model
91
Table a. 5: Hausman test result for agricultural productivity model
92
b) Instrumental variable regression and Hausman test results for off-farm rural
household income model
Table a. 6: Fixed effect IV regression output for off-farm household income model
93
Table a. 7: Random effect IV regression output for off-farm household income model
94
Table a. 8: Hausman test result for off- farm household income model
95
Appendix IV: The Regional Distribution of EAs, Sampled Households for Interview and Actual
Households Included in the Study from Rural Ethiopian Socioeconomic Survey
Data
Table a. 9: The regional distribution of EAs, sampled households and actual households
included in the study
Afar 10 100 3
Somali 20 200 37
Gambella 10 100 36
Harari 10 100 89
Source: own construction based on the Ethiopian socioeconomic survey data basic information documents of
2012, 2015 and 2017.
96