FARM MECHANIZATION, COMMERCIALIZATION AND CROP

DIVERSIFICATION IN CENTRAL AND SOUTHERN OROMIA REGION,

ETHIOPIA

PhD DISSERTATION

TAMRAT GEBISO CHALLA

MARCH 2024
HARAMAYA UNIVERSITY, HARAMAYA
 ii

Farm Mechanization, Commercialization and Crop Diversification in Central

and Southern Oromia Region, Ethiopia

A Dissertation Submitted to the Postgraduate Program Directorate

(School of Agricultural Economics and Agribusiness)
HARAMAYA UNIVERSITY

In Partial Fulfillment of the Requirements for the Degree of

DOCTOR OF PHILOSOPHY
IN AGRICULTURAL ECONOMICS

Tamrat Gebiso Challa

March 2024
Haramaya University, Haramaya
 Haramaya University
Postgraduate Program Directorate

I hereby certify that I have read and evaluated this Dissertation entitled “Farm Mechanization,
Commercialization and Crop Diversification in Central and Southern Oromia Region,
Ethiopia” prepared under my guidance by Tamrat Gebiso. I recommend that it be submitted as
fulfilling the dissertation requirement.

Mengistu Ketema (PhD, Professor) ______________ __________

Chairman, Advisory Committee Signature Date
Arega Shumetie (PhD, Assoc. Prof) ______________ __________
Member, Advisory Committee Signature Date
Getachew Legese (PhD) ______________ __________
Member, Advisory committee Signature Date

As a member of the Board of Examiners of the PhD Dissertation Open Defense Examination, I
certify that I have read and evaluated the dissertation prepared by Tamrat Gebiso and examined
the candidate. I recommend that the dissertation be accepted as fulfilling the dissertation
requirement for the degree of Doctor of Philosophy in Agricultural Economics.

Abebaw Shimelis (Ph.D.) _________________ _________________

Chairperson Signature Date

Legese Dadi (Ph.D.) __________________ _________________

External Examiner Signature Date

Jema Haji (Ph.D., Professor) ________________ _________________

Internal Examiner Signature Date
 iii

DEDICATION

This piece of my dissertation work is dedicated to my hero father, Gebiso Challa.

iii
 iv

STATEMENT OF THE AUTHOR

By my signature below, I declare and affirm that this Dissertation is my own work. I have
followed all ethical and technical principles of scholarship in the preparation, data collection,
data analysis and compilation of this Dissertation. Any scholarly matter that is included in the
Dissertation has been given recognition through citation.

This Dissertation is submitted in partial fulfillment of the requirement for PhD. degree at
Haramaya University. The Dissertation is deposited in the Haramaya University Library and is
made available to borrowers under the rule of the library. I solemnly declare that this
Dissertation has not been submitted to any other institution anywhere for the award of any
academic degree, diploma or certificate.

Brief quotation for this Dissertation may be made without special permission provided that
accurate and complete acknowledgement of the source is made. Requests for permission for
extended quotations from our reproduction of this Dissertation in whole or in part may be
granted by the Head of the School of Agricultural Economics and Agribusiness 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 Dissertation.

Name: Tamrat Gebiso Challa Signature ________________________

Date: ________________________
School: Agricultural Economics and Agribusiness

iv
 v

BIOGRAPHICAL SKETCH

Tamrat Gebiso was born in Seden Sodo district, Southwest Shewa zone, Oromia regional state
on February 08, 1982. After two years, his family moved to Nono district West Shewa zone and
he was raised there in a farming community. He attended his primary school (Grade 1 to 8) in
Tadele Dargie (Xaadallee Daargee) primary and junior secondary school. He joined Goro
General Secondary School (Gurage zone, Wolkitie town) in 1996 and attended his grade 9 and
10 up to 1997. Then he joined Dejazmach Geresu Duki Comprehensive General Secondary
School in 1998 and attended his grade 11 and 12 in Woliso, Southwest Shewa zone of Oromia
regional state. He took his Ethiopian School Leaving Certificate Examination (ESLCE) and
joined Haramaya University (the then Alemaya University) in 2001 and graduated from
department of Agribusiness Management (ABM) in July, 2005 with his BSc degree. After his
graduation, he was employed by Ministry of Agriculture as a Junior Instructor at Alage ATVET
college where he served for only one semester. Then he joined Oromia Agricultural Research
Institute (IQQO) based at Asella Agricultural Engineering Research Center as Junior
Socioeconomic researcher. He rejoined Haramaya University in September, 2009 for his
Master’s program and awarded his MSc. degree in 2011 in the field of Agricultural Economics.
Finally, he again joined the same University for his PhD degree in the field of Agricultural
Economics in September, 2019.

v
 vi

ACKNOWLEDGEMENTS

Above all, my special thank goes to Almighty God, who gave me the courage to pursue this
study and I know that everything in my life happened with his goodwill and the help of his
righteous specially mother of God St. Virgin Mary. My heartfelt thanks also go to my Advisors
Prof. Mengistu Ketema (the chairman of the advisory committee) and the member of the
advisory committee, Arega Shumetie (PhD) and Getachew Leggesse (PhD). Your supports at
each stage of my dissertation starting from conception and design, to the final write-up were
irreplaceable. It was really due to your efforts that I am here today. You gave me all kinds of
supports to finalize my work on time. I want to thank you all for your brotherly and friendly
approaches beyond the advisory work of dissertation and it has been a great opportunity to work
with you.

I would like to thank Oromia Agricultural Research Institute for giving me the chance to pursue
my PhD study and sponsoring the data collection. My next gratitude will go to Dr. Dagnachew
Lule (former AGP-II coordinator at IQQO) for the financial support in data collection. Mr.
Tilahun Geneti and Mr. Teha Mume also deserve special thanks for their support in every aspect
specially for inviting me to participate on CSPro software training that really made things easy in
data collection. Thank you Dr Fato Esmo (former IQQO DG) for your encouragement to
continue my study especially when I was planning to withdraw due to my sickness during the
course work (2012 E.C.). I would like to acknowledge the commendable efforts put by
researchers from Asella Agricultural Engineering Research Center, the data collection team:
Ephrem Boka, Aman Nebo, Getu Regassa, Abdisa Teshome, Degefa Weyessa, Webi Tafa,
Asnake Tilaye, and Rebirra Nigusse and our driver Kebede A. who were devoted to make data
collection successful. Our center director, finance head (Kuftu Mamiya) and her team and
research team leaders also deserve special thanks. My friends Dr. Tadele Mamo and Sintayehu
Fetene also deserve special thanks for their inspirations to pursue my study. Head of
Agricultural Economics and Agribusiness school and staff members also deserve especial thanks
for your positive cooperation. I am also very grateful to our batch’s fellow students and
colleagues with whom I have shared knowledge, experience and life at HU.

vi
 vii

I want to extend my deepest thank for my father Mr. Gebiso Challa who bringuped me with love
and affection, and is always inspiring me in my life especially in academic line considering all
my academic progress as his own success. Last but not the least, thank you my lovely wife
Haimanot Chanyalew for all kinds of support and taking all responsibilities of leading our family
and taking care of our children by scarifying all your interest for this long journey. My queens,
Egzihareya and Keniel, thank you for your patience and I always remember your question “Are
you going to work even on Sunday? So, when are you going to be with us?”. Now is the time to
spent with you.

vii
 viii

ACRONYMS AND ABBREVIATIONS

AE Allocative Efficiency
AIPW Augmented Inverse Probability Weight
AML Agricultural Mechanization Level
ARDU Arsi Rural Development Unit
ATA Agricultural Transformation Agency
CADU Chilalo Agricultural Development Unit
CEMA European Agricultural Machinery Industry
COM Crop Output Market
COCI Crop Output Commercialization Index
CRS Constant Returns to Scale
CSA Central Statistics Agency
DEA Data Envelopment Analysis
DHM Double Hurdle Model
DRF Dose-Response-Function
EE Economic Efficiency
FACASI Farm Mechanization & Conservation Agriculture for Sustainable Intensification
FAO Food and Agriculture Organization
FGD Focus Group Discussion
GDP Gross Domestic Product
GoE Government of Ethiopia
GPS Generalized Propensity Score
HIMIs High and Intermediate Mechanization Implements
IMR Invers Mills Ratio
IPW Inverse Probability Weight
KII Key Informants Interview
LOM Level of Mechanization
MI Mechanization Index
MSE Mean Squared Errors
OLS Ordinary Least Squares

viii
 ix

PS Propensity Score
PSM Propensity Score Matching
SDI Simpson Diversification Index
SFA Stochastic Frontier Analysis
SSA Sub-Saharan Africa
SUR Seemingly Unrelated Regression
TE Technical Efficiency
TLU Tropical Livestock Unit
UNIDO United Nations Industrial Development Organization
WFB World Fact Book

ix
 x

TABLE OF CONTENTS

STATEMENT OF THE AUTHOR IV

BIOGRAPHICAL SKETCH V
ACKNOWLEDGEMENTS VI
ACRONYMS AND ABBREVIATIONS VIII
TABLE OF CONTENTS X
LIST OF TABLES XIV
LIST OF FIGURES XVI
LIST OF TABLES IN APPENDIXES XVII
ABSTRACTS XVIII
1. INTRODUCTION 1
1.1. Background of the Study 1
1.2. Statement of the Problem 4
1.3. Objectives of the Study 7
1.4. Scope and Limitations of the Study 8
1.5. Significance of the Study 8
1.6. Organization of the Dissertation 9
2. LITERATURE REVIEW 10
2.1. Definitions of Basic Related Terms 10
2.1.1. Farm Mechanization and Its Adoption 10
2.1.2. Impacts of Adoption of Farm Mechanization 12
2.1.3. Agricultural Commercialization 14
2.1.4. Production Efficiency and Productivity 15
2.1.5. Crop Diversification 16
2.2. Theoretical Perspectives 17
2.2.1. Agricultural Development, Induced Innovation Theory and Farm Mechanization 17
2.2.2. Adoption of Farm Mechanization 20

x
 xi

TABLE OF CONTENTS (Continued)

2.2.3. Impacts of Adoption of Farm Mechanization on Farm Labor Employment, Income,
Economic Efficiency, Productivity and Input Use 23
2.2.4. Crop Commercialization 24
2.2.5. Productivity and Production Efficiency 29
[Link]. Input-oriented measures 31
[Link]. Output oriented measures 32
2.2.6. Crop Diversification 34
2.3. Analytical Framework 35
2.3.1. Measurement of Agricultural Mechanization Level 35
2.3.2. Economic Impact of Adopting Farm Mechanization 37
[Link]. Analytical framework for technology adoption 37
[Link]. Impact of farm mechanization on EE, income and intensity of input use: 38
2.3.3. Crop Commercialization 39
2.3.4. Measurement of Production Efficiency and Identifying its Determinants 41
2.3.5. Crop Diversification and Its Determinants 44
2.4. Review of Related Empirical Literatures 45
2.4.1. Agricultural Mechanization History in Ethiopia 45
[Link]. Agricultural mechanization during the imperial regime 45
[Link] Agricultural mechanization during the Dergue regime (1974-91) 46
[Link] Agricultural mechanization in Ethiopia after 1991 47
2.4.2. Economic Effects of Agricultural Mechanization 47
2.4.3. Factors Affecting Farm Mechanization Level 49
2.4.4. Agricultural Commercialization: Level and Its Determinants 50
2.4.5. Production Efficiency and Its Determinants 52
2.4.6. Empirical review on Effects and Determinants of Crop Diversification 53
[Link]. Economic effects of crop diversification 53
[Link]. Review of determinants of crop diversification 54
2.5. Conceptual Framework of the Study 55
3. METHODOLOGY OF THE STUDY 59

xi
 xii

TABLE OF CONTENTS (Continued)

3.1. Description of the Study Area 59

3.2. Sampling Design and Sample Size Determination 61
3.3. Data Sources, Types and Methods of Collection 63
3.4. Methods of Data Analyses 64
3.4.1. Descriptive Statistics 64
3.4.2. Estimating the Farm Mechanization Level 64
3.4.3. Econometric Models 65
[Link]. Determinants of farm mechanization level: Tobit model 65
[Link]. Economic efficiency analysis 67
[Link]. Impact of farm mechanization on EE, yield, income and inputs use intensity 72
[Link]. Measurement and determinants of households’ crop commercialization 75
[Link]. Measurement and determinants of crop diversification 78
[Link]. Inter-linkage among farm mechanization, commercialization and diversification 80
3.4.4. Hypotheses and Definitions of Variables 82
[Link] Determinants of the level of farm mechanization 83
[Link]. Definition of variables for impact of farm mechanization level: AIPW 86
[Link]. Variables definition for economic efficiency-SFA model 87
[Link]. Variables definition for crop output commercialization: Heckman’s model 89
[Link]. Variables definition for determinants of crop diversification 93
4. RESULTS AND DISCUSSION 97
4.1. Descriptive Analysis 97
4.1.1. Demographic Characteristics of Sample Households 97
4.1.2. Sample households’ characteristics along mechanization level 98
4.1.3. Asset Ownership of the Sample Households 100
4.1.4. Types and Level of Farm Mechanization Technology Adoption 101
4.1.5. Major Constraints to Mechanize Agriculture in the Study Area 103
4.1.6. Crop Commercialization and Diversification 104
4.1.7. Socio-economic, Institutional Characteristics and Crop Commercialization 107
4.1.8. Level of Crop Diversification and Potential Reasons 110

xii
 xiii

TABLE OF CONTENTS (Continued)

4.2. Results of Econometric Models 112

4.2.1. Determinants of Farm Mechanization Level: Tobit Model Results 112
4.2.2. Economic Efficiency of Wheat and Barley Production 122
4.2.3. Impacts of Farm Mechanization on Input Use Intensity, Economic Efficiency, Income,
and Productivity of Households 123
[Link]. Post-estimation test for model validity 123
[Link]. Impact of farm mechanization on agricultural input use intensity 124
[Link]. Effect of farm mechanization on economic efficiency of wheat and barley producers
128
[Link]. Impact of farm mechanization on household income 130
[Link]. Farm mechanization impact on wheat and barley productivity 131
4.2.4. Determinants of Smallholder Farmers’ Crop Commercialization 132
[Link]. Determinants of crop output market participation 132
[Link]. Determinants of crop commercialization level 137
4.2.5. Determinants of Crop Diversification 140
4.2.6. Interlinkage among Farm Mechanization, Crop Commercialization and Crop
Diversification 146
[Link]. Results of OLS estimation 150
[Link]. Results of SUR Estimation 150
5. SUMMARY, CONCLUSIONS AND RECOMMENDATIONS 139
5.1. Summary 139
5.2. Conclusions and Recommendations 142
6. REFERENCES 149
7. APPENDICES 188
Appendix 1. Appendix Tables 188
Appendix 2: Stata Commands and Estimation Procedures 194
Appendix 3. Sample Households Survey Interview Schedule 198

xiii
 xiv

LIST OF TABLES

Table Page
Table 1. Distribution of the sample by zones, districts, and Kebeles 63
Table 2. Demographic and socioeconomic characteristics of sample households 98
Table 3. Socioeconomic variables along level of farm mechanization level 99
Table 4. Asset ownership and labor availability variables across sex of household head 100
Table 5. Farm mechanization technology adoption by sample districts 101
Table 6. Types and main reasons behind adopting farm mechanization technologies 102
Table 7. Land allocation and level of mechanization for wheat and barley production 103
Table 8. Major constraints in using farm mechanization technologies and their severity 104
Table 9. Production goals, sale types and commercialization status of sample households 106
Table 10. Equality test for socio-economic and institutional characteristics of respondents and
their participation in crop market 109
Table 11. Landholding by agro-ecological zones in the study area 110
Table 12. Tobit estimate and marginal effects of determinants of level of farm mechanization 120
Table 13. Economic and technical efficiency of wheat and barley producers (bc* estimated) 123
Table 14. Results of the treatment effects estimation for labor employment 126
Table 15. Results of the treatment effects estimation for chemical fertilizer application 127
Table 16. Results of the treatment effects estimation for agrochemicals (ln transformed) 128
Table 17. Results of the treatment effects estimation for wheat and barley economic efficiency
129
Table 18. Results of the treatment effects estimation for household income (ln transformed) 130
Table 19. Results of the treatment effects on wheat productivity (qt/ha): AIPW estimation 131
Table 20. Farm mechanization impact on barley productivity: AIPW estimation 132
Table 21. Heckman selection model two-step estimates the result of market participation 134
Table 22. Odds ratios for Ordered Logit Model estimation 142
Table 23. Marginal effects estimate for crop diversification level after ologit 145
Table 24. Models R-squares and Root MSE valued for SUR and OLS models 146
Table 25. Correlation matrix of residuals for SUR and OLS models 147

xiv
 xv

Table 26. Estimation of determinants of mechanization, commercialization and crop

diversification using OLS model 148
Table 27. Estimation of determinants of mechanization, commercialization and crop
diversification using SUR model 152

xv
 xvi

LIST OF FIGURES
Figure Page

Figure 1. Input-oriented technical, allocative and economic efficiencies 32

Figure 2. Technical and allocative efficiencies from an output orientation 33
Figure 3. Conceptual framework of the study 58
Figure 4. Geographical location of the study area (Arsi and West Arsi zones) 60
Figure 5. Boxplots of crop diversification by AEZs, land fragmentation and mechanization level
112
Figure 6. Histogram farm mechanization per household 119
Figure 7. Propensity score overlap for level of farm mechanization 124

xvi
 xvii

LIST OF TABLES IN APPENDIXES

Appendix Table Pages

Appendix Table 1. Household distribution across level of FMI, COCI, and CDI 188
Appendix Table 2. Crop types grown and mean farm land allocation (ha) 188
Appendix Table 3. Level of crop diversification index (CDI) 188
Appendix Table 4. Farmers’ perception on their crop production system 189
Appendix Table 5. Linktest result for Tobit model misspecification test 189
Appendix Table 6. Collinearity Diagnostics for commercialization explanatory variables 189
Appendix Table 7. Covariate balance summary for impact estimation 190
Appendix Table 8. Collinearity diagnostics for inputs used in barley production 190
Appendix Table 9. Collinearity diagnostics for input price in barley production 191
Appendix Table 10. Collinearity diagnostics for wheat production inputs 191
Appendix Table 11. Model fitness test (parallel regression assumption and collinearity diagnosis)
192
Appendix Table 12. Mean distribution for FMI, COCI and CDI across each level (percent) 192
Appendix Table 13. Breusch–Pagan/Cook–Weisberg test for heteroskedasticity of
commercialization studies 193
Appendix Table 14. Test for endogeneity of level of farm mechanization: 193
Appendix Table 15. Conversion factors to compute tropical livestock unit (TLU) 194

xvii
 xviii

Farm Mechanization, Commercialization and Crop Diversification in Central

and Southern Oromia Region, Ethiopia

ABSTRACT

In Ethiopia, the agricultural sector contributes the largest employment opportunities (70%) and
is the major source of export earnings (90% of export value). Despite efforts made to modernize,
mechanize, commercialize, and transform the Ethiopian agricultural sector from subsistence to
production of diversified and high value crops, the sector’s performance has been lower. In light
of the problems and the research gaps identified, this study seeks to address and generate
information on the level of farm mechanization, factors affecting the level of farm mechanization,
commercialization, and crop diversification, and impacts of farm mechanization on input
utilization, economic efficiency (EE), and household income in the central and southern parts of
Oromia region, Ethiopia. A total of 390 sample farm households were randomly drawn from 8
Kebeles in four districts, through a multi-stage sampling technique. Descriptive statistics and
econometric models including Tobit, Stochastic Frontier Analysis (SFA), Augmented Inverse
Probability Weight (AIPW) model, Heckman’s two-stage, Ordered Logit, and Seemingly
Unrelated Regressions (SUR) models were employed to analyze the data. Results of this study
show that the mean farm mechanization index was 35.14%. The Tobit model result pointed out
that farm experience, education, off/non-farm activities, cultivated landholding, market
participation and access to road were positively and significantly affect the level of farm
mechanization. However, sex, zonal location, livestock possession, land fragmentation, distance
to mechanization centers, and social capital were negatively and significantly affecting the level
of farm mechanization. The stochastic frontier analysis results also indicated that the mean
technical and economic efficiency of wheat farmers were 94 and 72% respectively whereas the
barley producers’ mean technical and economic efficiency were 73 and 91% respectively. The
relatively higher technical efficiency (TE) and EE scores could be due to enterprise
specialization and government attention to such crops in the area. The result of the AIPW model
revealed that higher level of farm mechanization has positive effect on the economic efficiency of
wheat and barley producers, amount of chemical fertilizer applied, household income, and wheat
farm productivity. On the other hand, farm mechanization has a significant negative effect on the

xviii
 xix

amount of farm labor employed and agrochemical applied. The Heckman’s two-stages selection
model result showed that the decision to participate in crop marketing was significantly and
negatively affected by age, market distance, and instrumental variable for the level of farm
mechanization. Contrary to this, it was affected significantly and positively by livestock size,
agroecological zone, ownership of equines, annual income, and access to market information.
Second tier result of the model also showed that intensity of commercialization was positively
and significantly affected by annual income, access to market information, and number of oxen;
while it was negatively and significantly affected by cultivated land, crop diversification level,
and market distance. The ordered Logit model result indicated that the crop diversification was
positively and significantly affected by variables like family size, farmland, level of farm
mechanization, and land fragmentation but negatively affected by agro-ecological location. The
SUR model results revealed that there is a positive and significant causation between farm
mechanization and crop output commercialization with coefficient of 12.45% while there was
also positive and significant causation between farm mechanization level and crop
diversification index (SDI). However, crop commercialization and diversification had negative
causation with a coefficient of 11%. Hence, commercialization and crop diversification were
negatively affecting each other. In general land consolidation, availing infrastructural facilities
and facilitating adult education and short-term trainings are important to enhance the level of
farm mechanization level in the study area. Improving the resource endowment, and minimizing
transaction costs by improving access to market centers, market information, and means of
transportation can further enhance commercialization. Improving access to farm mechanization
and enhancing productivity in highland areas are also issues that shall get policy and
development practitioners’ focus. Furthermore, increasing the level of farm mechanization by
availing the technologies can maximize the households’ welfare through reducing inputs cost
and increasing efficiency and household income. Finally, from the result of the inter-linkage
among farm mechanization, crop commercialization and diversification it can be recommended
that improved level of farm mechanization can promote level of crop commercialization and
crop diversification.

Keywords: Crop Commercialization; Crop Diversification; Economic Efficiency; Farm

Mechanization; Impact; Land Fragmentation.

xix
 1

1. INTRODUCTION

1.1. Background of the Study

Within the last 50 years, world agricultural production was grown threefold while areas of
production expanded by only 11% (Chauvin et al., 2012; Pretty et al., 2011). But productivity
in African agriculture is only around 56% of world average while more than 50% of world
uncultivated arable land is found in Africa. This low productivity of African agriculture in
general and that of Sub-Saharan African (SSA) in particular, is attributed to low application of
science and technologies which is the main source of agricultural productivity growth (OECD,
2012). Despite African leaders’ Maputo Declaration to spend at least 10% of their national
budget to support the agriculture (Diao et al., 2016), this did not happen in the continent and
important inputs to transform and modernize the sector such as farm mechanization and other
technologies, were neglected in the region for long time (Liu and Tian, 2009; Sims et al.,
2016). However, recently the demand for farm mechanization in Africa as a whole and in SSA
in particular is increasing and there are policy changes towards promoting mechanization
(Diao et al., 2016; Daum, 2020).

Ethiopian agricultural sector continued to grow (5.30%) in both Growth and Transformation
Plans (GTP I and GTP II) between 2011 and 2020. The sector continued as a backbone of the
national economy by contributing around 33% of the national GDP. Moreover, it is employing
around 70% of national workforce, and it is a major source of exports which accounts for
around 90% of export value (ITA, 2022). Furthermore, the average growth rate of agricultural
sector is well above population growth rate and this contributed to reduction of population
below poverty line from 30.4% in 2015 to 25.6% in 2020 (WFB, 2022; NBE, 2020). This
considerable and continuous growth of the sector was attributed to factors such as use of
modern farm inputs, rapid expansion of arable land, increased labor productivity,
government’s investment in extension system and an improved road network (Pauw, 2017),
which is mainly due to intensification and area expansion similar to that of other SSA
 2

countries (Guush et al., 2017). However, there are also efforts to sustainable farm
mechanization in the country (Malabo, 2018).

Generally, increase in real rural wage was among the most important factors to promote use of
farm mechanization (Pingali et al., 1987) and among the drivers there are two phenomena
realized in Ethiopian economy recently: the rise in rural wages and rise in cost of animal
traction. The rise in wage is as a result of schooling and massive migration in (urban) and
outside the country (which increased rural wage by around 54%) (Fantu et al., 2016) or an
increase in living cost could also be a reason. On the other hands, cost of animal traction is
fueled by increase in cost of maintaining traction animals due to scarce grazing land.

The use of tractors for tillage and engine-driven threshers and combine harvesters were
common in the central highlands of Arsi and Bale since the 1950’s of emperor’s regime. The
first and second five-year development plans of the Imperial government of Ethiopia (1957-
1968) were favoring large scale farms to import duty free machineries, spare parts and
subsidized fuels (Stahl, 1973; Cohen, 1987). These privileges encouraged the large private
farms (usually owned by landlords and foreigners) to increase their profit margins and
commercialization was promoted in the area. However, the action that gave attention to only
large-scale farms led to massive tenant eviction around Arsi and Shewa which end up with
conflicts and that in turn stagnated mechanization process of the country as a whole (Stahl,
1973; Cohen, 1987).

The military socialist government of Dergue nationalized all large private farms of emperor’s
period and transferred them to state farms while trying to maintain their level of
mechanization but private farm mechanization was not encouraged (Mohammed et al., 2000;
Yared and Moti, 2019). After the overthrow of the Dergue government, small scale farms in
particular and the agriculture sector as a whole has been given better attention from policy
makers (FACASI, 2016; Yared and Moti, 2019). As a result, even though the attention given
was not equivalent with that of other inputs like fertilizer and improved seed, small-scale
farms mechanization was included in the policy agenda of the Ethiopian government
(FACASI, 2016).
 3

Enterprise specialization in the study area mainly targeted wheat which was principally
promoted by Emperor Haile-Selassie’s policy to make the country food self-sufficient
following the 1950s’ national food shortage (Cohen, 1987) and the study area are designated
as the ‘wheat-belt’ of the country (Guush, 2014; Guush et al., 2017). According to CSA
(2021a), the three zones namely Arsi, west Arsi and Bale only accounts for 53% of the total
wheat grown area and 56% of the total wheat production of Oromia region. Level of farm
mechanization in Ethiopia is very low compared to SSA and other African countries’ average
(Seife, 2022). Studies shown that 18% of African countries are using tractor-mounted
machinery (Kirui, 2019) but only 9% of Ethiopian farmers are using either tractor, combine
harvesters or threshers for their farm operations with high variation among regions while
Oromia region is the leading with 11% of households using the technologies (Guush et al.,
2016).

Pingali and Rosegrant (1995) defined farm commercialization as a process that is

accompanied by economic growth, urbanization and withdrawal of labor from the agricultural
sector. Therefore, farm mechanization is a catalyst for both agricultural commercialization and
enterprise specialization. According to Guush et al. (2017) and Guush (2014) there are
reduced labor requirement and increase in yield due to mechanization (specifically combine
harvesting) in some parts of Ethiopian agriculture.

In agriculture dependent economy, crop diversification is one way of developing a resilient

agricultural production system to the changing climate which is applied as one practice of
climate smart agriculture to improve farmers’ livelihoods (Njeru, 2013; Makate et al., 2016). It
is also one of the most ecologically feasible, cost-effective and rational ways of reducing
uncertainties in agriculture, especially among small-scale farmers. Furthermore, crop
diversification helps to maintain biodiversity on the farm and increases its resilience (Njeru,
2013). Crop diversification has also positively impacted different nutrition outcome indicators
that include child growth, household nutrient production gaps, diet quality, food intake, and
diet diversity (Wondimagegn, 2021). Crop diversification can also enable smallholder farmers
to manage prices and production risks (Di Falco and Chavas, 2009). Farm level crop
 4

productivity can also be improved by using crop diversification especially in moisture-stressed

and fragile ecology and it can also contribute to efforts to alleviate food insecurity (Di Falco
and Chavas, 2009; Mango et al., 2018). Winters et al. (2006) summarized farmers’ motives for
crop diversification into three as: i) managing risk, ii) adopting to heterogeneous agro-
ecological production conditions; and iii) satisfy market demands.

Urbanization and the withdrawal of labor from the agricultural sector increase the demand for
farm mechanization and food in urban areas. The increase in farm mechanization boosts
production and productivity, and the demand for food in urban areas lead to
commercialization. On the other hands, since mechanizing all enterprises is difficult at a time,
usually crops which are easy for mechanization are selected for farm mechanization and this
encourages farmers to shift towards those mechanized crops and that promotes less crop
diversification (Sims et al., 2016; Malabo, 2018; Guush et al., 2017). The argument of
agricultural commercialization advocates is that specialization leads to commercialization
(Moti et al., 2009; Juma, 2010; Shrestha, 2011). The government of Ethiopia (GoE) is also
currently promoting smallholder farmers crop commercialization to improve their livelihood
sustainably through cluster farming, optimal resource allocation and mobilization thereby
stimulate them to improve their farm output (Jony, 2011). Empirical studies of these issues at
the national level in general, and in the study area in particular, is limited.

1.2. Statement of the Problem

Farm mechanization has multi-dimensional impacts in agricultural production like: level of

input use, economic efficiency (EE) and life quality of farming community by reducing work
drudgeries (Amirani, 2001; Houmy et al., 2013). Its introduction can also increase the
productivity of the farming system significantly especially where agriculture is more
dominated by traditional technologies. Mechanization reduces production cost thereby
increasing profit and reducing food costs (Cossar, 2019; Bello, 2012). Depending on the
structure of the national economy, farm mechanization can create unemployment or release
labor for productive work outside the agricultural sector that can be explained based on
induced innovation theory of Hayami and Ruttan (1971).
 5

Farm mechanization’s advantages (specifically high and intermediate mechanization

implements which include tractor, combine harvesters and threshers) can be explained in terms
of acquired or desired work quality and maintained timeliness of work accomplishment which
can contribute to production quality (Bello, 2012; Houmy et al., 2013). Level of
industrialization and the development of agricultural economy of a nation are also directly
related to farm mechanization (Singh, 2006). As a result, it is needed to assess the drivers and
estimate impacts of farm mechanization on productivity and other economic variables (Singh,
2006; Hormozi et al., 2012). Cognizant of farm mechanization’s multi-dimensional impact
and its more than 70 years history in Ethiopia, there is dearth of study on such issues.
Moreover, drivers and constraints of farm mechanization are not quantified.

To the best of researcher’s knowledge, the only quantitative studies conducted on this area
were Guush et al. (2016) and Guush et al. (2017). These studies were limited to comparing
modern input utilization level and land rental difference between tractor adopters and non-
adopters. But issues such as drivers and constraints; impacts of farm mechanization; the
interaction effects of adoption of different mechanization technologies on income, farm
productivity, economic efficiency and intensity of input use were not covered. Moreover, even
though mixed use of draught animal and tractors is common (Guush, 2014), all tractor users
were considered as “farm mechanization adopters” irrespective of their level of adoption.
Furthermore, due to the better attention given to mechanized farming by Oromia regional state
in the last few years, there is significant progress regarding supply of machineries in the study
area. Hence, good policy directions that can enhance availability of farm mechanization may
have positive impact on adoption of technologies and its associated benefits. In general, these
issues necessitated further studies on mechanization in the study area.

On the other hand, the GoE is also promoting agricultural commercialization through different
means with the main objective of transforming the subsistence supply-led agriculture sector to
organized, high-tech, safe, and demand-driven agriculture. To achieve this goal, the GoE is
expanding the integrated agro-industrial parks and promoting cluster-based mechanization
supported production system and this action is believed to accelerate the commercialization of
 6

smallholder agriculture and support the structural change of the national economy as a whole
(Brasesco et al., 2019). The expansion of agro-industrial parks, such as Bulbula and Dodola, in
and around the study areas are examples of such actions. However, the level and determinant
of smallholders’ commercialization were not studied in the study areas.

Crop diversification increases agricultural production, enhances nutrition security, and support
sustainable agricultural transformation. It was also recognized as a strategy to improve
nutrition and climate risk coping mechanism by the GoE (Wondimagegn, 2021). In spite of the
national level aspiration on nutritional security and increase in agricultural production through
the use of crop diversification, the study area is specialized with cereal crops and known as the
wheat belt of the country, which might create favorable conditions for agricultural
mechanization (Guush et al., 2017). Moreover, farm commercialization has also an impact on
crop diversification, and is triggered by different factors (Lone, 2013). Therefore, level of crop
diversification, its determinants and linkage with mechanization at household level needs to be
studied.

Empirical findings also revealed that there is interlinkage among the three economic variables
where one affects the other and vis-versa. For instance, the finding in a province of China
shows that the level of mechanization has a significant positive impact on the cost, output
value, income and rate of return in all crop types (Peng et al., 2022). The integrated use of
irrigation technologies and other farm mechanization technologies like tractor and combine
harvesters, increases cropping intensity and thereby improve production and
commercialization level of farms (Tan, 1981). For instance, Adetule et al. (2021) analyzed the
nexus of agricultural commercialization among farm households in Nigeria and they found
that farm mechanization has significant impact on crop commercialization level. It was also
reported that lack of farm machinery is having a significant impact on farm productivity in
Kyrgyz Republic (Guadagni and Fileccia, 2009). Commercialization of crop can also have
significant impact on household’s welfare and affects the income and technological adoption
of households and crop diversification at national level. Hence, the study of the interlinkages
among these three variables, farm mechanization, crop commercialization and crop
diversification are very crucial for policy making and development interventions.
 7

In a nutshell, generation of information on level of farm mechanization, its driving factors, and
impacts; level of commercialization and its determinants; level of crop-diversification and its
determinants and the linkage among the three agricultural phenomena (level of agricultural
mechanization, commercialization and crop-diversification) is vital for policy making and
development planning purposes and needs to be studied. Hence, based on the problem stated
above, this study is intended to address the following research questions:
• What is the level of farm mechanization and its determinants at household level?
• What is the effect of agricultural mechanization on farm labor employment, economic
efficiency of farmers, household income and intensity of farm inputs use?
• What is the level of crop output commercialization and what are its determinants at
household level?
• What is the level of crop diversification and what are its determinants at household
level?
• What is the inter-linkage among farm mechanization, commercialization and crop
diversification?

1.3. Objectives of the Study

In general, this study focused on farm mechanization of major crops, commercialization, and
crop diversification and their inter-linkages in the central and southern parts of Oromia
regional state while the specific objectives are:

• To estimate the farm mechanization level and identify its determinants in the study
area for major crops;
• To estimate the level of economic efficiency of farming households;
• To estimate the impacts of farm mechanization on farm labor employment,
productivity (yield), economic efficiency, household income and intensity of farm
inputs use for the major crops of the area;
• To measure and identify the determinants of crop commercialization;
• To estimate the crop diversification level and identify its determinants; and
 8

• To analyze the inter-linkage among farm mechanization, commercialization and crop

diversification in the study area.

1.4. Scope and Limitations of the Study

The scope of a research can be explained in terms of spatial, technical, methodological, and
theoretical aspects. Accordingly, the objectives of this research are limited to assessing the
level of farm mechanization and its impact on some agricultural activities and analyze farm
commercialization level and its determinants while the third main theme is to estimate crop
diversification level and identify its determinants. Only High and Intermediate Mechanization
Implements (HIMIs) technologies following ATA (2014) that include mainly tractors,
combine harvesters and engine-driven threshers were considered in this study. The
geographical coverage was also limited to Arsi and West Arsi zones due to time and financial
constraints with cross-sectional data. Due to the nature of data, which is cross sectional, the
overtime improvement of the situation of farm mechanization, crop commercialization and
diversification nature were not studied. Community level impacts of farm mechanization like
deforestation, erosion and others were not covered in the study.

1.5. Significance of the Study

Agricultural research in the SSA region as a whole has given more attention to seeds, nutrients
and water than mechanization (Baudron et al., 2019). The tradeoffs between positive impacts
of farm mechanization and its social costs like labor displacement were not studied so far in
detail. Furthermore, issues of farm mechanization like drivers and its impacts in general in
Ethiopia for evidence-based policy decisions are scarce. Hence, knowing the demographic,
socioeconomic and institutional factors that affects the use of farm mechanization will help to
design policy best fits to the households. Studies on commercialization of smallholders, and
level of crop-diversification and their determinants were also scarce. Therefore, this study can
be used as a tool to formulate a policy that can improve households’ commercial orientation. It
can also be used to provide information for development practitioners and policy makers to
 9

make well informed policy decisions regarding farm mechanization, commercialization, and
crop diversification.

1.6. Organization of the Dissertation

The dissertation is organized in to five chapters. The first chapter introduces the subject matter
of study and problem statement justifying the importance of the study. Under this chapter,
research questions, objectives, and significance of the study were briefly presented. The
second chapter deals with literature review which presents the concept and definition,
discusses the theoretical and analytical framework of the research, related empirical literature
review and finally, presents conceptual framework of the study. Chapter three is about the
methodology of the research describing the study area, data collection procedures and data
analysis methods. Results of the study are presented and discussed in the fourth chapter.
Finally, the fifth chapter summarizes the major findings of the study followed by conclusion
and recommendations based on the results of the study.
 10

2. LITERATURE REVIEW

This chapter of the document reviews the theoretical, methodological, analytical frameworks
and empirical studies conducted on smallholder farmers’ farm mechanization level, its multi-
dimensional impact and effects, crop commercialization, and crop diversification. The
theoretical review provides the background that offers justification of the research problem
and used to construct theoretical framework to be used in the study. The analytical framework
leads to a specific method and deemed to have clarity on the analytical techniques and
procedures. The empirical literature provides evidence from past studies related to the subject
and the gaps to be filled.

2.1. Definitions of Basic Related Terms

2.1.1. Farm Mechanization and Its Adoption

The term agricultural mechanization and farm mechanization are used inter-changeably in this
document. Agricultural mechanization has been defined in different ways but the central
element of mechanization which is “increasing agricultural production through land and labor
productivity improvement and reducing human drudgery and increasing leisure time” is
unchanged. The use of agricultural mechanization is intended to enhance the effectiveness and
productivity of human labor (FAO and UNIDO, 2008). The Malabo Montpellier Panel (2018)
defined agricultural mechanization as a process by which human labor in the process of
agricultural production along the entire value chain from production to marketing is replaced
by other sources of energy, such as animal power, fossil energy or renewable energy. It can
also be defined as the introduction and use of any mechanical tools, implements and
machinery as agricultural production inputs to make production activities more efficient
(Clarke, 2000; Loon et al., 2020). In this sense, agricultural mechanization/farm
mechanization refers to tools, implements and machineries that are applicable for both crop
and livestock production.

Mechanization can be classified into three based on their power sources: human power based,
animal power based, and mechanical power-based mechanization (Malabo, 2018). ATA
 11

(2014) also classified agricultural mechanization technologies in to two main categories based
on their power sources and affordability. The first category is human or animal powered farm
implements called ‘Low Mechanization Implements (LMIs)’ while the second category
includes larger implements which require additional power from external sources like engine,
electric motors or tractors and these are referred as ‘High and Intermediate Mechanization
Implements (HIMIs)’.

Farm mechanization: Following ATA (2014), farm mechanization is defined as the use of
second category (HIMIs) for this study. Therefore, farm mechanization in this study included
tillage, planting, cultivation and harvesting activities by using tractors, walking behind
mowers or combine harvester, and stationary engine driven threshers.

Level of mechanization (LOM): Nowacki (1974) developed mechanization index based on

the matrix of use of animate and mechanical energy inputs expressing as the ratio of machine
energy to the sum of machine, animal and human energy. Accordingly, LOM in this study is
defined in terms of mechanization index (MI) and expressed as the ratio of cost of work done
using mechanization technology to the sum of the costs of work done using mechanization,
animal and human labor per hectare (Ebenehi and Lawal, 2019).

Technology adoption: The study of adoption of innovation was started by pioneer work of
Griliches (1957). Adoption of agricultural innovations is essential in ensuring food security
and poverty reduction, by potentially increasing the income of farming households as well as
reducing the market price of staple foods as the contribution of new technology to economic
growth can only be realized when and if the new technology is widely diffused and used (de
Janvry and Sadoulet, 2008). Since adoption of new agricultural technology is important to the
realization of agriculture’s full potential, study of the adoption rate/intensity of agricultural
innovation and technology has drawn attentions of the researchers and policymakers over the
years. The adoption study also contains adoption rate/intensity and factors that
motivate/promote or hinder the adoption of that technology. In order to adopt a technology, it
has to fulfill at least three conditions: the availability of sufficient information, the existence of
 12

a favorable attitude towards the innovation, and the physical availability of the
innovation/technology (Yapa and Mayfield, 2006).

Adoption at individual level was defined differently for divisible and non-divisible
innovations. Final adoption of divisible innovation (like high yielding varieties, chemical
fertilizers and etc.) at the level of the individual farmer is implied as the degree of use of a new
technology in long-run equilibrium when the farmer has full information about the new
technology and its’ potential (Feder et al., 1985). To measure the extent of adoption at the
farm level in a given period for non-divisible innovations like farm mechanization
technologies (tractor for instance), Feder et al. (1985) suggested to use necessarily a
dichotomous (use/no use) method but, in the aggregate, the measure becomes continuous (e.g.,
the percentage of farmers using tractors). However, other authors used the same approach to
that of the former definition assuming mechanization as divisible innovation by transforming
use of farm mechanization into monetary terms at regional level (Singh, 2006). Similarly, this
study adopted this method to measure adoption at household level.

2.1.2. Impacts of Adoption of Farm Mechanization

Mechanization of farming is expected to play a key role in bringing about growth in

agriculture in general. Level of farm mechanization is assumed to have impacts on level or
intensity and types of agricultural inputs like improved seed, chemical fertilizers and labor.
According to Pingali (2007), the impact of switching to mechanical technologies for
agricultural operations on productivity can be measured in terms of changes in yields, labor
savings, area expansion (in terms of increases in cropping intensities), and quality
enhancement of the marketed output.

The timeliness of operations has assumed greater significance in obtaining optimal yields from
different crops, which has been possible by way of mechanization (Zhang and Hanlin, 2014;
Sims et al., 2016). Furthermore, the quality and precision of the operations are equally
significant for realizing higher yields. Hence, the various operations such as land leveling,
irrigation, sowing and planting, fertilizers applications, plant protection, harvesting and
 13

threshing need a high degree of precision to increase the efficiency of the inputs and reduce
the losses. On the other hand, cropping intensity is determined by the time taken to perform
sequence of operations which can be improved by the use of farm mechanization technologies.

Farm mechanization is also used as land augmenting technology especially when there is land
shortage (Binswanger, 1978) and this is more facilitated by the use of irrigation mechanization
technologies (Dey and Haq, 2009). However, high cropping intensity may lead to high soil
degradation and higher intensity of input use to compensate the yield losses due to nutrient
loss (Dey and Haq, 2009). Higher productivity of land and labor is another factor, which
clearly justifies farm mechanization. Mechanization increases output per hour and at the same
time reduces total labor required. Mechanization also increases commercialization level of
farm households as it increases agricultural production which increases the surplus. The action
of farm mechanization also increases farm profitability as it reduces production cost since it
saves both human and animal labors. Furthermore, mechanization increases scale of
production and economies of scope which have positive effects on reducing per unit cost of
production (Dey and Haq, 2009; Takeshima et al., 2020).

Other study findings attested that land productivity, total agricultural production and
profitability are affected positively by the use of farm mechanization due to timeliness of
operation. The timeliness of operation is more important specially during the planting and
harvesting period which has short window time and better quality of work done and more
efficient utilization of inputs (Singh and Singh, 1980; Daum et al., 2020; Ahmed et al., 2020;
Caunedo and Kala, 2021; Ramya and Muruganandham, 2016; Baudron et al., 2019). In
Ethiopia it was indicated that use of tractor has no effect on yield while that of combine
harvester has positive and significant impact on yield (Guush et al., 2017).

The use of farm mechanization also positively affects the intensity of agricultural inputs like
improved seed, chemical fertilizers and agrochemicals. Most authors even argued that the
higher production and productivity were not directly attributed to the application of
agricultural mechanization technologies (especially tractor) but to the use of higher level of
such farm inputs, and other indirect effects like timeliness of operation, better quality of work
 14

done and more efficient utilization of those inputs due to farm mechanization (Singh and
Singh, 1972; Binswanger, 1978; Herdt, 1983; Ramya and Muruganandham, 2016).

2.1.3. Agricultural Commercialization

Agricultural commercialization can be viewed in many forms and defined in different ways. It
can be considered as static form or as a dynamic process over time. Considering the static
form of smallholder farm commercialization, it could be seen as the measure of the strength of
the linkage between farm households and markets at a given point in time. This household-to-
market linkage could relate to output or input markets either in selling, buying or both
including labor (Moti et al., 2009). Considering farm commercialization as dynamic process, it
could be seen as a process at what speed the proportion of outputs sold and inputs purchased
are changing over time at household level. Based on this premises agricultural
commercialization occurs when agricultural enterprises and/or the agricultural sector as a
whole rely increasingly on the market for the sale of produce and for the acquisition of
production inputs (Poulton, 2017). Generally, three approaches can be followed to define farm
commercialization: in terms of type of crops grown, participation in input and output markets
and goals of production.

Based on types of crops grown a farm household is considered as commercialized if it

produced a significant amount of cash commodities, allocating a proportion of its resources to
marketable commodities, or selling a considerable proportion of its agricultural outputs
(Immink and Alarcon 1993; Strasberg et al., 1999). However, defining commercialization
based on types of crops grown may be misleading as the farmers are not marketing only the
so-called cash commodities and there is high probability of marketing traditionally classified
as food crops especially during the transitional period from subsistence to semi-
commercialized farming and also there may be considerable amount of home consumption of
cash-crops (Pingali and Rosegrant, 1995; von Braun, 1995; Dawit et al., 2006).

From the inputs side, commercialization refers to valuing both traded and non-traded inputs
like family labor and manure in terms of their market prices while from the output side it is
defined as the proportion of agricultural production that is marketed (Pingali, 1997; Timmer,
 15

1997). The degree of participation of a household in output market with much focus on cash
incomes is widely used to define agricultural commercialization (Leavy and Poulton, 2007).
Agricultural commercialization goes beyond producing and allocating considerable proportion
of farm product for market and rather it refers to the choice of what to produce and decision
about the input use based on the principles of profit maximization both for staple cereals and
high-valued commercial crops (Pingali and Rosegrant, 1995; von Braun, 1995).

Considering the dynamic process, Pingali and Rosegrant (1995) argue that as degree of
participation of farmers in commercialization increases, it passes through process of
commercialization from subsistence system to semi-commercial system and finally to
commercial system. Then, the objective of production changes from food self-sufficiency to
surplus production and finally to profit maximization. Inputs are non-traded and household
generated in first stage and it is mixed of traded and non-traded in second stage and finally in
commercial system inputs are predominantly traded. Based on this argument, Sokoni (2008)
defined agricultural commercialization as “a process involving the transformation from
production for subsistence to production for the market”. Hence, in this study the dynamic
process of commercialization was adapted.

2.1.4. Production Efficiency and Productivity

Terms such as efficiency, productivity, technology growth, and economic growth are very
widely used and even interchangeably in economics. However, they are not equivalent terms
(Jema, 2008). Among others, efficiency is the most widely employed and it is measured by
comparing the observed output against the feasible (frontier) output. The difference between
productivity and efficiency is that productivity is the ratio of total output produced to the total
inputs used in production while efficiency is the ratio of the value of outputs produced to the
value of inputs used. By saying ‘efficiency’ we always refer to economic efficiency (EE) and
Farrell (1957) was the pioneer economist to introduce the method to decompose EE into
technical efficiency (TE) and allocative efficiency (AE).

Technical Efficiency (TE): It involves producing farm products with the best combination of
resources or inputs. It is expressed as attaining the maximum level of output from a given level
 16

of inputs or, alternatively, a given output with the use of minimum inputs. TE is measured by
taking the physical relationship between inputs and outputs into accounts. Similarly, TE can be
defined as producing the maximum amount of output from a given amount of input or
alternatively, producing a given output with minimum input quantities (Lovell, 1993; Hensher,
2001; Hollingsworth, 2008).

Allocative Efficiency (AE): Given the input prices that prevail in input selection, AE involves
selecting that mix of inputs such as labor and capital that produces a given quantity of output
at a minimum cost (Coelli et al., 2005). AE measures whether the maximum output production
is attained by the use of different proportion of factors of production with a given market
prices (Farrell, 1957). In AE in addition to the process of production (how it is produced), the
proportion of outputs and inputs used in the processes is important (Hensher, 2001).

Economic Efficiency (EE): It implies an economic state in which every resource is optimally
allocated to serve each individual or entity in the best way while minimizing waste and
inefficiency. The concept of EE or overall efficiency was founded in seminal work of Farell
(1957). EE is the product of TE and AE (TE*AE) and the value always lies between 0 and 1 as
both TE and AE are lying between 0 and 1. Zero means no production while one implies the
most efficient producer.

Productivity: Productivity is defined as output per inputs used. In this study, productivity is
defined as the yield (per hectare output) of the crops under study and it does not refer to the
broad productivity definition of the total output per total factor of production.

2.1.5. Crop Diversification

Crop diversification is considered as major economic decision that has a strong bearing on the
farmer’s income level and food security (Pope and Prescott 1980). Conceptually,
diversification is considered as the re-allocation of some of a farm’s productive resources,
such as land, capital, labor, and farm equipment, into new farm activities. Crop diversification
is usually viewed as a shift from traditionally grown, less profitable crops to newer, more
 17

profitable crops (Pal and Kar, 2012). Diversification can take place in two ways either through
area augmentation or by crop substitution. It is also a strategy that is used to maximize the use
of land, water, and other resources for the overall agricultural development in a country.

For the purpose of this study, crop diversification means producing a variety of crops
involving intensity of competition amongst field crops for arable or cultivable land (Pal and
Kar, 2012). Therefore, crop diversification is a measure of portfolio of crop enterprises grown
by a household.

2.2. Theoretical Perspectives

2.2.1. Agricultural Development, Induced Innovation Theory and Farm Mechanization

The problem that earliest economists faced was what interrelations population growth and
food production have. The two entirely different approaches to answer this question are: to ask
“what is the effect of agricultural conditions on population demography i.e., putting population
as dependent variable of agricultural conditions” or to put agricultural conditions as dependent
and ask “how the change in population affects agricultural production and productivity”. The
Malthusian theory is based on the first approach and tried to explain how population growth is
checked by agricultural conditions or production (Malthus, 1798) while the next theory
developed by Boserup (1965) stated population growth as autonomous factor that forces
agricultural intensification and promotes the whole economic and sociological development.

The pre-dualistic development economics theory considered agriculture as obstacles, systems

which had to be reduced and eventually eliminated if modernization was to succeed. However,
the dualistic theory of development model in long-run growth process model emphasized that
for the growth in industrialization to be sustainable, it must be balanced with agricultural
productivity and unless and otherwise the development of a modern manufacturing sector
would be limited (Self and Grabowski, 2007). The dualistic development theories constructed
by Lewis (1954), Fei and Ranis (1961) and Jorgenson (1961), classified the less developed
economy into traditional and modern sector. According to them, the modern sector was driven
by profit maximization and the accumulation of physical capital while the traditional sector
 18

was subsistence oriented and dominated by peasant agricultural production and characterized
by output sharing mechanisms rather than profit maximization. It further presumed that the
traditional sector is labor abundant and as the surplus labor is withdrawn to modern
industrialized sector, the primary sector will be stagnated. Building on this theory, Johnston
and Mellor (1961) argued that agriculture supplied the labor necessary to the modern sector
firms, the food necessary to feed that labor and serve as a market for the produce of the
modern sector.

Hayami and Ruttan (1971) introduced induced innovation theory which states that whenever
there is a shift in relative factor prices, it will initiate the technological development or
improvement towards the replacement of the scarce factor by relatively abundant factor and
that will make the farmer to operate at socially optimal production cost. Development of
agricultural technologies and increase in agricultural productivity are keys in economic
transformation (Gollin et al., 2002; Olsson and Hibbs, 2005) and reducing poverty (Lipton,
1977; Datt and Ravallion, 1998; Ravallion and Datt, 1999; Mellor, 2001; Thirtle et al., 2003).

Generally, there are two ways of explaining the machine-labor substitution process in
agricultural sector. The first one is when technical change in agriculture caused mechanization
to become introduced and labor become unwanted in agriculture. This process of technology
introduction causes labor displacement and creates unemployment higher social cost. In this
case the real mechanization price is high due to the relative low farm wage rates. In most cases
this kind of mechanization especially in less developed countries happened when
mechanization process is subsidized. Literatures on African and Asian agriculture during
1960s and 70s showed that there was government subsidized tractorization but discontinued
due to huge labor displacement. The subsidized mechanization process in Ethiopia during the
same period also falsely reduced machine cost relative to human and animal labor and the
process caused tenant eviction and labor displacement (Henock, 1972; Holmberg, 1972; Stahl,
1973; Cohen, 1987). The second explanation is that when labor exit is led by market
phenomena: i.e., labor market outside the agricultural sector is attractive and there is voluntary
exit of the labor from agriculture as there is high opportunity and advantages to leave the
sector (Kislev and Peterson, 1981).
 19

According to induced innovation theory of Hick’s (1932) “The Theory of Wages”, any
rational competitive firm always strive to substitute a less expensive and abundant factor for
more expensive factor. According to this theory of induced innovation, whenever there is an
exit of labor from agricultural sector due to attractive wage rate in industrial sector, labor wage
in the sector will increase and this will force the farmers to search for alternative factor which
is less expensive. As a result, the research and development will work on technologies which
are more labor-saving and farmers are induced to use those technologies which have low
machine-labor factor price ratio (Hayami and Ruttan, 1971). The hypothesis of these authors
was that technical change is guided along an efficient path by price signals in the market;
when there is no distortion of the market implies that the prices efficiently reflect changes in
the demand and supply of products and factors and there is an effective interaction among
farmers, public research institutions, and private agricultural supply firms.

According to Hayami and Ruttan (1971) not all mechanical innovations are necessarily
motivated by labor saving incentives which means that innovation will always follow price
signals in such a way that it can solve the problem of the sector. For instance, when the land-
labor ratio is low innovation will direct to invent land augmenting technologies and when
there is low machine-labor price ratio, labor-augmented technologies will be invented. The
case of horse-plowing and irrigation technologies are an example of land augmented
inventions which were introduced to increase production through improvement of productivity
and production intensity respectively. Binswanger (1978) also agreed that introduction of
agricultural mechanization technologies into the sector is based on the relative factor price
ratio.

Agricultural intensification is defined as the increased application of labor and other inputs per
unit of land (inputs intensification) and more frequent cropping of land through reducing
fallow periods (land intensification) and it is induced by increased population pressure and
rising demand for agricultural products (Boserup, 1965; Ruthenberg, 1980). According to the
theory of long-term evolution of farming system by Ruthenberg (1980) the intensified use of
lands which are being fallowed and even used for forest and other purposes was also included
 20

as agricultural intensification while most agricultural economists considered only intensive use
of inputs as agricultural intensification whereas the act of putting more additional lands under
cultivation rather referred as agricultural extensification (Tachibana et al., 2001).

Agricultural mechanization is also one of the agricultural inputs that help in both
intensification and extensification of agriculture production (Diao et al., 2016). According to
Hayami and Ruttan’s (1970; 1985) induced technological change/innovation theory, the
development and application of new technology is endogenous to the economic system
(Ruttan, 2002). Therefore, one can assess the demand for mechanization technology as any
adoption process of agricultural technology. The considered agricultural technology
innovation and adoption as a continuous sequence initiated to save the production limiting
factor (usually land or labor) as the relative scarcity of land or labor endowment is reflected in
the change in their relative prices.

Developing countries developed agricultural mechanization strategies with the ultimate

objectives of increasing the income of farm households and create positive dynamics and
opportunities for economic growth in rural areas through increasing the efficiency of
agricultural production, reducing post-harvest losses, increasing value added to agricultural
raw materials, maintaining the integrity and quality of farm products (CEMA, 2014) even
though studies on farm mechanization impacts on different aspects of agriculture vary from
place to place. Some researches magnify the positive effects of mechanization on labor
absorption, higher yield while others focused on reduced working drudgery and improved
working quality. Baudron et al. (2019) found that use of tractor has negative significant effect
on land productivity in Ethiopia.

2.2.2. Adoption of Farm Mechanization

The adoption of new technology/innovation always follows the assumption of utility

maximization principle. The assumption in this study is also that households are always
rational in decision making and decide to adopt farm mechanization if only they feel the new
technology (farm mechanization) adoption will improve their welfare. Hence, the farmers’
 21

preference of adoption decision which is motivated by random utility function will be

expressed as (1):

U i =  j X i + ei (1)

Where U j is the utility to be obtained from adopting farm mechanization technologies; j = 1,

2, 3…N; X i is the institutional factors and farm household characteristics;  j s are

parameter to be estimated and ei is the disturbance term that is assumed to be distributed

normally.

If farmers’ utility by adopting and not adopting farm mechanization technologies are
represented by U1 and U 0 , respectively; then the utility function of farmer subject to the

farmers’ resource endowment constraints (z), and other observable attributes of the farm
household can affect the adoption decision of farmers (Y) and can be described as follows (2):

U1 = (1, Z , Y )andU0 = (0, Z , Y ) (2)

Base on the theoretical specification of the utility function, we can assume and express an
additively separable utility function in the deterministic and stochastic component (3):

U 1 = U (1, z, y ) = D1 (1, z, y ) +e1

and (3)
U o = U (0, z, y ) = D0 (0, z, y ) + e0

Where, U j (•) is the utility obtained, D j (•) is the deterministic part of the utility and ei is the

stochastic component known to the farmer but unobservable to the researcher. The
household’s decision process is modeled using the random utility framework. From the utility
theoretical standpoint, a farmer prefers to adopt farm mechanization technologies if the
farmer's return with the technology adoption minus its cost of adoption is at least as great as
the return without adopting the technology (4).
 22

U1 (•)  U 0 (•)
(4)
D1 (1, z − c; y ) + e1  D1 (0, z − c; y )

Where, c is the cost associated to adoption of technology that includes the implicit and explicit
cost to the farmers while the existence of the stochastic component allows us to apply
probabilistic distribution about a decision maker’s behavior. Similarly, the probability function
of adoption and non-adoption can be specified as in (5) and (6):

P1 = P(choice) = Pr(D1 (1, z − c; y) + e1  Pr(D0 (0, z, y) + e0 (5)

P0 = Pr(D0 (0, z, y) + e0  Pr(D1 (1, z − c; y) + e1 (6)

The choice of the farmer to adopt the farm mechanization technology in terms of the utility
function of a probability distribution can be expressed as (7):

Pr(choice) = Pr(U1 )  Pr(U 0 ) (7)

By assuming that the deterministic component of the utility function is linear in the
explanatory variables, the utility functions in (3) can be expressed as;

U1b' X i + e1 , andU0 = b' X i + e0 and probabilities in equations (5) and (6) can be given as:

Pr (to adopt) = Pr(U i  U 0 ) .

Pr( i ' X i − ei   0 ' X i + e0

(8)
Pr( i ' X i − ei   i ' X i − e0
 23

2.2.3. Impacts of Adoption of Farm Mechanization on Farm Labor Employment,

Income, Economic Efficiency, Productivity and Input Use

A household can only decide to use a given farm mechanization technology if and only if the
associated utility ( U M ) is higher than that of non-mechanization utility ( U NM ). Specifically, if

‟D*” is assumed to be the difference between these utilities i.e., D* is the difference between
utility from using mechanization and not using mechanization technologies, a household will
use mechanization technologies when D*  0 . However, since utility is not measurable in
direct way then D* is not observable variable. Utilities gained by applying farm mechanization
technologies can be analyzed as function of farm net returns and household characteristics.
Hence, D * can be expressed as a function of utility components (9):

1, 𝑖𝑓 𝐷𝑖 ∗> 0
𝐷 ∗= 𝛼𝑍𝑖 + 𝜀𝑖 , 𝑤𝑖𝑡ℎ 𝐷𝑖 = { } (9)
0, 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒

Where, D is a continuous variable (lying between 0 and 1) that shows index of level of farm
mechanization; Z is a vector that includes households and farm-level characteristics (including
farm net return);  is the error term, and  is the vector of parameters to be estimated.

As the increase in level of farm mechanization use is supposed to increase household welfare
through increasing farm income, productivity, economic efficiency, intensity of agricultural
inputs use and reducing farm labor employment assumes that these variables to be considered
in this study (household income, productivity, economic efficiency, intensity of agricultural
inputs use and farm labor employment) are linear function of level of agricultural
mechanization technologies used and other explanatory variables (10):

Yh = X h + Dh +  (10)

Where, Yh standing for outcome variables in the study; total income, economic efficiency and
intensity of agricultural inputs; X is a vector of explanatory variables;  is the error term, 
and  are parameters to be estimated. The accurate impact of farm mechanization on the
 24

outcome variables is measured by  if households are randomly assigned to different level of

farm mechanization adoption (Dh) However, given a set of household characteristics, access to
technologies and others, level of farm mechanization adoption is not a random process among
households as the farmer can make a decision (on his/her own) to use certain level of farm
mechanization. And as a result, a systematic difference among households may arise based on
their level of farm mechanization use; implying D and X are not orthogonal.

2.2.4. Crop Commercialization

The household is an important but complex unit of analysis which is defined and analyzed in
different ways. Anthropologists have long studied and debated what is actually meant by the
household while economists focus on the activities and behavior of the household (Chen and
Dunn, 1996). In terms of functional definition household can be defined as the peoples who
share collective production, consumption, investment, or ownership of resources. Chowdhury
(2005) also described a household as units of biological reproduction, as property owning
entities or as initiators of a labor process. The peasant agricultural household has peculiar
characteristics; it is both a family and an enterprise; i.e., it simultaneously engages in both
consumption and production (Ellis, 1993). In general household are a basic unit of
interpersonal interaction, generally reflecting both biological and economic communalities
(Haddad et al., 1997). The basic economic model of household formation is built on Becker
(1974) theory.

Neoclassical economic theory recognizes two key units of analysis at the microeconomic
level: consumers and firms. In neoclassical theory, however, consumers and firms are treated
independently; all consumption activities are modeled in terms of the household (or individual
consumer), while all production activities are modeled in terms of the firm. The complete
separation of consumption from production is not especially useful for modeling the
household economic portfolio, particularly in developing countries where most households
engage in a mix of market and nonmarket production. On the other hand, the more useful,
integrated models of the household economy rely heavily on many of the basic concepts and
 25

conclusions from these parallel neoclassical theories of the consumer and the firm (Chen and
Dunn, 1996).

Decision making models of households are historically studied by classifying into two as the
Unitary Model and the collective models of households. The unitary models for household-
level analysis assume that the household behaves "as if" it was a single entity. It is the simplest
form of a hypothetical household (Chowdhury, 2005). A single welfare function represents the
household's preferences. All household resources (capital, labor, land, and non-labor income)
are pooled and all expenditures are made out of pooled income. It is characterized by absence
of conflict and overall collective utilities-which enable complete social harmony (Matilla-
Wiro, 1999). According to this model assumption all household members jointly make
decisions to maximize their collective interest (Ellis, 1988). This assumption of unitary model
of single welfare function has strong theoretical basis like that outlined by Samuelson (1956),
which states that the household welfare function reflects a consensus amongst members; and
that of Sen's (1966) model of cooperatives to the household. Here, family welfare is the
weighted sum of the net utility of all members. The third theoretical base is the possibility that
there exists a household dictator capable of imposing his or her preferences on other members

Relaxing some strong assumptions of unitary model, collective model can be adopted. These
models do not require any unique household welfare index to be interpreted as a utility
function. This allows the index to be dependent on prices and incomes, as well as "tastes"
(Chiappori, 1992). There are two broad types of collective model: cooperative and non-
cooperative. The collective model is limited in its use due to large data needs, fear of
difficulties of getting reliable data against each member of the household and complexity in
applying appropriate analytical models (Ellis, 1988; Haddad, 1994; Haddad et al., 1997). The
collective models explicitly consider the household as a collective entity, but with more than
one decision-making unit. These models concentrate on the question- how can individual,
dissimilar preferences lead to a collective choice (Alderman, 1995).

Based on the objective functions to maximize with a set of constraints, there are three (four)
alternative economic theories of peasant household behaviors. The first two models ignored
 26

the consumption aspects of the household production (Kello, 1992). Since the process of
decision making of peasant household involves both production and consumption aspects,
these models ignore the major side of peasant household decision making process.

The Chayanovian/utility maximizing model is among the earliest proposed model to integrate
production and consumption decisions in the analysis of the peasant household by Chayanov,
Russian agricultural economist (Thorner et al., 1966). In this model, the household seeks to
maximize its utility, where utility is derived from the consumption of both on farm produced
and purchased goods and leisure. By combining utility maximization from consumption theory
with production function from production theory, the Chayanov model provides a foundation
for the integrated models of household decision making, including the farm household model.
Then the new farm household models are also derived from Chayanovian or utility
maximizing model with different assumptions and wider scope and predictive powers (Kello,
1992).

The choice between subsistence and commercial production needs decision of farm household
to allocate resources between the two. This resource allocation decisions between subsistence
and commercial production are influenced by demographic, socio-economic, institutional and
natural factors. In developing countries insuring household members against hunger and
destitution is the main goal of a household (Dasgupta, 1993). Within the standard expected
utility approach, the introduction of risk in peasant production choices has entailed including
household preferences toward risk. The risk behavior of agents is determined not only by
preferences but also by the availability of institutions that facilitate risk bearing (Roumasset,
1976). Institutional arrangements provide imperfect insurance, households will self-protect by
exercising caution in their production decisions (Morduch, 1995). The three or (four)
alternative economic theories of peasant household behaviors are discussed as follows:

Profit maximizing peasant theory: Schultz’s (1964) in his book Transforming Traditional
Agriculture, hypothesized that the peasant production mode as profit-maximization behavior.
He refers explicitly to allocative efficiency (optimal combination of units or amount of inputs
so as to maximize profit; i.e., equating marginal value product (MVP) to marginal factor cost
 27

(MFC), and implicitly to technical efficiency (the manner in which the same unit or amount of
inputs are combined in order to maximize profit). According to this efficiency is defined in a
context of perfect competition (i.e., all producers apply the same prices, workers are paid
according to the value of their marginal product, inefficient firms go out of business or
become competitive by improving their efficiency, and entrepreneurs display non-diminishing
marginal utility of money income). Several studies have adopted the AE criterion to test
whether peasants were or were not efficient (i.e., whether they were profit-maximizers or not)
with some contradictory results where some shows that peasant households are efficient
(Hopper, 1965) other show they are not efficient (Bliss and Nicholas, 1982). Moreover, his
model is criticized for its focus on AE and overlooked TE which is the major aspect of EE as
its methods are using average production function opposed to best practice firm condition
(Kello, 1992). Most importantly, this approach ignores risk and uncertainty in agricultural
production (Lipton, 1968).

Utility maximization theory: The analysis of this theory is based on Chayanov’s ‘theory of
peasant economy’. According to this theory, a peasant farm household works till it achieves an
equilibrium between the increasing drudgery of family labor and the decreasing marginal
utility of goods produced (Chayanov, 1966). Chayanov’s theory has been developed into a
neoclassical framework by Tanaka (Nakajima (1986). Nakajima called it the ‘subjective
equilibrium theory of the farm household’. He developed the theory so as to make it relevant
for the analyses of the farm firms, commercial farms, farm households and subsistence farms.
The neoclassical farm household model explains the behavior of farm households in
simultaneous decision making about consumption and production. This model typically
incorporates the notion of full household income and conceives of the household as a
production unit that converts purchased goods and services as well as its own resources into
use values or utilities when consumed (Becker, 1965). Thus, the household maximizes utility
through the consumption of all available commodities (i.e., home-produced goods, market-
purchased goods, and leisure), subject to full income constraints. The model shows that if all
markets exist and all goods are tradable, prices are exogenous and production decisions are
taken independently of consumption efficiency infers the nature of the former by investigating
on the latter. In such conditions the decision-making process could be regarded as recursive
 28

(or separable), because time spent on leisure and time used in production becomes
independent; utilization of household labor will be directly linked to the market-determined
wage rate, and income is singled out as the only link between production and consumption
(Singh et al., 1986).

In reality, households operating in developing countries are likely to face more than one
market imperfection, which prevents first-best transactions and investments from taking place.
Empirical analyses of recursivity in farm household decision making have generally produced
negative results (Bardhan and Udry, 1999). Therefore, as indicated in Mendola (2007), there is
no separability between consumption and production. The decision process becomes circular
as consumption affects income and income affects consumption. Hence, the validity of
recursive modeling of household resource allocation depends on the household being a price
taker and the absence of missing or imperfect markets (for output or input, including labor and
capital).

The risk-averse peasant: Ellis (1992) outlined that peasant produce under very high levels of
uncertainty induced by natural hazards; market fluctuations; and social uncertainty (insecurity
associated with control over resources, such as land tenure and state interventions, and war)
which pose risks to peasant production and make farmers very cautious in their decision
making (Walker and Jodha, 1986). It is not surprising, therefore, that farmers (in common with
most other decision makers) are generally assumed to exhibit risk aversion in their decision
making. Lipton’s (1968) criticism of the profit approach sought to show how the existence of
uncertainty and risk eroded the theoretical basis of the profit-maximizing model. Small
farmers are risk-averse because they have to secure their household needs from their current
production or face starvation. However, some studies on the risk-taking behavior of the
households showed that smallholder farmers are risk-averse (Norman, 1974; Binswanger and
Sillers, 1983) while others concluded that they are not (Roumasset, 1976). Further argument is
that the theory of the risk-averse peasant holds if the sources of risks are removed through the
introduction of measures to mitigate ravages of nature like introducing irrigation (Lipton,
1968; 1977).
 29

The new household farm models theory: This became popular since 1960s and the model
starts by dropping two major assumptions of Chayanov: non-existence of labor market and
unlimited supply of land. It usually incorporates the “new home economics” theory developed
by Becker (1965) to the basic structure that has been developed by Chayanov (1966).
Consequently, the utility function of the household represents its preference ordering between
a range of final characteristics of home-produced goods and services. Hence, the rather than
considering a household as consumption unit, the model considered the household as
production unit that convert purchased goods and services as well as its own resources into use
values or utilities in the case of consumption.

According to Nakajima (1986) the degree of own labor utilization and own output
consumption are an identification criterion for degree of subsistence and commercial farming
categories. When all outputs produced are consumed and only own labor are used for
production, it is complete subsistence and when all outputs are sold and all labor used are
employed externally, it is termed as complete commercialization while all other households
are categorized in between these two extremes. Farm decision making process could be
divided into production and consumption according to this model. When the household is
price taker and labor market exist, the decision flows from production to consumption and
family labor utilization decision will be directly linked to the market determined wage rate
(Kello, 1992).

In this research, a neoclassical utility maximization theory was adopted where it was assumed
that households are rational and they make decision whether to participate or not to participate
in commercialization process based on the utility they can gain from the scenarios.

2.2.5. Productivity and Production Efficiency

The basic theory of production is based on the assumption that firm is unit of analysis and
seeks to maximize its profits by selecting the optimal levels and combinations of inputs and
outputs. Production function also shows the physical relationship between production inputs
(usually termed as factors of production) and the outputs. According to neoclassical theory of
competitive market model, it is assumed that firms pay certain amount of money (price) to
 30

purchase inputs and receive certain amount of money in selling their output (produce). The
model has certain basic assumptions which are: first, the purchases as much inputs it can and
also sell unlimited amount of its outputs (unlimited purchase of inputs and selling of outputs);
second there is no limitation of capital for the production of outputs for firms; and third the
optimal combination of inputs and outputs occurs where the contribution to profit of each
additional unit of an input is equal to the price of the input and the contribution to profit of
each additional unit of output is the same (Chen and Dunn, 1996).

According to economic models of household decision making evolution, household is

considered as both producer (firm) and consumer unit. It is known that the neoclassical
economic models are based on three assumptions: 1. People have rational preferences among
outcomes. 2. Individuals maximize utility and firms maximize profits. 3. People act
independently on the basis of full and relevant information (Chowdhury, 2005). Based on
these, there are several important concepts from neoclassical production theory that are useful
in building a model of the household economic portfolio. First, production theory provides a
framework for understanding how the household would (or should) allocate its resources to the
various economic activities in order to maximize profits. The household's productive resources
correspond to the inputs, or factors of production, in this theoretical framework. In addition,
production theory provides the basis for clear conceptual definitions of technical, allocative,
and EE. This provides us with a benchmark for the profit maximizing household and allows us
to predict how the household might react to changes in prices or technology (Chen and Dunn,
1996).

For the empirical research on productivity measurement, the economic theory of production
has provided the analytical framework. Production function is the foundation of the theory and
it postulates a well-defined relationship between output and factor inputs. Mainly two sources
are seen as means to achieve productivity improvement; first, technological change (like use
of improved practices of production such as mechanical technologies, fertilizers, pesticides,
improved seeds, etc.) which pushes the production frontier upward; and second, improving the
farmer’s skills in using the existing techniques of production (Gezahegn et al., 2006).
 31

Economic theory assumes that producers optimize both from a technical and economic
perspective: (1) from a technical perspective, producers optimize by not wasting productive
resources and (2) from an economic perspective producers optimize by solving allocation
problem involving prices (Kokkinou, 2010). According to Singh (1986) productive efficiency
constitutes three components: TE, AE, and over all individual production (EE) efficiencies.
Other economists like Färe et al. (1985) and Antle and Capalabo (1988) have tried to include
scale efficiency in addition to technical and allocative as components of production efficiency.
Scale inefficiency is the situation where the firm is unable to operate at its constant returns to
scale (CRS), at a long-run competitive equilibrium. AE reflects the ability of a firm to use the
inputs in optimal proportions, given their respective prices and the production technology.

The pioneer work of Farrell (1957) is the starting point for the theoretical basis of economic
efficiency. According to his work to measure efficiency two approaches: input-orientated or
output orientated approaches can be used. The input-oriented approach answers the question
“by how much can input quantities be proportionally reduced without changing the quantity of
output produced” while the latter addresses the question “by how much can output be
increased without increasing the amount of input but by more efficiently utilizing the given
input”. Later on, Coelli et al. (2005) elaborated these two approaches based on Farrell (1957)
with a separate graph for both input-oriented and output-oriented methods as precisely
presented subsequently.

[Link]. Input-oriented measures

To demonstrate this approach, Farrell (1957) made two important assumptions (constant
returns to scale and knowledge of the unit isoquant of fully efficient firms, represented by SS'
in Figure 1). For simplicity, he used example of firms which use only two inputs (X1 and X2)
to produce a single output (q) for the measurement of technical efficiency. Based on Coelli et
al. (2005) and Farrell (1957), if a given firm uses quantities of inputs, defined by the point P to
produce a unit of output, the technical inefficiency of that firm could be represented by the
distance QP, which is the amount that all inputs could be proportionally reduced without a
reduction in output. The usual ways of expressing this idea are in percentage term by the ratio
 32

QP/OP to show the percentage by which all inputs need to be reduced to achieve technically
efficient production. The TE of a firm is commonly measured by the ratio TE = OQ/OP, or (1-
QP/OP).

Figure 1. Input-oriented technical, allocative and economic efficiencies

Source: Coelli et al. (2005)

Farrell (1957) demonstrated that the cost efficiency of a firm can be estimated if inputs price
data of firms does exist. For example, if the price ratio is denoted by the iso-cost line, AA′ in
Figure 1, then the cost efficiency of the firm can be defined as the ratio of OR/OP. With the
same information, the AE is defined by the ratio: AE = OR/OQ, TE = OQ/OP and the overall
efficiency or economic efficiency (the product of AE and TE) is: EE = AE*TE =
OR/OQ*OQ/OP = OR/OP; which is equal to the cost efficiency. It should be noted that all
efficiencies (TE, AE and EE) take a value between zero and one, and hence provides an
indicator of the degree of inefficiency of the firm where the value one implies the firm is fully
efficient.

[Link]. Output oriented measures

 33

This approach is to address a question ‟By how much can output quantities be proportionally
expanded without altering the input quantities used?” Unlike input-oriented measure Farrell
(1957) and Coelli et al. (2005) considered the case where production involves two outputs (q1
and q2) and a single input (x) for illustration. Assuming a constant return to scale, one can
represent the technology by a unit production possibility curve in two dimensions. The curve
ZZ' in Figure 2, is the unit production possibility curve and point ‘A’ corresponds to an
inefficient firm. It can be clearly noted that an inefficient firm operating at point ‘A’ which lies
below the curve, as ZZ' represents the upper bound of the production possibilities.

In the Farrell output-oriented efficiency measure, the distance AB (in Figure 2) represents
technical inefficiency (the amount by which outputs could be increased without using
additional input). The TE is the ratio of OA/OB. If price information is known, from the iso-
revenue line, DD', the AE and TE measures can be defined as: AE = OB/OC and TE =
OA/OB. Furthermore, the overall revenue efficiency or EE can be obtained as the product of
TE*AE, which is OA/OC.

Figure 2. Technical and allocative efficiencies from an output orientation

Source: Coelli et al. (2005)
 34

Though production efficiency measurement is possible using either of output oriented or input
oriented approach when all the necessary data are available, however, in developing countries
where output shortfalls is more concerned than input overuse, output-oriented measures of
efficiencies seem to be logical and more frequently used (Solomon, 2014). For the same
reason, output-oriented approach is used in this study.

2.2.6. Crop Diversification

Ellis (1993) argues that in any given locality farm household decisions regarding what to
produce, how much to produce and how to produce is the basic economic theory of farm
production and management. Basically, these decisions call for crop choices and cropland
allocation and have implications for farm household’s crop diversification level and crop
rotation practices (Nkonya et al., 2005). Cropping system (level of crop diversification) is one
of the crop choices and cropland decision making in farm household which is the resource
allocation process among enterprises to maximize household welfare. Economists have also
adopted the use of agricultural household model in determining resource allocation for
production (Singh et al., 1986; Mwaura and Adong, 2016). The model considers agricultural
households to be involved in maximization of both production and consumption. The
decisions of what to produce and consume are simultaneously made with the aim of
maximizing profit and utility, considering the constraints presented by resource availability.
The use of profit and utility maximization approach yields better results in situations where
households operate in a complete market. However, under incomplete markets, the agricultural
household model predicts production decisions to be a factor of preference and endowment of
households (Mwaura and Adong, 2016).

Therefore, what crop to produce and cropland allocation (level of crop diversification) are
simultaneous decisions to be made by farm households under the consideration of incomplete
market (Hua and Hite, 2005). These decisions can be influenced by both farm household
characteristics and government policies such as subsidy program (Di Falco and Perrings, 2005;
Westcott and Young, 2004; Wu and Brorsen, 1995). Similarly, crop choice and cropland
 35

allocation decisions determine agricultural production levels, and as well affect land resource
conditions, crop diversification, levels of farm income and household food security and
welfare. As a result, these decisions are becoming current issues of concern for both
smallholder farmers and policy makers (Wu et al., 2008; Hua and Hite, 2005). Farm
households make crop choice and cropland allocation decisions within their own production
risk and uncertainty management strategies, income diversification strategies and market
access constraints (Zeller et al., 1998; Chavas and Holt, 1990; Collender and Zilberman,
1985). Hence, similar to other objectives, utility maximization theory was also adopted for this
objective where a household can select a given mix of crop portfolio based on the expected
utility it can gain from that bundle.

2.3. Analytical Framework

2.3.1. Measurement of Agricultural Mechanization Level

Different authors used different methods to measure the level of farm mechanization. Wawire
et al. (2016) and Özpınar (2020) used qualitative description method to show level of farm
mechanization. To assess mechanization level Nowacki (1974) first categorized work on a
given farm based on their sources of energy as work performed by human power (labor),
animal power and machine. Then, he calculated mechanization indicator (mechanization
index) as percentage of work done by machine (LM) in the total amount of work in a given
process, i.e., in the total amount of work performed by workers/labor (LR), by animal (LA) and
by machine (LM).

Zangeneh et al. (2010) also defined Mechanization Index (MI) and LOM, to characterize
farming system of potato in the Hamadan province of Iran. Their index was based on the prior
calculated value of regional mechanization index. It is used to analyze mechanization index of
a given location (province) relative to the larger region, for instance, MI of Ethiopia compared
to east Africa or that of Oromia region relative to Ethiopia. According to Almasi et al. (2000)
and Maheshwari and Tripathi (2019) the LOM is calculated as the ratio of total energy power
(Kwh or MJha-1) that has been exerted by use of tractors and other machines to total land
 36

cultivated in the study area. However, this approach requires the estimation of energy used by
each and every power sources (human, animal and machine) based on their standard
equivalence. As it was used in Abbas et al. (2017), the energy associated with fuel, human and
animal working hours can be estimated directly by multiplying their equivalences, while the
machine energy can be calculated directly by using equation (11) as follows:

(Wm xEeq xtm )

Machine.E = (11)
Tem

Where, MechE is the machine energy (MJha-1); Wm is the mass of machine (kg); Eeq is the
energy equivalent for machinery; tm is the time that machine used per unit area (hha-1); Tem is
the economic life of machine (h).

Machinery Energy Ratio (MER) is also another method used by Collado and Calderon (2000)
which indicates the investment in machinery energy in comparison with the other input energy
sources required for crop production. In this study, mechanization index (MI) (IE) expressed
by the percentage of machine work EM to the sum of manual EH, animal EA and machine work
EM expressed in energy units, as suggested by Nowacki (1978) and extended by Collado and
Calderon (2000) has been proposed for model forecasting by incorporating the production cost
factors using the following equation (12):

EM
MI( I E ) =
( E H + E A + EM ) (12)

Where; IE, is Mechanization Index (MI) in MJha-1, EM, EH and EA are manual, animal and
machine work expressed in energy units (MJha-1); while the application of the rest of the
approaches listed above which have been used by other authors will not be feasible mainly due
to data availability both at household and at aggregate level.

Factors affecting level of farm mechanization can also be identified by using different methods
based on data availability, target of technology under consideration and research objectives. In
 37

this study, since the dependent variable (level of mechanization) is in an index form and
continuous, Tobit (censored limited dependent) model was used.

2.3.2. Economic Impact of Adopting Farm Mechanization

It is assumed that households are free to select themselves into different levels of farm
mechanization based on their demographic and socioeconomic characteristics (Takeshima,
2017; Ma et al., 2018; Amoozad-Khalili et al., 2020; Tesfaye et al., 2021; Zhou and Ma,
2022). And this situation will lead to a sample selection bias to estimate impacts of farm
mechanization on outcome variables. Self-selection and other measurement errors are common
problems of observational data that have been tried to be overcome by Propensity Score
Matching methods like pscore, Inverse Probability Weighting (IPW), and others. However,
these estimators are not both robust and consistent. Hence, an estimator known as Augmented
Inverse Probability Weighted (AIPW) was developed (Robins et al., 1994; Robins, 1999;
Scharfstein et al., 1999) and applied recently by social researchers (Glynn and Quinn, 2009;
Linden and Yarnold, 2016; Ma et al., 2022; Zhou and Ma, 2022).

It is expected that adoption of farm mechanization has positive and/or negative impact on
different parameters in a given farm household. Literatures revealed that use of farm
machinery can improve household welfare by increasing household income, time for leisure,
and reducing work drudgeries. Similarly, it can also have impact on economic efficiency,
amount of labor required, amount and types of inputs like use of improved seed, chemical
fertilizer and pesticides.

[Link]. Analytical framework for technology adoption

There are two basic adoption study models: static and dynamic models. The static model
assumption is the analysis of adoption extent and intensity of use of the new technology at a
given time. However, even though we are estimating adoption extent at a given time, it is the
result of dynamic processes. Decisions of the farmer in a given period are assumed to be
derived from the maximization of expected utility (or expected income, productivity etc.)
 38

subject to household characteristics, land availability, credit, and other constraints. Income,
productivity etc. are functions of the farmer's choices level of farm mechanization technology
level in each time period. It therefore depends on its discrete selection of a technology and its
level of farm mechanization technologies from a mix including the traditional technology and
a set of components of the modern technology package (Feder et al., 1985).

[Link]. Impact of farm mechanization on EE, income and intensity of input use:

When treatment is binary (whether to participate or not in a given program or adopt or not to
adopt a given technology) assessing the impact is too simplistic, since farmers usually
participate or adopt differently in the program or technology adoption, resulting in
considerable differences in their net returns. However, when the program or adoption of
technology provides a different “level” of treatment (or dose t) to treated - ranging from 0
(absence of treatment) to 100 (maximum treatment level): two groups of units are thus formed:
(i) untreated, whose level of treatment (or “dose”) is zero, and (ii) treated, whose level of
treatment is greater than zero. In this case we are interested in estimating the causal effect of
the average treatment (t) on an outcome y by assuming that treated and untreated units may
respond differently to observable confounders (x) we wish to estimate a Dose-Response-
Function (DRF) of y on t. Hence, in this study, DRF models like Generalized Propensity Score
(GPS) (Hirano and Imbens, 2004), IPW and AIPW are possible for continuous treatment cases
(level of farm mechanization-MI) to capture the impact on household’s income, amount of
hired labor, adoption of inputs (improved seed, chemical fertilizer, agrochemicals) and
economic efficiency.

However, both regression and inverse probability weighting methods are vulnerable to bias.
But the AIPW estimator combines both the properties of the regression-based estimator and
the IPW estimator and is therefore a ‘‘doubly robust’’ method in that it requires only either the
propensity or outcome model to be correctly specified but not both (Kurz, 2022). The
regression adjustment and IPW are modeling the outcome and treatment probability
respectively while AIPW estimator models both the outcome and the treatment probability.
Due to the property of AIPW, that it only needs the correct specification of one of the two
 39

models (either outcome or treatment probability) to consistently estimate the treatment effects,
it is called doubly robust estimator.

2.3.3. Crop Commercialization

Degree of commercialization can be measured from output side or input side. Even though the
household may not produce based on market price signal it may also have high market
participation because of surplus production due to various reasons, including favorable
weather conditions, although it may not be market oriented, which is a good conclusion for
framing Key et al. (2000). In this study market participation in crop production or
commercialization was measured by the proportion of crop output sold.

Crop output commercialization index (COMPI): The output side of commercialization can
be seen at household or community level. In the case of household level, one definition of
agricultural commercialization is the ratio of the value of agricultural outputs sold to the total
value of agricultural outputs produced by a household (Von Braun and Kennedy, 1994;
Strasberg et al., 1999; Berhanu and Moti, 2010). Hence, crop output side commercialization
index for a household (COMPI) can be computed as (13):

N
S ki
COMPI k =  ; Qki  S ki
i =1 Qki
(13)

Where, COMPIk is the proportion of crop k sold (Ski) to the total amount produced (Qki)
aggregated over the total sample households in a given location.

The index measures the extent to which household agricultural production is oriented towards
the market. A value of zero would signify a totally subsistence-oriented household; the closer
the index is to 100, the higher the degree of commercialization. Based on suggestion of some
authors like (Samuel and Ludi, 2007; Samuel and Sharp, 2007; Bedaso et al., 2012) a
household is commercial oriented if market orientation index is a minimum of 65% whereas
values in between 30% and 65% termed as semi-commercial and otherwise non-commercial.
 40

The most widely used models to investigate household’s decision for market participation and
its level of participation are double hurdle model (DHM), Heckman’s two stage model, Tobit
model and multinomial logit models (Efa et al., 2016; Abafita et al., 2016; Geoffrey, 2014).
However, each model has its own limitations and advantages while choice of one over the
others depend on objectives of study and data type. The censored regression or Tobit (Tobin,
1958) model is appropriate when the dependent variable is censored at some upper or lower
bounds while in Heckman’s model the estimation of the first stage equation of the model as a
probit model assumes that the errors are Homoscedastic (Winship and Mare, 1992).

In any cases, Heckman, Tobit and DHM are the standard models. The Tobit model is
sufficient to accommodate the zero observed figures alongside other positive values if the
simultaneity of decisions is assumed; i.e., both the decision to participate and the level of
participation are determined by the same variables and with the same sign (Wooldridge, 2002).
Hence, decision to participate in the market and the intensity of market participation are jointly
determined and influenced by the same parameters. This restriction of the variables and
coefficients in the two decisions in Tobit model pushed the econometricians to develop
alternative approaches.

Alternative for Tobit model is the Heckman’s two stage model. In this case even though this
model assumes that the decision to participate in market (either in output or input) and the
intensity of market participation may not necessarily be jointly determined (Humpheys, 2010),
Heckman selection model is appropriate if there is a censoring process in measuring the
intensity of participation. This implies that the Heckman procedure assumes that there are
some potential market levels in the sample population but are not observed due to sample
selection problems. In general, Heckman's sample selection model is designed to account for
the fact that the observed sample may be non-random.

A model which incorporates relevance of the participation decision to the Tobit model with
the probability of participation and the intensity of participation being determined by separate
processes was proposed by Cragg (1971) and is a two-tiered process known as the double
 41

hurdle model. According to this model it is assumed that farmers faced with two hurdles in
any agricultural decision-making process (Humpheys, 2010). Accordingly, the decision to
participate in an activity is made first and then the decision regarding the level of participation
in the activity follows. In this study decision to use either DHM or Heckman’s two stage was
determined based on test for inverse mills ratio (IMR) for Heckman’s two stage model and
accordingly, since IMR was insignificant, Heckman’s two stage model was employed.

2.3.4. Measurement of Production Efficiency and Identifying its Determinants

Generally parametric and non-parametric are the two analytical approaches to address
empirical studies of economic efficiency. Parametric approach employs either distance
function (DF) or stochastic frontier analysis (SFA) along with econometric methods in
estimating the parameters of the function while the latter uses the Data Envelopment Analysis
(DEA) with mathematical programming as a non-parametric approach. Parametric approach
has an assumption that the functional form of the production function is known ahead (Coelli
et al., 2005) but the non-parametric approach has no assumption. However, both approaches
have their own advantages and disadvantages.

The SFA approach has advantages because it: 1) introduces stochastic random noises that are
beyond the control of the farmers; 2) allows statistical test of hypothesis relating to the
goodness of fit of the model and test the degree of inefficiency. However, it has limitations in
that: 1) It imposes explicit restriction on functional form and 2) It requires specification of the
technology, which may be restricted in most cases (Coelli et al., 2005).

The non-parametric approach has advantage as it does not require the specification of a
particular functional relationship between inputs and outputs but the production frontier is
primary estimated using field data and measure efficiency relative to the constructed frontier
using linear programming approaches. However, it has certain drawbacks which are: First, it is
deterministic rather than stochastic which assumes that all deviations from the frontier are
attributed to inefficiency. This means it ignores the possible influences of external factors or
random shocks like measurement errors and other noises in the data which makes the approach
 42

to be sensitive to outliers (extreme observations) and data measurement errors. Second, it does
not enable to estimate parameters of the model and hence impossible to test hypothesis
concerning the performance of the model. Finally, the estimated function has no statistical
properties, and hence the estimated production frontier has no statistical properties to be
evaluated upon as it uses mathematical programing procedure that produces estimates without
standard error, t-ratio, and other important statistics (Coelli and Battese, 1996).

Both parametric and non-parametric approaches have been widely employed in the past
empirical studies. However, after the work of Coelli and Battese (1996) the application of non-
parametric (deterministic) approach for efficiency analysis in the field of agriculture was
challenged for not accounting exogenous factors as well as measurement errors which are
inherently associated in agricultural production activities specially in rain-fed production
system like Ethiopia. As a result, the parametric approach which accounts for those limitations
of non-parametric approaches was suggested for agricultural sector efficiency analysis.

The stochastic frontier model developed by Aigner and Chu (1968), Aigner et al. (1977), and
Meeusen and Broeck (1977), recognizes the effect of random errors and data noise on
agricultural production and assumes that deviations from the production frontier may not be
totally under the control of the producers. This method permits random variation of the
frontier across firms, and captures the effects of measurement errors caused from outside the
firms’ (farmers’) control, random shocks and other statistical “noise” and the one-sided
component captures the effects of inefficiency relative to the stochastic frontier.

Any kind of forms of production functions such as linear, normalized quadratic, generalized
Leontief, constant elasticity of substitution, Cobb-Douglas and Transcendental Logarithmic
(Translog) production functions are possible to be assumed in parametric stochastic frontier
approach. Nonetheless, it is quite common to use the last (Cobb-Douglas and Transcendental
Logarithmic (Translog)) production functions in field of agricultural production economics.
Their nature which enables us to find a model which is linear in the logs of the inputs and thus
easily imparts itself to econometric estimation methods, make the two functional forms
attractive. It is clear that Cobb-Douglas production function is a special form of the translog
 43

production function where the coefficients of the squared and interaction terms of input
variables are assumed to be zero. Furthermore, the Cobb-Douglas function assumes CRS and a
unitary elasticity of input substitution which is the imposition of severe prior restriction on the
technology of production (Wilson et al., 1998; Coelli et al., 2005). Besides this, the Translog
production function has advantages over the Cobb-Douglass production function as it
estimates second order functional form that accounts for the effect of input interaction on
output. However, its susceptibility to serious multicollinearity problem unlike that of Cobb-
Douglass is its major disadvantage (Headey et al., 2010).

Kopp and Smith (1980), Taylor et al. (1986), Krishna and Sohota (1991), and Banik (1994),
argued that unless the researcher is interested in the analysis of general structure of the
production function, the functional form has a little impact on the measure of technical
efficiency. As a result, either of the two, Cobb-Douglas or Translog stochastic frontier models,
can be employed to measure economic efficiency in agriculture and the choice depends on the
objectives of the research. However, in most recent literatures, Translog is being widely
employed when Cobb-Douglas tested to fail to represent the data at hand using relevant testing
mechanisms. However, both of them can also be used simultaneously to compare the results
(Hassan and Ahmad, 2005).

There are two approaches to analyze determinants factors of inefficiency namely two-stage
and one-stage approaches. The two-stage approach estimates the level of efficiency score in
the first stage and then, using the estimated level of efficiency score as a dependent variable,
regress it with the hypothesized independent inefficiency factor variables in the second step.
This two-step modeling approach has been subjected to criticisms by opponents such as
(Battese et al., 1996; Battese and Coelli, 1995; Coelli et al., 1998); Kumbhakar and Lovell,
2000). The criticism was justified by the fact that the two-step approach violates one of the
basic assumptions of ‘identically independently distributed technical inefficiency effects in the
stochastic frontier’. The second approach proposed by these economists is a one-stage
modeling approach in which technical inefficiency effects are function of various observable
factors that are assumed to affect efficiency of producers and has been widely implemented
(Battese et al., 1996; Battese and Broca, 1997). However, this approach is best suited to
 44

estimate only determinants of technical inefficiency as in Battese and Coelli (1995) and unable
of estimate determinants of allocative and economic inefficiencies (Bravo-Ureta and Rieger,
1991). Therefore, to analyze of economic efficiency, the usual two-stage procedure is more
appropriate.

Efficiency score in two-stage estimation procedure can be estimated by using either parametric
or nonparametric approaches in the first stage and the estimated efficiency score is taken as
dependent variable in the next step and, regressed on explanatory variables using either
Ordinary Least Squares (OLS) or two-limit Tobit models. However, problem of
heteroscedasticity is serious in such cases since the efficiency scores (dependent variable in
the second stage) is between 0 and 1 which makes the use of OLS model difficult (Gujarati,
1995). As a result, Maddala (1983) suggested two-limit Tobit model for such types of
dependent variables
2.3.5. Crop Diversification and Its Determinants

Different diversity measuring indices methods have been employed to measure crop
diversification by different authors. Several methods can be used to examine the extent of crop
diversification at a given point in time such as: Entropy Index (EI), Modified Entropy Index
(MEI), Herfindahl Index (HI), Transformed Herfindahl Index (THI), Margalef index (MI), and
Simpson Index (SI) methods (Pal and Kal, 2012). As a rational decision maker, a household
should always select a portfolio of crop enterprises that fulfill certain criteria like high
yielding, provides food security, reduce/minimize risk of crop failure, having good market
demand based on their local conditions.

The choice of level of crop diversification will follow a utility maximization theory based on
certain features of crops from available crops types. A household can maximize its utility by
increasing household income, secure food security, maximize leisure time, minimizes risk of
crop failure due to different diseases and climatic conditions, etc. (Inoni et al., 2021). A
household may choose and decides to grow a given combination of crops (bundles) measured
by crop diversification index (CDI) if the perceived utility is greater than the utility, he gets
from other portfolio choices (CDI) available. Let Uj and Uk represent utility from two CDI for
‘j’ < ‘k’, then random utility models can be given by (14):
 45

U j =  ' X i +  j andU k =  ' X i + k, (14)

Where: Xi is a vector of explanatory variables that determine CDI; βj, βk - represent parameters
to be estimated and εj, εk – are error terms. Thus, the household will decide to adopt CDIj if
perceived utility (Uij) is greater than other cropping options or CDIk (Uik).

2.4. Review of Related Empirical Literatures

This part of the document will provide empirical literatures that are related to sub-topics:
history of farm mechanization in Ethiopia; level of mechanization; effects of mechanization on
selected economic variables of household; level of agricultural commercialization and its
determinants; and finally, level of crop diversification and its determinants.

2.4.1. Agricultural Mechanization History in Ethiopia

[Link]. Agricultural mechanization during the imperial regime

Unlike most other African countries, Ethiopia is one of the ancient users of animal powered
mechanization technologies especially for tillage purposes (Solomon et al., 2006; Firew, 2015)
which have been performed by using an ox-drawn breaking plough, known as ard plough –
maresha. Similarly, during 1970s animal drawn wooden sledge locally called “hobollo” was
introduced for transportation of bundle of non-threshed crops (Starkey, 2000).

However, like other African countries, the introduction of inanimate farm mechanization
technologies was started after the 1950s national food shortage when, medium and large scale
mechanized commercial farms were given attention. As a result, few lowland irrigations based
and other few highland rain-fed commercial farms appeared in the mid of 1960s (Cohen,
1987). The agricultural sector part of first (1957-62) and second (1963-1968) five years
imperial’s development plan focus on generating foreign exchange through increased exports
and raising local food production to reduced food import. As a result, commercial farms were
 46

established by joint government ventures, foreign and local investors (Kline et al., 1969;
Firew, 2015).

However, the third (1968-73) and fourth (which was neither published nor implemented due to
revolutionary actions happened in the country) five years development plans sought to deal
more directly with bottlenecks of agricultural development but due attention was given to
large commercial farms as the key to agricultural progress of the nation (Cohen, 1987; Stahl,
1973). The government launched Chilalo Agricultural Development Unit (CADU) and Arsi
Rural Development Unit (ARDU) programs and started the promotion of agricultural
mechanization along with other inputs like improved seed and chemical fertilizers. The policy
directly favored the medium and large farmland holdings giving privileges of duty-free
tractors, fuels and spare parts imports, access of loan. As a result, landlords evicted their
tenants and other merchants with no agricultural background from cities were attracted by this
lucrative business and joined the sector through contractual farming. This also increased land
rental price and created pressure on smallholder farmers (Cohen, 1987; Stahl, 1973).
Similarly, the program called ‘minimum packages’ expanded large mechanized and
commercialized farming to other part of Shewa like Bako, Tullu-Bollo and others (Henock,
1972; Cohen, 1987; Stahl, 1973). Hence, due to reported negative consequences of the
program CADU/ARDU decided to stop promotion of such technologies (Holmberg, 1972).

[Link] Agricultural mechanization during the Dergue regime (1974-91)

The military Dergue government has performed the redistribution of land and changed the
land tenure system that changed owner entitlement and all tenants become landowner (Jemma,
2004). However, most of the development activities were centrally planned and all large and
medium commercial farms were owned by state. The main agricultural mechanization
technologies users were producers’ cooperatives and state farms (Abenet et al., 1991;
Mohammed, 2000). A few producers’ cooperatives owned their own tractors and their
accessories while others were renting where they got access and otherwise using draught
animal power (Yared and Moti, 2019).
 47

[Link] Agricultural mechanization in Ethiopia after 1991

Following the political and economic reform of 1991, the producers’ cooperatives have been
dismantled while the large and medium state farms continued under government’s control.
Even though, the government has no clear agricultural mechanization policies until the recent
period it supported the state-owned machinery assembling and service render plants and
enterprises. In 2004, under Regulation No. 97/2004, the Council of Ministers declared the
establishment of Agricultural Mechanization Service Enterprise (AMSE) to render
mechanization and maintenance services on rental basis and sale farm implements and spare
parts (FNG, 2004).

In 2010, the tractor assembling plant established by the preceding government was transferred
to the Metals and Engineering Corporation (METEC). The relatively free market policy allows
investment on agricultural mechanization and the number of mechanization service providers
is also increasing (Yared and Moti, 2019; Mohammed et al., 2000). As a result, small-scale
farmers have started benefiting from hiring agricultural machinery both from government and
private-owned agricultural machinery centers. In the proposed study areas state farms and the
Ethiopian Seed Enterprise (ESE) were also providing services to surrounding farmers mainly
to establish good relationships with the surrounding farmers (Mohammed et al., 2000).

The increased number of agricultural machinery service providers around Arsi and west Arsi
zones has encouraged some individuals (e.g., civil servants, traders) in towns and well-to-do
farmers to aggressively work on farming ventures through land and machinery renting
(Mohammed et al., 2000). Based on the recommendation of ATA and MoA the ministry of
finance (MoF) also approved tax free imports of agricultural machineries in 2019 (ATA,
2019). Moreover, ATA and MoA developed comprehensive agricultural mechanization
strategies.

2.4.2. Economic Effects of Agricultural Mechanization

Agricultural mechanization has effects on different parameters or aspects of the farm

households such as factor productivity, household income, economic efficiency, and inputs
 48

(improved seed, chemical fertilizer, agrochemicals and labor) utilization and so on. In
smallholder farming system working animals are the most important asset that a household
used for various agricultural activities. The adoption of farm mechanization like tractors,
power tiller and combine harvester is expected to reduce the number of these working animals
definitely. Study findings somewhere in the world attested that the use of farm machineries
like power tillage drastically reduced number of working animals (Jabbar et al., 1983).

At every stage of agricultural production, mechanization contributes to availability of food,

reduce production cost and improve working condition and lifestyle of the farmers (Baudron et
al., 2015). Pre-harvest mechanizations enable farmers to cultivate more land when there is
abundant land; increases intensification when irrigation and favorable moisture exist; timely
operation of plowing, sowing, weeding and field management and inspection (Malabo, 2018;
Baudron et al., 2015). It can also enhance efficient use of agricultural inputs (Tamrat et al.,
2017; Reid, 2011). Similarly, harvest and postharvest mechanization can enhance productivity
by improving the timeliness and efficiency of farm operations, reduces postharvest losses and
maintains production quality (Malabo, 2018; Reid, 2011).

Farm mechanization is the key for farm modernization that revolutionized agriculture
throughout the world. It has played irreplaceable role in increasing total factor productivity
and determines the efficiency and productivity of all other inputs used in crop production such
as improved seed, fertilizers, agro-chemicals, water, labor and time and decreased cost of
production there by increase farm profitability (Bello, 2012; Ramya and Muruganandham,
2016; CEMA, 2014; Reid, 2011). Moreover, level of farm mechanization is reported to affect
technical efficiency of producers positively (Vortia et al., 2019; Lei et al., 2016; Hormozi et
al., 2012).

Agricultural mechanization is considered as labor-saving technology (Pingali et al., 1987;

Braman and Deka, 2019; Billing and Singh, 1970; Srisompun et al., 2019; Caunedo and Kala,
2021). However, there are inconsistencies in some findings like Cossar (2019). His finding
showed that there is no change in the labor use for land preparation, and potentially an
increase in labor use for other operations due to use of tractor plowing. Insignificant labor
 49

displacement was also reported by Ramya and Muruganandham (2016) while animal power
displacement was significant accounting for 50 to 100% and even the displaced labor may be
compensated by new employment in machinery service provision areas.

Other study findings also attested that land productivity, agricultural production and
profitability are affected positively by the use of farm mechanization due to timeliness of
operation especially, planting and harvesting and efficient utilization of inputs (Singh and
Singh, 1980; Daum et al., 2020; Ahmed et al., 2020; Caunedo and Kala, 2021; Ramya and
Muruganandham, 2016; Baudron et al., 2019). However, study in Ethiopia showed that use of
tractor has no effect while combine harvester has significant positive impact on yield (Guush
et al., 2017).

2.4.3. Factors Affecting Farm Mechanization Level

Several factors that affect the level of agricultural mechanization have been investigated by
different researchers. According to Zhang et al. (2017), Yiang et al. (2006) and Hou (2008)
agricultural management, transfer of agricultural force, income policy, income level of
farmers, development level of agricultural machinery industry, and the cost of using
agricultural machinery products have significant impacts on the development of agricultural
mechanization. In Chinese agriculture it was concluded that the level of economic
development is the most important contributing to the development of mechanization (Liu and
Tian, 2009; Yan, 2014).

The study in Pakistan investigated that the level of education, public agricultural machinery
(external support for machinery), and subsidies had significant effects on increasing product
performance (Abbas et al., 2017). Potential agricultural productivity and other socioeconomic
indices that improve the sustainability of agricultural systems have also direct effect on level
of mechanization (Zangeneh et al., 2010). Another key factor affecting successful
mechanization includes socio-economic infrastructure and technical skills (Olaoye and Rotimi,
2010).
 50

Other mechanization constraints include awareness, knowledge and skill gaps, access to the
technologies, less extension service, less policy emphasis, high costs of agricultural machinery
and equipment due to fragmented, scattered, small sized, and traditional type of farms, lack of
improvement in industrial development planning of the country in order to meet the
mechanical needs of the agricultural sector, lack of proper structure and legal requirements for
mechanization development, lack of access to finance, and lack of quantitative and qualitative
mechanized trade unions and services (Tamrat, 2016; Guush, 2014; Guush et al., 2017).

2.4.4. Agricultural Commercialization: Level and Its Determinants

Different studies have been conducted both in Ethiopia and other countries where most of
them are specific in terms of commodity and study areas. Different econometric models were
also employed to identify both decision for participation in commercialization and level of
participation or commercialization level along with their determinants. Variables like farm
size, access to land markets, agro-dealers and traders with heavy carriage vehicles, household
head education status, sex of the household heads, household head age, extension contact,
credit use, the value of crops produced, perception on lagged crop price, geographical location
were found to exert positive and significant influence on market orientation indices of
households (Aromolaran et al., 2020; Alula et al., 2022; Tariku et al., 2021). von Braun et al.
(1991) also categorized determinants of commercialization as internal and external factors.

The external factors are beyond the smallholder’s control including population growth and
demographic change, technological change and introduction of new commodities,
development of infrastructure and market institutions, development of the non-farm sector and
the broader economy, rising labor opportunity costs, macroeconomic, trade and sectoral
policies affecting prices and other driving forces (von Braun et al., 1991; Pingali and
Rosegrant, 1995). Likewise, input and output markets, institutions like property rights and
land tenure, market regulations, factors affecting consumption preferences of households like
cultural and social factors, opportunities and constraints in production and marketing, agro-
climatic conditions, and production and market related risks are other external factors (Pender
et al., 2006). Internal factors include resource endowments such as land and other natural
capital, labor, physical capital, human capital and others.
 51

Goetz (1992) used Tobit and second-stage endogenous switching regression models to analyze
household food marketing behavior in SSA (Senegal) and found that commodity prices, grain
processing technologies, social cast, access to market information and dependency ratio were
determinant factors for market participation. Abafita et al. (2016) also analyzed smallholder
cereals farmers’ market orientation and participation in Ethiopian. They found that market
participation is enhanced by market orientation strongly. Furthermore, their result revealed
that level of crop production, land size, access to credit and all-weather roads enhanced market
participation whereas, age of household head and family size reduced participation. Berhanu
and Moti (2012) also analyzed the determinants of market orientation and market participation
in Ethiopia, and examined if market orientation translates into market participation. The result
revealed that Market orientation of smallholders in the study area is moderate. According to
Taye et al. (2019) market orientation of smallholder barley producers was significantly
affected by membership in cooperative, households’ livestock possession, land allocated for
barley, land fragmentation, access to market information and extension service in Ethiopia.

Pender and Dawit (2007) found that most producers are either autarkic or net buyers of food
crops while net buyers and autarkic households are poorer in many aspects than net sellers.
Their result attested that size of production was the most important factor for
commercialization level while access to roads, land, livestock, farm equipment, and traders
can improve production and commercialization. According to Dawit et al. (2006) market
participation is highest for cereal crops while commercialization level is highest for oil crops
followed by cereals and horticultural crops in Ethiopia whereas majority of households are net
sellers for all crop types. Oluwatayo and Rachoene (2017) reported that farm size, access to
mechanization technologies, government subsidies, monthly expenditure of farmers on farm
activities and number of hired labors were significant variables in influencing
commercialization in South Africa.

Other authors also found that family size, household head’s educational background, livestock
ownership, access to credit, production size of the crop under study, family labor, distance to
all-weather road, involvement in extension programs, amount of rainfall, altitude, distance to
 52

local and main market, market information access and land fragmentation were variables that
affect market orientation of a household significantly in Ethiopia (Berhanu and Moti, 2010;
Wondmagegn, 2013; Tadele et al., 2016). Household size also affected market participation
negatively and significantly in Vietnam, Guatemala and Nigeria (Rios et al., 2009; Awotide et
al., 2016) while in Tanzania only land size positively and significantly affects market
orientation (Rios et al., 2009). Awotide et al. (2016) also reported that household head
education, access to improved inputs and credit have significant and positive effect on
decision to participate in market.
2.4.5. Production Efficiency and Its Determinants

Different factors have been reported to affect the efficiency of smallholder farmers’ allocative,
technical and economic efficiencies. Efficiency analyses were conducted for different
commodities at different levels (in terms of geographical coverage) for both livestock and
crop. Major determinant factors of agricultural production efficiency were demographic,
socioeconomic, institutional factors like access to market information and credit, extension
services etc. and infrastructural facilities like road and market access.

Study result by Tadie et al. (2019) for red pepper farmers in Ethiopia attested that age,
education status, land size and fragmentation, extension service, credit access and market
information were significant factors that affect the level of TE. Other study revealed that sex,
education, extension visit, farm fragmentation, training, frequency of farm visit, farming
experience, road quality and credit amount, distance of farm from residence, ownership of living
home and livestock, off-farm income, household size and cooperative and other farmer’s
association membership were significant sources of technical, allocative and economic
inefficiencies in Ethiopia (Desale, 2017; Kifle et al., 2020; Wollie et al., 2018, Tolesa, 2021;
Adino and Tessema, 2020).

In other African countries like Uganda, Nigeria and Kenya AE, TE and EE were mostly
affected by variables like age, educational status, farm size, use of improved inputs,
experiences in farming, distance to trading and extension service centers, crop diversification,
use of hired labor, level of farm mechanization, and access to credit (Okello et al., 2019;
Okoye et al., 2007; Mburu et al., 2014; Vortia et al., 2019). In Brazil EE of broiler breeder
 53

poultry production was significantly affected by cost of electricity, as well as area of occupied
land, production scale, and feed intake per hatching egg of the broiler breeder farms (Carvalho
et al., 2015).

2.4.6. Empirical review on Effects and Determinants of Crop Diversification

[Link]. Economic effects of crop diversification

Crop diversification has been practiced for its different purposes like maximization of utility
gained from the use of resources like land, water and other; and for the overall agricultural
development in the country or at household level. It provides the farmers with optimal viable
options to grow different crops on their land (Ojo et al., 2014). The practice of crop
diversification has also advantages of avoiding production risk and uncertainty due to climatic
and biological vagaries (Saraswati et al., 2011). It can also be used to improve the incomes, to
provide gainful employment and to stabilize the income flow (Birthal et al., 2015). Besides in
several circumstances diversification or the introduction of a new cropping system is needed to
restore the degraded natural resource base or to enhance the value of natural resources
principally land and water (Faurès et al., 2013).

There is also the claim that diversification tends to stabilize farm income at a higher level.
This happens when the pattern of diversification is to accommodate more and more rewarding
crops, which is particularly important for the small farmers who strive to make their farms
viable (Saleth, 1995, Maggio et al., 2018; Arslan et al., 2018; Reardon et al., 2007).
Diversification can be used as an instrument to augment net farm income, reduce
unemployment, reduce poverty and conserve water resources and valued soil (Pingali and
Rosegrant, 1995). Generally, to increase farmers’ income, to curtail unemployment and
stabilize the income flow, diversification of crops emerges as a major important strategy (Sati,
2012).

Generally, this practice enhances farm income, minimizes risk in production and prices,
improve soil fertility, optimizing nutrients availability throughout seasons and marketing of
products at competitive national and international markets, supporting food security and
 54

employment, supplying diversified nutrient rich food over its demand, etc. Gbenga et al.
(2020) points that crop diversification has positive and significant effects on dietary
diversification of rural Nigeria. Furthermore, crop diversification, especially towards high
value crops such as vegetables and fruits, also reduces smallholder farmers’ poverty
incidences significantly (Birthal et al., 2015).

[Link]. Review of determinants of crop diversification

Seng (2014) divided determinant factors of diversification into: price factors, farmer’s
characteristics, farmer’s endowments, production risk and characteristics, and institutional
factors. In general, there are both push and pull factors of crop diversification (Barrett and
Reardon, 2000; Maggio et al., 2018). The push factors include inputs and output markets high
transactions costs, adverse shocks, including weather shocks, imperfect or absent credit and
insurance markets and stagnation in the agriculture sector (Barrett et al., 2001; Arslan et al.,
2018). Risk adverse households with limited resources to cope with production variability
associated with climate shocks may be more likely to be pushed into diversification strategies
than better endowed households (Arslan et al., 2018). In situations where push factors
dominate, crop diversification may not improve productivity or incomes, but may help to
stabilize them (Arslan et al., 2018; Barrett et al., 2001). On the other hand, the pull factors of
diversification are common under conditions of crop sector dynamism, where new market
opportunities or practices/technologies induce risk-taking farmers starts raising of new crops.

The results of the researches concerning the influence of socio-economic features of

smallholder farmers on the degree of crop diversification make it impossible to adopt a
uniform approach to determining the relationship between these features and the studied
phenomenon. Studies showed that variables such as age, education, farm size, off-farm income
generation, availing credit and family size found to be the most important factors that
positively affect crop diversification. However, tenure status, training program attended and
rainfall show significant and negative relationship with crop diversification, indicating that all
these three factors led to specialization. Moreover, social classification (caste-based division)
of society also affect diversification where those favored as superior are more diversifiers
 55

(Birthal and Hazrana, 2019; Mazzocchi et al., 2019; Kurdys-Kujawska et al., 2021).
Combination of factors such as year of schooling, labor equivalence of household size, tropical
livestock unit and number of parcels were the most important determinants that affect crop
diversity significantly (Dembele et al., 2018; Yemane et al., 2009) in Ethiopia. The study by
Benin et al. (2004) also attested that farm physical features and household characteristics such
as wealth and labor stocks have significant effects on crop diversity. Use of tractors/power
tillers, credit and per capita income from agriculture are significant determinant factors that
affect crop diversification (Kumar, 2020).

2.5. Conceptual Framework of the Study

Agricultural mechanization is important in production and productivity improvement and

symbol of agricultural modernization (Sims et al., 2016; Knickel et al., 2017; Liu and Tian,
2009). The modernization and mechanization of agriculture has multifaceted impacts on
different aspects of agricultural development which in turn has great role in economic growth
of a given nation. Farm mechanization is an innovation favored by policy support or induced
by shortage of labor and need for farm modernization at broader level. It can also be affected
by level of economic development of the nation, investment on the development of
agricultural mechanization, agricultural machinery supply, costs of agricultural machinery and
equipment and price proportion of agricultural output and machineries. Status of
mechanization at household level can be determined by different internal (demographic,
socioeconomic and knowledge and technical skills) and external (institutional, sectoral
policies and infrastructural) factors (Tamrat, 2016; Guush, 2014; Guush et al., 2017; Olaoye
and Rotimi, 2010).

It is conceptualized that farm mechanization has an impact on intensity of agricultural inputs

use which has positive impact on farm production and productivity. In addition, it has an
impact on production and productivity through timely and precise operation of agricultural
activities and efficient utilization of agricultural inputs. It is also hypothesized that
mechanization increases labor productivity and reduced per unit cost of labor and avail more
 56

time for leisure and other economic activity. Furthermore, it is conceptualized that the increase
in production and productivity will have positive impact on household income (Dey and Haq,
2009).

Farm commercialization is affected and driven by population size and demographic change;
urbanization; development of infrastructure and market institutions; development of the non-
farm sector and broader economy; rising labor opportunity costs; and macroeconomic, trade,
and sectoral policies affecting these forces at broader economic level (von Braun et al., 1994;
Pingali and Rosegrant, 1995). Agroclimatic conditions and risks; access to markets and
infrastructure; community and household resource and asset endowments; local commodity
development; input and factor markets; laws and institutions; and cultural and social factors
affecting consumption preferences, production, and opportunities and market constraints are
among the factors that affect commercialization at household level (Pender et al., 2006).

It is also hypothesized that smallholder agricultural commercialization is a process; primarily

pushed by key interrelated components called driving forces; determined by internal and
external factors (enablers and constraints); and is measured by input and output market
orientations. Market orientation and crop productivity are derived by internal and external
factors. Smallholder resource endowments, labor, physical and human capital are household
specific internal determinants factors (Moti et al., 2009).

Crop diversification is derived by factors related to institutions, resources, technologies,

socioeconomic, and infrastructure. Institutions related factors include credit, insurance,
Minimum Support Prices (MSP) while resource related factors include rainfall, irrigation, and
fertilizers. Technology related factors are High Yielding Variety (HVY) seeds and household
related factors like education, income from agriculture, attitude towards risk while
infrastructure related factors include telecommunication facilities, road and markets (Hazra,
2001; Inoni et al., 2021). It is hypothesized that both farm mechanization and
commercialization negatively affect crop diversification (Jha et al., 2009; Van den Berg et al.,
2005) i.e., both of them promote crop specialization towards more mechanized and
commercialized enterprises.
 58

Commercialization drivers
Demand promoting factors Determinants/enablers/
Internal factors (urbanization, population, constraints
Labor shortage income …etc) *External
Resource endowment
environmental factors
Household demography Environmental factors pushing *Farm level factors
to new production system *Household level
Efficiency promoting factors factors
Mechanization drivers
*Policy factors
(subsidy, cluster
farming, …)
*Access to machinery
*Economic Farm mechanization Farm commercialization Market
development participation
*Industrialization
*Higher wage

Diversification drivers
Crop diversification *Government policies
*Resource availability (land,
Mechanization impacts water, etc)
*Inputs use intensity (seed, *Technology availability
chemicals …) (mechanization, improved
*Household income seed . . . )
*Economic efficiency *Physical factors (soil,
ecology etc )
*Attitude towards risk

Figure 3. Conceptual framework of the study

Sources: Adapted from Li et al. (2018), Zhou et al. (2018) and illustrated based on literature
 59

3. METHODOLOGY OF THE STUDY

This third chapter of the dissertation provides details of the methodology of the research with
different subsections such as the description of the study area, sampling design, data type,
sources and method of collection, methods of data analyses. Each subsection is also
subdivided into other subsections for more elaboration when it is need to do so. The choice of
analytical models and variables are made in view of the empirical and analytical reviews
discussed under chapter two.

3.1. Description of the Study Area

Oromia is the largest national regional state in Ethiopia both in terms of population and land
size by having 35.1% and 32% of total national population and land area, respectively
(Aynalem, 2021). Geographically, it is located between 30 18’ 43’’ and 100 09’ 24’’ N latitude
and 340 18’03’’ and 430 04’ 33’’E longitude with diverse agro-ecological zones that are highly
suitable for agriculture (Drucza et al., 2021). Agriculture is the mainstay for Oromia and it is
the main employment source for more than 87% of its population. Agricultural products from
Oromia are partly supplied to national market. Of grain produced in Ethiopia, around 50% is
produced in Oromia region. Cereals are the most important crops in the region, which covers
around 84% of total grain crops land. Teff, maize, wheat and sorghum are the most important
cereals in terms of their area coverage and volume of production. Red peppers, Ethiopian
cabbages and green peppers are widely grown while root crops like potatoes, sweet potatoes
and onions are among the major vegetables in terms of area coverage (CSA, 2021a). In the
Oromia region, only 21.4% of cereal crops produced is marketed while largest proportion
(62.72%) is consumed at household level. Rice (36.84%), teff (32.2%), wheat (28.12%) and
oat/‘aja’ (23.78%) are the most commercialized cereals while oil seeds of which around 50%
is sold are highly commercialized crops followed by pulses (CSA, 2021b).

The study area, Arsi and west Arsi zones, are found in central and southern part of Oromia
national regional state, respectively. Geographically, Arsi zone lies between 60 45ʹN to 80
58’N latitude and 380 32 E to 40050ʹ E longitude while west Arsi zone lies between 6012ʹ29”
 60

to 7042ʹ55” latitude and 38004ʹ04” to 39046ʹ08” longitude. Because of its great diversity in
altitude, Arsi zone has great physiographic diversity. Based on the altitude there are four major
identified physiographic divisions which are cool; cool temperate; warm temperate and
lowlands agro-ecological zones (AEZs) (Tamrat et al., 2019; OBOFED, 2011). Similarly, west
Arsi zone also possesses three major AEZs - highland, midland and lowland (OBOFED,
2011).

Figure 4. Geographical location of the study area (Arsi and West Arsi zones)

Both zones have complex and interlinked and diversified mixed crop-livestock farming
systems. For instance, Tamrat et al. (2019) broadly categorized farming typologies of Arsi into
seven, which are lowland agropastoral, coffee-khat tree based, irrigation based, highland
barley-root crop based, maize-sorghum based, mechanized/semi-mechanized wheat belt and
non-mechanized wheat belt. However, there is also dynamism of farming system in the zones
 61

and following current government policy change in the sector, introduction of mechanization
is becoming common. Both zones are commonly known for cereal and root crops production
and huge potential of livestock with general mixed farming systems.

In Arsi zone, 39.81% of the land under cereal crops is covered by wheat while teff, barley and
maize are the next important crops in terms of area coverage. In line with this, faba bean and
field peas are the major pulse crops in terms of area coverage, while linseed covers 91.2% of
area under total oilseeds. Vegetables like Ethiopian cabbage and green peppers are widely
grown while onion and garlic are known from root crops (CSA, 2021a). In west Arsi cereals
are the major crops, which cover around 90% of grain covered area. Among others wheat is
the widely grown crop in terms of area coverage by covering 41.59% of total cereal crops and
37.5% of grain crops grown. Barley, maize and teff are other major cereals while red haricot
bean and faba bean are major pulse crops in the zone. Ethiopian Cabbage and Potatoes are
other major vegetable crops in this zone (CSA, 2021a).

About 28% grain crops and 24.38% of cereal crops are commercialized in Arsi. White and red
chickpeas, white haricot beans and field peas are produced and sold largely from pulse crops
and from oilseed crops largely produced and sold crops are groundnut and linseeds in both
zones. Red pepper, Ethiopian cabbage and head cabbage are vegetables grown widely and sold
in large proportions while garlic and potatoes are among the root crops (CSA, 2021b).

3.2. Sampling Design and Sample Size Determination

The study considered households as a decision-making unit and considered farm household
heads as key respondents and decision points of the household’s farm and non-farm activities.
Hence, the unit of analysis in this study was a household. A multi-stage sampling procedure
was followed to draw the target sample households. In the first stage, two zones namely Arsi
and West Arsi were purposively selected as they represent the central and southern parts of
Oromia region and where adoption of farm mechanization is considerably high. Due to
similarity in terms of agroecological zones and farming systems, the two zones were highly
homogenous and sampling of districts and kebeles were designed accordingly. The study
 62

mainly focused on highland and midland districts of the study area where adoption of farm
mechanization is relatively high. In the second stage, four districts, two from each zone,
namely, Kofele and Gedeb-Hasasa from West Arsi and Lemu-Bilbilo and Hetosa districts
from Arsi were selected randomly. Kofele and Lemu-Bilbilo are highland districts while
Hetosa and Gedeb-Hasasa are midland districts. At the third stage, systematic simple random
sampling procedure was employed to draw two kebeles from each district and accordingly a
total of eight kebeles were selected. Finally, the total sample size was determined using a
formula, which provides a representative size to ensure the desired precision using the formula
given by Kothari (2004) as follows:

Z 2 pq (1.96) 2 (0.5)(0.5)
N= =  385
e2 (0.05) 2

Where, N is the desired sample size; Z is the standard cumulative distribution that corresponds
to the level of confidence with the value of 1.96; e is desired level of precision; p is the
estimated proportion of an attribute present in the population with the value of 0.5 as
suggested by Israel (1992) to get the desired minimum sample size of households at 95%
confidence level and ±5% precision; q=1-p. Accordingly, a sample of 385 were proposed and
finally 390 respondents were interviewed using systematic random sampling with probability
proportional to size (Table 1).
 63

Table 1. Distribution of the sample by zones, districts, and Kebeles

Districts Total HHs Kebeles Total HHs Sampled HHs

Hetosa 17788 Shaki 906 42

Boru-lencha 981 46
Lemu-Bilbilo 22455 Dawa-bursa 825 64
Chiba-Michael 684 51
Kofele 22486 Gurmicho 626 41
Guchi 758 42
Gedeb-Hasasa 27842 Dabara-walta’i 1015 50
Huruba 1062 54
Total HHs 90571 6,857 390

Source: Author’s computation based on secondary data from respective district (2022).

3.3. Data Sources, Types and Methods of Collection

The study was conducted based on data from primary and secondary sources. Primary data
sources were primarily peasant households, mechanization service renders (cooperatives or
private machinery owners) and government’s policy documents. Secondary data was collected
from published and unpublished sources like CSA, official reports, and journals. Primary data
was collected by face-to-face personal interviews using structured and semi-structured
questionnaires. The collected data types determined based on the need of each and every
research objective. Both qualitative and quantitative data that are pertinent to the study was
collected by using both open and close-ended types of questions. Given the complexity of the
constraints in Ethiopia’s agriculture modernization process, sole approach of research method,
either qualitative or quantitative, is not enough to grasp the major causes and effects of the
subjects under study. As a result, when quantitative approach has no enough power to do so, it
is recommended to supplement by qualitative methods (Doorenbos, 2014). Hence, in addition
to quantitative method, qualitative data was collected by using Participatory Rural Appraisal
 64

(PRA) method through FGD and KII. Production cost and level of farm mechanization data
were collected for major crop.

Questionnaire was pre-tested on 20 randomly selected farmers from Kubsa kebele in Gedeb
Asasa districts prior to the formal survey for testing appropriateness and relevance. FGD at
each kebele and KII was conducted to enrich and triangulate the data from the questionnaires
and the field survey was conducted by researchers from Asella agricultural research center
after training and thorough briefings.

3.4. Methods of Data Analyses

Both simple descriptive and econometric models are employed. Accordingly, in accordance
with each objective, the necessary analysis methods are discussed in subsequent sub-sections.

3.4.1. Descriptive Statistics

To analyze the data, both descriptive analysis and econometric models were used. Descriptive
analysis included mean, standard deviation, frequency and percentage. Furthermore,
inferential statistics such as t-test, chi square test, where it is applicable, ANOVA tests are
employed to substantiate the previous descriptive statistics.

3.4.2. Estimating the Farm Mechanization Level

Level of farm mechanization was estimated by using Mechanization Index (MI). Following
Singh (2006) MI based on the matrix use of animate and mechanical energy inputs could be
given by incorporating cost factors in equation (15):

CMi
MI = (15)
CHi + C Ai + CMi

Where MI is mechanization index in %; CMi, CHi, and CAi are costs of using machinery, human
labor and animal power by ith household per hectare, respectively.
 65

3.4.3. Econometric Models

[Link]. Determinants of farm mechanization level: Tobit model

Level of mechanization is an index that lies between zero (non-mechanized) and 1 completely
mechanized. Therefore, to identify factors that determines level of farm mechanization, OLS
will be biased and inconsistent since some values of MI will be zero (Nasrin et al., 2018), and
hence instead, a censored regression model developed by Tobin (1958) called Tobit, is
employed in this study. Therefore, following Maddala (1983), the Tobit model can be
specified as (16):

𝑌𝑖∗ = 𝛽′𝑋𝑗𝑘 + 𝑢𝑖 (16)

If Yi is denoted as the observed dependent (censored) variable, then it can be given as (17):
𝑌𝑖∗ = 𝐿 𝑖𝑓 𝑌𝑖∗ ≤ 𝐿
𝑌𝑖 = {𝑌𝑖∗ = 𝐿 𝑖𝑓 𝐿 < 𝑌 ∗ < 𝑈} (17)
𝑌𝑖∗ = 𝑈 𝑖𝑓 𝑌 ∗ ≥ 𝑈
Where: Yi is the observed dependent variable, in this case level of farm mechanization of
household i (unobserved for values smaller than 0 and greater than 1), Xjk is a vector of
explanatory variables for household k (l = 1, 2, …..., j) and ui an error term that is

independently and normally distributed with mean zero and variance  2 which is assumed to
be independent of Xjk. The distribution of the dependent variable in equation (16) is not a
normal distribution because its value varies between 0 and 1. The likelihood function of this
model, following Maddala (1983) is specified as (18):

L1 j −  * X j yi −  * X j L2 j −  * X j
L(  ,  / y j X j L1 j L2 j ) =  yj − L1 j ( ) yj = y . (
*
) yj − L2 j  ( ) (18)
 j
 

Where: L1j = 0 (lower limit) and L2j = 1 (upper limit) are normal and standard density
functions.
 66

The regression coefficients of the two-limit Tobit regression model cannot be interpreted like
other OLS regression coefficients that give the magnitude of the marginal effects of change in
the explanatory variables on the expected value of the dependent variable. However, in a two-
limit Tobit model, each marginal effect includes both the influence of explanatory variables on
the probability of the dependent variable to fall in the uncensored part of the distribution and
on the expected value of the dependent variable conditional on it being larger than the lower
bound. Thus, the total marginal effect takes into account that a change in explanatory variable
will have a simultaneous effect on the probability of being highly mechanized and value of
farm mechanization level scores for a given household.

For the reason already mentioned, the coefficients of the Tobit model cannot be interpreted
directly, and as result estimating marginal effect is important. Based on McDonald and Moffitt
(1980) decomposition techniques of total marginal effects, from the likelihood function of the
model stated in equation (18), the three marginal effects of the Tobit model are:

1. The unconditional expected value of the dependent variable (19):

E ( y* ) E ( y* ) [ ( ZU ) −  ( ZU )] (1 −  ( ZU )
= [ ( ZU ) −  ( Z L )] + + (19)
x j x j x j x j

2. The expected value of the dependent variable conditional upon being between the limits
(20):

E ( y* )  {Z  ( Z L ) − ZU  ( ZU )} {( Z L ) − ( ZU )}2 

=  1 + L −  (20)
x j  { ( ZU ) −  ( Z L )}   { ( ZU ) −  ( Z L )}2 

3. The probability of being between the limits (21):

[ ( ZU −  ( Z L )]  k
= [ ( Z L ) −  ( ZU )] (21)
X j 
 67

Where (.) is the cumulative normal distribution, (.) is the normal density function, ZL=
Z L = − ' X /  and ZU = (1 − X ) /  are standardized variables that come from the likelihood
function given the limits of y*, and  is the standard deviation of the model.

[Link]. Economic efficiency analysis

A Stochastic Frontier Analysis (SFA) approach is employed to address the objectives of

measuring economic efficiency and its determinants. SFP model was first proposed by Aigner
et al. (1977) and then adopted by Battese and Coelli (1995). The model is given as (22):

Yi = f ( X i ,  ) + i −u i (22)

Where: Yi is the quantity of crop output of the ith household, Xi is a vector of the inputs used
by ith household on the selected specific crop, 𝛽 are a vector of unknown parameters, f (Xi;
𝛽) is a suitable production functional form. The random disturbance term,  i captures the
effects of the stochastic noise, is assumed to be Independently and Identically Distributed
(IID) and expressed as N (0,  u2 ). On the other hand, the disturbance term ui, captures

technical inefficiency and is assumed to be independent of the random term  i and are
assumed to be iid as half-normal, u ~ |N (0,  u2 )|. The variance parameters are expressed as

 u2
 =  u2 + 2 and  = . The values of  ranges from 0 to 1, such that the value of zero
2
corresponds with the traditional average response function for which the nonnegative random
variable, ui, is absent from the model (perfect efficiency in production) while the value of one
indicating that all the deviations from the frontier are entirely due to technical inefficiency (the
random error on production is zero).

The parameters of stochastic frontier production model are commonly estimated using the
Maximum Likelihood Estimation (MLE) techniques. The assumptions made on the statistical
distributions of  and u enables us to calculate the conditional mean of ui given as (23):
 68

 i =  i − i (23)

TE of an individual household for specific selected crop is estimated as the ratio of the
observed output (for crop considered) Yi to the corresponding frontier output Yi f , both are

given in the original units. That is, the measure of technical efficiency is given as (24):

Yi f ( X i ;  ).exp( i − ui )
TE = = = exp(−ui ) \(24)
Yi f
f ( X i ;  ).exp( i )

The SFP model estimated using equation (22) only measures the TE. However, to measure EE,
the measurement of AE is important since EE is the product of TE and AE. The measurement
of allocative (and hence economic efficiencies) can be investigated through applying
efficiency decomposition techniques, which works under the framework of SFP model first
proposed by Kopp and Diewert (1982) and developed by Bravo-Ureta and Rieger (1991). The
method is used to decompose the overall efficiency into TE and AE. This technique is
generally based on the duality between production and cost functions. The parameters of the
SFP model are actually used to derive the parameters of the duality cost function. From the
original SFP model given in equation (22), if  i is subtracted from both sides of equation (22),

we obtain equation (25):

Yi * = f ( X i* ;  ) − ui = Yi −  i (25)

Where: Yi * is the ith household observed output adjusted for the statistical noise captured by

 i , f (•) is the deterministic frontier output and u and  are the inefficiency and random
component of overall deviation from the frontier, respectively. The adjusted output Y* is used
to derive the TE input vector, X*. It is possible to derive the technically efficient input vector
for ith household for a crop by simultaneously solving equation (24) and an observed input
ratio X1/Xi=k (i>1) where ki is equal to the observed ratio of the two inputs in the production
of Yi * . This derivation of the technically efficient input vectors paves the way for deriving the
TE measures by taking the ratio of vector of the norms of the efficient and observed input
 69

quantities. On the other hand, the adjusted output is used to derive AE and EE employing the
dual cost frontier function that is analytically derived from the stochastic frontier production
model.

If it is assumed as usual that the production function in equation (22) is self-dual (as it is true
for the Cobb-Douglas production function), the corresponding parameter of the dual cost
frontier can algebraically be derived and written in a general form in equation (26) as:

𝑙𝑛𝐶𝑖= (𝑤𝑖, Yi * ; 𝛼) (26)

Where: Ci is the minimum cost of the ith crop farmer associated with the output Yi * , wi is a

vector of input prices for the ith household (for specific crop), 𝛼 is a vector of parameters to be
estimated from the primal function. The economically efficient input vector for the ith
household, X ie can be derived by applying Shephard’s Lemma and substituting the

household’s input prices and adjusted the output level into the resulting systems of input
demand equations (27).

Ci
= X ie (Wi , Yi* ; ) (27)
Wi

Where: i=1, 2, ..., n are inputs used for production.

Given that the cost of the household actual operating input combination as Wa' X i , the

observed and economically efficient costs of production of the ith household are equal to
Wa' X i and Wi ' X ie , respectively which are used to compute TE and EE induces for the ith

household crop production as given in equations (28) and (29), respectively.

Wi ' X it
TEi = (28)
Wi ' X i
 70

Wi ' X ie
EEi = (29)
Wi ' X i

Finally, since TE and EE are known, it is easy to derive AE from equations (28) and (29)
based on Farrell (1957) which is given as:

Wi ' X ie
AEi = (30)
Wi ' X it

Production efficiency estimation

Following the Aigner et al. (1977) method, the analysis of efficiency will be estimated by the
SFP model that has been specified in equation (22) earlier. A Cobb-Douglas production
function form which describes the transformation relationship from input to output
(Kumbhakar and Horncastle, 2015) is employed for the crop producer households. Cobb-
Douglas production is widely used production function in agricultural production (Zhang et
al., 2015; Imad et al., 2019, Min et al., 2021) which is mainly due to its simplicity over
translog. A translog function has problems of multicollinearity and degrees of freedom
problems that can arise due to a substantial number of parameters to be estimated. Hence, use
of translog function especially with input variables of more than three is difficult for
computation. Moreover, parameters of not directly interpreted because of the second-order
terms involved in the function and additional calculations are needed to get the partial outputs
elasticities of individuals. Hence, production technology for wheat/barley crop is defined as
follows:

ln Yi = ln  0i + 1 ln X 1i +  2 ln X 2i + ... + 8 ln X 8 +  i − ui (31)

Y, represents amount of specific crop produced in (kg); X1 to X8 represent amounts of

production inputs which are farmland (ha), labor in person days, seed (Kg), agrochemicals
(litters), fertilizers (Kg), oxen labor (Oxen-days), tractor rent (Birr/ha), and combine harvester
rent (Birr/qt). ln is natural logarithm; βj’s are unknown parameters to be estimated; vi’s are
symmetric component of the error term and assumed to be independent and identically
 71

distributed with distribution of N (0, σ2); the ui’s are the inefficiency component of the error
term, that are assumed to be independently distributed such that ui is defined by truncation (at
zero) of the normal distribution with mean μi and variance σ2 (Temesgen and Ayalneh, 2005),
where μi is defined as (32):

ui =  0 +  j =1 j Z ji + Wi
n
(32)

Where,  j is parameter to be estimated Zj are socioeconomic, institutional and demographic

variables that are assumed to determine inefficiency component of the model.

The Cobb-Douglas cost frontier function for selected crop production is specified as (33):
ln C =  0 + 1 ln P1 +  2 ln P2 + ... + ln P8 + i + ui (33)

Where, C is total cost of production of a given crop (ETB);

P1 = is cost of land rent (ha)
P2 = is cost of labor per hectare
P3 = is cost of seed for the crop under consideration per hectare
P4 = cost of agrochemicals used in crop production per hectare
P5 = cost chemical fertilizer applied for the crop production per hectare
P6 = cost of tractor renting per hectare
P7= combine renting price per hectare
P8= thresher renting price per hectare
𝛼j, j = 0, 1..., 6 are parameters to be estimated;

If a firm is both allocatively and technically efficient, it is then cost efficient (economic
efficient/overall efficient) (Badunenko et al., 2008). Hence, cost frontier approach of
efficiency analysis was directly used in analyzing economic efficiency.
 72

[Link]. Impact of farm mechanization on EE, yield, income and inputs use intensity

Information and data related to impacts of farm mechanization on different outcome variables
like farm inputs (chemical fertilizer, agrochemicals and labor) use intensity, economic
efficiency of wheat and barley production, wheat and barley productivity and household
income was analyzed and synthesized using different statistical and econometric tools. An
estimator for average treatment effects (ATEs) known as the AIPW estimator method was
used to analyze impacts of treatment variable level of farm mechanization.

Specification of AIPW estimator

The hypothesis to be tested under this section is whether level of farm mechanization has a
significant contribution on intensity of agricultural inputs (seed, labor, chemical fertilizer and
agrochemicals) use, household economic efficiency and households’ income. The treatment
variable (level of farm mechanization-MI) is continuous variable lying between 0 (zero) and 1
(one) which implies that it involves dose-response treatment. The outcome variables EE is
ratio (index) variable, household income and intensity of inputs use (fertilizer, amount of labor
and agrochemicals) are interval.

It is obvious that self-selection and other measurement errors are common problems of
observational data that has been tried to be overcome by PS matching methods like pscore,
IPW and others. However, these estimators are not both robust and consistent. Hence, an
estimator known as AIPW was developed (Robins et al., 1994; Robins, 1999; Scharfstein et
al., 1999) and under use recently by social researchers (Glynn and Quinn, 2009). AIPW
estimators combine aspects of regression-adjustment and inverse-probability-weighted
methods. AIPW estimators have the double-robust property. The estimator ‘teffects aipw’
accepts a continuous, binary, count, fractional, or nonnegative outcome and allows a
multivalued treatment. The AIPW estimator has a property called ‘‘double robustness,’’
meaning that it is consistent (i.e., it converges in probability to the true value of the parameter)
for the ATE if either the propensity score model or the outcome model is correctly specified. It
also solves the problem of poor performance in IPW especially with smaller data size (Raad et
al., 2020).
 73

AIPW estimator instead of modeling either the outcome, like regression adjustment (RA), or
the treatment probability, like IPW, models both the outcome and the treatment probability.
An interesting part of this estimator is that only one of the two models must be correctly
specified to consistently estimate the treatment effects, a property of the AIPW estimators
known as being “doubly robust” (Robins et al., 2000). A three-step approach to estimating
treatment effects in AIPW are: 1. Estimate the parameters of the treatment model and compute
IPW. 2. Estimate separate regression models of the outcome for each treatment level and
obtain the treatment-specific predicted outcomes for each subject. 3. Compute the weighted
means of the treatment-specific predicted outcomes, where the weights are the inverse-
probability weights computed in first step. The contrasts of these weighted averages provide
the estimates of the ATEs. In AIPW estimation we are interested in three parameters: the
potential-outcome mean (POM)  t = E( yt ) , 2. The average treatment effect (ATE)

 t = E ( yt − y0 ); and 3. The average treatment effect on the treated (ATET)

 t = E( yt − y0 ) | t = ~t .

For the different level of treatment ranging from 0 to G; where zero typically indicating
control group and 1 to G indicating different level of treatment, then wi(treatment level) takes
a value in {0, 1, . . . G}. Hence, there are G+1 counterfactual outcomes that can be denoted for
random draw i, {yig: g=0,1, . . .G}. The observed response, yi, can be expressed as (34):

yi = 1[wi = 0] yi 0 + 1[wi = 1] yi1 + .... + 1[wi = G] yiG (34)

Let also Xi be a set of covariates and  g = E ( yig ) a population means of counterfactuals. And

the conditional mean independence assumption for identifying the means is (35):

( yi | wi ,  i ) = ( yig |  i ), g = 0,1,...., G. (35)

Under this above assumption, the following will be true (36):

 74

E( yi | wi , X i ) = 1[wi = 0]E( yi 0 | X i ) + 1[ E( yi1 | X i + ...... + 1[wi = G]E( yiG | X i ) (36)

Finally, following Robins et al. (1994), Wooldridge (2010) and Glynn and Quinn (2010) the
AIPW model can be estimated as follows (37):

( X i−ˆ ( Zi )
 (1− X i )Yi
[ ˆ ( Zi ) − 1−ˆ ( Zi ) ] − ˆ ( Zi )(1−ˆ ( Zi ))
X iYi


ATE AIPW = 
1
n  (37)
ˆ ˆ
[(1 − ˆ ( Z i ))(Yi | X i = 1) + ˆ ( Z i )(Yi | X i == 0, Z i )]
 

Where Z contains information about the probability of treatment and predictive information
about the outcome variable. The first line of equation (4) corresponds to the basic IPW
estimator, and the second line adjusts this estimator by a weighted average of the two
regression estimators.

For the test of model validity, covariate balance test and treatment effect overlap check were
also conducted. These two assumptions are generalized as ignorability assumptions.
Ignorability is a combination of the selection-on-observables assumption (covariate balance)
and a no-empty-cell assumption (treatment effect overlap) and it allows us to recover the
parameters of the unobserved unconditional distribution from the observed conditional
distribution. The first assumption implies that the distribution of each potential outcome y(j) is
independent of the random treatment t(j), conditional on the covariates x and mathematically
specified as (38):

Y(j) ⊥t(j)|X (38)

where “⊥” means “independent of” and “|” denotes “conditional on”. The second assumption
of ignorability states that for every possible x in the population, there is a strictly positive
probability that someone with that covariate pattern could be assigned to each treatment level
and mathematically expressed as (39):

Pj ( X ) = P ( w = j | X ) (39)
 75

Both assumptions, covariate balance and treatment effect overlap, were checked by using
‘tebalance’ and ‘teoverlap’ commands respectively and the model was valid for both
assumptions.

[Link]. Measurement and determinants of households’ crop commercialization

Major crops grown in the study areas including two dominant ones, wheat and barley, and
other pulse crops (faba bean, peas, and chickpeas) and potato from root crops were considered
for this objective based on their area coverage and marketability. The level of crop output
commercialization at a household level was computed as the ratio of the value of agricultural
outputs sold to the total value of agricultural outputs produced by a household (Strasberg et
al., 1999; Berhanu and Moti, 2010; von Braun and Kennedy, 1994). For this study, major
crops: wheat, barley, potato, and pulse crops (faba bean, peas, and chickpeas) were considered.
Hence, the crop output side commercialization index for a household (COMPI) was computed
as (40):

COMPIk =  P− k ; Qki  S ki
6Sik
i =1
P Qik (40)

Where; COMPIk is the proportion of crop k sold (Ski) to the total amount produced (Qki) by a
−
household i evaluated at an average price ( P) given and reported by a household during

selling.

The index measures the intensity of a household’s participation in the market. A value of zero
would signify a totally subsistence-oriented household; the closer the index is to 100, the
higher the degree of commercialization. Based on suggestions of some authors (Samuel and
Ludi, 2007; Samuel and Sharp, 2007; Bedaso, 2012) a household is commercially oriented if
the crop output commercialization index (COCI) is a minimum of 65% whereas values in
between 30% and 65% termed as semi-commercial and otherwise non-commercial.
 76

Household level market participation (COMP) and intensity of market participation are
modeled as a function of household characteristics (sex, age, and educational background of
household head), household resource endowment, and technology use which are family labor
supply, household size, landholding, land cultivated, working animal availability (draft power,
equines, and other livestock, level of mechanization), access to physical infrastructures
(distance to main market and access to all-weather road), access to institutional services
(extension, credit, and market information), and land fragmentation, the spatial location of the
household (zone), level of crop diversification, total annual household income.

Specification of the Heckman two stage model

The most widely used models to investigate a household’s decision for market participation
and its level of participation (level of commercialization) are the double hurdle model (DHM),
Heckman’s two-stage model, Tobit model, and multinomial logit models (Efa et al., 2016;
Geoffrey, 2014; Tobin, 1958). However, each model has its own limitations and advantages
while the choice of one over the others depends on the objectives of the study and data type.
The censored regression or Tobit (Winship and Mare, 1992) model is appropriate when the
dependent variable is censored at some upper or lower bounds. Tobit model is criticized due to
the fact that it cannot separate the participation and intensity of participation decisions. It
assumes that factors affecting market participation also affect the intensity decision in the
same way. Second it assumes the zero values in the intensity equation as a corner solution.
However, the zero value of the level of commercialization may be due to a discrete choice of
not to participate in the commercialization decision. Hence, Heckman two-step and Cragg’s
models are the two best options for this analysis.

There are also critical assumption differences between the Heckman’s model and Craggit
model. In Heckman model the estimation of the first stage equation of the model as a probit
model assumes that the errors are Homoscedastic (Rufino, 2016). Furthermore, Heckman
assumes that there is no zero observation in the dependent variable of the second stage once
the first stage is passed whereas the Craggit model still considers that there might be a
 77

possibility of zero observation. However, in this study once a household decide to participate
in the crop output market, there is no possibility that the amount of crop sold can be zero.
Hence, based on this assumption Heckman two-step is better for this data. The second
assumption of Craggit model is that there is independence of hurdles while Heckman’s two-
step assumes that there is dependence of the hurdles (Humpheys, 2010). The dependence
between the two hurdles can be tested based on the significance of Inverse Mill’s Ratio (IMR).

In this case, even though this model assumes that the decision to participate in the market and
the intensity of market participation may not necessarily be jointly determined (Nuri et al.,
2016), the Heckman selection model is appropriate if there is a censoring process in measuring
the intensity of participation. This implies that the Heckman procedure assumes that there are
some potential market levels in the sample population that is not observed due to sample
selection problems. In general, Heckman’s sample selection model is designed to account for
the fact that the observed sample may be non-random and the market participation equation
can be specified as (41):

Pr(Zi = 1 | wi , ) = (h(wi , )) +  i (41)

Where Zi is an indicator variable equal to unity for households that participate in crop sales; Φ
is the standard normal cumulative distribution function; wi is a vector of factors affecting
market participation; α is a vector of coefficients to be estimated; and ε i is the error term
assumed to be distributed normally with a mean of zero and variance of σ2. The variable Zi
takes the value of 1 if the marginal utility that the household i gets from participating in the
crop market is greater than zero, and zero otherwise. So, we have equation (42):
Z i* = wi +  i (42)

Where Zi* is the latent level of utility that the household gets from crop market participation
(43),  i ~ N(0,1) and,
1 𝑖𝑓 𝑍𝑖 = 1 𝑖𝑓 𝑍𝑖∗ > 0
𝑍𝑖 = { } (43)
0 𝑖𝑓 𝑍𝑖 = 0 𝑖𝑓 𝑍𝑖∗ ≤ 0
 78

In the second stage, the Inverse Mills Ratio (IMR) will be added as a regressor in the sales
function or level of participation to correct for potential selection bias. To get rid of
endogeneity problem this selection bias, Heckman suggested first to estimate  via
maximum likelihood estimation (MLE) probit, typically using an exclusion restriction (only
households who participate in the market are included in the second stage), then to estimate an
IMR, which essentially tells us the probability that household decide to participate in the
market over the cumulative probability of a household’s decision, i.e. (44).
^
^  (h( wi , )
= ^
(44)
( wi , )

Where ϕ (.) is the normal probability density function. The second-stage (level of sales)
equation is then given by (45):


 (h( wi ,  ))
E (Yi | Z = 1) = f ( X i ,  ) +   (45)
( wi ,  )

Where E is the expectation operator, Y is the level of crop commercialization, X is a vector of

independent variables affecting sales, and β is the vector of the corresponding coefficients to
be estimated. Then, Yi can be explained as (46):

^
Yi =  ' X i +   + i (46)

Where, u ~ N (0,   ), Yi* is only observed for crop sellers (Zi=1), in which case Yi=Yi*

[Link]. Measurement and determinants of crop diversification

To identify determinants of crop diversification, first households’ level of crop diversification

was estimated using indices, which measures evenness (equitability). For this purpose,
Simpson Diversification Index (SDI) was selected due to its simplicity for computation
(Khatun and Roy, 2012), robustness (Sahal and Baha, 2010) and it is also widely used in the
 79

literature of crop diversification (e.g., Vani and Pavithra, 2021; Aneani et al., 2011; Nagpure
et al., 2017, Uddin, 2019). SDI is the difference between one and the sum of squares of all the
proportion of a particular crop involved in a particular household represented as:

n
SDI = 1 −  Pi 2 (47)
i =1

n
ai
The crop diversification index can be defined as: P =  (48)
i =1 A

Where ai = amount of land involved in a particular crop item produced by household i in a

given time period, and A= total land owned by household i. Value of SDI ranges from zero to
one, where value near to zero means the household is specialized while one means fully
diversified household. The diversification status of the households will be classified based on
the rating given by Sahal and Baha (2010) low (0 to 0.38), medium (0.39 to 0.63) and high
(above 0.63). To identify the determinant factors of level of diversification Ordered Logit
model is proposed.

The central idea in Ordered Logit is that there is a latent continuous metric underlying the
ordinal responses observed by the analyst. Thresholds partition the real line into a series of
regions corresponding to the various ordinal categories. The latent continuous variable, y* is a
linear combination of some predictors, x, plus a disturbance term that has a standard Normal
distribution. The dependent variable will be SDI value labeled 1, 2 and 3 representing low,
medium and highly diversified respectively. And the model can be specified as (49) and
estimated using maximum likelihood estimation (MLE) method.

y * = X i '  +  i ,  ~ N (0,1), i = 1,...N (49)

Where yi, the observed ordinal variable, takes on values 0 through 3 according to the following
scheme:
𝑦𝑖 = 𝑗 𝑖𝑓 𝑎𝑛𝑑 𝑜𝑛𝑙𝑦 𝑖𝑓 𝛼𝑗−1 < 𝑦𝑖∗ ≤ 𝛼𝑗
The probability that a household ‘i’ will fall in an alternative j is (50):
 80

Pij = P ( yi = j ) = P ( j −1  yi*   j ) = F ( j − X i '  ) − F ( j −1 − X i '  ) (50)

For the ordered logit, F is the logistic cumulative distribution function F (Z ) = e z /(1 + e z ) .
The interpretation of coefficient in ordered logit is similar to normal logit model where the
sign of the coefficients shows whether the latent variable y* increases with the regressor. The
magnitude of the coefficients has no meaning and marginal effect value is important to
estimate the likelihood of a given observation to be in a certain category such as low, medium
or higher. The marginal effect of an increase in a regressor xr on the probability of selecting
alternative j is shown in (51):

Pij / xri = [{F ' ( j −1 − X i '  ) − F ' ( j − X i '  )} r (51)

Where the marginal effects of each variable on the different alternatives sum up to zero. The
marginal effects interpretation follows: each unit increase in the independent variable
increases/decreases the probability of selecting alternative j by the marginal effect expressed
as a percent.

[Link]. Inter-linkage among farm mechanization, commercialization and diversification

There is strong hypothesis about the effects of one variable on the other among level of farm
mechanization, crop output commercialization and crop diversification. The review of
empirical literatures on this area also revealed that there is interlinkage among the three
economic variables where one affects the other and vis-versa. For instance, the finding in a
province of China shows that the level of mechanization has a significant positive impact on
the cost, output value, income and return rate of all types of crops (Peng et al., 2022). The
integrated use of irrigation technologies and other farm mechanization technologies like
tractor and combine harvesters, increases cropping intensity and thereby increase production
and commercialization level of farms (Tan, 1981). Adetule et al. (2021) analyzed the nexus of
agricultural commercialization among farming households in Nigeria and they found that farm
mechanization has significant impact on level of crop commercialization in the study area. it
 81

was also reported that lack of farm machinery is having a significant impact on farm
productivity in Kyrgyz Republic (Guadagni and Fileccia, 2009). However, the experimental
finding of Caunedo and Kala (2021) in Karnataka didn’t show any significant impact of farm
mechanization on productivity. Commercialization of crop can also have significant impact on
household’s welfare at household level and macroeconomic level that affects the income and
technological use of households and crop diversification. Detail discussion of the impact of
commercialization on technology use and crop diversification was given by Moti et al. (2009).

SUR is a method of estimating two or more linear regression models simultaneously when
there is correlation among error terms in OLS model. SUR estimator is a model developed by
Zellner and it accounts for the correlations in the error terms for OLS models and allows
different dependent variables to have different sets of independent variables. In SUR method,
the parameters of all equations are estimated simultaneously by taking information provided
by the other equations into account (Zeller, 1962). The SUR model output will be efficient
than OLS and the efficiency will increase with the increase in correlation among the equations
of OLS and larger sample size and higher multicollinearity between the independent variables
(Judge et al., 1988; Yahya et al., 2008).

The SUR model is a generalization of multivariate regression using a vectorized parameter

model. The OLS estimation assumes the non-correlation of error terms of different equations
which may not be true in most cases. However, in the presence contemporaneous correlation
among the error terms of different OLS equations, one has to take the presence of this
correlation and follow an estimation procedure that takes it into account (Heidari et al., 2017).
In this case, the SUR estimator leads to efficient parameter estimates (Yahya et al., 2008).
There are two cases where the SUR model is equivalent to OLS method: when the errors are
uncorrelated between the equations, so that they are truly unrelated, and when each equation
contains exactly the same set of regressors (Saraceno et al., 2021).

Following Greene (2012) a Seemingly Unrelated Regressions (SUR) model is proposed to be

applied for the estimation of the interdependence between the extents of farm mechanization,
 82

commercialization level and crop-diversification of smallholder farmers of the study area and
their determinant factors. The model will be specified as (52):

AMLi = X 1 ' 1 +  1i 

COMM i= X 2 '  2 +  2i  (52)
CROPDi = X 3 '  3 +  3i 


𝜈1𝑖 0 1 𝑃12 𝑃13

Where (𝜈2𝑖 ) ~𝑁 [(0) (𝑃21 1 𝑃23 )] (53)
𝜈3𝑖 0 𝑃31 𝑃32 1

( | X ) = 0 (54)

Var( | X ) = 1 (55)

Cov( | X ) = P (56)

Where, AMLi, COMMi, and CROPDi are agricultural mechanization level (index), level of
commercialization (index) and crop diversification (index) of a given household respectively;
the X’s are the respective vectors of covariates determining the endogenous variables; and the
β’s are their vectors of parameter coefficients. V’s are correlated disturbances in a seemingly
unrelated regression (SUR) model; and P’s are correlations between endogenous variables.
The assumption of disturbances to be strictly exogenous, homoscedastic, nonzero correlation
among equation but zero correlation among observation will be also considered. Moreover, the
model goodness of fit was also tested by using McElroy (1997) procedures.

3.4.4. Hypotheses and Definitions of Variables

Hypothesis and variable definitions for all econometric models will be given under subsequent
sub-sections below based on theories and previous findings. Views of the researchers were
also considered to include variables under each sub-section.
 83

[Link] Determinants of the level of farm mechanization

The dependent variable for Tobit model is a continuous index number (i.e., farm
mechanization level index- MI). Adoption level of farm mechanization is determined by
different factors.

Sex of household head (SEX): It is a dummy variable taking a value of 1 for male and 0
otherwise. Female headed households in most parts of Ethiopia are resource poor, have no full
power on resources and may be deprived of technology (Theis et al., 2019; Arun et al., 2019).
Therefore, it is hypothesized that this variable will affect farm mechanization level positively.

Experience in farming (EXPERIENCE): Farming experience is also a continuous variable

measured in years and hypothesized to have positive influence on AML. Experienced farmers
may have accumulated knowledge and skill on importance of farm mechanization
technologies. Similarly, Moniruzzaman at al. (2021) also found that experience has a positive
effect on adoption decision of potato mechanization in Bangladesh. Therefore, it is
hypothesized that this variable will affect farm mechanization level positively.

Household head education (EDUCATION): It is continuous a variable measured in years of

schooling. Education is a base for any economic or social development and has positive effect
on any technology adoption (Moniruzzaman at al., 2021; Tamrat, 2016; Rajkhowa et al.,
2020). Accordingly, it is hypothesized that education influence the AML positively.

Family labor availability (LABOR): This is a non-hired labor size available in a given
household measured in man-equivalent. Households with larger family labors are expected to
have low level of agricultural mechanization (Schmitz and Moss, 2015) because if a household
has enough labor force to operate its agricultural activities, it may not need farm
mechanization technologies. Therefore, it is hypothesized that this variable will affect farm
mechanization level negatively.

Operational farm size (OPERATIONAL FARM): This is total farm size operated
(cultivated) by the household in ha. If a household has larger size of farm land, it needs more
 84

farm machineries to cover its agricultural operation timely. Hence, this variable is
hypothesized to have positive impact on AML (Daum et al., 2020; Moniruzzaman at al., 2021;
Rasouli et al., 2009).

Social capital (SOCIAL): will be measured in terms of household head’s participation in

social affairs in a society like leadership role in the PA and cooperative, participation in debo,
and wonfel, etc. Social capital stands for the ability of actors to secure benefits by virtue of
membership in social networks or other social structures, is becoming a critical factor (input)
in understanding differences in economic outcome (Portes, 1998). As a result, a socially active
household is expected to get different benefits from society like getting free labor or working
together commonly known as “jigi” and ‘wonfel’. Hence, the variable is expected to influence
AML negatively.

Land fragmentation (FRAGMENTATION): This variable was measured in terms of

number of land plots that a given farm household is cultivating. The more the farmlands are
consolidated, the more it is convenient for mechanization (Arun et al., 2019; Rasouli et al.,
2009). Hence, it is hypothesized that the small the number of plots, the more the level of farm
mechanization.

Distance from main market center (DMAIN MARKET): It is a continuous variable.

Household’s access to main market center will affect AML through affecting the farm
productivity (amount of farm income) and marketability. It implies that outputs are marketed
more which will motivate the household to use more technologies in order to be more
productive. Hence, the variable is expected to affect AML negatively.

Distance from mechanization service rendering centers (MECH DIST): This variable is a
proxy to measure household’s access to mechanization technologies. If a household has more
access to farm mechanization centers, implying located at less distance to the service
providers, it will create good opportunities to use the technology. Therefore, it was continuous
and measured in km and expected to have negative effect on level of farm mechanization.
 85

Livestock possession (TLU): Livestock size is a good indicator for resource richness in rural
areas. Strong ownership of capital base assets would be a good reflection of the total resources
available to the farmers to invest and mechanize his/her farm (Akram et al., 2020). Hence, it is
hypothesized that livestock possession (in TLU) will affect AML positively. Livestock size is
computed in to standard Tropical Livestock Unit based on Stock et al. (1991).

Distance to all-weather road (ROAD ACCESS): Availability of infrastructures like road is

one of the most important factors to enable farmers use farm mechanization (Akram et al.,
2020). Access to all-weather road, can facilitate easy mobility of machinery (tractors and
combine harvesters). Hence, distance to all-weather road is measured in kilometers and it was
expected to influence AML positively.

Participation in non-farm income (OFF_FARM): Arun et al. (2019) found differences in

level of investment in mechanizing farm for different income sources. They found that level of
mechanization has generally positive relation with proportion of non-farm income. This is
because of the fact that as a household participate in non-farm income generating activities, it
will have less time to work on agricultural activities. Hence, the household will tend to use
more labor-saving technologies (farm mechanizations). As a result, the variable is expected to
affect AML positively.

Spatial location (LOCATION): It is defined as administrative zones of the household in this

study and it is a dummy variable defined 1 = West Arsi and 2 otherwise. Based on the
previous history of farm mechanization and access to technologies like presence of state farms
and land suitability for mechanization, it was hypothesized that West Arsi zone will have more
farm mechanization level.

Market participation (MKT-PART): It is a dummy variable taking one if the household

participates in a market by selling crop output and zero otherwise. Technology adoption and
market participation have complex inter-relations. The hypothesis is that if a household is
using more farm mechanization technologies (adoption of technologies), the technology
 86

adoption will improve its productivity and it will produce more surplus output to supply for
market and will have more crop output market participation probability (Barret, 2008).

Crop diversification (SDI): It was hypothesized that crop diversity has a positive impact on
farm machinery technology adoption. Higher crop diversity on their farms could motivate
farmers to adopt more agricultural machinery technologies and use them on different crops to
improve machinery use efficiency (Quan and Doluschitz, 2021) and that will reduce labor
shortage pressure during the pick agricultural activities.

[Link]. Definition of variables for impact of farm mechanization level: AIPW

There are three main categories of variables under this sub-section: treatment, outcomes and
control (covariates) variables and each of them are defined as follows:

Treatment variable: This variable refers to level of agricultural mechanization measured in

terms of mechanization index estimated by using equation (15).

Outcome variables: Five outcome variables namely household’s total income, economic
efficiency, yield (per hectare productivity), amount of labour employed and intensity of inputs
use were considered. The agricultural inputs considered were chemical fertilizer and
agrochemicals.

Control variables (covariates): As propensity score is the probability that an agent takes
treatment given covariates, selection of covariates is one of the most important criteria in
observational or non-experimental impact studies (Li and Fraser, 2015). It is about identifying
and controlling confounding variables. Seeger et al. (2007) and Patorno et al. (2013) put that
the general rule-of-thumb is to include all covariates that associate with both treatment and
outcome variables. However, inclusion of covariates that have association with outcome
variable regardless of its association with treatment variable is useful in reducing variance of
estimated treatment effect (Rubin, 1997; Brookhart et al., 2006). This is because there is a
chance that a variable related to the outcome is also related to treatment (Garrido et al., 2014).
Even though there is a fear for introduction of too much “noise” especially for small dataset by
 87

including too many covariates (Brookhart et al., 2006), the richness of covariates may
strengthen the assumption of unconfoundedness (Kluve et al., 2012) in dose-response function
analysis. However, studies strongly suggest excluding covariates that might be affected by the
treatment (Imbens 2004; Ho et al., 2007). Therefore, considering all these issues variables that
are defined under [Link] sub-sections were considered for control variables (covariates) in this
study.

[Link]. Variables definition for economic efficiency-SFA model

The variables that are used in the production functions for the stochastic frontier analysis
(SFA) are here defined briefly as follows. All inputs and outputs were transformed to their
corresponding log values in estimating the Cobb Douglas production function. As the log
value of zero is undefined, for the variables out of which production is possible zero values in
the data set were changed to nearly zero (0.0001) value before transforming the data to log
form.

Output: It is the dependent viable in the estimation of production functions, are measured in
standard measurement unit, Kilogram (Kg). The data were collected using different local
measurement units of output. However, for uniformity it was changed to the commonly used
measurement unit, Kg. Therefore, output measured in Kg was used in the analysis. Two
production functions were fitted in this study independently while the variable definitions are
same for both functions. The production functions were fitted for wheat and barley producers
separately.

Inputs: This refers to explanatory variables used in the estimation of production functions.
There are eight variables used in production function in this study and are explained as
follows:

1. Land (LAND): This refers to the area of plot of land allotted for the specific (wheat/barley)
crop production. The unit of measurement for area is also different in different parts of the
 88

country; hence the data was changed to hectare for smoothness. Accordingly, hectare of plot of
land used for crop production was used in the analysis.

2. Chemical fertilizer (FERTILIZER): The two most commonly used fertilizer types in
Ethiopia are Urea and DAP. These are important inputs for the crop production in Ethiopia
(Kefyalew, 2011). Hence, total chemical fertilizer applied on plot of land measured in Kg was
used in this study.

3. Seed (SEED): This refers to the amounts of seed used in the production of the wheat or
barley output measured in Kg.

4. Human Labor (LABOR): This input captures all human labor from any kinds of sources
like family, shared and hired, that is used in production of the specific crop under study. Since
age and sex of the labor that are used in the production system are common, all human labors
are computed to the standard labor measurement in terms of man-equivalent based on (Storck
et al., 1991 as cited in Arega and Rashid, 2005). Hence, all labor amounts used in crop
production from land preparation to harvesting was considered.

5. Oxen labor (OXEN): Oxen labor is an important input in production of crop in Ethiopia.
Agricultural operations like land preparation, planting, fertilizer application (in some cases)
and threshing (trampling) are commonly performed by using animal powers in most cases
oxen. Hence, number of animals (OXEN) used in the production of the crop was used in this
analysis.

6. Agrochemicals (CHEMICALS): There are different types of agrochemicals used in wheat

and barley production in Ethiopia for control of pests and diseases. Hence, these chemicals are
an important input in crop production. Therefore, amount of chemicals measured in litters for
the production of the specific crop was used in the analysis.

7. Tractor rental cost (TRACTOR): Currently, the use of tractor for land preparation and
planting (either row planting or broadcasting) is becoming important in most parts of Oromia
 89

region and this makes tractors one of the most important inputs. Hence, tractor rental cost for
the production of the crops under study is considered as one of the inputs in this study.
Because of the absence of data on tractor-hours (machine-hours), we are forced to use cost as a
proxy.

8. Combine harvester rental cost (COMBINE): In most parts of central and southern
Oromia, wheat and barley crop harvesting is performed by combine harvester that makes it
one of the important agricultural inputs for crops production. As a result, combine harvester
rental cost per quintal was used in this specific analysis.

[Link]. Variables definition for crop output commercialization: Heckman’s model

The objective of this hypothesis is to test households’ crop output market participation
decision and level of sales and internal and external factors that affect decision and level of
participation. The first dependent variable is market participation decision in crop output
which is a dummy variable taking 1 (one) if a household participate in market of crop selling
and 0 (zero) otherwise. The second dependent variable for Heckman’s two stages model is
proportion of value of crop sell to total value of crop produced. The variable is a ratio ranging
from 0 to 1 and becomes continuous.

Sex of the household head: This will be a dummy variable that takes 1 if a household head is
male and 0 otherwise. Male headed households are expected to have more access to
information and inputs that could boost their production. Therefore, based on pervious
empirical studies, it is hypothesized that being male will increase probability of market
participation and level of commercialization positively (Hlomendlini, 2015; Sigei, 2014;
Agete, 2014; Leykun and Jemma, 2014; Adino et al., 2021; Atinkugn, 2020).

Age of the household head: This is a continuous variable measured in years. Aged farmers
may have more resources like land and livestock and have better knowledge and skill in
production and this may increase their probability of market participation and intensity (Sigei,
2014); or on the other hand aged farmers may be deprived of resources and information and
 90

dependency on others also increases and as a result their market participation and intensity
many decrease (Guta et al., 2020). Hence, the sign of this variable is not determined a prior.
Educational level of household head: Literacy is assumed to improve households’ skills,
access to information and ability to process information, and thus expected to affect positively
the marketing decision and level of market participation (Adino et al., 2021; Birara et al.,
2020; Sigei 2014; Guta et al., 2020; Omiti et al., 2009). Therefore, it is expected that
education has a positive effect on CC in this research.

Family size: In subsistence farming, family is source of labor and can be assumed to affect
market participation and intensity positively (Hlomendlini, 2015; Agete, 2014). On the other
hand, the larger the household, the greater its consumption demand and can be also assumed to
have negative effect (Guta et al., 2020; Efa et al., 2016). Hence, the effect of family size is not
predetermined a priori in this study.

Size of total farmland owned (FLAND_OWN): This refers to size of land under cultivation
and a continuous variable measured in hectares. The variable is expected to influence market
participation and level of commercialization positively. The result from empirical study also
showed that the variable has positive effect on market participation and level of
commercialization (Efa et al., 2016; Birara et al., 2020; Hlomendlini, 2015; Atinkugn, 2020).

Cultivated land (ha): this refers to operational farmland of a household after deducting
farmland rented/shed-out and adding those rented/shared-in. The hypothesis is that if a
household operates on larger farmland, it will have more crop production (Tekilu and Tesfaye,
2020). Hence, this variable is posited to have positive effect on commercialization of
smallholders.

Credit access: This is a dummy variable taking value 1 if the household received credit within
the last two years and 0 otherwise. Based on previous studies (Hlomendlini, 2015; Atinkugn,
2020; Tewoderos et al., 2020), the hypothesis is that credit will facilitate to access input and
enable the household operate its agricultural activities and is expected with positive sign.
 91

Distance to market place: It is measured in kilometers and a continuous variable. Based on

studies previous studies (Omiti et al., 2009; Birara et al., 2020; Efa et al., 2016; Guta et al.,
2020) the variable is hypothesized to negatively influence decision and level of participation.

Access to market information: This is a dummy variable taking 1 if a household has market
information prior to selling and 0 otherwise which enables farmers to make informed decisions
on when to sell and what price to receive. Therefore, based on previous studies (Kyaw et al.,
2018; Guta et al., 2020; Agete; 2014; Sigei, 2014) it is hypothesized that the variable will
positively influence the decision and level of participation.

Access to extension service: It is number of extensions contact per month. It is assumed that
more contact with the extension agents will create better understanding on new technologies
and encourage more production and hence enhances market participation and level of
participation as well (Kyaw et al., 2018; Agete, 2014; Hlomendlini, 2015; Birara et al., 2020).

Total livestock holding (excluding oxen): Livestock is an important production shifter as

they enhance capacity of a household to produce more and hence improves market
participation and level of commercialization. But Efa et al. (2016) and Kyaw et al. (2018)
found that it has negative effect and hence, negative effect is expected.

Number of owned oxen: This is a count data, and oxen are the main source of working force
in agricultural activities that facilitate timely operation and enhances production and
productivity of the household thereby commercialization (Birara et al., 2020; Kyaw et al.,
2018). Hence, number of oxen owned by a household were hypothesized to influence
commercialization positively.

Access to all-weather road: It is a dummy variable taking 1 if a household has access to all-
weather road and 0 otherwise. Access to roads helps a farmer to access market information and
transportation (Kyaw et al., 2018; Guta et al., 2020). The variable has a direct impact on
transaction cost. Therefore, it is posted as a factor that positively influenced both the volume
of crop sold and the decision to sell.
 92

Household labor (man-equivalent): The higher the household labor available in a given
household, the lower the problem of agricultural labor shortage faced which could increase
productivity and production. Hence, based on previous studies like Gebreslassie et al. (2015)
in Ethiopia, which reported that lack of labor affected level of commercialization negatively,
we hypothesized that availability of household labor will have positive effects both on
probability and level of commercialization.

Agroecological zone: Agroecological zone determine types of crops grown, inputs

availability, production and productivity, and hence, probability of commercialization and its
level (Seo et al., 2009). There is a suspect that being in a highland area may affects the
probability and level of crop commercialization by affecting access to some important inputs
like farm mechanization that improve production and productivity. However, the wider
production of commercial crops in the highland areas like malt barley, potatoes and other
vegetables may affect commercialization positively. Hence, the effect of AEZs on
commercialization is not predetermined in this study.

Road access: It is a dummy variable with value of 1 if the household has an access to road and
0 otherwise. Access to market and infrastructures like road are important factors in increasing
both productivity and production. The improvement in production has a direct effect on
improving level of commercialization (Amrit, 2000). Rural infrastructures like access to roads
in close coordination with institutional infrastructure, such as credit institutions, agricultural
research and extension determines the nature and the magnitude of agricultural output and
thereby encourage farmers to supply their surplus for output market (Wharton, 1969; Noad
and Bamlaku, 2017). Hence, access to road is expected to positively affect both crop output
market participation and level of commercialization in this study.

Crop diversification (SDI): It is measured in terms of mix of crop grown by a household

relative to landholding using Simpson diversification index. Different studies revealed that if a
household is producing more diversified crops, it will have more probability of
 93

commercialization (Mekonnen, 2019). Hence, it is hypothesized that level of crop

diversification will have positive effects on commercialization.

Equines (EQUINES): Equine animals are the major means of transportation in rural Ethiopia
and it is also well understood that access to transportation means are important factors in
determining transaction costs and then level of commercialization and probability of market
participation (Berhanu and Moti, 2010). Hence, it is hypothesized that a household owing
equine animals is more likely to participate in output market and will have more level of
commercialization.

Household total Income (INCOME): different studies identified that income has positive
effects on level and probability of crop output commercialization. This could be if a household
has more income from different sources, it invests on crop production and produce more
surplus for market. Hence, it was hypothesized that this variable will positively affect
commercialization.

[Link]. Variables definition for determinants of crop diversification

Environmental, socioeconomic and cultural factors as well as policy and political climate are
the most important factors to influence farmers’ decision of level of crop diversification (Sati,
2012). Based on previous research outputs, a total of fourteen explanatory variables like
socioeconomic conditions, household demographics, farmer experiences, and community-
level characteristics (Rehima et al., 2013; Jarvis and Hodgkin, 2000; Neill and Lee, 2001;
Ryan et al., 2003; Degrande et al., 2006) that are expected to have significant effects on crop
diversification level at a household level were included in the model.

Sex of household head: This is a dummy variable taking 1 if male and 0 otherwise. Females
are mostly resource poor, have no full power on resources and may be deprived of technology
(Theis et al., 2019; Arun et al., 2019). Therefore, being a female household head is
hypothesized to negatively influence diversification.
 94

Age of household head: This variable could serve as a proxy for farm experience. The
assumption is that elderly farmers consider farming as way of life, while youngesrs may
consider it as business opportunity (FAO, 2012). However, some also found positive effects of
age on diversification and justified that from experience, older farmers adopt diversification to
mitigate unwarranted climatic variability, price shocks and total crop failure, to protect the
welfare of the household (Inoni et al., 2021). Hence, effect of age in this study is not
predetermined.

Level of education of the head: It is a continuous variable in years of schooling. Education is

hypothesized to enable a farmers understand the institution of agriculture more and helps
farmers to diversify more towards high value crops (Longpichai, 2013; Ngunjiri et al., 2016;
Inoni et al., 2021).

Land fragmentation: This refers to the total number of plots that a household has which is a
continuous and expected to influence crop diversification positively. The greater number of
plots a household owned, it will create opportunity of growing different types of crops as it
will give an opportunity to have different soil types with different fertility levels. According to
Benin et al. (2004), the more farm plots a farmer has, the more he or she is able to diversify.

Farm size: The arable land size possessed by a household is a crucial variable in determining
number of crops to be grown. Previously conducted studies attested that it has positive effects
on crop diversification (Seng, 2014; Ngunjiri et al., 2016; Jha, 2009; Inoni et al., 2021;
Sichoongwe et al., 2014; Wuletaw and Kindu, 2018) and the same is hypothesized in this
study too.

Access to credit: It is posited that this variable will enable a household to purchase necessary
inputs and promote more crop diversification. Previous results also attested positive relation
supports the hypothesis (Kumar, 2020; Inoni et al., 2021).

Extension contacts: This is measured in terms of number of extensions contact per month.
The hypothesis is that the more the farmer has extension contact, he/she will get trainings and
 95

the better he/she will diversify. Previous studies also attested similar results (Longpichai,
2013; Wuletaw and Kindu, 2018; Inoni et al., 2021).

Family size: It is expected to be positively related with crop diversification. The larger the
household size, the more likely that it will diversify to increase its food production levels. The
result of previous studies also supports this supposition (Weiss and Briglauer, 2000; Benin et
al., 2004).

Dependency ratio: Crop diversification is one of the risk management strategies to sustain
household livelihood and food security (Adnan et al., 2020; Chhatre et al., 2016). Hence, a
household with more dependency ratio is expected to diversify to sustain the household food
security demand. Accordingly, the variable, estimated based on the number of household
members below or above working age (age < 14 and age > 64), is one of the variables
anticipated to drive the production objectives towards diversification to minimize risks of food
security. Hence, crop diversification is a means to manage food insecurity.

Livestock holding: It refers to size of livestock holding of a household measured in TLU.

Wuletaw and Kindu (2018) found that this variable has positive impact on crop-livestock
diversification in Ethiopia. This study also hypothesized the same (positive) effect of
livestock.

Level of farm mechanization: It is index of farm mechanization level in terms of cost of

mechanization relative to other costs. It is posited that the variable will have positive
influences on level of crop diversification. Other studies also found similar results (Kumar,
2020; Seng, 2014; Jha, 2016; van den Berg et al., 2005; Sichoongwe et al., 2014).

Distance to market: It is a continuous variable and hypothesized to affect crop diversification

positively or negatively. Some authors justify that as farmers are distant from market, they
diversify more to meet their subsistence household need for food (Kankwamba, et al., 2012;
Sichoongwe et al., 2014); while others tried to justify that proximity to main roads and
 96

markets are important for development of other farm enterprises (Benin et al., 2004; Jha,
2016).

Agroecological zones (AEZs): This variable is important in determining types (mix) of crops
to be grown by a household in a given area (Kankwamba, et al., 2012). Prevalence of pests
and diseases is also notably determined by AEZs and in turn determines the level of
diversification. Hence, in areas (AEZs) of wider pest and diseases prevalence, crop
diversification level is expected to be low as there may be a smaller number of alternative
crops.

Non/off-farm income: It is a dummy variable indicating whether a household participates in

non-farm or off-farm income generating activities. The effect of participation in nonfarm
and/or? Off/non-farm income generating activities depends on the root-cause for participation.
If a household is participating in nonfarm and/or off-farm activities due to shortage of
farmland, it may have negative effect and if it is due to large income from farm activities and
diversifying towards other activities, it may have positive effects as a household with large
nonfarm/of-farm income can invest on agricultural activities more and more. Hence, the effect
of this variable is not predetermined in this study.
 97

4. RESULTS AND DISCUSSION

This chapter presents the results and discusses the main findings of the study. Hence, it is
presented in two sections. The first section provides descriptive analyses on the demographic,
socio-economic and institutional characteristics of sample farm households. The descriptive
analysis further extended to discuss the farm households’ farm mechanization, crop
commercialization, crop diversification, productivity, efficiency characteristics. The second
section presents econometric analyses of determinants of farm mechanization adoption,
economic efficiency, impacts of farm mechanization on economic efficiency, productivity,
labor employment, income and input uses, crop output commercialization, and crop
diversification. It further discusses the findings of the study against the earlier related research
results.

4.1. Descriptive Analysis

4.1.1. Demographic Characteristics of Sample Households

The households’ socioeconomic characteristics (sex, age, educational status, physical resource
endowment, income sources) and institutional characteristics like membership to different
economic and social institutions, proxy and access to institutions are the most important
factors in determining household’s decision regarding technology adoption, commercialization
and crop diversification. Nearly 96% of households were headed by men, and 74% of them
had formal education. The average farming experience was about 23 years, with a mean age of
44 years. It was revealed in the dependency ratio figure that each economically active member
of the household was providing support for approximately 1.40 persons. The mean household
size was 6.59 with minimum and maximum of 1 and 13 persons, respectively. Each household
has 2.34 man-equivalent of family labor while the mean family members working full time on
farm activities were 3.66%.
 98

Table 2. Demographic and socioeconomic characteristics of sample households

Education category Frequency Percent Cumulative

Illiterate 64 16.42 16.42
Read and write 18 4.62 21.04
Informal (religious school) 4 1.03 22.07
Formal education (college/university) 16 4.10 26.17
Formal education (1-12) 288 73.85 100.00

Sex of the household head Frequency Percent Cumulative

Female 16 4.10 4.10
Male 374 95.90 100.00
Source: Author’s computation from sample survey data (2022).

4.1.2. Sample households’ characteristics along mechanization level

Demographic and socioeconomic characteristics of the households were discussed based on

their farm mechanization (MI) level (Table 3). The sampled households were categorized into
three based on their level of mechanization (low if MI<20%, medium if 20%  MI <45% and
high if MI45%). It is observed that 29.97% of the sample households had low MI, while
39.80 and 30.23% had medium and high MI, respectively. This category was made by
dividing the MI at the 30th and 70th percentile distribution, which roughly divided the sample
households into three equal groups (Hirano and Imbens, 2004).
 99

Table 3. Socioeconomic variables along level of farm mechanization level

Mechanization level category Overall

Variable Low Medium High Mean F/2-value Prob>F/2

Age 44.12 43.54 44.40 44.00 0.15 0.858

Experience 22.64 22.21 22.10 22.60 0.15 0.860
Family size 8.10 8.03 8.00 8.04 0.01 0.986
Labor force 2.44 2.34 2.24 2.34 0.48 0.617
Education 5.61 5.84 6.00 5.84 0.36 0.966
Dependency 1.58 1.40 1.17 1.39 2.70 0.069*
Land size 1.89 1.79 1.97 1.88 0.61 0.546
Cultivated land 1.43 1.46 1.60 1.5 0.77 0.464
Social capital 0.53 0.46 0.44 0.47 6.37 0.002***
Market distance 5.27 4.66 5.07 4.97 0.79 0.456
Coop distance 4.40 4.43 4.82 4.54 0.51 0.598
Road distance 1.59 1.90 1.71 1.75 0.94 0.392
FTC distance 2.42 2.38 2.62 2.46 0.41 0.661
MMKT distance 7.00 6.93 6.91 6.95 0.01 0.987
Mechan distance 6.67 6.87 6.54 6.71 0.07 0.937
DA distance 2.50 2.54 2.56 2.53 0.03 0.968
Livestock 5.74 5.19 4.61 5.18 2.23 0.108
Oxen 1.37 1.40 1.50 1.43 15.58 0.000***
Diversification 0.51 0.48 0.43 0.47 2.61 0.075*
Income 81567 94249 122417 98688 7.10 0.001***
Livestock income 23271 23553 28882 25080 0.76 0.466
Crop value 124225 149827 227534 165691 1.62 0.199
Commer index 0.40 0.42 0.51 0.44 5.64 0.004***
Barley farm 0.65 0.80 1.25 0.88 5.48 0.005***
Barley output 30.47 27.18 10.52 23.13 21.76 0.000***
Wheat farm 0.54 0.62 1.32 0.82 15.41 0.000***
Wheat output 22.87 26.48 52.56 33.29 14.68 0.000***
Source: Author’s computation from sample survey data (2022).
 100

The F-value test results (Table 3) shows that dependency ratio and social capital are
statistically significant at 10 and 5%, respectively. As the dependency ratio increases the level
of farm mechanization also increases from low to higher MI. This implies that if the household
members capability to work on farm activities is getting less, then the probability of adopting
and using farm mechanization technologies will increase since the technologies will substitute
the human labor. Furthermore, the result revealed that more mechanized farms are more
diversified and most commercialized type, which are significant at 10% and 1%, respectively.
Similarly, the household’s annual income could be high enough if the farm is mechanized. The
table also shows that farm size and production of wheat and barley have significant interaction
with farm mechanization.

4.1.3. Asset Ownership of the Sample Households

Table 4. Asset ownership and labor availability variables across sex of household head
Variables Mean
Male Female Pooled mean t-value
Number of oxen 1.43 1.11 1.43 0.801
Livestock (TLU) 5.14 5.46 5.15 0.241
Family labor (man equivalent) 2.33 1.94 2.32 0.722
Total landholding (ha) 1.93 1.79 1.93 0.325
Educational level (grade) 6.08 3.00 6.01 2.284**
Mechanization Index (MI) 0.35 0.46 0.35 1.408
** t-value is significant at 5% level
Source: Author’s computation from sample survey data (2022).

The mean landholding of a household was around 1.94ha. The result showed that male-headed
households have a larger number of oxen and livestock (TLU) in aggregate. Similarly,
landholding size and family labor are more for male-headed households. Contrary to this, farm
mechanization is higher for female-headed households though the difference with male-headed
ones was statistically insignificant. The higher MI for female headed households could be
 101

attributed to a shortage of family labor and oxen shortage which are substitutes for farm
mechanization (Table 4).

4.1.4. Types and Level of Farm Mechanization Technology Adoption

Respondent households are using different farm mechanization technologies, whereas

agricultural farm implements that are operated by animal power are owned by almost all of the
households. The result from secondary data in the respective districts and zonal offices of
agriculture also revealed that around 84% of the households in Arsi and West Arsi zones are
using at least one mechanization technology (Table 5). The majority of the households
adopted either four-wheel tractors (86.15%) or combine harvesters (77.83%) while there are
two-wheel tractors and engine-driven stationery threshers adopted by a few households. The
result further shows that 87.66% of the households are using any of inanimate agricultural
mechanization technologies. According to the official report from districts and zonal offices of
agriculture, the majority of the households in the study areas are using farm mechanization.
The result further shows that more than 70% of the households are using combine harvesters
and tractors or either of the two technologies together (Table 6). The most common sources of
the farm machineries are private owners and unions such as Galema and Hetosa. Oromia seed
enterprise and Ardayta ATVET college also provide rental services for neighboring
smallholder farmers during their slug time.

Table 5. Farm mechanization technology adoption by sample districts

Thresher/ Mean of Percent

District Total HHs Tractor users combine users the two of users

Hetosa 17,788 16,700 16,365 16,532.5 92.94

Lemuna Bilbilo 22,455 16,240 19,982 18,111 80.65
Kofele 22,486 16943 14810 15876.5 70.60
Gedeb-Hasasa 27,842 23,098 27,707 25,402.5 91.23

Total 90,571 72,981 78,864 75,922.5 83.83

Source: Own computation based on zonal and district level agricultural office annual report
 102

Four main reasons have been identified behind adopting farm machineries. Accordingly, a
significant proportion of the respondents (64%) argued that the high cost of keeping
agricultural working animals both for land tillage and threshing purposes is the main reason
that forced smallholders to adopt farm machineries. Given this, agricultural labor shortage was
the second main driving factor towards farm mechanization among household in the study
area. Moreover, family and hired labor shortage is becoming a critical problem due to
schooling and migration of youths into urban areas in search of better life. Hired labor is
neither available nor affordable due to high wage rate. The continuous wage rate increment
may be due to an increase in the educational status of the society and hence they are either not
willing to do such drudgery works or demanding better payments. On the other hands, the ever
increase in inflation which fueled living cost could also be the main reason. About 37% of the
households replied that demanding better (non-drudgery) life and modernization of agriculture
was the other main driving factor of farm mechanization (Table 6).

Table 6. Types and main reasons behind adopting farm mechanization technologies

Description of variables Frequency Percent

Farm mechanization type used by households:
Two-wheel tractor 80 20.15
Four-wheel tractor 342 86.15
Engine-driven threshers 55 13.85
Combine harvester 309 77.83
Used either of mechanization technologies 348 87.66
Reasons to look at mechanization options:
Labor shortage 200 50.38
High wage rate for hired agricultural laborer 197 62.97
High animal feed cost 254 63.98
To reduce work drudgery (better life) 145 36.52
Source: Author’s computation from sample survey data (2022).
 103

Level of farm mechanization was computed for the two major crops namely wheat and barley
which covers around 81% of the total cropland in the study area. Households of the study area
on average allocated 1.39ha to producing barley and wheat crops, while they had 1.88ha
average cropland. The average wheat and barley farm land allocation of the households was
0.86ha and 0.53ha, respectively. Level of farm mechanization (Mechanization Index-MI) was
calculated by using equation (15). Results revealed that wheat MI ranges from zero to 100%
while its mean value is 38.46%, which implies that around 39% of the wheat production
operation are performed by machineries such as tractors and combine harvesters. In line with
this, mean barley MI is 34.89% while the compound MI for the two crops is 35.13%. Hence,
the result reveals that wheat farming is the more mechanized practice of the area (Table 7).

Table 7. Land allocation and level of mechanization for wheat and barley production

Land coverage Mean value St. Deviation Min Max

Total land (ha) 1.88 1.30 0.02 10
wheat farm size (ha) 0.86 1.69 0 10
Barley farm (ha) 0.53 0.59 0 4
Ratio of major crop to total land 0.81 0.84 0 10
Major crop farm size (ha) 1.39 1.75 0 10
Mechanization Index for major crop
Variable Mean value (%) St. Deviation Min Max
MI_Wheat 38.46 0.25 0 100%
MI_Barley 34.89 0.31 0 100%
MICPD_Major crops 35.14 23.17 0 100%
Source: Author’s computation from sample survey data (2022).

4.1.5. Major Constraints to Mechanize Agriculture in the Study Area

There are different factors that hinder the use of farm mechanization. Accordingly, 73.55%
and 46.22% of respondents indicated that high price and access to finance are the main
challenges for farm mechanization, respectively (Table 8). In addition to these, limited supply
 104

of the machineries is the next problem to easily adopt the technologies. Significant (more than
72%) proportion of the households replied that the topography of the area is convenient to use
mechanized farming, hence the topography is not a problem at all in the process of adopting
the technology. However, in the highland areas such as Kofele and Lemuna Bilbilo districts,
inconvenient topography and limited supply of the technology are among the most important
constraints in the process of farm mechanization.

Table 8. Major constraints in using farm mechanization technologies and their severity

Constraints Severity of the constraints

(Variables) Highest Medium Low

High price 292 (73.55) * 71 (17.88) 26 (2.55)

Access to finance 171 (46.22) 106 (28.65) 64 (17.30)
Technology supply 90 (22.67) 147 (37) 86 (21.66)
Awareness 60 (15.11) 155(38.04) 147 (37.03)
Ignorance 43 (10.83) 129 (32.49) 163 (41.06)
Topography 32 (8.10) 78 (19.75) 93(23.76)
Poor extension 37 (10.00) 199 (53.78) 99 (26.76)

*Numbers in parenthesis are percentage of the respondents responding ‘yes’

Source: Author’s computation from sample survey data (2022).

4.1.6. Crop Commercialization and Diversification

Commercialization in this context refers to crop commercialization defined as the proportion

of crop outputs marketed and valued in monetary terms at market price from the totally
produced within a specific production season. The major crops that are commercialized are
cereals specially wheat and barley (mainly malt barley), which accounts about 94%, while
faba bean from pulses and potato from root crops are the other crops produced for commercial
purposes. However, pulses and root crops are not common, only 22.42% and 13.10% of the
households participate in pulse (faba bean and field pea) and potato crop production,
respectively. The mean output commercialization index for potato and pulse crops were
 105

61.39% and 53.39%, respectively (Table 9). Around 91% of the households are participant in
crop output marketing even though more than 69% of the households said that their sell is
distressed. The majority of the households are producing mainly for consumption purposes
while around 22% of sample households produced mainly or entirely for market (commercial)
purposes. The mean commercialization index was around 44% implying about 44% of the
crops produced by sample households is supplied to market with maximum value of 99.85%
and a minimum of nil. Based on the Sahal and Baha (2010) classification, the level of
household commercialization for the two major crops is medium while only 27.69% of the
households were classified under high commercialization index (Appendix Table 1).

Most farmers do not have experience of storing crops for future selling mainly due to absence
of surplus marketable products (30%) and immediate need of money (46%) for agricultural
inputs loan repayment and other social commitments. Some portions of the respondents do not
store their products due to lack of appropriate, conducive and enough storage space. The
survey result revealed that only 49% the households are storing some proportion of their crop
output for five months on average to get a better price. Around 48% of the households, who
are storing their products, do not have separate store or warehouse and are using their living
house to store their products while 56% of them are using warehouses. The use of improved
storage (goiter) is very minimum in which about 14% of the households are still using local
unsafe storages. The most important and widely used transportation means to take crops to the
market is animal drawn carts, which may be either rented or owned. Out of the total
households only 58% have equine animals, mostly used to draw carts for transportation
purpose. But only 6% and 7% of the households are using vehicles and back animals,
respectively, to transport their agricultural products to market (Table 9).
 106

Table 9. Production goals, sale types and commercialization status of sample households

Variables description Frequency Percentage

Crop sell type: Distressed type 254 69.40
Surplus types 112 30.60
Production goal: Totally for consumption 57 15.36
Mainly for consumption 142 38.27
Equally for consumption and sale 89 23.99
Mainly for market 80 21.56
Entirely for market purpose 3 0.81
Crops produced mainly for market purpose: Cereals 339 94.43
Pulses 6 1.67
Vegetables 14 3.90
Participant households in COM 361 90.93
Mean (%) Std Dev. Min Max

SDI 47.46 0.26 0 1

Major crops commercialization Index 44.10 0.27 0 99.85
Pulse crop commercialization index ( 89) 53.00 0.28 0 0.98
Potato commercialization index (57) 61.00 0.29 0 1
Store months 5.20 2.29 1 12

Storage facilities for those store grain Frequency Percent Cumulative percent

Local storage ‘gotera’ 28 14.36 14.36

Improved storage ‘gotera’ 3 1.54 15.90
Warehouse using sack 71 36.41 52.31
Living house using sack 93 47.69 100.00
Store grain for latter sale 189 52.50
Have equines to transport grain 196 58.33
 107

Table 9. Production goals, sale types and commercialization (continued)

Means of transportation for agricultural product

Carts 282 86.24 92.97
Equines back 22 6.73 6.73
Vehicles 21 6.42 99.39
Human’s back 2 0.61 100.00
If you do not store your product, why?
I do not have enough (surplus) product 84 29.68 29.68
I do not have conducive (appropriate) area 42 14.84 44.52
I do not have enough spaces for storage facilities 21 7.42 51.94
Because I need money for immediate obligations 130 45.94 97.88
I do not see the importance of storing 6 2.12 100.00
Source: Author’s computation from sample survey data (2022).

4.1.7. Socio-economic, Institutional Characteristics and Crop Commercialization

Socioeconomic and institutional characteristics of sample farm household like a physical

resource endowment, annual income and infrastructural facilities that are in line with farm
activities are the most important phenomena that determine farm mechanization level, crop
output market participation and level of crop diversification. Physical resource endowment
includes landholding, livestock holding, farm labor and other assets. The institutional and
infrastructural facilities that affect the farm mechanization level, crop commercialization and
diversification are access to development agents, primary cooperatives or unions, market
information, distance to farmers training centers (FTC), market areas and the like.

The total cultivated land size is slightly less than regional landholding and greater than the
national level (1.54ha and 1.12ha, respectively) according to CSA (2021a). The result revealed
that the total livestock owned, the amount of non/off-farm income, land cultivated, total land
possessed, total household income and total crop value (in Birr) are positively related to crop
 108

output market participation. This implies that market participants are in a better position in
terms of income and resource endowments. The mean differences between market participants
and non-participants of cultivated land and total land owned are statistically significant at 5%
while total household income is significant at the 1% level of significance (Table 10).

Distance to different service provider centers like primary cooperatives or unions, FTC, main
market areas, mechanization service centers and development agents’ offices influence the
COM participation. Hence, the distant the household from those centers, the less it becomes a
participant in crop output market. The first three variables are significant at 5% each, while the
last two are at 10% and 1% level of significance, respectively. The respondents are getting
income from different sources, mainly cereal crops (wheat and malt barley) and livestock
production, root crop production mainly potato and carrot and pulse crops like Fabian bean.
Though it is not significant, crop output market participants are earning more annual income
from livestock compared to non-participants while annual total household income is
significantly higher for market participants (1% level). Moreover, market participant
households have highest mechanization index (significant at the 1% level) which may reveal
that mechanization is the promoter for agricultural commercialization. According to the result
(Table 6) more than 94% of the respondents answered that the main crop types they produce
for marketing purpose are cereals and followed by vegetables (4%) and pulses (1.7%).

As expected, the landholding is also different across agro-ecological zones of the study area
and, accordingly, highland respondents have average cultivated land and total landholding of
1.26ha, 1.69 ha respectively. Similarly, respondents from mid-land AEZ have 1.76ha and
2.09ha of cultivated and total landholding, respectively. The difference in land size among
households from mid-land and highland areas is statistically significant at the 1% level (Table
11).
 109

Table 10. Equality test for socio-economic and institutional characteristics of respondents and
their participation in crop market

Mean Values of Crop Output Market (COM)

Participant non-participant Combined t-Value

Variable Description (n=356) (n=34) (n=390)

Livestock owned (TLU) 5.27(.22)a 4.31(.58) 5.18(.21) -1.32

Number of oxen 1.61(.06) 1.17(.18) 1.57(.06) -2.15**
Non/off-farm income 25,866(2205) 17,184(2935) 25,079(2026) -1.23
Income from livestock 25,867(2205) 17,184(2935) 25,080(2025) -1.23
Total income 107,423(4655) 28,469(6897) 100,586(4450) -5.17***
Total cultivated land 1.53(.06) 1.12(.13) 1.49(.06) -2.05**
Total land possessed 1.94(.07) 1.20(.11) 1.88(.06) -3.29***
Total crop value 175,775(25661) 59,320(9534) 165,691(23514) -1.39
Distance to local market 4.898(.21) 5.65(.94) 4.97(.21) 1.04
Distance to cooperative 4.41(.18) 5.85(.81) 4.54(.18) 2.28**
Distance to main road 1.72(.10) 2.06(.39) 1.75(.09) 1.02
Distance to FTC 2.36(.10) 3.44(.69) 2.46(.11) 2.82**
Distance to main market 6.75(.25) 8.92(.98) 6.95(.25) 2.56**
Mechanization distance 6.49(.37) 8.88(1.84) 6.71(.38) 1.81*
Mechanization index 0.29(.01) 0.14(.03) 0.28(.01) -3.97***
Distance to DA office 2.43(.10) 3.59(.71) 2.53(.12) 2.92***
Distance to nearest road 0.88(.02) 0.86(.06) .88(.02) -0.35
DA contact per month 2.22(.01) 2.61(.21) 2.25(.07) 1.41
Note: *, ** and *** represent statistical significance at 10, 5 and 1% levels respectively;
COM = Crop Output Market; a Figures in the parenthesis are Standard Error of the mean.
Source: Author’s computation from sample survey data (2022).
 110

Table 11. Landholding by agroecological zones in the study area

Agro-ecological zones (AEZs)

Variable description Highland Mid-land Combined mean t-Value
Cultivated land 1.26(0.07) 1.76(0.09) 1.49(0.06) -4.40***
Crop diversification 0.52(0.02) 0.43(0.02) 0.48(0.01) 3.60***
Farm MI 27.34(0.01) 43.37(0.02) 34.74(0.01) -7.26***
Commercialization index 41.83(0.02) 45.23(0.02) 43.40(0.01) -1.20
Total landholding 1.69(0.08) 2.09(0.10) 1.88(0.06) -3.14***
Source: Author’s computation from sample survey data (2022).

4.1.8. Level of Crop Diversification and Potential Reasons

Around 72% of cultivated land of the sample households was allocated for the two major
cereals namely wheat and barley. This showed that these crops are the most dominant in the
study area. Crop diversification index is estimated by using the Simpson Diversification Index
(SDI). The results of this study revealed that mean household crop diversification level was
48% (Appendix Table 3). Based on Sahal and Baha (2010) classification, the result further
showed that around 25% of the households are low diversified while around 44% and 32% are
producing medium and highly diversified crop portfolios (bundles of crops), respectively.
Even though most households considered themselves as diversified, the result of SDI showed
that the diversification level in the study area is very low even compared to other locations in
Ethiopia (Appendix Table 3). For instance, the finding in the northern part of Ethiopia, showed
that the majority of the households were high level crop diversifiers (46%) and 11% of them
were very high diversifies with CDI of 46 to 60% and 61 to 75% respectively (Chanie, 2019).
Hence, this can justify that crop diversification in the study area is low compared to other parts
of Ethiopia as expected. The main lock-in constraints to diversify were discussed as lack of
improved varieties of alternative crops, lack of mechanization technologies, inputs like
pesticides and land unavailability which are similar with other findings in some parts of
Europe (Meynard et al., 2018).
 111

The boxplot graph was used to compare the Simpson diversification index of crop
diversification along three variables: agro-ecological zones (highland vis-à-vis mid-lands),
land fragmentation (number of plots of land owned) and the level of farm mechanization. The
figure clearly depicted that highland area’s farmers, higher mechanization level users and
farmers possessing a greater number of farm plots are more diversified (Figure 5).

The result further showed that diversification is used as the main mitigation strategy of market
and price fluctuation and means of food security and self-sufficiency. Risk mitigation strategy
in case of crop failures was the third reason, whereas labor diversification to reduce the
pressure that can be occurred during peak agricultural seasons was the fourth reason for
diversification of crop enterprises. However, as it can be seen from the result, most of the
farmers are focusing more towards specialization (Appendix Table 4). The FGD and KII
results revealed that most farmers prefer to produce cereals (mostly wheat and barley) due to
ease of production and availability of inputs. In most cases, mechanization technologies like
harvesting and threshing are available for cereals only. Improved seed for alternative crops
like pulses, fruit and vegetables are hardly available due to the absence of seed producers. The
perishability of root crops like potato, carrot, beetroots, and fruits and vegetables was also
another obstacle that was raised by farmers. Finally, land availability was raised as one of the
lock-in constraints to diversify their production as the farmers consider cereals important for
food self-sufficiency and prefer its production.
 112

Figure 5. Boxplots of crop diversification by AEZs, land fragmentation and mechanization

level

4.2. Results of Econometric Models

4.2.1. Determinants of Farm Mechanization Level: Tobit Model Results1

The main objective of this section is to identify the factors that determine the level of farm
mechanization. The dependent variable in Tobit model was level of farm mechanization
expressed as index calculated using equation 15. As a result, Ordinary Least Square (OLS)
could be biased and inconsistent (Tobin, 1958), and hence instead, a censored regression
model developed by (Maddala, 1983) called Tobit, is employed. The likelihood function of the
model for level of the farm mechanization index is highly significant (F (15, 382) =7.43; Prob
> F = 0.000), which evidenced a strong explanatory power of the independent variables. A
total of twelve socioeconomic and demographic variables is significant in determining the
level of farm mechanization index (Table 12).

1
This chapter is published on Elsevier Ltd Heliyon (Heliyon 9 (2023) e18390).
 113

Household characteristics such as sex of the head, educational status, experience in farming,
family labor, and social capital are statistically significant variables affecting farm
mechanization level. Variables that determine access to institutions like location of household
(i.e., zone of the household), access to all-weather roads and distance to farm mechanization
service centers are also statistically significant enough in affecting the farm mechanization
level of households. Variables that indicate households’ economic activities, types and
endowments of resources like participation in markets and off-farm activities, land size (ha),
land fragmentation and livestock size (TLU) are also statistically significant in determining the
level of farm mechanization in the study area.

Before running the model, test such as Lagrange Multiplier (LM) or score test suggested for
the normality assumption in the Tobit model was performed. The LM-statistic test was carried
out for testing the Tobit specification, against the alternative of a model that is non-linear in
the regressors and contains an error term that can be heteroskedastic and non-normally
distributed following Vincent (2010). The test was carried out by taking a Box-Cox
transformation of the dependent variable (yλ-1)/ λ and testing whether the parameter λ equal to
1 or not. A rejection of the null suggests that the Tobit specification is unsuitable, as an
alternative value for λ would be required to return the linearity, homoscedasticity and
normality assumptions that are necessary for consistent estimation. Critical values are obtained
via the parametric bootstrap, where the regressors are assumed to be stochastic. The result
revealed that bootstrap critical values displayed for 1%, 5% and 10% level tests were 9.06,
4.46 and 3.05, respectively, and these values are less than the LM-statistic (47.92). Hence, the
LM test suggests that the Tobit model is suitable for the data. Normality was also tested using
histogram (Figure 6). The linktest approved that there is no model misspecification problem
(Appendix Table 5).

Accordingly, the result of Tobit model (Table 12) shows that male-headed households are
significantly less likely to use farm mechanization. This finding is opposite to the findings of
Theis et al. (2019) and Arun et al. (2019) who hypothesized that female-headed households
are resource-poor and have low power on resources and hence, deprived of technology
adoption. However, opposite to the hypothesis, being female-headed household increases the
 114

level of farm mechanization by 10.59% while it increases the level of farm mechanization by
0.092 and 0.032 for the whole sample and for those who already started using farm
mechanization, respectively ceteris paribus. Hence, sex significantly affects adoption of farm
mechanization positively. Previous studies, like Mottaleb et al. (2016) and Kirui (2019) found
that women were less knowledgeable about effects and advantages of mechanization due to
some factors such as less formal education, an inability to attend extension services, and
limitations of women’s movements outside the household, which finally result in they were
low adopters. However, females may tend to use farm mechanization technologies to
overcome family labor shortage and working animals like oxen as it is the case indicated under
descriptive analysis result where female-headed households have a smaller number of oxen
and family labor (Table 4). As females are resource poor and possess less livestock (oxen) and
low family labor adopting mechanized farming could be the only way-out for survival in this
area.

Experience in farming significantly and positively explained level of mechanization at 5%

level. The result of marginal effects depicted that if experience in farming increases by one
year the level of farm mechanization (index) increases by around 0.26%, keeping other
variables constant. It also revealed that increasing farm experience by a year will increase
level of farm mechanization for whole population under study and for those already started
using farm mechanization technologies by 0.0022 and 0.0013, respectively. Similar studies
also showed that age (proxy for farming experience), positively affects the use of tractor
powered machineries in African countries like Kenya, Niger, Zambia, and Zimbabwe and also
for pooled sample in general (Kirui, 2019) significantly. Ai et al. (2023) also found that age
has positive effect on farm mechanization service purchase of rice farmer in China. However,
Aryal et al. (2021) found that the elderly farmers are less likely to adopt harvesting machines
may be due to nonrealization of direct impacts on the yield and net profit of the machine.

Households’ head and other family members (especially spouse) educational status is an
important factor in enhancing farm technology adoption. It is believed that education, whether
formal or informal, will improve awareness, knowledge and skill of a farmer on importance
and how to use of the farm machineries (Aryal et al., 2021). The result from this finding also
 115

showed that educational status of a household head affects farm mechanization level positively
at 5% level of significance. A unit change in educational status will change level of farm
mechanization by around 0.70%, whereas it increases the level of farm mechanization by
0.0056 and 0.0033 for whole population domain and for those who are already using farm
mechanization technologies, respectively. According to studies in some African countries
including Ethiopia and South Asia, household head education level is statistically significant
in explaining the use of tractor-powered mechanization (Paudel et al., 2019; Kirui, 2019;
Aryal et al., 2021). Similarly, Owombo et al. (2012) also found that an increase in household
heads’ education increased the probability of adoption of mechanization practices.

Spatial location of household (zone) is negatively affecting level of farm mechanization and is
significant at 10% level of significance. According to the marginal effects result, farms in
West Arsi are more mechanized and the model output indicated that shifting a household’s
location from West Arsi to Arsi will reduce level of farm mechanization by around 5%.
Similarly, being in Arsi zone can reduce the farm mechanization level for the whole
population and for those already practicing the technologies by 0.0419 and 0.0247,
respectively. This could be due to the topography of the farm land, which is plain especially
around Gedeb Asasa district and convenient for mechanization and number of federal and
regional state farms in and nearby districts like Ardayta agricultural technical and vocational
college, Oromia seed enterprise and Gofer farms. The presence of these farm organizations
can have positive impact in mechanizing private smallholders’ farm by providing machinery
rental service and promoting mechanization. Others also reported that farm mechanization
technologies ownership and adoption are affected by spatial location (Akram et al., 2020;
Aryal et al., 2019; Mottaleb et al., 2016).

Distance from household’s residence to farm mechanization service providing center is

statistically significant at 1% level of significance and is negatively affecting level of farm
mechanization. The marginal effect coefficient revealed that if the center of mechanization
service provider is far from a household residence by a kilometer, the probability of using
farm mechanization will be reduced by about 0.94% while level of farm mechanization will be
reduced by 0.0079 for population understudy and by 0.0047 for technology users.
 116

Household’s wealth indicators like landholding and size of livestock are the most important
significant variables in determining level of farm mechanization. Accordingly, total land
cultivated (hectare) affects level of farm mechanization positively and significantly. According
to the result, as cultivated land size increased by one hectare, probability of using farm
mechanization can be increased by 2.70% while the level of farm mechanization for the whole
sample understudy and those started using farm mechanization will increase 0.0227 and
0.0134, respectively. Similar study by Kirui (2019) in Ethiopia and somewhere in African
countries showed that probability of using animal-powered mechanization increases with
landholding size. Similarly, farm size significantly increases the probability of using tractor-
powered mechanization in Egypt, Niger, Zambia and Zimbabwe (ibid) and somewhere else
(Albiero et al., 2019; Rasouli et al., 2009; Quan and Doluschitz, 2021).

Livestock possession measured in Tropical Livestock Unit (TLU) is another variable that
explained level of farm mechanization significantly at 5%, but negatively, which may be due
to the fact that livestock are major sources of agricultural working power and mechanization
technologies are substitute for animal power. This finding is also consistent with other
findings somewhere in the world wherein increment in farm mechanization level could
significantly decrease the use of draught animal power (Zhou et al., 2021). The marginal effect
results showed that increasing size of livestock holding by one TLU would decrease the
probability of using farm mechanization by 0.80%, whereas it reduces the level of farm
mechanization by 0.0067 and 0.0040 for the whole sample households and for those who have
already started using farm mechanization, respectively.

Land fragmentation measured by number of farm plots possessed is another variable that
affects farm mechanization level negatively at 1% level of significance. This result also agrees
with the findings of Lai et al. (2015), Gürsoy et al. (2021), Rasouli et al. (2009) and Gavgani
et al. (2021). Result of marginal effects shows that the increase in number of farm plots by one
unit will reduce the probability of using farm mechanization by 5% while it decreases the level
of farm mechanization by 0.0421 and 0.0249 for the whole sample households and for those
who started using farm mechanization technologies, respectively.
 117

Total family labor in man-equivalence has negative and significant effect on farm
mechanization level. This result is in line with the assumption that labor availability within a
household will reduce level of farm mechanization due to the substitution nature of the two
inputs and it is similar with others findings (Caunedo and Kala, 2021; Barman and Deka,
2019; Kirui, 2019). The marginal effects result revealed that an increase of family labor by one
unit decreases the probability of using farm mechanization by around 0.60% whereas it
reduces the level of farm mechanization by 0.0049 and 0.0029 for the sample households and
current users of the technologies respectively.

Another important finding from this study was the positive and significant relationship
between crop output market participation and the level of farm mechanization among farming
households, which is statistically significant at a 1% level. Our finding is also similar to others
finding (Owusu and Iscan, 2020). According to the findings, participation in crop output
market can increase households’ probability of farm mechanization use by around 9%. The
marginal effects further showed that increasing market participation by a unit can increase
level of farm mechanization by 0.0747 and 0.0593 for population of study and current users
respectively.

There is a multidimensional concept of social capital (Sabatini 2005; Kaasa 2009).

Accordingly, index of social capital is constructed from seven dimensions. Social capital is
crucial for technology adoption and wellbeing as it is supposed to improve social networks
and access to information; hence, it is expected to influence technology adoption positively.
However, in our finding, membership to different village institutions is negatively affecting
level of farm mechanization at 5% significance level. This is because having stronger social
capital and bond enables households to get access to pooled labor through other means like
debo, and wonfel and can access working animals (oxen and horses) easily and this result is
consistent with the finding of Kaasa (2009). The result of marginal effects showed that the
increase in index of social capital decreases the decision to use farm mechanization by around
13% whereas, it decreases the level of farm mechanization by 0.1099 and 0.0649 for the whole
population understudy and for those already using the technologies respectively.
 118

Access to all-weather road is also one of the variables that affect probability of using farm
mechanization and level of farm mechanization at 5% level of significance. The finding of
Mottaleb et al. (2016) also supports our findings. The marginal effects show that the
probability of using farm mechanization increases by 7.07% for those who have access to all-
weather roads as compared to those who do not have access while level of farm mechanization
increases by 0.0581 and 0.0452 for the whole population and for users of farm machineries
respectively.

Participation of households’ member in off-farm economic activities is significant at 10% in

explaining level of farm mechanization. It increases, ceteris paribus, the probability of using
farm mechanization by 4.40% while the result of marginal effects further revealed that
increase in off-farm activity participation will increase level of farm mechanization by 0.0373
and 0.0190 for whole population and current user sample households respectively. Similar
findings were reported by Kirui (2019) and Akram et al. (2020). As per the assumption, use of
farm mechanization by a household can save more time to participate in more nonfarm
activities while the participation will in turn generate additional income to invest on farm
mechanization technologies and they are supplementing each other.
 119

Histogram of Farm mechanization index per household

2.5
2
1.5
Density

1
.5
0

0 .2 .4 .6 .8 1
Mean mechnanization index for two major crops (barley and wheat)

Figure 6. Histogram farm mechanization per household

 120

Table 12. Tobit estimate and marginal effects of determinants of level of farm mechanization

E ( y* | x) E ( y | x)  x   Pr( y  0 | x)  x  
Variable = Std. Err. P>t =    =  
x x   x   
Sex of head (Male=1) -0.1059 0.0081 -3.98*** -0.0920 -0.0322
Farm experience (years) 0.0026 0.0005 2.39** 0.0022 0.0013
Education of head (grade) 0.0066 0.0015 2.23** 0.0056 0.0033
Location (W/Arsi=0/Arsi=1) -0.0498 0.0134 -1.85* -0.0419 -0.0247
Off farm activities (yes=1) 0.0438 0.0101 1.89* 0.0373 0.0190
Cultivated landholding (ha) 0.0270 0.0058 2.32** 0.0227 0.0134
Livestock (TLU) -0.0080 0.0016 -2.46** -0.0067 -0.0039
Number of plots (count) -0.0501 0.0060 -4.18*** -0.0422 -0.0249
Family labor (man-equiv.) -0.0058 0.0035 -0.83 -0.0049 -0.0029
 121

Table 12. Tobit estimate and marginal effects of determinants of level of farm mechanization (continued)

Market participation (yes=1) 0.0911 0.0230 2.58*** 0.0747 0.0593

Access to road (Yes=1) 0.0707 0.0222 2.04** 0.0580 0.0452
Market distance (km) 0.0009 0.0012 0.38 0.0008 0.0005
Mech. center distance (km) -0.0094 0.0013 -3.71*** -0.0079 -0.0047
Crop diversification (SDI) 0.0470 0.0262 0.89 0.0396 0.0234
Social capital (Index) -0.1307 0.0280 -2.32** -0.1099 -0.0649
Constant 0.0802 5.67***
Number of observations = 390; Uncensored = 375
Left-censored observations = 10; Right-censored observations = 5 y = Pr (0<MI<1) = 0.9486
Log pseudolikelihood = 30.3159; Pseudo R2 = 2.6158 y = E(MI|0<MI<1) = 0.3709
F (15, 382) = 7.43; Prob > F = 0.0000 y = E(MI*|0<MI<1) = 0.3529
*, **, and *** imply the p-value is significant at 10, 5, and 1% significance respectively
MI=mechanization index
Source: Author’s computation from sample survey data (2022).
 122

4.2.2. Economic Efficiency of Wheat and Barley Production

Economic/cost and technical efficiency of the farming households is estimated by using

stochastic frontier model and the efficiency scores were further used in the AIPW model for
evaluating impact of farm mechanization on economic efficiency of households. Cost
inefficiency is nothing but the product of allocative and technical inefficiency Coelli et al.
(2005) (CE/EE= AE*TE). An ‘sfcross’ STATA command was used to estimate economic
efficiency of wheat and barley producer households for each crop. Finally, the economic
efficiency scores were generated following jlms predict procedures/command ‘predict
ee_jlms_wt, jlms/predict ee_bc_barly’ (Jondrow et al., 1982).

The result revealed that the mean economic efficiency of wheat production is found to be 72%
with a value ranging from 5% to 95%, which implies that the overall lose in producing wheat
due to economic inefficiency ranges from 95 to 5%. Similar research findings have been
reported in the different parts of Ethiopia (Milkessa et al., 2019; Mesay et al., 2013).
Households of the study area had 73% mean EE in producing barley, while efficiency ranges
from 2% to 96% indicating a wider difference among individual producers. The estimation
result showed that there is a huge cost inefficiency in barley production that is sourced from
high technical and economic inefficiencies. According to this result, the barley production loss
of households due to economic inefficiency ranges from 98% to 4%, which is consistent with
other findings (Sime et al., 2022; Mustefa and Jema, 2020). The estimation result presented
below reveals that households of the study area can further improve their EE of producing
wheat and barley by 28% and 27%, respectively, without incurring additional input costs
(Table 13).
 123

Table 13. Economic and technical efficiency of wheat and barley producers (bc* estimated)

Variable Observation Mean Std. dev. Min Max

Technical efficiency (wheat) 232 0.94 0.05 0.69 0.99

Economic efficiency (wheat) 232 0.72 0.19 0.05 0.95
Technical efficiency (barley) 257 0.91 0.09 0.15 0.98
Economic efficiency (barley) 257 0.73 0.21 0.02 0.96

* Battese and Coelli estimation method

Source: Author’s computation from sample survey data (2022).

4.2.3. Impacts of Farm Mechanization on Input Use Intensity, Economic Efficiency,

Income, and Productivity of Households

Under this sub-topic, impact of farm mechanization level on total household income,
households’ economic efficiency (EE) for the major crops (wheat and barley), per hectare farm
productivity for wheat and barley crops, and agricultural inputs (chemical fertilizer,
agrochemicals and labor) use intensity were discussed. An AIPW was used for estimating the
impacts. AIPW estimates instead of modeling either the outcome, like regression adjustment
(RA), or the treatment probability, like IPW, models both the outcome and the treatment
probability. The most important strong side (advantage) of this estimator is that only one of
the two models must be correctly specified to consistently estimate the treatment effects, a
property known as “doubly robust” (Robins et al., 2000).

[Link]. Post-estimation test for model validity

The overlap assumption that states ‘each individual has a positive probability of receiving each
treatment level’ was tested by ‘teoverlap’ command on STATA version 17. The ‘teoverlap’, a
post-estimation command, is used to showed that plots the estimated densities of the
probability of getting each treatment level are used to inspect whether the assumption is
violated.
 124

Figure 7. Propensity score overlap for level of farm mechanization

Figure 7 depicted that neither of the plots indicates too much probability mass near 0 or 1, and
the three estimated densities have most of their respective masses in regions wherein they
overlap each other. Hence, there is no evidence that the overlap assumption was violated in
our data.

Second, a command known as ‘tebalance summarize’ was employed to test whether a

‘teffects’ estimate command has balanced the covariates over treatment levels. According to
Austin (2009) the covariates are held to be balanced if the covariate has a weighted
standardized difference of equal or closest to zero and a weighted variance ratio of one or
nearer. The result for covariate balance test is given and the result showed that all of the
covariates are balanced for the matching of the three MI levels (Appendix Table 7).

[Link]. Impact of farm mechanization on agricultural input use intensity

Under this section, results for impacts of farm mechanization level on intensity of agricultural
inputs use which are labor, chemical fertilizer, and agrochemicals were presented. The
potential mean outcome, percentage changes, and average treatment effects of farm
mechanization on labor are also reported (Table 14). The result revealed that increase in MI by
one unit in the lower category of MI will decrease the amount of labor employed in the farm
by 3.81 times and it is significant at 1% level. Similarly, the change in MI in the medium and
 125

higher MI categories will affect the of labor employment negatively by 3.43 and 3.19 times,
respectively and both are statistically significant at 1%.

The Average Treatment Effects (the average causal effect of farm mechanization) (ATEs))
results further reveal that adoption of both medium and high-levels of farm mechanization
reduced the amount of labor employed on farms. The results indicate that relative to the low-
level of mechanization, adoption of farm mechanization at the medium level tends to reduce
the amount of labor employment by 4%. Similarly, relative to low-level mechanization, the
use of high-level of mechanization will reduce level of labor (in natural log form) use by 16%,
while adopting farm mechanization at high-level MI will reduce ln of labor amount by 13%
relative to medium level MI. Generally, the model results revealed that as the level of farm
mechanization increases, the amount of labor employed in a given farm will decrease
significantly and that confirms the labor substitution nature of farm mechanization. This result
is in line with the findings of different researchers like Barman and Deka (2019), Wang et al.
(2016), and Chhun et al. (2015) who found that farm labor employment and investment in
farm mechanization (the use of the service) have inverse relationships. Similarly, Kirui (2019)
reported that mechanization (especially using a tractor) has a labor displacement effect in
Ethiopia too.

However, other researchers argued that farm mechanization does not mean labor displacement
in absolute terms. For instance, Reimer (1984) argued that farm mechanization effects as farm
labor reduction are only for short period, and in the long run, mechanization increases the farm
labor required. Hassan and Kornher (2019) also found that farm mechanization does not
displace agricultural labor and lower wage rate in rural Bangladesh. But the non-displacement
of labor as a result of mechanization is mainly due to the increase in cropping intensity
especially in areas of irrigation facilities. Some also count for other job opportunities that are
created due to mechanization like machinery operators, dealers of spare parts, service and
maintenance and etc., and argue that farm mechanization does not mean labor displacement
necessarily. However, in Ethiopian cases where there is low access to irrigation facilities,
mechanization cannot have a positive effect on cropping intensity and therefore farm labor
displacement is inhibitable. And also, as most rural laborers are unskilled, mechanization of
 126

farms can hardly provide them job opportunities in the service sector like operation and
maintenance.

Table 14. Results of the treatment effects estimation for labor employment

ATT estimates Percentage change in ATE

Level of mechanization Coefficient z-value Coefficient z-value
Medium MI vs low MI -0.38(0.12) -3.23*** -0.04(0.01) -3.31***
High MI vs low MI -0.63(0.15) -4.10*** -0.16(0.04) -4.57***
High MI vs Medium MI -1.09(0.32) -3.42*** -0.13(0.04) -3.43***
PO means of labor by MI Coefficient z-value
Low level MI 3.81(0.10) 36.37***
Medium level MI 3.43(0.05) 63.40***
High level MI 3.19(0.11) 28.87***
Note: PO=potential outcome; robust standard errors are in parenthesis; ***<0.001;
ATT=Average Treatment effect on the Treated; ATE=Average Treatment Effect
Source: Author’s computation from sample survey data (2022).

The second input considered in the analysis was chemical fertilizer utilization. The
econometric results showed that use of chemical fertilizer increases with the increase in the
level of farm mechanization. The teffects results from aipw estimator method show that the
potential outcome means for ln of the amount of fertilizer applied by a household for lower
level, medium level, and high-level mechanization index were 4.76, 4.99, and 5.34,
respectively. The results further demonstrated that both medium-level and high-level
mechanization-indexed households have a higher application of chemical fertilizers compared
to low-indexed ones. The ATE results (Table 15) show that relative to low-mechanized
farming, medium and high-level mechanized farming increases the ln of the amount of
chemical fertilizer application by 14% and 21% respectively. Relative to medium-level
mechanization, high-level mechanization adoption will increase chemical fertilizer application
by 7%. This result is in line with the result of Kirui (2019) which evidenced that tractor-
powered mechanization is accompanied by increased use of fertilizer in Burkina Faso,
 127

Ethiopia, Ghana, Kenya, Niger, and Senegal. Other researchers also reported higher fertilizer
application for a higher level of mechanization (Shields, 1985; Salam, 1981).

Table 15. Results of the treatment effects estimation for chemical fertilizer application

ATT estimates Percentage change in ATE

Level of mechanization Coefficient z-value Coefficient z-value
Medium MI vs low MI 0.61(0.20) 3.06*** 0.14(0.05) 2.76***
High MI vs low MI 0.93(0.20) 4.56*** 0.21(0.05) 3.97***
High MI vs Medium MI 0.33(0.15) 2.17** 0.07(0.03) 2.12**
PO means of labor by MI
Low level MI 4.41(0.17) 25.57***
Medium level MI 5.02(0.09) 53.34***
High level MI 5.35(0.12) 44.77***
PO=potential outcome; robust standard errors are in parenthesis; ***<0.001
Source: Author’s computation from sample survey data (2022).

The third agricultural input considered in this study is the intensity of agrochemicals
application. The ln of potential output value for relatively low mechanized, medium
mechanized, and high mechanized farms are 9.07, 8.63, and 7.37 respectively. Generally, there
can be two theoretical assumptions behind the relation between the level of farm
mechanization and amount of agrochemicals applications. The first assumption is that use of
precision technology in chemical application reduces wastage; while the second assumption is
using farm mechanization technologies in land preparation and other agricultural activities
such as better-quality tillage reduces weed population, and growth, which can reduce pest
incidence and ultimately reduces chemical application. Hence, different authors found a
similar result that level of farm mechanization has negative and significant effects on the
intensity of agrochemical application (Zhang et al., 2019; Zhou et al., 2020; Su et al., 2022;
Afridi et al., 2020).
 128

The ATE results (Table 16) also showed that farm mechanization has a negative impact on the
intensity of agrochemicals used by households. According to the result, relatively medium-
level mechanized and high-level mechanized farms are using fewer amounts of chemicals
compared to low-mechanized farmers. Accordingly, the shift from low MI to medium and
from low MI to high MI will decrease the ln of agrochemicals by 4% and 16% respectively.
similarly, relative to medium MI, the high MI farms are using less amounts of agrochemicals
by 13%.

Table 16. Results of the treatment effects estimation for agrochemicals (ln transformed)

ATT estimates Percentage change in ATE

Level of mechanization Coefficient z-value Coefficient z-value
Medium MI vs low MI -0.44(0.13) -3.20*** -0.04(0.01) -3.31***
High MI vs low MI -1.69(0.38) -4.49*** -0.16(0.04) -4.57***
High MI vs Medium MI -1.09(0.32) -3.41*** -0.13(0.04) -3.43***
PO means of agrochemical Coefficient z-value
Low-level MI 9.07(0.12) 75.13***
Medium level MI 8.63(0.06) 144.29***
High level MI 7.37(0.36) 20.43***
Note: Robust standard errors are in parenthesis; *** = significant at 1% level
Source: Author’s computation from sample survey data (2022).

[Link]. Effect of farm mechanization on economic efficiency of wheat and barley

producers

The study examines effect of farm mechanization level on the EE of wheat and barley
producer households. Three observations are found to have propensity scores below the
minimum overlap of the common region while four got above the maximum overlap of the
common support region hence a total of seven observations were excluded from the analysis.
 129

The results of treatment effects on wheat and barley producers’ EE revealed that wheat
producers who are in the category of medium and a higher level of farm mechanization are
economically more efficient than those having a low MI (Table 17). Relative to the lower MI,
medium-level MI, and high-level MI are 37% and 49% more economically efficient,
respectively. Similarly, relative to medium-level MI, farmers with high-level of MI are 7%
more economically efficient. Moreover, barley producer farmers with higher farm MI are also
more efficient. Accordingly, compared to farmers with low MI, medium and high MI farms
are more efficient by 20% and 25%, respectively. Relative to the medium farm mechanization
level, barley producer farmers with high MI are more efficient by 13%. This finding is also
consistent with other findings somewhere in the world that show farm mechanization has
positive and significant impacts on the EE (Vortia et al., 2019; Min et al., 2021; Soliman,
1992).

Table 17. Results of the treatment effects estimation for wheat and barley economic efficiency

a. For wheat producers ATT estimates Percentage change in ATE

Mechanization level Coefficient z-value Coefficient z-value
Medium vs Low 0.15(0.03) 5.76*** 0.37(0.13) 2.88***
High vs Low 0.08(0.03) 2.28** 0.49(0.14) 3.51***
High vs Medium MI 0.07(0.02) 4.65*** 0.07(0.01) 4.65***
b. For barley producers, ATT estimates Percentage change in ATE
Level of mechanization Coefficient z-value Coefficient z-value
Medium MI vs Low MI 0.12(0.04) 2.68*** 0.20(0.16) 1.19
High MI vs low MI 0.25(0.04) 6.32*** 0.25(0.08) 3.05***
High MI vs Medium MI 0.13(0.02) 5.33*** 0.13(0.02) 5.33***
Note: Robust standard errors are in parenthesis; *** = significant at 1% level
Source: Author’s computation from sample survey data (2022).
 130

[Link]. Impact of farm mechanization on household income

Total household income, which is the sum of on-farm, off-farm, and non-farm earnings is
considered in this study. Results of the treatment effects from aipw estimation model show
that a high level of farm mechanization has positive and significant effects on household
income (Table 18). This result is in line with findings of other studies in different areas of the
world such as Philippines, China, Indonesia and the like (Lim, 1985; Peng et al., 2022; Berg et
al., 2007; Sang et al., 2023, Feryanto et al., 2022). The result of our findings also revealed that
the potential outcome means of income (in natural log) for a low, medium, and high level of
mechanization are 10.97, 11.20, and 11.45, respectively and all are statistically significant.
The result further shows that medium and high MI farmers have higher income compared to
low-level farm mechanization users. The shift from low to medium MI and from low to high
MI will increase the ln of household income by 2% and 4%, respectively while the shift from
medium to a high level of MI will increase the ln of income by 2%, ceteris paribus.

Table 18. Results of the treatment effects estimation for household income (ln transformed)

ATT estimates Percentage change in ATE

Mechanization level Coefficient z-value Coefficient z-value
Medium MI vs low MI 0.23(0.10) 2.38** 0.02(0.01) 2.34**
High MI vs low MI 0.48(0.12) 3.87*** 0.04(0.01) 3.79***
High MI vs Medium MI 0.25(0.11) 2.30** 0.02(0.01) 2.28**
PO means fraction
Low-level MI 10.97(0.09) 121.45***
Medium-level MI 11.20(0.07) 169.34***
High-level MI 11.45(0.09) 122.54***
Note: Robust standard errors are in parenthesis; **, *** = significant at 5% and 1% level
Source: Author’s computation from sample survey data (2022).
 131

[Link]. Farm mechanization impact on wheat and barley productivity

The potential yield of wheat for farmers with low, medium, and high mechanization index is
36.78, 46.25, and 54.55 quintals per hectare, respectively (Table 19). As it is hypothesized, the
result showed that the level of farm mechanization has positive and significant effects on
wheat productivity. According to the model output, compared to low-level mechanized farms,
both relatively medium-level and high-level mechanized farms are more productive. This
finding is consistent with other studies in Ethiopia (Guush et al., 2017) and the outside world
(Zhou and Ma, 2021; Mather and Belton, 2018; Belton et al., 2021; Kirui, 2019; Roy et al.,
2022; Peng et al., 2022; Feryanto et al., 2022), as well as with evidence from our focus group
discussion and key informant interview in these areas. The result further shows that relative to
the low mechanized farm, adoption of the medium mechanization level increases wheat farm
productivity by 26%. Relative to the low-mechanized farm, the high-level mechanized farm is
more productive by 48%. Similarly, relative to medium-level MI, the high-level MI farms are
more productive by 18%.

Table 19. Results of the treatment effects on wheat productivity (qt/ha): AIPW estimation

ATT estimates Percentage change in ATE

Mechanization level Coefficient z-value Coefficient z-value
Medium MI vs low MI 9.48(4.96) 1.91* 0.26(0.17) 1.53
High MI vs low MI 17.77(5.19) 3.42*** 0.48(0.20) 2.39**
High MI vs Medium MI 8.30(2.18) 3.80*** 0.18(0.05) 3.58 ***
PO means by the MI level Coefficient z-value
Low MI 36.78(4.81) 7.64***
Medium MI 46.25(1.23) 37.52 ***
High MI 54.55(1.79) 30.47***
Note: Robust standard errors are in parenthesis; *, **, *** = significant at 10, 5 and 1% level
Source: Author’s computation from sample survey data (2022).
 132

The result further shows that potential mean outputs of barley are 42.36, 45.42, and 49.82
quintals per hectare for low, medium, and high levels of farm mechanization, respectively
(Table 20). The result shows that farm productivity is increasing with an increase in the level
of farm mechanization implying a direct relationship between the level of mechanization and
productivity, but the result is not statistically significant. The insignificant result may be due to
the fact that combine harvesting, which is the main contributor in increasing productivity by
reducing postharvest losses, is not widely used in barley production.

Table 20. Farm mechanization impact on barley productivity: AIPW estimation

Treatment-effects estimation for barley productivity; Number of observations = 240

ATE estimates Percentage change in ATE
Level of farm mechanization Coefficient z-value Coefficient z-value
Medium MI vs low MI 3.06(4.58) 0.67 0.07(0.12) 0.63
High MI vs low MI 7.46(6.06) 1.23 0.18(0.16) 1.12
High MI vs Medium MI 4.41(4.29) 1.03 0.10(0.09) 1.02
PO means by the level of MI
Low mechanization index 42.36(4.39) 9.64***
Medium mechanization index 45.42(1.22) 37.38***
High mechanization index 49.82(4.13) 12.08***
Note: Robust standard errors are in parenthesis; *** = significant at 10% level
Source: Author’s computation from sample survey data (2022).

4.2.4. Determinants of Smallholder Farmers’ Crop Commercialization2

[Link]. Determinants of crop output market participation

In this and the next subsequent sub-section, factors that determine participation and level of
smallholder farmers’ crop commercialization are analyzed using Heckman’s two-stages

2
This part is published on the journal: Springer Nature ‘Discover Food (2023) 3:15’
 133

selection model. Prior to running the model, the data was tested whether it is suffering from
multicollinearity, heteroscedasticity and endogeneity problem. According to the test result, the
data is free from serious problem of multicollinearity and heteroscedasticity (Appendix Tables
6, 13). Similarly, endogeneity was tested for level of farm mechanization (MI) using
Hausman’s specification test method. However, based on the result (Appendix Table 14), we
were unable to reject for the endogeneity problem. Hence, instrumental variables (farm
mechanization technologies unavailability, price of the technologies, land topography,
participation on training and demonstration on farm mechanization) were selected as
instrumental variables and index of farm mechanization was regressed on these variables
including other variables in the structural model. Then prediction of farm mechanization
(index) was taken and included in the original model (Heckman two-step).

The Heckman two-stages model was fitted with eighteen (18) explanatory variables.
According to the first step model output, the household head’s age, livestock ownership
(TLU), AEZs, household residence to main market (km), availability of equine animals, total
household income, instrumental variable for farm mechanization, and access to market
information were significant in determining crop output market participation. The value of
Wald 2(18) equals 123.80 with a Chi-square (2) value significant at 1% and lambda ()
significant at 10% level of significance implied that data best fits Heckman’s two-step model
(Table 21).

It was believed that aged households are more deprived from up-to-dated production
technologies, labor and others. Risk aversion nature of older farmers to accept new
technologies and ideas may be possible arguments behind the low market participation
(Rapsomanikis, 2015; Pender and Dawit, 2007). Hence, their focus is mainly for food self-
sufficiency. In line with this assumption, the finding revealed that age of the household head
has a negative effect on crop market participation. Similar results were reported by researchers
in other rural parts of Ethiopia for bulla and pineapple crops (Sigei et al. 2014; Kusse et al.,
2022) while similar result was also found in Zambia (Chalwe, 2011).
 134

Table 21. Heckman selection model two-step estimates the result of market participation

Market participation Intensity of participation

Determinant factors Coefficient Z-value Coefficient Z-value

HHH sex(1=male) 0.821 (0.575) 1.43 - -

HHH Age (year) -0.020 (0.008) *** -2.34 -0.001(0.001) 0.36
HHH education (year) -0.038 (0.026) -1.46 -0.006(0.003) 1.51
Landholding (ha) -0.021 (0.107) -0.19 0.009(0.014) 0.59
Cultivated land (ha) 0.016 (0.131) 0.12 -0.031(0.016)** 1.95
HH family size (No.) -0.004 (0.029) -0.13 -0.003(0.004) 0.82
HH labor (man-eqvt) -0.079 (0.084) -0.94 0.007(0.012) 0.58
Number of oxen 0.089(0.088) 1.00 0.019(0.011)* 1.81
Other livestock (TLU) 0.074(0.034) ** 2.15 -0.004(.004) 1.07
AEZs (0=high) 0.651(0.277) ** 2.35 0.036(0.045) 0.80
Road access 0.344(0.312) 1.10 0.053(0.041) 1.28
Main market distance -0.065(0.019) *** -3.46 -0.007(0.004) ** 1.98
Crop diversification (SDI) -0.113(0.388) -0.29 -0.099(0.053)* 1.87
Instrument for MI -2.007(1.239) * -1.64 -0.177(0.180) -0.98
Equines (1=Yes/0=No) 0.348(0.203) * 1.71 0.001(0.029) 0.02
Income(ln) 0.381(0.074) *** 5.10 0.180(0.021) *** 8.40
DA contact 0.273(0.207) 1.31 -0.027(0.032) 0.85
Credit access -0.208(0.261) -0.79 -0.017(0.035) 0.50
Market information access 0.547(0.193) *** 2.82 0.124(0.038) *** 3.24
Constant -3.116(1.029) *** -3.03 -1.462(0.247) *** -5.92
Mills/lambda 0.222(0.122) 1.82*
Rho 1.000
Sigma 0.222

Note: *, **, *** values significant at 10, 5, and 1% level of significance respectively; Market
participants = 315; non-participants = 75; Wald 2(18) = 123.80; Prob > chi2 = 0.000;
Standard Errors given in parenthesis;
Source: Author’s computation from sample survey data (2022).
 135

Farm mechanization can enhance crop output market participation through increasing
production and productivity. Hence, any factor that affects level of farm mechanization either
positively or negatively can affect market participation the same way (Peng et al., 2022). Five
instrumental variables (IV), (technology unavailability, high price, topography, training and
demonstration participation on farm mechanization) were used in the ordinary least square
regression to estimate their effect on level of farm mechanization (the endogenous variable)
and the test for endogeneity showed that farm mechanization was endogenous variable. Hence,
Heckman two-step model was fitted by instrumenting the variable by taking the predicted
value for farm mechanization. The IV was found to be significant in determining probability
of participation. Hence, the result revealed that the application of farm mechanization
technologies especially, tractors and combine harvesters, significantly affects probability of
participation in crop output commercialization.

Transaction cost is one of the important determinant factors for market participation (Berhanu
and Moti, 2010. Distance of household’s residence from the market center has a direct
implication on marketing cost and access to means of transportation, which directly affects
market participation (Berhanu and Moti, 2010; Key et al., 2000; Eleni, 2001; Barrett, 2008;
Pender and Dawit, 2007; Getachew et al., 2022). The result revealed that distance to the main
market center negatively and significantly affects crop market participation significantly. The
result implies that the nearer the household is to the main market center, the higher the
probability of participating in crop marketing. This result is consistent with other findings in
Ethiopia and somewhere in the world where distance of market center from household’s
residence affects the probability of participation in a market (Holloway and Ehui S. 2002;
Berhanu et al., 2009; Rehima and Dawit, 2012; Kassa and Kibreab, 2017; Alula et al., 2022).

Lack of access to means of transportation in rural areas is one of the hindering factors in
market participation (Moti et al., 2009). As it is known, equine animals (donkey, horse and
mules) are the most important means of transportation both for goods and human beings in
rural Ethiopia (Berhanu and Yoseph, 2011) and it is also hypothesized that its presence can
positively influence market participation (Berhanu and Moti, 2010). Hence, its availability in a
 136

given household will definitely affect probability of market participation. The result also
showed that the probability of market participation is positively and significantly affected by
means of transportation availability, which is the equine animal in our cases. According to the
finding, households having equine animals have more probability of participation in the crop
market. According to this finding, households possessing equines have better probability of
crop output market participation significant at 10%. Similar researches (Alene et al., 2008;
Bedaso et al., 2012) also reported that the availability means of transportation mainly, equines,
was among the significant variables in determining teff market participation in Ethiopia and
maize crop output market participation in Kenya, respectively.

As expected, the amount of a household income per annum has also a positive and significant
impact on the probability of a household’s participation in the crop market. This can also be
explained as if a household has more income, it will have more capital to invest in crop
production to purchase inputs like fertilizer and improved seeds, to rent machinery for farm
mechanization, and even rent more land that will enable it to produce a surplus product for
market purposes. Hence, the model result tallies that the higher the amount of household's
annual income, the higher the probability of participation in crop market and the result is
statistically significant at 1% level of significance. This result is also in conformity with the
result of Getahun et al. (2017).

In the study area, households mainly accumulate their asset, which are sources of inorganic
fertilizer, manure, that can increase production and productivity as well as the probability of
market participation. The result showed that livestock size other than oxen, measured in
Tropical Livestock Unit (TLU) is another important variable that positively determined the
probability of crop market participation significantly at a 1% level of significance. The result
implied that farmers who owned a larger unit of livestock have better probability of crop
market participation. This result is consistent with the findings of others (Getahun et al., 2017;
Sultan, 2016) who found that households owning a greater number of livestock are more
participant in crop marketing in the Oromia region, Ethiopia. However, (Dagmawe et al.,
2022) found that livestock ownership has a negative impact on market participation in Afar
 137

region, Ethiopia. This result may be attributed to the production system in the area where
households having more size of livestock are pastoralists and less involved in crop production.

Households in mid-land agroecological zones are tending to specialization towards cereals

especially, wheat, and therefore, they have to sell some amount of their produce at least to buy
some grain that they do not produce for their own consumption (Table 11). Hence, as
expected, market participation is significantly affected by agroecological location of the
households at 5% level of significance. According to the result, a shift from highland
agroecological zones to mid-land, can increase probability of market participation
significantly. This finding is also in lined with other research findings which revealed that
there is difference in probability of market participation across spatial location which may
affect the production potential, access to input or transportation and others (Efa et al., 2016).
The finding of this author also showed that mid-land farmers are less commercialized and
have low probability of market participation in South West Oromia region.

[Link]. Determinants of crop commercialization level

Once a household decides to participate in a crop output market, the next issue is by how
much to participate or to answer what is the intensity of participation. The intensity of
participation was estimated based on the ratio of value of crops sold to the value of total crop
produced by a farmer. For this objective, market participation and level/intensity of crop
output market participation was calculated by taking the major crops of the area namely wheat,
barley, potato and pulse crops (faba bean, field bean, chickpea and haricot beans). Different
factors are found to affect levels of crop output market participation. The result of Heckman’s
second step showed that six variables explained the intensity of crop output commercialization
in this study. The result revealed that education of the household head, total cultivated land,
distance to the main market center, crop diversification level, annual income, and access to
market information are significantly affecting the intensity of crop output market participation.
Negative signs for the explanatory variable cultivated land size (ha), was unexpected finding
of this model that can be explained differently opposite to the hypothesis (Table 21).
 138

According to the model output, Lambda (IMR) or selectivity bias correction factor has a
positive impact on farm households’ crop output market participation at a 10% level. The
existence of unobserved factors that positively influence both participation decision and level
of marketed crop output is indicated by the positive sign of the IMR. Moreover, the positive
relationship (correlation) between the unobservable factors is justified by positive sign of rho.

A household operating on larger farm size is expected to produce more surplus and its supply
is expected to be more. And hence, the size of operational farm size is expected to have
positive effect on level of commercialization. The finding of other authors also revealed that
land size has positive effects on level of commercialization of wheat and chickpea (Sultan,
2016; Solomon et al., 2010) respectively. However, unexpectedly, result of this study
revealed negative effects of size of cultivated land on the level of market participation. The
variable is significant at the 10% level. The plausible argument for this finding is that in most
cases large land size is associated with large age, and larger family size which have negative
impacts on market participation and level of commercialization implying that younger
households have less access to land. Similar ways, even though it was not statistically
significant Chernet et al. (2021) also reported similar findings.

In most cases of less developed countries, crop diversification is practiced for the purposes of
food self-sufficiency. The result from descriptive statistics (Table 11) also showed that
highland areas are more diversified and at the same time are less commercialized. Crop
diversification, measured in Simpson Diversification Index (SDI), as expected has negative
significant effect on the intensity of crop output market participation at a 10% probability level
of significance. Assessments in different countries also revealed that commercialization at
national level leads to diversified crop production while it ends with specialization at regional
and household level (Pingali and Rosegrant, 1995). Hence, the result posited that an increase
in SDI value will decrease market participation intensity significantly. The result from KII and
FGD also revealed that in highland areas, farmers are producing everything to feed their
families and striving not to buy anything from outside.
 139

The annual income of a household comes from different sources like sale of crops, livestock,
and non-farm income sources such as remittance, working on others farm as labor etc. The
higher the household’s income implies, the more it can invest on agricultural production
activities to purchase input, employe labor and farm machineries and that increases surplus
production to be supplied to the market. Accordingly, total annual income of the household
showed a positive and significant effect on the volume of crop output supplied to the market
with a p-value significant at the 1% level. This result is also consistent with other findings
(Getahun et al., 2017; Kyaw et al., 2018). The coefficient of the variable revealed that an
increase in ln of annual income by one unit will increase the volume of crop output supplied to
the market by 18% ceteris Paribus.

Access to market information: The presence of intermediaries in the agricultural value chain
will increase transactions and that will lower the actual price of the products for the farmers.
Access to direct market information will have the advantages of bypassing middlemen and
direct marketing with agricultural product buyers and will reduce transaction costs. That will
also enhance the market participation and intensity of market participation (Strasberg et al.,
1999). Hence, the result of our finding revealed that access to market information has a
positive and significant effect on the intensity of market participation where the value is
significant at a 10% probability. The finding of this research is consistent with other findings
(Chalwe, 2011; Alula et al., 2023; Kyaw et al., 2018; Mzyece, 2016; Anwar, 2020).

Households’ distance to the main market area (center) is another determinant of market
participation intensity which is significant at a 10% probability level with a negative sign as
expected. Distance from market center has a direct implication on transactions which affects
intensity of market participation negatively. Hence, the result implied that as a household
distance from market center increases, volume of surplus supplied to market decreases and
vis- versa. This finding is also consistent with other findings (Berhanu et al., 2009; Solomon et
al., 2010; Kyaw et al., 2018; Kusse et al., 2022; Alula et al., 2023).

The major sources of traction power in Ethiopian agriculture are oxen. As a result, households
having more units of oxen can perform their agricultural operation and meet the average
 140

cultivation rate of crop production than their correspondent. Hence, having a greater number
of oxen implies that the household has more potential to produce more crops and participate
intensively in crop output market. The finding from econometric model also revealed that
intensity of crop commercialization was positively and significantly affected by number of
oxen owned. The result was also similar with the findings of other researches in rural Ethiopia
(Tadele et al., 2016; Engida et al., 2021; Kusse et al., 2022).

4.2.5. Determinants of Crop Diversification3

Before embarking on crop diversification analysis, the diversification level was estimated first
using the Simpson Diversification Index (SDI). Households are then categorized into low,
medium and high diversification based on Sahal and Baha (2010) index. Accordingly, low
crop diversification households are those with SDI values of 0 to 38%, medium are those with
39% to 63% SDI value and high crop diversification households are those with above 63%
SDI. Then, ordered logistic regression model was used to identify the determinants of the crop
diversification level in the study area. Ordered logistic regression and ordered probit were the
two possible models. To select the model that best fits to determine level of crop
diversification, model specification tests was carried out using Akaike’s Information Criteria
(AIC). The AIC values for ordered logit and probit were 657.84 and 660.47, respectively,
which indicate that ordered logit is preferable than ordered probit because value of AIC is
more preferable.

Prior to running Ordered Logit, multicollinearity test was done for explanatory variables using
a ‘collin’ command and all the variables have a Variance Inflation Factors (VIF) of less than
two and the mean value was 1.26 (Appendix Table 11), which implies that there is no problem
of multicollinearity. After running the model, the central assumption of the ordered logit
model commonly known as proportional odds assumption was tested. This assumption says
that the gaps among the various scales of the ordering are equidistant to each other or the
effect of each predictor across the categories of the ordinal dependent variable is the same.

3
Published on Cogent Food & Agriculture Online
([Link]
 141

This assumption was tested by running post-estimation test called Brant Wald’s test (Brant,
1990). Brant test of parallel regression assumption revealed that the effect of each explanatory
variable does not vary across different cutting points of the ordered outcome variable
(Appendix Table 11); hence, the ordered logit model is an appropriate to address the problem.

Table 22 presents the results of Ordered Logit model. According to results indicated (Table
22), Chi2-test for LR (14) of 157.83 with a probability of less than 1% is highly significant,
indicating best fit of the model used in the analysis. The Pseudo R2 value also indicates that
20.14% of the variation in the outcome variable (level of crop diversification) is explained by
those explanatory variables included in the model. A total of fourteen (14) explanatory
variables were included in the model, wherein five of them were found statistically significant
in determining the crop diversification level. The significant variables are family size, total
farm landholding, level of farm mechanization (index), level of land fragmentation (number of
plots) and spatial location (agroecological zones) of the farming households (Table 22).

The result from odd ratios (Table 23) revealed that family size of a household is positively and
significantly affecting the level of crop diversification. According to the result, for one unit
increase in family size, the odds ratio of higher level of crop diversification index versus the
combination of medium and low diversification categories are 1.05 times greater, given that all
other variables in the model are kept constant. The marginal effects value also shows that a
one unit increase in family size of the household, will associate with the 0.7% less likely to be
in the lower level of crop diversification status and 0.93% more likely to be in the higher level
of crop diversification status. The logical justification for this could be that as a household has
more family member, it will be more concerned about food self-sufficiency and try to be
secured by diversifying its production. The other most important reason is that more family
member means more labor to work on farm and less labor shortage to be faced if there is crop
diversification. Hence, the household will not have problem of labor to work on different
fields at a time, which could enhance probability of having higher crop diversification. This
result is also similar with other authors (Rahman and Kazal, 2014; Dereje et al., 2022;
Rahman, 2008; Culas, 2006; Baba and Abdulai, 2020), who revealed that availability of labor
and large family size could increase the crop diversification level.
 142

Table 22. Odds ratios for Ordered Logit Model estimation

Crop Diversification level Coefficient Std. Error z-value

HH heads’ sex 0.7492 0.544 -0.40

Farm experience (years) 0.9980 0.010 -0.19
Education (years) 0.9929 0.031 -0.22
Family size 1.0472* 0.027 1.76
Dependency ratio (ratio) 0.9337 0.071 -0.90
Total farm landholding (ha) 1.9286*** 0.227 5.57
Livestock hold (TLU) 0.9824 0.032 -0.54
Contact with DA (count) 0.9797 0.253 -0.08
Accessed credit (yes/no) 1.3335 0.402 0.95
Mechanization level (index) 2.6455* 1.429 1.80
Land fragmentation (no. of plots) 2.7155*** 0.347 7.82
Non/off farm income (yes/no) 1.1678 0.337 0.54
Agroecological zones (AEZs) 0.4491*** 0.123 -2.93
Distance to main market (Km) 1.0193 0.024 0.80
/cut1 1.2151 (0.895)
/cut2 3.6576 (0.917)
Note: *, **, *** significant at 10%, 5% and 1% level of significance; LR chi2(14) = 157.83,
Prob > chi2 = 0.000, Log likelihood = -312.92; Pseudo R2 = 0.2014
Source: Author’s computation from sample survey data (2022).

Size of farm land is also a variable that has positive and significant effect on household’s crop
diversification level. Farm landholding is the most important factor to diversify crop. The
plausible logic behind this is that if a household has larger farmland, it can allocate it for a
greater number of crops but with limited farmland, it will allocate the land for limited crops
that are more important for food self-sufficiency. The result shows that higher landholding
leads to higher likely to be in the high category of crop diversification status. The odds ratio of
ordered logit model showed that for one unit increase in landholding, the odds of higher crop
 143

diversification level (measured in index) versus the combined medium and low categories are
1.9286 times greater, given other variables constant. Similarly, the odds of the combined
medium and high categories versus low diversification are 1.93 times greater than fixing other
variables constant in the model. As to the marginal effects, a one unit increase in the farmland
size will be associated with the 9.84% and 3.37% less likely to be in a low crop diversification
status and a medium crop diversification status respectively, while it is 13.21% more likely to
be in a higher crop diversification status. Other studies also revealed that landholding size has
positive relationship with crop diversification in Ethiopia and somewhere in the world (Ashfaq
et al., 2008; Ibrahim et al., 2009; Sichoongwe et al., 2014; Benmehaia and Brabez, 2016;
Lawin and Tamini, 2019; Mekonnen, 2019, Baba and Abdulai, 2020; Li et al., 2021).

Level of farm mechanization affects probability of being in a higher crop diversification index
and there were two arguments and hypotheses about the effect of this variable on the crop
diversification level (Daum, 2023). The first hypothesis was that farm mechanization leads to
crop enterprise specialization at least at household level since farmers may prefer crops for
which mechanization technologies are available. The second hypothesis was that highly
mechanized farmers may diversify more because the mechanization will solve problem of
labor shortage. For instance, some studies supported that farm mechanization facilitates crop
diversification as it facilitates operation of labor intensive and drudgery activities (Daum et al.,
2022; Dahlin and Rusinamhodzi, 2019). Similarly, at an early stage of mechanization,
households may shift to easily mechanized crops as it was the case in Ghana and Ethiopia
(Kansanga et al., 2018). Hence, the second hypothesis held true in this research. Accordingly,
the odds ratio of the model result, a one unit increase in the farm mechanization level will
increase the odds of higher crop diversification level versus the combined medium and low
categories by 2.65 times given that all of the other variables in the model remain constant.
Likewise, the odds of the combined medium and high diversification versus low category of
diversification were 2.65 times greater when other variables are constant. The marginal effects
result also showed that a one unit increase in farm mechanization level increases the less
likeliness of a household to be in low and medium level of crop diversification by 14.58% and
4.99%, respectively and increases the more likeliness of a household in higher crop enterprise
 144

diversification status by 19.57%. This result is also in conformity with the findings (Ashfaq et
al., 2008; Wondimagegn et al., 2011; Abro, 2012).

Land fragmentation measured in terms of number of farm plots is one of important variables
that affect probability of households’ level of crop diversification positively and significantly
as it was hypothesized. Having a larger number of farm plots at different places may enable
the household to have different soil types and fertility level, and even agroecological zones
which favor different crop types. Hence, this clearly initiates a farmer to have more diversified
crop portfolio. The result indicated that, for one unit increase in number of farm plots, the odds
ratios of higher crop diversification level versus the combined medium and low categories
were 2.72 times greater, keeping other variables constant. In the same manner, the odds of the
combined medium and high crop diversification categories compared to low diversification is
2.72 times greater given that all other variables in the model are held constant. The marginal
effect results also revealed that a one unit increase of number of plots will associate with less
probability to be in a low and medium crop diversification level by 14.97% and 5.12%,
respectively. On the other hand, households will be 20.09% more likely to be in a high level of
crop diversification. This result was also in consistent with the findings of Lawin and Tamini
(2019).

Agro-ecological zone is an important factor that determines crop type to be grown by a farmer.
According to the result, being in highland agroecological zone increases the probability of
being in high diversification level status significantly. For one unit increase in AEZs (shift
from 0 to 1/ from highland to mid-land) the odds of high crop diversification level versus the
combined medium and low categories are 0.45 times greater, ceteris paribus. Likewise, the
odds of the combined medium and high categories versus low diversification are 0.45 times
greater given that all other variables in the model are held constant. The result of marginal
effects values also revealed that a shift from highland agroecological zone to mid highland
will associate with a 12% and 4.12% more likely to be in a low and medium level of crop
diversification. However, this shift from highland to mid-land will be associated with the
16.10% less likely being in a high level of crop diversification status. In general, highland
dwellers have more probability for higher diversification. Studies also evidenced that crop
 145

diversification level varies across locations (Sichoongwe et al., 2014; Rahman and Kazal,
2015).

Table 23. Marginal effects estimate for crop diversification level after ologit

Ologit ME for Ologit ME for Ologit ME for

Low CD level Medium CD level High CD level

Crop Diversification level (dy/dx) (dy/dx) (dy/dx)

HH heads’ sex 0.0400 (0.090) 0.0218 (0.072) -0.0613 (0.162)

Farm experience (years) 0.0003 (0.001) 0.0001 (0.001) -0.0004 (0.002)
Education (years) 0.0011 (0.005) 0.0004 (0.002) -0.0014 (0.006)
Family size -0.0069 (0.004)* -0.0024 (0.002) 0.0093 (0.005)*
Dependency ratio (ratio) 0.0103 (0.011) 0.0035 (0.004) -0.0138 (0.015)
Total farmland (ha) -0.0984 (0.018)*** -0.0337(0.015)** 0.1321(0.025)***
Livestock hold (TLU) 0.0027 (0.005) 0.0009(0.002) -0.0036(0.007)
Contact with DA (count) 0.0031 (0.038) 0.0012(0.014) -0.0041(0.052)
Accessed credit (yes/no) -0.0404 (0.039) -0.0200(0.027) 0.0603(0.066)
Mechanization level (index) -0.1458(0.081)* -0.0499(0.034) 0.1957(0.109)*
Number of plots (counts) -0.1497(0.020)*** -0.0512(0.0221)** 0.2009(0.027)***
Non-farm income (yes/no) -0.0225(0.040) -0.0093(0.0202) 0.0319(0.060)
AEZs (highland/mid) 0.1200(0.041)*** 0.0412(0.021)* -0.1610(0.055)***
Main market distance (Km) -0.0029(0.003) -0.0010(0.001) 0.0038(0.005)
Y=Pr(low CDI =1) =18.35%; y= Pr(medium CDI =2) =53.75; and y=Pr(high CDI=3) =
27.89%.
Source: Author’s computation from sample survey data (2022).
 146

4.2.6. Interlinkage among Farm Mechanization, Crop Commercialization and Crop

Diversification

The main aim of this sub-section is to estimate the interlinkage among the level of farm
mechanization, crop output commercialization and crop diversification in the study areas using
Seemingly Unrelated Regression (SUR) model. To proceed with the analysis, first, model fit
parameters for OLS and SUR models were assessed. The outputs for the two models are given
in Table 24. The result showed that the root Mean Squared Errors (MSE) value of the SUR
model for each equation is less than the corresponding root MSE values in OLS models
implying that using SUR model introduces less error terms to the equations. Furthermore, the
results showed that McElroy R-squares SUR model are all greater than that of their
corresponding Adjusted R-squared in OLS models which implied that the explanatory
variables better explained the dependent variables in SUR model. Hence, the use of SUR
model over OLS models is well justified here also.

Table 24. Models R-squares and Root MSE valued for SUR and OLS models

OLS model SUR model

Parameters MI COCI SDI MI COCI SDI

Adj. R-sq (R2) 20.04 24.93% 29.93% 25.34% 28.77% 35.23%

Root MSE 0.206 0.242 0.215 0.199 0.235 0.211
COCI= crop output commercialization index; MI=mechanization index and SDI= crop
diversification index measured in Simpson Diversification Index
Source: Author’s computation from sample survey data (2022).

The model goodness of fit showed that as a model shifted from OLS to SUR models, the
model error terms decreased by 3.52, 2.98 and 1.90% for level of farm mechanization (MI),
crop output commercialization index (COCI) and crop diversification (SDI), respectively.
Similarly, the R-squares were increased by 5.30, 3.84 and 5.30, respectively for MI, COCI and
SDI respectively.
 147

Table 25. Correlation matrix of residuals for SUR and OLS models

Variables correlated SUR OLS

Mech. Index vs Com. Index 12.45% 11.65%

Mech. Index vs SDI 7.99% 7.61%
Com. Index vs SDI -10.79% -11.13%
Breusch–Pagan test of independence (SUR): chi2(3) =11.026; Pr. =0.0116
Source: Author’s computation from sample survey data (2022).

The application of SUR model improves the correlation among equations except for
commercialization index vis-à-vis crop diversification index (SDI). In the first two cases; the
correlation between the error terms for farm mechanization index and commercialization index
increased from 11.65% to 12.45% as we shift the model from OLS to SUR, while it increased
from 7.61% to 7.99% for farm mechanization versus crop diversification correlation.
However, in the case of correlation between crop commercialization and crop diversification,
the coefficient decreased from 11.13% to 10.79% (in absolute value). In general, the use of
SUR model over OLS model is preferred as the Breusch-Pagan test of independence for the
SUR model is significant at 5% with 11.026 chi-square value where we can strongly reject the
null that the equations are independent. This implies that the correlations between the residuals
of multivariate model with the SUR estimator indicate that all equations are moderately
interrelated. Hence, the values showed that the equations are better determined together as the
error terms are strongly correlated. Hence, the use of SUR model is well justified (Table 25).
 148

Table 26. Estimation of determinants of mechanization, commercialization and crop diversification using OLS model

Farm mechanization level Commercialization level Crop diversification level

Variables Coeff. t-ratio Coeff. t-ratio Coeff. t-ratio

Sex -0.108(0.08) -1.40 0.038(0.091) 0.42 -0.082(0.076) -1.09

Age 0.002(0.001) 1.51 -0.003(0.001) -2.28** 0.001(0.001) 1.42
Education 0.007(0.003) 2.24** -0.005(0.004) -1.15 0.002(0.003) 0.52
Family size 0.002(0.003) 0.85 -0.002(0.003) -0.74 0.001(0.003) 0.22
Dependency ratio _ _ - - -0.004(0.008) -0.49
Landholding 0.008(0.010) 0.82 0.014(0.012) 1.19 0.054(0.010) 5.62***
TLU -0.002(0.003) -0.76 -0.002(0.004) -0.45 -0.001(0.003) -0.42
Oxen (no.) 0.015(.009) 1.66* 0.018(0.012) 1.68* -0.002(0.009) -0.17
Topography (reference is high)
Medium -0.007(0.046) -0.16 - - 0.120(0.047) 2.53**
Low sloppy 0.017(0.04) 0.38 - - 0.050(0.046) 1.09
Flat -0.0001(0.04) -0.00 - - 0.097(0.046) 2.12**
AEZs 0.146(0.03) 5.11*** 0.049(0.031) 1.58 -0.113(0.028) -4.05***
Near mkt dist. 0.001(0.003) 0.33 -0.007(0.004) -1.75* - -
Main mkt dist. -0.002(0.003) -0.87 -0.009(0.004) -2.46** 0.003(0.002) 1.55
Mech. Dist -0.007(0.002) -2.96*** - - - -
 149

Table 26. Estimation of determinants of mechanization, . . . (Continued)

Coop dist - - 0.001(0.004) 0.28 - -

Road dist - - -0.013(0.008) -1.73* - -
FTC dist - - -0.003(0.007) -0.45 - -
Off farm - - 0.012(0.035) 0.35 - -
Lag price percept - - 0.059(0.030) 1.98** - -
Income (log) - - 0.064(0.007) 9.06*** - -
Social capital -0.060(0.06) -1.02 - - - -
Medium risk percept - - - - 0.017(0.032) 0.55
High risk percept - - - - -0.024(0.028) -0.85
Plots numbs -0.044(0.01) -4.14*** - - 0.078(0.010) 7.54***
Prod. Goal -0.002(0.01) -0.15 - - - -
Constant 0.396(0.11) 3.61*** -0.240(0.138) -1.74* 0.167(0.096) 1.73*

Source: Author’s computation from sample survey data (2022).

 150

[Link]. Results of OLS estimation

The result of OLS model showed that educational background of the household’s head, number
of oxen, agroecological zone (being in a mid-land agroecological zone) and production goal i.e.,
producing for market purposes were positively affecting (influencing) the level of farm
mechanization. in contrast, distance to mechanization service providing centers and land
fragmentation measured in terms of number of plots were variables that affect level of farm
mechanization (Table 26).

On the other hands, level of crop output commercialization was positively affected by number of
oxen, perception on lagged price of their product (those perceiving the previous year price was
less than this year’s price), and total annual income. The same was negatively affected by age of
the household heads, distance to nearer and main markets, distance to main roads.

Crop diversification index (measured as SDI) is also positively affected by size of landholding,
topography and land fragmentation measured in terms of number of plots. Medium slope and flat
slope positively affect crop diversification compared to highly slopped farm lands, implying high
slope associated with low diversification. The result also showed that highland areas are more
diversified than mid-land which is more coinciding with researchers’ observation where mid-
land areas are more wheat dominant areas.

[Link]. Results of SUR Estimation

The estimated parameters and summary statistics based on SUR estimator for the farm
mechanization level, crop output commercialization and crop diversification level were
discussed (Table 27). The result of SUR model showed that educational background of
household heads, number of oxen possessed and agroecological zone (mid-land) were variables
that affect level of farm mechanization. in opposite to this, distance of farm mechanization
service rendering centers, social capital and land fragmentation (number of plots) were affecting
farm mechanization’s level negatively.
 151

Crop output commercialization intensity was negatively influenced by age of households,

distances of household’s residence from near and main market centers and main (all-weather)
road. On the other hand, variables like number of oxen, agroecological zone (being mid-land),
lagged price perception and amount of household’s annual income were positively affecting
intensity of crop commercialization. Landholding, topography of plots (lower slop) and land
fragmentation (number of plots) have positive impacts on level of crop diversification while,
agroecological zone (being in mid-land compared to highland) have negative significant impacts
on crop diversification level.

Comparison of OLS and SUR revealed that in some cases variables which were insignificant in
the OLS equation became significant in the SUR for instance, social capital in level of farm
mechanization. However, the SUR estimates provide relatively lower standard errors and, hence,
provide more efficient estimates than OLS estimates and, in most cases, the t-ratios are relatively
higher for SUR models. The standard errors’ reduction was observed for the intercepts of level of
farm mechanization and crop output commercialization where it was reduced by 0.92 and 3.33%
respectively between OLS and SUR estimation methods.
 152

Table 27. Estimation of determinants of mechanization, commercialization and crop diversification using SUR model

Farm mechanization level Commercialization level Crop diversification level

Variables Coeff. t-ratio Coeff. t-ratio Coeff. t-ratio
Sex -0.095(0.07) -1.27 0.035(0.09) 0.40 -0.108(0.08) -1.37
Age 0.001(0.001) 1.43 -0.003(0.00) -2.35** 0.001(0.00) 1.04
Education 0.006(0.003) 1.75* -0.004(0.00) -1.12 0.002(0.00) 0.61
Family size 0.001(0.01) 0.05 -0.002(0.00) -0.77 0.001(0.00) 0.35
Dependency ratio - - - - -0.002(0.01) -0.22
Landholding 0.009(0.01) 0.85 0.015(0.01) 1.27 0.050(0.01) 4.83***
TLU -0.001(0.00) -0.16 -0.002(0.00) -0.41 0.002(0.00) 0.56
Oxen (no.) 0.017(0.01) 1.84* 0.018(0.01) 1.68* -0.006(0.01) -0.60
Topography (reference=high)
Medium 0.002(0.04) 0.04 - - 0.096(0.05) 2.00**
Low sloppy 0.019(0.04) 0.43 - - 0.059(0.05) 1.28
Flat -0.010(0.04) -0.33 - - 0.088(0.05) 1.91*
AEZs 0.163(0.03) 5.56*** 0.050(0.03) 1.65* -0.089(0.03) -2.91**
Near mkt dist. -0.001(0.00) -0.20 0.007(0.003) 2.00** - -
Main mkt dist. -0.003(0.00) -1.09 -0.009(0.003) -2.61** 0.002(0.00) 0.63
Mech. Dist -0.007(0.00) -3.06*** - - - -
 153

Table 28. Estimation of determinants of mechanization, . . . (continued)

Coop dist - - 0.001(0.00) 0.23 - -

Road dist - - -0.013(0.01) -1.74* - -
FTC dist - - -0.003(0.01) -0.49 - -
Off farm - - 0.009(0.03) 0.26 - -
Lag price percept - - 0.067(0.03) 2.35** - -
Income (log) - - 0.061(0.01) 9.11*** - -
Social capital -0.095(0.05) -1.92 - - - -
Medium risk percept - - - - 0.019(0.03) 0.57
High risk percept - - - - -0.018(0.03) -0.62
Plots numbs -0.046(0.01) -4.26*** - - 0.089(0.01) 7.81***
Prod. Goal 0.003(0.01) 0.22 - - - -
Constant 0.411(0.11) 3.79*** -0.231(0.13) -1.73* 0.169(0.09) 1.71*
Source: Author’s computation from sample survey data (2022).
 139

5. SUMMARY, CONCLUSIONS AND RECOMMENDATIONS

5.1. Summary

Given the contribution of agriculture to the national economy in Ethiopia, and the infancy of
other sectors, one can say that the sector may continue as backbone in the economic
development of the country for the coming years. Hence, the policy direction for economic
development of Ethiopia should put agricultural modernization at pivotal. Farm mechanization
is often considered as agricultural modernization, while crop output commercialization and
crop diversification are crucial strategies for risk management. This study investigated the
level of farm mechanization, determinants of farm mechanization, its use and impacts on
household income, labor displacement, inputs use intensities and economic efficiencies of
wheat and barley producers. Furthermore, the study investigated the level of crop
commercialization, diversification and the interlinkage among farm mechanization, crop
commercialization and diversification in central and southern Oromia regional state of
Ethiopia.

For these investigations, informal and formal surveys were conducted in 8 kebeles selected
from 4 districts in Arsi and West Arsi zones. Following multistage sampling technique, 390
household heads were randomly drawn. Structured and pre-tested questionnaire were used to
collect household level data through face-to-face personal interview by well-experienced and
trained enumerators from research centers. Published and unpublished secondary sources were
used to gather data on mechanization, crop output commercialization and crop diversification
status of each district. Furthermore, FGD and KII were used to enrich the quantitative data and
get deep insight of the subject matter. Descriptive statistics and econometric models were used
to analyze the data. Econometric models including Tobit model, Stochastic Frontier Analysis
model, AIPW model, Heckman two-step selection model, Ordered Logit model, and
Seemingly Unrelated Regression (SUR) model were employed.

The descriptive analyses provided the sketch of issues related to the socioeconomic and
institutional characteristics of sample farm households. Moreover, indicators and level of farm
 140

mechanization, efficiency, commercialization, and crop diversification were also discussed

using the descriptive statistics and mathematical computation of some indexes like SDI, and
others.

The empirical analysis of determinants of smallholder households’ farm mechanization level

was computed by using Tobit model. The result showed that sex, spatial location, livestock
size, land fragmentation, distance from farm mechanization service centers, and social capital
negatively affected level of farm mechanization. However, farming experience, education,
participation in nonfarm/off-farm income activities, size of land cultivated, market
participation, and access to road affected level of farm mechanization positively.

The SFA model result for TE and EE of wheat and barley producers showed that the mean TE
and EE of the wheat producers were 94 and 72%, respectively. The mean TE and EE for
barley producers were 91 and 73%, respectively. The result also shows that the potential mean
outputs of barley are 42.36, 45.42, and 49.82 quintals per hectare for low, medium, and high
levels of farm Mis, respectively and it shows that farm productivity is increasing with an
increase in the level of farm mechanization. The AIPW model output revealed that as the use
of farm mechanization was increasing, the number of labors employed and amounts of
agrochemicals were also decreasing, implying the negative impacts of farm mechanization on
these two outcome variables. On the other hands, the result further implied that farm
mechanization has a positive impact on economic efficiency of wheat and barley producers,
sample smallholder households’ income, amounts of chemical fertilizer applications, and
wheat and barley land productivities.

Heckman two-stages selection model was employed to examine smallholder farmers’ crop
output market participation decision and level of participation. Farm household’s decision to
participate in crop output market was significantly influenced by livestock holding (TLU),
agroecological zones (being in mid-land), equine holding (dummy), amounts of income and
access to market information positively. Age of household head, distance from main market
center and instrumental variable for farm mechanization were negatively and significantly
affecting households’ market participation. In addition to this, farm households’ intensity of
 141

crop output market participation was positively influenced by number of oxen holding, amount
of annual income, and access to market information while it was negatively influenced by
distance from market center, level of crop diversification and size of cultivated land.

Ordered Logit model was employed to analyze the factors that affect the level of crop
diversification. Probability of a households’ falling in the range of highly diversified crop
producers was 28% while the majority of the householders were medium level diversified. The
result of the model showed that family size, total farmland, level of farm mechanization and
land fragmentation negatively affect the probability of being in lower-level diversified crop
production system. On the other hands, agroecological zone (being in mid-land AEZ)
negatively affected the probability of highly diversified crop production while affected
probability of low-diversified production positively. Similarly, the above four variables
positively affected probability of highly diversified crop production.

As the Breusch–Pagan test of independence favored the Seemingly Unrelated Regression

(SUR) model, it was used to analyze the interlinkage among level of farm mechanization, crop
commercialization level and level of crop diversification. The correlation coefficient between
mechanization index and commercialization index, and mechanization index vis-à-vis crop
diversification index were positive and significant while the coefficient between
commercialization and crop diversification index were negative and significant. The SUR
model revealed that education, AEZ, oxen number possessed, distance from mechanization
service centers, and land fragmentation were significantly determining level of farm
mechanization. Furthermore, it showed that age, number of oxen, AEZs, near market distance,
distance from main road, lag price of crop and income amounts were important variables in
determining level of crop commercialization. Finally, the model output disclosed that
landholding, plot topography, AEZs, and land fragmentation were determinants of crop
diversification.
 142

5.2. Conclusions and Recommendations

Even though it is clear that agricultural modernization is important to feed the ever-growing
population, most agricultural activities of the country are managed by animal and human
power, implying the level of farm mechanization in the study area is still in infant stage. There
are different factors that deter the level of farm mechanization. Hence, improving farmers
knowledge and understanding through formal and informal education and experience-sharing
among themselves using experienced farmers can enhance the farm mechanization level.
Topography, distance from mechanization service provider centers, access to all-weather roads
and spatial location of the household also matters in the process of farm mechanization.
Hence, selection and demonstration of appropriate technologies according to compatibility to
the existing topography is an important intervention issue to be considered in the future.
Moreover, land clustering, introduction of land augmenting mechanization technologies and
land consolidation are important to increase farm mechanization level of the area. Therefore,
recent government initiatives such as cluster farming should be scaled up as landholding size
is the most important variable that positively affects the level of farm mechanization.

Furthermore, in case where land is the limiting factor for farm mechanization technology
adoption, the use of land augmenting mechanization technologies such as irrigation
technologies can also improve the use of other farm mechanization technologies like tractors
and harvesting and threshing technologies. In the near-future, the best approach to mechanize
Ethiopian agriculture is through custom-hiring service provision. In such cases, improving
credit facilitation and establishing nearby custom service providing centers can enhance access
to technologies and thereby improve the level of farm mechanization. Market participant
households were also more mechanized. Hence, policies should aim at improving households’
market participation to improve the level of farm mechanization.

The economic inefficiency of wheat production ranges from 5 to 95% having an average
inefficiency of 28% implied that there is a large potential to be utilized by smallholders
without incurring additional costs and inputs. Hence, by improving the EE of wheat producers,
it is possible to increase the current wheat production and productivity by around 28%.
 143

Similarly, the economic inefficiency of smallholder barley producers varies from 2to 98%
with average inefficiency of 27%. As a result, it is also possible to improve the existing
production and productivity of barley producers by around 27% with the existing resources in
use. In addition to production improvement, production cost of wheat and barley can also be
reduced without reduction in the amount of current production. Hence, to take policy and
development intervention actions, further investigation shall be done to identify the main
causes for the inefficiency of the producers.

Higher level of farm mechanization has negative impacts on agricultural labor employment of
households in the area, which is expected. This implies that farm mechanization has labor
displacement effects in Ethiopian agriculture. This may continue to be true at the national level
where the use of farm mechanization cannot improve the actual cropping intensity due to low
irrigation facilities. However, there is also a paradox where there is an increase in agricultural
labor wage through time which could be due to ever increase of living cost in rural areas.
Hence, the increase in the level of farm mechanization would be the best solution to the
alarmingly increasing agricultural wage. In general, land augmenting technologies that can
improve landholding such as irrigation, which can increase cropping intensity, and improving
the level of farm mechanization have to be planned parallelly during policy design and
development interventions.

This study found that higher level of farm mechanization also encourages use of chemical
fertilizers. However, the current rate of chemical fertilizer application is below the
recommended rate on all levels (national, regional and household levels), and hence there is a
need for improvement. In general, the result indicated that the use of farm mechanization,
amounts of chemicals (synthetic fertilizers) and agricultural productivity goes together. Hence,
to improve productivity in Ethiopia, the use of optimum inputs like fertilizer is important and
policies, and development intervention should aim at improving the level of farm
mechanization to improve chemical fertilizer utilization intensity and thereby increase
production and productivity.
 144

Farm mechanization has also significant effects on the agrochemical utilization level to control
diseases and pest. The low level of chemical application may be attributed to the low
infestation of diseases and pests. For instance, the use of tractors may reduce weed infestation.
Studies also unveiled that weed infestation caused significant yield losses besides impairing
product quality. Therefore, the use of farm mechanization has multiple advantages: first, it
reduces the amounts of chemicals to be applied which reduces the production cost; and
second, it reduces the weed infestation and contributes to the production of quality grains.
Moreover, the wider use of farm mechanization, especially precision agriculture, can reduce
the long run residual effects of chemicals on the environment and human health. Less
application of agrochemicals may also have sounded implications on environmental
protection.

More application of farm mechanization has also positive and significant impacts on economic
efficiency of smallholder wheat and barley producers. The per hectare productivity of wheat
also showed significant improvement, but the value for barley producers was not statistically
significant even though there is positive change. The insignificance for the low productivity of
barley producers may attributed to the low level of farm mechanization in barley producer
smallholder farmers. Barley production is more dominant in highland and where the
topography is somehow inconvenient. Hence, improving levels of farm mechanization should
be one of the strategies to enhance economic efficiencies and productivities of wheat and
barley farmers. Smallholder farming household income was another outcome variable that was
positively and significantly affected by the higher level of farm mechanization.

In general, the AIPW model results clearly indicate that improving the level of farm
mechanization has significant impact on reduction of amounts of labors required that can solve
labor shortage problem, and enhances the application of chemical fertilizer. This implies that
farm mechanization can directly or indirectly improve production and productivity, reduce the
amount of agrochemicals by reducing pests and disease infestation, and improve productivity
and economic efficiencies of wheat and barley producers. Finally, the increase in productivity
directly boosts household income which can improve household welfare. Hence, policy design
 145

should focus on how to improve the level of farm mechanization in order to make
smallholders more beneficial.

The findings of this study pointed out that the age of the household head is negatively and
significantly associated with the probability of crop market participation. The plausible
reasons are that aged household heads are facing labor shortage to produce a surplus and are
less educated and less market oriented. On the other hands, households with aged household-
heads are possessing larger farm size. Hence, to increase the number of market participant
households, targeting labor saving technologies to solve labor shortage, development of
business models that can link aged households with energetic, market oriented and innovative
farmers, could be some of the future intervention to solve problem of low market participation
among older age farmers.

Resource ownership like livestock is one of the important variables in the areas of agricultural
development. Livestock is an important source of income, fertilizer as a manure, and more
importantly a draught power. As a result, those households possessing more size of Tropical
Livestock Units (TLU) are expected to produce more surplus grains to market. Hence,
improving households’ resource possession like livestock ownership through improved
livestock production system can enhance market participation in the future. Agroecological
zones (AEZs) determine the types crops and production systems of the area. It also affects
productivity and production since natural factors of production like rainfall, soil types and
severity of pest infestation, length of rain seasons. In most cases, AEZs in the study area is
also associated with topography in which the highland areas are poor in physical
infrastructures like a road which in turn directly affect the market participation of a household
negatively. In addition to these, availability of marketable crops, soil fertility status and other
factors associated with AEZs should be examined and get solution in order to enhance market
participation of the households especially in highland areas.

The variables that directly affect market participation; access to market information, distance
from households’ residence to market centers, and possession of equine animals for
transportation purposes were important in determining probability of market participation. The
 146

result clearly unveiled that market participation is highly determined by market infrastructures.
Therefore, to enhance market participation, establishment of market information system and
availing market information, enhancing the use of mobile phone and increasing farmers’
awareness to use the available market information at wereda (district) level, establishing small
village-level markets at an accessible area, and promotion of rural transportation means like
carts in addition to the use equine animals need to getting proper attention from development
practitioners.

Level of farm mechanization is one of the important variables that affect both participation and
level of market participants. This is due to fact that amount of crop production is positively
affected by level of farm mechanization which affects the surplus that is supplied to the
market. Hence, policy designs and development intervention must work towards improving
the level of farm mechanization in order to improve market participation probability and level
of market participation. The income level of a household is also affecting market participation.
The apparent and valid reason could be that households generating more income from either
farm, or nonfarm sources or both, have more probability of investing on their farm activities
and make their farms more productive and produce surplus for the market. Therefore,
activities that enhance income of farming households both farm and nonfarm sources should
get attention to make farmers more market oriented.

The result showed that land is in the hands of elders who are less market oriented and have a
less commercialization level. Hence, developing business models that can allow the youth to
use land found in the hands of such older farmers can improve the level of commercialization.
Market distance from a household’s residence, and access to market information were also
important factors in determining commercialization level. Therefore, policy design shall focus
on improving access to market information by smallholder farmers. Adoption of more
diversified crop production system was also an important factor in boosting crop
commercialization levels. The plausible reason could be that farmers have some high valued
crops which are primarily produced for commercial/market purpose. Hence, identifying such
crops and improving crop diversification level can improve market participation and enhance
commercialization level of the households.
 147

Resource ownership base of the households like oxen are an important factor that enables a
household to produce a surplus for market purposes. Moreover, as oxen are important sources
of power in agricultural activities, possession of more oxen implies that the household has no
problem of animal power shortage and can perform its agricultural activities timely. So, policy
should focus on improving the resource bases of households especially on improving livestock
production systems. Smallholder farms households having larger farmland can relax more in
order to grow more crop types. Hence, possession of larger farmland is important for the
enhancement of crop diversification. Therefore, land augmenting technologies like high
yielding varieties, irrigation technologies and other inputs shall be the focus of policy
designers to solve land shortage constraints in agricultural production.

Crop diversification was positively and significantly affected by level of farm mechanization.
Diversified crop production demands more labor and intensive crop field management to
control the different incidence of diseases and pests. Hence, the use of more mechanization
technologies can solve problems of labor shortage. Hence, policies shall focus on improving
access to farm mechanization technologies and demonstration of alternative appropriate
mechanization technologies. Agroecological zone (AEZ), notably being in highland areas
positively affected the level of crop diversification. The most probable reason could be
availability of alternative crops for diversification especially vegetable crops like beetroots,
carrots, potatoes, and cabbages and etc., and higher land fragmentation in the areas. Hence,
improving such conditions like access to farm mechanization, availing alternative crops
through agricultural research centers can enhance the crop diversification level in mid-land
agroecological zones.

Even though land fragmentation is not desirable in order to adopt other modern agricultural
production systems like large farm mechanization at least at smallholder level, it is positively
associated with crop diversification. The credible reason is that, farmers having fragmented
plots of farms can have plots with different soil types, fertility status and even the slightest
difference in agroecological zones that bring variations in time of sowing and harvesting, crop
types etc. Generally, all these factors can initiate a household to produce diversified types of
crops. Hence, encouraging smallholder farmers to work on different farm plots may enhance
 148

the level of crop diversification and helps them to minimize risk and improve their nutritional
security. Furthermore, introduction of competitive alternative crop types by agricultural
research institutes and extension may enhance crop diversification. On the other hands, deep
investigation on crop choice of more diversified households and scaling-up of their experience
can also raise levels of crop diversification.

The results from the SUR model revealed that there were significant and positive associations
between level of farm mechanization and crop output commercialization and farm
mechanization vis-à-vis crop diversification level. Hence, to improve levels of crop output
commercialization and crop diversification, it is important to improve adoption of farm
mechanization technologies by working on recommended policy actions on factors that
enhance the level of farm mechanization in previous paragraphs of this document and vise-
versa. On the other hands, level of crop output commercialization and crop diversification
were correlated negatively and significantly. The result may be an indication of the fact that
commercialized farmers were more specialized and vise-versa. However, that may not be due
to the less market orientation (understanding level) of specialized (less diversified)
households. The reason could be some others like less farmland size and others. Hence, policy
design of the future should focus on how to boost parallelly both level of commercialization
and crop diversification in the study area.

It is clear that everything cannot be addressed in one study and a result is important to suggest
future research direction in the areas of farm mechanization, crop output commercialization,
and crop diversification in Ethiopia. The scope of this study is limited to specific areas and
cross-sectional data. It is necessary to investigate the factors that contribute to farm
mechanization, crop commercialization, and crop diversification using a larger area that
represents national agroecological zones. Additionally, time series data can provide superior
results for improved policy recommendations by understanding the impact of farm
mechanization on overtime improvements. Farm mechanization's impact on community level,
including deforestation, erosion, and other impacts, was not covered in the study. Therefore,
future research that focuses on the impact/effects of farm mechanization on the community
level can provide good directions for policy makers and development practitioners.
 149

6. REFERENCES

Abafita, J., Atkinson, J. and Kim, C. S. 2016. Smallholder commercialization in Ethiopia: market
orientation and participation. International Food Research Journal, 23(4).
Abbas, A., Minli, Y., Elahi, E., Yousaf, K., Ahmad, R., and Iqbal, T. 2017. Quantification of
mechanization index and its impact on crop productivity and socio-economic factors.
International Agricultural Engineering Journal, 26(3).
Abenet Belete, Dillon, J.L. and Anderson, F.A. 1991. Development of agriculture in Ethiopia since
the 1975 land reform. Agricultural Economics, 6(2):159-175.
Abro, A.A. 2012. Determinants of Crop Diversification towards High Value Crops in Pakistan.
International Journal of Business, Management and Economics Research, 3(3): 536-545.
Adetule, F.S., Owoeye, R.S., and Sekumade, A.B. 2021. Nexus of Agricultural Commercialization
among Farming Households in Southwest, Nigeria. International Journal of Research, 7(6),
16-24.
Adino Andaregie, Tessema Astatkie and Fentaw Teshome. 2021. Determinants of market
participation decision by smallholder haricot bean (phaseolus vulgaris l.) farmers in Northwest
Ethiopia. Cogent Food & Agriculture, 7(1), 1879715.
Adino Andaregie, Tessema Astatkie. 2020. Determinants of technical efficiency of potato farmers
and effects of constraints on potato production in Northern Ethiopia. Experimental
Agriculture, 56(5), 699-709.
Adnan, K. M., Ying, L., Ayoub, Z., Sarker, S. A., Menhas, R., Chen, F. and Yu, M. M. 2020. Risk
management strategies to cope catastrophic risks in agriculture: the case of contract farming,
diversification and precautionary savings. Agriculture, 10(8), 351.
Afridi, F., Bishnu, M. and Mahajan, K. 2020. Gendering Technological Change: Evidence from
Agricultural Mechanization. IZA – Institute of Labor Economics, Discussion paper series IZA
DP No. 13712.
Agete Jerena Kabeto. 2014. An Analysis of Factors Influencing Participation of Smallholder
Farmers in Red Bean Marketing in Halaba Special District, Ethiopia. MSC Thesis, University
of Nairobi, Nairobi, Kenya.
Ahmed, S., Bagal, Y.S., Mahajan, R. and Sharma, L.K. 2020. Impact of Farm Mechanization on
Crop Productivity in Sub-Tropical Areas of Jammu and Kashmir. International Journal of
 150

Current Microbiological and Applied Science, 9(5): 1168-1173.

Ai, J.; Hu, L.; Xia, S.; Xiang, H.; Chen, Z. 2023. Analysis of Factors Influencing the Adoption
Behavior of Agricultural Productive Services Based on Logistic—ISM Model: A Case Study
of Rice Farmers in Jiangxi Province, China. Agriculture, (13): 162.
[Link]
Aigner, D. and Chu, S. 1968. On Estimating the Industry Production Function. American Economic
Review, 58, 826-839.
Aigner, D.J., Lovell, C.K. and Schmidt, P. 1977. Formulation and estimation of stochastic frontier
production function models. Journal of Econometrics, 6, 21-37.
Akram, W.W., Akram, N., Wang, H., Andleeb, S., Rehman, K.U., Kashif, U. and Hassan, S.F.
2020. Socioeconomics Determinants to Adopt Agricultural Machinery for Sustainable Organic
Farming in Pakistan: A Multinomial Probit Model. Sustainability, 12 (9806).
Albiero, D., Xavier, R. S., Garcia, A. P., Marques, A. R., and Rodrigues, R. L. 2019. The
technological level of agricultural mechanization in the state of ceará, brazil. Engenharia
Agrícola, 39, 133-138.
Alderman, H. 1995. Unitary Versus Collective Models of the Household: Is it Time to Shift the
Burden of Proof? The World Bank Research Observer, 10 (1), 1-19.
Alene, A.D., Manyong, V.M., Omanya, G., Mignouna, H.D., Bokanga, M. and Odhiambo, G. 2008.
Smallholder Market Participation under Transaction Costs: Maize Supply and Fertilizer
Demand in Kenya. Food Policy, 33(2008), 318-328, Elsevier.
Almasi, M., S. Kiani, and N. Loui-mi. 2000. Principles of Agricultural Mechanization. Ma’soumeh
(PBUH) Publication. Ghom, Iran. PP. 19-40.
Alula Tafesse, Gezahagn Gechere, Alemayehu Asale, Abrham Belay, John W. Recha, Ermias
Aynekulu, Zerihun Berhane, Philip M. Osano, Teferi D. Demissie and Dawit Solomon. 2023.
Determinants of maize farmers market participation in Southern Ethiopia: Emphasis on
demographic, socioeconomic and institutional factors. Cogent Food & Agriculture, 9(1),
2191850.
Alula Tafesse, Guta Regasa Megerssa and Bogale Gebeyehu. 2020. Determinants of agricultural
commercialization in Offa District, Ethiopia. Cogent Food & Agriculture, 6 (1), 1816253.
Amirani, [Link]. E. 2001. Economic factors and their relationship with the promotion of agricultural
mechanization. Jihad Scientific Monthly. No. 239-238, April and May 2001.30.
 151

Amoozad-Khalili M, Rostamian R, Esmaeilpour-Troujeni M, Kosari-Moghaddam A. 2020.

Economic modeling of mechanized and semi-mechanized rainfed wheat production systems
using multiple linear regression model. Information Processing in Agriculture, 7(1), 30–40.
[Link] inpa.2019.06.002
Amrit, P. 2000. Infrastructure for Agriculture & Rural Development in India. Need For A
Comprehensive Program & Adequate Investment.
Aneani, F., Anchirinah, V. M., Owusu-Ansah, F. and Asamoah, M. 2011. An analysis of the extent
and determinants of crop diversification by cocoa (Theobroma cacao) farmers in Ghana.
African Journal of Agricultural Research, 6(18), 4277-4287.
Antle, J.M. and Capalbo, S.M. 1988. An introduction to recent developments in production theory
and productivity measurement. pp.17-95. In: Capalbo, S.M. and Antle, J.M. (eds.),
Agricultural Productivity Measurement and Explanation, Resources for the Future,
Washington D.C.
Anwar Muzemil. 2020. Analysis of market participation of small dairy farmers in Gubre town,
SNNPR, Ethiopia. International Journal of Agricultural Science, 6(1), 058–067.
doi:10.17352/2455-815X.000056.
Arega, D and Rashid, M. 2005. The efficiency of traditional and hybrid maize production in
Eastern Ethiopia: An extended efficiency decomposition approach. Journal of African
Economies, 15(1), 91-116.
Aromolaran, A., Muyanga, M., Jayne, T., Obayelu, A. E., Awokuse, T., Ogunmola, O. O. and Issa,
F. O. 2020. Drivers of Market-Oriented Land Use Decisions among Farm Households in
Nigeria.
Arslan, A., Cavatassi, R., Alfani, F., Mccarthy, N., Lipper, L. and Kokwe, M. 2018. Diversification
under climate variability as part of a CSA strategy in rural Zambia. The Journal of
Development Studies, 54(3), 457–480.
Arun, G.C, Yeo, J., Ghimire, K. 2019. Determinants of Farm Mechanization in Nepal. Turkish
Journal of Agriculture - Food Science and Technology, 7(1), 87-91.
Aryal, J. P., Rahut, D. B., Maharjan, S. and Erenstein, O. 2019. Understanding factors associated
with agricultural mechanization: A Bangladesh case. World Dev. Pers., 13(2019), 1–9.
[Link]
Aryal, J. P., Rahut, D. B., Thapa, G. and Simtowe, F. 2021. Mechanisation of small-scale farms in
 152

South Asia: Empirical evidence derived from farm households survey. Tech. in Society,
Ashfaq, M., Hassan, S., Naseer, Z. M., Baig, A., & Asma, J. 2008. Factors affecting farm
diversification in rice–wheat. Pakistan Journal of Agricultural Sciences, 45(3), 45–47.
ATA and MoA 2014. Ethiopian National Agricultural Mechanization Strategy: Vision, Systemic
Challenges and Strategic Intervention. The Federal Democratic Republic of Ethiopia Ministry
of Agriculture and Agricultural Transformation Agency, working document.
ATA. 2019. Agricultural Transformation Agency. A quarterly newsletter: October-December.
Addis Ababa, Ethiopia.
Atinkugn Assefa Belete. 2020. Determinants of Market Participation of Smallholder Sorghum
Farmers and Strategies for Improving Their Participation; the Case of Moretna Jiru District,
Ethiopia. Research Square.
Austin, P. C. 2009. Balance diagnostics for comparing the distribution of baseline covariates
between treatment groups in propensity-score matched samples. Statistics in Medicine, 28:
3083–3107. [Link]
Awotide, B. A., Karimov, A. A. and Diagne, A. 2016. Agricultural technology adoption,
commercialization and smallholder rice farmers’ welfare in rural Nigeria. Agricultural and
Food Economics, 4, 1-24.
Aynalem Adugna. 2021. OROMIYA: Demography and Health. Available at
[Link]. accessed on 15, September 2021.
Baba, A.R. and Abdulai, A. 2020. Determinants of Crop Diversification and Its Effects on
Household Food Security in Northern Ghana. Journal of Economic Theory and Practice; 1–19.
Badunenko, O., Fritsch, M. and Stephan, A. 2008. Allocative efficiency measurement revisited—
Do we really need input prices? Economic Modelling, 25(5), 1093-1109.
Banik, A. (1994). Technical Efficiency of Irrigated Farms in a Village of Bengladesh. Indian
Journal of Agricultural Economics, 49(902-2018-3272), 70-79.
Bardhan, P. and Udry, C. 1999. Development Microeconomics. Oxford: Oxford University Press.
Barman, S. and Deka, N. 2019. Effect of Farm Mechanization in Human Labour Employment.
International Journal of Agricultural Science, (4):16-22.
Barrett, C. B. and Reardon, T. 2000. Asset, activity, and income diversification among African
agriculturalists: some practical issues. Available at SSRN 257344.
Barrett, C.B. 2008. Smallholder market participation: Concepts and evidence from Eastern and
 153

Southern Africa. Food Policy, 33(2008): 299–317.

Barrett, C.B., Reardon, T. and Webb, P. 2001. Nonfarm income diversification and household
livelihood strategies in rural Africa: concepts, dynamics, and policy implications. Food policy,
26(4): 315–331.
Battese, G. E. and Broca, S. S. 1997. Functional forms of stochastic frontier production functions
and models for technical inefficiency effects: a comparative study for wheat farmers in
Pakistan. Journal of productivity analysis, 8, 395-414.
Battese, G.E. and Coelli, T.J. 1995. A model for technical inefficiency effects in a stochastic
frontier production function for panel data. Empirical Economics, 20, 325–332.
[Link]
Baudron, F., Blackwell, J., Sims, B., Justice, S., Kahan, D.G., Rose, R., Mkomwa, S., Kaumbutho,
P., Sariah, P., Moges, G. 2015. Re-examining Appropriate Mechanization in Eastern and
Southern Africa: Two Wheel Tractors, Conservation Agriculture, and Private Sector
Involvement. Food Security, 7: 889–904.
Baudron, F., Misiko, M., Getnet, B., Nazare, R., Sariah, J. and Kaumbutho, P. 2019. A farm- level
assessment of labor and mechanization in Eastern and Southern Africa. Agronomy for
Sustainable Development, 39: 17.
Becker, G.S. 1965. A Theory of the Allocation of Time. The Economic Journal, (75) 299: 493-517.
Becker, G.S. 1974. A Theory of Social Interactions. Journal of Political Economy, (82): 1063-
1094.
Bedaso Taye, Wondwosen Tsegaye and Mesfin Ketema. 2012. Commercialization of Ethiopian
smallholder farmer’s production: Factors and Challenges behind. Paper presented on the Tenth
International Conference on the Ethiopian Economy, Ethiopian Economics Association, July
19-21, 2012, Addis Ababa, Ethiopia.
Bello, S.R. 2012. Agricultural Machinery & Mechanization: Basic Concepts. DPS Dominion
Publishing Services, USA.
Belton, B., Win, M. T., Xiaobo Zhang, X. and Filipski, M. 2021. The rapid rise of agricultural
mechanization in Myanmar. Food Policy, 101(2021) 102095.
Benin, S, M Smaleb, J Pender, GM Berhanu, and Ehui. 2004. The economic determinants of cereal
crop diversity on farms in the Ethiopian Highlands. Agri. Econ. 31(2004): 197-208.
Berg, M.M.V., Hengsdijk, H., Wolf, J., Ittersum, M.K.V., Guanghuo, W. and Roetter, R.P. 2007.
 154

The impact of increasing farm size and mechanization on rural income and rice production in
Zhejiang province, China. Agricultural Systems, 94 (2007) 841–850.
Berhanu Admassu and Yoseph Shiferaw. 2011. Donkeys, horses and mules - their contribution to
people’s livelihoods in Ethiopia. The Brooke, Addis Ababa, Ethiopia.
[Link]
Berhanu Gebremedhin and Moti Jaleta. 2010. Commercialization of smallholders: Does market
orientation translate into market participation? Improving Productivity and Market Success
(IPMS) of Ethiopian farmers project Working Paper 22. Nairobi, Kenya, ILRI.
Berhanu Gebremedhin and Moti Jaleta. 2012. Market orientation and market participation of
smallholders in Ethiopia: Implications for commercial transformation. Selected Paper
prepared for presentation at the International Association of Agricultural Economists (IAAE)
Triennial Conference, Foz do lguacu, Brazil, 18-24.
Berhanu Gebremedhin, Moti Jaleta and Hoekstra D. 2009. Smallholders, institutional services, and
commercial transformation in Ethiopia. Agricultural Economics, 40(s1):773–87. https:// doi.
org/ 10. 1111/j. 15740 862. 2009. 00414.x.
Billings, M. H., and Singh, A. 1970. Mechanisation and the wheat revolution: Effects on female
labour in Punjab. Economic and Political Weekly, A169-A174.
Binswanger, H.P. 1978. Measured Biases of Technical Change: The United States. pp. 215-242. In:
Binswanger, H.P. and Ruttan, V.W. (eds.), Induced Innovation, Technology, Institutions and
Development. Baltimore, Md., and London: Johns Hopkins University Press.
Binswanger, H.P. and Sillers, D.A. 1983. Risk aversion and credit constraints in farmers' decision‐
making: A reinterpretation, The Journal of Development Studies, 20:1, 5-21.
Birara Endalew, Mezegebu Aynalem, Fenta Assefa, and Zemen Ayalew. 2020. Determinants of
Wheat Commercialization among Smallholder Farmers in Debre Elias Woreda, Ethiopia.
Hindawi Advances in Agriculture, 2020, Article ID: 2195823.
Birthal, P. S. and Hazrana, J. 2019. Crop diversification and resilience of agriculture to climatic
shocks: Evidence from India. Agricultural systems, 173, 345-354.
Birthal, P.S., Roy, D. and Negi, D.S. 2015. Assessing the Impact of Crop Diversification on Farm
Poverty in India. World Development, 72: 70–92.
Bliss, C. J. and Nicholas, S. 1982. Palanpur: the economy of an Indian village. Oxford University
Press, Oxford, UK.
 155

Boserup, E. 1965. The Conditions of Agricultural Growth: The economics of agrarian change
under population pressure. New Brunswick, USA: Transaction Publishers.
Brant R. 1990. Assessing proportionality in the proportional odds model for ordinal logistic
regression. Biometrics; 46(4):1171–8.
Brasesco, F., Asgedom, D., Sommacal, V., Casari G. 2019. Strategic analysis and intervention plan
for wheat and wheat products in the Agro-Commodities Procurement Zone of the pilot
Integrated Agro-Industrial Park in Central-Eastern Oromia, Ethiopia. Addis Ababa. FAO. 104
pp. Licence: CC BY-NC-SA 3.0 IGO.
Bravo‐Ureta, B. E. and Rieger, L. 1991. Dairy farm efficiency measurement using stochastic
frontiers and neoclassical duality. American Journal of Agricultural Economics, 73(2), 421-
428.
Bravo‐Ureta, B. E. and Rieger, L. 1991. Dairy farm efficiency measurement using stochastic
frontiers and neoclassical duality. American Journal of Agricultural Economics, 73(2), 421-
428.
Brookhart, M., Schneeweiss, S., Rothman, K., Glynn, R., Avorn, J. and Stürmer, T. 2006. Variable
Selection for Propensity Score Models. American journal of epidemiology, 163(12): 1149-
1156.
Byerlee, D., De Janvry, A., Sadoulet, E., Townsend, R., and Klytchnikova, I. 2008. World
development report 2008: Agriculture for development.
Carvalho, E.H., Carvalho, E.H., Zilli, J.B., Mendes, A.S., Morello, G.M., Bonamigo, D.V., Zilli,
J.B., Mendes, A.S, Morello, G.M. and Bonamigo, D.V. 2015. Main Factors that Affect the
Economic Efficiency of Broiler Breeder Production. Brazilian Journal of Poultry Science,
17(1): 11-16.
Caunedo, J. and Kala, N. 2021. Mechanizing Agriculture Impacts on Labor and Productivity.
Working Paper 29061. National Bureau of Economic Research. 1050 Massachusetts Avenue
Cambridge, MA 02138 July 2021. Accessed on 3/11/2023.
CEMA. 2014. Advancing Agricultural Mechanization (AM) to Promote Farming & Rural
Development in Africa. CEMA (European Agricultural Machinery).
Chalwe, S.2011. Factors Influencing Bean Producers‟ Choice of Marketing Channels in Zambia.
An MSc. Thesis Presented to the School of Graduate Studies of Zambia University.
Chanie Ejigu Berhie. 2019. The Role of Crop Diversification to Household Food Security: The
 156

Case of Enderta Woreda, Tigray, Ethiopia. MSC Thesis Submitted to Addis Ababa University
College of Development Studies Center for Food Security Studies, Addis Ababa- Ethiopia.
Chauvin, N. D., Mulangu, F. and Porto, G. 2012. Food Production and Consumption Trends in
Sub-Saharan Africa: Prospects for the Transformation of the Agricultural Sector. United
Nations Development Programme, Working Paper 2012-011.
Chavas, J. and Holt, M. 1990. Acreage decisions under risk: the case of corn and soybeans.
American Journal of Agricultural Economics, 72 (3): 529-538.
Chayanov, A. V., 1966. The Theory of Peasant Economy, ed. and introduction by Daniel Thorner,
Basile Kerblay and R. E. F. Smith, Homewood, III.
Chen, M. A. and Dunn, E. 1996. Household economic portfolios. Assessing the Impact of
Microenterprise Services (AIMS). Management Systems International, Washington, D.C.
20024-2488.
Chernet Worku, Marlign Adugna and Essa Chanie Mussa. 2021. Determinants of market
participation and intensity of marketed surplus of smallholder chickpea producers in Este
woreda. Wiley Legume Science; Legume Science. 2022;4: e132. Legume Science.
2022;4:e132.
Chhatre, A., Devalkar, S. and Seshadri, S. 2016. Crop diversification and risk management in
Indian agriculture. Decision, 43, 167-179.
Chhun, C., Bora, B. and Sothy, E. 2015. Effect of labour movement on agricultural mechanisation
in Cambodia. Cambodia Development Resource Institute.
Chiappori, P. A. 1992. “Collective" Models of Household Behavior: The Sharing Rule Approach.
DELTA.
Chowdhury, J. 2005. Contested History of Neo-Classical Economic Models of Households in
Development. BRAC University Journal, (2)2: 23-33.
Clarke, L. J. 2000. Strategies for Agricultural Mechanization Development: The roles of the private
sectore and the Government.
Coelli T.J., Rao, D.S.P. and Battese, G.E. 1998. An Introduction to Efficiency and Productivity
Analysis. Kluwer Academic Publishers, USA.
Coelli, T.J., Rao, D.S.P., O’Donnell, C.J. and Battese, G.E. 2005. An Introduction to Efficiency and
Productivity Analysis, 2nd Edition. Springer Science + Business Media, Inc. New York, USA.
Cohen, J.M. 1987. Integrated Rural development: The Ethiopian experience and the Debate. The
 157

Scandinavian Institute of African Studies, Uppsala, Sweden.

Collado, M, Calderón E. 2000. Energía de utilizada para la producción agrícola en el estado de
Guanajuato: Cuantificación y análisis. Memorias del III Congreso Latinoamericano de
Ingeniería Agrícola. Guanajuato. México. (In Spanish - cross referenced).
Collender, R. and Zilberman, D. 1985. Land allocation under uncertainty for alternative
specifications of return distributions. American Journal of Agricultural Economics, 779-786.
Cossar, F. 2019. Impact of mechanization on smallholder agricultural production: evidence from
Ghana. Contributed Paper prepared for presentation at the 93rd Annual Conference of the
Agricultural Economics Society, University of Warwick, England, 15 – 17 April 2019.
Cragg, J. 1971. Some statistical models for limited dependent variables with application to the
demand for durable goods. Econometrica, 39: 829-844.
CSA (The Federal Democratic Republic of Ethiopia Central Statistical Agency). 2021a.
Agricultural Sample Survey 2020 / 2021 (2013 E.C.) (September – January 2020/2021)
volume vii report on area and production of major crops (private peasant holdings, Meher
season).
CSA (The Federal Democratic Republic of Ethiopia Central Statistical Agency). 2021b.
Agricultural Sample Survey 2020 / 2021 (2013 E.C.) (September – January 2020/2021)
volume VII report on crop and livestock product utilization (private peasant holdings, Meher
season).
Culas, R. J. 2006. Causes of farm diversification over time: An Australian perspective on an
Eastern Norway model. Australian Farm Business Management Journal, 3, 1–9.
Dagmawe Menelek Asfaw, Sirage Mohammed Shifaw and Atinkugn Assefa Belete. 2022.
Determinants of Market Participation Decision and Intensity among Date Producers in Afar
Region, Ethiopia: A Double Hurdle Approach, International Journal of Fruit Science, 22:1,
741-758, DOI: 10.1080/15538362.2022.2119189.
Dahlin, A.S. and Rusinamhodzi, L. 2019. Yield and labor relations of sustainable intensification
options for smallholder farmers in sub-Saharan Africa. A Meta-Analysis. Agron Sustain Dev
39(3):1–18. https:// doi. org/ 10. 1007/ s13593- 019- 0575-1.
Dasgupta, P. 1993. An inquiry into well-being and destitution. Oxford, Clarendon Press.
Datt, G. and Ravallion, M., 1998. Farm productivity and rural poverty in India. J. Dev. Studies
34(4), 62–85.
 158

Daum, T. 2023. Mechanization and sustainable agri-food system transformation in the Global
South. A review. Agronomy for Sustainable Development, 43(1), 16.
Daum, T., Adegbola, Y. P., Kamau, G., Kergna, O. A., Daudu, C., Zossou, R. C., Crinot, G. F.,
Houssou, P., Mose, L., Ndirpaya, Y., Wahab A. A., Kirui, O., A. Oluwole, F. A. 2020.
Impacts of agricultural mechanization: Evidence from four African countries. Hohenheim
Working Papers on Social and Institutional Change in Agricultural Development. 003-2020.
University of Hohenheim.
Daum, T., Baudron, F., Birner, R., Qaim, M., and Grass, I. 2023. Addressing Agricultural Labour
Issues is Key to Biodiversity-Smart Farming. Biological Conservation, 284, 110165.
[Link]
Dawit Alemu, Eleni Gabre-Madhin, and Samson Dejene. 2006. From Farmer to Market and Market
to Farmer: Characterizing Smallholder Commercialization in Ethiopia. Paper submitted for
ESSP 2006 Policy Conference on “Bridging, Balancing, and Scaling up: Advancing the Rural
Growth Agenda in Ethiopia” 6-8 June 2006, Addis Ababa, Ethiopia.
de Janvry, A., Fafchamps, M. and Sadoulet, E. 1991. Peasant Household Behavior with Missing
Markets: Some Paradoxes Explained. The Economic Journal, 101(409): 1400-1417.
Degrande, A., Schreckenberg, K., Mbosso, C., Anegbeh, P., Okafor, V., Kanmegne, J. 2006.
Farmers’ fruit tree-growing strategies in the humid forest zone of Cameroon and Nigeria.
Agroforest. Syst. 67 (2), 159–175.
Dembele, B.; Bett, H.K.; Kariuki, I.M.; LeBars, M.; Ouko, K.O. 2018. Factors influencing crop
diversification strategies among smallholder farmers in cotton production zone in Mali. Adv.
Agric. Sci. 2018, 6, 1–16.
Dereje Derso, Degefa Tolossa and Abrham Seyoum. 2022. A double hurdle estimation of crop
diversification decisions by smallholder wheat farmers in Sinana District, Bale Zone, Ethiopia,
CABI Agriculture and Bioscience (2022) 3:25; [Link]
Desale Gebretsadik. 2017. Technical, Allocative and Economic Efficiencies and Sources of
Inefficiencies among Large-scale Sesame Producers in Kafta Humera District, Western Zone
of Tigray, Ethiopia: Non-parametric approach. International Journal of Scientific &
Engineering Research, 8(6).
Dey, N. C., and Haq, F. 2009. Study of the Impact of Intensive Cropping on the Long-Term
Degradation of Natural Resources in Some Selected Agroecological Regions of
 159

Bangladesh. Center for agri-research and sustainable environment and entrepreneurship

development (CASEED), 29.
Di Falco, S. and Chavas, J.P. 2009. On crop biodiversity, risk exposure and food security in the
highlands of Ethiopia. Am J Agric Econ.; 91(3):599–613.
Di Falco, S. and Perrings, C. 2005. Crop biodiversity, risk management and the implications of
agricultural assistance. Ecological Economics, 55 (4): 459-466.
Diao, X., Silver, J., and Takeshima, H. 2016. Agricultural Mechanization and Agricultural
Transformation. Background Paper for African Transformation Report 2016: Transforming
Africa’s Agriculture. International Food Policy Research Institute.
Doorenbos, A.Z. 2014. Mixed Methods in Nursing Research: An Overview and Practical
Examples. Kango Kenkyu, 47(3): 207-217.
Drucza, K., Aregu, L., and Tsegaye, M. 2021. Agency, Gender and Development in Oromia,
Ethiopia. Cadernos de Estudos Africanos, (41), 15-75.
Ebenehi, E. L. and Lawal, R. A. 2019. Assessment of Agricultural Mechanization and its’
Productivity Functions in Ibaji Local Government Area of Kogi State, Nigeria. Paper
Presentation at 4th National Conference, School of Engineering. The Federal Polytechnic
Ilaro, Ogun State. 25th- 28th November, 2019.
Efa Gobena Turaa, Degye Goshub, Tinsae Demisiec and Tadesse Kenead. 2016. Determinants of
Market Participation and Intensity of Marketed Surplus of Teff Producers in Bacho and Dawo
Districts of Oromia State, Ethiopia. Journal of Economics and Sustainable Development, 7(1).
Eleni Zaude Gabre-Madhin. 2001. Market Institutions, Transaction Costs, and Social Capital in the
Ethiopian Grain Market. Research Report 124. International Food Policy Research Institute,
Washington D.C.
Ellis, B. H. 1992. The effects of uncertainty and source credibility on attitudes about organizational
change. Management Communication Quarterly, 6(1), 34-57.
Ellis, F. 1988. Peasant Economics, Farm Household and Agrarian Development. Newcastle,
Oxford University Press.
Ellis, F. 1993. Peasant Economics: Farm management in Agrarian development. Cambridge
University Press.
Engida Gebre, Agegnehu Workiye and Kusse Haile. 2021. Determinants of sorghum crop
commercialization the case of Southwest Ethiopia. Heliyon, 7(7),
 160

[Link].1016/[Link].2021.e07453.
Engle Warnick, J. C., Escobal, J. and Laszlo, S. C. 2011. Ambiguity Aversion and Portfolio Choice
in Small-Scale Peruvian Farming. The B.E. Journal of Economic Analysis &Policy, 11(1):1-
56. [Link]
FACASI project. 2016. Development of Agricultural Mechanization in Ethiopia and the Role of
National Policies, mimeo.
Fantu Bachewe, Guush Berhane, Bart Minten, and Alemayehu Seyoum Taffesse. 2016. Non-farm
income and labor markets in rural Ethiopia. Ethiopia Strategy Support Program, working
paper 90. Accessed on June, 2021 and available at
[Link]
FAO (Food and Agriculture Organization of the United Nations). 2012. Sustainable Crop
Production Intensification. Twenty-third Session. Rome.
FAO and UNIDO. 2008. Agricultural mechanization in Africa … Time for action: Planning
Investment for Enhanced Agricultural Productivity. Report of an expert group meeting jointly
held by FAO and UNIDO in Vienna on 29–30 Nov. 2007.
Färe R., Grosskopf S. and Lovell, C.A.K. 1985. Technical efficiency of Philippine agriculture.
Applied Economics, 17: 205 – 214.
Farrell, M.J. 1957. The measurement of productive efficiency. Journal of Royal Statistical Society
120, 253-290.
Faurès, J.M., Bartley, D., Bazza, M., Burke, J., Hoogeveen, J., Soto, D., Steduto, P. 2013. Climate
smart agriculture sourcebook. FAO, Rome, p 557.
Feder, G., Just, R.E., and Zilberman, D. 1985. Adoption of Agricultural Innovations in Developing
Countries: A Survey. Economic Development and Cultural Change, 33(2), 255–298.
Fei, J., and Ranis, G., 1961. A theory of economic development. Am. Econ. Rev., 51(4), 533–565.
Feryanto, Herawati, Rifin, A. and N Tinaprilla, N. 2022. Does mechanization have an impact on
increasing the productivity and income of narrow-land corn farmers in Indonesia? In IOP
Conference Series: Earth and Environmental Science, 1107(1): 012054. IOP Publishing.
Firew Kelemu. 2015. Agricultural Mechanization in Ethiopian: Experience, Status and Prospects.
Ethiop. J. Agric. Sci., 25: 45-60.
Garrido, M.M., Kelley, A.S., Paris, J., Roza, K., Meier, D.E., Morrison, R.S. and Aldridge, M.D.
2014. Methods for Constructing and Assessing Propensity Scores. Health Services Research,
 161

49(5): 1701-1720.
Gavgani, S., Zamani, D. M. and Par-Shokohi, M.G. 2021. Effect of Fragmentation of Land on
Agricultural Mechanization Development using AHP Technique. J. of Agr. Mach., 11(1): 43-
53. [Link]
Gbenga, O., Opaluwa, H. I., Adedeji, S. O., and Abdulrahman, S. 2020. Effect of Crop Diversity on
Rural Farming Households’ dietary Diversity in Kogi State, Nigeria. Journal of Asian Rural
Studies, 4(2), 218-229.
Gebreslassie Hailua, Kebede Manjureb and Kiros-Meles Aymut. 2015. Crop commercialization and
smallholder farmers livelihood in Tigray region, Ethiopia. Journal of Development and
Agricultural Economics, 7(9), 314-322.
Geoffrey, KS. 2014. Determinants of Market Participation among Small-Scale Pineapple Farmers
in Kericho County, Kenya.
Getachew Eshetu Gidelew, Tesfaye Lemma Tefera and Chanyalew Seyoum Aweke. 2022. From
staple food to market‑oriented crop: commercialization level of smallholder teff (Eragrostis
teff) growers in Jamma District, Ethiopia. CABI Agriculture and Bioscience, (2022) 3:53.
Getahun Kassa, Eskinder Yigezu and Desalegn Alemayehu. 2017. Determinants of smallholder
market participation among banana growers in Bench Maji Zone, southwest Ethiopia. Int.
Journal of Agricultural Policy Research, 5, 169–177.
Gezahegn Ayele, Mekonnen Bekele and Samia Zekeria. 2006. Productivity and efficiency of
agricultural extension package in Ethiopia. The Ethiopian Development Research Institute
(EDRI) Research Report 5, Addis Ababa, Ethiopia.
Glynn, A. N. and Quinn, K. M. 2010. An introduction to the augmented inverse propensity
weighted estimator. Political analysis, 18(1), 36-56.
Glynn, A.N and Quinn, K.M. 2009. An Introduction to the Augmented Inverse Propensity
Weighted Estimator. Political Analysis (18):36–56.
Goetz, S. 1992. A Selectivity Model of Household Food Marketing Behavior in Sub-Saharan
Africa. American Journal of Agricultural Economics, 74(2): 444-452.
Gollin, D., Parente, S., Rogerson, R., 2002. The role of agriculture in development. Am. Econ. Rev.
92(1), 160–164.
Greene, W. 2012. Econometric Analysis. 7th Edition. Prentice Hall, Upper Saddle River.
Griliches, Z., 1957. Hypbrid corn: An exploration in the economics of technological change.
 162

Econometrica 25, 501±523.

Guadagni, M. and Fileccia, T. 2009. The Kyrgyz republic: Farm mechanization and agricultural
productivity.
Gujarati, D.N. 1995. Basic Economics. 3rd Edition. McGraw-Hill, Inc., New York.
Gürsoy, S., Kara, A. and Akın, S. 2021. Factors Affecting the Farmers’ Decision-Making on
Tractor Power Selection in Pistachio Farms: The Case of Siirt Province in Turkey. J Agron
Technol Eng Manag 2021, 4(3): 591-597. [Link] › v4_3 › 2. _Songül_et_al_2021,
_4(3), _591-[Link].
Guta Regasa Megerssa, Rijalu Negash, Adugna Eneyew Bekele and Diriba Bane Nemera. 2020.
Smallholder market participation and its associated factors: Evidence from Ethiopian
vegetable producers. Cogent Food & Agriculture, 6:1, 1783173.
Guush Berhane Tesfay. 2014. Emerging demand for tractor mechanization in Ethiopia. Conference
paper on “Mechanization and Agricultural Transformation in Asia and Africa: Sharing
Development Experiences”. June 18-19. International food policy institute and Ethiopian
development research institute. Beijing, China.
Guush Berhane, Hirvonen, K., and Minten, B. 2016. Agricultural mechanization in Ethiopia:
Evidence from the 2015 feed the future survey. International food policy research institute
(IFPRI)-Ethiopia strategy support program (ESSP) research note 48.
[Link] agricultural-mechanization-Ethiopia-evidence -
2015-feed-future-survey.
Guush Berhane, Mekdim Dereje, Bart Minten, and Seneshaw Tamru. 2017. The rapid – but from a
low base – uptake of agricultural mechanization in Ethiopia: Patterns, implications and
Challenges. Ethiopian Strategy Support Program. (ESSP working paper 105). International
Food Policy Research Institute.
Haddad, L. Hoddinott, J. and Alderma, H. 1997. Intrahousehold Resource Allocation and
overview. The World Bank Policy Research Department Poverty and Human Resource
Division, Working paper 1255.
Haddad, L. 1994. Strengthening food policy through intra-household analysis. Food Policy, 19(4):
347-356.
Hassan, F. and Kornher, L. 2019. Lets get mechanized - labor market implications of structural
transformation in Bangladesh. Available at SSRN: [Link]
 163

Hassan, S. and Ahmad, B. 2005. Stochastic Frontior Production Function, Application and
hypothesis testing. International Journal of Agriculture and Biology, 7(3), 427-430.
Hayami, Y and Ruttan, V.W. 1985. Agricultural development: an international perspective, revised
and expanded. Johns Hopkins University Press, Baltimore.
Hayami, Y. and Ruttan, V.W. 1971. Induced Innovation in Agricultural Development. Discussion
Paper No.3. Center for Economics Research Department of Economics University of
Minnesota, Minneapolis, Minnesota, 55455.
Hayami, Y. and Vernon Ruttan, V. 1970. Factor Prices and Technical Change in Agricultural
Development: The United States and Japan, 1880-1960, Journal of Political Economy, 78, (5),
1115-41.
Hazra, C. R. 2001. Crop Diversification in India, in Papademetriou, M. K. and Dent, F. J. (Ed),
Crop Diversification in the Asia-Pacific Region, Bangkok, Thailand: Food and Agriculture
Organization of the United Nations Regional Office for Asia and the Pacific.
Headey, D., Alauddin, M. and Rao, D. P. 2010. Explaining agricultural productivity growth: an
international perspective. Agricultural Economics, 41(1), 1-14.
Heidari, S., Keshavarz, S., and Mirahmadizadeh, A. 2017. Application of seemingly unrelated
regression (SUR) in determination of risk factors of fatigue and general health among the
employees of petrochemical companies. Journal of Health Sciences & Surveillance
System, 5(4), 180-187.
Henock Kifle. 1972. Investigations on mechanized and its effects on peasant agriculture. CADU
publication No. 74.
Hensher, M. 2001. Financing health systems through efficiency gains. CMH Working Paper Series,
Paper No. WG3: 2.
Herdt, R.W. 1983. Mechanization of rice production in developing Asian countries: Perspective,
evidence and issues”. In: Consequences of Small Farm Mechanization. International Rice
Research Institute, LosBaños, Laguna, Philippines.
Hicks, J.R. 1932. The Theory of Wages. Macmilian and Co. London 1932. 247 S.
[Link]
Hirano, K. and Imbens, G. W. 2004. The propensity score with continuous treatments. Pp. 73-84.
In: Gelman, A. and Meng, X. (eds.), Applied Bayesian Modeling and Causal Inference from
Incomplete-data Perspectives. New York, Wiley.
 164

Hlomendlini, P.H. 2015. Key Factors Influencing Smallholder Market Participation in the Former
Homelands of South Africa: Case Study of the Eastern Cape. [Link]. Thesis, Stellenbosch
University.
Ho, D. E., K. Imai, G. King, and E. A. Stuart. 2007. Matching as Nonparametric Preprocessing for
Reducing Model Dependence in Parametric Causal Inference. Political Analysis, 15: 199–236.
Hollingsworth, B. 2008. The measurement of efficiency and productivity of health care delivery.
Health Economics, 17(10): 1107–1128.
Holloway G. and Ehui S. 2002. Expanding market participation among smallholder livestock
producers: a collection of studies employing Gibbs sampling and data from the Ethiopian
highlands, 1998–2001. Socio-economic and policy research working paper no. 48. Nairobi:
International Livestock Research Institute; P. 1–88.
Holloway, B.B., Wang, S., Parish, J.T., 2005. The role of cumulative online purchasing experience
in service recovery management. Journal of Interactive Marketing, 19 (3), 54–66
Holmberg, J. 1972. Master plan for the evaluation of CADU. CADU publication No. 81.
Hopper, W.D. 1965. Allocation efficiency in traditional Indian agriculture. Journal of farm
economics, 47(3): 611-624.
Hormozi, A.A., Mohammad Amin Asoodar, M.A. and Abdeshahi, A. 2012. Impact of
mechanization on technical efficiency: A case study of rice farmers in Iran. International
Conference on Applied Economics (ICOAE). Procedia Economics and Finance 1(2012) 176-
185.
Hou, F. Agricultural Mechanization Propulsion Mechanism Influencing Factors Analysis and
Policy Implications. China Rural Survey. 2008, 5, 42–48.
Houmy, K., Clarke, L.J., John. E. Ashburner, JE., and Kienzle J. 2013. Agricultural Mechanization
in Sub-Saharan Africa: Guidelines for preparing a strategy. Integrated Crop Management
(22): 2013. Plant Production and Protection Division. Food and Agriculture Organization of
the United Nations. Rome, 2013.
Hua, W. and Hite, D. 2005. Assessing the relationship between crop choice and land use change
using a Markov Model. Selected paper for prepared for presentation at the American
Agricultural Economics Association Annual Meeting, Providence, Rhode Island, July 24-27.
Humpheys, R. B. 2010. Dealing with zeros in economic data. University of Alberta, department of
economics, Edmenton, Canada.
 165

Ibrahim, H., Rahman, S., Envulus, E., & Oyewole, S. 2009. Income and crop diversification among
farming households in a rural area of North Central Nigeria. Agro-Science: Journal of
Tropical Agriculture, Food, Environment and Extension, 8(2), 84–89.
Imad A, Huo X X, Imran K, Hashmat A, Khan B, Sufyan U.K. 2019. Technical efficiency of
hybrid maize growers: A stochastic frontier model approach. Journal of Integrative
Agriculture, 18, 2408–2421.
Imbens, G. W. 2004. Nonparametric Estimation of Average Treatment Effects under Exogeneity: A
Review. Review of Economics and Statistics, 86 (1): 4–29.
Immink, M.D.C and Alarcon, J.A. 1993. Household income, food availability, and commercial crop
production by smallholder farmers in the western highlands of Guatemala. Economic
Development and Cultural Change, 41: 319-342.
Inoni, O.E., Gani, B.S. and Sabo, E. 2021. Drivers of Crop Diversification: Evidence from
Smallholder Farmers in Delta State Nigeria. Acta Universitatis Agriculturae et Silviculturae
Mendelianae Brunensis, 69(1): 59–70.
Israel, G.D. 1992. Sampling the evidence of extension program impact. Program evaluation and
organizational development, IFAS, University of Florida, PEOD-5.
ITA (International Trade Administration). 2022. Ethiopia- Country Commercial Guide. Accessed
on 06 November, 2023, Available on [Link]
guides/ethiopia-agricultural-sector.
Jabbar, M. A., Bhuiyan, M. S. and Bari, A. M. 1983. Causes and consequences of power tiller
utilization in two areas of Bangladesh (No. 453-2016-36323, pp. 71-82).
Jarvis, D. and Hodgkin, T. 2000. Farmer decision making and genetic diversity: linking
multidisciplinary research to implementation on-farm. In S. B. Brush (Ed.), Genes in the Field:
On-Farm Conservation of Crop Diversity (p. 288). IDRC. Retrieved from
[Link]
Jema Haji. 2008. Economic Efficiency and Marketing performance of Vegetable Production in the
Eastern and Central Parts of Ethiopia. Doctoral Thesis, Swedish University of Agricultural
Sciences, Uppsala.
Jemma Haji. 2004. The politics of land tenure in Ethiopian History: experience from the south.
paper for XI World congress of rural sociology, Trondheim, Norway.
Jha, B., Tripathi, A. and Mohanty, B. 2009. Drivers of Agricultural Diversification in India,
 166

Haryana and the Greenbelt Farms of India. Institute of Economic Growth University of Delhi
Enclave North Campus, Working Paper Series No. E/303/2009.
Johnston, B. and Mellor, J., 1961. The role of agriculture in economic development. Am. Econ.
Rev., 51(4), 566–593.
Jondrow, J., Lovell, C. K., Materov, I. S., and Schmidt, P. 1982. On the estimation of technical
inefficiency in the stochastic frontier production function model. Journal of
econometrics, 19(2-3), 233-238.
Jony Girma Meshesha. 2011. Impact of contract farming on household income of smallholder
farmers: The case of organic honey production in South West Ethiopia, Sheka Zone. M. Sc.
Thesis, Wageningen University.
Jorgenson, D., 1961. The development of a dual economy. Econ. J., 282, 209–334.
Judge, G.G, Hill, R.C., Griffiths, W.E., Lutkepohl. H. and Lee, T.C. 1988. Introduction to the teory
and practice of econometrics. 2 editions. Wiley, New York.
Juma, C. 2010. The New Harvest: Agricultural Innovation in Africa, Oxford University Press.
Kaasa, A. 2009. Effects of different dimensions of social capital on innovative activity: Evidence
from Europe at the regional level. Technovation 29 (2009) 218–233;
[Link]
Kankwamba, H., Mapila, M. A. T. J. and Pauw, K. 2012. Determinants and spatiotemporal
dimensions of crop diversification in Malawi. Project Report produced under a co-financed
research agreement between Irish Aid, USAID and IFPRI, Paper, 3.
Kansanga, M., Andersen, P., Kpienbaareh, D., Mason-Renton, S., Atuoye, K., Sano, Y., Antabe, R.,
and Luginaah, I. 2018. Traditional agriculture in transition: examining the impacts of
agricultural modernization on smallholder farming in Ghana under the new Green Revolution,
International Journal of Sustainable Development & World Ecology, DOI:
10.1080/13504509.2018.1491429.
Kassa Tarekegn and Kibreab Yosefe. 2017. Determinants of Poultry Market Participation
Decisions: The Case of Producers in Kaffa and Bench Majji Zones, Southern Ethiopia.
Journal of Economics and Sustainability, 8(3):23-29. 2017.
Kefyalew Endale. 2011. Fertilizer consumption and agricultural productivity in Ethiopia. Ethiopian
Development Research Institute Addis Ababa, Ethiopia. Available at
[Link]
 167

Kello, A.B. 1992. The peasant economy: A review of different theories. Ethiopian journal of
economics, 1(2):53-78.
Key N, Sadoulet E and de Janvry A. 2000. Transaction costs and agricultural household supply
response. American Journal of Agricultural Economics, 82(2):245–259.
Khatun, D., and B.C. Roy. 2012. Rural livelihood diversification in West Bengal: Determinants and
constraints. Agricultural Economics Research Review, 25(No.1): 115–124.
Kifle Degefa, Getachew Biru, and Gelmessa Abebe. 2020. Economic Efficiency of Smallholder
Farmers in Tomato Production in BakoTibe District, Oromia Region, Ethiopia. J Agri Sci
Food Res, 11: 273.
Kirui, Oliver Kiptoo, K.O. 2019. The agricultural mechanization in Africa: micro-level analysis of
state, drivers and effects. Sixth International Conference, September 23-26, 2019, Abuja,
Nigeria 295819, African Association of Agricultural Economists (AAAE).
Kislev, Y. and Peterson, W. 1981. Induced Innovations and Farm Mechanization. American
Journal of Agricultural Economics, (63): 3. 562-565.
Kline, C. K., Green, D. A. G., Donahue, R. L. and Stout, B. A. 1969. Agricultural mechanization in
Equatorial Africa. Agricultural mechanization in Equatorial Africa., (6).
Kluve, J., Schneider, H., Uhlendorff, A. and Zhao, Z. 2012. Evaluating Continuous Training
Programmes by Using the Generalized Propensity Score. Journal of the Royal Statistical
Society, 175(2): 587-617.
Knickel, K., Ashkenazy, A., Chebach, T.C. and Parrot, N. 2017. Agricultural modernization and
sustainable agriculture: Contradictions and complementarities. International Journal of
Agricultural Sustainability, 15(4).
Kokkinou, A. 2010. A note on theory of productive efficiency and stochastic frontier models.
European Research Studies, 13(4): 109-118.
Koopmans, T.C. 1951. An analysis of production as an efficient combination of activities. In:
Koopmans, T.C. (ed.), Activity Analysis of Production and Allocation. Cowles Commission
for Research in Economics Monograph No. 13. New York: John Wiley and Sons.
Kopp, R. J., & Diewert, W. E. 1982. The decomposition of frontier cost function deviations into
measures of technical and allocative efficiency. Journal of Econometrics, 19(2-3), 319-331.
Kopp, R. J., and Smith, V. K. 1980. Frontier production function estimates for steam electric
generation: A comparative analysis. Southern Economic Journal, 1049-1059.
 168

Kothari, C.R. 2004. Research Methodology: Methods and Techniques, 2nd Edition. New Age
International, New Delhi, India.
Krishna, K. L., and Sahota, G. S. 1991. Technical efficiency in Bangladesh manufacturing
industries. The Bangladesh development studies, 89-105.
Kumar, C.R. 2020. Crop diversification and its determinants: a comparative study between Cuttack
and Kandhamal districts of Odisha, India. Journal of Global Resources, 6(02): 43-48.
Kumbhakar, S. C. and Horncastle, A. P. 2015. A practitioner's guide to stochastic frontier analysis
using Stata. Cambridge University Press.
Kumbhakar, S.C. and Lovell, C.A.K. 2000. Stochastic Frontier Analysis. Cambridge University
Press, Cambridge. [Link]
Kurdys-Kujawska, A., Strzelecka, A. and Zawadzka, D. 2021. The Impact of Crop Diversification
on the Economic Efficiency of Small Farms in Poland. Agriculture 2021, 11, 250. Agricultural
Diversification, 177.
Kurz, C. F. 2022. Augmented Inverse Probability Weighting and the Double Robustness Property
(vol 42, pg 156, 2022). Medical Decision Making, 42(2), NP1-NP1.
Kusse Haile, Engida Gebre and Agegnehu Workye. 2022. Determinants of market participation
among smallholder farmers in Southwest Ethiopia: double‑hurdle model approach.
Kyaw, N.N., Ahn, S. and Lee, S.H. 2018. Analysis of the Factors Influencing Market Participation
among Smallholder Rice Farmers in Magway Region, Central Dry Zone of Myanmar.
Sustainability, 2018 (10): 4441.
Lai, W., Roe, B. and Liu, Y. 2015. Estimating the Effect of Land Fragmentation on Machinery Use
and Crop Production. Selected Paper prepared for presentation at the 2015 Agricultural &
Applied Economics Association and Western Agricultural Economics Association Annual
Meeting, San Francisco, CA, July 26-28. DOI: 10.22004/[Link].205280.
Lawin, K.G. and Tamini, L.D. 2019. Determinants of Crop Diversification in Burkina Faso What
is the Impact of Risk Preference? Available at [Link]
[Link], accessed on 18/07/2023.
Leavy, J. and Poulton C. 2007. Commercialization in agriculture: A typology. Paper presented at
the Fifth International Conference on the Ethiopian Economy, Ethiopian Economic
Association (EEA), USA. pp. 79-99.
Lei, D., Ruimei, W., Weisong, W. and Jingjie, Z. 2016. Farm Size, Agricultural Mechanization and
 169

Technical Efficiency: An Empirical Study on Grape Producers in China. International

Conference on Education, Sports, Arts and Management Engineering (ICESAME 2016),
Atlantis Press.
Lewis,W. A., 1954. Economic development with unlimited supplies of labour. The Manchester
School, 22(1), 139–191.
Leykun Demeke and Jemma Haji. 2014. Econometric analysis of factors affecting market
participation of smallholder farming in central Ethiopia. Available: Online at
[Link] and accessed on 15 September 202.
Li, C.; Chen, X.; Jiang, A.; Lee, M.-B.; Mammides, C.; Goodale, E. 2021. Socioeconomic
Determinants of Crop Diversity and Its Effect on Farmer Income in Guangxi, Southern China.
Agriculture, 11, 336. [Link]
Li, J. and Fraser, M.W. 2015. Evaluating Dosage Effects in a Social-Emotional Skills Training
Program for Children: An Application of Generalized Propensity Scores. Journal of Social
Service Research, 41(3): 345-364.
Li, W., Wei, X., Zhu, R., and Guo, K. 2018. Study on factors affecting the agricultural
mechanization level in China based on structural equation modeling. Sustainability, 11(1), 51.
Lim, P.C. 1985. Effects of Agricultural Mechanization on Farm Income Patterns. Journal of
Philippine Development, 12(1):198-210.
Linden, A., and Yarnold, P. R. 2016. Combining machine learning and matching techniques to
improve causal inference in program evaluation. Journal of evaluation in clinical
practice, 22(6), 868-874.
Lipton, M. 1968. The theory of the optimizing peasant. Journal of Development Studies 4(3): 327–
351.
Lipton, M. 1977. Why do poor people stay poor? Urban bias in world development. Temple Smith,
London.
Liu, Y,.and Tian, Z. 2009. Analysis on Household income levels affects the demand for farm
machinery and equipment. Chin. Rural Econ., 12, 44-55.
Lone, R. A. 2013 Agricultural Diversification towards High Value Commodities In South Asia,
International Journal of Trade & Global Business Perspectives, Vol. 2, No. 4, pp. 688-694
Longpichai, O. 2013. Determinants of Adoption of Crop Diversification by Smallholder Rubber
Producers in Southern Thailand: Implications on Natural Resource Conservation. Kasetsart J.
 170

(Soc. Sci), 34: 370 – 382.

Loon, V. J., Woltering, L., Krupnik, T. J., Baudron, F., Boa, M., & Govaerts, B. (2020). Scaling
agricultural mechanization services in smallholder farming systems: Case studies from sub-
Saharan Africa, South Asia, and Latin America. Agricultural systems, 180, 102792.
Lovell, C.A.K. 1993. Production frontiers and productive efficiency. In: Fried, H.O., Lovell C.A.K.
and Schmidt S.S. (Eds.), The Measurement of Productive Efficiency: Techniques and
Applications. Oxford: Oxford University Press, New York, 3-67.
Ma, H., Pang, W., Sun, L., and Xu, W. 2022. Augmented weighting estimators for the additive
rates model under multivariate recurrent event data with missing event type. Statistics in
Medicine, 41(22), 4285-4298.
Ma, W., Renwick, A. and Grafton, Q. 2018. Farm machinery use, off-farm employment and farm
performance in China. Australian Journal of Agricultural and Resource Economics, 62(2),
279–298. doi:10.1111/1467-8489.12249.
Maddala, G.S. 1983. Limited Dependent and Qualitative Variables in Econometrics. Cambridge
University Press, Cambridge.
Maggio, G., Sitko, N. and Ignaciuk, A. 2018. Cropping system diversification in Eastern and
Southern Africa: Identifying policy options to enhance productivity and build resilience. FAO
Agricultural Development Economics Working Paper 18-05. Rome, FAO. Licence: CC BY-
NC-SA 3.0 IGO.
Maheshwari, T.K. and Tripathi, A. 2019. Quantification of Agricultural Mechanization Status for
Etawah District of Uttar Pradesh, India. International Journal of Current Microbiology and
Applied Sciences, 8(5): 659-666.
Makate, C., Wang, R., Makate, M., and Mango N. 2016. Crop diversification and livelihoods of
smallholder farmers in Zimbabwe: adaptive management for environmental change.
SpringerPlus., 5(1):1–18.
Malabo Montpellier Panel, 2018. Mechanized: Transforming Africa's Agriculture Value Chains.
International Food Policy Research Institute (IFPRI) and Malabo Montpellier Panel, Dakar,
Senegal.
Malthus, Th. 1798. An Essay on the Principle of Population. Electronic Scholarly Publishing
Project, London.
Mango, N., Makate, C., Mapemba, L. and Sopo, M. 2018. The role of crop diversification in
 171

improving household food security in central Malawi. Agriculture and Food Security 7, 7.
[Link]
Mather, D. and Belton, B. 2018. Mechanization and Crop Profitability, Productivity and Labor Use
in Myanmar’s Dry Zone. Research Paper 103. July 2018, East Lansing, Michigan State
University.
Matilla-Wiro, P. 1999. Economic Theories of the Household. A Critical Review.
Mazzocchi, C., Orsi, L., Ferrazzi, G., & Corsi, S. (2020). The dimensions of agricultural
diversification: a spatial analysis of Italian municipalities. Rural Sociology, 85(2), 316-345.
Mburu, S., Ackello-Ogutu, C. and Mulwa, R. 2014. Analysis of Economic Efficiency and Farm
Size: A Case Study of Wheat Farmers in Nakuru District, Kenya. Economics Research
International, Hindawi Publishing Corporation.
McDonald and Moffitt (1980): The uses of Tobit analysis. The Review of Economics and Statistics
62, 318-321.
McElroy, M. B. 1977. Goodness of fit for seemingly unrelated regressions: Glahn's R2y. x and
Hooper's r̄ 2. Journal of econometrics, 6(3), 381-387.
Meeusen, W., and J. van den Broeck. 1977. Technical efficiency and dimension of the firm: Some
results on the use of frontier production functions. Empirical Economics 2: 109–122.
Mekonnen Sime Kidane. 2019. The nexus of crop and income diversification, commercialisation
and household welfare: empirical evidence from Ethiopia. Doctoral dissertation, University of
KwaZulu-Natal, Pietermaritzburg, South Africa.
Mellor, J., 2001. Faster more equitable growth-agriculture, employment multipliers, and poverty
reduction. Paper prepared for USADD/G/EGAD.
Mendola, M. 2007. Farm household production theories: A review of institutional and behavioral
responses. Asian Development Review, 24(1): 49-68.
Mesay Yami, Tesafye Solomon, Bedada Begna, Fekadu Fufa, Tolesa Alemu and Dawit Alemu.
2013. Source of technical inefficiency of smallholder wheat farmers in selected waterlogged
areas of Ethiopia: A translog production function approach. African Journal of Agricultural
Research, 8(29): 3930-3940.
Meynard, J. M., Charrier, F., Fares, M., Le Bail, M., Magrini, M. B., Charlier, A., and Messéan, A.
2018. Socio-technical lock-in hinders crop diversification in France. Agron Sustain Dev 38:
54.
 172

Milkessa Asfaw, Endrias Geta and Fikadu Mitiku. 2019. Economic Efficiency of Smallholder
Farmers in Wheat Production: The Case of Abuna Gindeberet District, Oromia National
Regional State, Ethiopia. International Journal of Environmental Sciences & Natural
Resources, 6(2): 41-51.
Min, S., Paudel, K.P. and Feng-bo, C. 2021. Mechanization and efficiency in rice production in
China. Journal of Integrative Agriculture 2021, 20(7): 1996–2008.
Mohammed Hassena, Regassa Ensermu, Mwangi, W., and Verkuijl, H. 2000. A Comparative
Assessment of Combine Harvesting Vis-à-vis Conventional Harvesting and Threshing in Arsi
Region, Ethiopia. International Maize and Wheat Improvement Center (CIMMYT) and
Ethiopia Agricultural Research Organization (EARO), Mexico, D. F.
Moniruzzaman, Rahman, S., Sujan, H.K. 2021. Determinants of Small-Scale Mechanization for
Potato Farming: A Case from Bangladesh. Journal of Nepal Agricultural Research Council,
(7): 75-82.
Morduch, J. 1995. Income smoothing and consumption smoothing. Journal of Economic
Perspectives, 9(3):103–14.
Moti Jaleta, Berhanu Gebremedhin and Hoekstra, D. 2009. Smallholder Commercialization:
Processes, Determinants and Impact. Discussion Paper No. 18. Improving Productivity and
Market Success (IPMS) of Ethiopian Farmers Project. ILRI: Nairobi, Kenya.
Mottaleb, K. A., Krupnik, T. J. and Erenstein, O. 2016. Factors associated with small-scale
agricultural machinery adoption in Bangladesh: Census findings. Jour. of Rural Studies 46
(2016) 155e168. [Link] Elsevier.
Mustefa Bati and Jema Haji. 2020. Economic Efficiency in Barely Production: The Case of Chole
District, East Arsi Zone, Oromia National Regional State, Ethiopia. Journal of Resources
Development and Management, 66:25-36.
Mwaura, F.M. and Adong, A. 2016. Determinants of Households’ Land Allocation for Crop
Production in Uganda. Journal of Sustainable Development, (9)5.
Mzyece, A. 2016. Effect Of Buyer Type on Market Participation of Smallholder Farmers in
Northern Ghana. [Link]. Thesis Submitted to Kansas State University Manhattan, Kansas.
Available at [Link]
Nagpure, S.C., Deshmukh, R.G., Sharma, P.K. and Ingole, D.N. 2017. Pattern of crop
concentration and crop diversification– An economic analysis. Maharashtra Jn. of Agril.
 173

Economics, 20(2): 128-132.

Nakajima, Ch. 1986. Subjective Equilibrium Theory of the Farm Household. Elsevier Publishers
H.V., Amsterdam, The Netherlands:
Nasrin, M., Bauer, S. and Arman. 2018. Assessing the impact of fertilizer subsidy on farming
efficiency: A case of Bangladeshi farmers. Open Agriculture, 2018 (3): 567–577.
NBE (National Bank of Ethiopia). 2020. National Bank of Ethiopia, 2019/20 Annual Report.
[Link]
Neill, S. P. and Lee, D. R. 2001. Explaining the adoption and disadoption of sustainable
agriculture: the case of cover crops in northern Honduras. Economic development and cultural
change, 49(4), 793-820.
Ngunjiri, W.S., Lagat, J.K, and Bett, H.K. 2016. Drivers of Crop Distribution and Duration
Diversification among Small holder Farmers in Kenya. Sch J Econ Bus Manag, 3(9):471-480.
Njeru, E.M. 2013. Crop diversification: a potential strategy to mitigate food insecurity by
smallholders in sub-Saharan Africa. J Agric Food Syst Community Dev., 3(4):63–9.
Nkonya, E., Pender, J., Kaizzi, C., Edward, K. and Mugarura, S. 2005. Policy Options for
Increasing Crop Productivity and Reducing Soil Nutrient Depletion and Poverty in Uganda.
International Food Policy Research Institute, Washington DC.
Noad Mekonnen and Bamlaku Alamirew. 2017. The Effect of Rural Road Transport on Crop
Commercialization: Evidence from the Ethiopian Living Standard Measurement
Survey. Ethiopian Journal of the Social Sciences and Humanities, 13(2), 81-108.
Norman, D.W. 1974. Rationalising mixed cropping under indigenous conditions: The example of
Northern Nigeria. The Journal of Development Studies, 11(1): 3-21.
Nowacki, T. 1974. Example of technical economic analysis of mechanized process in various agro-
technical conditions: Economic commission for Europe AGRI/MECH/32.
Nowacki, T. 1978. Methodology used by ECE countries in forecasting mechanisation
developments. United Nations Economic Commission for Europe, AGRI/MECH Report No.
74.
Nuri Lefebo, Jema Haji, Endrias Geta and Lemma Zemedu. 2016. Determinants of Status and
Extent of Market Participation among Bulla Producers in Hadiya Zone, Southern Ethiopia.
Journal of Biology, Agriculture and Healthcare, 6(7):84-92.
OBOFED (Oromia Bureau of Finance and Economic Development) 2011. Physical and Socio-
 174

Economic Profile of Arsi Zone and Districts. The National Regional Government of Oromia,
Bureau of Finance and d Economic Development, Regional Data and Information Core
Process.
OECD (The Organization for Economic Co-operation and Development). 2012. Agricultural policies
for poverty reduction. OECD Publishing, Paris. [Link] 902-en.
Ojo M.A., Ojo A.O., A.I. Odine and Ogaji A. 2014. Determinants of Crop Diversification among
Small – Scale Food Crop Farmers in North Central, Nigeria. PAT, 10 (2): 1-11.
Okello, D.M., Bonabana-Wabbi, J. and Mugonola, B. 2019. Farm level allocative efficiency of rice
production in Gulu and Amuru districts, Northern Uganda. Agricultural and Food Economics,
7:9.
Okoye, B. C., Onyenweaku, C. E. and Asumugha, G.N. 2007. Economic Efficiency of Small-
Holder Cocoyam Farmers in Anambra State, Nigeria: A Translog Stochastic Frontier Cost
Function Approach. National Root Crops Research Institute, Umudike, Abia State, Micheal
okpara University of Agriculture, Umudike, Umuahia, Abia State.
Olaoye, J. O. and Rotimi, A. O. 2010. Measurement of agricultural mechanization index and
analysis of agricultural productivity of farm settlements in Southwest Nigeria. Agricultural
Engineering International: CIGR Journal, 12(1).
Olsson, O. and Hibbs, D. A., 2005. Biogeography and long-run economic development. Eur. Econ.
Rev., 49(4), 909–938.
Oluwatayo, I.B. and Rachoene, M.A. 2017. Effect of Agricultural Commercialization on Food
Security among Smallholder Farmers in Polokwane Municipality, Capricorn District of
Limpopo Province, South Africa. Journal of Agribusiness and Rural Development, 1(43):
143–156.
Omiti, J., Otieno, D., Nyanamba, T. and Mc Cullough, E. 2009. Factors influencing the intensity of
market participation by smallholder farmers: A case study of rural and peri-urban areas of
Kenya. Afjare, 3(1).
Owombo, P.T., Akinola, A.A., Ayodele, O.O. and Koledoye, G.F. 2012. Economic impact of
agricultural mechanization adoption: evidence from maize farmers in Ondo State, Nigeria, J.
Agric. Biodivers. Res., 1(2012) 25–32. Available Online at
[Link]
Owusu, O., Iscan, T.B. 2020. Drivers of farm commercialization in Nigeria and Tanzania. Agric.
 175

Econ. 52, 265–299. [Link]

Özpınar, S. 2020. Mechanization and agricultural farm structure in the agricultural area of the
Dardanelles region. International Journal of Agriculture, Environment and Food Sciences, 4
(1): 39-56.
Pal, S. and Kal, S. 2012. Implications of the methods of agricultural diversification in reference
with malda district: drawback and rationale. International Journal of Food, Agriculture and
Veterinary Sciences, 2 (2): 97-105.
Patorno, E., Grotta, A., Bellocco, R. and Schneeweiss, S. 2013. Propensity Score Methodology for
Confounding Control in Health Care Utilization Databases. Epidemiology, Biostatistics and
Public Health, 10(3).
Paudel, G.P, DB KC, Rahut DB, Justice SE, McDonald AJ. 2019. Scale-appropriate mechanization
impacts on productivity among smallholders: evidence from rice systems in the mid-hills of
Nepal. Land Use Policy, 85 (2019) 104–113. [Link]
Pauw, S. 2017. Agricultural Commercialization in Ethiopia: A Review of Warehouse Receipts in
the Maize, Wheat, Sorghum and Tef Value Chains. USAID/Ethiopia Agriculture Knowledge,
Learning, Documentation and Policy Project, Ababa Addis.
Pender, J and Dawit Alemu. 2007. Determinants of smallholder commercialization of food crops:
Theory and evidence from Ethiopia. Discussion Paper No. 75. IFPRI (International Food
Policy Research Institute), Washington, DC, USA.
Pender, J., Ehui, S. and Place, F. 2006. Conceptual framework and hypothesis. In: Pender J, Place F
and Ehui S (eds), Strategies for sustainable land management in the East African highlands,
IFPRI (International Food Policy Research Institute), Washington, DC, USA.
Peng, J., Zhao, Z. and Liu D. 2022. Impact of Agricultural Mechanization on Agricultural
Production, Income, and Mechanism: Evidence from Hubei Province, China. Front. Environ.
Sci. 10:838686. doi: 10.3389/fenvs.2022.838686.
Pingali, P., Bigot, Y. and Hans P. Binswanger, H.P. 1987. Agricultural Mechanization and the
Evolution of Farming Systems in Sub-Saharan Africa. The Johns Hopkins University Press
Baltimore, Maryland 21211, U.S.A.
Pingali, P.L, and Rosegrant, M.W. 1995. Agricultural commercialization and diversification:
Process and polices. Food Policy, 20(3):171-185.
Pingali, P.L. 1997. From subsistence to commercial production system: The transformation of
 176

Asian agriculture. American Journal of Agricultural Economics, 79 (2): 628-634.

Pingali, P.L. 2007. Agricultural mechanization: adoption patterns and economic impact. Handbook
of Agricultural Economics, 3: 2779-2805.
Pope, R., and Prescott, R. 1980. Diversification in Relation to Farm Size and Other Socioeconomic
Characteristics. American Journal of Agricultural Economics,62(3), 554-559.
Portes, A. 1998. Social capital: Its origins and applications in modern sociology. Annual review of
sociology, 24(1), 1-24.
Poulton, C. 2017. What is agricultural commercialization, why is it important and how do we
Measure it? Agricultural policy research in Africa, Working paper 06.
Pretty, J., Toulmin, C., and Williams, S. 2011. Sustainable intensification in African agriculture.
International Journal of Agricultural Sustainability, 9(1), 5–24.
Quan, X., and Doluschitz, R. 2021. Factors influencing the adoption of agricultural machinery by
Chinese maize farmers. Agriculture, 11(11), 1090.
Raad, H., Cornelius, V., Chan, S., Williamson, E. and Cro, S. 2020. An evaluation of inverse
probability weighting using the propensity score for baseline covariate adjustment in smaller
population randomised controlled trials with a continuous outcome. BMC Medical Research
Methodology, 20, 1-12.
Rahman, S. 2008. Determinants of crop choices by Bangladeshi farmers: A bivariate probit
analysis. Asian Journal of Agriculture and Development, 5(1), 29–41.
Rahman, S. and Kazal, M.M.H. 2014. Determinants of crop diversity in the regions of Bangladesh
(1990–2008). Singapore Journal of Tropical Geography 36 (2015) 83–97;
[Link]
Rajkhowa, A., Barman, I. Das, P. K., Deka, S. D. and Sonowal, A. 2020. An Analysis of Extent of
Farm Mechanization in North Bank Plains Agro-Climatic Zone of Assam. Asian Journal of
Agricultural Extension, Economics & Sociology, 38(11): 81-90.
Ramya, P. and Muruganandham, V. 2016. Impact of Agricultural Mechanization on Production,
Productivity and Employment of Labour. Shanlax International Journal of Commerce, 4(3)
2016-222.
Rapsomanikis, G. 2015. The economic lives of smallholder farmers: An analysis based on
household data from nine countries. Food and Agriculture Organization of the United Nations
Rome.
 177

Rasouli, F., Sadighi, H. and Minaei, S. 2009. Factors Affecting Agricultural Mechanization: A Case
Study on Sunflower Seed Farms in Iran. J. Agric. Sci. Technol., 11: 39-48.
Ravallion, M. and Datt, G., 1999. When is growth pro-poor? Evidence from the diverse experience
of India’s states. World Bank policy research working paper series, 2263, World Bank,
Washington, DC.
Ravisankar, N., Ansari, M. A., Shamim, M., Prusty, A. K., Singh, R., Panwar, A. S. and Bal, S. K.
2022. Sustainable livelihood security of small farmers improved through a resilient farming
system in the semiarid region of India. Land Degradation & Development, 33(15), 2830-2842.
Reardon, T., Berdegué, J., Barrett, C. B. and Stamoulis, K. 2007. Household income diversification
into rural nonfarm activities. Transforming the rural nonfarm economy: opportunities and
threats in the developing world, 115-140.
Rehima Mussema, Belay Kassa, Dawit Alemu and Rashid, S. 2013. Factors affecting farmers’
crops diversification: Evidence from SNNPR, Ethiopia. International Journal of Agricultural
Sciences, 3(6), 558-565.
Rehima, Musema. and Dawit, Alemu. 2012. Red pepper marketing in Siltie and Alaba in SNNPRS
of Ethiopia: factors affecting households' marketed pepper. International Research Journal of
Agricultural Science and Soil Science, 2(6): 261-266.
Reid, J.F. 2011. The Impact of Mechanization on Agriculture. In: The Bridge. National Academy of
Engineering, 41(3).
Reimer, B. 1984. Farm Mechanization: The Impact on Labour at the Level of the Farm Household.
The Canadian Journal of Sociology, 9(4): 429-443.
Rios, A.R., Shively, G. E. and Masters, W. A. 2009. Farm Productivity and Household Market
Participation: Evidence from LSMS Data. Contributed Paper prepared for presentation at the
International Association of Agricultural Economists Conference, Beijing, China, August 16-
22, 2009.
Robins, J. M. 1999. Robust Estimation in Sequentially Ignorable Missing Data and Causal
Inference Models. Proceedings of the American Statistical Association Section on Bayesian
Statistical Science pp. 6–10.
Robins, J. M., Rotnitzky, A. and Zhao, L.P. 1994. Estimation of Regression Coefficients When
Some Regressors are Not Always Observed. Journal of the American Statistical Association
89:846–866.
 178

Robins, J. M., Rotnitzky, A., and van der Laan, M. 2000. Comment. Journal of the American
Statistical Association, 95:477–482.
Roumasset, J. 1976. Rice and risk: Decision making among low-income farmers. Amsterdam,
North-Holland Publishing.
Roy, P., Hansra, B.S., Burman, R.R., Bhattacharyya, S., Roy, T. and Rouf Ahmed, R. 2022. Can
farm mechanization enhance small farmers’ income? Lessons from Lower Shivalik Hills of
the Indian Himalayan Region. Current Science, (123) 5:667-676.
Rubin, D. 1997. Estimating Causal Effects from Large Data Sets Using Propensity Scores. Annals
of Internal Medicine, 127(8): 757-763.
Rufino, C. 2016. Microeconometric Analysis of the Eating-out Behavior of Modern Filipino
Households: The Tobit, the Craggit and the Heckit Models. DLSU Bus. Econ. Rev. 26(1):50-
69.
Ruthenberg, H. 1980. Farming Systems in the Tropics. Oxford, Clarendon Press.
Ruttan, V. 2002. Productivity growth in world agriculture: Sources and constraints. Journal of
Economic Perspectives.
Sabatini, F. 2005. Social Capital as Social Networks. A New Framework for Measurement (June
2005). Available at [Link]
Sahal, B., and R. Baha. 2010. Livelihood diversification pursued by farmers in West Bengal. Indian
Research Journal of Extension Education, 10 (2): 2.
Salam, A. 1981. Farm Tractorization, Fertilizer Use and Productivity of Mexican Wheat in
Pakistan. The Pakistan Development Review, 20(3): 323-345
Saleth, R. M. 1995. Agricultural Diversification in Tamil Nadu: Potentials and Prospects. Institute
of economic Growth, Delhi, 710.
Samuel Gebreselassie and Ludi, E. 2007. Commercialization of smallholder agriculture in selected
Tef-growing areas of Ethiopia. Ethiopian Journal of Economics, 16(1): 57-88.
Samuel Gebreselassie and Sharp, K. 2007. Agricultural commercialization in coffee growing areas
of Ethiopia. Ethiopian Journal of Economics, 16(1): 89-118.
Samuelson, P. A. 1956. Social Indifference Curves. Quarterly Journal of Economics, 70(1):1-22.
Sang, X., Luo, X., Razzaq, A., Huang, Y. and Sahar Erfanian, S. 2023. Can agricultural
mechanization services narrow the income gap in rural China? Heliyon, 9 (2023) e13367.
Saraceno, G., Alqallaf, F., and Agostinelli, C. 2021. A Robust Seemingly Unrelated Regressions
 179

for Row-Wise and Cell-Wise Contamination. arXiv preprint arXiv:2107.00975.

Saraswati, P. A., H. Basavaraja, L.B. Kunnal, S.B. Mahajanashetti and A.R.S Bhat. 2011. Crop
Diversification in Karnataka: An Economic Analysis. Agricultural Economics Research
Review, (24) 351-357.
Sati, V. P. 2012. Agricultural diversification in the Garhwal Himalaya: A Spatio Temporal
Analysis, Sustainable Agriculture Research, 1 (1): 77-86.
Scharfstein, D. O., Rotnitzky, A. and Robins, J. 1999. Rejoinder to Adjusting for Nonignorable
Drop-out Using Semiparametric Nonresponse Models. Journal of the American Statistical
Association 94:1135–1146.
Schmitz, A. and Moss, C.B. 2015. Mechanized Agriculture: Machine Adoption, Farm Size, and
Labor Displacement. AgBioForum, 18(3): 278-296.
Schultz, T. W. 1964. Transforming Traditional Agriculture, New Haven: Yale University Press.
Seeger, J., Kurth, T. and Walker, A. 2007. Use of Propensity Score Technique to Account for
Exposure-Related Covariates: An Example and Lesson. Medical Care, 45(10): S143-S148.
Seife Ayele. 2022. The resurgence of agricultural mechan- isation in Ethiopia: Rhetoric or real
commitment? The Journal of Peasant Studies, 49(1), 137–157.
Self, S. and Grabowski, R. 2007. Economic development and the role of agricultural technology.
36(3), 395–404.
Sen, A. K. 1966. Labour allocation in a cooperative enterprise. The Review of Economic
Studies, 33(4), 361-371.
Seng, K. 2014. Determinants of Farmers’ Agricultural Diversification: The Case of Cambodia.
Asian Journal of Agriculture and Rural Development, 4(8)2014: 414-428.
Seo, S. N., Mendelsohn, R., Dinar, A., Hassan, R., and Kurukulasuriya, P. 2009. A Ricardian
analysis of the distribution of climate change impacts on agriculture across agro-ecological
zones in Africa. Environmental and Resource Economics, 43, 313-332.
Shields, M. 1985. The Impact of Mechanization on Agricultural Production in Selected Villages of
Nueva Ecija. Journal of Philippine Development, 21(12): 182-197.
Shrestha, S. 2011. Status of agricultural mechanization in Nepal. In United Nations Asian and
Pacific Center for Agricultural Engineering and Machinery (UNAPCAEM); Nepal
Agricultural Research Council: Lalitpur, Nepal, Volume 4. (accessed on 20 July 2021).
Sichoongwe, K., Mapemba, L., Ng’ong’ola, D. and Tembo, G. 2014. The determinants and extent
 180

of crop diversification among smallholder farmers: A case study of Southern Province,

Zambia. Malawi Strategic Support Program; Working Paper (05). International Food Policy
Research Institute (IFPRI).
[Link]
Sigei, G., Bett, H., and Kibet, L. 2014. Determinants of market participation among small-scale
pineapple farmers in Kericho County, Kenya. available on [Link]
[Link]/56149/1/MPRA_paper_56149.pdf.
Sigei, G.K. 2014. Determinants of Market Participation among Small-Scale Pineapple Farmers in
Kericho County, Kenya. [Link]. Thesis, Egerton University, Njoro (Kenya). MSC thesis,
Egerton University, Njoro, Kenya
Sime Shiferaw, Jema Haji, Mengistu Ketema and Million Sileshi. 2022. Technical, allocative, and
economic efficiency of malt barley producers in Arsi zone, Ethiopia. Cogent food and
agriculture, 8: 2115669. [Link]
Sims, B., Hilmi, M. and Kienzle, J. 2016. Agricultural mechanization: A key input for Sub-
Saharan African smallholders. Food and Agriculture Organization of the United Nations,
Rome.
Singh, G. 2006. Estimation of a Mechanisation Index and Its Impact on Production and Economic
Factors-a Case Study in India. Biosystems Engineering, 93 (1), 99–106.
Singh, I., Squire, L., and Strauss, J. 1986. Agricultural Household Models: Extensions,
Applications, and Policy. Baltimore, Johns Hopkins University Press.
Singh, R. and Singh, B.B.1972. Farm Mechanization in Western Uttar Pradesh - Problems of Farm
Mechanization Seminar Series-IX, Indian Society of Agricultural Economics, Bombay, Feb.
Singh, R.V. and Singh, L.R. 1980. Mechanization and Employment Degeneration on Farms in
Eastern Uttar Pradesh. Indian Journal of Industrial Relations, 16 (1): 115-124.
Singh, S.R. 1986. Technical parameters in agricultural production function. Ashish Publishing
House, New Delhi.
Sokoni, C.H. 2008. Commercialisation of smallholder production in Tanzania: implications for
sustainable resources management. Geographical Journal, 174(2): 149–175.
Soliman, I. 1992. Agricultural Mechanization and Economic Efficiency of Agricultural Production
in Egypt. International Conference on Agricultural Engineering & Rural Development Beijing
China, Beijing University of Agricultural Engineering, October, 1992, Volume I.
 181

Solomon Asfaw, Bekele Shiferaw and Franklin Simtow. 2010. Does technology adoption promote
commercialization? Evidence from chickpea technologies in Ethiopia. In CSAE 2010
conference on Economic Development in Africa, University of Oxford, UK.
Solomon Bizuayehu. 2014. Technical efficiency of major crops in Ethiopia: Stochastic frontier
model. Academia Journal of Agricultural Research, 2(6): 147-153.
Solomon Gebregziabher; Abdul Mounem Mouazen; Hendrik Van Brussel; Herman Ramon; Jan
Nyssen; Hubert Verplancke; Mintesinot Behailu; Jozef Deckers; Josse De Baerdemaeker.
2006. Animal drawn tillage, the Ethiopian ard plough, maresha: A review., 89(2), 129–
[Link].1016/[Link].2005.08.010
Srisompun, O., Athipanyakul, T. and Isvilanonda, S. 2019. The adoption of mechanization, labour
productivity and household income: Evidence from rice production in Thailand (No. WP-016).
TVSEP Working Paper.
Stahl, M. 1973. Contradictions in agricultural development: A study of three minimum package
projects in southern Ethiopia. Research report No. 14, The Scandinavian Institute of African
Studies, Uppsala.
Starkey, P. 2000. The history of working animals in Africa. pp 478-502. In: Blench R.M. and
MacDonald K. (eds), The origins and development of African livestock: archaeology, genetics,
linguistics and ethnography. University College London Press, London, UK.
Storck, H., Bezabeh Emana, Berhanu Adenew, Borowiecki, A. and Shimeles W/Hawariat. 1991.
Farming systems and farm management practices of the 149 smallholders in the Hararghe
highlands: A baseline survey. Farming Systems and Resource Economics in the Tropics,
11:1991-195.
Strasberg, P.J., Jayne, T.S. Yamano, T. Nyoro, J. Karanja D. and Strauss, J. 1999. Effects of
agricultural commercialization on food cop input use and productivity in Kenya. MSU
International Department of Agricultural Economics Development Department of Economics.
Working paper number 71. Michigan, USA: MSU.
Su, M., Heerink, N., Oosterveer, P. and Feng, S. 2022. Upscaling farming operations, agricultural
mechanization and chemical pesticide usage: A macro-analysis of Jiangsu Province, China.
Journal of Cleaner Production, 380(2): 135120.
Sultan Usman. 2016. Analysis of wheat value chain: The case of Sinana District, Bale Zone,
Oromia region, Ethiopia. MSc thesis presented to Haramaya University. Harar, Ethiopia.
 182

Tachibana, T., Nguyen, T. M., & Otsuka, K. 2001. Agricultural intensification versus
extensification: A case study of deforestation in the northern-hill region of Vietnam. Journal of
Environmental Economics and Management, 44-69.
Tadele Mamo, Wudineh Getahun, Agajie Tesfaye, Ali Chebil, Tesfaye Solomon, Aden Aw-Hassan,
Tolessa Debele and Solomon Assefa. 2016. Analysis of wheat commercialization in Ethiopia:
The case of SARD-SC wheat project innovation platform sites. African Journal of
Agricultural Research, 12(10), 841-849.
Tadie Mirie Abate, Abebe Birara Dessie and Taye Melese Mekie. 2019. Technical efficiency of
smallholder farmers in red pepper production in North Gondar zone Amhara regional state,
Ethiopia. Journal of Economic Structure, (2019) 8:18.
Takeshima, H. 2017. Custom‐hired tractor services and returns to scale in smallholder agriculture: a
production function approach. Agricultural Economics, 48(3), 363-372.
Takeshima, H., Hatzenbuehler, P. L., and Edeh, H. O. 2020. Effects of agricultural mechanization
on economies of scope in crop production in Nigeria. Agricultural Systems, 177, 102691.
Tamrat Gebiso Challa, Aman Nebo Tibesso, Ashebir Tsegayie Mamo. 2019. Farming System
Characterization of Arsi zone: Case of Small-Scale Farming. American Journal of
Environmental and Resource Economics. 4 (1): 12-24.
Tamrat Gebiso Challa. 2016. Prospects and Challenges of Agricultural Mechanization in Oromia
Regional State-Ethiopia, Policy Perspectives. American Journal of Agriculture and Forestry,
(4) 5: 118-127.
Tamrat Gebiso, Ayalew Bekele, and Ephrem Boka. 2017. Role of Agricultural Mechanization
Technologies in Transforming Agriculture: The Case of Wheat Row Planter in Ethiopia.
American Journal of Biological and Environmental Statistics, (3)1: 10-19.
Tan, Y. L. 1981. The impact of farm mechanization on small-scale rice production. Unpublished.
MS Thesis. UPLB.
Tariku Ayele, Dagnaygebaw Goshme and Haile Tamiru. 2021. Determinants of cereal crops
commercialization among smallholder farmers in Guji Zone, Ethiopia. Cogent Food &
Agriculture, 7(1), 1948249.
Taye Melese Mekie, Gizachew Desalegn, Abebe Birara Dessie, Tadie Mirie Abate. 2019. Market
orientation and Market participation of smallholder barely producers in North Gondar Zone:
Implications for Commercial Transformation. African Journal of Agricultural Research, 13,
 183

20-30.
Taylor, T. G. and Shonkwiler, J. S. 1986. Alternative stochastic specifications of the frontier
production function in the analysis of agricultural credit programs and technical
efficiency. Journal of development economics, 21(1), 149-160.
Tekilu Tadesse and Tesfaye Melaku. 2020. Impact of Rural Infrastructure on smallholder Output
Market Participation in South West Ethiopia. Horn of African Journal of Business and
Economics (HAJBE), 3(2), 29-46.
Temesgen Bogale and Ayalneh Bogale. 2005. Technical efficiency of resource use in the
production of irrigated potato: A study of farmers using modern and traditional irrigation
schemes in Awi Zone, Ethiopia. Journal of Agriculture and Rural Development in the Tropics
and Subtropics (JARTS), 106(1), 59-70.
Tesfaye W, Blalock G, Tirivayi N. 2021. Climate-smart innovations and rural poverty in Ethiopia:
exploring impacts and pathways. Am J Agric Econ. 103(3):878–899. doi:10.1111/ ajae.12161.
Tewoderos Meleaku, Degye Goshu and Bosena Tegegne. 2020. Determinants Sorghum Market
among Smallholder Farmers in Kafta Humera District Tigeray Ethiopia. South Asian Journal
of Social Studies and Economics, 8(1): 1-13, 2020.
The Federal Negarit Gazettee of the Federal Democratic Republic of Ethiopia (FNG). 2012.
Council of Ministers Regulation Number No. 270/2102, Regulation on Investment Incentives
and Investment Areas Reserved for Domestic Investors. 19th Year No.4
Theis, S., Krupnik, T.J., Sultana, N., Rahman, S., Seymour, G., and Abedin, N. 2019. Gender and
Agricultural Mechanization: A mixed-methods exploration of the impacts of multi-crop
reaper-harvester service provision in Bangladesh. IFPRI Discussion Paper 01837 May 2019.
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 Dev., 31(12), 1959–1975.
Thorner, D., Kerblay, B. and Smith, R.E.F. 1966. Chayanov on the Theory of Peasant Economy.
Homewood, IL: Richard D. Irwin.
Timmer, C. P. 1997. Farmers and markets: The political economy of new paradigms. American
Journal of Agricultural Economics, 79(2): 621–627.
Tobin, J. 1958. Estimation of Relationships for Limited Dependent Variables. Econometrica,26,
24-36.
Tolesa Tesema 2021. Analysis of Technical, Allocative and Economic Efficiency In Maize
 184

Production In Ethiopia Evidence From Low Land of Gudeya Bila: Stochastic Frontier
Approach. available at [Link]
792476/v1_covered.pdf?c=1631876542.
Uddin, M.R. 2019. Crop Diversification for Dietary Diversity and Nutrition: Evidence from
Bangladeshi Farm Households. Bangladesh Institute of Development Studies, Dhaka.
Van den Berg, M., Hengsdijk, H., Wolf, J., van Ittersum, M., Guanghuo, W. and Rotter, R. 2005.
The impact of increasing farm size and mechanization on rural income and rice production in
Zhejiang province, China. Paper presented at first Asia-Europe workshop on sustainable
resource management and policy options for rice ecosystem (SUMAPOL 2005), 11-14, 2005,
Hangzhou, Zhejiang province, China, P.R. China.
Vani, N.R. and Pavithra, R.H. 2021. Trends and determinants of crop diversification in
Chikkaballapur District of Karnataka: An Economic Analysis. Journal of Research in
Agriculture and Animal Science, 8 (10): 16-21.
Vincent, D. (2010, July). Bootstrap LM tests for the box cox tobit model. In BOS10 Stata
Conference (Vol. 9). Stata Users Group.
von Braun J., Bouis H. and Kennedy, E. 1994. Conceptual framework. pp. 11-36. In: von Braun, J.
and Kennedy, E. (eds.). Agricultural Commercialization, Economic Development, and
Nutrition. The Johns Hopkins Press Ltd., Maryland.
von Braun, J. 1995. Agricultural commercialization: impacts on income, and nutrition and
implications for policy.
von Braun, J. and Kennedy, E. 1994. Agricultural Commercialization, Economic Development and
Nutrition. The Johns Hopkins University Press, Baltimore and London.
von Braun, J., de Haen, H. and Blanken, J. 1991. Commercialization of agriculture under
population pressure: Effects on production, consumption and nutrition in Rwanda. Research
Report 85. IFPRI (International Food Policy Research Institute), Washington, DC, USA.
Vortia, P., Nasrin, M., Bipasha, S. K. and Islam, M. M. 2021. Extent of farm mechanization and
technical efficiency of rice production in some selected areas of Bangladesh. GeoJournal, 86,
729-742.
Walker, T., and Jodha, N. 1986. How Small Farmers Adapt to Risk. In: Hazell, P., Pomareda, C.
and Valdez, A. (eds.), Crop Insurance for Agricultural Development. Baltimore. Johns
Hopkins University Press.
 185

Wang, X., Yamauchic, F. and Huang, J. 2016. Rising wages, mechanization, and the substitution
between capital and labor: evidence from small scale farm system in China. Agricultural
Economics 47 (2016) 309–317.
Wawire, Nw., Charles Bett, C., Ruttoh, R.C., Wambua, J., Omari, F.O., Kisilu, R., Kavoi, J.,
Omari, J., Wanyonyi, N.W. and Ketiem, P. 2016. The Status of Agricultural Mechanization in
Kenya.
Weiss, C. R. and Briglauer, W. 2002. Determinants and dynamics of farm diversification (No. 723-
2016-48972).
Westcott, P. and Young, C. 2004. Farm program effects on agricultural production: coupled and
decoupled programs. ME Burfisher and J. Hopkins. (eds.) Decoupled Payments in a Changing
Policy Setting, Agricultural Economic Report, 838.
Wharton, C. R. (Ed.). 1969. Subsistence agriculture and economic development. Transaction
Publishers.
Wilson, P., Hadley, D., Ramsden, S., Kaltsas, I., 1998. Measuring and explaining technical
efficiency in UK potato production. Journal of Agricultural Economics, 49 (3), 294-304.
Winship, C. and Mare, R. 1992. Models for sample selection bias. Annual review of sociology.
Volume 19, pp 327-350.
Winters, P., Cavatassi, R. and L. Lipper, 2006. Sowing the Seeds of Social Relations: The Role of
Social Capital in Crop Diversity. FAO, ESA Working Paper No. 06-16, Rome. Italy.
Wollie, G., Zemedu, L. and Tegegn, B. 2018. Economic efficiency of smallholder farmers in
barley production in Meket district, Ethiopia. Journal of Development and Agricultural
Economics, 10(10): 328-338.
Wondimagegn Mesfin, Bekabil Fufa and Jema Haji. 2011. Pattern, trend and determinants of crop
diversification: empirical evidence from smallholders in Eastern Ethiopia. Journal of
Economics and Sustainable Development 2(8): ISSN 2222-2855.
Wondimagegn Tesfaye. 2021. Crop Diversification Increases Household Diets and Improves Child
Growth in Ethiopia. African Economic Research Consortium, Policy brief No. TR007.
Wondmagegn Tafesse Tirkaso. 2013. The Role of Agricultural Commercialization for Smallholders
Productivity and Food Security - An Empirical Study in Rural Ethiopia. Agricultural
Economics and Management – Master’s Programme Degree thesis No 827 · ISSN 1401-4084
Uppsala, available at: [Link]
 186

Wooldridge, J. M. 2002. Introductory Econometrics A Modern Approach, 2002. New York: South-
Western College Publishing.
Wu, J. and Brorsen, B. 1995. The impact of government programs and land characteristics on
cropping patterns. Canadian Journal of Agricultural Economics/Revue canadienne
d'agroeconomie, 43 (1): 87-104.
Wu, W., P. Yang, C. Meng, R. Shibasaki, Q. Zhou, H. Tang, and Y. Shi. 2008. An integrated model
to simulate sown area changes for major crops at a global scale. Science in China Series D:
Earth Sciences 51: 370–9.
Wuletaw Mekuria and Kindu Mekonnen. 2018. Determinants of crop livestock diversification in
the mixed farming systems: evidence from central highlands of Ethiopia. Agric & Food Secur
(2018) 7:60.
Yahya, W.B., Adebayo,[Link],[Link],B. Sanni. 2008. Effects of orthogonality
efficiency of seemingly unrelated regression models.
Yan, C. 2014. Determinants of Agricultural Machinery Equipment Level in China. Hebei Agric., 9:
45–46.
Yapa, L. and C. Mayfield, 2006. Non-adoption of innovations: evidence from discriminate analysis.
Economic Geography, 54(2): 145-156.
Yared Deribe and Moti Jaleta. 2019. Review of National Policies Influencing the Expansion of
Agricultural Mechanization in Ethiopia. Farm Mechanization and Conservation Agriculture
for Sustainable Intensification (FACASI) project. Technical Report.
Yemane Tsehaye, Zeratsion Abera, Afewerk Kebede and Berhane Ghebremichael. 2009. A
Dynamic Sorghum (Sorghum Bicolor (L.) Moench diversity Management in Situ and
Livelihood Resilience in South and Central Tigray Region, Ethiopia. MEJS, 1(2): 67-94.
Yiang, Y.; Liu, P.; Li, N. Analysis on Affecting Factors and System of Agricultural Mechanization
Development in Northeast China. Journal of Agrotech. Econ. 2006, 5, 28–33.
Zangeneh, M., Omid, M. and Akram, A. 2010. Assessment of agricultural mechanization status of
potato production by means of artificial neural network model. Australian Journal of Crop
Science, 4(5), 372-377.
Zeller, M., Diagne, A. and Mataya, C. 1998. Market access by smallholder farmers in Malawi:
Implications for technology adoption, agricultural productivity and crop income. Agricultural
Economics, 19 (1-2): 219-229.
 187

Zellner, A. 1962. An efficient method of estimating seemingly unrelated regressions and tests for
aggregation bias. Journal of the American Statistical Association.
Zhang, C, Shi, G.M., Shen, J., Hu, R.F. 2015. Productivity effect and overuse of pesticide in crop
production in China. Journal of Integrative Agriculture, 14, 1903–1910.
Zhang, J., Wang, J., Zhou, X. 2019. Farm machine use and pesticide expenditure in maize
production: health and environment implications. International Journal of Environmental
Research and Public Health, 16(10):1808. doi:10.3390/ijerph16101808.
Zhang, W.Z. and Hanlin, S. 2014. Empirical Research of Agricultural Mechanization on the Effect
of Increasing Famers’ Income. Management Science and Engineering, 8 (2), 7-13.
Zhang, X., Yang, J., and Thomas, R. 2017. Mechanization outsourcing clusters and division of
labor in Chinese agriculture. China Economic Review, 43, 184-195.
Zhou, X., and Ma, W. 2022. Agricultural mechanization and land productivity in China.
International Journal of Sustainable Development & World Ecology,
[Link]
Zhou, X., Ma, W. and Li, G. 2018. Draft animals, farm machines and sustainable agricultural
production: Insight from China. Sustainability, 10(9), 3015.
Zhou, X., Ma, W. and Li, G. 2021. Draft Animals, Farm Machines and Sustainable Agricultural
Production: Insight from China. Sustainability 2018, 10(9),
3015; [Link]
Zhou, X.; Ma, W.; Li, G.; Qiu, H. 2020. Farm Machinery Use and Maize Yields in China: An
Analysis Accounting for Selection Bias and Heterogeneity. Australian Journal of Agricultural
Resource Economics, 64(1282–1307): 1-26.
 188

7. APPENDICES

Appendix 1. Appendix Tables

Appendix Table 1. Household distribution across level of FMI, COCI, and CDI

Level FMI COCI CDI

Low 30.51 43.33 25.13

Medium 39.51 28.97 43.59
High 29.74 27.69 31.28
FMI=farm mechanization index, COCI=crop output commercialization index, CDI=crop
diversification index

Appendix Table 2. Crop types grown and mean farm land allocation (ha)

Crop type land size Percent Crop type land allocated Percent
Wheat 0.97 50.00% Vegetables* 0.04 2.06%
Barley 0.43 22.17% Potatoes 0.04 2.06%
Teff 0.08 4.12% Oilseeds 0.02 1.24%
Faba bean 0.22 11.86% Lentils 0.04 1.84%
Field pea 0.08 4.64% Chickpeas 0.02 1.24%
Total 1.94 100%
*Vegetables (Ethiopian & head cabbage & red pepper; **oilseeds (rapeseeds & linseeds)

Appendix Table 3. Level of crop diversification index (CDI)

Category of CDI Frequency Percent Cumulative
Low (0-0.38) 98 25.13 25.13
Medium (0.39-0.63) 170 43.59 68.72
High (>0.63) 122 31.28 100.00
Variable Mean Std. dev. Min. Max
Crop diversification index 48% 0.26 0 100%
 189

Appendix Table 4. Farmers’ perception on their crop production system

Crop production system Frequency Percent Cumulative

Diversified 318 81.54 81.54
Specialized 72 18.46 100.00
Reasons for diversification Score Rank
To mitigate risk of market and price fluctuation 233(73.27%) 1
To be food self-sufficient and food secured 199(62.58%) 2
To mitigate risk of crop failure 196(61.64%) 3
To diverse labor pressure during peak agricultural seasons 55(17.30%) 4

Appendix Table 5. Linktest result for Tobit model misspecification test

Farm MI Coefficient Std. err. t

_hat 0.031 0.389 0.08

_hatsq 1.430 0.823 1.24
_cons 0.150 0.090 1.67
var(e.mi_cpd) 0.041 0.003 0.04
LR chi2(2) = 99.89, Prob > chi2 = 0.000, Log likelihood = 67.86, Pseudo R2 = -2.79

Appendix Table 6. Collinearity Diagnostics for commercialization explanatory variables

Variable VIF Tolerance Variable VIF Tolerance
Age 1.67 0.5981 DA contact 1.09 0.9167
Education 1.35 0.7431 Land cultivated 2.41 0.4152
Landholding 2.22 0.4504 Agroecology 1.46 0.6854
Family labor 2.49 0.4014 Family size 2.40 0.4160
Road access 1.12 0.8904 Number of oxen 1.05 0.9511
Crop diversification 1.26 0.7947 Main market distance 1.10 0.9130
Number of equines 1.15 0.8702 Mechanization level 1.28 0.7822
Access credit 1.05 0.9487 Income (ln) 1.13 0.8852
Livestock (TLU) 1.50 0.6662 Access market information 1.06 0.9412
Mean VIF 1.49
 190

Appendix Table 7. Covariate balance summary for impact estimation

Medium MI High MI
Weighted Weighted Weighted Weighted
Covariates Std difference Variance ratio Std difference Variance ratio
Zone 0.059 1.000 0.102 0.987
Sex 0.040 0.785 0.113 0.455
Education status 0.023 1.034 0.109 0.981
Farming experiences 0.041 0.894 0.149 1.462
Mechanization distance -0.046 0.902 -0.093 0.943
Cultivated land 0.076 1.096 0.025 1.036
No. of plots -.003 1.210 0.145 1.399
Family labor 0.018 0.892 0.092 1.085
Household size 0.001 1.311 0.149 1.262
Market participation -.008 1.024 -0.187 1.519
Livestock (TLU) -.062 1.145 0.208 2.284
Road access 0.054 0.879 -0.038 1.085
Main market distance -0.041 1.013 0.111 1.171
Crop diversification (SDI) -0.004 0.890 0.039 0.983
Mechanization distance 0.054 0.953 0.064 0.943
Dependency ratio 0.039 1.291 .023 0.757
Social capital 0.023 0.848 0.007 0.793

Appendix Table 8. Collinearity diagnostics for inputs used in barley production

Variable VIF SQRT-VIF Tolerance R-Squared
LCOST_SEED 1.07 1.04 0.93 0.0686
LCOST_FER 1.04 1.02 0.96 0.0356
LCOST_LBR 1.09 1.04 0.92 0.0834
LCOST_CHM 1.08 1.04 0.92 0.0768
LCOST_COMB 1.07 1.03 0.94 0.0647
LCOST_TRAC 1.04 1.02 0.96 0.0354
LCOST_LND 1.03 1.01 0.97 0.0260
Mean VIF 1.06
 191

Appendix Table 9. Collinearity diagnostics for input price in barley production

Variable VIF SQRT-VIF Tolerance R-Squared

LP_OXEN 1.09 1.04 0.92 0.08

LP_SEED 1.01 1.00 0.99 0.01
LP_FERTLIZER 1.01 1.01 0.99 0.01
LP_LBOOR 1.02 1.01 0.98 0.02
LP_CHEMICALS 1.03 1.01 0.98 0.02
LP_LAND 1.06 1.03 0.95 0.05
LP_THRESHING 1.06 1.03 0.94 0.06
LP_TRACTOR 1.05 1.03 0.95 0.05
Mean VIF 1.04

Appendix Table 10. Collinearity diagnostics for wheat production inputs

Variable VIF SQRT-VIF Tolerance (1/VIF) R-Squared

lnoxday 1.48 1.22 0.6742 0.3258

ln_fert 1.14 1.07 0.8781 0.1219
Ln_lnd 2.15 1.47 0.4641 0.5359
ln_lbr 1.73 1.31 0.5785 0.4215
lchm 1.19 1.09 0.8385 0.1615
lcomb 1.12 1.06 0.8912 0.1088
ltrac 1.31 1.14 0.7643 0.2357
lseed 1.12 1.06 0.8947 0.1053
Mean VIF 1.41
 192

Appendix Table 11. Model fitness test (parallel regression assumption and collinearity
diagnosis)

Test types Chi-square value Significance (probability)

omodel test 7.75 0.9558 (not significant)

brant test 8.42 0.866 (not significant)
gologit2 test 8.98 0.9141 (not significant)
Collinearity Diagnostics

Variables VIF Tolerance Variables VIF Tolerance

Household head sex 1.07 0.9335 DA contact 1.09 0.9209
Farming experience 1.52 0.6560 Access to credit 1.05 0.9520
Educational background 1.37 0.7306 Mechanization level 1.28 0.7829
Family size 1.29 0.7770 Number of plots 1.19 0.8376
Dependency ratio 1.14 0.8784 Off-farm income 1.05 0.9503
Total landholding 1.32 0.7551 AEZs 1.48 0.6764
Livestock (TLU) 1.53 0.6527 Main market distance 1.19 0.8424

Mean VIF 1.26

Appendix Table 12. Mean distribution for FMI, COCI and CDI across each level (percent)
Level FMI COCI CDI

Low 20.15 16.04 9.10

Medium 50.01 52.02 52.34
High 76.91 77.20 71.84
 193

Appendix Table 13. Breusch–Pagan/Cook–Weisberg test for heteroskedasticity of

commercialization studies

Assumption: Normal error terms

Variable: Fitted values of commercialization index

H0: Constant variance

Chi2(1) = 1.78
Prob > chi2 = 0.1823

Since probability of Chi-squared (1.78) is insignificant with p-value of 0.1823, we are not able
to reject the null hypothesis of normal error terms. Hence, we conclude that there is no series
problem of heteroscedasticity in this data.

Appendix Table 14. Test for endogeneity of level of farm mechanization:

Coefficients

(b) (B) (b-B) sqrt(diag(V_b-V_B))

Dependent variables bIV BOLS Difference Std. err.

AGE 0.0003 0.0010 0.0010 0.0005

EDUCATION -0.0012 0.0043 -0.0055 0.0017
AEZS 0.0459 0.1468 -0.1009 0.0229
TOTAL LAND 0.0381 -0.0222 0.0603 0.0069
LAND_CULTD -0.0228 0.02589 -0.0487 0.0080
FAMILY_SIZE -0.0058 0.0054 -0.0112 0.0019
FAMILY_LABOR 0.0113 -0.0125 0.0238 0.0055
OXEN 0.0211 0.0146 0.0065 0.0050
LIVESTOCK (TLU) -0.0012 -0.0066 0.0055 0.0018
ROAD_ACCS 0.0143 0.0622 -.0479 0.0193
DIST_MROAD -0.0107 -0.0033 -0.0074 0.0012
CROP_DIVRS (SDI) -0.1051 -0.0077 -0.0974 0.0227
EQUINES 0.0091 -0.0193 0.0284 0.0116
TOTAL INCOME 0.0489 0.0212 0.0278 0.0041
DA_CONTACT -0.0656 0.0081 -0.0737 0.0128
 194

ACCS_CREDIT -0.0291 -0.0116 -0.0175 0.0152

ACCESS_MI 0.1596 0.0004 0.1592 0.0119
b = Consistent under H0 and Ha; obtained from regress.
B = Inconsistent under Ha, efficient under H0; obtained from regress.
Test of H0: Difference in coefficients not systematic
chi2(17) = (b-B)'[(V_b-V_B)^(-1)](b-B) = 773.17; Prob > chi2 = 0.0000

Appendix Table 15. Conversion factors to compute tropical livestock unit (TLU)

Livestock Type Conversion factors (TLU per head)

Camel 1.25
Horse 1.10
Cow and Oxen 1.00
Heifer 0.75
Donkey (adult) 0.70
Donkey (young) 0.35
Weaned calf 0.34
Calf 0.25
Sheep and goat (adult) 0.10
Sheep and goat (young) 0.06
Chickens 0.01

1 TLU (Tropical Livestock Unit) is calculated for an animal of 250kg live-weight.

Source: Storck et. al., 1991.

Appendix 2: Stata Commands and Estimation Procedures

Determinants of level of farm mechanization: Tobit model
• tobit mi_cpd zone off_farm sex grade_educ exprce_year mech_dist LandCultvatd plots_numb
tm_eqvt mkt_partcipt tlu near_road MMKT_DIST SDI Social_Cap, ll(0) ul(1) vce(robust)
• mfx compute, predict(p(0,1)); mfx compute, predict(e(0,1)) and
• mfx compute, predict(ys(0,1))
 195

1. Model diagnostic tests made to check the appropriateness of Tobit model:

a. LM test of Tobit specification

• tobit mi_cpd zone off_farm sex grade_educ exprce_year mech_dist LandCultvatd plots_numb
tm_eqvt mkt_partcipt tlu near_road MMKT_DIST SDI Social_Cap, ll(0) ul(1) vce(robust)
• bctobit, nodots OR tobcm, pbs reps(500)
b. Normality
• tobit mi_cpd zone off_farm sex grade_educ exprce_year mech_dist LandCultvatd plots_numb
tm_eqvt mkt_partcipt tlu near_road MMKT_DIST SDI Social_Cap, ll(0) ul(1) vce(robust)
• histogram mi_cpd, normal bin(19) xline(0.52631579)
• histogram mi_cpd, discrete freq
Linktest is used if there is model specification errors and to conduct the test
• tobit mi_cpd zone off_farm sex grade_educ exprce_year mech_dist LandCultvatd plots_numb
tm_eqvt mkt_partcipt tlu near_road MMKT_DIST SDI Social_Cap, ll robust
• linktest
2. Estimation of Economic and Technical Efficiency for wheat and barley producers
• sfcross ly llnd loxday lfer llbr lchm lcomb ltrac lseed, model(bc) dist(tn) ort(o) emean(sex
exprce_year grade_educ tm_eqvt LandCultvatd Ext_Mech Lnd_Frag MMKT_DIST
mech_dist ratio_dep tlu oxen_Num Accss_road SDI) robust
• sfcross lntc_wt_qt lnoutput_wt lnP_seed_wt lnP_lnd_wt lnP_tract lnP_lbor lnP_Oxren
lnP_chem lnP_fert lnP_thresh_wt, model(bc) dist(tn) ort(o) emean(sex exprce_year
grade_educ LandCultvatd Ext_Mech near_road MMKT_DIST mech_dist ratio_dep tlu
off_farm oxen SDI) cost robust
• predict ee_bc_wt, bc
• predict ee_jlms_wt, jlms
• sum ee_bc_wt te_jlms_wt

3. Augmented Inverse Probability Weight (AIPW) result for Impact analysis

• teffects aipw(ee [Link] age grade_educ sex tm_eqvt LandCultvatd Lnd_Frag ratio_dep tlu,
linear) (fraction [Link] age exprce_year grade_educ sex tm_eqvt LandCultvatd Ext_Mech
Social_Cap Lnd_Frag nmkt_dst MMKT_DIST mech_dist , probit), aequations ate
 196

• teffects aipw(ee [Link] age grade_educ sex tm_eqvt LandCultvatd Lnd_Frag ratio_dep tlu)
(mi_rate [Link] age exprce_year grade_educ sex tm_eqvt LandCultvatd Ext_Mech
Social_Cap Lnd_Frag nmkt_dst MMKT_DIST mech_dist , logit), aequations ate
• teffects aipw(ee [Link] age grade_educ sex tm_eqvt LandCultvatd Lnd_Frag ratio_dep tlu)
(mi_rate [Link] age exprce_year grade_educ sex tm_eqvt LandCultvatd Ext_Mech
Social_Cap Lnd_Frag nmkt_dst MMKT_DIST mech_dist, logit), aequations osample(OLvar)
ate
• drop if OLvar ==1
• teffects aipw(ee [Link] age grade_educ sex tm_eqvt LandCultvatd Lnd_Frag ratio_dep tlu)
(mi_rate [Link] age exprce_year grade_educ sex tm_eqvt LandCultvatd Ext_Mech
Social_Cap Lnd_Frag nmkt_dst MMKT_DIST mech_dist , logit), aequations ate
• tebalance summarize

4. Determinants of Crop Commercialization- Heckman two-stage models

• heckman com_index age grade_educ ecology tland LandCultvatd HH_Size tm_eqvt oxen tlu
near_road MMKT_DIST SDI equines lnincome cont_da get_crdt get_mi mi_cpdHat,
select(mkt_partcipt= sex age grade_educ ecology tland LandCultvatd HH_Size tm_eqvt tlu
oxen near_road MMKT_DIST SDI equines lnincome cont_da get_crdt get_mi mi_cpdHat)
twostep

5. Determinants of level of crop diversification

Model diagnostic tests made to check the appropriateness of ordered logit model:
• ologit CDI_Catg sex exprce_year grade_educ HH_Size ratio_dep tland tlu cont_da get_crdt
mi_cpd plots_numb none_farm Ag_Eco MMKT_DIST near_road Social_Cap mech_dist
mkt_orient get_mi Com_Index mkt_partcipt mi_cpd T_Income Ag_Eco Social_Cap
Wald's test for test whether the coefficients of the variables in the M - 1 equations are all
equal (which means parallel lines holds assumption).
• gologit2 CDI_Catg sex exprce_year grade_educ HH_Size ratio_dep tland tlu cont_da get_crdt
mi_cpd plots_numb none_farm Ag_Eco MMKT_DIST mkt_partcipt T_Income
• test [#1 = #2]
Omodel test
 197

• omodel CDI_Catg sex exprce_year grade_educ HH_Size ratio_dep tland tlu cont_da get_crdt
mi_cpd plots_numb none_farm Ag_Eco MMKT_DIST mkt_partcipt T_Income
ologit postestimation test for proportional odds assumption/parallel lines holds assumption:
• brant
• brant, detail
• gologit2 CDI_Catg sex exprce_year grade_educ HH_Size ratio_dep tland tlu cont_da get_crdt
mi_cpd plots_numb none_farm Ag_Eco MMKT_DIST
• test [#1 = #2]
• ologit CDI_Catg sex exprce_year grade_educ HH_Size ratio_dep tland tlu cont_da get_crdt
mi_cpd plots_numb none_farm Ag_Eco MMKT_DIST, or
• test [#1 = #2]
• ologit CDI_Catg sex exprce_year grade_educ HH_Size ratio_dep tland tlu cont_da get_crdt
mi_cpd plots_numb none_farm Ag_Eco MMKT_DIST, or
• graph box SDI, over(fraction) over(ecology) over(plots_numb) title(Boxplots of Crop
Diversification Level by variables)

Model specification tests of ologit againist oprobit

• oprobit CDI_Catg sex exprce_year grade_educ HH_Size ratio_dep tland tlu cont_da get_crdt
mi_cpd plots_numb none_farm Ag_Eco MMKT_DIST
• estat ic
• ologit CDI_Catg sex exprce_year grade_educ HH_Size ratio_dep tland tlu cont_da get_crdt
mi_cpd plots_numb none_farm Ag_Eco MMKT_DIST
• estat ic
• omodel ologit CDI_Catg sex exprce_year grade_educ HH_Size ratio_dep tland tlu cont_da
get_crdt mi_cpd plots_numb none_farm Ag_Eco MMKT_DIST
6. SUR model stata command
• sureg ( mi_cpd sex age grade_educ HH_Size tland tlu oxen [Link] ecology nmkt_dst
MMKT_DIST mech_dist soil_fertlty Social_Cap plots_numb prodn_goal) (Com_Index sex
age grade_educ HH_Size tland tlu oxen ecology nmkt_dst MMKT_DIST coop_dist road_dist
ftc_dist contrac_farm off_farm price_comp lnincome) (SDI sex age grade_educ HH_Size
ratio_dep tland tlu oxen [Link] ecology MMKT_DIST i.risk_att plots_numb), corr isure
 198

Appendix 3. Sample Households Survey Interview Schedule

Farm Mechanization, Commercialization and Agricultural Practices in the Highlands

and Mid-lands of Central and Southern Oromia Region, Ethiopia

PhD Dissertation Research Survey Questionnaire

General instruction for Enumerators

1. Please introduce yourself before starting the interview according to local custom.
2. Inform the purpose of the study, which is to collect the first-hand data that will be used only
for research activity writing on the above indicated topic only for knowledge purpose.
3. Please be careful on the estimates when exact measurements are not available.
4. Carefully read and follow the instructions while you are starting the interview.
5. Do not forget to thank when you finish your interview

Dissertation objectives
This study was initiated with the following specific objectives:
To estimate the farm mechanization level and identify its determinants for major crops
To estimate the level of economic efficiency of farming households
To estimate the impacts of farm mechanization on farm labor employment, economic
efficiency, household income, and intensity of farm inputs used for major crops of the area;
To identify determinants of farm commercialization;
To estimate crop diversification level and identify determinants of crop diversification; and
To analyze the inter-linkage among farm mechanization, commercialization, and crop
diversification in the study area.

Household ID ________________; Date ____________________

Questionnaire ID: _____________________; Administrative zone code: ___________;
District code _________ Kebele code _______; Enumerator’s name (code): ___________
 199

Part I: Household Characteristics

1. Name of the respondent (Household head) _____________________________
2. Sex of the household head: 1. Male 0. Female
3. Age of respondent _____________ years
4. Marital status:
1) Single 2) Married 3) Divorced 4) Widow
5. Educational status of household head/respondent: 1. Illiterate 2. Read and
write 3. Informal (religious school)
6. If the household head is literate what is his/her years of schooling? ________
7. Family size of the household
S. N. Age category Number of males Number of females
1 Bellow 10 years
2 10-14 years
3 14-16 years
4 17-64 years
5 Above 64 yrs

8. What is the number of family members working full-time in the farm? ____________
9. What is your experience in farming (years): ___________________ years
Part II: Access to Services
1. Distance to infrastructures/service centers:
No. Infrastructure/service center Distance (Km) from residence
1 Nearest market
2 Farmer cooperative center
3 Main road
4 FTC
5 Main market
5 Mechanization service providing center
6 Major (district’s) town
7 DA office
 200

2. Do you have an all-weather road near your residence? 1) Yes 0) No

3. Do you have contact with DA? 1) Yes 0) No
4. If ‘Yes’, how often do you contact DA per month? ________________ times per month.
5. Do you have access to credit (can you borrow if you need it)? 1) Yes 0) No
6. If yes, for what purposes did you borrow money for 2012/13 or 2013/14 production
seasons?
No. Purpose of credit Amount Source
1 Fertilizer Relatives/friends
2 Seed Ekub
3 Agrochemicals purchase Saving and credit
associations
4 To participate on off/non-farm activities [it Banks
includes fattening, petty trading etc.]
5 To fulfill basic needs (food, clothes) Cooperatives/unions
6 Children’s school
7 To pay for/buy farm mechanization
technologies services

7. Are you a member of any of these associations in the last two years? (list them)
Type of associations For how many years? Role in the institution? Still member
(Use code A) (Use code B) now? (Use code
C)

Code A Code B Code C

1. Input Supply/service coops /union 1. Elected official 1. Yes
2. Crop producer and marketing coops 2. Ordinary member 0. No
 201

3. Local administration (Kebele) 3. Agricultural cadre

4. Women’s Association 4. Model farmer
5. Youth Association 5. Cashier
6. Mosque/Church Association 6. Coordinator
7. Saving and credit group 7. Secretary
8. Funeral Association (Idir) 8. Militia
9. Seed producers’ cooperatives 9. Store keeper
10. Water User’s Association 10. Executive committee
member

8. Main roofing material of your main residential house? 1. Grass thatched 2. Iron
sheet
Part III: Household Land Holding and Land Use Pattern
Land possession during the 2013/14 E.C cropping season
No. Landholding Size/area in ha Value of land in year basis*
1 Total cultivated land
2 Total own landholding (from PA)
3 Grazing land
4 Homestead land
7 Land rented in
8 Land shared in
9 Land rented out
10 Land shared out
11 Total land
*Take rental value since it is the best proxy for price of land
2. What is the number of plots/parcels of your agricultural land? ______________.
3. What is the size of each plot?
Plot number Plot size (ha) Plot number Plot size (ha)
1 3
2 4
4. In general what is the fertility status of your farmlands?
 202

1) Good 2) Medium 3) Poor

Part IV: Off/non-farm activities
1. Do you or/any member of your family have an off-farm/non-farm job? 1. Yes 0. No
2. If yes, indicate the type of off/non-farm activities and annual income for the year 2013/14
S/N Off/non-farm activities 1. Yes Payment mode Total annual
0. No 1. In kind 2. Cash income(birr/annual)
1 Livestock trading
2 Crop trading (grain, vegetable
…)
3 Petty trade (consumer goods)
4 Hired laborer (on/non-farm
activities)
5 Rental income (land, house...)
6 Income from aid (safety net)
7 Salaried employment
8 Remittance

Part V: Cost of Production for the Selected Major Crops

1. Calculating the costs of production for wheat crop: your total wheat farm size is _____ ha.

Means of operation (amount and unit cost)

I Type of Agricultural operations MD U-cost OD U-cost Machine
cost/ha
1 Primary tillage
2 Secondary tillage
3 Tertiary tillage
4 Fourth tillage
5 Fifth tillage
6 Sixth tillage
7 Planting
 203

II Field Management
1 Mechanical cultivation (shogoru)
2 Mechanical weeding MD U-cost OD U-cost Machine
cost
3 1st round
4 2nd round
5 3rd round
III Weedicide Application MD U-cost OD U-cost Machine
cost
1 1st round
2 2nd round
3 3rd round
IV Insecticide application MD U-cost OD U-cost Machine
cost
1 1st round
2 2nd round
V Fungicide application (kan waagii) MD U-cost OD U-cost Machine
cost
1 1st round
2 2nd round
3 3rd round
VI Types of fertilizer used MD U-cost OD UC Machine
cost
A DAP
B UREA
C NPS+
VII seed used: Improved seed Quantity: ____kg Total cost: _________Birr
Local seed Quantity: ____kg Total cost: _________Birr
1 Harvesting
2 Threshing
 204

3 Total production ______ quintal Farm gate price:

______Birr/100kg
4 Price at your nearest market ___________Br/100kg (qt)
Key: MD=number of man days; U-cost=unit cost; OD= number of oxen-days

1. Calculating the costs of production for barley crop: your total barley farm size is _____ ha.
Means of operation (amount and unit cost)
I Type of Agricultural operations MD U-cost OD U-cost Machine
cost/ha
1 Primary tillage
2 Secondary tillage
3 Tertiary tillage
4 Fourth tillage
5 Fifth tillage
6 Sixth tillage
7 Planting
II Field Management
1 Mechanical cultivation (shogoru)
2 Mechanical weeding MD U-cost OD U-cost Machine
cost
3 1st round
4 2nd round
5 3rd round
III Weedicide Application MD U-cost OD U-cost Machine
cost
1 1st round
2 2nd round
3 3rd round
IV Insecticide application MD U-cost OD U-cost Machine
cost
 205

1 1st round
2 2nd round
V Fungicide application (kan waagii) MD U-cost OD U-cost Machine
cost
1 1st round
2 2nd round
3 3rd round
VI Types of fertilizer used MD U-cost OD UC Machine
cost
A DAP
B UREA
C NPS+
VII seed used: Improved seed Quantity: ____kg Total cost: _________Birr
Local seed Quantity: ____kg Total cost: _________Birr
1 Harvesting
2 Threshing
3 Total production ______ quintal Farm gate price:
______Birr/100kg
4 Price at your nearest market ___________Br/100kg (qt)

Calculating the costs of production for barley crop: your total barley farm size is _____ ha.
Key: MD=number of man days; U-cost=unit cost; OD= number of oxen-days
14. Do you use crop rotation? (Please explain shortly about crop rotation) 1) Yes 0) No

Part VI: Crop Production, marketing, and Crop Diversification

1. Crop production and income from crops in 2013/14 cropping season
Crop Cultivated T. The Quantity Did you Quantity Selling
types Area (ha) produced quantity saved sell? 1. sold (kg) price
grown (kg) consumed for Yes 0. No (birr/kg)
kg) seed(kg)
 206

Wheat
Barley
Maize
Sorghum
Teff
Common
bean
Field pea
Chick
peas
Faba
bean
Khat
Potato
Sweet
potato
Onion
Tomato
Hot
pepper
Garlic
Carrot
Others:-

2. In general, the sale of your crop is distressed or surplus? 1. Distressed 2. Surplus

3. Your production goal or target is: 1. Totally for household consumption
2. Mainly for consumption but to some extent for market 3. Equally for both
consumption and market 4. Mainly for market purposes but for consumption to some
extent 5. Entirely for market purpose
4. Crop produced mainly for market purposes is 1. Cereals 2. Pulses 3. Vegetables
 207

5. Your Production goal is market-oriented; do you agree? 1. Strongly disagree

2. Disagree 3. Neither disagree nor agree 4. Agree 5. Strongly
agree
6. Do you store your agricultural product for later sale? 1. Yes 0. No.
7. If you store your product, for how many months do you store it? _____________ months.
8. What criteria do you consider while selecting crop for production? (Rank them 1 to 7)
No. Selection criteria Rank No. Selection criteria Rank
1 food types accustomed to the 5 Suitability of the crop for
family/society rotation
2 Suitability of the crop to my land 6 Availability of inputs
3 Availability of market/good price 7 Availability of its
mechanization technologies
4 Availability of land

9. If you do not store your product for later sale, why?

1. I do not have enough (surplus) products for sale
2. I do not have conducive (appropriate) and safe storage facilities
3. I do not have enough space for storage
4. Because I need money for immediate obligations (purposes)
5. I do not see the importance of storing products for future sale
10. What are the two more important constraints in marketing your agricultural products
(crops)?
1. Low demand for products 2. Price fluctuation
3. Middlemen’s problems 4. Absence of market information
5. Absence of road access 6. The absence of transportation means
11. To whom do you sell your agricultural products?
1. Directly to rural consumers 2. Directly to urban consumers 3.
To retailers and assemblers 4. To wholesalers
5. Agricultural cooperatives
12. Do you have mobile phone? 1. Yes 0. No
 208

13. Do you have equines to transport your agricultural products to market? 1. Yes 0.
No
14. What means of transportation do you use to transport your agricultural products to market?
1. Equines back 2. Cart 3. Vehicle 4. Human back
15. Do you get market Information (like price) for your agricultural products? 1. Yes 0.
No
16. List two crops that you mainly grow for market purposes: first ______; Second ________.
18. Do you have a contract farming agreement for the marketing of any of your crops?
1. Yes 0. No
19. If you do have a contract farming agreement, for which crop? ____________________.
20. If you do have a contract farming agreement, who is the contracting party?
______________.
21. In general do you use the improved seed for your major crops regularly? Yes, 0.
No
22. Is your crop production diversified or specialized? 1. Specialized 2. Diversified
23. If it is diversified, why? 1. To mitigate the risk of crop failure 2. To mitigate the
risk of price fluctuation 3. To get food self-sufficiency
4. To diversify labor pressure during peak time
24. What is (are) the main reason for not to diversify your crop production? (Rank them)
No. Reason for not to diversify Rank
1 I need crops that are more commercialized
2 I need more mechanized crops
3 My land size is too small and limits my diversification needs
4 Inputs like improved seed availability restrict my need

25. Which crop types cover most parts of your land? 1st crop ______=_________%; 2nd crop
_______= ________%; 3rd crop ________ = ______% and 4th crops ______ = _______%.

Part VII: Agricultural Mechanization-Related Issues

1. Did you use any farm mechanization technologies (listed in Q.2 below) in the (2022) 2013/4
E.C. Production season? 1. Yes 0. No
 209

2. Which Farm mechanization are you using currently (this year)?

No. Farm Mechanization type Response Supplier*
1 Two wheels tractors 1. Yes 0. No
2 Four wheels tractors 1. Yes 0. No
3 Walking behind harvesters 1. Yes 0. No
4 Engine-driven threshers 1. Yes 0. No
5 Combine harvesters 1. Yes 0. No
*1. Own 2. Cooperatives/Union, 3. Government enterprises; 4. Private owners
3. How often do you use this farm machinery? 1. Rarely 2. Sometimes 3.
Every year (regularly)
4. What are the constraints you face in using farm mechanization in general? (Rank 1-4)
No. Constraints to use farm mechanization technologies Rank
1 Unavailability of the technologies in the area
2 High price of the service or technologies to purchase
3 Technical incapability on how to use the technology
4 Ignorance to the technology

5. Do you participate in training on farm mechanization issues? 1. Yes 0. No

6. Do you participate in demonstrations of farm mechanization? 1. Yes 0. No
7. Do you participate in field days of farm mechanization? 1. Yes 0. No
8. Do you get extension services on farm mechanization? 1. Yes 0. No
8. If your answer is yes, how do you rate extension service provision on farm mechanization?
1. Poor 2. Neither poor nor good (medium) 3. Good
9. Properties of your farm plot 1. Slopy/rocky and inconvenient for mechanization
2. It is neither too sloppy nor rocky to use farm mechanization

Part VIII: Livestock ownership and annual income from livestock

No. Type of livestock Current owned Sold within a year Sales amount (ETB)
since September
1 Oxen
2 Cow
 210

3 Calf
4 Weaned calf
5 Heifer
6 Horse
7 Mule
8 Donkey (adult)
9 Donkey (young)
10 Sheep (adult)
11 Sheep (young)
12 Goat (adult)
13 Goat (young)
14 Chicken /poultry
birds
15 Beehives Bee swarm
Honey
(kg)

Thank you