,_ p; 1‘1: :
v r “E"
JNLR‘“

 

x.

F vi»

\

"Cm.
.134 ‘ A 3'
1 ’.

1
A

v .

(Can

‘
l A
wk

 

THESvs

MICHIGAN STATE UNIVE EIRS TYL

I ll Milli} W H ’HI’MJIIIIW

3 12 301033 2215

i!

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

This is to certify that the

dissertation entitled

An Economic Analysis of Smallholder Wheat Production
and
Technology AdOption in the Southeastern Highland;
of Ethiopia

presented by

Mulugetta Mekuria

has been accepted towards fulfillment
of the requirements for

Ph.D. degree m Agricultural
Eccnomics

 

 

 

Cdffm

Major professor

Date March 28, 1994

MSU is an Affirmative Action/Equal Opportunity Institution 0- 12771

 

LIBRARY
Michigan State
Unlverslty

 

 

 

PLACE ll RETURN BOX to mnovo this checkout from your rocord.
TO AVOID FINES rotum on or bdoro onto duo.

DATE DUE DATE DUE DATE DUE

 

 

 

 

' 513.39%

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU I. An Attirmotivo Action/Equal Opponunlty Institution
W

AN ECONOMIC ANALYSIS OF SMALLHOLDER WHEAT PRODUCTION
AND
TECHNOLOGY ADOPTION IN THE SOUTHEASTERN HIGHLANDS
OF ETHIOPIA
By

Mulugetta Mekuria

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Agricultural Economics

1994

ABSTRACT

AN ECONOMIC ANALYSIS OF SMALLHOLDER WHEAT PRODUCTION
AND
TECHNOLOGY ADOPTION IN THE SOUTHEASTERN HIGHLANDS OF
ETHIOPIA

By

Mulugetta Mekuria

The countries of Sub Saharan Africa are faced with lagging food
production, pervasive poverty, high rates of population grth and degradation of
natural resources. Most of these countries, including Ethiopia, meet their food
requirements from domestic production, imports and food aid. Wheat imports
account for 75 percent of wheat consumed in Africa. In Ethiopia about half of
the wheat consumed is imported. Increasing wheat production is thus a high-
priority topic in Sub Saharan Africa and in Ethiopia. This study is motivated by
the need to better understand the economics of smallholder wheat production and
technology adoption in Arssi, a major wheat growing region in southeastern
Ethiopia.

The study tests the hypothesis that farmers’ characteristics, economic and
institutional factors, and farmers’ perceptions of recommended technologies
significantly influence farmers’ adoption decisions. Partial budgets and marginal
returns analysis are used to compare the profitability of recommended
technologies and four alternative production packages. Descriptive statistics, logit,

ordered probit, Tobit and discriminant analysis models were used to identify and

0

estimate the quantitative impact of variables influencing adoption, and to classify
farmers into adopter categories. The primary data were collected during the
1990/91 crop season from a sample of 426 wheat farmers in five wheat-growing
districts of Arssi region.

All the sample farmers planted improved wheat varieties and 87 percent of
the respondents reported using fertilizer. However, only 48 percent were using 50
kg/ha or less fertilizer and only 5-7 percent of them applied the recommended
rate of 100 kg/ha. A fifth of the farmers used herbicide. The results of the net
benefit and marginal rate of return analysis indicated that wheat production
technologies are profitable but inputs are used suboptimally. The econometric
model results showed that institutional variables (input availability, credit access
and extension contact) significantly affect the incidence of adoption while
economic factors (farm size, oxen ownership, labor availability) influence the
intensity of use. The analysis of adoption revealed that 17, 46 and 37 percent of
the farmers are low, moderate and advanced adopters, respectively. Different
factors distinguish the adopter categories in each district. The implications of
these results for agricultural research, extension and development policy are

summarized.

DEDICATED

In living memory of my mother, Zenebework Seyonm who was always eager to
see her children attain the highest possible level of education but who sadly

passed away just three weeks before the completion of this dissertation,

to my father Mekuria Asfaw, who is proud to hear and see that his son has finally

achieved the academic dream he always wanted him to realize,

to my dearest wife Nigist Bekele whose encouragement, sacrifice and love made

this all possible.

ACKNOWLEDGMENTS

I would like to express my gratitude to Professors Carl Eicher, my major
professor and Eric Crawford my dissertation adviser whose support and guidance
was crucial to the successful completion of my graduate program. I would also
like to thank other members of my committee, Professors Richard Bernsten,
James Bonnen and Carl Liedhom for their support, understanding and guidance
at the various stages of my graduate program.

I deeply acknowledge the Fellowship Award and financial support of the
International Development Research Center of Canada and also for allowing me
to pursue my training at a university of my choice, Michigan State University. My
field work would not have been possible without the financial support of
CIMMY'I’S Economics program and the continued assistance of its Director, Dr.
Derek Byerlee. My Special thanks goes to him. I wish to acknowledge the support
of the Institute of Agricultural Research for granting me the study leave and
providing me the necessary facilities for my field research.

I am thankful to my colleagues in the Department of Agricultural
Economics at IAR, specially to Hailu Beyene, Aleligne Kefyalew, Senait Regassa
and Meskerem Haile Mariam, who were able to provide me all they could during
my field work and data entry. I am thankful to Amanuel Gorfu, Manager of

Kulumsa Research Center, Debela Dinka, Head of the Arssi Regional

Agricultural Development Department, Semu Getahun and Teshome Negussie,
Head and agricultural economist of the Regional Planning Office, respectively,
who gave me access to the facilities of their respective institutions and their time
under difficult security situation. My deep gratitude goes to the farmers and
extension agents in the study areas and the ten student enumerators from
Alemaya University of Agriculture.

My recognition goes to Dr. Seme Debela, General Manger of LAR (1984-
92) whose leadership and vision of institution building greatly contributed to my
perspective of research management and interdisciplinary research. I feel proud
and privileged to have worked under him. My recognition goes also to many of
my colleagues and friends in the IAR and specially Dr. Steven Franzel (now with
ICRAF) and Dr. Girma Begashaw of the IMF who always had faith in me and
supported me throughout my graduate program. I am grateful to Pat Eisele,
Sherry Rich, Ann Robinson and the staff of the computer service of the
Department of Agricultural Economics at MSU for giving me the necessary
support whenever I asked for it. Professor Carl and Shirley Eicher have always
made us feel at home at Michigan State under difficult times and we are grateful
for their support and hospitality.

My greatest debt and appreciation are to my wife Nigist, who offered
patience, encouragement, love and care throughout our respective graduate

programs we both successfully completed.

TABLE OF CONTENTS

Page
LIST OF TABLES ............................................ xii
LIST OF MAPS AND FIGURES ................................. xiv
CHAPTER 1 INTRODUCTION ............................. 1
1.1 Background ................................... 1
1.2 Problem Statement and Justification for the Study ...... 2
1.3 Wheat Production and Consumption in Africa and
Ethiopia .................................... 4
1.4 Research Objectives ............................ 5
1.4.1 Specific Objectives ........................ 7
1.5 Organization of the Dissertation ................... 7
CHAPTER 2 PROBLEM SETTING .......................... 9
2.1 Country Profile and the Agricultural Sector ........... 9
2.2 The Resource Endowments ..................... 10
2.2.1 The Natural Setting ...................... 10
2.2.2 Production Potentials and limitations ........ 13
2.3 Structure of the Agricultural Sector ................ 16
2.3.1 Land Tenure and Peasant Organizations ....... 16
2.3.2 Institutions: Input Use, Distribution and Credit
Services ............................... 20
2.4 Agricultural Development Policy : Revisited ...... 22
2.4.1 Agricultural Development Policy: Pre 1974
Period ................................ 23
2.4.2 Agrarian Socialism ....................... 24
2.4.3 Agricultural Policy in the Transition Period . . . . 28
25 Agricultural Technology Generation and Transfer . . . . 29

2.5.1 Agricultural Research in Ethiopia: Evolution

and Development ........................ 29
2.5.2 Agricultural Extension Service .............. 32
2.5.3 Agricultural Research and Extension linkage . . . 34
2.5.4 Experiences in Integrated Rural Development . . 35
2.5.5 Constraints in Project Activities ............. 38

vii

2.6

CHAPTER3

3.1
3.2
3.3

3.4

3.5
CHAPTER 4

4.1

4.2

4.3

4.4

4.5

Summary ................................... 41
CONCEPTUAL FRAMEWORK, LITERATURE

REVIEW AND RESEARCH DESIGN ............. 42
Conceptual Framework of Technology Adoption ...... 42
Technology Adoption Studies .................... 44
Research Design and Methodology ................ 48
3.3.1 The Study Area ......................... 48
3.3.2 Crop Production ..... ‘ .................... 50
3.3.3 Farmers’ Organizations .................... 50
Survey ..................................... 51
3.4.1 Locations and Sampling Methods ............ 51
3.4.2 Questionnaire Preparation and Enumerator
Training ............................... 52
3.4.3 Data Collection ......................... 53
Summary ................................... 54

SMALLHOLDER WHEAT PRODUCTION IN ETHIOPIA6

Introduction ................................. 5 6
4.1.1 Wheat in Ethiopian Agriculture ............. 56
4.1.2 Regional Production Trends ................ 59
4.1.3 Wheat Production, Area and Importance in

Arssi ................................. 62
Crop Production Technology Use in Arssi ........... 64
4.2.1 Integrated Rural Development and Technology

Adoption .............................. 64
4.2.2 Adoption of Commercial Fertilizer .......... 71

4.2.3 Adoption of High Yielding Varieties of Wheat . . 75
4.2.4 Adoption of Herbicides in the study districts . . . . 78
Farmers’ Perceptions of Yield Increasing

Technologies. .............................. 82
Characteristics of the Households ................. 87
4.4.1 Household Size and Labor Availability ........ 87
4.4.2 Age Categories of Household Heads .......... 89
4.4.3 Educational Levels of Household Heads ....... 89
4.4.4 Farm Sizes of Households .................. 91
Summary ................................... 93

viii

CHAPTER 5

CHAPTER 6

CHAPTER 7

5.1

5.2

5.3

6.1

6.2

6.3
6.4

7.1

TECHNOLOGY ADOPTION MODEL ............ 96
The Economics of Adoption and Conceptual Model . . . 96
5.1.1 The Linear Probability Model (LP) ........... 99
5.1.2 Logit and Probit Models ................... 100
5.1.3 Multinomial Logit and Probit Models ......... 103
5.1.4 The Tobit Model ........................ 105
Empirical Model Specification .................... 109
5.2.1 The Binomial Logit Model ................. 110
5.2.2 The Ordered Probit Model ................. 110
5.2.3 The Tobit Model ........................ 111
5.2.4 Dependent and Independent Variables ........ 113
Adopter Categories ........................... 120
5.3.1 Discriminant Analysis ..................... 120
EMPIRICAL RESULTS ........................ 123
Profitability Analysis of Technologies .............. 123
6.1.1 Partial Budget Analysis .................... 123
6.1.2 Marginal Analysis ........................ 133
6.1.3 Comparison of Alternative Wheat Production

Packages .............................. 139
Econometric Analysis .......................... 143
6.2.1 Binomial Logit Estimates .................. 144
6.2.2 Ordered Probit Analysis ................... 147
6.2.3 Tobit Analysis of Adoption and Use of

Fertilizer .............................. 150
6.2.4 Logit Analysis of Herbicide Adoption ......... 156
Discriminant Analysis and Adopter Categories ....... 158
Summary ................................... 164
6.4.1 Profitability of Production Technologies ....... 164
6.4.2 Econometric Analysis of Incidence and Intensity

of Adoption ............................ 165
6.4.3 Adoption Categories ...................... 169

SUMMARY, CONCLUSIONS AND IMPLICATIONS 170

Introduction ................................. 170
7.1.1 Background of the Study ................... 170

7.1.2 Research Objectives ..................... 171

7.13 Technology Adoption: An Overview .......... 172
7.2 Summary of Results ........................... 173
7.2.1 Household Characteristics .................. 173
7.2.2 Profitability of Production Technologies ....... 174
7.2.3 Analysis of Farmers’ Production Practices and
Technology Adoption ..................... 176
7.2.4 Econometric Analysis of Incidence, Intensity
and Determinants of ...................... 178
7.2.5 Adoption Categories ...................... 182
7.3 Policy Implications ............................ 183
7.3.1 Implications for Agricultural Research ........ 183
7.3.2 Implications for Agricultural Extension ........ 184
7.3.3 Implications for Agricultural Development
Policy ................................. 185
7.4 Limitations of the Study and Recommendation for
Further Research ............................. 186
APPENDIX I ................................................ 188
APPENDIX 11 ............................................... 204
REFERENCES ..... i ......................................... 215

Table 2.1

Table 2.2
Table 2.3

Table 2.4
Table 2.5
Table 3.1

Table 4.1

Table 4.2
Table 4.3
Table 4.4

Table 4.5
Table 4.6

Table 4.7
Table 4.8

Table 4.9

Table 4.10
Table 4.11

Table 4.12

LIST OF TABLES

Page
Ethiopia: Average Production, Yield and Area of Major Cereals,
1987-1989 ........................................ 14
Ethiopia: Number of Domestic Animals (’000) ............. 16

Ethiopia: Area, Yield and Production of Major Crops by

Sector, 1988-1989. ................................ 18
Ethiopia: Yields of Major Cereals by Types of Producers,
1987/88 (kg/ha) ................................... 19
Consumption of Fertilizer in Ethiopia Compared to that in
other Countries. ................................... 21
Peasant Associations and Household Heads Interviewed in
the Farmer Survey in Arssi Districts, 1990/91. ............. 52
Production, Consumption and Research Investment Indicators
for Selected Wheat Producing Countries in Sub Saharan
Africa, 1983-92 .................................... 58
Wheat Area, Production and Yield by Farm Types in
Ethiopia, 1986 ..................................... 59
Ethiopia: Cereal and Wheat Area and Production by Regions
(1987/88) ........................................ 60
Relationship Between Arssi Ecological Zones and Crops
Considered Most Important By Farmers ................. 63

Amount of Fertilizer Distributed in Arssi Districts 1975-1991. . 66
Amount of Improved Wheat Seed Distributed to Farmers in

Arssi Districts 1975-1991. ............................. 67
Use, Experience and Farmers’ Knowledge of Fertilizer in
Arssi Districts. ..................................... 72
Percent of Farmers in Different Fertilizer Application
Categories, 1990-1991. ............................... 73

Percent of Farmers and Fertilizer use Levels in Arssi, 1990-

1991. ........................................... 74
Adoption of Wheat Varieties, Average Area Planted, Seed
Rate and Yield in Arssi Districts, 1990-1991. .............. 75
Average Quantity of Wheat Harvested, Sold and Retained,
Farm Gate Prices and Seed Source in Arssi, 1990/91. ....... 78
Farmers’ Weed Control Practices in Arssi Districts, 1990/91 . . 81

xi

Table 4.13

Table 4.14
Table 4.15

Table 4.16
Table 4.17

Table 4.18
Table 4.19

Table 5.1
Table 5 .2
Table 6.1
Table 6.2
Table 6.3
Table 6.4
Table 6.5
Table 6.6
Table 6.7

Table 6.8
Table 6.9

Table 6.10
Table 6.11

Table 6.12

Farmers’ Assessment and Perceptions of Wheat Production
Technologies in Arssi ............................... 84
Farmers’ Criteria for Wheat Varietal Preferences in Arssi . . . . 85
Family Size Categories by percent of Households in Study

Districts of Arssi ................................... 88
Age Categories of Household Heads in Arssi Districts ....... 88
Levels of Education of Household Heads in Arssi Districts,
1990/ 91 .......................................... 90
Farm Size of Households in Arssi Districts, 1990/ 91 ........ 92
Number and Mean Values of Selected Household
Characteristics ................................... 93
List of Economic, Social and Institutional Variables Affecting
Adoption Decision ................................. 112
Technology Types, Econometric Models, Dependent and
Independent Variables Used. ........................ 115
Partial Budget Analysis of Wheat Production Technologies
Under Farmers’ Practices in Arssi, 1990/ 91 ............... 126
Adoption of Wheat Production Technologies in Arssi
Districts, 1990/91. .................................. 130
Partial Budget Analysis of Wheat Yield Under Selected
Fertilizer Levels in Arssi, 1990 ......................... 133
Marginal Analysis of Wheat Yields Under Different Fertilizer
Levels in Arssi, 1990/91 ............................. 135
Adoption, Fertilized Area and Rates of Fertilizer Use in
Arssi Districts, 1990 ................................ 138
Economic Optimum Rate of Fertilizer for Wheat Production
in Arssi, 1986-89. ................................... 139
Partial Budget and Marginal Analysis of Alternative
Production Packages for Wheat Farmers in Arssi ........... 142
Logit Estimates of Fertilizer Adoption in Arssi, 1990/ 91. ..... 145
Relative Effects of Significant Variables on Probability of
Fertilizer Adoption in Arssi, 1990/ 91 .................... 146
Ordered Probit Analysis for Different Rates of Fertilizer Use
in Arssi, 1990/91. .................................. 148
Tobit Analysis of Adoption and Intensity of Fertilizer Use in
Arssi, 1990/91. .................................... 152

Tobit Elasticity Decomposition for Changes in the
Explanatory Variables on Fertilizer Adoption and Intensity of
Use ............................................. 154

xii

 

 

Table 6.13 Logit Analysis of Herbicide Adoption in Arssi, 1990/ 91 ....... 157
Table 6.15 Discriminant Estimates for Adopter Categories in Arssi

Districts, 1990/91 .................................. 160
Table 6.16 Discriminant Model Prediction and Actual Classification of
Adopter Categories: Percent of Farmers ................. 162

xiii

Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6

Figure 7

LIST OF MAPS AND FIGURES

Page
Map of Administrative Regions of Ethiopia, 1991 .......... 12
Map of Arssi Region ................................ 49
Map of Wheat Producing Regions of Ethiopia ............. 61
Fertilizer Distribution to Smallholders in Arssi Districts,
1975-1991 ...................................... 68
Improved Wheat Seed Distribution to Smallholders in Arssi ‘
Districts 1975-1991 ................................. 69
Response Curve for wheat under selected fertilizer rates,
Arssi, 1990/91 ..................................... 136
Net Benefit Curve for Wheat and fertilizer Response in
Arssi, 1990/91. .................................... 137

xiv

CHAPTER 1

INTRODUCTION

1.1 Background

Agriculture occupies a pivotal position in the present and future of African
development. In Sub Saharan Africa, agriculture accounts for an overwhelming
share of the gross national product, employment, and exports. While some
improvement in the productivity of this sector has occurred in some countries, in
general agricultural production has been stagnant or declining, and per capita
food production has fallen in many countries. Also the Gross Domestic Product
(GDP) has stagnated in many Sub Saharan African countries and population
continues to grow rapidly (2.5 - 4% per annum) placing a burden on the
agricultural sector and its fragile resource base. Increasing the productivity of the
agricultural sector is critical in improving food security, providing more
employment in the rural areas, increasing exports for foreign exchange,
stimulating development in other sectors, and raising the overall standard of
living.

The food security of rural households can be improved by 1) increasing
food/cash crop production, 2) expanding rural employment opportunities, and/ or

3) greater household access to transfers. The first option, food production can be

2

increased by 1) area expansion, 2) increasing cropping intensity or 3) increasing
yields.

1.2 Problem Statement and Justification for the Study

Given the rapidly growing population in Africa and continued degradation
of the natural resources, the opportunity to increase production through area
expansion is very limited. Since irrigation in Africa is limited, the greatest
potential for increasing agricultural productivity is likely to come from increasing
yields on rainfed cultivated land. Yields can be increased through more intensive
application of labor, more intensive application of existing technologies or
adoption of new technologies. Farmers can adopt new inputs and new
management practices in their production. Adoption of agricultural production
technologies in developing countries is influenced by a wide range of economic,
social, physical and technical aspects of farnring. Recent adoption studies in Asia
(Duraisamy, 1989; Lin, 1991; Jansen et al., 1990), Africa (Polson and Spencer,
1991; Kebede et al., 1990; Adesina and Zinnah, 1993, Green and N’ogolala, 1993
and Hassan and Paid, 1993) have identified farmer and farm specific, technology
specific, institutional and policy variables and environmental factors to explain the
patterns and intensity of adoption. It is important to investigate the role and
contributions of these socioeconomic and institutional factors in the adoption of
agricultural technologies. Such studies are justified as their results will:

a) improve the efficiency of technology generation and transfer,

3

b) strengthen the links between institutions responsible for technology

development and policy formulation, and

c) help to measure and assess the impact of investments in technology

development and transfer (CIMMYT, 1992).

Agriculture in Ethiopia constitutes 45% of the GDP, provides employment
for 85% of its population and generates 90% of the foreign exchange. The
country has an area of 1.24 million square kilometers of which 69% is estimated
to be suitable for agriculture. But less than 15% is under crop production and
only 4% of the irrigable land is currently irrigated. Despite 25 years of research
and rural development projects, only 10% of the Ethiopian farmers are using
fertilizers and only 2% are planting improved seeds. The national average rate of
fertilizer application has not exceeded 20 kg per ha and the use of improved seeds
has not been more than 5 kg per ha (Debela and Gebre-Mariarn,1990). The
national average yield for the major cereals is not more than 1 ton per ha, yet, the
potential for attaining higher yields have been demonstrated by several on-farm
trials conducted by the Institute of Agricultural Research (IAR) and the Ministry
of Agriculture (MOA). This discussion helps explain why the study of technology
adoption is of high priority for researchers in Ethiopia. The main objective of this
study is to examine the social and economic factors associated with the adoption
of agricultural technologies by smallholders in one of the major wheat growing

regions of Southeastern Ethiopia.

4

1.3 Wheat Production and Consumption in Africa and Ethiopia

Sub-Saharan African countries have potential wheat-growing area of about
20 million ha and it is estimated that only 1.3 million hectares are under
cultivation. In most of the wheat growing countries of Sub Saharan Africa, wheat
is a more recent introduction which is produced using imported technologies, high
levels of purchased inputs and mechanized production practices (Morris and
Byerlee, 1993). Over half of this area (0.687 million ha.) is cultivated in Ethiopia,
which is the largest producer of wheat in the region. Ethiopia is also the center
of origin for durum wheat. Traditional production practices and lack of modern
inputs contribute to low wheat yield in Ethiopia, which is 1.3 t/ ha compared to 5.0
t/ha. yield of Zimbabwe. Three-fourths of all wheat currently consumed in Africa
is imported. Wheat production is lagging behind demand and at the same time
per capita wheat consumption is increasing. For the period 1961-90 wheat per
capita consumption increased at an annual rate of 4.2% (Morris and Byerlee,
1993). Imports increased by more than 600% in most Sub-Saharan African
countries (Morris, 1989).

The findings are consistent with wheat production and consumption
patterns in Ethiopia. Ethiopia was self-sufficient in wheat production in the
19605, but by 1980 it was producing only 70% of its requirements. In 1987 only
53% was produced locally while 47% of the wheat consumed was imported (both
purchased and food aid). In the 1980s, the growth rate of wheat per capita

consumption has been 3.5% per year. A recent IAR-CIMMYT study on wheat

5

production constraints in Ethiopia reveals that smallholders account for 76% of
the total wheat harvested (Beyene et al. 1989). In their review of wheat
consumption and production issues in Sub Saharan Africa, Morris and Byerlee
(1993) emphasized the need for more economic analysis of wheat production that
would help researchers and policy makers design appropriate research and
development priorities.

1.4 Research Objectives

Many adoption studies have viewed adoption of a technological practice in
isolation from other related practices. Interdependencies between different
technological practices have been ignored. Moreover the intensity of adoption of
technological practices has not been emphasized since adoption behavior has been
considered a discrete choice of adoption or non adoption. A number of studies
have examined the importance of household and environmental factors in the
adoption of agricultural technologies in LDCs. However, only a few studies have
included the intensity of adoption as a dependent variable in the adoption
decision model. Modelling the incidence of adoption does not provide adequate
information about possible production decisions farmers make. Previous studies
of adoption have identified the following factors influencing farmers’ decisions to
adopt new technologies: the availability of technical and financial assistance,
tenurial status, risk attitudes, farm size, education levels and age of head of
household and income. Factors that influence the incidence of adoption can be

different from those that influence the intensity of adoption. Several of the

6

adoption models have been under specified as they considered very few
explanatory variables.

In Ethiopia, 3 limited number of adoption studies (Tecle, 1973; Waktola,
1975; Aklilu, 1980 and Kebede et al., 1990) have been conducted. These studies
have not dealt with the relative importance of the various constraints listed above.
There have been very few commodity-related studies undertaken by local
economists.

The research questions addressed by this study are the following:

1. What are the farmers’ resources, household characteristics,

production practices and technologies used by smallholders growing

wheat in Southeastern Ethiopia?

2. How profitable and acceptable are the recommended technologies?
3. What is the relative importance of adoption determinants?
4. What are the socioeconomic characteristics of farmers who adOpt

new wheat production technologies, and how do advanced adopters
differ from moderate and low adopters?
5. What policy and institutional changes are necessary to increase the
rate of technology adoption to expand wheat output?
The general objective of this study is to examine the social, economic, and
institutional factors that influence the adoption of selected crop production
technologies in Arssi, a major wheat production region of the Southeastern

highlands of Ethiopia. Components of the production technology to be studied

7

include high-yielding varieties (HYVS), chemical fertilizers, herbicides, and

recommended crop husbandry practices for wheat.

1.4.1 Specific Objectives

The specific objectives of the dissertation are:

1.

2.

To present descriptive information on smallholders in Arssi region;

. To evaluate the economic profitability and acceptability of

recommended wheat production technologies;

To generate quantitative estimates of the factors influencing the
pattern and intensity of adoption using econometric models of logit,
probit and Tobit;

To identify constraints on the adoption of recommended
technologies for wheat;

To classify farmers into adopter categories based on adoption
pattern and socioeconomic characteristics;

To draw implications for agricultural research, extension, and policy

changes needed to overcome the constraints on adoption.

1.5 Organization of the Dissertation

The study is presented in seven chapters. Chapter 2 describes the problem

setting by discussing the roles of agriculture in the Ethiopian economy, production

constraints and the socio-political environment that created the existing agrarian

structure of smallholder, cooperative and state farms.

8

Chapter 3 discusses the conceptual framework, literature review on
adoption studies and the research design and methodology employed in the study.
Description of the survey areas, survey methods, sampling procedures,
questionnaire design and data collection procedures are outlined in the chapter.

Chapter 4 reviews the wheat subsector and presents descriptive statistics on
the characteristics of smallholder wheat producers in the study areas. The chapter
also reviews the experiences in technology development and transfer and
integrated rural development.

Chapter 5 examines the technology adOption model and the econometric
analytical techniques used. The applications of logit, probit and Tobit models,
model specification issues, selection of dependent and independent variables and
the application of discriminant analysis are discussed.

Chapter 6 presents the empirical results of the profitability analysis and the
econometric models developed in the previous chapter. In addition adoption
patterns, determinants, and relative importance of selected socio economic factors
are synthesized in this chapter.

Chapter 7 summarizes the results and draws conclusions and the policy
implications for national agricultural research, extension and development policy.
Finally the limitations of the study are discussed and further areas of research are

recommended.

CHAPTER 2

PROBLEM SETTING

2.1 Country Profile and the Agricultural Sector

Ethiopia is located in the "Horn of Africa " and is the second largest
country in Sub Saharan Africa with an area of 1.223 million sz and a
population of 52 million. Agriculture has always been the backbone of the
country and today about 86% of the population, engaged in agriculture, produce
40% of Ethiopia’s gross domestic product and 90% of the export earnings and
supply the raw material for the limited agro-industries. The manufacturing and
services sectors contribute 40% and 19%, respectively. The economy suffers from
weak infrastructure, heavy dependence on a single agricultural export of coffee, a
small industrial base, and shortage of skilled labor. Ethiopia ranks as one of the
poorest countries in the world, with a per capita GNP at $120 (World Bank,
1993).

The performance of Ethiopia’s agriculture deteriorated from an average
annual growth rate of 2.6% between 1965-75 to less than 1% between 1975-90.
The problem of food insecurity and subsequent famines has come to the world’s
attention because of the publicized 1984 drought that resulted in the loss of many
lives. Drought in Ethiopia, a result of limited rainfall, is enormously complicated

by economic, political, and other environmental factors.

10

In addition, with the current annual growth rate of 3.1%, population is
projected to reach 70 million by year 2000 and double by year 2015. Thus, food
production in general is not keeping pace with population growth. Arable land
under cultivation is said to be only 40% of the potential with only 100, 000 ha of a
possible 2.25 million ha of irrigable land developed. A complex set of factors
contribute to the low productivity of agriculture. Yet Ethiopia has the potential
to produce adequate food and increase its agricultural export earnings.

2.2 The Resource Endowments

2.2.1 MUM

Ethiopia is a large and diverse country, both culturally and agro-
ecologically. Traditional names have been assigned to zones on the basis of
altitude and temperature, e.g., kola, weinadega, dega and wurch. However, the
amount of rainfall and its periodicity are also important in defining zones (Hurni
1986). Often reference is made to various types of farming systems, for example
wheat-tef, or sorghum or coffee-based, which are determined by temperature,
altitude, rainfall, and soil conditions.

The highland areas cover the regions of Shewa, Gojam, southwestern Welo,
southern Gonder, and eastern Welega, which amount to 47% of the total area of
the country, where 74% of the population lives, 14% of the cultivated land, lying
between 1800 and 3000 m and receiving 950-1500 mm rainfall, produces 93% of
the country’s food. The northern Ethiopian highlands (southern Eritrea and

western Tigray) have 9% of the cultivated land but only 4% of the production.

11

The yields are less by 50% compared to those of the other regions because of soil
erosion, war, and drought. Sorghum is the most important crop. Southwest
Ethiopia, encompassing most of Kefa, Ilubabor, and western Welega, is at about
1500-2400 m and maize is the most important crop. The eastern highlands,
including Sidamo, Bale, Arssi, and Harerge, are characterized by having 950-1500
mm of rainfall, an average elevation of 1800 m, 16% of the cultivated land, and
19% of the total production. Sixty percent of Sidamo’s cultivated area is under
maize. Wheat and barley are the major crops of Arsi and Bale, and Harerge
specializes in sorghum (60% of production) and maize (23% of production)
(EMA 1988).

It is notable from these statistics the large extent to which food production in
general and various major crops specifically are regionalized and furthermore,

what a small area of the country produces the majority of the marketed surplus

food.1

 

1For details on an overview of Ethiopia’s Agriculture see Stroud and Mulugetta Mekuria
(1992).

 

 

 

 
  
  
   

W/

W
t ‘ ‘~.-_-: .
Dfihoullm
‘ Repuflkm
I W
W
/
Dire Dav:
_. il/ \ '\ _
ll .5 I

  
   

 

 

. SOMALI!)
_' ' if»? , ////////._...._. /,
F Sure ' ‘ ia,1991

i Ma of Adnrinistratrve Regions of Ethiop

 

13
2,22 Production Potentials and Limitations

Ethiopia’s agriculture is dominated by a number of cereal crops, which
account for about 69% of the calories in the Ethiopian diet, namely maize, tef
(Eragrostis tef), sorghum, barley, and wheat. Tef, a small indigenous cereal is the
most important crop in terms of area. Maize ranks first in total production and
yield (Table 2.1). Enset (Ensete edule) or false banana, is a major indigenous root
crop for approximately 8 million people in the central, southern and western
Ethiopia. These crops are augmented by a variety of pulses such as faba bean
(Vicia faba), chickpea (Cicer an’etinum), field pea (Pisum sativum), haricot bean
(Phaseolus vulgar-is), lentil (Lens esculenta) and rough pea (Lathyrus sativus). On
an area and production basis, faba beans, chickpeas, and field peas are ranked (in
descending order) as the most important pulse crops. Noug or niger seed
(Guizotia abyssinica) is the most important oilseed crOp, followed by linseed.

Generally, crop production has been stagnant over the last 15 years and
dropped in the drought years (1975 and 1984). There was an annual deficit of
around 350,000 tonnes of food crops during this period, which in 1987 increased
to 500,000 tonnes. Since 1979 with food aid has provided 8.5% of the calories
consumed (Faught 1988; ONCCP 1987). Food imports increased from an annual
average of 60,000 tonnes over the period from 1979 to 1983 to 738 800 tonnes
over the period from 1985 to 1987. In 1988 food aid reached 825,300 tonnes
(UNDP/World Bank 1989).

14
Table 2.1 Ethiopia: Average Production, Yield and Area of Major Cereals, 1987-

1989

 

 

 

 

 

 

 

Production Area Yield Average Yield

('000 t) (‘000 ha) (t/ha) Rank
Crop Type 1987 1988 1989 1987 1988 1989 1987 1988 1989 1987-89
Tet 977 923 1201 1212 1161 1354 0.81 0.79 0.89 0.83 5
Barley 834 758 726 745 615 639 1.12 1.23 1.14 1.61 3
Wnoat 585 509 635 526 508 531 1.11 1.17 1.19 1.15 4
Maize 1467 1504 1364 887 783 727 1.65 1.92 1.87 1.81 1
Sorghum 922 846 911 715 726 688 1.29 1.16 1.32 1.25 2

 

 

 

 

Euros: Price Studies and Policy Institute, July 1989

 

Cash crops for Ethiopia include coffee, tobacco, cotton, and other fibers
(sisal, enset, and kenaf). For small farmers, coffee, pepper (Capsicum frutescens),
chat (Catha edulis, a stimulant), and some vegetables are the main cash crops.
Arabica coffee is produced in Kefa, Ilubabor, Welega, Gamo Gofa, Sidamo,
Shewa, and Harerge regions. State farms have about 8,000 ha out of a total of
450,000 ha of cultivated coffee and 150,000 ha of forest coffee. About 20% of
government revenue comes from coffee and coffee revenue amounts to 4%-5% of
the GDP. Forty percent of coffee is consumed in the country and the rest is
exported, accounting for 60% of the foreign exchange earnings. Ethiopia is the
sixth major producer and tenth major exporter of coffee in the world (EMA
1988). Coffee produced a moderately good income for farmers in the 19705 and

19805 until 1989/90 when the international coffee cartel and the quota system

15

collapsed, which resulted in a more competitive market and lower coffee price.
Foreign exchange earnings from other export crops including pulses, oilseeds,
oilcake, fruits, and vegetables has also dropped in recent years.

Natural resources have been subject to degradation for centuries, a
phenomenon particularly well documented in the highland areas, and this is
negatively affecting production and the quality of life. Erosion is decreasing soil
depth, water-holding capacity, and fertility and it is also increasing the frequency
of drought in marginal areas. Deforestation has taken place at a dramatic
rate—200,000 ha per year. The area under forests is 4%, down from 16 percent in
the 19505. Fuel is a problem for most families, who resort to using dung, weeds,
and crop residues, practices that further erode the soil fertility and organic matter
base. There have been many tree planting and soil and water conservation
projects in which farmers and their families take part; however, they are seldom
left in charge of these resources and the lack of ownership discourages judicious
management (Faught 1988, Stroud 1989).

Livestock (cattle, goats, sheep, donkeys, poultry, horses, and camels) are also
an important component of all farming systems, although concentrated in the
highlands. More numerous than in any other country in Africa, livestock numbers
are increasing gradually over time (1970—1988) (Table 2.2). Livestock makes up
17% of the total annual export revenue including hides and skins (sheep, goats,
and cattle) and live animals (cattle and goats). Private traders export 57%; the

rest is exported through government corporations (EMA 1988). Livestock

16
productivity indices rank one-third lower than the tropical African average (World

Bank 1987).

Table 2.2 Ethiopia: Number of Domestic Animals (’000)

—

 

 

 

 

 

 

Ammal 1979/1981 1986 1987 1988
Cattle 26,000 30,000 30,000 31 .000
Sheep 23,250 23,000 23,200 23,400
Goats 17,177 17,000 17,300 17,500
Camels 980 1,000 1 .500 1 ,060

 

 

 

Scurce: FAO estimate, FAO, 1988.

2.3 Structure of the Agricultural Sector
2.3.1 Laod fl'onoro and Poasant Qrganizatioos

The 1975 land reform abolished the old tenancy system including all private
ownership of land, without compensation, making land collectively owned by all
(Provisional Military Government of Ethiopia, 1975). The land was divided into
800-ha units called peasant associations(PA), which served as the basis for
administering land-use directives from the government, administering and
conserving public property, and settling land cases. The government felt that PA

formation would give people more self-administration based on a group approach

17
and would eliminate fragmented land holdings (EMA 1988). By 1986, 20, 157

PAS, including 5, 594, 000 households or 65% of the rural population, had been
formed Although the original policy was that a family could farm up to 10 ha, in
fact most holdings are only 1-2 ha. The peasants’ new relationship to the land
has proved to be neither fixed nor secure: peasants can be moved to
accommodate expanding cooperatives, state farms, the creation of villages, or for
other reasons (MOA, 1988).

Smallholders dominate agricultural production in terms of the population
involved, the amount produced, and the area cultivated, accounting for over 94%
of each category in 1983. Producer cooperatives which are collective farms
ranging in size from a few to several hundred families, accounted for 1.8 % of
cultivated area for the same period. State farms, which are large, usually
mechanized operations managed by government agencies, accounted for 3.5% of
cultivated area. By 1988-89 smallholders still accounted for over 91% of the
cultivated area; and the share of the cooperatives had risen to 6-7% while state
farms had declined to 3 - 2.7% (Table 2.3).

Table 2.4 illustrates that for major food crops, smallholders have produced
higher yields than the cooperatives, despite the fact that the supply of inputs and
market pricing structure for inputs and outputs is less favorable for smallholder
sector. In some cases (tef and sorghum) yields of the private farms exceed those

of the state farms.

18

Table 2.3 Ethiopia: Area, Yield and Production of Major Crops by Sector, 1988-

1989.

 

 

 

 

 

Sector 1988 1989
Area Yield Production Area Yield Production
('000 ha) % (t/ha) ('000 t) % (’000 ha) % (t/ha) (’000 t) %
Smallholder 6492 90 0.98 6384 87 6676 90.8 1.03 6904 88.8
Cooperatives 494 7 1.11 549 7.5 479 6.5 1.06 508 6.5
State Farms 208 3 1.99 415 5.5 197 2.7 1.83 361 4.7

 

 

Source: Debela and Gebre-Mariam, 1990

Table 2.4 Ethiopia: Yields of Major Cereals by Types of Producers, 1987/88

 

 

 

 

 

(kg/ ha)
Crop Smallholders Producers“ cooperatives State Farms
Maize 1923 1659 3032
Barely 1234 977 1219
Wheat 1170 1015 1429
Sorghum 1 163 939 973
Tef 795 738 172

 

 

 

Source: CSA19898

19

In 1979 government policy encouraged the formation of producers’
cooperatives with a target to be one per peasant association. A producers’
cooperative is similar to a collective rather than a cooperative in that the land is
pooled and the members’ income is to be based on the share of labor and
resources contributed. Individual households retain 0.1-0.2 ha for their own use.
The government’s aim in 1979 was to have half of the cultivated land organized
into producers’ cooperatives by 1994. Incentives were offered to encourage
farmers to join this system: any land in a peasant association could be allocated to
a producers’ cooperative; lower taxes were to be paid; improved seed, fertilizer,
credit, and extension services were to be readily available; free labor could be
obtained from PAS during peak periods; the Agricultural Marketing Cooperation
(AMC) would purchase produce for higher prices; and there would be priority
access to consumer goods, training, and building materials. By 1989 there were
3,741 producers’ cooperatives involving 321, 324 households or about 4% of all
rural households. The government introduced service cooperatives with the aim
of selling farm inputs, purchase locally produced cereals and pulses, provide loans
at fair interest rates, provide storage and savings services, supply basic consumer

goods, educate members in socialist philosophy, supply tractor hire services,

20

collect self-help contributions, provide flour milling services, and promote cottage
industries.
232 WWW

Input use (fertilizers, improved varieties, pesticides, veterinary drugs) by
smallholders is generally restricted for a number of reasons. Distribution has
been through the Agriculture Input Supply Corporation, which receives an
estimation of input requirements and arranges for seed production, importation,
credit, and distribution. Distribution has often been inefficient and late; inputs in
general are expensive relative to produce market prices, and in many instances
have not been economic (Franzel et a1. 1989). In addition inputs have been
allocated to state farms and cooperatives on a priority basis; credit is scarce and is
available only through service cooperatives; inputs are not manufactured locally
but require foreign exchange, which is limited, or rely on irregular donations.
Total fertilizer use has been estimated to be less than 4 kg/ha in Ethiopia.
Ethiopia consumes much less fertilizer than many other countries (Table 2.5).

From 1974 to 1980 the cost and use of fertilizer increased moderately.
However, the price doubled over the following two years and fertilizer use
dropped drastically in response. In the 1981 and 1982 crop seasons, the AMC
prices for crops also dropped. From 1979 to 1985, 64% of the improved seed was
allocated to state farms, 15% to settlement areas, and 19% to smallholders. Only

2% of the farmers use improved seed. Seed costs up to five times more than the

21
Table 2.5 Consumption of Fertilizer in Ethiopia Compared to that in other

 

Countries. .
—
Country Nitrogen fertilizer Total amount
(kg/ha) (kg/ha)
Ethiopia 1.3 3.5
Somalia 1.7 2.3
Kenya 13.4 37.6
USSR 44.3 98.8
USA 53.2 104.4

 

Source: Awoke and Hailu 1986

crop’s marketed price (Faught 1987). In 1986, 63% of the fertilizer2 went to the
peasant sector, 25% to state farms, and the rest to other government
organizations. Fertilizer supplied to the peasant sector is directed to surplus-
producing areas.

Institutional credit remains insufficient as funds are short and priority is given
to the State farms. Only legally recognized service cooperative can supply loans to
peasant associations on behalf of Agricultural and Industrial Development Bank
(AIDB). Most of the credit available is used for purchasing fertilizer (15% of the
principal; 9.5%-11% interest must be paid in cash in advance and the rest repaid
within a 9-month period). Loans are also used for purchase of oxen, equipment,

and infrastructure development such as stores. One restrictive repayment policy

 

2 In 1986 fertilizer use went up 72.8% and in 1987 another 13% from the previous year These
increases may have been due to increased supplies from donor agencies in response to the 1984—85
drought (MOA 1988c).

22

that was adopted by the AIDB and service cooperatives is that if a peasant
association does not repay its loans, all farmers in the association will be ineligible
to receive additional credit the following year. Smallholders also borrow money
from money lenders or relatives.

Taxes include a direct tax of about 20 birr per year per family and indirect
taxes where farmers are required to contribute to national or local campaigns. In
some areas farmers are required to provide labor for public works (e.g., road
building, building schools, farming militia men’s farms, which can amount to
several days per week).

2.4 Agricultural Development Policy : Revisited

Governments in many countries intervene in the management of the national
economy in varying degrees. There is common agreement that the government
has an important role to play in regulation and/ or taxation and provision of public
goods. The Governments of Ethiopia in the past and present had and have
adopted national development and central planning as a means to direct the
economicc development of the country.

2.4.1 Ago'oiltoral Dovelooment Poligy; Pro 1974 Poriod

The Imperial Government of Ethiopia under the late Emperor Haile Selassie
I lasted for almost half a century. The Imperial Government adopted three
successive Five Year Plans (1957-62, 1963-68, and 1968-73). The first two plans
emphasized the infrastructure and industrial development, respectively. Urban

and rural population growth out stripped food production and the efforts in

23

industrial development did not yield the intended results. The Third Five Year
Plan had to explicitly emphasize agricultural development. During the plan
period comprehensive rural development projects were established in Chilalo,
Wolayita and Ada districts.

The Plan’s strategy to promote smallholder agriculture was labelled a
"package approac It aimed to concentrate financial and human resources on
the development of a few promising highland areas. Essentially an integrated
rural development strategy, the package approach argued for combined
investment in agricultural research and extension with initiatives to improve the
distribution of seeds and fertilizer, provide credit, develop market facilities, diffuse
better farm implements, expand storage facilities, promote rural health, and raise
frmctional literacy3. The projects were found too expensive to duplicate in other
parts of the country. In 19705, they were followed by the Minimum Package
Programs. The objectives of the package approach were to provide smallholders
with improved seeds, fertilizer and farm implements. The government also gave
incentives to the emerging commercial farms growing cash and export crops. The
agricultural sector grew at a rate of 2.1% per annum during 1965-73, barely
keeping abreast of population growth.

Based on the experiences of the three Plans a Fourth Five Year plan (1974-

79) was prepared and it incorporated the lessons from the integrated rural

 

3 For a detailed review of agricultural development policy in the 19605 and evaluation of
Integrated Rural Development in Ethiopia see Cohen (1987).

24

development and minimum package programs. The strategy clearly identified the
need to overcome the following constraints facing agricultural development in
Ethiopia: insufficient government finance, lack of trained technical and
administrative manpower, inaccessibility of much of the subsistence farming areas,
shortage of technical and economic data, and lack of proven agricultural
technologies (Cohen, 1987). When the 1974 revolution erupted the Fourth Five
Year plan was abandoned. During the Imperial era progress was made in
building the institutional capacity to support a modernizing smallholder sector.
The MOA was established as an organization, and agricultural training and
research institutions were established.
2.4.2 Agrarian Sooialism

The Provisional Military Government that replaced the Imperial regime in
September 1974, declared socialism as its guiding ideology. In 1975 many key
financial and manufacturing enterprises were nationalized. Rural and urban land
reforms were put into effect and the first references were made to ‘villagization’
and ‘agrarian socialism’. Most peasant farmers have never had secure use of their
farms under either the old system or the new. Between 1976 and 1981 central
planning and state control over the economy were strengthened, in particular the
control over production, distribution, and exchange. From the 1974 revolution
until 1990, the Ethiopian government followed a policy of agrarian socialism, an
approach based on collective ownership of the means of production, villagization,

resettlement, group farming and state farms, and government control of rural

25

marketing. The arguments for this approach included wider sharing of the rural
economy and political power, more equal access to land; more efficient use of
labor for agricultural production and rural activities, a reduction of rural-urban
migration, more efficient implementation of policies for development with greater
control, and economies of scale in agricultural production and marketing (Cohen
and Isaksson 1988).

The Development of State Farms: In 1979 the government formed the Ministry of
State Farm Development (MSFD). The purpose of the State Farms was to supply
urban areas and government organizations with food and to provide some
commodities for export. State farms were allocated 222, 000 ha in 1983, 3.5% of
the agricultural land; by 1988 cultivated area under state farms was 214, 000 ha.
The average size of a single state farm is 15, 000 ha. Tractors, improved varieties,
chemical fertilizers, herbicides, and insecticides are used. They absorbed 40% of
the government expenditure on agriculture during 1980-1985 period but
contributed only 4%-5% to the total production. In 1983 they received 76% of
the fertilizer, 95% of the improved seed, and 80% of the credit. The state farms
have increased the country’s production of cereals, coffee, sugar, cotton, and
tobacco but have had problems with mechanization, uneconomic use of inputs,
labor shortages, and financial viability (Cohen and Isaksson 1988). Most of the
state farms operated at a loss.

Villagization and Resettlement: The objectives of Villagization were to conserve

natural resources by promoting a better land-use plan, enhance delivery of

26

extension services, give greater access to public services, and strengthen security
and self-defense. Each village has between 200 to 300 households, with 1000 m2
(0.10 ha) allocated for each family. Villagization was started in 1986 as Phase 1
in Shewa, Arsi, Harerge, and to a limited extent in Gojam, Welega, Kefa, and
Ilubabor. The campaign was easier to implement in areas where (1) annual crops
rather than perennial crops had been common, (2) houses had been constructed
of wood, mud, and thatch but not stones, and (3) farmers had rights to the land
for centuries, as compared to areas where land had been owned by landlords.
Villagization brought further disruption. For most of the farmers, walking to and
from their fields takes a considerable amount of time. Grazing areas are often
too far from home. Bringing hundreds of families together without adequate
health facilities has led to frequent outbreaks of human and animal disease. A
large number of farm animals died immediately after moving to the new villages.
Most farmers had to move their livestock back to their old villages to escape the
plagues. Increased livestock and wildlife attacks on crops because of the distance
of the fields from the house, overgrazing near the village leading to more erosion,
and more pressure on water supplies and tree resources are the potential
problems of the Villagization effort. Moreover, the government has lacked
resources to provide necessary services such as water.

Resettlement was initially recommended by the World Bank as a solution to
overcrowded areas where the resource base could no longer support the

population. The underlying reasons for resettlement were population growth,

27

exploitative farming practices, energy shortages, overgrazing, stagnating yields,
limited off-farm employment, and low economic growth (World Bank 1987). In
1984 the government planned to resettle 1.5 million people affected by the
drought to areas that were supposed to be more productive; 800, 000 were moved
in a hurried fashion, without the necessary amount of time and effort put into
planning. Most of the settlers were moved from the northern highlands to areas
of low population density in the western lowlands.

Recent review of Ethiopia’s agrarian experience clearly indicate that the 1975
radical land redistribution program was Successful as the poor and underprivileged
acquired land. However, the subsequent socialist policies launched by
administrative fiat, often hurriedly or secretively, and without consultation with
peasants did not promote the development of smallholder agriculture. The
reform transferred all land to public ownership, prohibited all forms of private
property which gave rise to insecurity of holdings and replaced landlords with the
state. It ultimately transformed rural Ethiopia into a society of self-laboring
peasants (Dessalegn, 1992).

2.4.3 Ago'ooltoral Boligy in the Transition Period

In 1990 the government of Mengistu Haile Mariam initiated reform measures
to liberalize the economy in general and the agricultural sector in particular.
Agricultural trade restrictions, grain delivery quotas, fixed prices for the major
crops were lifted. Producers’ cooperatives members were allowed to make

decisions on either to continue to farm collectively or abandon the cooperatives.

28

Almost all decided to dissolve the cooperatives. The government also attempted
to encourage the private sector to invest in commercial agriculture. These
measures did not have any significant impact on the already troubled economy.
In May 1991, the current the Transitional Government of Ethiopia (TGE) took
power. It then introduced a New Economic Policy of the Transitional Period in
November, 1991. In terms of the agricultural sector, the new policy included the
1990 reform measures introduced by the defunct regime. Recently the TGE
adopted macroeconomic policy reforms and devalued the currency from 2.07
Ethiopian Birr(EB) to 5.00 Birr to a US dollar. Other World Bank prescriptions
for implementing the structural adjustment program are being adopted by the
government. However, the government does not intend to privatize land which is
still public property by decree. To date the TGE owns the State farms and
agricultural input distributing enterprises. Appropriate agricultural policy
formulation would be the most important challenge for an in-coming democratic
government of Ethiopia.

2.5 Agricultural Technology Generation and Transfer

2.5.1 Agrjggltural Research in Ethiopia: Evolution and Dovelooment

The Institute of Agricultural Research (IAR), established in 1966 as a semi-
autonomous public institution, is the national organization mandated to
coordinate and execute agricultural research in EthiOpia. The Alemaya University
of Agriculture (AUA), Addis Ababa University and different units of the

Ministries of Agriculture, State Farms, Coffee and Tea Development also

29

undertake agricultural research. The IAR has established 28 research centers and
sub-centers in the different agro-ecological zones of the country.

In the last 25 years IAR adopted different approaches to organizing its
research programs. Initially expatriate staff were assigned to conducted their own
experiments but gradually commodity programs evolved. Because of the need to
strengthen the different agricultural disciplines, departmental approaches
followed. Departments of field crops, coffee, horticulture, soil science, animal
sciences, agricultural engineering and food science, and agricultural economics
were established each having a department head coordinating the program
nationally and based at an appropriate research center. Only the agricultural
economics program is coordinated from headquarters.

Recent reorganizations led to a two-pronged approach to the generation and
transfer of technologies i.e., commodity and zonal/regional research approach.
There are eight zonal research programs with one or more research centers in
each zone. With this set- up, the centers are responsible for tackling the
researchable problems of a particular agro-ecological region. The research
program of each center is generally grouped into three major categories: crop
production, animal production and natural resources management. All supporting
disciplines participate in the research projects.

The national commodity programs are designed to tackle the production
problems of selected crops and livestock that are accorded top priority in the

national agricultural development objectives: achieving food self-sufficiency,

30

supplying raw materials for industry and earning foreign exchange. Research
problems are identified by a multidiscplinary team consisting of breeders,
agronomists, crop protection specialists, soil scientists, agricultural economists and
food science experts. By 1991 the IAR has identified 23 commodity research
programs to work on selected crops and livestock. The wheat, maize, sorghum,
cotton, coffee, highland and lowland pulses and oil crops programs are
operational as commodity research programs. The dairy, sheep, beef and animal
nutrition and feeds programs are also functional. Farm implements, irrigation,
soil and water management research programs are being strengthened. Core and
cooperating research centers for each commodity have been designated. The
implications of such an expansion in terms of financial and human resources and
level of research capability has to be carefully evaluated. The size of research
centers and program content is a key issue currently under consideration by the
institute and funding agencies.

To build a sustainable NARS a set of necessary and sufficient conditions have
to be met. The IAR was able to survive and attain its current status because most
of the conditions were met. These conditions were:

1. Continuous government funding and support,

2. Continuous autonomy and de-politicized environment

3. Strong scientific leadership for 18 of the 25 years

4. Investment in and steady supply of human capital (continuous graduate

training opportunities despite staff attrition)

31

5. Articulated and defined requests for international support, including well
prepared terms of reference for expatriate stafi role 9
6. Necessary organizational developments and changes without disruption of
on going programs (rarely the case in other public institutions).
7. Control of the research agenda (i.e., not donor driven).
The IAR had a total of 592 research personnel (25 PhD, 87 MS., 6 DVM.,
191 B.Sc. and 284 Diploma holders and additional 63 on study leave as. of 1990.)
(Debela and Gebre-Mariam, 1990).

The recent World Bank comparison of Asian and African NARS indicated
that no African NARS is to the standard of Asian NARS. The study however,
identified that of the 47 NARSS in Africa only 9 meet the minimum standard.
Based on the Bank’s rating score of 0 to 5, a score of 3 is considered to meet the
minimum acceptable standard and the IAR was rated to meet the standard
(Cleaver, 1993).4

The contributions of any national research service are measured by how much
its outputs increase the productivity and production of the agricultural sector.

5

The IAR has released a number of crop varieties , crop management practices,

improved breeds of animals, farm implements and other kinds of improved

 

4For details on evaluation of African Agricultural research institutions see Cleaver (1993).
According to the evaluation the nine NARSs in Sub Saharan Africa that meet the minimum
standards are: Cote d’Ivoire, Ethiopia, Ghana, Kenya, Madagascar, Nigeria, Senegal and Zimbabwe.
Mauritius 15 rated above average.

5 The IAR released 55 varieties of cereals (10 wheat varieties) and oil crops during 1970-1986

period. Disease resistant coffee varieties were also selected and distributed to coffee state farms and
farmers (IAR, 1988).

32

agricultural technologies. It has been both a producing and borrowing (creating
and evaluating capability) NARS having strong collaboration with the IARC56.
25.2 Warm

The MOA and the Ministry of Coffee and Tea Development operate
extension systems to serve smallholders and PCs. An extensionist’s role in
Ethiopia is to demonstrate technology, distribute inputs, carry out soil and water
conservation projects, villagize farmers, and promote afforestation, among other
duties. Although there have been extension-strengthening projects in the past,
relatively frequent reorganization, little in-service training of development agents
(DAs—those closest to the farmer), frequent transfers, and few incentives including
lack of pay raises and transportation to do their job, have resulted in a generally
unmotivated staff.

With assistance from the World Bank, the Government introduced a training-
and-visit system (T&V) approach to extension in 1986. Under this approach
subject matter specialists in agronomy, soil and water conservation, crop
protection, socioeconomics, cooperatives, livestock, and home economics are
located at district centers. The ideal is to have one DA for each service
cooperative (one for every 1,600 households, at the least). Under the T&V
system extensionists would consistently visit farmers on a fortnightly schedule and
receive in-service training regularly. However, extension agents have not had

access to good, appropriate information and do not generally have the training or

 

6On the experiences of building sustainable NARS in Ethiopia, see Mekuria (1992).

33

the ability to adapt information to the farmer’s situation. Many have no
background in farming; coming from more urban backgrounds, they lack empathy
with the farmer, and their formal training does not give them appropriate
communication and technical skills. Furthermore, farmers often'view the
extension agent with distrust and suspicion because they want to minimize
government interference in their lives. However, with some encouragement and
training, extension agents could no doubt become much more sensitive and useful
to the farmer.
2.5.3 Ago'ooltural Research and Extension Linkage

Historically, there has been a weak linkage between extension and
agricultural research, which is coordinated by the Institute for Agricultural
Research. For example, extensionists have not been formally involved in IAR’S
on-farm research activities. The number of forums or appropriate publications
where research information is passed to the extension worker has been limited.
There have been few attempts by extensionists and researchers to collect farmer
feedback concerning newly introduced technologies. In 1986 the Research and
Extension liaison Committee (RELC) was formed with members representing
both IAR and MOA. Although active in some sites, it has not been effective in
others. With a new reorganization of the regions and MOA, perhaps RELC will
have more of a role in linking the researcher, the extensionist, and the farmer.
Farming practices, in Spite , of extension advice, have remained traditional and

relatively unchanged over past years. Many argue that there has been little

34

incentive to change, given the policy environment and land tenure structure both
before and after the revolution.

Farmersmay plow the land anywhere from two to six times before seeding
depending upon the crop to be sown, the soil type, and rainfall. In the highland
areas, tef is given priority, and because of its small seed and stature needs a
carefully prepared seedbed. Cereal crops, especially tef, are given priority for
receiving inputs, e.g., fertilizer, pesticides, and labor. Weeding is usually done by
hand, less frequently with a hoe, or by slashing. Herbicides have not been readily
available, although the farmer demand in some areas has been high, because of
the government’s decision to concentrate foreign exchange allocation on items
other than herbicides, for which it feels hand labor can be substituted. As a
result, weeding is one of the farmer’s major problems in many crops. Many crops
are broadcast rather than row seeded, a practice that the farmer feels justified in
using because it is faster and because inputs such as fertilizer are generally not
used (row planting facilitates application of inputs). Most farmers do not use

intercropping and farming generally tends to be extensive rather than intensive
(Stroud and Mekuria, 1992).
2.5.4 Exporienoes in Integrated Rural Development
In 1967 the first integrated rural development project in the country was
launched in Chilalo district (Arssi region) known as the Chilalo Agricultural
Development Unit (CADU). The project was designed to stimulate agricultural

change by concentrating resources in high potential areas to provide all the

35

necessary ingredients in an integrated manner. Chilalo was chosen (from several
proposed sites) because of its favorable farming conditions and availability of
transport and marketing facilities. CADU was jointly financed by the Swedish and
the Ethiopian governments. The main objectives of CADU, as stipulated in the
agreement, were: 1) to bring about economic and social development in the
district; 2) to give the local population an increased awareness of and
responsibility for development work; 3) to verify methods of agricultural
development; and 4) to train staff not only for the project itself but for other
similar efforts. It was agreed to introduce only intermediate technologies
appropriate for small farmers.

CADU provided a combination of different services that were believed to
enhance productivity. Its primary functions were: 1) to carry out adaptive
research on crops and livestock; 2) to supply inputs; 3) to disseminate proven
technologies; 4) to provide credit and marketing services; 5) to promote soil and
water conservation and forestry; 6) to develop improved farm implements; and 7)
to improve health services.

The amount of resources required to undertake all these activities was clearly
enormous and the project was heavily dependent on the generous assistance of the
Swedish government. In 1976, the operation area of CADU was expanded to
cover the whole of Arssi, it was hence renamed the Arssi Rural Development
Unit (ARDU). The project continued to assist 50% of the total fertilizer

distribution in the peasant sector of the country. Between 1968 and 1973, cereal

36

yields doubled for farmers using fertilizer and improved seeds. The gain in wheat
yield was particularly significant. It increased from an average of 7 quintals / ha. in
1966 to 16.3 (232%) in Chilalo, 13.3 (190%) in Arba-Gugu and 12.8 (183%) in
Ticho m (district) in 1981. On the project’s own seed farm, wheat and
barley yields were as high as 40 quintals in 1981. The potential for further
increase is evident from the wide yield gap between the seed farm and the
peasant farms, nearly three times higher on the former. With improved
management practices, the peasant sector could easily double or even triple its
output.

Arssi has increasingly become a major regional contributor to the national
marketed surplus of cereals. In 1986, the share of Arssi in the total planned
cereal purchase of AMC was 31%. The AMC used to impose a fixed amount of
grain be supplied to it by each region and each farm household had to deliver the
required amount. A total of 1.78 million quintals was imposed on the farmers of
the region, the highest per capita quota share in the country. The rising grain
quota sharply contrasted with the fall in yield levels.

New innovation in the livestock sector took the form of upgrading local
breeds through crossbreeding with Friesian and Jersey breeds. It is estimated that
more than 1,750 crossbred heifers were sold to farmers between 1967 and 1981.
The average milk yield of a crossbred cow is reported to be 6 to 10 times more
than the average local cow. Few other parts of the country have had access to

such improvement opportunity. Significant progress was also made in the

37

construction of rural roads and water bore holes. As of 1981 a total of 330 kms
of roads were built and about 126 bore holes were dug to supply water for home
and animal consumption. CADU/ARDU’s attempt to introduce new farm
implements was not, however, successful. Few of the improved ploughs, barrows
and ox-drawn carts were adopted. The implements were heavy (for being pulled
by oxen), costly to repair and expensive to buy. Further research was not
undertaken to make the implements lighter and cheaper. The contribution of the
project in forestry research and soil and water conservation was also minimal7.
2.5.5 r 'n in Pr ' A ivi ie

A major problem in reaching small farmers before 1975 was the land tenure
system. Many of the small farmers were tenants of absentee landlords. The share
of the landowners varied from one-third to half of the total produce. There was a
further levy of one-tenth of the value of output, free labor and various kinds of
tributes by the landlords. The tenants were unable to benefit from the increasing
yields, resulting from the use of new inputs, as land rents increased in many
places. The proportion of tenants in Chilalo was estimated at about 52% of all
farmers. For many peasants, the threat from commercial farming was more
serious than the rise in land rents. Several small farmers were forced to leave
their land as many landlords, motivated by the CADU’s demonstration of

technologies and government subsidy for agricultural machinery, started to buy

 

7Cohen (1987) gives an excellent analysis of the evolution, achievements and the constraints of
CADU-ARDU.

38

tractors or sell/rent their land to commercial farmers. It is estimated that as
many as 5,000 tenant households or 20,000 people were evicted from Chilalo
during the period from 1968 to 1974. The absence of written leases meant that
the tenants had no protection against arbitrary sudden eviction. Government
officials who were landowners themselves offered little help to improve the
condition of the small farmers.

Another factor that adversely affected the use of new technology in the early
years of CADU/ARDU was low and fluctuating grain prices. With less than 10%
of the population living in the urban areas, the demand for grains failed to keep
up with the increase in production. The project attempted to raise and stabilize
prices through its marketing division. But, the volume of grain handled was too
small to make a significant contribution.

The land reform legislation of 1975 resolved the problem of tenancy in Arssi.
Tenancy was abolished and peasants were freed from all kinds of obligation to the
landlords. Many evicted tenants were able to return to their old villages and
farmland. But the reform was not accompanied by improved provision of farm
inputs and better output prices. The emphasis of CADU/ARDU changed from
technical farming problems to activism and social mobilization. In 1986, about
20% of the farmers in the region were members of cooperatives. The proportion
is nearly six times the national average. Research and technical innovation were
neglected. The resources (staff, vehicle and finance) of the project were used to

promote cooperatives and assist political and fund-raising campaigns, and

39

collection of compulsory quota and national contributions. Most of the
experienced professionals left the project.

The number of varieties released by the research department of
CADU/ARDU declined from ten during the period of 1967-74 to only four
during 1975-83. The number of barley varieties released fell from six to just one
during the same period. Most of the new varieties were abandoned in the early
19805. CADU/ARDU was unable to replace old seeds that had outlived their
genetic potential and became susceptible to, disease.

The contribution of CADU/ARDU is also undermined by unfavorable input-
output price relations. The introduction of compulsory delivery quotas and price
control has left the farmers without incentive. A quintal of fertilizer costs the
equivalent of 2.7 quintals of wheat in the 19805, compared to 1.5 quintals in
1971/72.

The marketing services of the project was taken over by the Agricultural
Marketing Corporation (AMC) in 1976. In order to ensure complete monopoly of
the grain trade by the AMC, all grain merchants in Arssi were banned and
movement of grain by individuals made illegal. With no large markets to
generate sufficient demand, the open market price in the region has been
depressed. According to a recent survey, which covered three major grain
supplying administrative regions in 1985, AMC quotas are exceptionally high in

Arssi.

4o

Arssi is one of the most fertile regions and one of the least drought-affected
areas in the country. The inhabitants of the region have benefited from the
Green Revolution technologies introduced by CADU/ARDU. The region is a
major supplier of grain to the national market. But development has ceased in
recent years because of inadequate resource allocation,low prices, excessive
quotas, and disruptions arising from the rapid pace of agrarian transformation.

2.6 Summary

Chapter 2 reviewed the agricultural sector and presented the production
potential in the different regions of Ethiopia. The agrarian structure resulting
from three different policy eras are discussed. The three major agricultural policy
eras are: the Imperial period until 1974, the agrarian socialism period under the
Marxist military regime (1974-91) and the current transitional period (1991-
present) under the Transitional Government.

Efforts to build a national capacity in technology generation and transfer by
the IAR and the extension service are also discussed. The problems related to
research and extension linkage and the institutional arrangements made to rectify
these problems are briefly presented.

Ethiopia has experienced different kinds of agricultural development
strategies with varying degrees of success and failure. The CADU-ARDU
experience in integrated rural development, technology development and transfer

are reviewed this chapter. In spite of the socio-political problems of the Ethiopian

41

economy, the experiences of the integrated rural development project have been
positive.

In light of this, the study examines technology adoption issues related to
wheat production in Arssi, the home of the first integrated rural development
project in Ethiopia. The lessons learned from the project have significantly

contributed to the design of smallholder agricultural development strategy.

CHAPTER 3

CONCEPTUAL FRAMEWORK, LITERATURE REVIEW AND

RESEARCH DESIGN

3.1 Conceptual Framework of Technology Adoption

Agricultural technology includes one or more of the following aspects:
mechanical (tractor plowing, harvesting), biological (high - yielding varieties),
chemicals (fertilizers, pesticides) and management methods (optimum agronomic
practices). The technology may consist of a package of several components which
may be adopted simultaneously or independently depending upon whether the
specific practice is complementary or not. The technologies can be considered
divisible (eg., hybrid seeds, fertilizers, pesticides, etc.) and non-divisible or lumpy
(eg., tractor, mechanical harvesters, etc.). Most countries have attempted to
encourage the adoption of new technologies in order to increase factor
productivity in their agricultural sector assuming that increases in the productivity
of scarce resources such as land and labor will increase aggregate output.

At the conceptual level, the adoption of technOlogy is represented as an
upward shift in the production function. A shift in the production function will
increase both the marginal and average product of the variable input.

The terms "new technology" and ”innovations" are used interchangeably in

much of the literature. Rogers (1983) makes a distinction between adoption and

42

43

diffusion. Adoption refers to an individual decision to use a new practice on a
regular basis and is synonymous with intra-firm diffusion (Stoneman, 1983).

Diffusion often refers to the communication of ideas which are not necessarily
accepted and implemented whereas adoption always implies the acceptance and
implementation of new ideas and practices. Adoption follows sequential and/or
overlapping stages of awareness, evaluation, trial and final adoption. Moreover,
different groups of farmers adopt innovations depending on their varying amounts
of human capital. Schultz (1975) contends that when new technologies are
introduced at the farm level, resources are not utilized efficiently by the farms and
disequilibrium exists in the firm. 'A new equilibrium can be obtained only through
the process of learning. With this contention, Feder et a1. (1985) define farm-
level final adoption as the degree of use of new technology in long-run
equilibrium when the farmer has full information about the new technology and
its potential.

Farmers may not achieve full adoption of a package of improved technology
in the short run, because it often takes mm for farmers to experiment with one or
more components of a technological package. Adoption may be at the aggregate
or farm level. Aggregate adoption involves the spread of an innovation within a

region and is measured by the aggregate level of use of a specific new technology

44

within a given geographical area. The farm-level adoption involves adoption by
individual farmers.

3.2 Technology Adoption Studies

Several studies have empirically established the strategic role of technological
innovation in economic growth (Manning 1984). Moreover studies conducted by
Herdt (1984), Sanderson (1984), Thomas (1982), Paulino & Mellor (1984), and
Ruttan (1986) have shown that new agricultural technology is a major source of
agricultural development in the Third World. The role of technology in achieving
a sustained increase in food production in Sub-Saharan Afiica is documented by
Herdt (1988), Delgado et a1. (1987), and Eicher (1990).

Griliches’ study of hybrid corn adoption (1957) and other diffusion studies in
the U.S. indicate that the proportion of farmers adopting a new technology usually
follows an S-shaped path, with the proportion of adoption on the vertical axis and
time on the horizontal axis. Griliches approximated these S-shaped paths with
logistic functions, then estimated the relationship between the logistic function
parameter corresponding to the rate of adoption and various profitability
variables. The study, however, did not reveal why producers did not adopt
technologies immediately, even if profitable.

Studies of technology adoption in different LDCs have identified several
specific sets of explanatory variables: farm size, tenancy, human capital, capital,
labor availability, profitability, public policy and socioeconomic variables. Akinola

and Young (1985) and Akinola (1987) in their study of adoption of tractor hiring

45

services schemes in Nigeria found farm income, family labor size, information and
distance to input supply points to be the explanatory variables.

In their study of farmers’ step-wise adoption of technological packages for
barley in Mexico, Byerlee and de Polanco (1986) found that initial capital
required, risk and uncertainty associated with rainfall and output prices were
crucial in farmers’ adoption decisions. Based on these factors farmers followed a
step-by-step-adoption process over a five year period of time. A study by Jansen
et a1. (1990) confirmed that the adoption of coarse cereals in India was mainly
affected by infrastructural and environmental (agro-climatic) variables. Jarvis
(1981) studied the adoption of improved pasture techniques in Uruguay, pointing
out that profitability of the innovation is not constant as is assumed in most
logistic function analysis but is a function of varying livestock prices.

ILCA’s recent study of technology adoption in Ethiopia investigated the
adoption of ILCA’s single-oxen technology, fertilizer and pesticides and found that
farm size and farming experience were the most important variables explaining
farm adoption (Kebede et a1. 1990). A recent study on the economics of wheat
production in the Sudan revealed that limited access to inputs has restricted the
adoption of wheat production technologies. Farmers that have access to inputs
delivered by the private sector have faster adoption rates than farmers getting
inputs from government controlled distribution systems (Hassan and Faki, 1993).
Although farm and farmer specific factors are considered crucial in

influencing adoption, a recent study on the adoption of modern mangrove rice

.46

varieties in West Africa indicated that these factors were not significant in
determining adoption. Farmers’ perception of the varietal characteristics were the a
major factors determining adoption of the rice varieties (Adesina and Zinnah,

1993).

Two alternative paradigms have been used to explain adoption lags. Rural
sociologists have used a communication network or epidemic model, while
economists have used a learning process model. The difference between the "
paradigms is that adoption in the first model occurs when the farmer gets the
information from the right messengers. In the second model the farmer combines
information with previous information and evaluates it with respect to the
probability of favorable outcomes and his / her decision rules.

The question of why farmers do not adopt profitable new technology
immediately has been investigated by numerous social scientists. Studies of the
adoption of agricultural innovations in the Third World have shown that
immediate and uniform adoption of innovations in agriculture is quite rare (Feder
et al.1985, Byerlee and de Polanco, 1986). In most cases adoption behavior
differs across socioeconomic groups and over time.

Preliminary studies have shown that farmers in Ethiopia have been very
reluctant to adopt simple recommendations such as high yielding varieties and
fertilizers for a number of years. The yield gaps between research stations, on-
farm trials and farmers’ fields are as high as five and three folds to one. The

possible explanations for the slow adoption of improved technology by farmers

47

include lack of credit, lack of location specific research recommendations, poor
extension service and rural infrastructure and repressive pricing policies. This
large and diverse list of constraints has to be examined to identify the most
limiting ones and to determine their relative importance. It is also important for
policy makers to know these critical factors for efficient allocation of the country’s
meager resources for research and development programs. Furthermore, national
research and extension services need to understand factors that induce or inhibit
the adoption of technology.
Many of these studies provided evidence to support that:
a) the Sigmoid diffusion curve approximates a cumulative normal
distribution curve,
b) different categories of adopters have differing adoption lags,
c) adopters have different sources of information;
d) demographic and education characteristics are important;
e) relative profitability, complexity and riskiness of the innovations have
significant effects on adoption rates and lags and
f) environmental factors influence adoption or no adoption of new

technologies.

48
3.3 Research Design and Methodology

3.3.1 W

Arssi Administrative Region is located in the southeastern highlands of
Ethiopia and covers an area of about 24,000 5q.km.. It is the smallest (in land
size) of the 14 administrative regions of the country8. It is made up of twelve
districts (awrajas). The bulk of the landmass is highland with elevation of over
1,500, above sea level. Most districts receive sufficient rainfall, over 900 mm per
year. The rainfall pattern is bimodal, with the small rainy season in February-
April and the main season in June-September. The amount of rain increases with
altitude. Soil characteristics vary from acidic, clay and low phosphorus content in
the highlands to mildly alkaline, sandy loam and medium phosphorus levels in the .
lowlands. The region is generally regarded as one of the most fertile in the
country. Unlike in the northern parts of the country, the soil has not been
severely exhausted by over-cultivation.

According to the 1984 census, about 1.7 million people live in Arssi and 92%
of these are rural. It is estimated that 57% of the inhabitants are Moslem and
43% Christians. The ethnic composition consists of Oromos, Amharas and other

minority groups.

 

8 In the mid 19805 a new regional classification was issued by the Military government, based
on ethnic, linguistic and agro-ecology resulted in 29 administrative regions and five autonomous
regions. This was again changed by the new Transitional Government to 14 ethnic/nationality based
regions.

49

 

ARSSI REGION
Legend:
- Amie bound
— worede ' ‘

I Worede capitals
- Al weather road
~--- Dry weather road

 

 

 

 

 

  
 
 
 
 
 
   

 

 

  

LIST OF WOREDAS

Chilalo Amja Arbagugu Amja
1. Sire . 13. ago

2. Dodote 14. Gun:

3. Hctoea 15. 0010ch

4. flyo to. Choir.

5. Zn” 8: Dugda

8. Dlgelu a rue Ticho Awraja
7. Manta l7. Sud:
8.1mm a: Bilbao re. Amlgne

9. Gedcb 19. Sen:

IO. Kofele 20. Ten:
11.55:» 21. Robe

12. Matt 22. Shirk:

 

Figure 2 Map of Arssi Region

 

50
3.3.2 Crop Production

Crop cultivation is the primary occupation of the rural population. The major
cultivated crops are barley, wheat, pulses, tef, maize, sorghum and oil crops.
Wheat and barley are grown in most altitudes and account for about 40% of the
cultivated area. Maize and sorghum are grown at the lower and medium
altitudes. The total arable land in Arssi is about 504,414 hectares with a per
capita holding of a little over 2 hectares. Nearly all farmers keep some livestock
which serve as a source of draught power for cultivation, additional income and
home consumption.

33.3 W

All farmers are members of peasant associations (PAS). There are 1,0349
PAS with 243,224 household members (1986). All PAS belong to service
cooperatives (SCs) totalling 163 in the region. These cooperatives are active in
the distribution of some consumer goods and farm inputs and handle grain quota
deliveries for the state owned Agricultural Marketing Corporation. The number
of producers cooperatives (PCs) was reported to be 409 with a total membership
of 37,242 households in 1989 (Arssi Agriculture Dev. Dept, 1989).

The Ministry of State Farms maintains a heavy presence in the region. The
Arssi Agricultural Development Enterprise runs six state farms with a total
cultivated Area of about 36,000 hectares. Wheat and malt barley are the major

crOps grown.

5 1
3.4 Survey

3.4.1 MW

Multistage sampling was used for the study. First, five of the twelve districts
in the region were identified as major wheat producers and included in the survey.
The remaining seven districts are either marginal wheat growers or they are not
suitable for wheat production because of their agro-ecology. Each district is served
by a number of Rural Development Centers (RDCs) depending on the size of the
farming population in the district. Each RDC has categorized its PAS into high,
medium or low wheat producers. The second stage was to select a sample of PAS
based on their productivity from each of the five districts. Five or six PAS per

district (2 high, 1 medium and 2 low producers) were identified.

52

Table 3.1 Peasant Associations and Household Heads Interviewed in the Farmer

Survey in Arssi Districts, 1990/91.

 

 

District Number PAS Number of
of PAS Selected Households
Interviewed
Keleta 156 6 100
Chilalo 87 5 82
Galema 107 5 97
Gedeb 107 5 74
Ticho 93 5 99
Total 443 26 452

 

Source: Field Survey, 1991.

—-——__

This stratified sampling was based on representativness and accessibility of
the PAS. The assistance of local extension agents in the selection process was
critical. A proportional sample of household beads was randomly selected from
each sample PA (15-20 Households, HHS per PA). A total of 26 PAS and 452
HHS were selected for the survey and 426 questionnaires are used for the analysis.
Table 3.1 indicates the distribution of PAS and HHS in the survey districts.

3.4.2 Quostionnairo fio’paration and Enumorator Troining

Since several farm surveys have been conducted in the region, it was possible

to prepare a relevant questionnaire in a short period of time. Questionnaire

pretesting was done to modify some of the questions.

53

Enumerators fi'om the study districts were available for employment and
subsequent training. All ten enumerators were students of the Alemaya
University of Agriculture which was closed because of the political changes the
country was going through. The questionnaire was designed to collect data on
farmer characteristics, farm resources, production practices, particularly on the
adoption of HYVS, fertilizers, herbicides and other recommended crop production
practices. Farmers’ opinions on the technologies were asked. Respondents were
asked to identify availability, purpose and sources of credit. Extension contact,
frequency of visits, and different kinds of farmer training programs and field days
are some of the mechanisms through which farmers get access to modern
production technologies. Farmers were also asked to identify the institutional
support, marketing arrangements and off farm employment opportunities.
Information on wheat quantities produced, sold and retained, number and value
of oxen, cows and sheep for each respondent was collected. However farmers
were reluctant to release information related to their income or wealth status.
The enumerators asked if the respondent was willing to disclose such
information.

3.4.3 Data Colloction

Data used for this study was collected during the eight-month field work
undertaken in February-September 1991. Available published data are used to
understand the smallholder production practices in the region. The National

Wheat Research Center, Kulumsa is located in Chilalo district and has substations

54

and trial sites in the other four districts. Information from the on-farm trials are
used to assess the profitability of the recommended technologies. Previous
Farming Systems Research surveys are used as background information for this
study. CADU/ARDU publications gave a historical perspective on pilot
integrated rural development in Ethiopia. Other statistical information on
agriculture in Ethiopia and particularly in the study areas is used for describing
the farming systems. The regional planning office has published statistical
abstracts for the 12 districts.

3.5 Summary

This chapter presented the conceptual framework on technology adoption and
reviewed the literature. Chapter 3 also presented the survey methods used to
collect primary data for the study.

The types of agricultural technologies (mechanical, biological, chemical and
management methods), the distinction between adoption (an individual decision
to use a new practice on a regular basis) and diffusion (communication of ideas
which are not necessarily accepted or implemented) are mentioned.

Studies in LDCS have attempted to explain why producers do not adopt
profitable technologies. The factors responsible for adoption or no adoption of
technologies vary across countries, regions, farms and the kinds of enterprises
(food or cash crops) under study. Various studies have documented that
economic, social, institutional and environmental variables as well as technology

specific characteristics influence farmers’ adoption decisions.

55
During the 1990/91 crop season the field survey was conducted in five major

wheat producing districts (Keleta, Chilalo, Galema, Gedeb and Ticho). From a
total of 443 peasant associations in the five districts 26 were selected for the
survey. A total of 426 household heads were interviewed. Data on farmer
characteristics, farm resources, production practices, adoption (incidence and
intensity) of wheat varieties, fertilizer and herbicide and farmers’ perceptions of
recommended technologies was collected. The questionnaire used for the study is

presented in appendix 1.

CHAPTER 4

SMALLHOLDER WHEAT PRODUCTION IN ETHIOPIA

4.1 Introduction
4.1.1 WW

Ethiopia is the largest wheat producer in Sub Saharan Africa with an average
annual area of 687,000 ha. which is half of the total area (1.3 million. ha) under
wheat in the region. Total wheat production for the period 1990-92 was 886,000
tonnes/year which is 41 percent of the regional total production. The Sudan and
Zimbabwe rank second and third with 661,000 and 222,000 tonnes per year,
respectively (CIMMYT, 1993). National average wheat yield for recent years is
about 1.3 t/ha compared to 1.7 t/ha yield for the rest of Sub Saharan Africa.

In Ethiopia, wheat is the fifth most important cereal crop, in both area and
production, grown using traditional varieties and practices. In other African
countries it is of most recent introduction and is grown using improved varieties
and production practices. The most important wheat growing areas of Ethiopia
are the highlands of the central, southeastern and northwestern regions of the
country. Wheat is grown at altitudes ranging from 1700 to 2900 meters, and
rainfall in these areas is bimodal and varies from 600 to 2000 mm. Although the
possibility for irrigated wheat production in the Rift valley has been explored,

currently all wheat is produced under rain fed conditions. Most of the wheat is

56

57

grown during the main rainy season of June to September. In areas where the
short rains of March to May are adequate farmers have the Opportunity to
produce two crops of wheat.

The two major wheat species grown in Ethiopia are durum and bread wheat.
Ethiopia is the center of origin and diversity for durum wheat which accOunts to
60% of the total wheat area in the country. Bread wheat has been introduced

recently and covers 40% of the total wheat area.

58

Table 4.1 Production, Consumption and Research Investment Indicators for

Selected Wheat Producing Countries in Sub Saharan Africa, 1983-92

 

 

 

 

 

 

 

 

 

 

Indiators Ethiopia Kenya Sudan Tanzania Zambia Zimbabwe SSAa
Area Harvested, 1990- 687 102 368 46 14 41 1,308
92 (000 ha)
Yield, 1990-92 (t/ha) 1.3 1.9 1.8 1.7 .. 55 1.7
Production, 1990-92 886 195 661 76 .. 222 2.160
(000 t)
Growth rate Yield 2.8 1.4 5.3 3.3 .. -0.7 3.7
1983-92 (%)
Growth rate 3.2 0.1 21.4 1.6 .. 4.8 7.0
Production 1983-92
(%)
Wheat area as % of 13 6 6 2 2 3 5
total cereal area 1990-
92 .
Growth rate of per 35 -0.1 4.9 0.0 -10.0 6.5 2.6

capita consumption
1983-91 (%/yr)

Percent area under 12 100 95 100 100 100 47
modern wheat
varieties, 1990

Number of wheat 35 34 8 16 14 25 138
varieties released,

1966-90

Number of full-time 13 4 2 .. 5 1 27

wheat improvement
researchers, 1992

Farm price of wheat, 338 196 .. .. 233 195

1991-92 (USS/t)

Fertilizer applied /ha 6 47 4 14 17 57 10
Of arable land, 1988-90

(kg nutrients/ha)

 

 

 

Source: CIMMYT 1992/93 World Wheat Facts and Trends. The Wheat Breeding Industry in Developing Countries:

An Analysis of Investments and Impacts.
aSSA refers to Sub Saharan African countries and the figures are aggregates for major and minor wheat producing
countries.

—

59
Table 4.2 Wheat Area, Production and Yield by Farm Types in Ethiopia, 1986

—

 

Area Production Yield
Sector ’000 ha % ’000 t % (t/ha)
Smallholders 527 82 585 76 1.11
Producers’ cooperatives 49 8 53 7 1.08
State farms 63 10 129 17 2.04
Total 639 100 767 100

 

Souroo: CSA 1987

Wheat is produced by three types of farms in Ethiopia: smallholders, state
farms and producer cooperatives. Government policy changes since 1990 have
resulted in the dismantling of producer cooperatives. Smallholders account to
82% of the total wheat area producing 76% of the total wheat harvest with an
average yield of 1.1 t/ha. Table 4.2 reveals the relative contributions of the three
types of farms in wheat area, production and yield.

4.1.2 Rogionfl Production Trends
Wheat is produced in different parts of the country and the expansion of state

farms9 (1979-81) coupled with improved wheat production technologies have

 

9 The majority of the State farms are commercial farms nationalized in 1975. The total area
under the state farms in 1975/76 was 64,000 ha expanded to 222,000 ha in 1983 and 214,000 ha in
1988. About 70 % of their farm land is under cereal crops (3% of the total cereal crop land) and
the remaining are planted with cotton, fruits and coffee (Mirotchie and Taylor, 1993). The farms
received 76 % of the fertilizer, 95% of the improved seeds and 80% of the credit in 1983 (Cohen
and Isakson, 1988). There are eight state farms in Arssi cultivating 33,974 ha and producing 654,504
quintals of different crOps annually. Wheat and barley cover about 75 of the total area of the state
farms in Arssi (Arssi Regional Atlas, 1990).

60

contributed to the concentration of production in the southeastern and central
highlands. Durum wheat is mainly grown in the other zones where the vertisols
predominate.

Table 4.3 Ethiopia: Cereal and Wheat Area and Production by Regions
(1987/88)

 

 

 

 

 

Highlands (Regions) Area Production % of
(’000 ha) (’000 t) cereal
Cereal Wheat Cereal Wheat prom
Southeastern (Arssi-Bale) 606 237 895 341 38.1
Central (Shewa) 1272 218 1451 245 16.8
Northwestern (Gojam-Gonder) 1259 102 1280 92 7.2
Northeastern (Welo) 383 49 491 31 6.4
Eastern (Harerge) 327 14 368 18 5.0
Western (Welega-Kefa-Illubabor) 817 25 1085 25 2.3
Southwestern (Sidamo-Gamo Gofa) 375 14 503 27 5.4
, Total 5039 659 6073 779 12.8
59315;; CSA 1988.

Table 4.3 indicates that about 75% of the wheat produced in the country
comes from the Southeastern and Central highlands and these regions constitute
69% of the national wheat area and have a share of 38% of the total cereal

production in the country.

61

 

 

--nalmm
Jaw-um“

634m:
”fim W
.1.:..mm emu
....0_Om mu!

- -.._mn

o manna-on mm
immune an LG.
'®.mnu nu.
'Qnm u.

'0 -uuu "nu me.
00-nou me.

I@_slmu me.

IO -uutu ac.

 

 

 

 

 

 

 

Figure 3 Map of Wheat Producing Regions of Ethiopia

62
4.1.3 Wheat flgdugign, Area and Importance in Arssi

The major crops grown in Arssi are barley, wheat, faba bean, field pea, maize
and sorghum. Oil crops such as linseed and rape seed are grown in some parts of
the region. Cereal production has expanded at the expense of grassland and
fallow land. From 1967 to 1980 area under wheat per farm increased from 16%
to 27% while barley area remained unchanged (44%). Area under maize
increased from 3.3% to 12% and Tef area increased from 1% to 5% and flax area
declined from 12% to 2.5% for the same period. In general non-cereal crops area
dropped from 33.8% to 9.8% (Bengston, 1983).

The large number of state farms operating in the region and increase in the
proportion of wheat area in the smallholder sector have made Arssi region the
largest wheat supplier to the Agricultural Marketing Corporation (AMC).

According to AMC’s classification there are 210 wheat producing sub districts
(woredas) in the country and 33 of these are surplus producing and 16 (48%) are
in Arssi region (AMC, 1989). A 1980 survey by ARDU and 1984 by Dejene
revealed 34 and 60 percent of the farmers in Arssi considered crop production as
a major occupation. The same sources suggest that crop farming is the single
most important source of cash earnings to Arssi farmers, particularly wheat

followed by barley. In the high altitude areas of Arssi about 55 percent of the

63
cropped land is under wheat and barley and in the mid-altitudes and lower areas
these crops occupy almost 71 and 34% respectively“).
The following table shows that wheat is the primary crop (55%) in the
medium altitude areas and the second crop (38%) in the highland zone. Barley is

a major crop only in the highland zone of the region.

Table 4.4 Relationship Between Arssi Ecological Zones and Crops Considered
Most Important By Farmers

Percent of Farmers Reporting Most Important Crops

 

Ecological

Zone Wheat Barley Tef Sorghum Maize Others
Highland 38 4o 8 o 4 10
Medium 52 6 2O 11 0 11
Altitude

Lowland 4 4 0 44 48 0

 

Source: Dejene Survey, 1984.

 

10According to IAR/FSR study in Chilalo district, nearly 96 and 93% of the farmers in the
high altitude zone grow barley and wheat, respectively and in the medium altitude zone all the
farmers grow both crops (Chilot et a1. 1989).

54
4.2 Crop Production Technology Use in Arssi

4.2.1 W

Farmers in Arssi have been using improved crop production technologies
since 1967 when the CADU project became operational. During the early phases
of the project land owners and progressive farmers benefited from the
technologies. The then land tenure system had hindered the adoption of
technologies by the small and tenant farmers in the region. Land owners and/ or
bigger farmers obtained significantly larger proportions of the benefits compared
to tenants and /or small farmers. Fertilizers and improved seeds were given out
by the project in direct proportion to area cultivated, i.e., a one-hectare cultivator
was entitled to one quintal of fertilizer while a 20-hectare cultivator was entitled
to 20 quintals of fertilizer (Tecle, 1973). As absentee landlords started to fully
mechanize their farms and use improved technologies a substantial number of
tenants were evicted.

Tables 4.5 and 4.6 present time series data on the volume of fertilizer and
improved wheat varieties distributed to small farmers in the five survey districts of
Arssi for the period 1975-1991. The data were compiled during the field research
from different files at ARDU headquarters. A comparative analysis of the data
reveals that the mean annual fertilizer use was the highest in Galema 18,338
quintals/year (31%), followed by Keleta 14,025 quintals/year (24%), Gedeb
12,318 quintals/year (21%), Chilalo 10,877 quintals/year (18%) and Ticho 4,812

quintals/year (8%).

65

Use of improved wheat varieties in the study areas showed different pattern.
Gedeb has the highest annual use of improved wheat seeds; 1,969 quintals/year
(29%), Keleta has 1,820 quintals/year (27%), .Galema farmers used 1,242
quintals/year (18%), Chilalo used 1,229 quintals/year (18%) and Ticho had used
only 621 quintals/year or 9% of the annual total distributed in the region during
1975-91 period. Farmers in Galema, the highest fertilizer users are third in the
use of improved wheat and Gedeb farmers are first in wheat seed use but third in
fertilizer use. This does not suggest that higher fertilizer use necessarily imply
higher seed use or vice versa. However, Keleta, Chilalo and Ticho are second,
fourth and fifth respectively in both kinds of input utilization. Districts with larger
numbers of producers cooperatives received the highest share of the inputs.
Although the aggregate time series data show the above trend, it is important to
further investigate differences in input use across farmers in each district to
establish a relationship between farmer-specific socioeconomic characteristics,

their levels of input use and natural and institutional constraints.

66

Table 4.5 Amount of Fertilizer Distributed in Arssi Districts 1975-1991.

 

 

 

Keleta Chilalo Galema Gedeb Ticho Arssi
Total
Year 3 Amount of Fertilizer(DAP) distributed
in quintals

1975 15510 17350 16109 6178 N/A 55147
1976 3796 7616 18881 3454 N/A 33747
1977 13429 10918 14607 8906 N /A 47860
1978 11212 7045 20529 5753 N/A 44539
1979 6819 7210 17941 17565 2661 52196
1980 8040 8022 17823 7259 2310 41796
1981 10503 8474 14583 12476 1882 47918
1982 9732 3775 14767 13078 1859 70211
1983 10214 6545 13588 9234 1795 41376
1984 11130 7132 18030 12930 4421 53643
1985 16645 8149 17860 10928 4244 57826
1986 9585 6029 10586 5533 1596 33329
1987 16806 10556 20703. 16449 6692 71206
1988 25132 16944 26404 27140 9359 104979
1989 25086 10892 23354 18637 9360 87529
1990 22805 10741 23873 22169 11034 90622
1991 21988 10503 22107 11712 7002 73312
Mean 14025 10876 18338 12318 4812 59237

 

_ 3.932%; Compiled by Authorii'om ARDU files, 1991.

67

Table 4.6 Amount of Improved Wheat Seed Distributed to Farmers in Arssi
Districts 1975-1991.

 

 

 

Keleta Chilalo Galema Gede Ticho Arssi
b Total
Year Amount of Improved Wheat seeds distributed
in quintals
1975 241 342 435 135 N /A 1153
1976 3256 1577 580 910 N /A 6323
1977 1491 3239 2430 4518 N/ A 11678
1978 2334 332 1110 141 N/A 3917
1979 1045 989 1080 676 411 4201
1980 N/A N /A N /A 339 652 991
1981 246 189 855 5099 402 6851
1982 489 219 933 4453 399 6493
1983 704 N/A 306 4161 466 5637
1984 1868 826 114 2296 793 5897
1985 2910 4500 1964 886 468 10728
1986 1900 1300 1947 N/A N/A 5147
1987 896 924 151 236 249 2456
1988 3075 1143 4049 5968 1055 15290
1989 3075 1143 3949 1345 1055 10567
1990 1460 1775 1210 2310 1400 8155
1991 5957 2396 N/A N/A 729 9082
Mean 1820 1229 1241 1969 621 6735

52m; Compiled by Author from ARDU files, 1991.

—

68

 

USEOIAREBIEHRETS

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1975-'991

120
«£100
5 so 27
3
a a; §§\
5 , $115 a t \M
g 40 § \\
I: \\ \
E 20. ........... .... \\

 

75 76 77 78 79 80 8182 83 84 85 86 87 88 89 90 91

 

33%; KELETA - CHLALo GALEMA
EEEE GEDEB ncuo

 

 

 

 

 

 

 

Figure 4 Fertilizer Distribution to Smallholders in Arssi Districts, 1975-1991

69

 

WROVEDWI-EATUSENARSSDISIBCTS

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

     

 

 

 

 

   
   

 

 

 

       
 
    

 

 

 

 

 

 

 

 

 

 

 

1975-1981
16
9 12 .. :
§A10 :1 E 1i [7]
Cl i—u- I—tt-
3 v —:~ ~-
>- S 8 :: '"
E i :5 ‘ s
\ \ \
t- s
2 4... .................... .........
2...-..............I.
......... -~ g;
o- (0‘.

   

KELETA - ChlALO GALEMA
E GEDEB TICHO

 

 

 

 

 

 

 

Figure 5 Improved Wheat Seed Distribution to Smallholders in Arssi Districts
1975-1991

70

The two graphs (Fig.4 and 5) illustrate that there was a moderate annual
increase in the use of both inputs in the four districts except Ticho for the period
1975-78, and 1979-82 for all the districts. The period 1983-86 did not show a
marked increase and 1986 and 1980 have the lowest amount of fertilizer and
wheat seed distribution, respectively. In 1980 no wheat seed was distributed for
three of the five districts.

In 1987 the government declared a new policy to attain food self-sufficiency .
by adopting an agricultural development strategy known as the Peasant
Agricultural Development Program (PAD-EP). The program concentrates
development efforts in selected high potential regions and districts to boost food
production in a short period of time. Input distribution, extension service and
adaptive research programs are to be directed to 153 districts selected that were
identified as high grain-producing areas. A new agricultural marketing and
pricing policy that removed the restrictions on private grain trade and a modest
increase in farm gate prices for the major crops was instituted in 1988. As a
result in the period 1987-1990 both fertilizer and improved wheat seeds
distribution increased by more than 50% and 60% respectively in the five districts.
The following sections describe farm level adoption of yield increasing inputs and
farmers’ perception and assessment on the usefulness and/ or problems in using

these yield increasing inputs.

71
4.2.2 Adpption pf memercial Fertilizer

Commercial fertilizers are widely used as 86% and 87% of the sample
farmers in the region reported that they applied Diammonium Phosphate
(DAP)11 in their wheat fields in 1990 and 1991 crop seasons, respectively. The
average level of fertilizer use for wheat was 63 and 64 kg per household in the
two seasons. Table 4.7 indicates percent of farmers in each district who applied
fertilizers on their wheat fields in 1990 and 1991 crop seasons, average amount
used per household; number of years of awareness; number of years since first
used; farmers’ knowledge of recommended rate; assessment on fertilizer
availability; average rate of fertilizer applied and percent area fertilized per farm.
Categoric data were cross tabulated by districts and a Chi-Square test of
significance was calculated. All variables show significant difference between the
five districts at the 0.05 level. Values for the descriptive statistics of the
continuous variable also suggest statistically significant differences between the
study areas. The adoption of yield increasing technologies in Arssi is expected to
be higher than in other parts of the country given the many years of integrated
rural development programs of ARDU. A closer look at the rate of fertilizer
application (kg/ ha) across districts reveals that about half of the sample farmers
in the region apply less than 50 kg/ha which is less than half of the recommended

rate of 100 kg/ha DAP and 50 kg/ha of urea for mid altitudes and 150 kg/ha

1 1DAP and urea are the two types of fertilizers distributed in Ethiopia. DAP is widely used in
Arssi and Urea is used by less than 2% of the farmers.

72
DAP and 50 kg/ha urea for higher altitudes. The average rate of DAP

application on wheat is 71 kg/ha (1990) and 72 kg/ha (1991) and ranged from 7

to 200 kg/ha among the sample farmers.

Table 4.7 Use, Experience and Farmers’ Knowledge of Fertilizer in Arssi

Districts.

 

Keleta Chilalo Galema Gedeb Ticho Arssi

Fertilizer use % 87 97 87 83 78 87
of farmers

Average Use 66 85 65 65 39 63
kg/hh

Awareness mean 18 20 19 18 15 18
years

Number of years 14 18 16 15 11 18
used

Use fertilizer 69 89 85 76 51 73
yearly %
farmers

Knowledge of 69 78 89 75 77 78
recommended

rate % of

farmers

On time 0 44 26 48 51 32
availability of

fertilizer % of

farmers

Rate of 67 82 82 60 65 72
application mean

kg/ha-

% area fertilized 64 68 44 43 55 55
mean per h

Sm Field Survey, 1991.
—

 

 

73

Tables 4.8 and 4.9 present categories of fertilizer application rates and levels,

and the percent of farmers in each category by districts.

Table 4.8 Percent of Farmers in Different Fertilizer Application Categories, 1990-
1991.

Keleta Chilalo Galema Gedeb Ticho Arssi

 

Application Percent of Farmers

Rate kg/ha.

less than 25 28 13 6 31 28 21
26 - 50 29 19 22 33 33 27
51 - 75 23 22 17 17 16 19
76 - 100 14 32 50 15 18 26
More than 6 14 5 4 5 7
100

Mean 68 82 81 60 65 72

 

Sm; Author’s Survey, 1991.

The data further illustrate the difference between the districts in the rate of
fertilizer application. Almost 30% of the farmers in each of the three districts of
Keleta, Gedeb and Ticho used very low rates of fertilizer, i.e. less or equal to 25
kg/ha. Only 11% of the farmers in Chilalo and 5% in Galema applied less than
26 kg/ha and 45 and 55% of them applied more than 75 kg/ha.

The two districts of Galema and Chilalo are the high production potential
districts. Farmers have been aware of fertilizer for almost 20 years while more

than 85 % have been using fertilizer yearly (Table 4.7).

74

Table 4.9 Percent of Farmers and Fertilizer use Levels in Arssi, 1990-1991.

 

Levels of Galema Chilalo Keleta Gedeb Ticho Arssi
Application

Low 28 32 57 64 62 49
50 kg/ha or less

Medium 17 22 23 17 16 19
51-75 kg/ha

High 55 46 20 ' 19 22 32
more than 75 ‘

kg/ ha
SEES; Fiefi Survey, fil

 

 

Farmers in the three low level fertilizer user districts (Keleta, Gedeb and
Ticho) have reported average years of awareness of 18, 17 and 15 years,
respectively. Ticho has the least number of farmers using fertilizer yearly and it is

not more than eleven years since farmers started to use commercial fertilizers.

75

Table 4.10 Adoption of Wheat Varieties, Average Area Planted, Seed Rate and
Yield in Arssi Districts, 1990-1991.

 

 

 

Wheat District % Farmers Mean area Seed rate Yield
Variety planting planted kg/ ha kg/ ha
ha/hh

Dashen Keleta 63 0.71 210 1057
Chilalo 34 0.59 232 934

Enkoy Chilalo 66 0.62 231 1330
Keleta 13 0.50 211 1217
Galema 89 0.72 207 1441
Gedeb 68 1.01 148 1068
Ticho 65 0.49 192 805

Bulk K6295 Chilalo 18 1.06 155 1063
Gedeb 29 1.86 148 1025
Ticho 56 1.14 137 965

 

Source: Field Survey, 1991

4.2.3 Adpptipn pf High Yielding Varieties of Wheat
The adoption of wheat HYVs in Ethiopia is limited as only 12% of the total

wheat area in the country is planted to modern wheat varieties while in other

major wheat producing countries in Sub Saharan Africa 100 percent of their total

76
wheat area is planted with HYVs (CIMMYT, 1993)12. Use of the local wheat

varieties, which are generally low yielding, have been identified as limiting
constraint to increase wheat production in Ethiopia.

Table 4.10 depicts the adoption, average area, seed rate and yield of three
bread wheat varieties in each district. The survey data indicates that the most
popular variety grown by farmers in all of the 5 districts is Enkoy followed by
Bulk-K6295 produced in 3 of the districts. Variety Dashen is grown in Keleta and
Chilalo. Dashen, released in 1984, has been susceptible to diseases; farmers are
no longer planting it in the high altitude areas. It is still widely grown in the mid
altitude zones of Chilalo and Keleta. The widespread outbreak of yellow rust in
1988/89 almost wiped out the high yielding variety Dashen over a large area in
Arssi and Bale regions (Debela, 1990).

Research station yield of 3810 and 4670 kg/ha for Enkoy and Dashen (Hailu,
1991) confirm that there is a potential for increased wheat production through the
adoption of improved production technologies. Yields from a number of on—farm
trials of four bread wheat varieties ranged 2099 to 2545 kg/ha ( Dereje et al.,
1990). State farms reported yields were 1450 and 1680 kg/ha for 1989 and 1990
seasons. Despite the widespread use of the three varieties, the yields reported by

the respondents in the survey suggest that farmers are not getting more than 1.5

 

12CIMMYT 1992/93 World Wheat Facts and Trends data indicate that the major wheat
producing countries i.e., Kenya, Tanzania, Zambia and Zimbabwe have about 100% of their wheat
areas under improved varieties while in the Sudan and Ethiopia 95 and 12 % of their wheat fields
are planted with HYVs, respectively. Ethiopia is the origin and center of diversity for wheat and
most of the fields are under traditional wheat varieties.

77

t/ha13. Compared to the research station, on-farm trial and yields from the

state farms the average yield for the sample farmers clearly indicate the presence
of yield gaps. Further explanations of the factors responsible for the yield gaps is
required.14 Table 4.15 shows that the variety Enkoy is grown by 89% of the
sample farmers in Galema where the other two varieties are not grown at all. In
Gedeb and Ticho districts this variety is planted by 64-70% and in Keleta by 8%
(1990) and 18% (1991) of the farmers in each district. Bulk variety is popular -in.
Ticho where 55-58% of the farmers are growing it, and in Chilalo and Gedeb 29
and 19% of the respondents have adopted it.

Mean area under these three varieties per household in each of the five
districts for the 1990 and 1991 seasons ranges from 0.43 in Chilalo to 2.9 ha in
Gedeb for Bulk. The seed rates range from 86 to 240 kg/ha. For most of the
cases the seed rates are on the high side relative to recommended rates of 175
kg/ha for broadcast sowing. Farmers believe that higher seed rates will help the
crop to out compete weeds and also minimizes the risk of poor germination for
seeds purchased from the market or from own store. Table 4.11 reveals that the
average quantity harvested per household for the region is between 613 and 812

kg/hh. Enkoy gave the highest mean production of 1001 kg/hh in Gedeb, and

 

13 It is expected that farmers under report their harvest for they are suspicions that
government institutions might use the study to increase the agricultural tax burden. The reported
harvested quantity and yield during the survey (0.8 to 1.44 t/ha) is hence under estimated at least by
25 %.

14 Yield gap 1 is defined as the difference between current yield at research stations and
researcher managed trials and Yield gap 2 is the difference between yields obtained by farmers and
yields from on-farm trials (Dillon and Hardaker, 1984).

78
Bulk gave the lowest quantity of 467 kg/hh in Ticho. Quantities sold and retained

by each household are shown in the table and prices in Keleta are the highest (93
EB/q) and lowest in Ticho (57-60 EB/q). Most farmers used their own seed
retained from previous years. The farmers in Ticho (11%), Galema (18 %) and
Keleta (23 %) got their wheat seed from the MOA which is an indicator of the
inadequate supply of improved seeds and subsequent low yields.

4.2.4 Adoption of Herbicides in the study districts

Smallholder agriculture in Ethiopia is characterized by the prevalence of
many enterprises where crop and livestock systems dominate and compete for
available farm resources. Farm operations in general and crop production
activities in particular are labor intensive and compete for farm labor during the
critical periods of weeding and harvesting. Farm families have to make rational
decisions in allocating their limited resources between the different farm
enterprises. Weed infestation has been recognized as a major constraint to wheat
production both at smallholder and state farm sectors. The latter mainly use
herbicides to control weeds in their large farms and most of the small farmers use
hand weeding.

Surveys on weed control practices have revealed that farmers prepare a fine
seed bed preparation to reduce weed infestation. A second practice, high seed
rate, facilitates the establishment of a dense stand of wheat to compete with
weeds. In many regions wheat fields remain unweeded or weeded very late

during the crop cycle after the weeds have affected the crop stand.

79
Table 4.11 Average Quantity of Wheat Harvested, Sold and Retained, Farm Gate

Prices and Seed Source in Arssi, 1990/ 91.

—

 

 

 

 

 

 

 

Keleta Chilalo Galema Gedeb Ticho Arssi
Quantity
harvested
Dashen 721 673 714
Enkoy 650 769 970 1001 411 812
Bulk 669 809 467 613
Sold
Dashen 393 395 390
Enkoy 350 383 457 380 183 404
Bulk 281 377 184 284
Retained ,
Dashen 553 571 561
Enkoy 550 517 711 758 389 614
Bulk 517 567 542 558
Price EB / q
Dashen 93 76 88
Enkoy 93 66 69 64 57 67
Bulk 64 62 60 62
Seed
bought- Kg 114 104 86 130 86 104
Price EB / q 117 90 39 90 95 93
Seed Percent of Farmers
sources
MOA 23 41 18 30 11 23
Market 19 24 12 16 39 2
Own seed 56 32 69 55 47 53

 

Spurce: Field Survey, 1991

 

The farmers’ explanation for suboptimal weeding include the following:
shortage of labor, insufficient cash to hire labor during the peak flush of weed
growth, intense rain fall which can limit the ability to enter fields and hand weed

wheat, and the overlapping of the optimal time of weeding with other crops

80
(Tanner and Giref, 1991; Birhanu, 1983, 1985; Franzel and Mekuria, 1987).

Different research programs have addressed the weed control issue and identified
recommendations of time and frequency of hand weeding, a combination of pre-
emergence herbicides and hand weeding, pre-emergence and post-emergence
herbicides to control different weed species.

Recently the economics of herbicides has been studied by economists in IAR.
The objective of the studies is to verify whether family labor is a constraint during
critical farm activities such as weeding. The government has been discouraging
the use of herbicides by smallholders under the assumption that family labor is
available for weeding. However the studies have confirmed that a) in terms of
weed control cost per hectare, most of the herbicides cost less than labor, b)
herbicides are more efficient and increase yield, c) they release labor for other

crop production activities (Hailu, 1990).

81

Table 4.12 Farmers’ Weed Control Practices in Arssi Districts, 1990/ 91

 

 

 

 

 

 

 

 

Keleta Chilalo Galema Gedeb Ticho Arssi
% Farmers
applying
herbicides
1990 48 52 4 0 8 23
1991 42 21 13 7 19 21
Wheat area
sprayed ha
1990
1991 0.75 0.96 0.94 2.00 057 0.89
0.83 0.96 0.75 0.87 0.68 0.82
Herbicide
cost EB/lit
1990 28 26 32 15 26
1991 30 32 27 29 26
herbicides
disadvantageo
us %
Yes 39 43 13 4 13 23
No 61 57 47 96 87 ‘77
Number of
weedings %
of farmers:
1990 1991 1990 1991 1990 1991 1990 1991 1990 1991 19901991
once
twice 6483 9496 8592 5963 8191 7685
three times 29 14 6 4 15 8 38 37 17 6 21 13
7 3 3 2 3 3 2
Weeding
starts-
Number of
weeks after
planting %
3 30 19 ' 12 20 15 19
4 43 36 44 55 41 43
S 18 28 20 6 17 18
6 or later 9 17 fi 19 27 20

 

 

 

Source: Field Survey, 1991.

82

Table 4.12 presents farmers’ weed control practices in the study areas. Only
21-23 percent of the 426 households applied herbicides either in 1990 or 1991.
About half of the sample farmers in both Keleta and Chilalo appear to use
herbicides and few farmers have adopted in Galema, Gedeb and Ticho. Average
area of wheat sprayed does not exceed 1 ha for most of the locations, except in
Gedeb where one farmer sprayed his 2 ha field. The cost of herbicides depending
where they get from varies from EB 26 to 32 per liter. Most of the respondents
(60-90%) weed their fields only once and weeding starts three to four weeks after
planting. Farmers in Chilalo and Galema tend to weed their fields twice and
some farmers start the first weeding four and five weeks after planting. Studies in
the region have found that the largest yield increases have been obtained with
early hand weeding during the interval between 15 and 30 days post emergence
of the wheat crop. The weed control practices of the farmers is expected to
explain the yield differences between households.

4.3 Farmers’ Perceptions of Yield Increasing Technologies.

In their recent study on technology characteristics, farmers’ perceptions and
adoption in Sierra Leone, Adesina and Zinnah (1993) discuss the three paradigms
for adoption decisions: the innovation-diffusion, the economic constraint and the
adopters’ perceptions paradigms. They assert that farmers’ perceptions of
technology-specific attributes of the mangrove rice varieties are major

determinants of farmers’ adoption decisions in Sierra Leone.

(M‘f

“54., M
I

16C:

f?

B)

{it

83

Farmers were asked to identify specific reasons or factors they consider are
important in deciding to use or not to use fertilizer, wheat variety or herbicides.
In addition they were requested to list the advantages or disadvantages of the
technologies. Farmers’ criteria for varietal preference were recorded as such
information is indispensable for the design of farmer-oriented research programs.
Byerlee and Franzel (1993), Mekuria et al.(1992) and Franzel (1992) have
reported the contributions of farmers’ feedback in influencing technology
development and transfer and consequently improving the internal and external
efficiency of the research system in Ethiopian agriculture.15

Farmers who have not used fertilizers indicated that unavailability and high
fertilizer prices were the two most important reasons (32 and 24%) for not
adopting. The same factors are identified contributing to lower rates of fertilizer
applications. About 72 percent ranked unavailability of the inputs forced them
use suboptimal levels.

The respondents recognized grain yield (97%) and straw yield (29%) and
quality improvement (42%) as the advantages of fertilizers. Some farmers (10
and 17%) indicated that fertilizers damage or burn the crop and favors weed
growth. Fertilizer use increases the demand for farm labor.

Less than a quarter of the respondents use herbicides in the region and those

not using at all reported that herbicides are not available even if they want to

 

15 For detailed analysis refer to Mulugetta Mekuria et al (1992); Franzel (1992), Byerlee and
Franzel (1993)

34
Table 4.13 Farmers’ Assessment and Perceptions of Wheat Production

Technologies in Arssi

 

 

 

 

 

 

 

Fertilizer Herbicide

Not using Lower Rates Not Using
Farmers’ reasons Percent of Farmers
Unavailable 32 29 58
Expensive 24 56 53
Late delivery 7 2
Not heard of it 3
Damages/ burns crop 5
Not effective
Others 32 9

 

 

Source: Field Survey, 1991.

 

apply. Higher prices are considered as disadvantages of herbicides by farmers and
are aware of the labor saving advantages (85 %).

Farmers’ selection criteria among available wheat varieties differs between
farmers and regions of the country. Farmers give different weights to the grain
color, size, taste in different foods or the price they receive in the local markets.
In terms of production potential of the varieties, criteria includes higher yields,
resistance to certain plant diseases and pests, height of the crop as residues are
fed to cattle, earliness, weed competition and any other specific factors the

farmers consider are important.

85

Table 4.14 provides information on farmers’ criteria for varietal preferences
in the study areas. About 75 and 84 percent of the respondents said their
preference for Dashen and Enkoy is influenced by the high yielding attributes of
the varieties. Half of the farmers who chose Bulk also indicated the yield as a
major criteria. For Dashen farmers consider grain color (16%), and taste in
bread as their first criteria. The second set of criteria indicated include grain
color (31%), taste in bread (23%), taste in Injera (15%), higher prices (23%) and
lodging resistance (8%). Enkoy’s disease resistance ability (20%), lodging
resistance (14%), earliness (10%), its taste in Injera and bread (25 and 15%) are
the second group of criteria that farmers identified.

The variety Bulk has been preferred by some farmers for its disease
resistance (9%), and its taste in Injera and bread (12 and 8%). Farmers’
perceptions of the varieties are clearly indicated in their responses. Apart from
the yield consideration farmers prefer Enkoy on its agronomic characteristics
(disease and lodging resistance, earliness) relative to the other two varieties.
Dashen has also gained popularity in its grain color, taste in Injera and higher
market prices.

The development and transfer of wheat production technology requires that
farmers’ evaluation of recommended technologies and practices be included as
important set of criteria in the planning stages of the research programs.

Subsequent sections of the study will examine the appropriateness and

profitability of the technologies in order to identify adoption constraints.

86

Table 4.14 Farmers’ Criteria for Wheat Varietal Preferences in Arssi
—

 

 

 

 

 

 

Dashen Enkoy Bulk 6290
Criteria Criteria Criteria Criteria
First Second First Second First Second
Higher yield 74 84 3 55
Disease 1 20 9
resistance
Lodging 8 2 14 4
resistance
Earliness 2 10 1
Grain size 1 1
Grain color 16 31 1 4
Weed 2
competition
Taste in 15 2 25 12
’Injera’
Taste in 10 23 2 15 8
bread
Higher price 23 1 5 3
Other 6 5 3
factors

 

 

 

ource: Field Survey, 1991

87

4.4 Characteristics of the Households
4.4-1 mm

The size of the household is an important variable which determines total
labor available and the number of individuals the household has to supply
necessary requirements such as food, shelter and clothing. Larger households put.
pressure on the household resources and at same time larger household size
implies greater access to labor particularly during critical farm operations. Labor
availability has been an important issue in the technology deve10pment and
transfer. Several FSR studies undertaken by IAR researchers have concluded that
despite larger family sizes (5-7 persons per family) in rural households, labor
availability during weeding and harvesting is a major constraint in rural
Ethiopia16. For the study region the average family size is 7.3 persons per
household. Table 4.15 shows that 29% of the sample farmers have a family size
of less than 5 while 58% and 13% of the households have 5 to 10 and more than
ten members, respectively. There is no significant difference in family size across
the 5 districts surveyed. It is expected that family size differences between
households will explain the difference in production decisions that farmers make

and it will be examined in subsequent chapters.

 

16 For detailed analysis of farm level constraints under different farming systems in western,
north central, southeastern and semi-arid areas of Ethiopia, see Franzel and van Houten (1992).

88

Table 4.15 Family Size Categories by percent of Households in Study Districts of

Arssi

 

Family Keleta Chilalo Galema Gedeb Ticho Total
Members

Percent of Farmers
Less than 5 8.2 4.2 7.3 4.7 4.5 28.9
5-10 12.5 10.1 125 9.2 13.4 57.6
Greater than 2.6 2.6 2.6 2.1 3.5 13.4
10
Mean Family 6.8 7.5 7.0 7.7 7.6 7.3
size per HH

 

MField Survey, 1991.

1

Table 4.16 Age Categories of Household Heads in Arssi Districts

 

 

Category Keleta Chilalo Galema Gedeb Ticho Arssi
Percent of Farmers

15 -25 12.0 6.9 12.6 16.2 1.1 9.6

26 - 35 19.0 23.6 22.1 23.5 15.4 20.4

36 - 55 40.0 37.5 38.9 36.8 64.8 44.1

56 and 29.0 31.9 26.3 23.5 18.7 25.8

Older

Mean HH 44.5 47.5 43.8 43.0 47.8 45.3

age yrs.

 

m Field Survey, 1991.

89
4.4.2 Age Categpries of Household Heads

About 65% of the sample farmers are in the age categories of 26-55 and a
quarter of them are more than 55 years of age while close to 10% are young
farmers under 25 years old. The mean household head age for the sample is 45
years. There is no significance difference in age categories between household
heads across the districts.

4.4.3 W

The country embarked upon a vigorous national literacy campaign in the
early 19805 and succeeded in reducing the illiteracy rate from 80% to 35%. All
high school graduates were required as a national service to participate in the
literacy campaign for 4-5 months before the beginning of the next academic year.
The campaigners were living in the rural communities and farmers associations
were providing food and shelter while the Ministry of Education was covering
transport cost and other teaching materials. The literacy data from Arssi indicate
that 17% of the sample farmers are illiterate (percentage of population age 15 or
older who cannot read and write a short simple statement about every day life).
Half of the respondents have participated in the national literacy campaign while
22% have elementary and 10% of them have secondary level education. The
literacy education was given in the local language of the area. According to the
World Bank (1992) Ethiopia is in the 20%-39% category of its classification of

illiteracy level.

90

Table 4.17 Levels of Education of Household Heads in Arssi Districts, 1990/ 91
_

 

 

 

Level Keleta Chilalo Galema Gedeb Ticho Arssi
Percent of Household Heads

Illiterate 18.0 11.1 15.8 22.1 18.7 17.1
Literacy 47.0 51.4 49.5 35.3 65.9 50.5
Campaign

Elementary 27.0 30.6 20.0 26.5 9.9 22.3
Secondary 8.0 6.9 14.7 16.2 55 10.1
Total 100 100 100 100 100 100

 

M Field Survey, 1991.

 

Given more than 25 years of rural development project experience and the
recent national efforts on literacy education the regional literacy level is higher
compared to the national average. Among the districts in the region Chilalo has
only 11% illiteracy rate while the other four districts have between 18% and 22%
illiteracy rate.

As Chilalo is the district where the regional capital Assela is located, with
more educational opportunities being available it is not surprising to find higher
rates of literacy among the sample households in the district. About 65 % of the
respondents in Ticho district and half of the sample farmers in Chilalo and
Galema indicated that they benefited from the national literacy campaign. Forty
seven and 35% of the farmers in Gedeb and Keleta, respectively, have also
acquired literacy campaign education. The data reveal that the national literacy

program has reduced the illiteracy rate in a very short period of time. Farming

91

experience of household heads ranges from 25 to 30 years and there appears no
significant difference between districts.
4.4.4 Fag; Sizes of Households

Average farm sizes per household in Gedeb and Galema are 4.1 ha. and 3.1
ha. while Ticho, Keleta, and Chilalo have 2.0, 2.2 and 2.5 ha. respectively.
Differences in farm sizes across districts and between farmers in each district will
further be examined to study the relationships with technology adoption. Regional.
mean farm size is 2.71 ha. and 71% of the sample farmers own less than 3 ha.
Nearly 16% and 13% of the farmers in the study area own 3.1 - 4 ha. and more
than 4 ha respectively. A quarter of the respondents in Ticho own up to 1 ha.
while no farmer in Galema and Gedeb own less than 1 ha. A significant number
of the sample farmers in Keleta, Ticho and Chilalo have farm size of 1.1 - 2 ha.
Gedeb and Galema districts are characterized by relatively larger farm sizes as
60% of the farmers in Gedeb and 45% in Galema have farm sizes greater than 3
ha. Low population density and loWer soil fertility in these districts contribute to
larger farm sizes.

A chi-square test of the cross tabulation of farm size category by districts
suggested a significant difference in farm sizes across the districts. Moreover an
F-test of the mean farm and arable area sizes showed a statistically significant

difference across districts.

92

Table 4.18 Farm Size of Households in Arssi Districts, 1990/ 91
—

Farm Size Keleta Chilalo Galema Gedeb Ticho Arssi

 

 

Total
Percent of Households

Up to 1 Ha 5.0 83 0 0 24.2 7.7
1.1 - 2 52.0 37.5 17.9 13.2 41.8 33.6
2.1 - 3 31.0 29.2 37.9 27.9 20.9 29.6
3.1 - 4 9.0 9.7 27.4 27.9 6.6 15.7
More than 4 3.0 15.3 16.8 30.9 6.6 13.4
Farm size 2.2 2.5 3.1 4.1 1.9 2.7
hit/11h

Arable area 1.9 1.8 2.0 2.6 1.4 1.9
htil/hh

 

m Field Survey, 1991

93

Table 4.19 Number and Mean Values of Selected Household Characteristics
—

 

Keleta Chilalo Galema Gedeb Ticho
Number of 100 72 95 68 91
Households
Family Size 6.9 7.6 7.0 7.7 7.6
Age of Household 45.0 47.5 43.8 43.1 47.8
Head
Level of Education 1.3 1.3 1.3 1.4 1.0
Farming Experience 26.9 28.7 26.2 25.4 29.9
Years
Farm Size per HH 2.2 2.5 3.1 4.1 2.0
in ha.

 

WField Survey, 1991.

4.5 Summary

Chapter 4 discussed the importance of wheat in Ethiopian agriculture in
general and in Arssi in particular. Farmers’ household characteristics, wheat
production practices and technology adoption patterns are presented in the
chapter.
Family size /Age: The mean family size for the region is about 7.3 persons and 57
percent of the families have 5 to 10 members. The family sizes in the region are
above the national average of 5 persons. The mean household age is 45 years.
The age distribution indicate that 30 percent of the household heads are under
the age of 36 and 44 percent are between 36 to 55 while 26 percent belong to the

56 or older age category.

94

Educational Levels: Since rural development project in the study area started in
1967 farmers appeared to have benefitted from the early years of literacy
campaign and also from the national literacy program the country launched in the
1980s. Only 17 percent of the sample farmers reported illiterate compared to 20-
39% illiteracy level for the country.
Farm Size: The mean farm size and arable area for Arssi are 2.7 and 1.9 ha.
There is significant difference in farm sizes across the study districts. The
regional data illustrate 40 percent of the households cultivate 2 ha or less and the
other 45 percent have farm sizes between 2.1 - 4 ha. Only 13 percent of the
farmers own 4 or more hectares of farm land.
1 A ' n i i

Historical and current data on the use of improved wheat varieties, and
fertilizer in the region and current adoption rates in the selected districts are
presented (Tables 4.5 and 4.6, Figures 4 and 5). Although the initial benefits of
the integrated rural development project were biased in favor of land-owning class
in Arssi, the radical land reform of 1975 gave the small farmers access to land and
improved production technologies. The following evidence from the survey
support the above assertion. Compared to farmers in other regions of Ethiopia
farmers in Arssi a) have high rates of fertilizer adoption (87%), b) on the average
they apply 71 kg fertilizer per ha for wheat e) have 55% of their arable area
fertilized per household and d) have more than 15 years of awareness and

knowledge about fertilizer use (Table 4.12). Almost half of the sample farmers in

95

the region use 50 or less kg of fertilizer per ha. About 60% of the farmers in the
districts of Ticho and Gedeb apply 50 kg or less fertilizer per ha while the same
proportion of farmers in Chilalo and Galema use relatively higher rates (75-100
kg/ha).

In terms of varietal adoption, Enkoy is grown in all the districts and is the
highest yielder except in Ticho. Bulk K6295 and Dashen are popular in three and
two of the districts, respectively. Dashen, though recently released and adopted in
a short period of time has been susceptible to yellow rust epidemic in the higher
altitudes of Arssi. Enkoy which is well adapted to in the different zones of the
region has a yield gap of 45%, between research station and on-farm trials, and a
30% yield gap between on-farm trials and farmers’ reported yields. In Chilalo
and Keleta 52 and 48% of the farmers reported using herbicides (in the other
areas the number of users is insignificant).

Farmers’ opinions on specific attributes and their criteria for adopting the
recommended wheat production technologies were collected and summarized
(Tables 4.13 and 4.14). Unavailability, late delivery and higher prices are the
most frequently reported reasons for not using of fertilizer, herbicide and HYVs
in the districts. For instance, 75 and 84 and 55% of the respondents said yield is
their criteria for selecting Dashen, Enkoy and Bulk. Other factors such as grain
color, grain price, taste in preparing local foods, resistance to lodging and diseases

are listed as important criteria in varietal preference by farmers.

CHAPTER 5
TECHNOLOGY ADOPTION MODEL

5.1 The Economics of Adoption and Conceptual Model

The adoption of technology is an economic decision based on expected
marginal benefits and costs. Most empirical specifications deal with a variety of
models of farmer or farm household optimization: maximizing expected profits,
expected utility of profits or expected utility of consumption and leisure subject to
I production function and time. For example, discounted expected profits is the
difference in discounted expected value of production of all crops and livestock
with and without the adoption of a particular technology, minus the difference in
costs. Farmers’ adoption decision making is analyzed using qualitative response
(QR) models also known as binary, discrete or dichotomous models. Applications
of QR models are appropriate for analyzing relationships involving a discrete
dependent variable. Based on the assumption that farmers will make the
adoption decision based upon an objective of utility maximization, a technology t2
is preferred to technology t1 as long as utility derived from t2 is greater than
utility from t1.

Following Rahman and Huffman (1984), the utility function that ranks the ith
farmers preference of technology is represented as: U(Rfi, Afi ), where utility U

depends on a vector Rt that describes the distribution of net returns for

96

97

technology t, and a vector At of other attributes associated with the technology.
The variables Rti and Afi are unobserved and unavailable, but a linear
relationship is postulated for the ith farmer between the utility derived from the
ith technology and a vector of observed farm specific characteristics X1 and a zero
mean random disturbance term et :
Uti = Xit + et, 1‘ = 1, 2 andi = 1,..., n. (1)
The ith farmer adopts t2 if Ut2 exceeds Ut1 and the adoption decision can be
represented by a qualitative variable Y.
Y = 1 if Utl < Ut2’ new technology t2 is adopted replacing old
technology t1
0 if Utl 2 Ut2’ old technology t1 is continued. (2)
The probability that Yi is equal to one is expressed as a function of farm/farmer
specific characteristics: 1
Pi = Pr(Yi = 1) = Pr(Ut1 < Ut2)
= Pr(xi°‘1 +81: < Xiaz + 321')
= 1311811” 322' < Xi(°‘2 '“1”
= PM < xis )= min). (3)
Where Pl.(.) is a probability function,
“i = 311' - 321' is random disturbance term,

[3 = a2 «:1 is a coefficient vector and

98
F(Xifl) is a cumulative distribution function for pi evaluated at XiB. The

exact distribution of F depends on the distribution of the random term “i (if “i is
normal then F is a cumulative normal).
The marginal effect of a variable X]- on the probability of adopting the new
technology can be calculated by differentiating Pi with respect to )8 :

aPi/axij = f(XiB) - Bi, (4)

Where f(-) is the marginal probability density function of “i and j = 1,2,...J is
the number of explanatory variables.

Amemiya (1981) also defined a dichotomous random variable y which takes
the value of 1 if the event occurs and 0 if it does not. It is also assumed that the
probability of an event depends on a vector of independent variables X and a
vector of unknown parameters 6.

The general form of the univariate dichotomous choice model is expressed as

Pi = Pi(yi = 1) = G(Xi,6), (i = 1,2,...n) (5)
Equation (5) states that the probability that the ith farmer will adopt a technology
is a function of the vector of explanatory variables Xi, and the unknown parameter
vector, 6. The functional form of G is unknown and it is too general for practical
applications. In order to specify G explicitly, three alternative functional
relationships are commonly used by researchers : Linear Probability (LP), probit

and logit models.

99
5.1.1 Windham

One option of specifying G in Equation (5) is by the linear probability
functional form which has been frequently used in econometrics applications. The
general equation used in the LP model is the form:

yl- = X'iB + el- (6)
Where yi, the dependent variable takes the value of 1 if farmer i adopts a
technology and

0 if the farmer does not adopt.

Xi is a matrix of independent variables (regressors) with n observations

and k estimable coefficients;

[3 is a k x 1 vector of parameters and

el- is the ith identically and independently distributed random disturbance

with zero mean.

In this model the observed dependent variable is assigned either the value 1
or 0 and the OLS technique is used to estimate the unknown parameters. While
the LP model is computationally and conceptually easier than the logit and probit
models, its specification creates estimation problems and the non-normality of the
disturbance terms makes the use of traditional tests of significance (the t-test and
F-test) inappropriate. Akinola (1988) explains the factors contributing to these
problems and the limitations of the LP functional form as:

1) it gives a heteroscedastic regression model and that its variance-covariance

matrix varies systematically with the independent variables,

100

2) the predicted value of X’B is not constrained to lie between 0 and 1 which is
inconsistent with the definition of yl- as a conditional probability. This is
because the linear function is unbounded and there are certain sets of values
of the regressors (Xis) for which yi = X’iB exceeds 1 and others for which it is
less than 0,

3) studies have shown that adoption decision functions are curvilinear rather
than linear. The OLS could not be used because it would lead to inefficient
parameter estimates (Judge et al. 1988, Kennedy, 1992).

1 5.1.2 mg; and Probit Models

In an attempt to constrain the estimated probabilities between 0 and 1 range,
alternative functions are developed. The two most popular are the logistic and
the cumulative normal functions creating the logit and probit models respectively.
Univariate and multivariate logit and probit models and their modified forms
have been used extensively to study the adoption behavior of farmers. The
probability that a given farmer will adopt a new technology can be expressed as a
function of :

P (Y = 1) = F(XB) as in equation (6). According to the logit model, the
probability of an individual household adopting a new technology, T2, given
economic, social and physical characteristics (X) can be specified as:

P(T2IX) = exr) (XB + e) / {1 + eXP (X13 + 6)} (7)

where e is a random disturbance term

101

The probability of continuing to use the traditional technology (i.e., not adopting
the new technology is then:

P(T1|X) = 1 - P(T2|X)
1-[eXP(XB +e)/{1 + exr3(XB +e)}

1 / {1 + eXP (XB + e)} (8)

The relative odds of adopting versus not adopting a new technology are given by

P(T2 I X)’ P(T1 I X) = [CXP(XB+e)l {1+6XP(XB+e)}/ll+eXP(XB+e)l

= exp (XB + 6)
Taking the logarithm of both sides:
In [sz lX)/P(T1IX)l = xn + e (9)

The basic merit in the logit model is that the parameter estimates are linear
and thus estimation is computationally simpler (Kennedy, 1992). The estimates
only require transformation in exponential form to arrive at the probabilities.
Assuming a normally distributed error term (E) the logit maximum likelihood
(LML) estimation procedure is used to obtain consistent, efficient and
asymptotically normal estimators. The estimates will indicate the effects and
significance of the independent variables on the adoption of specific technology
(Judge et al.1988).

Probit analysis is a procedure that takes account of heteroscedasticity of the
disturbances and restricts predictions to values between 0 and 1. In the probit
model, the probability of observing a response in this study whether a farmer

adopts improved wheat production technologies or not, is defined in terms of the

102

level of an observed index. The standard cumulative normal distribution is used
to transform the index into probability value. The relationship between the index
and the probability follows a sigmoid curve. The index may take on any value
between -oo and + co, but the transformation ensures that all corresponding
probability values lie between 0 and 1.

The ftmctional form can be expressed as:

 

2
F(w)=¢(w)= _W 1 e-: du.

(12—fa:

Where (w) = xie (10)
The linear equation in (6) is used and parameters are estimated using the
Maximum Likelihood Estimationn. An estimated 8 value in a logit or probit does
not estimate the change in the dependent variable due to a unit change in the

relevant independent variable. The effect of an explanatory variable on the
dependent variable is given by the partial derivative of the expression for Prob

(y= 1) with respect to B which is not equal to 8.17

 

17The marginal effects of the independent variables in a logit model is calculated as:
[Pron = 1)][1 - Prob(y = 1)B, which is usually reported by estimating it at the mean value of the
explanatory variables. This formula is appropriate for continuous variables only and it is

recommended to measure the effects of categoric variables by estimating the difference before and
after the change (i.e. when X=1 and X=0).

103
5.1.3 Multinomial Logit and Prpbit Mpdels

The preceding section addressed the problem of binary, or dichotomous
variables for which there are only two choice categories. Categorical variables
that can be classified into many categories are called polychotomous variables. A
farmer may be presented with a choice of planting one of the three wheat
varieties or choices of different levels of fertilizer application (from the lowest to
the highest rate). Estimation of the probability of adoption of any of the varieties/l

or rate of fertilizer use can be undertaken by means of a generalization of the ) ,

‘a- "r
I

/' ‘\

binomial logit or probit models, called, respectively the multinomial logit and th
multinomial probit models. These generalizations employ the random utility
model where the utility to an individual farmer of an alternative is specified as a
linear function of the characteristics of the farm household and the attributes of
the alternative plus an error term.

The probability that a farmer will choose a particular alternative can be given
by the probability that the utility of that alternative to that farmer is greater than
the utility of all other available alternatives. If the error terms are assumed to be
independently and identically distributed we will end up with the multinomial logit
model. If however the distribution is assumed to be distributed multivariate—
normally, the multinomial probit model results. Kennedy (1992) discusses that
multinomial logit model has a disadvantage of the independence of irrelevant
alternatives property. When an identical alternative is included in the choice set,

one would expect that as a result the probability from this model of choosing the

104

duplicated alternative would be cut in half and the probabilities of choosing the
other alternatives would be unafiected. The multinomial logit model does not
overcome such a problem and it will be inappropriate to use the multinomial logit
whenever two or more of the alternatives are close substitutes. The probit model
allows the error terms to be correlated across alternatives and thereby
circumvents the problem of the independence of irrelevant alternatives mentioned
earlier. Some multinomial-choice variables are inherently ordered and the
individual producer or consumer has to make the selection from the ranked sets
of 'choices. In this study examples of ordered choices include different rates of
input use (i.e. no use, very low, low, medium and high levels). Another example
is farmers’ opinions related to different technologies ranked in order of
preference. The ordered probit and logit models are used for analyzing such
ordered responses. According to Greene (1990), such a model is built around a
latent regression in the same manner as the binomial models such as:

Y* = B ’X + e
where y: is unobserved and we observe :

y=0ifY* $0, (11)

= 1 if 0 s Y' < ”1

=1 ithsY’.

105

Where the alternative choices are represented by j = 0,1,2,..J,
the n’s are unknown parameters to be estimated with [3,
the e is normally distributed across observations and mean and variance of e
are zero and one.
Given the normal distribution we have the following probabilities for each of the
choices:
Prob[Y=0] = <1>(-B’X),
Prob[Y=11 = «>(tt1-B'X)- ¢(-B'X). (12)

Prob[Y=2l = 4W2 - B'X)- @011 - B’X)

Prob[Y=J] = 1-<I>(nJ_1 - B'X).
Where ID is the cumulative standard distribution function.
In order for the probabilities to be positive the following must hold: 0< '11 < [.12 <

. < u. J-1' The maximum likelihood techniques are used to estimate the

coefficients and calculate the marginal effects of the regressors on the
probabilities of the different alternatives.

5.1.4 The Tphit Model

Once the models are used to predict the probability of adoption of a certain

technology, it will be useful to estimate the intensity of adoption by different
categories of farmers. The degree of adoption can be divided into very low, low,

moderate and high. Depending on the type of technology a specific level of

106

quantity or range will be defined to indicate the quantity of use or the intensity.
In their recent study, Adesina and Zinnah (1993) used the Tobit model to
measulgbmh the probability of adoption and the intensityof use_of modern
mangrove rice varieties in Sierra Leone. Shakaya and Flinn (1985) applied the
Tobit model for analyzing the adoption of modern rice varieties and fertilizer use
in Nepal and referred the Tobit model as a more general case of Probit model or
a "hybrid Probit". Sureshwaran et al. (1992) and Norris and Batie (1987) also
employed Tobit analysis to study the intensity of adoption and factors affecting
soil conservation decisions by farmers in the Philippines and Virginia, respectively.
Following the linear probability model (equation 6), where the probability of
technology adoption is given as a function of a vector of explanatory variables and
of unknown parameters and error term:

Pi = Fi(XiB), the functional form of F can be specified with a Tobit model, where
pi is an independently, normally distributed error term with zero mean and

constant variance 02:

Yi =xip,iri’=xip+tti>r (13)
=0ifi’=xip+ pisT
Where Yi is the probability of adopting (and the intensity of use of) the
technology (fertilizer or improved wheat varieties); i: is a non-observable latent
variable, and T is a non-observed threshold level. The Tobit model (Tobin, 1958)
measures not only the probability that a farmer will adopt the new technology

(the decision to adopt), but also the intensity of use (the effort to continue using
if .
/_ i t,

1‘ ’ f .4" ‘ \‘J
'l y 'l ‘ ‘.

, U I“ 7 CC

107

the technology) of the technology once adopted. The above equation is a
simultaneous and a stochastic decision model. If the non-observed latent variable
i‘: is greater than T, the observed qualitative variable Yi becomes a continuous
function of the independent variables, and 0 otherwise. The value of a Tobit
coefficient does not represent the expected change in the dependent variable
given a one unit change in an explanatory variable. The model estimates a vector
of normalized coefficients which can be transformed into the vector of first
derivatives (Adesina and Zinnah, 1992). McDonald and Moffitt (1980) explain
that Tobit effects can be decomposed into a) changes in the probability of being
above the limit and b) changes in the value of the dependent variable if it is
already above the limit. Given such a decomposition we can use the means of the
explanatory variables to calculate the elasticity of adoption and elasticity of
intensity /effort once adoption occurs. The model below illustrates the
decomposition process and the transformation of the coefficients into a vector of
first derivatives and subsequent calculation of the elasticity of adoption.
The basic Tobit model (as in equation 13) is

Y=XB+eifXB+e>0

Y = 0 if X13 + e _<. 0
The expected value of Y in the model (Tobin, 1958) is given as:

Em = xs1=<z)+ are),
Where 2 = XB/o, f(z) is the unit normal density, and F(z) is the cumulative

normal distribution function.

108

The expected probability of being above a certain value once adoption occurs is
given as

E(Y‘) = E(YIY> 0),

E(Y") = xp + o f(z)/F(z)18
The relationship between the expected value of all observations E(Y), the
expected value conditional upon being the limit E(Y'.) and the probability of
being above the limit F(z) is represented as

E(Y) = F(z)*E(Y’). (14)
To calculate the marginal effects of the explanatory variables and to decompose
the total effect we differentiate the following:

aE(Y)/ax = F(z)[aE(Y‘)/ax1 + E(Y‘)[aF(z)/ax1
Where, E(Y*)/8X represents the effect of the variable X on the intensity of use
and aF(z)/6X is the effect on the probability of adoption.
Manipulating the above equation (multiplying both sides by X/E(Y)) we can
calculate the total elasticity of change due to changes in the level of any of the

explanatory variables.19 Adoption elasticity in this study can be decomposed

 

18McDonald and Moffitt(1980) reported that the expected value of Y for the observations
above the threshold or limit here referred as E(Y ) is XB plus the expected value of the truncated
normal error term.
E(Y ) = E(YlY > 0)
= 13008 > -XB)
= X8 + af(z)/F(z).
E(Y)= F(z)E(Y)

19 Given that E(Y)= F(z)E(Y )1. then F(z) = ram/E(Y" ) and E(Y ) = E(Y)/F(z)
Given dram/ax= F(z)[aE(Y )/ax1 + E(Y )[6F(z) /ax1 and multiplying by X/E(Y)
{6E(Y)/aX}X/E(Y)= F(Z){aE(Y )JBX}X/E(Y) ‘l' E(Y ){3F(Z)/3X}X/E(Y)
and substituting for F(z) and E(Y ) we get
aE(Y)/aX}X/Em= E(Y)/E(Y )(aE(Y )/aX}X/E(Y) + EM/F(z){aF(z)/aX}X/B(Y)

109

into a) the change in the elasticity of the use intensities of technologies and b) the
change in the probability of being an adopter.
The elasticity equation is:

{6E(Y)/6X}X/E(Y) ={aE(Y*)/6X}X/E(Y*) + {aF(z)/aX}X/F(z) (15)

5.2 Empirical Model Specification

The probability of adoption of HYVs, fertilizers, herbicides and agronomic
practices for wheat will be specified as a function of economic, social, and physical
variables and it is hypothesized to depend on a wide range of explanatory
variables. Three kinds models will be used to study the probability of adoption,
probability of choosing an alternative and to measure the intensity of adoption.
The binomial logit and probit models will be employed to study the factors
affecting the adoption of fertilizers, herbicides and improved wheat varieties. The )_
ordered probit model will be used to determine the variables explaining the I
adoption of different rates of fertilizer applications and different wheat varieties
in the study areas. The third model, the Tobit model will be used to estimate
both the probability of adoption and the level of intensity of fertilizer application

as this input appears to vary among and across farmers in the region.

 

aEtYl/aX}X/E(Y) = {aE(Y’)/aX}X/E(Y’> + {aF(z)/aX}X/F(z)

110

The data on which the empirical models are based is discussed in chapter 3.
The estimated empirical models, derived from the binomial logit model
(equations 7-9), ordered probit model (equations 11- 12) and the Tobit model
(equation 13 ) are presented below. The various farm and farmer specific

characteristics and the associated variables are also listed in Table 5.1.

5.2.1 The Binomial Logit Model
P(T2|X) = CXP (X13 + e) / {1 + 6X1) (X11 + 8)} (7)
The probability of continuing to use the traditional technology T1 (i.e. not
adopting the new technology is then:
P(T1|X) = 1 - P(T2|X)
1 - [exp (X3 + e)/{1 + CXP(XB + 8)}

1/{1 + CXP (X11 + e)} (8)

The relative odds of adopting versus not adopting a new technology are given by
P (TZIX)IP(T1IX) = [exp (X9 + 6)] {1 + exp (X13 + e)}’[1 + CXP(XB + 6)]
= exp (XB + 6)
Taking the logarithm of both sides:
In [P (T2 IX) /P(T1 00] = xv) + e (9)
5.2.2 The Ordered Probit Model
Y=0 iiY‘ 50, (11)

1if0sY’< 111

2 ifttle’<n2

111

= J if a“ s Y‘ .
Prob[Y = 0] = <I>(-B’X),
Prob[Y = 1) = «>61 -B'X) - «(1)20, - (12)

Prob[Y = 2] = @012 - B'X)- @011 - B'X)

Prob[Y = J] = 1-<I>(uJ_1 - B'X).
52.3w
Yi =xib,ifi"=xip+tti>r (13)
=0ifi" =xip+ uisT
orY=XB+eifXB+e>0
Y = 0 ifXB + e s 0
The elasticity of adoption is given by equation 15 as discussed earlier:

{aE(Y)/aX}X/E(Y) = {aE(Y")/aX}X/E(Y’) + {aF(z)/aX}X/F(z) (15)

112

Table 5.1 List of Economic, Social and Institutional Variables Affecting Adoption

Decision

 

 

 

 

Institutional Factors

 

 

Fertilizer availability, use,
rates, farmers’ perception on
advantages and disadvantages
of fertilizers

Herbicide availability, use,
rates, farmers’ perception on
advantages and disadvantages
of herbicide

Credit availability, source and
purpose

Extension service, farmer
training, Field days and
demonstrations

Information access

Distance to nearest market,
town

Own radio

Listen to ageducation

programs

GTF'IMLKWRFI'RT, FI‘AWYR,
FTUSY R, USFI'YRL, RNNOFAPl,
RLFI‘APIADVFI‘I, DADVFTI,
APHBW90, THKHBDAV

GETCRDT, SRCCRDTI,
SRCCRDTZ, PRPCRDTl,
PRPCRDTZ

DAVST90, ATFLDD, ATFTRNG,
NRMKKMFI'DSTKM, OWNRDIO,
LSNAGED

Category Variable Name Variable Code Measureme-
nt/Value
Economic Factors Farm Size per HH FARMSZHA Hectares
Arable Area per HH ARBLHA Hectares
Oxen owned OXNUM Numbers
Cow owned CWNUM numbers
Sheep owned SHPNUM Numbers
Wheat Production per DSHPRODKG, ENKPRODKG Quintals
Variety BLKPRDKG Quintals
DSHYLDKGENKYLDKGBLKYL Kg/ha.
Yield of each Variety DKG Kg/ha.
DSHPRQ, ENKPRQ Kg/ha.
Wheat prices by variety BLKPRQ ETB/qnt
ETB/qnt
Social Factors Age of Household Head(l-IH) AGEHH Years
Sex of HH SEXHH M/F (1,2)
Level of Education of HH LVLEDHH (14)
Farming Experience FARMEXP Years
Family Size FMLSZ Numbers
Crop production technology Seed availabilty and sources SRCWSD1,SRWSD2 Yes/No (1,0)
applications Reasons for not using RDSHPRFI, RENKPRFI, Coded Choices
improved seeds RBLKPRF1
sources of improved seeds RDSHPRFZ, RENKPRFZ,
Varietal preference criteria RBLKPRF‘Z

Distance in Km.
Yes/No

 

 

 

 

 

113
5.2.4 Dependent and Independent Variables

The preceding sections highlighted the econometric theory and the
conceptual models underlying the proposed analysis and the models to be used.
This section attempts to define the dependent and explanatory variables used in
the binomial logit, ordered probit and Tobit models discussed above. The
dependent variables used for the adoption models estimated in the study:

a) Fertilizer adoption or no adoption in 1990 and 1991 crop seasons (FW90 and
FW91) Yes/No response coded 1 or 0 ,

b) Herbicide application on no application in 1990 and 1991 (APHBW90 and
API-IBW91) Yes/No response coded 1 or 0,

c) Farmers’ choice among the three widely used varieties of wheat, i.e., Dashen,
Enkoy and Bulk (DSHHAP, ENKHAP and BLKHAP coded 0,1,2,
respectively) ,

d) Fertilizer application rates group (FAPGP) the farmer belongs: response
coded 0 if fertilizer application rate less than 25kg/ha.,

1 if greater than 25 and less or equal to 50 kg/ha.,

2 if greater than 50 and less or equal to 75 kg/ha., y K i

3 if greater than 75 and less or equal to 100 kg/ha. and 1"

4 if greater than 100 kg/ha.

e) Fertilizer adoption and intensity of use: Quantity of fertilizer used per farm
and rate of fertilizer use in the two crop seasons (WDAPKG90, WDAPKG91,

FRTRW90 and FRTRW91) and the response are coded:

114
0 if no fertilizer is used,

total quantity reported in kg.,

and rate of fertilizer application (kg/ha.) calculated by dividing total quantity

by wheat area.

The binomial logit model is used for the binary responses in a and b above
while for c and d the ordered probit model is employed to estimate the empirical
model. The Tobit model is applied to investigate both the probability and

intensity of fertilizer adoption.

Tabl

Var)

”‘1
'fl
1
<1
. :1
II

 

 

115

Table 5.2 Technology Types, Econometric Models, Dependent and Independent
Variables Used.

 

 

 

 

 

 

Type of Econometric Dependent Independent Variables
Technology Models Used Variable(s)
Fertilizer Use Binomial FW90, AGEHH,LVLEDHI-LFMLSZ,
(Y es/No) Logit (Fertilizer use FARMSZHA,OXNUM90, GTFTML,
in 1990/91) GETCRDT, KWRFI‘RT,
USFI'YRL, APHBW90, DAVST90
Herbicide Use Binomial APHBW90, AGEHH,LVLEDHI-LFMLSZ,
(Y es/No) Logit (Herbicide FARMSZHA, OXNUM90,GETCRDT,
application) FW90, DAVST90
Choice of Wheat Ordered WHVCH AGEHH,LVLEDHI-I,FMLSZ,
Varieties Probit (Wheat FARMSZHA, OXNUM90, GTFI'ML,
(Dashen, Enkoy Variety GETCRDT, FW90, KWRFTRT,
and Bulk) Choice) USFI‘YRL, APHBW90,DAVST90
Fertilizer Ordered FAPGP90 AGEHH,LVLEDHI-I,FMLSZ,
Application Probit FAPGP91 FARMSZHA,O)G\IUM90,GTFI'ML,
Rates (0,1,2,3,4) (Fertilizer GETCRDT, KWRFI'RT, USFI'YRL,
rates groups APHBW90, DAVST90, DSHPHA90,
used) ENKPHA90, BLKPHA90
Fertilizer Tobit WDAPKG90, AGEHHLVLEDHH,FMLSZ,
Adoption and FRTRW90, FARMSZHA,OXNUM90, GTF'I'ML,
Intensity of use (Fertilizer GETCRDT, KWRFI‘RT,
(Kg/F arm, quantity and USFTYRLAPI-IIBW90, DAVST90
Kg/ha) rates) DSHPHA90,ENKPHA90,BLIG’HA90,

 

 

 

 

 

 

116
Explanatory Variables of Technology Adoption: Most technological adoption

studies have under-specified the models of adoption behavior by using only a few
decision variables. A farmer’s decision making process in adopting is complex and
should be examined taking into account economic, social, cultural and institutional
factors. In this study as indicated in Table 5.2 different categories of explanatory
variables are used to study the relationship between the farmer’s decision (to
adopt or not to adopt fertilizer and/ or herbicides, which wheat varieties to
choose, which rate of fertilizer to use and what quantity of fertilizer to apply and
their social and economic characteristics, farm specific factors and institutions that
affect adoption.

Farm size has been shown to positively affect adoption decisions and it is
hypothesized that farm size will have a positive sign in the empirical model.
Earlier studies indicate that older farmers tend to be risk averse and are reluctant
to adopt new production techniques. Younger farmers have access to education
and appear more knowledgeable about new production technologies. It is then
hypothesized that farmer’s age (AGEHH) is negatively related while level of
education of household head (LVLEDHH) is positively related to adoption.
Formal schooling enhances the farmer’s ability to perceive, interpret, and respond
to new events in the context of risk. Several empirical studies have verified the
link between higher levels of educational attainment and early adoption of new
technologies. A large family (FMLSZ) is expected to have a large number of

working members and this might positively influence the use of labor using

117

technologies and negatively affect the adoption of labor saving technologies.
Adoption of herbicides is expected to be related negatively with family size as
family labor (for hand weeding) can substitute for herbicide use.

In most cases adoption of new technologies requires the purchase of farm
inputs such as fertilizers, improved seeds, herbicides and hiring of farm labor
when family labor is not adequate. The role of credit in accelerating adoption
and agricultural development is widely discussed. If farmers do not generate their
own savings or have no access to credit their ability to adopt new technologies will
be constrained. The variables that indicate credit availability, sources and
purposes of credit (GETCRDT, SRCCRDT, PRPCRDT ) and the wealth variable
oxen number (OXNUM90, OWUM91) are used as proxies for access to capital.
Availability of credit and number of livestock owned are expected to positively
affect adoption decisions.

. Farmers’ perceptions on fertilizer availability (GT'FI'ML), knowledge of
fertilizer recommended rates (KWRFI'RT), frequency of fertilizer use
(USFRTYRL) are important to enhance the adoption process. Proximity to
marketing centers will obviously reduce transport cost to acquire fertilizer and
other purchased inputs and sale of farm produce. Distance to the nearest market
(NRMKKM) is expected to negatively affect the adoption of purchased inputs.
Access to information is represented by extension agent visit (DAVST90),
farmer’s attendance at farmer training programs, field days, ownership of radio,

farmers’ listening to agricultural education programs (ATFTRNG, ATFLDD,

118
OWNRDIO, ISNAGED). Contact with extension and farmers’ access to these

sources of knowledge will stimulate adoption.

Preference to either of the three wheat varieties is assumed to depend on
their adaptability in each district, respective yields, prices and specific agronomic
characteristics ( AWRAJA, DSHYLDKG, ENKYLGKG, BLKYLDKG, DSI-IPQ,
ENKPRQ, BLKPRQ, RENKPRFI, RBLKPRFl). The use of complementary
inputs such as fertilizer and herbicides is also an important adoption decision.
(FW90, FW91, APHBW90. APHBW91). It is hypothesized that some farmers can
adOpt both inputs simultaneously, while others adopt fertilizers first and then
adopt hebicides at a latter stage, an illustration of the ’step-wise adoption
behavior’ of farmers that Byerlee and Polanco found in Mexicozo.

The complete empirical models specified as above and their dependent
independent variables defined in Tables 5.1 and 5.2 are listed below.

I. Binemial Qgit Mpdel

FW90 = BO + BIAGEHH + BZLVLEDHH + [33sz + B4FARMSZHA

+ BSOXNUM90+ B6GTFTML + B7GTCRDT + BSKWRFI‘RT +

89APHBW90 + B 10DAVST90

APHBW90 = [30+ BIAGEHI-H BZLVLEDHH+ B3FMLSZ+

B4FARMSZHA+ BSOXNUM90 B6GTFTML + B7GTCRDT +

BSKWRFTRT + pgrwslo + plonAvsruo

 

20Byerlee and de Polanco (1985) found out that because of capital scarcity and risk
considerations, Mexican farmers did not adopt a given package of technologies, rather they adopted
individual components of a package of technology. Adoption followed a stepwise process, i.e.,
farmers first adopted improved barley varieties, fertilizer and then herbicide.

119
II. RDE D PROBIT M DEL

FAPG90 = 1’0 + BIAGEHH + BZLVLEDHH + [33sz +
84FARMSZHA + [SSOXNUM90 + [36GTFI‘ML + [37GTCRDT +
BgKWRFTRT + BgAPHBW90 + BIODAVST90

In. TQBIT MQDEL

WDPKG90 = (30 + BIAGEHH + BZLVLEDHH +B3FMLSZ +
B4FARMSZHA + BSOXNUM90 + B6GTFI'ML + B7GTCRDT +

BSKWRFI'RT + 89AP1~IBW90 +810DAVST90 + BllNRMKKM

+ filZWHVCH + 813ATFLDD + [314LSNAGED

120
5.3 Adopter Categories

One of the objectives of the study is to classify farmers into adopter
categories based on adoption pattern and socioeconomic characteristics.
Incidence of adoption is represented by the use or non use of fertilizer and
herbicide, and intensity of input use are used to classify the farmers to different
adopter categories. The first stage dealt with users and non users While the
second stage attempted to identify farmers into : low adopters, moderate adopters
and advanced adopters. Farmers using less than 25 kg of fertilizer per hectare
and not using herbicide at all are categorized as low adopters. The moderate
adopters category are farmers who apply 50 - 75 kg of fertilizer per hectare and
also have used herbicide but not on a regular basis. Advanced adopters are those
who apply fertilizer at a rate of 75 kg/ha or more and at the same time apply
herbicide regularly if available. The survey data on farmers’ use of herbicide and
fertilizer is then used to identify the sample farmers into the different groups of
adopters.

5.3.1 Diseg'minant Analysis

The general objective of this study has been to identify the socioeconomic
factors that are responsible for technology adoption in the region. In this section
a statistical model is developed based on farmers’ socioeconomic characteristics
and institutional factors to see if the variables can discriminate statistically
between the different adopter categories. Discriminant analysis is one of the

statistical techniques often used to identify the characteristics that differentiate

121
observations in one group from those in another group. In discriminant analysis a
linear combination of independent variables is formed to serve as a basis for
assigning sample cases to groups. Information contained in multiple independent
variables is sWarized in a single score (Norusis, 1990). The technique suggests
that a variable having the largest absolute discriminant weight possesses the
greatest power in discriminating one group from another. The discriminant model
used to classify farmers into different adopter categories is :
Dg = BO + 131 x1 +132x2 +133x3 +B4X +. .. + 3an (16)

Where Dg = Discriminant score for each group (D1, Dm, Da for low, moderate
and advanced adopter category), B’s are coefficients to be estimated from the
data and the X’s are the values of the independent variables.

After the discriminant function is determined, the independent variables for
each observation are multiplied by the discriminant coefficients to obtain a single
discriminant score. The score is then used to classify each case to one of the
categories of adopters. The results of the discriminant analysis in which the
standardized coefficient, Wilk’s lambda, the ratio of the with-in groups sum of
squares to the total sum of squares and the level of significance for the estimates
by district and the region are presented in the empirical results chapter. The
interpretation of the coefficients is similar to that in multiple regression.

Variables with large coefficients are thought to contribute more to the overall

discriminating function. The relative importance of the variables can be

122

21. Large values of lambda show that

measured by the standardized coefficients
group means do not differ significantly while smaller values indicate that the
group means are different. Lambda only gives a test of the null hypothesis that
the population means are equal.‘ It is important to note that even though Wilk’s
lambda may be statistically significant it does not provide adequate information
on the effectiveness of the discriminant function in classification. Comparing

actual and predicted classification of the cases into different categories is used to

test the effectiveness of the model.

 

21The magnitude of the unstandardized coefficients is not a good index of the relative
importance when the variables differ in their units of measurement. Coefficients are standardized to
adjust for the unequal means and standard deviations of the independent variables. In addition the
signs of the coefficients are arbitrary and only determine which variable values result in large or
small discriminant score (Dg)' For details and procedures in discriminant analysis see Norusis
(1990).

CHAPTER 6
EMPIRICAL RESULTS

This chapter is divided into three sections. The first section examines the
profitability of wheat production technologies by presenting results of partial
budgets and marginal returns analysis. The second section describes the
econometric analysis of adoption in the study areas. The binomial logit, ordered
probit and Tobit analysis and the respective results of the models will be
presented. Moreover, the determinants of adoption and the relative contributions
and effects of identified and significant variables affecting adoption are examined
in the chapter. The third section deals with the results of the discriminant analysis.

6.1 Profitability Analysis of Technologies

6.1.1 Mammalian

A partial budget summarizes the value of cost and revenue items that would
change when one production activity or several production activities are replaced
by another. Costs and returns that do not vary from one activity to another are
not included in the analysis as they do not affect the profitability of adopting an
activity. The decision to invest in production technologies (such as purchasing
improved wheat varieties, fertilizers and/or herbicides) depends on the farmer’s
expected increase in yields and reduction in production costs. Partial budget

analysis will consider the added benefits (increased revenues and reduced costs),

123

124

and increased costs and reduced revenues if any as a result of adopting all or part
of a given package of production technologies.

In the study areas almost all farmers reported using improved wheat varieties.
The intensity of fertilizer and herbicide use (as discussed in chapter 4) varies
among farmers and across districts. Farmers’ decisions to use or not to use these
inputs depend on the expected net benefits from each level of input use. When
making adoption decisions, small farmers will consider not only an increase in
expected net benefits and/or profit but also want to make sure that family food
security is attained. Empirical studies have shown that subsistence farmers are
risk averse and hence tend to adopt technologies that will reduce production
and/ or income variability. Farmers continue to use traditional production
practices which they are sure will secure them the minimum level of production to
sustain the family’s well-being. Small farmers employ a safety-first strategy before
making adoption decision on new production techniques.

Economic analysis of crop responses to fertilizers has been carried out by
IAR scientists in the different agro-ecological zones of the country. In Arssi
CADU/ARDU, MOA and IAR agronomists and recently agricultural economists
have conducted on-station and on-farm fertilizer, variety and other agronomic
experiments. The results of the research programs have helped to reorient on-
station research programs and develop a set of recommended production

technologies for farmers in the region.

125

Partial budget analysis is used to compare the profitability of alternative
production practices. Partial budget analysis for the three wheat varieties of
Enkoy, Dashen and Bulk in the five districts is presented in Table 6.1.

Gross benefits are calculated by multiplying adjusted yield (allowing a 10%
harvesting and storage loss) by the average farm gate prices22 that farmers

reported. The farm gate price of wheat is considered to reflect the opportunity

cost of seed used from farmers’ own previous harvests.

 

22Consumers prefer Dashen to other varieties (for its baking qualities), hence commands

higher price. Enkoy is the second preferred and has higher farm gate price except in Ticho.

126
Table 6.1 Partial Budget Analysis of Wheat Production Technologies Under

Farmers’ Practices in Arssi, 1990/ 91
—

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Keleta Chilalo Galema Gedeb Ticho
Item Unit Enkoy Dashen Enkoy Dashen Bulk Enkoy Enkoy Bulk Enkoy Bulk
Average Kg/ha 1217.00 1057.00 1330.00 934.00 1063.00 1441.00 1068.00 1025.00 805.00 965.00
Yield
Adjusted Kg/ha 1095.30 95130 1197.00 840.60 956.70 1296.90 961.20 92250 72450 86850
Yield (10%
Loss)
Crop Price EB/kg 0.93 0.93 0.66 0.76 0.64 0.69 0.64 0.62 0.57. 0.60
Gross EB/ha 1018.63 884.71 790.02 63886 612.29 894.86 615.17 571.95 41297 521.10
Benefits
Inputs and
Costs
Labor” Mandays/ha 64.73 6253 66.29 60.84 62.62 67.81 62.69 62.09 59.07 61.27
Labor Cost EB/ha 12429 120.06 127.27 116.82 12022 130.20 12036 119.22 113.41 117.64
Oxen Oxendays/ha 35.26 32.86 36.95 ' 31.01 32.95 38.62 33.02 32.38 29.08 31.48
Cost OXEN EB/ha 105.77 98.57 110.85 93.03 98.84 11585 99.06 97.13 87.23 94.43
Seed Rate Kg/ha 196.00 210.00 240.00 222.00 215.00 206.00 147.00 148.00 190.00 188.00
Seed Cost EB/HA 182.28 195.3 158.4 168.72 137.6 142.14 94.08 91.76 1083 112.8
Fertilizer KG/ha 6450 6450 81.00 81.00 81.00 81.90 59.00 59.00 66.90 66.90
Rate
Cost of EB/kg 0.91 0.91 0.91 0.91 0.91 0.91 0.91 0.91 0.91 0.91
Fertilizer
Fertilizer EB/ha 58.70 58.70 73.71 73.71 73.71 7453 53.69 53.69 60.88 60.88
Cost [ha
Total EB/ha 471.03 472.62 470.23 452.28 430.37 462.72 367.19 361.80 369.82 385.74
Variable Cost
Net Benefit EB/ha 547.60 412.08 319.79 186.58 181.92 432.14 247.98 210.16 43.15 135.36

 

 

 

aLabor and oxen use for harvesting and threshing is estimated using ARDU(1980) studies in which 1.375 Man days
and 1.5 oxen days per 100 kg yield are required and valued at their respective opportunity costs

127

Labor and oxen use are assumed not to vary either between the districts and
across the three varieties. Introduction of improved production practices will
eventually require additional oxen power and labor23 at least to harvest the
incremental yield. The opportunity costs for these inputs are the local wage and
rental rates, respectively. The average rate of fertilizer application is significantly
different across the study areas. Farmers do not use different rates of fertilizer
for the varieties. The variety Enkoy, grown in all the study districts gave the
highest and lowest yields in Galema and Ticho, respectively. It generated the
highest net benefit of 548 EB/ha in Keleta district during the 1990 cropping
season. The same variety also gave the second highest net benefit of 432 EB/ha
at Galema. Dashen gave the third highest net benefit of 412 EB/ha at Keleta.
The yields for both Enkoy and Dashen are not the highest at Keleta but a 40%
higher farm gate price resulted in the high net benefits indicated.

The variety Bulk had a net benefit range of 135 to 182 EB/ha in the three
areas. At Ticho Enkoy’s net benefit is the lowest (43.EB/ha) but in the other four
districts it out-yielded all the varieties and generated higher net benefits. At
Galema the only variety farmers grow is Enkoy where the farmers’ reported yield
was the highest. Farm gate prices and reported yields are the lowest at Ticho

contributing to lower net benefits. Variable costs (mainly fertilizer24,labor and

 

23 Based on the ARDU (1980) study, labor and oxen power requirement for the increased
yield is estimated as 1.375 Mandays and 150 Oxendays per 100 kg grain.

24Ethiopia has the potential to produce fertilizers but does not yet have a fertilizer
manufacturing plant. There is no subsidy for fertilizer. All fertilizer is imported and the retail prices
have generally been in line with the import parity price.

128

seed costs) range from 362 to 471 EB/ha. Given farmers’ estimate of yield, prices
and the lack of detailed data on labor, draft power use, the partial budget figures
have to be considered as proxy indicators of profitability of the farmers’ wheat
production practices.

Research and extension services have developed a set of recommended
production packages to increase crop production in the region. Despite more
than 20 years of regional agricultural development effort farmers have not yet
adopted all the recommended practices. Preliminary analysis of the data (as
presented in the descriptive statistics in chapter 4) reveal that about 21% the
sample farmers apply below 25 kg/ha and almost 48% of them use 50 or less
kg/ha of fertilizers. However, there is a significant variation amongthe five
districts. Chilalo and Galema farmers apply higher rates (81 kg/ha), followed by
Ticho and Keleta (65 and 70 kg/ha) while farmers in Gedeb use an average of 59
kg/ha. Improved seeds are widely used in the region but extension agents express
their concern on the quality of the improved seeds that farmers use. The varieties
released more than ten years ago have been mixed up with the local seeds and
farmers plant uncleaned seeds. The majority of the sample farmers (53%) use
their own seed from the previous harvest while 22% purchase seeds from the local
markets. Only 23% of the farmers purchase improved wheat seeds fiom their
local MOA office. Farmers in Chilalo (41%) are the major beneficiaries and 11%

of the farmers in Ticho bought seeds from MOA. In 1990/91 seasons about 77

129
and 79% of the farmers did not apply any herbicides at all, although herbicides

are considered very effective and economical in controlling weeds in the region.

130

Table 6.2 Adoption of Wheat Production Technologies in Arssi Districts, 1990/91.
—

 

 

 

 

 

 

 

Variety Keleta Chilalo Galema Gedeb Ticho
ENKOY
Average Yield kg/ha 1216 1330 1441 1068 805
Adoption Rate(% of 74 41 88 69 49
Farmers)
DASHEN
Average Yield kg/ha 1057 934 900
Adoption Rate(% of 6 51 2
Farmers)
BULK
Average Yield kg/ha 1333 1950 1025 965
Adoption Rate(% of Farmers) 15 2 28 43
Fertilizer Adoption Percent of Farmers
% of Farmers Not Using 13 3 14 13 26
% of Farmers Using 87 97 86 87 74
Rates of Fertilizer kg/ha
Less than 25 kg/ha 30 11 5 29 28
GT 25 LE 50 kg/ha 32 18 19 34 34
GT 50 LE 75 21 25 20 15 13
GT 75 LE 100 11 36 53 19 19
GT 100 6 10 3 3 4
Use of Herbicides
Not Using 52 47 96 100 92
Using 48 53 4 8

 

 

 

Source: Field Survey, 1991.

131
A recent study by the Ministry of Agriculture’s National Fertilizer and Inputs

Unit (NFTU) supported by FAO Fertilizer Program reports that fertilizer
consumption in Ethiopia has concentrated in Arssi, Gojam and Shewa regions
amounting to 75 % of the country’s fertilizer consumption. These regions produce
half of the total cereal production cultivating 45% of the national area under food
crops. In these regions, the study show that about a third of the farmers have
adopted fertilizer use, implying a 15 to 20 percent adoption rate at the national
level (Fischer and Shah, 1990). In order to select and subsequently recommend a
viable package of production practices for small farmers we need to compare
alternative technical packages using the partial budgets and marginal rate of
return analysis. Analysis for the farmers’ reported production practices is
presented in the last table.

The MOA /FAO study on crop yield response and fertilizer policies
estimated fertilizer response functions (using a quadratic polynomial response
curve)25 for six cereal crops including wheat, and three categories of oilseeds
and pulses for the different agro-ecological zones of Ethiopia. The estimated
models provide a useful set of values for important parameters that can be used
in arriving at economically optimal levels of fertilizer. According to the wheat

yield response function, a maximum yield of 3227 kg/ha is attained at the level of

 

25Quadratic polynomial response curve is given as: y = A + B‘F + C"F2
where y is yield per ha., F is fertilizer use per ha. and A,B,C are yield function parameters. Further
mathematical proofs show that parameters A,B and C measure maximum yield, the share of yield
explained by fertilizer use and maximum level of fertilizer use respectively. For further details see
Fischer and Shah, 1990.

132
143 kg/ha fertilizer. Fertilizer use contributed 75% of the increase in wheat yield.

The economically optimal rate is 131 kg/ha. The following Tables (6.2 and 6.3)
show the partial budget and marginal analysis of wheat yield response under zero
level and six incremental levels of the most commonly used and available fertilizer
(DAP) in Arssi. Farmers’ use of fertilizers ranges from none to more than 100
kg/ha. Accordingly farmers’ levels of fertilizer use are grouped into:

a) 0 - 25 kg/ha,

b) 26 - 50 kg/ha,

c) 51 - 75 kg/ha,

d) 76 - 100 kg/ha and

c) more than 100 kg/ha. The grouping is similar to the levels indicated in
Tables 6.2 and 6.3. Other on-farm trials have used different rates and sources of
nitrogen and phosphorus.26 A response function estimated from fertilizer and
variety on-farm trials recommended for the region’s predominant soil type and
color is employed to estimate the wheat yield for the different levels of fertilizers
used in the analysis. The yields range from 1582 kg/ha for the control plot to
2776 kg/ha for the 150 kg/ha fertilizer level. The highest rate of 200 kg/ha gave

a yield of 2739 kg/ha as shown on the response curve of Figure 3.

 

26For more detailed explanations on alternative sources of Nitrogen and Phosphorus for wheat
producing areas see Tanner et al " Developing Technologies to Improve Soil Fertility, Weed Control
and Wheat Varieties" In Franzel and van Houten(1992).

133
Table 6.3 Partial Budget Analysis of Wheat Yield Under Selected Fertilizer

Levels in Arssi, 1990
—

 

 

 

 

 

 

 

 

DAP Yield Gross benefit Fertilizer Cost Net Benefit
Kg/ha Kg/ha EB/ha EB/ha EB/ha
0 1582 965 0 965
25.00 2241 1367 24.43 1343
(45:11.5)
50.00 2420 1476 48.86 1427
(9 :23)
75.00 2563 1564 73.29 1490
(13.5345)
100.00 2670 1629 97.72 1531
(18:46)
150.00 2777 1694 146.58 1547
(27:69)
200.00 2739 1671 195.44 1476
(36:92)
ource: le ta 0 tam om e A A 1 er r an arrety on- arm‘
Trials for Southeastern Ethiopia

6.1.2 Marginal Analysis

Marginal rate of return (MRR) is the ratio of marginal net benefits to
marginal variable cost expressed as a percentage. In this particular study MRR is
calculated for different rates of fertilizer applications to determine alternative
recommendations for farmers. After estimating the net benefits of the different
fertilizer levels as shown in the partial budget analysis in Table 6.2, we need to

compare their relative profitability from the farmers’ perspectives. Marginal

134

analysis helps to determine the rate of increase in net benefits as investments
increase.
Boughton et al. (1990) list the four steps in marginal analysis:
1) dominance analysis (identification and elimination of inferior treatments)
2) construction of a net benefit curve,
3) calculation of the marginal rate of return (MRR) between treatments of
incremental cost,
4) comparison of MRR to the minimum rate acceptable to farmers (SO-100%).
The 200 kg/ha level has a higher cost of 195 EB/ha and a net benefit of 1475
EB/ha. The next lower level of 150 kg/ha has a higher net benefit of 1547
EB/ha with a cost of 146.58 EB/ha. The 200 kg/ha level is therefore a
dominated treatment when compared to the next lower rate. Moving from no
fertilizer application to 25 kg/ha gave the highest MRR(1546%) while the move
from 100 to 150 kg/ha treatment generated only 33% MRR. Although the net
benefit increases as one moves from the lower to the next higher rate, the MRR
is decreasing. The results of this analysis suggest that given current yield, wheat
and fertilizer prices, farmers’ application rates should not exceed the 100 kg/ha
level as the next higher level gives an MRR of only 33%, which is lower than the
generally accepted minimum return of 50 to 100%. Adoption of other production
practices to increase the yield, and changes in the crop-fertilizer price relationship,

are required to make fertilizer use economical and acceptable to small farmers.

135

Table 6.4 Marginal Analysis of Wheat Yields Under Different Fertilizer Levels in

Arssi, 1990/91
—

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Fertilizer Net Variable Dominated MRR
Level Benefit Costs
Kg/ha EB/ha EB/ha Yes/No %
200 1476 195 YES -147
150 1548 147 NO 33
100 1531 98 N O 168
75 1490 73 NO 257
50 1427 48 N 0 347
25 1343 24 NO 1546
0 965 0

 

 

136

 

RESPONSE CURVE FOR WHEAT

For Selected Fertilizer Roles

 

2800

26004

2000 7

1800‘

WHEAT YIELDS KG/HA

 

1600

 

 

 

1400 I I I I I T I I T
O 20 4O 60 80 100 120 140 160 180 200
FERTILIZER RATES KG/HA

 

 

 

Figure 6 Response Curve for wheat under selected fertilizer rates, Arssi, 1990/91

137

 

NET BENEFIT CURVE FOR WHEAT

For Selected Ferlilizer Rates

 

1600

 
 
  
  
 
 
 
  

1500“

14001

1300-

1200-

11007

NET BENEFITS EB/ HA

 

 

QOOI fi fi I I I I T I I
O 20' 4O 60 80 100 120 140 160 180 200

FERTILIZER COST EB/ HA

 

 

 

 

Figure 7 Net Benefit Curve for Wheat and fertilizer Response in Arssi, 1990/ 91.

Table 6.5 Adoption, Fertilized Area and Rates of Fertilizer Use in Arssi Districts,

1990

 

 

 

 

 

 

 

 

 

 

 

 

 

 

District Sample Adopters Average Area Application

Farmers Percent Fertilized Area Fertilized Rate

Number Per farm Ha Percent Kg/Ha
Keleta 103 99 0.79 97 89
Chilalo 54 100 0.64 100 91
Galema 65 100 0.88 97 94
Gedeb 20 95 0.71 98 69
Ticho 11 91 0.72 88 78
Arssi 344 98 0.78 98 90

EEEEErOp-Wlse F'ert'ilfir Usage Study, ng and Programmmm.

 

 

Tables 6.5 and 6.6 are compiled from MOA data on Crop-wise Fertilizer

Usage 1991 and Fertilizer Trials on Major Cereal Crops, 1986-89 published in

1991. The studies indicate that 98% of the sample farmers in the districts have

used fertilizer on their wheat fields. The average rate of fertilizer application for

the region is about 90 kg/ha.

The survey results report lower rates of adoption (87%) and 72 kg/ha rate of

fertilizer application. One possible explanation for the difference could be that

extension agents tend to report to their superiors at headquarters higher rates that

are closer to the recommended levels of 100-150 kg/ha. .

Table 616 Economic Optimum Rate of Fertilizer for Wheat Production in Arssi,

1986-89.

 

Table 6.6 Economic Optimum Rate of Fertilizer Levels for Wheat Production in

139

 

 

 

 

 

 

 

 

 

 

 

Arssi
Control Plot Average Yield Nutrient Level Increase in Profit VCRa
yield Range of Control Plot N P205 Yield EB/ha
k8/hal kg/ ha kg/ ha kg/ha
Less than 735 28 55 720 264 3.2
1100
1100 - 1700 1444 42 50 833 312 3.4
Greater than 2355 19 44 454 152 2.7
1700
A A ResUFs of F'er—tiliz_er Tue-1E Cdnductdd on Major Cerea‘

Crops(19861989), January 1991.
aVCR is Value Cost Ratio (the ratio of the value of yield increase to cost of fertilizer)

————————-

The economic optimum rates of fertilizer levels recommended for three
control yield levels indicate that farmers have to use about 136 kg/ha of DAP and
Urea combined to increase the less than 1100 kg/ha yield without fertilizer by 720
kg/ha. Use of 158 kg/ha fertilizer will increase yield from 1444 to 2277 kg/ha.
The potential areas where the control yield is greater than 1700 kg/ha are
recommended to use about 100 kg/ha fertilizer to push the yield level from 2355
to 2809 kg/ha. The value cost ratio of wheat yield value to fertilizer cost suggest
that farmers can benefit 2.7 to 3.4 Birr for every Birr they invest in fertilizer use.

6.1.3 Cemparispn of Alternative Whth Prpduetipn Peekeges
F our alternative wheat production packages will be compared to assess the

profitability of the technologies. The packages will also identify farmers’ adoption

140

categories and can be used to relate to specific farmer characteristics. The

packages identified have the following components:

a) HYV of wheat either retained or purchased from MOA office,
b) different levels of fertilizer use and
c) herbicide use or hand weeding.

Production Packages: The data from the study areas reveal that farmers’

production practices differ in terms of the levels of purchased inputs used in the .

production process. Four possible production practices are identified for the

analysis:

1) Low input package (LIP): in which farmers use 50 or less kg/ha of fertilizer,
own/retained wheat seeds, and family labor for hand weeding (no herbicides
are used);

2) Limited Inputs Package (LMIP): use of 50 - 75 kg/ha of fertilizer, purchased
wheat seeds and no herbicides but hand weeding using family labor;

3) Medium Inputs Package (MIP): 75 - 100 kg/ha of fertilizer, purchased seeds
and herbicides

4) High Inputs Package (HIP): greater than 100 kg/ha of fertilizer, purchased
seeds and herbicides.

Previous studies undertaken by CADU/ARDU indicate yield differentials of

23.9 %, 15% and 32% between farmers not weeding, hand weeding only, late

application of herbicides and early spraying of herbicides, respectively. Seed

purity is estimated to affect the yield level at least by 200 kg/ha (Tanner and

141

Giref, 1991). Calculations of the expected benefits and costs are presented in
Table 6.7. It is assumed that the averages for purchased improved Wheat,
fertilizer and herbicides are: 0.93 EB/kg, 0.91 EB/kg, 0.29 EB/lit, respectively.
Farm gate wheat prices are assumed to be 0.71 EB/kg while the opportunity cost
of labor for weeding at the peak season is imputed at 3 EB/day. Based on these
assumptions gross benefits, variable costs and net benefits are calculated for the
four alternative packages designated as LIP, LMIP, MIP and HIP reflecting the
intensity of input use. The HIP alternative gave the highest net benefit per ha
(1713 EB) while the minimum/low input LIP produced a net benefit of 1158

EB /ha. Prices, levels of fertilizer use, seed quality and source, use of herbicide
and time of application have contributed to the apparent difference in the yields,
costs and subsequently in the net benefits shown in Table 6.7. Similarly the
marginal rate of return analysis confirms that the MIP yields the highest MRR of
437% over the LMIP alternative. Farmers who have better access to resources
can invest in the HIP production alternative and earn 208% MRR. The
availability of high yielding varieties, level of input use, crop, fertilizer and
herbicide prices will obviously affect the net benefits and the marginal rates of
return of each alternative package. The magnitude and the relative impact of

possible and assumed changes can be calculated using sensitivity analysis.

142

Table 6.7 Partial Budget and Marginal Analysis of Alternative Production

Packagesa for Wheat Farmers in Arssi

—

 

 

 

 

 

 

 

 

 

 

Item UNIT LIP LMIP MIP HIP
Average Yieldb KG /HA 2330 2765 3284 3468
Adjusted Yield (10% KG /HA 2097 2488 2956 3121
Loss)
Crop Price EB /KG 0.70 0.70 0.70 0.70
Gross Benefits EB /HA 1468 1742 2069 2185
Inputs and Costs
Laborc MANDAYS/HA 32 38 46 48
Labor Cost EB /HA 96 114 138 144
Seed Rate KG /HA 240 206 190 175
Seed Cost /ha‘l EB/HA 168 191 177 163
Fertilizer Rate KG /HA 50 75 100 150
Cost of Fertilizer EB/KG 0.91 0.91 0.91 0.91
Fertilizer Cost/ ha EB /HA 45 68 91 136
Herbicide Rate /ha LITER/HA 0 0 1 1
Herbicide Unit Cost EB/HA 0 0 29 29
Herbicide Cost /ha EB /HA 0 0 29.00 29
Total Variable Cost EB /HA 309 374 435 472
Net Benefit EB/HA 1158 1368 1634 1713
MRR % 326 437 209

 

3‘ Alternative Packages are identified as Low, Limited, Medium and High Inputs Packages (LIP,

,MIPand HIP).

Average wheat yields are estimated using MOA/NFIU trial data and the yield response functions

for Arssi.

cLabor use includes harvesting, weeding and herbicide application and computed at its opportunity

cost during peak season.

dSeed used from previous harvest is assumed to have an opportunity cost of EB 0.70/kg which the

farm gate price.

 

 

143
6.2 Econometric Analysis

This section presents the econometric analysis that identify the determinants
of adoption and assess the effects of significant variables on the probability and
intensity of adoption. Econometric models discussed in chapter 5 are employed to
analyze technology adoption in the study region. Logit models are used to
identify factors responsible for adoption and no adoption of fertilizers and
herbicides by wheat growers. Ordered probit model is employed to study the

factors responsible for farmers’ choices of different rates of fertilizer application,

i.e, from no use to higher rates of fertilizer use. Tobit model is used to further 2 ,’ ~'. I -

l
I

analyze the variables contributing to the continued adoption and quantity of /

 

 

_..._. ”rm“.

fertilizer use. The econometric analysis on the choice of the different wheat 4w.” ,-

fl
varieties did not give a satisfactory result. This could be because farmers mostly

use their own seed from past harvests and even if they want to buy them the
improved seeds are not available. They do not have a choice but to use the
available variety that has adapted in their respective districts. However a survey
of farmers’ opinions on varietal preference indicate that variety-specific
characteristics are important in their adoption decision. Characteristics that
farmers consider important are yield, color, resistance to diseases, tastes in local

foods (’injera’), bread, price and adaptability in the area.

|__ .

144
6.2.1 31W
Farmers’ socio-economic circumstances and other institutional variables are

used in the logit analysis of fertilizer and herbicide adoption or nonadoption. The
dependent variable for fertilizer use is fertilizer used for wheat in 1990 (FW90),
coded as 0 or 1. Similarly the dependent variable for herbicide use is APHBW90,
coded 0 for not using and 1 for application of herbicides. Maximum likelihood
estimation techniques are used and the results are presented in the following
tables. The coefficient for farmer’s age category (AGEHCAT) as expected has a
negative sign implying that older household heads tend not to adopt fertilizer
technologies. Level of education of household heads, family size (FMLSZ) and

farm size (FARMSZHA) have positive effects on the probability of adoption.
’ LVLEDHH and FARMSZHA are not statistically significant while FMI.SZ is
significant at the 5 % level in this specific analysis.

Further analysis is useful to determine whether farm size will have an effect in
the intensity of use rather than on the incidence of technology adoption. The
Logit analysis results suggest that access to credit (GETCRDT'), herbicide use
(APHBW90) and timely availability of fertilizer (GTFI'ML) variables are the
most important determinants of fertilizer adoption and no adoption. Their
coefficients have the expected positive signs and are statistically significant at the
1%, 1% and 5% levels, respectively. Farrners’ knowledge of fertilizer use, number
of oxen owned and extension contact have positive signs as expected but are

nonsignificant.

145
Table 6.8 Logit Estimates of Fertilizer Adoption in Arssi, 1990/91.

Variables Coefficients Partials
(t-ratio)
Constant 0.5202 0.02386
(0.662)
AGEHCAT* -0.3151 -0.0144
(-1.641)
LVLEDHH 0.0585 0.0027
(0.289)
FMLSZ 0.1131 0.0052
(1.892)
FARMSZHA 0.0756 0.00346
(0.642)
OXNUM90 0.1601 0.0073
(1.079)
chrML** 0.6338 0.0289
*** (1.851)
GETCRDT 0.9958 0.0455
(3.071)
KWRFTRT 0.1034 0.0047
(0.385)
APHBW90*** 1.2758 0.0583
(2.537)
DAVST90 0.2641 0.0121
(0.504)

(“‘, ",‘ indicate level of significance at the 1,5,and 10% levels, respectively)

The model correctly predicted the probability of adoption and no adoption
for 85 per cent of the cases. The partial derivatives given in column 3 of Table
6.8 reflect the respective effects of the continuous explanatory variables on the
probability of adoption that are very minimal.

The marginal effects of the binary regressors (coded as 0 or 1) are calculated
by substituting their values ( 0 or 1) and the results are reported in Table 6.9.

The effects of technical and institutional variables on farmers’ decisions to use

or not use fertilizer are indicated in the table above.

146

Table 6.9 Relative Effects of Significant Variables on Probability of

Fertilizer Adoption in Arssi, 1990/91

 

VARIABLES Values of Selected Binary Variables Coded 0 or 1

 

 

GTFTML 0
GETCRDT 0
“KWRFTRT 0
APHBW90 O
DAVST90 O
PROB.OF

ADOPTION

l
0
0
O
0

0
1
0
0
0

0

0
l
0
0

0.7368 0.8407 0.8834 0.7563

0
O
0
1
0

0
0
0
0

1

P‘ rd )4 P‘

C) c: 14 P‘

0.9093 0.7847 0.9866 0.9346

 

RELATIVE EFFECTS ON PROBABILITY

NO TECHNICAL OR INSTITUTIONAL SUPPORT

EXTENSION SERVICE

ACCESS TO CREDIT

HERBICIDE AVAILABILITY

TIMELY FERTILIZER AND CREDIT AVAILABILITY

AVERAGE ADOPTER BASED ON MEAN VALUES

INPUTS AND SUPPORTS AVAILABLE

Since the farmers in the region have been aware of improved production
technologies since 1968, it is not surprising that the average farmer has an 88%

probability of using fertilizer. Those farmers having no access to extension

PROBABILITY OF ADOPTION

.7368

.7847

.8834

.9093

.9346

.8891

.9866

service, and no knowledge of the recommended rates of fertilizer and herbicides

 

147

rates) and institutional support (access to credit, timely availability of fertilizer
and herbicide) will increase the incidence of adoption to more than 90 per cent.
The complete package of inputs and services will raise the use of fertilizer to 98
per cent. It is worth mentioning that farmers’ decisions to use fertilizer are related
to their use of herbicide as these inputs are complementary in crop production.
Farmers’ resources of land, labor and number of oxen owned appeared less
important in determining the incidence of adoption. However, the Tobit analysis -
will identify the effects of economic, technical and institutional variables on both
the adoption and the intensity of fertilizer use.
6.2.2 Ordered Probit Analysis

Agricultural extension and research organizations have attempted to develop
fertilizer recommendations for different crops based upon agro-ecological zones
and soil types as discussed in the earlier sections. Given farmers’ resources and
environmental circumstances, researchers have to further refine their technical
recommendations to be economically and socially feasible and acceptable. The
ordered probit model is used to identify the factors influencing farmers’ decisions
to select any of the five rates of fertilizer levels ( no use or less than 25 kg, 25 -50
kg, 50 -75 kg, 75 - 100 kg, greater than 100 kg/ha). The ordered probit estimates
in Table 6.10 show that farmers’ knowledge of fertilizer recommended rates,
educational levels, and ability to get fertilizer on time are critical variables
(significant at 1 % level) influencing the decision to use higher rates of fertilizer

per hectares. Age category (young, middle age, old) and family size have a

. _ 148
Table 6.10 Ordered Probit Analysis for Different Rates of Fertilizer Use in Arssi,

b

Mable CQeffieient
(t-ratio)
Constant 0.19525
(0.634)
AGEHCAT -0.61320
(-0.820)
LVLEDHH” 0.20464
(2.569 )
FMLSZ -0.1005
(-0.511)
FARMSZHA 0.45490
' (1.066)
OXNUM90' 0.81287
(1.593)
GTFTML” 0.26073
(2.262)
GETCRDT' 0.17130
(1.527)
KWRFI‘RT'” 0.18847
(3.136)
APHBW90 0.20295
(1.517)
DAVST90 0.10004
(0.573)
MU( 1) 0.82230
(12.142)
MU( 2) 1.3355
(16.765)
MU( 3) 2.5586
(21.348)

(***, ",* indicate level of significance at the 1,5,and 10% levels, respectively)

negative effect on the dependent variable. The results confirm that older farmers
and larger families tend to avert risk by not investing in higher rates of fertilizer.

The coefficients are not equal to the marginal effects of the regressors.

149

The probabilities of adopting the five different fertilizer rates are computed

using equation 12”.

Prob[Y = 0] = <1>(-B’X),

Prob[Y = 1] = 4’04 -B'X) - 4’(-B'X), (12)

PIObIY = 2] = 9012 - B’X)- 4’04 - B'X)
PmbIY = 3] = @013 ‘ B'X)‘ ‘I’Il‘z ' B'X)

Prob[Y = 4] = 140., - B’X).

The predicted probabilities are computed by substituting the estimated values

for B’X and us and <1) values from Z table.

Predieted prehabilities
Prob[Y = 0] = 0.2005
Prob[Y = 1] = 0.2955
Prob[Y = 2] = 0.2025
Prob[Y = 3] = 0.2588

Prob[Y = 4] = 0.0427

Aetpel Relative Frequencies
0.21 127

0.27465
0. 18779
0.27465

0.05164

The model predictions are very close to the actual frequencies of the data and

it correctly predicts 76 % of the cases. Given the mean values for the explanatory

variables it is expected that only 4.2% of the farmers in the study area will use the

 

27The predicted probabilities for the five rates coded j=0, 1, 2, 3 and 4 are calculated using the
regression estimates for B’X and MUs given a normal distribution.

150
recommended rate of 100 kg/ha and 26 % of the farmers will apply 75 - 100

kg/ha.
6.2.3 Tobit Analysis of Adoption and Use of Fertilizer

The coefficients of the Tobit model used to investigate factors associated with
the adoption and the quantity of fertilizer used in wheat production are reported
in Table 6.11. The results of the Logit and Ordered Probit models suggest that the
institutional variables are more important explaining the incidence or choice of
adoption. The findings of the Tobit analysis confirm that resource endowments
and their availability play critical role in the adoption and intensity of fertilizer
use. Variables FARMSZHA, OXNUM90, APHBW90, and DAVST90 have a
significant effect on adoption and use of fertilizer. The regression equation is
computed using the mean values of the independent variables.

The predicted probability of adoption for a farmer with characteristics X (the
vector of explanatory variables) is estimated as28 :

F(X’B/or) = 0.85591,
where F is the cumulative standard normal distribution function.

According to these findings there is 85.6% chance that an average farmer
would apply fertilizer. Expected value of fertilizer use across all observations as

defined by the Tobit model (Tobin, 1958) is given as:

 

28The predicted probability is computed using results in Table 6.11:
Where X’B = 49.2974,
0 = 46.41346 then
F(X’B/a) = F(49.2974/46.41346) = F(1.062136),
from the 2 table F(1.06216) = 0.855913.

151

E(Y) = XBF(z) + of(z) and substituting the values for the estimates shown in
the table, E(Y) = 49.2974(0.85591) + 46.41346(0.2651609) = 54.50117 kg.
where z = XB/a and f(z) is the unit normal density.

Similarly the expected quantity of fertilizer used by an adopter is :

E(Y‘) = x0 + cf(z)/F(z) = 63.67628 kg.

152
Table 6.11 Tobit Analysis of Adoption and Intensity of Fertilizer Use in Arssi,

1990/91.
—
Xa_ri_abl¢_ Cnifimt
(t-ratio)
Constant -3.6448
* (0.292)
AGEHCAT -4.6776
(1.5 10)
LVLEDHH 4.1675
(1.294)
FMLSZ 0.64187
1. t . (0.703)
FARMSZHA 9.9668
1' j g (5.569)
OXNU M90 8.6339
(3.943)
GTFTML 0.65634
(0.130)
GETCRDT 5.0806
g (1.074)
KWRFI'RT 5.5388
‘g g g (1.550)
APHBW90 17.810
1‘ g (3.084)
DAVST90 14.557
(2.135)
a 46.4 13
(26.027)

(***, "3* indicate level of significance at the 1,5,and 10% levels, respectively)
—
These estimates show that the average (considering adopters and non adopters)
farmer is expected to use 55 kg of fertilizer while those farmers who have adopted
will use 64 kg of fertilizer per farm. This expected rate of application would be

less than 25 kg/ha given an average farm size of 2.7 hectares per household. This

153

suboptimal rate of fertilizer use hence calls for a comprehensive approach to
alleviate the constraints of adoption.

Furthermore, the results of the Tobit model can be used to identify the
effects of changes in an explanatory variable on the adoption and intensity of use.
McDonald and Moffitt (1980) present a Tobit decomposition approach to
separate two effects. The two effects are: changes due to likelihood of new
adoption and expected changes in intensity of adoption by those who have already
adopted. For the study the total probability of adoption (i.e, 0.85) is decomposed
to give probability of 0.55 and 0.30 for new adoption and intensity of use,
respectively. The decomposition of the effects is important to identify the
influence of adoption determinants on the sequential stages of adoption, i.e., to
adopt or not to adopt and then to continue using the technology.

Table 6.12 presents the elasticity of decomposition for changes in the
explanatory variables. Total elasticity of a change in the level of any of the
variables consists of two effects: elasticity of expected use intensity (E1) and

elasticity of adoption probability (E22).

66666666666-66666666666-6.666 .66666-66 .6666.66 1 6666666 a 66
6666.66 1 6666666666+6x t .66 .66666666 t 666666666
6666.6 t 66666666666-66666.66666-66

.3004 I u . manic: I a lo .NnoNd IAN:
annod I Anya
onaN.oc I ax

.2686 «0360326: no 66033666660 23 :6 veg p66. gene... 25 63 peace—«u: eue 9668033 2:.

6666616 66666666666 t 666666666 66 66666666666 66
6 666166666666666 t 6666666666 66 66666666666 66
6666666666 :6 66.666 666666626 66:

66666»666.6666 t

 

 

 

 

 

 

 

 

 

 

 

66 6
_— 6666.66 66 66666

6666.6 6666.6 6666.6 6666.6 6666.6 6666.6 6666.6 6666.66 6666.6 6666666
6666.6 6666.6 6666.6 6666.6 6666.6 6666.66 6666.6 6666.66 6666.6 663666<
6666.6 6666.6 6666.6 6666.6 6666.6 6666.6 6666.6 6666.6 6666.6 6666626
6666.6 6666.6 6666.6 6666.6 6666.6 6666.6 6666.6 6666.6 6666.6 6666666
6666.6 6666.6 6666.6 6666.6 6 666.6 6666.6 6666.6 6666.6 6666.6 626666
6666.6 6666.6 6666.6 6666.6 6666.6 6666.6 6666.66 6666.6 6666.6 6626266
6666.6 6666.6 6666.6 6666.6 6666.6 6666.6 6666.66 6666.6 6666.6 66662266
6666.6 6666.6 6666.6 6666.6 6666.6 6666.6 6666.6 6666.6 6666.6 666:6
6666.6 6666.6 6666.6 6666.6 6666.6 6666.6 6666.6 6666.6 6666.6 6666666
6666.6- 6666.6- 6666.6- 6666.6- 6666.6- 6666.6- 6666.66- 6666.6- 6666.6 “6666
6666.6- 6666 6- 6666.6 62666260

66 66 666. 666. 6666666666 6666666666 86 6.6.66 6666: 6.696653

agedoueau 6 6656...: 66666666

 

 

 

N
mm: «o >u6mcmucH can cofiumood umufiafiuuom

co moancfluc> xuoucccamxm ecu :6 mmmccco Mom cofiufimomaoooo muwofiumcam #6308 «6.6 dance

emma

155
Adding the two effects will give us the total elasticity. Given that there is a

high rate or incidence of adoption it is expected that any change in the
determinants of adoption will affect more the intensity than the probability of
adoption. The elasticities computed therefore confirm that marginal changes in
the independent variables listed increase the use intensities(E1) more than the
probability of adoption. Elasticities of adoption probabilities(FQ) are relatively
smaller than E1. Overall the elasticity estimates in Table 6.12 reflect inelastic
(E< 1) response to changes in the adoption variables”. Farm size has the
highest impact on the use intensity and probability of adoption with a total
elasticity value of 0.42. This value is divided into 0.24 and 0.18 for intensity and
adoption probability respectively. The implications of these estimates are useful
to evaluate proposed policy changes affecting technical and institutional variables
used in this study. For example a 10% change in farm size is expected to result in
about a 4.2% increase in adoption and use of fertilizer. The expected intensity
will increase by 2.4% while the probability of adoption will increase by 1.8%. In
addition the wealth variable used in the analysis, i.e., number of oxen owned, has
a total elasticity value of 0.26, which is composed of 0.14 and 0.11, the elasticities
of intensity and adoption respectively. The other variables are also important in

affecting the adoption and quantity of fertilizer use.

 

29Inappropriate government intervention based on the command economy model has
handicapped the performance of agriculture in Ethiopia as discussed in chapter 2. Consequently,
input and output prices were fixed for many years by the government. The price responsiveness of
small farmers and the effect of prices on technology adoption is not covered in this study. For details
on price responsiveness of smallholders in EthiOpia see Shire (1986).

156
6.2.4 Legit Analysis of Herbieide Adoption

Estimates of the logit model for herbicide use in Table 6.13 show that farm
size has a negative effect on the adoption of herbicide in the study area. This is
contrary to the previous literature on technology ad0ption. Farmers reported that
they were unable to buy the input in the market as it is rationed by the local
extension office. Government policy in the past discouraged the use of herbicides
by the smallholder sector while encouraging the state farms. Farmers with larger
plots were not able to get herbicide for all their fields and have to use family
and/ or hired labor for weeding. The coefficient for the extension variable
suggests that farmers who have regular contact with the local extension office
appeared to use herbicide. The variable DAVST90 has a significant and positive
effect on the dependent variable. As in the case of fertilizer use, herbicide use is
also enhanced by the use of fertilizer. Most farmers indicated that credit is used
primarily for fertilizer purchase. The complementarity of the two inputs is
confirmed by the positive effect of access to credit on herbicide use. More credit
enables farmers to buy fertilizer and more fertilizer use will encourage farmers to
adopt herbicides.

Family size was expected to be negatively affecting herbicide use as labor is a
substitute for hand weeding. The labor shortage during critical periods of
weeding and harvesting different crops that smallholders grow will not have a

substitution effect on herbicide.

157
Table 6.13 Logit Analysis of Herbicide Adoption in Arssi, 1990/91.

Maximum Likelihood Estimatesa

Veg’eple Qeefficient

(t-ratio)

Constant -3.0574
(0.8083)
AGEHCAT 0.46805
‘ (0.279)
LVLEDHH 0.2505
(1.470)

FMLSZ 0.5707
* .- (1. 172)
FARMSZHA -0.4261
‘ g .- (-3.744)

OXNU M90 0.3250
" (2.697)
GETCRDT 0.4527
H (1.813)
FW90 1.1432
n. (2.302)
DAVST90 0.9406
(3.018)

aMultinomial Logit Model

LOGIT;LHS = API-IBW90;RHS = X

2 Outcomes: APHBW9=0 APHBW9= 1

Coefficients for APHBW9= 0 set to ZERO.

("‘3 ",* indicate level of significance at the 1,5,and 10% levels, respectively)

158
The variable OXNUM90 has a positive relationship with herbicide use

indicating the relationship between resource endowment and input use. In this
particular analysis, the effects of land and labor did not conform to our theoretical
expectation. The results of the analysis do not support the contention that
farmers age negatively affect the use of herbicide. The model correctly predicts
78% of the cases and it is expected that an average farmer has only 19%
probability to use herbicides given the mean values of the regressors. Herbicide
use in the study area is very low as only 22 % of the sample farmers reported as
users at one time or another.

6.3 Discriminant Analysis and Adopter Categories

Estimates of the discriminating variables for the five districts and for the
region as depicted in Table 6.15, reveal the relative importance of the variables.
A total of twelve variables are identified to discriminate the adopter categories.
However, only seven of the variables are responsible for the difference among the
categories for the region. Farmers’ level of education, opinions about herbicide,
use of hired labor and access to credit appeared to be most important in
classifying the 426 sample farmers into different categories. The number of
discriminating variables are three in Galema, four each in Chilalo and Ticho, five
in Keleta and six in Gedeb district. Farm size and ownership of radio came out
to be important in four of the five districts suggesting the critical role of farmers’
resources and access to information in technology adoption. Level of education,

access to credit, oxen number, knowledge about herbicide and farmers’ attendance

159

of field days and demonstrations (ATFLDD) are relevant in two districts each
while the rest of the five variables appeared to be important in only one of the

five districts.

 

 

 

 

666. 666. 666.- 6666. 666. 666. 666666

6666. 666. 666.- 66626

6666. 666. 666. 66666 666. 666. 6666. 666. 666.- 6666. 666. 666.- 66662666

6666. 666. 666. 6666666

666. 666. 626666

666. 666. 666. 666. 666. 6666. 666. 666. 666. 666. 666. 6666. 666. 666. 6666236

666. 666. 666. 666. 666. 6666666

666. 666. 6666. 666. 666. 6666. 666. 666. 66666666

666. 666. 6666. 666. 666. 6666. 666. 666.- 6626266

666. 666. 6666. 666. 666.- 6666. 666. 666. 6666666

666. 666.6 6666. 666. 666. 6666. 666. 666 - 66666>6

. 6666. 666. 666. 6666666

66 6666.6 6668 66666 3.66.6 6668 66666 666.66 668.6 66666 35.6 6668 66666 665.6 6668 66666 3.8.6 6.6.8— 6.66866
6.666 _ 66666 - 6.6.6 666666 _ 66.6666 6666.6 _

 

 

 

 

N
Hm\omm6 .muofluumflo annum :6 mowuoaouco Monmocd you mounfiwumm ucccflawuomflo ma.m manna

00H

161

The accuracy of the classification procedure is measured by the per cent of
correctly predicted cases in each group. Actual and predicted classification of
farmers into fertilizer and herbicide users and three adopter categories is shown
in Table 6.16. The actual and predicted number of farmers in each of the
fertilizer and herbicide users and adopter category are converted to percent of
farmers in each district and the total sample. Accordingly the actual figures show
that 97 and 53 percent of the farmers use fertilizer and herbicide, respectively.
For the same district the model predicted 90 and 74 percent of the farmers to
apply fertilizer and herbicide respectively. The farmers in Keleta rank second in

the use of these inputs. Farmers in Gedeb do not use herbicide at all.

 

 

 

 

66 66 66 66 66 66 66 66 66 66 66 66 66.6666<-
vooco>u<
66 66 66 66 66 66 66 66 66 66 66 66 66.6666<
ouauovoz
66 66 66 66 66 66 6 6 6 6 66 66 6666666< :66
anon:
66 66 6 6 6 6 6 6 66 66 66 66 666666666
uuoufi
66 66 66 66 66 66 66 66 66 66 66 66 6666666666
anon-ado
66666666
666666666 66666< 666666666 66666< 666666666 66:6o< 6.6666666 66=6o< 666666666 666666.666666666 66:66<
66.66 66666 66666 65.666 6666660 . 6666.6

 

 

 

 

 

 

 

 

 

.muoauam

NO UnnOOHom "mQHHONOUQO HOUQOfid HO GOHUUOflHHmeHU H¢9UO< UGG GOHUOHV$HQ Hflmvoz UCUGHEHHOQHG ©H.© O—Dflr—L

N2

163

Ticho and Galema farmers are the least in fertilizer and herbicide use
respectively. In terms of adoption category for the entire sample, the actual
classifications indicate that about 17%, 46% and 37% of the sample farmers fail
into low, moderate and advanced adopters categories, respectively. Galema
farmers tend to use inputs more intensively as more than half of them are
classified (35% predicted) as advanced adopters. About 40 to 50 % of farmers in
each of the five districts and 46% in Arssi are classified as moderate adopters.
Among the low adopters, Gedeb and Ticho farmers constitute the largest shares.
The predicted figures are lower than the actual and this also suggests the problem
in the predicting ability of the model. The per cent of correct predictions are
better for the fertilizer and herbicide use classification. For example the model
correctly predicted 90% and 74%; 72% and 71% of fertilizer and herbicide users
in Chilalo and Keleta. The model predictions for the three adopter categories
ranged from 46 to 62% fOr the districts and is only 41% correct predictions for
the whole region. The implications of these classifications in the identification of
important variables and ranking them based on their relative contribution in
discriminating one group from the other group is critical in the design of research

and extension strategy.

164
6.4 Summary

In chapter six, the profitability of wheat production technologies, econometric
analysis of technology adoption and classification of adopter categories are
examined and the results of the analysis are presented.

6.4.1 WWW

Partial budget and marginal rate of return analysis techniques are used to
evaluate the profitability of wheat production technologies. The results are
presented for three scenarios : a) farmers’ production practices (Table 6.1), b)
selected levels of input use (Table 6.3 and 6.4) and c) comparison of four
alternative production packages (Table 6.7).

The variety Enkoy grown in. all the districts generated a net benefit of EB
548/ha in Keleta, and EB 432/ha in Galema district followed by Dashen with EB
412/ha in Keleta. The third variety Bulk gave lower net benefits ranging from BB
135 to 182/ha.

The survey data suggest the suboptimal rate of fertilizer application. A fifth of
the sample farmers in Arssi use less than 25 kg of fertilizer per hectare. Almost
half of the respondents use 50 kg or less and the highest users are farmers in
Galema and Chilalo (81 kg/ha). Results of the partial budget for the different
levels of fertilizer use reveal that the 100 kg/ha rate gave a net benefit of EB
1531/ha and the 200 kg/ha level had EB 1476/ha. A move from 100 kg/ha 6
150 kg/ha yields an MRR of only 33% which is not in the minimum acceptance
range of 50-100%.

165

Four possible production practices under different levels of input use, i.e.,
Low input package (LIP), Limited Inputs Package (LMIP), Medium Inputs
Package (MIP), High Inputs Package (HIP) are compared. Based on different
prices and opportunity cost assumptions, gross benefits, variable costs, net benefits
and marginal rate of returns are calculated for each alternative package. Prices,
levels of fertilizer use, seed quality and source, use of herbicide and time of
application have contributed to the apparent difference in the yields, costs and in
the net benefits.

The HIP alternative yielded the highest net benefit per ha (1713 EB) while
the minimum/low input LIP produced a net benefit of 1158 EB/ha. The MIP
gave the highest MRR of 437% over the LMIP alternative and the HIP yielded
209% MRR (Table 6.7). Farmers’ decision criteria to select any one of the
alternative packages will be influenced by their access to resources. Farmers
without a capital constraint to purchase inputs can choose the HIP package.

6.4.2 Economotrio Analysis of Incidenco and Intonsity of Adoption
Descriptive statistics on adoption rates of the three wheat varieties, fertilizer
(including proportion of farmers in different fertilizer application rates) and
herbicide use by district are presented. The results indicatethat the most p0pu1ar
variety Enkoy is adopted by farmers in the five districts. Galema farmers (88%)
grow Enkoy while 41% of the farmers in Chilalo adopted the same variety. At
the same time Dashen is grown by only 6 and 2% of the growers in Keleta and

Galema. Despite its high yield, the incidence of disease has restricted the use of

166

Dashen to Chilalo. The variety Bulk is more adopted by farmers in Ticho (43 %)
and Gedeb (28%).

Logit Analysis: The logit maximum likelihood estimation procedure was used to
obtain consistent, efficient and asymptotically normal estimators (Table 6.8). The
coefficient for farmer’s age category (AGEHCAT) has a negative sign implying
that older household heads tend not to adopt fertilizer technologies. Level of
education of household heads, family size (FMLSZ) and farm size (FARMSZHA)
have positive effects on the probability of adoption. LVLEDHH and
FARMSZHA are not statistically significant while FMLSZ is significant at the 5
% level in this specific analysis. The results suggest that access to credit
(GETCRDT), herbicide use (APHBW90) and timely availability of fertilizer
(GTFTML) variables are the most important determinants of fertilizer adoption.
Farmers’ knowledge of fertilizer use, number of oxen owned and extension
contact have positive signs as expected but are not significant. The model
correctly predicted the probability of adoption and no adoption for 85 per cent of
the cases.

The marginal effects of the explanatory variables are calculated by
substituting their values (0 or 1) and the results are reported in Table 6.9.
Provisions of technical support (extension) and institutional support (access to
credit, timely availability of fertilizer and herbicide) will increase the incidence of
adoption to more than 90 per cent. Availability of the complete package of inputs

and services will raise the use of fertilizer to 98 per cent. Farmers’ resources of

167

land, labor and number of oxen owned appeared less important in determining
the incidence of adoption.

Estimates of the logit model for herbicide use (Table 6.13) show that farm
size has a negative effect on the adoption of herbicide in the study area. This is
contrary to the previous literature on technology adoption. The coefficient for the
extension variable DAVST90 which is significant suggests that farmers who have
regular contact with the local extension office appeared to use herbicide. The
complementarity of fertilizer and herbicide is suggested by the positive effect of
access to credit on herbicide use. The expected substitution effect between labor
and herbicide was not observed in this study.

The model correctly predicts 78% of the cases and it is expected that an
average farmer has only 19% probability to use herbicides given the mean values
of the regressors. Herbicide use in the study area is very low as only 22 % of the
sample farmers reported being users at one time or another.

Probit Analysis: Ordered probit model was used for analyzing the factors
influencing farmers’ decisions to select any of the five rates of fertilizer levels (no
use or less than 25 kg, 25 -50 kg, 50 -75 kg, 75 - 100 kg, greater than 100 kg/ha).
The ordered probit estimates in Table 6.10 showed that farmers’ knowledge of
fertilizer recommended rates, educational levels, and ability to get fertilizer on
time are critical variables (significant at 1 % level) influencing the decision to use

higher rates of fertilizer per hectare. Age category and family size have a

negative effect on the dependent variable. The results confirm that older farmers

168

and larger families tend to avert risk by not investing in higher rates of fertilizer.
The predicted probabilities for selecting any of the five levels for an average
household are computed by substituting mean values of the explanatory variables
into the regression equations. The predictions are very close to the actual
frequencies of the data; the model correctly predicts 76 % of the cases. Given
the mean values for the explanatory variables it is expected that only 4.2% of the
farmers in the study area will use the recommended rate of 100 kg/ha and 26 % .
of the farmers will apply 75 7 100 kg/ha.
Tobit Analysis: The findings of the Tobit analysis revealed that resource
endowments play a critical role in the intensity of fertilizer use. Variables
FARMSZHA, OXNUM90, APHBW90, and DAVST90 have a significant effect
on adoption and use of fertilizer (Table 6.11). According to these findings there
is an 85.6% chance that an average farmer would apply fertilizer. These
estimates also indicated that the average farmer who is a new adopter is expected
to use 55 kg of fertilizer while those farmers who have already adopted will use
64 kg of fertilizer per farm. This expected rate of application would be less than
25 kg/ha given an average farm size of 2.7 hectares per household.

A Tobit decomposition approach was used to separate the two effects. For the
study the total probability of adoption (i.e. 0.85) is decomposed to give a
probability of 0.55 and 0.30 for new adoption and intensity of use, respectively.
Table 6.12 presented the elasticity of decomposition for changes in the

explanatory variables.

169
6.4.3 Adoption Categories

Classifying farmers into different adopter categories is undertaken by using
the technique of discriminant analysis. Estimates of the discriminating variables
for the five districts and for the region were depicted in Table 6.15. Twelve
variables are identified to discriminate the adopter categories. However, only
seven of the variables were responsible for the difference among the categories
for the region. Farmers’ level of education, opinions about herbicide, use of hired
labor and access to credit appeared to be most important in classifying the 426

sample farmers into the low, moderate and advanced adopter categories.

CHAPTER 7
SUMMARY, CONCLUSIONS AND IMPLICATIONS

7.1 Introduction
7.1.1 WM):

Ethiopia is one of the poorest countries in Sub Saharan Africa. The
agricultural sector employs 86% of the p0pulation, and accounts for 45% of the
GDP and 90% of the export earnings. While agriculture has significant potential,
its performance has been dismal.

Ethiopia is the largest wheat producer in Sub Saharan Africa with 687,000 ha
under wheat cultivation. The central, southeastern and northwestern highlands
have favorable climatic conditions for wheat production. Wheat area, production
and yield increased (0.4, 3.2 and 2.8% per annum, respectively) in the 1983-92
period due to the expansion of mechanized state farms and adoption of improved
technologies by state farms and smallholders. However, production lagged behind
the rapid increase in wheat consumption of nearly 35% per year. Average wheat
yields are lower than in other countries in Southern and Eastern Africa. Since,
Ethiopia is importing about half of its wheat requirements, it is important to
increase wheat production and save vital foreign exchange.

Ethiopia is in a transition to a market economy and it needs to formulate new

agricultural policies based on empirical studies. Although microeconomic studies

170

171

have been conducted for export crops and commodities, there are only a few
studies on the economics of wheat production in Ethiopia. This study is designed
to fill that gap.
7.1.2 Research Objectives

The general objective of the study has been to examine the social, economic,
and institutional factors that influence the adoption of selected wheat production
technologies in Arssi Region in the Southeastern Highlands of Ethiopia. Arssi is
a major wheat and barley producing region and the home of the first integrated
rural deve10pment project that was launched in 1968 with assistance from Sweden.

This study was undertaken in the Arssi Region in order to draw lessons from
the experiences of technology adoption in the region. The region has 12 districts;
the study was carried out in the five major wheat growing districts of Chilalo,
Keleta, Galema, Gedeb and Ticho.

The components of the production package include high-yielding varieties
(HYVS), chemical fertilizers and herbicides.

The specific objectives of the dissertation were to:

1) present descriptive information on wheat production by smallholders in

Arssi region;
2) determine the profitability of smallholder wheat production;
3) generate quantitative estimates of the factors influencing the incidence

and intensity of adoption of wheat production technologies;

172

4) identify constraints on the adoption of recommended technologies for

wheat;

5) classify farmers by adopter categories based on adoption patterns and

socioeconomic characteristics;

6) draw implications for agricultural research, extension, and policy changes

needed to overcome the constraints on adoption.

7.1.3 Technology Adoption: An Qerview

Studies of hybrid corn adoption in the United States in the 19305 and

subsequent studies by many economists and sociologists have shown that the path
of technology adOption approximates a cumulative normal distribution with an S-
shaped curve. Studies in the Third World have shown that the factors responsible
for adoption of technologies vary across countries, regions, farms and the type of
enterprises.

The literature reveals that technology adoption is a function of various social
and economic characteristics. Most studies have viewed adoption of a
technological practice in isolation from other related practices, ignoring the
interdependencies of different technological practices. Most studies have viewed
adoption as a discrete choice which assumes farmers will eventually adopt a full
range of technologies. Finally most studies have considered few explanatory
variables and hence the models are underspecified.

This study takes these limitations into account and examines the interrelated

technological practices of HYVs, fertilizer and herbicides and the incidence and

173

intensity of adoption of fertilizer. The study also includes a number of social,
economic and institutional variables and adopters’ perceptions of technologies.

A survey of 426 smallholders was conducted in the five major wheat-growing
districts of Arssi from February to September 1991. Data were collected on
farmer characteristics, farm resources, production practices, adoption of wheat
varieties, fertilizer and herbicide and farmers’ perceptions of recommended
technologies. Partial budgets and marginal returns analysis were used to measure
the profitability of technologies and alternative production packages identified.
Econometric models (logit, probit and Tobit) were estimated and statistically
significant variables were identified to measure the relative importance of the
variables influencing farmers’ adoption decisions.

7.2 Summary of Results

7.2.1 Household Characteristics

Farmers’ decision making in adopting technologies is complex and is
examined taking into account economic, social and institutional factors. Family
Size/Age: The mean family size for the region is about 7.3 persons which is
higher than the national average of 5 persons. This difference can be explained
by the fact that the majority of Arssi families are moslems and practice polygamy.
The mean household age is 45 years. '
Educational Levels: The literacy rate in the Arssi region is the highest in the
country. Seventeen percent of the sample were illiterate compared to 20-39%

illiteracy level for the country. Chilalo and Galema districts have 11 and 16

174

percent illiteracy rates, respectively. About half of the respondents have 3-4
months of literacy education and 22 percent had elementary-level education.
Farm Sizes: The land reform of 1975 resolved the inequitable ownership of land
and tenancy problems in most parts of Ethiopia. The mean farm size and arable
area for Arssi are 2.7 and 1.9 ha., respectively. There is a significant difference in
farm sizes across the study districts. The average farm size per household is 4.1
ha in Gedeb and 1.9 ha in Ticho. Forty percent of the households cultivate two
ha or less and the other 45 percent have farm sizes between 2.1 - 4 ha. Only 13
percent of the farmers in Arssi own more than 4 hectares of farm land.

7.2.2 Profitabilig of Prodoction Toohnologies

Partial budget and marginal rate of return analysis techniques are used to

evaluate the profitability of wheat production technologies. The results are
presented for three scenarios: a) farmers production practices, b) increased use of
fertilizer only, and c) comparison of four alternative packages of improved inputs.
The major findings from the profitability analysis are:

1. The wheat variety Enkoy, which was grown in each of the five surveyed
districts, generated a net benefit of Ethiopian Birr (EB) 548/ha in Keleta,
and EB 432/ha in Galema district. The Dashen variety generated a net
benefit of EB 412/ha in Keleta. The Bulk variety generated net benefits
ranging from BB 135 to 182/ha. The popularity and high net benefits of
Enkoy reveal that small farmers’ adoption decisions are influenced by

profitability.

175
2. The survey revealed a suboptimal rate of fertilizer application. A fifth of

the sample farmers in Arssi use less than 25 kg of fertilizer per hectare.

Almost half of the respondents use 50 kg or less and the highest users are

farmers in Galema and Chilalo (81 kg/ha). The partial budgets for the

different levels of fertilizer use reveal that the 100 kg/ha rate gave a net
benefit of EB 1531/ha and the 200 kg/ha level gave EB 1476/ha. An
increase from 100 kg/ha to 150 kg/ha yields a marginal rate of return

(MRR) of only 33% which is not in the minimum acceptance range of 50-

100%.

3. Based on the results of the MRR analysis of fertilizer levels used in the
study, we can conclude that given the 1991 season crop-fertilizer price
relationship farmers’ fertilizer application rates should not exceed the 100
kg/ha level.

4. The following four possible production practices under different levels of
input use are identified for the comparative analysis:

a) Low input package (LIP): in which farmers use 50 or less kg/ha
of fertilizer, own/retained wheat seeds, and relied on family labor
for hand weeding;

b) Limited input package (LMIP): use of 50 - 75 kg/ha of fertilizer,
purchased wheat seeds and hand weeding using family labor;

c) Medium input package (MIP): 75 - 100 kg/ha of fertilizer,

purchased seeds and herbicides;

176
d) High input package (HIP): greater than 100 kg/ha of fertilizer,

purchased seeds and herbicides.

The HIP alternative yielded the highest net benefit per ha (1713 EB) while
the LIP produced a net benefit of 1158 EB/ha. The MIP gave the highest MRR
of 437% and the HIP yielded a 209% MRR. Farmers’ decision criteria about the
alternative packages will be influenced by their access to resources. Farmers
without a capital constraint are able to purchase inputs and can adopt the HIP
package. This is different from the above recommendation (3) which was based
on fertilizer use only and did not include other package components (improved
seed, herbicide).

7.2.3 Analysis of Farmers’ Proopgion Pragiees and Technology Adoption

The initial benefits of the integrated rural development project in the Arssi
region went largely to the land-owning class in Arssi, but the radical land reform
of 1975 gave the small farmers access to land and improved production
technologies. The results of the study reveal that compared to farmers in other
regions of Ethiopia, farmers in Arssi:

a) have high rates of fertilizer adoption (87%),

b) apply an average of 71 kg fertilizer per ha for wheat,

c) fertilize 55 % of the cultivated area per household, and

d) have more than 15 years of awareness and knowledge about fertilizer use.

e) wheat has become an important food and cash crop (an average

household markets about half its wheat production).

177

In terms of fertilizer application rates, there was considerable variation across
the five districts. Almost half of the sample farmers in the region use 50 kg or
less of fertilizer per ha. About 60% of farmers in the districts of Ticho and
Gedeb apply 50 kg or less fertilizer per ha while the same proportion of farmers
in Chilalo and Galema use relatively higher rates (75-100 kg/ha).

Wheat Varietal Adoption and Herbicide Use: Enkoy is grown in all the districts
and is the highest yielder except in Ticho. Enkoy, which is grown throughout
different zones of the region, has a yield gap of 45 % between research station and
on-farm trials, and a 30% yield gap between on-farm trials and farmers’ reported
yields. Wheat gets priority in the use and allocation of purchased farm inputs. In
Chilalo and Keleta, 52 and 48% of the farmers reported using herbicides.
Adopters’ Perceptions of Technologies: Recent studies identify three paradigms
for technology adoption: a) innovation-diffusion, b) economic constraints and c)
adopters’ perceptions. The surveyed farmers reported that unavailability, late
delivery and high prices are the major reasons for not using fertilizer, herbicide
and HYVs in Arssi. Wheat varieties are preferred by farmers for their specific
attributes. For instance, 75 and 84 and 55% of the respondents reported that
yield is the most important criterion in selecting the three most popular varieties:
Dashen, Enkoy and Bulk. Other factors such as grain color, grain price, taste in
preparing local foods, resistance to lodging and diseases are listed as important

criteria in varietal preference by farmers.

178

7.2.4 B nom ri An 1 i f Inci en Intensi an Det rminants of
69mm

Farmers’ adoption decisions are analyzed using qualitative response models,
which constrain the estimated probabilities of adoption to a range between 0 and
1. The two most popular are the logit and probit models.
Logit Analysis: The logit is used to model yes/no adoption decision of farmers.
The logit analysis results suggest that access to credit, herbicide use and timely
availability of fertilizer variables are the most important determinants of fertilizer
adoption. Their coefficients have the expected positive signs and are statistically
significant at the 1% and 5% levels. The model correctly predicted the
probability of adoption and no ad0ption for 85 per cent of the cases. Farmers’
knowledge of fertilizer use, number of oxen owned and extension contact have
positive signs as expected but are not significant. The coefficient for farmer’s age
category has a negative sign implying that older household heads tend not to
adopt fertilizer technologies. Level of education of household heads, family size
and farm size have positive effects on the probability of adoption. Family size is
significant at the 5 % level in this analysis.

The marginal effects of the binary regressors are calculated by substituting
their values in the estimated logit model. Since farmers in the region have been
aware of improved production technologies since 1967/ 68, it is not surprising that
the average farmer has an 88% probability of using fertilizer. Farmers without

access to extension service and no knowledge of the recommended rates of

 

179
fertilizer and herbicides are expected to have a 73% probability of using fertilizer.

Results of the logit analysis further indicate that subsequent provisions of
technical support (extension) and institutional support (access to credit, timely
availability of fertilizer and herbicide) will increase the incidence of adoption to
more than 90 per cent. If the complete package of inputs and services is
available, fertilizer adoption will increase to 98 per cent.

Farmers’ decisions to use fertilizer are related to herbicide use because these
inputs are complementary in crop production. Farmers’ resources of land, labor
and number of oxen owned appeared less important in determining the incidence
of adoption. Estimates of the logit model for herbicide use show that farm size
has a negative effect on the adoption of herbicides in the study area. Farmers
reported that they were unable to buy herbicides in the market. Local extension
officers rationed herbicides to cooperative members only.

Most farmers indicated that credit is used to purchase fertilizer. The
complementarity of fertilizer and herbicide is suggested by the positive effect of
access to credit on herbicide use. Credit enables farmers to buy fertilizer and
more fertilizer use will encourage farmers to adopt herbicides. Family size was
expected to negatively aflect herbicide use since labor can be used for hand
weeding. Labor shortage during critical periods of weeding and harvesting of
different crops is a major constraint. Farmers use both labor and herbicide to

control weeds.

180

Ordered pmbit analysis: Farmers have a choice of using different levels of
fertilizer application. The ordered probit model was used for analyzing the
factors influencing farmers’ decisions to select one of the five levels of fertilizer.
The probit estimates showed that farmers’ knowledge of fertilizer recommended
rates, educational levels, and ability to get fertilizer on time are critical variables
(significant at 1 % level) influencing the decision to use higher rates of fertilizer
per hectare. Age and family size have a negative effect on the dependent
variable. Older farmers and larger families tend to avert risk by not investing in
higher rates of fertilizer. The predicted probabilities for selecting any of the five
levels for an average household are computed by substituting mean values of the
regressors into the regression equations. The predictions are very close to the
actual frequencies of the data; the model correctly predicts 76 % of the cases.
Given the mean values for the explanatory variables it is expected that only 4.2%
of the farmers in the study area will use the recommended rate of 100 kg/ha and
26 % of the farmers will apply 75 - 100 kg/ha.
Tobit Analysis: The Tobit model measures not only the probability that a farmer
will adopt the new technology, but also the intensity of use of the technology once
adopted.

The results of the logit and ordered probit models identified the importance
of institutional variables in explaining adoption. The findings of the Tobit
analysis, however, reveal that resource endowments play a critical role in the

adoption and intensity of fertilizer use. Farm size, number of oxen owned,

181

herbicide use and visit by extension agent variables have a significant effect on
adoption and use of fertilizer. According to these findings there is an 86% chance
that an average farmer would apply fertilizer. These estimates also indicated that
the average farmer who is a new adopter is expected to use 55 kg of fertilizer
while those farmers who have already adopted will use 64 kg of fertilizer per
farm. This expected rate of application would be less than 25 kg/ha given an
average farm size of 2.7 hectares per household for the region.

Elasticity of Adoption: The results of the Tobit model were also used to identify
the effects of changes in an explanatory variable on the adoption and intensity of
use. A Tobit decomposition approach was used to separate the two effects: a)
changes due to likelihood of new adoption and b) expected changes in intensity of
adoption by those who have already adopted. For the study the total probability
of adoption (i.e. 0.85) is decomposed to give a probability of 0.55 and 0.30 for
new adoption and intensity of use, respectively.

Farm size has the highest impact on the use intensity and probability of
adoption with a total elasticity value of 0.42. This value is divided into 0.24 and
0.18 for intensity and adoption probability, respectively. These estimates are
useful in evaluating proposed policy changes affecting technical and institutional
variables used in this study. For example, a 10% increase in farm size is expected

to result in about a 4.2% increase in adoption and use of fertilizer.

182
7.2.5 Adoption Categories

Discriminant analysis was used to classifying farmers into different adopter
categories. Twelve variables are identified to discriminate the adopter categories,
but only seven were responsible for the difference among the categories for the
region. Farmers’ level of education, opinions about herbicide, use of hired labor
and access to credit helped classify the 426 sample farmers into the low, moderate
and advanced adopter categories. The number of discriminating variables is three
in Galema, four each in Chilalo and Ticho, five in Keleta and six in Gedeb
district. Farm size and ownership of a radio were important in four of the five
districts, suggesting the critical role of farmers’ resources and access to
information.

In terms of adoption category for the entire sample, 17%, 46% and 37% of
the sample farmers are low, moderate and advanced adopters, respectively. More
than half of the Galema farmers are classified as advanced adopters. About 40 to
50% of farmers in each of the five districts and 46% in Arssi are classified as
moderate adopters. Among the low adopters, Gedeb and Ticho farmers

constitute the largest pr0portion.

183

7.3 Policy Implications

7.3.1 Implications for Agricultural Researeh

The descriptive information on wheat producers and the subsequent

econometric analysis of technology adoption shed some light on the crucial role of

agricultural research and the availability of appropriate technology. The following

major implications emerge from the results of the analysis:

1.

Although farmers in the study area are aware of the recommended
technologies, particularly fertilizer, HYVs and herbicides, they are using
suboptimal levels of these inputs because of the unavailability of inputs,
high input prices relative to grain prices, or farmers’ own reservations
about the technologies.

Since 20% of the farmers use less than 25 kg fertilizer per ha, it is
imperative to develop low input technologies. Agricultural researchers
have to focus on refining the fertilizer recommendations so that they can
tailor their recommendations for different farming systems and for
farmers with different levels of resources.

The three most popular varieties have their specific attributes that
farmers demand. The most widely used variety, Enkoy, gives a stable
yield and has disease resistance. Dashen is susceptible to disease and
Bulk has a lower yield. Future crop improvement programs have to
consider the specific attributes of varieties that influence farmers’

decisions.

184

The IAR has recently incorporated adopters’ perceptions of specific
characteristics of technology into its research agenda. This research
should be expanded. A demand-driven agricultural research program

requires feedback from the users.

7.3.2 Implications for Agp'cultural Extension

The major duties of the extension agents are input distribution, organizing

and managing demonstration sites and field days, and organizing and promoting

agricultural cooperatives. The implications of the study for the agricultural

extension service are summarized below.

1.

In the past 17 years, the extension service was preoccupied with
promoting collective agriculture, leaving little time for assistance to
smallholders. The study reveals that since the adoption of new
technology is influenced by the frequency of extension contact, the
extension service needs to design a cost-effective way of reaching the
smallholders who are located in the remote rural areas of the country.
The role of the extension service in creating the link between farmers and
researchers is critical in technology development and transfer. The study
found that farmers are not using the recommended packages of
technologies. With feedback from farmers, extension agents will be in a
better position to understand what factors are contributing to the

suboptimal use of recommended inputs. Extension feedback will also

185

contribute to the design of effective technology development and transfer
policies.

3. The study has demonstrated that higher levels of education of farmers
contributed to increased rates and intensity of adoption. The extension
program has to design technical training programs to provide farmers the
required skills.

7.3.3 Implications for Ago'cultural Development Polig

The challenges facing Ethiopia in the 19905 are the achievement of food
security, agricultural growth and sustainable development. Our findings in this
study reveal that smallholders are responsive to economic incentives and that the
smallholder road to development should form the centerpiece of agricultural
policy. The government should consider investing a higher percentage of its
resources in the smallholder sector while reducing the allocation for producer
cooperatives and state farms. Government resources are needed to strengthen
and expand agricultural research and extension services to enhance technology
generation and transfer.

Farmers in the study districts are not using the recommended levels of inputs
because of the physical unavailability of inputs, and the lack of credit and high
input prices. The lack of rural credit and input delivery institutions has been
identified as a major constraint on technology adoption. The current government

or project run input distribution system has not been effective. There is a need to

186

design an alternative input delivery system. The role of the private sector in input
delivery should be explored.

7.4 Limitations of the Study and Recommendation for Further Research
The main objective of the study was to examine the social and economic
factors influencing technology adoption in Arssi region. The ultimate goal of this

study was to contribute to the knowledge base on technology adoption by
smallholders. The limitations of this study are associated with the data. The data
were collected during the political upheaval in 1991. In this insecure atmosphere,
the author and the enumerators were unable to travel to some of the selected
villages. Farmers were hesitant in responding to questions related to past and
present yields, prices, wages, farm size and farmers’ opinions on government
institutions. These issues were found to be sensitive in the prevailing political
situation.

Time series data on farmers’ awareness and use of different varieties, and
quantities produced and used for home consumption and seed, are important in
analyzing adoption over time. Since this study used cross sectional data from one
season, the conclusions are only indicators of the technology adoption process and
the factors responsible for enhancing or constraining adoption in the study area.

Several related issues were beyond the scope of this study and merit
additional investigation:

1. The results of the econometric analysis of this study suggest that

institutional factors influence adoption decisions more than economic and

187
technical factors. Ethiopian agriculture is characterized by weak

research- extension linkage, marketing and input delivery systems.
Further research is needed on institutional innovations for smallholders.
This study provided evidence on the rates and intensity of adoption. The
results can be used to design future research on impact assessment to
strengthen technology development and transfer efforts. A follow-up rate
of return study on wheat research in Ethiopia is needed to formulate a
policy on agricultural research resource allocation.

Ethiopia is importing about half of its wheat requirements. Since the
profitability of wheat production is influenced by macro prices (wage,
exchange rate and interest rate) that are set by the government, a study
of the entire wheat sub sector will improve the empirical basis for
appropriate policy formulation.

There is a potential to increase wheat acreage and irrigated wheat
production. An in-depth economic analysis of irrigated wheat production
and other alternative/ competing enterprises would assist policy makers
and agricultural researchers in formulating priorities for wheat research

and development.

APPENDICES

188

APPENDIX I
Institute of Agricultural Research
Dept. of Agricultural Economics
Questionnaire on the Economics of Smallholder Wheat Production and
Technology Adaption in Southeastern Ethiopia.
Principal Investigator: Mulugetta Mekuria, Ph.D candidate,
Dept of Agricultural Economics, Michigan State University, East Lansing

Farmer's Name

 

PA

Awraja: Keleta-1 Chilalo-2 Galema-3e Gedeb-4. Ticho-5

 

AWRAJA

FNUHB

ENUH

Farmer Number

Enumerator

Date

 

Instruction to the Enumerator
Please introduce yourself before starting questioning the farmers by
name location, the institute you are working for and its purpose and
objective. Please ask each question patiently until the farmer gets the
point. Please fill up the questionnaire according to the type of
question. For open question fill the farmer response in short and for

closed one indicate by ticking (v). or encircling (1).

189

QUESTIONNAIRE

I HOUSEHOLD CHARACTERISTICS

1. Household head:

1.1 Male 1 Female 2

1.2 Age years.

1.3 Level of formal education: Illiterate

Literacy campaign 2 Elementary ed.

Secondary ed. 4.
1.4 If female, where is the husband not releyent-O
deed-1.

soldier-2. divorced-3, other—4

1.5. How long have you been farming?

 

 

 

 

 

 

 

 

years.
2. Household Members
2.1. Number of adults living in household
over 17.
2.2. Number of children 14-17
2.3 Number of children less than 14
2.4 Number working and living off farm
3. Land
3.1 Total farm size timad
3.2 Arable/cultivated area timad
3.3 Grazing area timad
3.4 Fallow timad
3.5 Other(specify) timad

 

SEXHH

AGEHH

()LVLEDHH

LOCHUSB

FMLSZ

 

ADT 17+

CHD 14-17

CHD 14-

NWLOFFH

FARHSZ

ARBLAREA

GRZAREA

FALHAREA

OTHERAREA

190

4. Land tenure of farmer

 

1. Own YES ___ 1 NO ___ 2 LDTNOVN
2. Family YES ___ 1 N0 ___ 2 LDTNFAH
3. Own/family YES ___ 1 NO ___ 2 LDTNOWF
4. Share cropping YES ___ 1 NO ___ 2 LDTNSHR
5. Rent YES .___ 1 NO ___ 2 LDTNRNT
6. Borrow YES ___ 1 NO ___ 2 LDBRW

7 If share cropping, area timad SHRAREA
8 If renting, area timad RNTAREA
9 If borrowing, area timad BRWAREA

5. Fertilizer
5.1 Did you apply fertilizer on the following crops in the 1990 and

1991 crop season?

 

 

 

 

 

1990 1992

Yes No Yes No
Wheat 1 2 l 2 FW90__ FW91__
Barley l 2 l 2 FB90__ FB91__
Faba bean 1 2 l 2 FFB90__ FFB9I__
5.2 If No. reasons RNNOFAP
Not heard of fertilizers/never tried before____ 1 RNNOFAP
Too expensive 2
Unavailable 3
Late delivery 4
Farmer believes that fertilizer doesn't increase yield 5

Other (specify) 6

191

5.3. If yes amount used

 

 

 

 

 

1990 1991
CROP Amount (kg) Amount (kg)
DAP Areg Ureg Area DAP Area Ureofi Areg
Wheat
Barley
Faba bean
5.4 Do you know the recommended fertilizer rate? KWRFTRT
by IAR or MOA Yes 1 No 2

5.5 If yes, what are the recommended rates for 1990/91?

DAP (kg/ha) Urea (kg/ha)

Wheat

 

Barley

 

Faba bean

 

5.6 If the recommended rate is higher than what the farmer had used

(Refer. Q. 8.3) ASK. Why the farmer did not apply more fertilizer?

 

 

 

 

 

 

RLFTAP
Too expensive 1
Unavailable 2
Late delivery 3 GTFTTHL
Higher rates damage the crop 4
Other (specify) 5
5.7 Do you get fertilizer on time? Yes 1

No 2

192

 

 

 

 

 

 

 

 

 

 

 

 

5.8 How far do you travel to buy fertilizers? FTDSTKH
Distance in km
5.9 Number of kilometers to the nearest market_____ NRHKKH
5.10. What is the length of time since you FTAWYR
first heard about fertilizers?
Years (No.years since the farmer was aware)
5.11 How many years have you been FTUSYR
using fertilizers?
5.12 Have you been using fertilizer every USFTYRL
year or discontinued using for some
reason? Use every year 1
Have not used every year 2
5.13 According to your view, what are the ADVFT
advantages of fertilizers?
Increase grain yield 1 ADVFT
Increase straw yield 2
Improves the quality of the crop 3
Other (specify) 4
5.14 Do you think fertilizers have
disadvantages? Yes 1 THKFTDA
No 2
5.15 If yes, what are the disadvantages
associated with fertilizer use? DADVFT
Favors weed growth 1 DADVFT
Burns/damages the crop 2

 

Doesn't increase grain yield 3

193

Expensive/unaffordable 4
Other (specify) 5

6. Weed control

 

 

6.1 How many times did you weed the following crops?
1989 1990 1991
Wheat NWDWH89 NWDWH9O NWDWH91
Barley NWDB89 NWDB90 NNDB91
Faba bean NWDFB89 NWDFB9O NWDFB91

 

6.2 How many weeks after planting do you normally weed your crop?

 

 

 

 

Wheat Berley Page bean
Three weeks after planting 3 3 3
Four weeks after planting 4 4 4
Five weeks after planting 5 5 5
Six weeks after planting 6 6 6

 

Other(specify)

 

6.3 Did you apply herbicides on the following crops in the 1990 and 1991

crop seasons?

 

 

 

 

1990 1991
Yes, No Yes No APHBW90_____
Wheat 1 2 l 2 APHBW91_____
APHBB90_____
Barley l 2 l 2 ABHBB91
ABHBFB90____

 

Faba bean 1 2 1 2 ABHBFB91

194

6.4 If yes, area sprayed in timad and cost

Wheat
Barley

Faba bean

 

 

 

 

w 1221
Area Birr Area Birr
WTARSP90_____ WHBCO90_____ WTASP91____ WTHBC091___
BARSP90_____ ' BHBCO90____ BARSP91____ BHBCO91____
FBARSP9O FBHBCO90____ FBARSP91____ FBHBCO9l___

 

6.5 If the farmer had used herbicide, what are the advantages of

herbicides?

Saves labor to do other tasks 1

 

Reduce weeding labor requirement
Controls weeds better than hand weeding

Increase yield

 

Gives more time to rest

 

Other (specify)

 

6.6. Do you think herbicides have disadvantages?

1 No

 

 

6.7. If yes, what are the disadvantages

associated with herbicides use?

Too expensive

 

Risk of not working if little moisture

Unavailability of sprayers

 

May burn or damage the crop

 

Lack of technical knowledge on use

Don't control weed adequately

 

Other (specify)

ADVHBUS
2
3
4
5
6
THKHBDAV
DADVHB
l
_____2
3
4
5
6
7

 

195

6.8. If the farmer did not use herbicides in both RNOHBVS
years, why did he not use herbicide?
Not heard of herbicides 1
Expensive/unavailable 2
Herbicides do not solve the weed problem __3
Farmer has enough labor 4
Other (specify) 5
7. Seed:

7.1 What kind of seed (variety) did you plant in the 1990 and 1991 crop

 

 

seasons?
1990 1991
Improved Local Improved Local

Wheatl l 2 1 2 WIHPS90____
WIHPS91____

Barley 1 2 1 2 BIMPS90____
BIHPS91____

Faba bean l 2 l 2 FIHP90_____

 

FIHP91

7.2

Crop

Wheat

Barley

Faba bean

196

If the farmer had planted improved varieties, what are

the varieties? (list)

Variety

Dashen

ENKOY

BULK

K6295 4A

Other

Variety

Variety

1990

Area

 

Area

 

Amount

Amount

1991

Area

 

Area

Amount

Amount

7.3 If the farmer had planted more than one variety of each or any of

the crops, which variety has he liked best?

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

RVRPREF]
RVRPREF2
RVRPREF3

Crop Variety Main reason for liking the best:
Wheat 1
2
3
4
Barley l
2
Faba bean l
2

 

 

LIST OF CRITERIA FOR VARIETY PREFERENCE:

High yield 1 Resistance to disease 2 Resistance to lodging 3
Earliness 4 Seed size 5 Seed color 6 Weed competition 7 Taste in
injera 8 Taste in bread 9

197

7.4 What was your source of seed ?

Source Whoop Barley Faba bean
Received from MOA 1 l 1
Received from Research Team 2 2 2
Purchased 3 3 3
Retained from own harvest 4 4 4
Other (specify) 5 5 5

7.5 If purchased, what was your source of purchase?

Wheat Barley Faba bean
Source Amount Cost Amount Cost Amount Cost

Service coops

 

 

Local Market

 

 

Producer Coops

 

 

Neighbor

 

 

Relatives

 

 

Others

 

 

7.6 If purchased, how far must you travel to buy improved seed?
Distance in km. . DBYISD

 

7.7. Quantity of improved seed harvested and disposed in 1990

Crop variety Harvested(qt) Sold(qt) Price/qt. Retained(qt)

 

Wheat Dashen
ENKOY
K6290 Bulk
K62954A
Barley

Faba bean

198

 

 

 

 

 

 

7.8 Did you plant all of ..... field to improved seed?
1990 1991
Yes No Yes No

All Helf One-third All Vgolf One-third
Wheat 10 12 13 20 10 12 13 20
Barely 10 12 13 2O 10 12 13 20
Eobogbeen 10 12 13 20 10 12 13 20
7.9. If No, why did you not plant all of your RNPIMPS

land to improved seed?

Too expensive

 

Unavailability of seed

 

Lack of fertilizer

 

Not aware of it

 

Because of previous bad experience

Other (specify)

 

6

7.10 If you have a question about improved varieties,

fertilizers or herbicides, whom do you ask first? SRCINFO

 

 

 

 

 

MOA development agent 1
Research team 2
Producer cooperative executive member___ 3
Service cooperative 4
Contact/Model farmer 5
Neighbor 6
Other 7

 

\ b6
1 “I A: 1
Kill“! Q, [)u‘l‘.‘ \ \lm

l
\

\/|

 

199

7.11. If the farmer has not used improved varieties of the three crops,

ask him/her why he did not plant improved varieties?

 

 

 

 

 

 

Not heard of improved varieties l RNOUSIV

Very expensive seed price 2

Improved varieties unavailable 3

Improved varieties are not better than local cultivars 4
Lack of fertilizer 5

Other (specify) 6

 

8. Access to Credit:

8.1. Do you get credit for your farming

 

 

 

 

 

 

operations? Yes 1 GETCRDT
No 2
8.2 If yes: Purpose: To purchase fertilizers___1 PRPCRDT
To purchase improved seed_____2
To buy oxen 3
To buy herbicides 4
Other(specify) 5

 

8.3 Source of credit :

 

 

 

 

MOA 1 SRCCRDT
Scs 2
Bank 3
Neighbor or relative 4
Local money lender 5

 

Other (specify) 6

200

8.4. Have you ever been disqualified from getting

credit because of lack of cash for credit down payment?

 

 

 

 

 

 

 

 

 

 

 

 

Yes 1 DISQCRD
No 2
9. Do you hire labor? Yes 1 HIRLABR
No 2
9.1 If yes for which operations?
Whegt Berlev Feboebeen
Plowing 1 l l
Planting 2 2 2
Weeding 3 3 3
Harvesting 4 4 4
Other(specify) 5 5 S

 

9.2 If no, why?
I have enough labor 1 RNOHIRL
Too expensive 2
No labor available for hiring 3
Other(specify) 4

10. Livestock:

10,1 Would you please tell me the number of livestock you hoye?

 

 

 

 

Type ' 1990 1991
Number ValuefBirr) Number V§lue(Birrl,
an
Cows

 

Male calves

Female calves

m

 

201

 

 

 

 

 

um

Mules

Sheen

10.2 Are your oxen enough for your farm operations? ENOXFOP
Yes 1
No 2

10.3 If No, how do you get additional oxen SRCADOX

when you need them?

 

 

Makanajo system 1
Hire from someone 2
Borrow from friend or relative 3
Other (specify) 4

 

ll. Off-farm activities

 

 

 

 

 

 

11.1 do you have a job off your farm? OFFARMJ
Yes 1
No 2
11.2 If yes, Type of work TYPOFJB
Type of work
11.3 Annual Income earned from off-farm work INCOFJB

 

11.4 Does any one else in the family living with you have job off the

 

 

farm? FAHOFJ
Yes 1
No 2
11.5. If yes, type of work JYPOFFJ

 

 

11.6. Annual income earned

11.7. Do you have family member(s) living and working LIVOFWK

11.8 If yes, annual remittance received from family ANREHRC

12.

12

12.

12.

12

12

13.

202

off-farm? Yes

 

No

 

members living and working off-f

1

2

arm

 

 

 

 

 

 

 

(birr)

 

 

 

 

2

1

'Access to Information:
.1 Have you ever attended a field day or
demonstration trial? Yes 1
No 2
2 Have you ever attended a farming
training course? Yes 1
No 2
3 Do you have a radio? Yes 1
No 2
.4 If yes, do you listen to any agricultural
education program? Yes 1
No
.5 If yes, is the coverage or content of the
program satisfactory? Yes 1
No 2
Did extension agent visit you last year?
Yes
No

 

2

13.1. Did extension agent visit you this year?

Yes

 

No

 

l

2

 

ATFLDD

 

ATFTRNG

 

OWNRDIO

 

LSNAGED

 

AGEDCNT

 

DAVST90

 

DAVST91

 

203

13.2. If yes, what time of the year or during which operation?

 

 

 

 

 

 

 

 

Plowing. 1 No.of visit
Plantingfi 2 No. of visit
Weeding, 3 No.visit
Harvesting 4 No.visit
14. Have you ever lived out side your village? LVDOVLG
Yes 1
No 2
14.1. If yes, for how long? years. LVDOVYR

 

14.2. How far is your house from the nearest town? Km.

00¢

>wmmchx HH
WmowmmmHoz wmm0fiam om FHZUMN wwoow>z

HowHH“ rmmuzcwfio0ouwmmIXHm

rHaHnma Umwmdaodn <mHHmUHm ZoamH - OMZmowmc HmmnmmmHo:
:mxHaca eroHHrooa mmnHamnmm

rom-fimeHHrooa.............. -H000.0H0

ardmmroHa <mHCmm mom n76 aoaan boson! 0.0000 cvvoHI+Hdeanw
<mHHmUHo oommmHOHmSn 060. 06606 n-6mnHo 0600_n_wx 26m: om x an.Um<.om x
oodmnmsn -0.0bb0 H0.b0 -0.000 0.00000
>ommo>a -b.0000 0 000 -H.0H0 0.HOHHO 0.00H0 0.0HHHb
r<rmcmm b.H000 0.000 H.000 0.H0000 H.0000 0.00000
marmN 0.0bH00 0.0H00 0.000 0.60HOH 0.00H0 0.0000
m>wszm> 0.0000 H.000 0.000 0.00000 0.0H0b H.000»
0x20200 0.0000 0.H00 0.000 0.00000 H.0000 H.Hb00
nawazr 0.0000» 0.000 0.H00 0.00000 0.00H00 0.60006
omanwca 0.0000 6.000 H.000 0.00000 0.60000 0.00000
Kawwawa 0.0000 0.00» H.000 0.HOHOH 0.00000 0.0b0Hb
>0:0€00 H0.0H0 0.000 0.00» 0.00006 0.00000 0.9H00p
0><0a00 Hb.000 0.000 0.H00 0.00000 0.HO0H0 0.00000
0 60.6H0 H.000 00.000 0.00000

H. ZmnHHx -V 0Hn0>waA0.H.HHV
AAAA a H
H

wos H -0.0bb00

wos 0 .9.00000

Nos 0 b.H00b0

wot b 0.09H00b

”08 0 0.00000

woz 0 0.00000

mos 0 0.000009

”08 0 0.00000

”08 0 0.00000

”02 H0 H0.0000

wOS H0 H0.0000
woz HH Hb.0000

0. zmnHHx -V waI3m>zAva

AAAA wa VVVV norczz
H
”08 H H.00000
woz 0 0.00H00
”08 0 H.00000
wot 9 0.00000
”02 0 0.0H0bb
woz 0 H.00000
”OS 0 0.00H000
wos 0 0.000000
woz 0 0.0000H0
W02 H0 0.000000
wot HH 0.Hb0H00

omHochanm -V 010

vvv o 60.6H066 AAA

omHochanm -V 0x0IUOHA0H.x0HV\o
vvv 0x6 H.000H06 AAA

omHochanm -V 30n20HAOx00\meA0xmv
vvv z: n c.060Hmo0 AAA

omHochanm -V mumeAmxmv
vvv 6 0.0006H0o AAA

omHochanm -V mlexAH-0x0*zc-KC>0v
vvv 6H t 0.0060060 AAA

omHochanm -V 00I20HwamvxwxM+20Hwa00630

VVV 00 n 0.00H0000 AAA

000

chanoEHmH romHn ZoamH
meHaca rHWmHHrooa anHamnmm

rom-ermHH:ooa..............
wmmnHHonma AmHovmmlov Hom-r.
03H-00cmnma AHov..
mHmsHmHomsom rm<mH.....

-H00.00H0
-H00.0000

00.00000
0.0000H00m-00

000

oommmHonsn mna. mnnon n-HmnHo wHoU_n_wx 26m: cm x mna.cm<.om x

CmHHmva

OOSmHmnn 0.00000
>00m0>a -0.0HOH0
fi<rmomm 0.00000m-0H
wzrmN 0.HHOH0
w>WZmNm> 0.00000m-0H
0x20300 0.H0000
nawHZF 0.0000H
omaowca 0.00000
Nzwwawa 0.H0000
>0mmz00 H.0000
0><0H00 0.000HH

0.
0.
0.
0.000Hm-0H
0.
.H000
.0000
.0000
.0000
.0000
0.

00000

0000
HO0H
0000

HH00

0000

0.000

-H.00H

0.000
H.000
0.000
H.000
H.00H
0.00H
0.000
0.000
0.000

meACmDOHom om moncmH m unmaHonoa oCnooamm
wHoQHonma OCnooBm rmm amxHaca UHOUWUHHHHV.

HOH>P
0H
000
000

norczz

0meHonma
>oncmH 0
0 H
H 0
Hoa>fi H
H. ZmnHHx -v x0n30>zAva
AAAA Xw VVVV
H
~08 H H.00000
”oz 0 0.00H00
NOS 0 H.00000
we: 0 0.00000
”02 0 0.0H000
wot 0 H.00000

0.00000
0.H0000
0.00000
0.00000
0.00000
0.00000
0.000H0
0.000H0
0.00000
0.0HHHO
0.0H000

0.00H0
H.0000
0.00H0
0.0H00
H.0000
0.00H00
0.00000
0.00000
0.00000
0.HO0H0

0.0HHHO
0.00000

0.0000

H.0000

H.H000
0.00000
0.00000
0.000H0
0.0H000
0.00000

wo2
”02
W02
”02
”02

omHochanm -V mlfiowfiuoan0.0vv
I 0.00000HH
omHochanm -V wnwaH-0v

VVV w

VVV W
omHochanm -V WIW02A00
vvv x I HH.ooooo
H. zmnHHx -V cmra>n060
AAAA Dmfia> VVVV
H
wo2 H 0.0H00000-0H
w02 0 -0.0H00000-0H
w02 0 0.0000000-00
w02 0 0.HHH0000-0H
w02 0 0.0000000-00
”02 0 0.H0000H0-0H
w02 0 0.0000000-0H
w02 0 0.00H0000-0H
wo2 0 0.HOHOH00-0H
w02 H0 0.H00000
w02 HH 0.0000000-0H
H. ZmnHHx -V cmra>lmxm
AAAA UMPH> VVVV
H
”02 H 0.0H00000-0H
w02 0 -0.0H00000-0H
”02 0 0.0000000-00
w02 0 0.HHH000m-0H
~02 0 0.0000000-00
w02 0 0.H0000Hm-0H

HOQQN

H
H

0.00H000
0.000000
0.0000H0
0.000000
0.HO0H00

00.0

AAA

I 0.00000H0m-0H AAA

AAA

000

wo2 0 0.0000000-0H
”02 0 0.00H000m-0H
w02 0 0.HOHOH0m-0H
w02 H0 0.H00000

w02 HH 0.0000000-0H

6. 26666x -v 616666626666.66-66666666_66.
6. zmnnwx -v 6:6666626666666-666666666_x6.

omanHmanm -v AH-0»0060
VVV wmmcHn I -0.00000H00-0H AAA

0. ZmnHHx -V ma>aficmra>.<>wcmra>v

ZmnHHx mnmnHmnHomH fimmcHnmu 00mmmH0Hmdnmlcmfie> <mnHm:omI<>wcmrH>
<mHHmch nommmHOHmdn 0na.mnnon n-HmnHo wnoc_n_wx 3mm: om x 0na.om<.om x
UmrH>lH 0.0H000m-0H 0.000H0-0H 0.000 0.0000H

Umra>l0 -0.0H000m-0H 0.H000m-0H -H.000 0.00000

00PH>I0 0200000m-00 0.H000m-0H 0.000 0.00000

Umra>l0 0.HHH00m-0H 0.0000m-00 H.000 0.000H0

0000>l0 0.000000-00 0.HH00m-0H 0.000 0.0H000

Umra>l0 0.H0000m-0H 0.H0000-0H H.000 0.00000

cmrH>l0 0.00000m-0H 0.0000m-0H H.000 0.00000

Umfia>l0 0.00H000-0H 0.000Hm-0H 0.H00 0.00H00

Umra>l0 0.HOH00m-0H 0.0000m-0H 0.000 0.00000

umra>lH0 0.H000H 0.0000m-0H 0.0H0 0.0000H

UmrH>IHH 0.00000m-0H 0.0H00m-0H 0.000 0.0H000

000

30000 00330206

000Han0zml>020200“wmmlem
chnHaoaHmH 0omHn ZommH

0 ocnooamm" >0:020I0 >0mw20lH
noommHonSnm mom >02020I0 mmn no wao.
meHaca 0meHH0000 manHamnmm
0om-0meHHrooa.............. -000.000H
wmmnHHoan AmHowmmlov 060-0. -000.00H0
00H-0ncmfima A 0v............ 00.0HO0H
0HmSHmHomsom 0m<mH.......... 0.00000000-00
<memva nommmHOHmSn 060. 06606 6-606Ho 0600_n_wx

GOSmHmSn -0.0000 0.0000 -0.000 0.000H0
>nmmn>a 0.00000m-0H 0.H000 0.000 0.00000
0<0mcmz 0.00000 0.HO0H H.000 0.H0000
0300N 0.00000m-0H 0.0000m-0H H.H00 0.000H0
w>w30Nm> -0.000H0 0.HH00 -0.000 0.000H0
0x20300 0.00000 0.H000 0.000 0.00000
nmaowca 0.00000 0.0000 H.000 0.00000
M200 H.H000 0.0000 0.000 0.00H00
0><0H00 0.00HHO 0.0HHO 0.0H0 0.00000

MHmQCmSOHmm om moncmH m UanHonma ocnooamm
wnmaHoan oCnooam 70m BmxHECS vwovmvHHHnw.

manHonma
>oncmH 0 H HOH>0
0 00H 0 000
H 00 HH 00

HOH>0 000 H0 000

36m: 00 x mna.cm<.om x

0.00H0
H.0000
0.00H0
0.0H00
H.0000
0.00000
0.0000H
0.HO0H0

0.0HHHO
0.00000
0.0000
H.0000
H.H000
0.00000
0.00000
0.00000

000

H. KmnHHx -V xmlzm>zAXHv

AAAA X0 VVVV 000032
H
~02 H H.00000
~02 0 0.00H00
~02 0 H.00000
~02 0 0.00000
~02 0 0.0H000
~02 0 H.00000
~02 0 0.000000
~02 0 0.000000
~02 0 0.HO0H00

omHochanm -v 09000A000Ax0.000
VVV 0 I 0.H000H00 AAA

anochanm -v wuwaH-wv
VVV w I 0.H000H00 AAA

omHochanm -V ~I~02va
VVV K n 0.000000 AAA

H. zmnHHx -V 0m0H>I060

AAAA 000H> VVVV 000032

H
~02 H -0.00HO0H
~02 0 0.00H000m-00
~02 0 0.000000m-0H
~02 0 0.000000m-00
~02 0 -0.000000m-0H
~02 0 0.0HOHOHm-0H
~02 0 0.0H0000m-0H
~02 0 0.H00000
~02 0 0.H00000

366666 -v 66>666666>.<>06666>V

ZmnHHx mnmnHmnHomH HmmcHnmn
<mHH00Ho

cmra>|6
66r6>|6
6666>I6
6666>r6
66r6>n6
6606>I6
6666>|0
6666>u6
6666>I6

oommmHonDn 0mm. MHHOH

-0.00H00
0.00H000-00
0.00000m-0H
0.00000m-00

-0.00000m-0H
0.0HOH0m-0H
0.0HO0Hm-0H
0.H0000
0.H0000

0.HH00

0.0000m-0H
0.0000m-0H
0.00H0m-00
0.HO0Hm-0H
0.H000m-0H
0.0000m-0H
0.00000-0H
0.0000m-0H

oommmHonsnmI000a>

00H

n-HmnHo 0600_n_wx
-0.00H 0.00000
0.000 0.00000
H.000 0.H0000
H.H00 0.00000
-0.000 0.00000
0.000 0.00000
H.000 0.00000
0.000 0.0H000
0.000 0.0000H

<mHHmsomI<>~000a>

Sow: cm x mna.cm<.om x

onawflma 0H0000 zoawH
meHaca 0wmeHsooa mmnHamnmm

000-0H00HH0000.............. -000.00H0
~mmnHHonma AmHovmmlov 000-0. -00H.0000
05H-mncmdma A 00............ 00.00000

mHmsHmHomdom 0040H.......... 0.H000HH00-00

omHH 0n00C0donm 008 05000500
anH 00Cdn ~0H.0Hmpr

0 00. 0.0HH00

H HHO. 0.00000

0 00. 0.H0000

0 HHO. 0.00000

0 00. 0.00H00
<mHH00H0 oommmHonsn 060. 00606 n-HmnH0 0600_n_wx
nonmnmdn 0.H0000 0.H000 0.000 0.00000
0<000=m 0.00H00 0.0H000-0H 0.000 0.0000H
0x20300 0.H0000 0.00000-0H 0.000 0.0H000
00H0~0H 0.00000 0.H000 H.000 0.00H00
000030 0.00000 0.HHOH 0.H00 0.00000
30A Hv 0.00000 0.000H0-0H H0.H00 0.00000
30A 00 H.0H00 0.00000-0H H0.000 0.00000
30A 00 0.0000 0.HH00 0H.0H0 0.00000
mamncmsonm om moncmH Q 0H00Honma osmooamm
whoaHoan canooam 000 smxHSCB unocmvHHHnw.

mamaHonma

>oncmH 0 H 0 0 0 HOH>0

0 00 00 0 00 0 00

H H0 00 0 00 0 HHO

0 0 00 0 00 0 00

0 0 00 0 0H 0 HHO

0 H 0 0 H0 0 00
HOH>0 0H H00 0 HOH 0 000

000

H.0000
H.0000
0.00000
0.00H00

Kmmz 00 x an.Um<.om x

0.00000

H.H000
0.00000
0.00000

000

amHOCHmanm -V wxlwAHO+0AOV$30~A0<00cmmv+0AOV$x0~A0xzc3000+0A0vxx0~A0000~00
0+0A0v*x0~A060630v
VVV 0% I 0.0000000 AAA

omHochanm -V 00I0-20HA-0xv
VVV 00 I -0.0000H00 AAA

OmHochanm -V 0HI20HA-0XV-20H30A0V-0xV
VVV OH I -0.HH00000 AAA

omHochanm -V 00I20H30A00-0x0-20HA0A00-0xv
VVV 00 I 0.00000000-0H AAA

anoCHmanm -V 00IZOHA0A0V-0NV-ZOHA0A0V-0xV
VVV 00 I 0.0000000 AAA

omHochanm -V 00I20HA0A0V-0x0-0
VVV 00 I 0.00000000-0H AAA
H. 3mnon -V 00H>I0>~HA0.H.0V

AAAA me> VVVV 000022
H
~02 H 0.H00000
~02 0 0.00H000
~02 0 0.H00000
~02 0 0.000000
~02 0 0.000000

0. 3mnHHx -V 00000XI00»00H>
AAAA 00000x VVVV 000032
H
~02 H -0.0000000-0H
~02 0 -0.00HOH00-0H
~02 0 -0.0000000-0H
~02 0 -0.0000000-0H
~02 0 -0.0000000-0H

”08
”OS
wos
”08
”02

mos
woz
”OS
”02
woz

woz
mos
wot
”08
mo:

”08
”08
”08
”OS
NOS

0.

ZmnflHx -V wcwHUXIUHfiwma>

AAAA waHUX VVVV OOFCKZ

MDWNH

0.

H
-0.H00000m-0H
-0.uHomomm-0H
-0.H000mmm-0H
-0.000m0um-0H
-0.0mmH00m-0H

zmnHHx -V 00000xlc0%wma>

AAAA 00000x VVVV OOFCZZ

m0wNH

m.

H
0.00HomHm-00
0.HHOmmom-0H
0.00000Hm-00
0.0000mmm-00
0.H00mu0m-0H
zmnHHx -v mcwwcxlcw*mma>

AAAA wcwwcx VVVV 000032

Uwat-J

m.

H
0.00000mm-0H
0.000000m-0H
0.00mmem-0H
0.00Hm0Hm-0H
0.000Hmom-0H

zmnHHx -V 0000UXI00*mma>

AAAA 00000x VVVV OOFCZZ

0'19me

H
0.HO0Hmum-0H
0.0mmoH0m-0H
0.Hoomwwm-0H
0.HomHHHm-0H
0.0ww0mum-0H

03

215
REFERENCES

Adesina, Akinwumi A. and Moses M. Zinnah. 1993. "Technology Characterstics,
Farmers’ perceptions and Adoption Decisions: A Tobit

Model Application in Sierra Leone." Ago'cultural Economics,
9:297-311.

ADD/NFIU/MOA/FAO.1991. Results of Fertilizer trials conducted on Major
ereal 19 -1 . Addis Ababa: MOA.

Akinola, Amos A. 1987. "An Application of Probit Analysis to the Adoption of

Tractor Hiring Services Scheme in Nigeria." Oxford Agrarian
Studies 16:70—82.

Akinola, Amos A and Trevor Young. 1985. "An Application of the Tobit Model
in the Analysis of Agricultural Innovation Adoption
Processes." Oxford Agrarian Studies 14:26-51.

Amemiya, T., 1981. "Qualitative Response Model: A Survey." Journal of Economic
Literature, 19:1483-1536.

Amemiya, T., 1984. "Tobit Models: A Survey." ournal of Econometrics, 24:3-61.

Audi, Patrick 0., and Dale Colyer. 1993. "Adoption of New Technologies by
Smallholder Farms in Semi-Arid Areas of Kenya." Paper
presented at the Annual Meeting of the American
Agricultural Economics Association, Orlando, Florida, August
1-4.

Bekure, Solomon. 1992. "Priming the Pump of Economic Growth: A Strategy for
Agricultural Development in Ethiopia." Paper presented at
Symposium on Economic Development in Ethiopia:
Economic Policy During the Transition Period. January 15-
18.

Beyene, Hailu., Steven Franzel, and Wilfred Mwangi. ”Constraints to Increased
Wheat Production in Ethiopia’s Smallholder Sector." In
Steven Franzel and Helen van Houten (eds). Research with
Farmers: Lessoos from Ethiopia. CAB International.

Boughton, Duncen., Eric Crawford, Mark Krause, and Bruno Henry de Frahan.
1990. "Economic Analysis of On-Farm Trials: A Review of
Approaches and Implications for Research Program Design."

216

Staff Paper No. 90-78. East Lansing: Dept. of Agricultural
Economics, Michigan State University.

Byerlee, Derek and Steven Franzel. 1993. ”Institutionalizing the Role of
Economists in National Agricultural Research Institutes."
CIMMYT Economics Working Paper No. 93-01. Mexico,
D.F.: CIMMYT.

Byerlee, Derek and Edith Hess de Polanco. 1986. "Farmers’ Stepwise Adoption
of Technological Packages: Evidence From the Mexican

Altiplano." American Journal of Agg'cultural Economios; 68:
520-527

CIMMYT Economics Program 1992. "The Adoption of Agricultural Technology:
A Guide for Survey Design." CIMMYT Economics Working
Paper (Draft). Mexico, D.F.: CIMMYT.

Cleaver, Kevin M., 1993. "A Strategy to Develop Agriculture in Sub-Saharan
Africa and a Focus for the World Bank." World Bank
Technical Paper Number 203 Africa Technical Department
Series. Washington DC: The World Bank.

Debela, Seme and Sentayehu G. Mariam. 1990 "A Review of the Status of
Agricultural Research, Extension and Training in Ethiopia."
Prepared for IGADD as part of the Agricultural Research
Planning Project, Addis Ababa.

Delgado, Christopher L.,John W.Mellor, and Malcolm L. Blackie. 1987. "Strategic
Issues in Sub Saharan Africa" In John W.Mellor, Christopher
L. Delgado,and Malcom J .Blackie (eds) Accelerating Food
Production in Sub-Saharan Africa BaltimorezJohn Hopkins
Press.

Eicher, Carl K, 1990. "Building African Scientific Capacity for Agricultural
Development." Agg'cultural Economios. 4:117-143.

Eicher, Carl K., 1992. "Revitalizing The CGIAR System and NARSs in the Third
World." Staff Paper No. 92-73. East Lansing: Dept. of
Agricultural Economics, Michigan State University.

Eicher, Carl K., 1994. "Zimbabwe’s Green Revolution: Preconditions for
Replication in Africa." Staff Paper No. 94-1. East Lansing:
Dept. of Agricultural Economics, Michigan State University.

217

Feder, Gershon., Richard E. Just, and David Zilberman. 1985. "Adoption of
Agricultural Innovation in Developing Countries: A Survey."

Economic Development and Cultural Change, 33:255-298.

Fischer,G.W and M.M. Shah. 1990. "Ethiopian Agriculture: Crop Yield Response
and Fertilizer Policies". Addis Ababa: FAO Fertilizer
Program, Project GCPE/ETH/O39/ITA, National Fertilizer
and Inputs Unit.

Green, D.A.G. and DH. Ng’ong’ola. 1993. "Factors Affecting Fertilizer Adoption
in Less Developed Countries: An Application of Multivariate

Logistic Analysis in Malawi." Journal of Agriculmral
Economics, 44(1):99-109.

Green, William H., 1990. Econometric Analysis. New York: Macmillan Publishing
Co.

Griliches, Zvi. 1957. "Hybrid Corn: An Exploration in the Economics of
Technological Change." Econometrica 25:501-522.

Hassan, Rashid M and Hamid Faki. 1993 "Economic Policy and Technology
Determinants of the Comparative Advantage of wheat
Production in Sudan." CIMMYT Economics Paper No.6.
Bangkok, Thailand: CIMMYT

Henery, Micheal C. 1988. "Risk, Uncertainty and Adoption of New Agricultural
Technology: An Empirical Application of Expected Utility
Maximization." "Oxford Agrarian Studies 17:78-94.

Henery, Micheal C. 1983. "Factors in the Adoption of New Farm Technology in a
Less Developed Country with Special Reference to Guyana. "
Oxford Agrarian Studies 12:163-176

Herdt, Robert W. 1988. "Increasing Crop Yields in Developing Countries: Sense
and Nonsense." paper presented at the American Agricultural
Economics Association Annual Meeting, Knoxville.

Herdt, Robert W. 1984 "Differing Perspective of the World Food Problem :

Discussion."Amerioan Journal of Agg’cultural Economios
66:186-187.

Institute of Agricultural Research. 1988. " Recommended Research Results" IAR
Publication (in Amharic): Addis Ababa

218

Jarvis, LS. 1981. "Predicting the Diffusion of Improved Pastures in Uruguay."
American Joornal of Ago‘culmral Eoonomics. 63:495-502.

Jansen, Hans G.P, Thomas S. Walker, and Randolph Barker. 1990. "Adoption
Ceilings and Modern Coarse Cereal Orltivars in India."

American Joornal of Ago'cultural Economics 72:655-663.

Judge, George W., R. Carter Hill, William E. Griffiths, Helmut Lutkephol and
Tsoung-Chao Lee.1988. Introduction to the Theory and

Practice of Econometrig 2nd. Edition New York: John
Wiley.

Kebede, Yohannes, Kisan Gunjal and Garth Coffin. 1990. "Adoption of New
Technologies in Ethiopian Agriculture: The Case of Tegulet-

Bulga District,Shoa Province" Agricultural Economics. 4:27-
43.

Kennedy, Peter. A Guide to Econometrics Cambridge: MIT Press, 1992

Kmenta, Jan. Elements of Econometrics 2nd.Edition, New York: Macmillan
Publishing Company, 1988

Maddala, G.S. Limited-Dependent and Qualitative Variables in Econometrics.
Cambridge University Press, 1983

Manning, Mervyn H."Future Technology and Foreign Competition" Ago'cultural
Engineering

Maredia, Mywish K, 1993. "The Economics of the International Transfer of
Wheat Varieties." Unpublished Ph.D Dissertation, East
Lansing, Department of Agricultural Economics, Michigan
State University.

McDonald, J .F. and Robert A. Moffitt. 1980. "The Uses of Tobit Analysis." Ibo
Review of Economics and Statistics 62:318-321.

 

Mekuria, Mulugetta. 1992, " Building Sustainable NARSs: The Ethiopian
Experience" unpublished paper, East Lansing: Dept of
Agricultural Economics, Michigan State
University

Mekuria, Mulugetta, Hailu Beyene and Steven Franzel. 1989. " Farming Systems
Research: Challenges and Issues in the IAR." IAR Research
Report No.11) IAR: Addis Ababa.

219

Mekuria, Mulugetta, Steven Franzel and Hailu Beyene. 1992. "Farming Systems
Research in Ethiopia: Evolution, Development and
Organization." In Steven Franzel and Helen van Houten

(eds). Research with Farmers: Lessons from Ethiopia. CAB
International.

Mekuria, Mulugetta and Steven Franzel. 1987. "Farming Systems Research in
Ethiopia: Evolution, Impact and Issues" Research Repor;
No.7. Institute of Agricultural Research, Department of

Agricultural Economics and Farming Systems Research:
Addis Ababa.

Mellor, John W. 1982. "Third World Development: Food, Employment and

Growth Interaction." American Journal of Ago'cultural
Economics 64:304-310.

Mellor, John W., Christopher L.Delgado and Malcolm J .Blackie. 1987. "Priorities
for Accelerating Food Production Growth in Sub Saharan
Africa". In John W. Mellor, Christopher L.Delgado and

Malcolm J.Blackie (eds) Accelerating Food Production in
Sub-Saharan Africa Baltimore: John Hopkins Press.

Mirotchi, Mesfin and Daniel B. Taylor. 1993. "Resource Allocation and
Productivity of Cereal State Farms in Ethiopia." Agg'cultural
Economics 8: 187-197.

MOA/NFIU. 1991. Crop-Wise Fertilizer Usage. Planning and programming Dept.
Addis AbabazMOA.

Morris, Micheal L. 1989. "Wheat Policy Options in Sub-Saharan Africa: The Case
' of Zimbabwe" Agg'cultural Economics 3:115-129.

Morris, Micheal and Derek Byerlee. 1993. "Narrowing The Wheat Gap in Sub
Saharan Africa: A Review of Consumption and
Production Issues" Economic Development and
Cultural change, 737-761.

Norris, RE and Sandra S. Batie. 1987, "Virginia Farmers’ Soil Conservation
Decisions: An Application of Tobit Analysis." Southem
Journal of Agricultural Economics, 19:79-89.

Norusis, Marija J ., 1990. SPSSZPQ+Advanced Statistics 4. SPSS Inc.

220

Oehmke, James F., and Eric W. Crawford. 1993. "The Impact of Agricultural
Technology in Sub-Saharan Africa: A synthesis of Symposium
Findings." Staff Paper Number 93-63. East Lansing: Dept. of
Agricultural Economics, Michigan State University.

O’Mara, Gerald T. 1971. "A Decision-Theoretic View of the Microeconomics of
Technique Diffusion in a Developing Country." Unpublished
Ph.D. dissertation. Stanford University.

ONCCP, Government of Ethiopia. 1989. Towards a Food and Nutrition Strategy
for Ethiopia. ONCCP: Addis Ababa.

Paolino, Leonard and John W.Mellor. 1984. "The Food Situation in Developing
Countries: Two decades in Review." Food Policy, 291-303.

Polson, Rudulph A. and Dunstan S. C. Spencer. 1991. "The Technology Adoption
Process in Subsistence Agriculture: The Case of Cassava in
Southwestern Nigeria." Agricultural Systems, 36(1):65-78.

Rahm, MR. and Huffman, W.E., 1984. "The Adoption of Reduced Tillage: The
Role of Human Capital and Other Variables." American
Journal of Agricultural Economics. November: 405-413.

Rahmato, Dessalegn. 1992. "The land Question and Reform Policy: Issues for
Debate." Paper presented at Symposium on Economic
Development in Ethiopia: Economic Policy During the
Transition Period. January 15-18.

Rauniyar, Ganesh P., 1990. "An Econometric Model of Rate of Adoption of
Technology for Developing Countries." Unpublished Ph.D
Dissertation, Department of Agricultural Economics and
Rural Sociology, The Pennsylvania State University.

Rogers, Everett M. 1983. Diffusion of Innovations. 3rd edition. New York: The
Free Press.

Ruttan, Vernon M. 1986 "Assistance to Expand Agricultural Production" World
Development 14:39-63.

Schultz, T.W. 1975. "The Value of the Ability to Deal with Equilibrium. "Jomal,
of Economic Liroratpro 13: 827-846.

221

Shakya, PB. and Flinn J .C. 1985. " Adoption of Modern Varieties and Fertilizer
Use on Rice in the Eastern Terrai of Nepal". Journal of
Ago'cultural Economics. 36:409-419.

Shire, Saad Ali 1986."The role of Prices in Agricultural Supply: A Household
Production Systems Approach for Smallholders in Ethiopia."
Unpublished Ph.D Dissertation, The Pennsylvania State
University.

Stoneman, Paul. 1983. The Economic Analysis of Technological Change. New
York: Oxford University Press.

Strauss, John. Mariza Barbosa, Sonia Teixeira, Duncan Thomas and Raimundo
Gomes Junior. 1991. "Role of Education and Extension in the
Adoption of Technology: A study of Upland Rice and
soybean Farmers in Central-West Brazil." Agg'cultural
Economics 52341-359.

Stroud, Ann and Mulugetta Mekuria. 1992. "Ethiopia’s Agricultural Sector: An
Overview." In Steven Franzel and Helen van Houten (eds).

Research with Farmers: Lessons from Ethiopia. CAB
International.

Sureshwaran, S., S.R. Londhe, P. Frazier, and NP. Pascual. 1992. "A Tobit
Analysis of Factors Affecting the Intensity of Adoption of a
Soil Conservation Technology." Paper presented at the annual
meeting of The American Agricultural Economics
Association, Baltimore.

Tanner, Douglas G., Amanuel Gorfu, and Kassahun Zewde. 1991. "Wheat
Agronomy Research in Ethiopia." In Hailu Gebre-Mariam,
D.G. Tanner and Mengistu Hulluka, (eds.). Wheat Research

in Ethiopia: A Historical Perspective. Addis Ababa:
IAR/CIMMYT.

Tanner, Douglas G. and Giref Sahile. 1991. "Weed Control Research Conducted
on Wheat in Ethiopia." In Hailu Gebre-Mariam, D.G. Tanner
and Mengistu Hulluka, (eds). Wheat Research in Ethiopia: A
Historical Perspective. Addis Ababa: IAR/CIMMYT.

Tanner, Douglas G., Amanuel Gorfu, Lemma Zewdie and Asefa Taa. 1992.
"Developing Technologies to Improve Soil Fertility, Weed
Control and Wheat Varieties." In Steven Franzel and Helen

van Houten (eds). Research with Farmers: Lessons from
Ethiopia. CAB International.

222

Thomas, John W. 1982. "Food Problems and Emerging policy responses in Sub-
Saharan Africa: Discussion." American Journal of

Agg'cultural Economigs, 64:907-908.

Tobin, J., 1958. "Estimation of Relationships for Limited Dependent Variables."
Econometrica, 26:29-39.

Yirga, Chilot., Hailu Beyene., Lemma Zewde and Douglas G. Tanner. 1992.
"Farming Systems of the Kulumsa Area." In Steven Franzel
and Helen van Houten (eds). Research with Farmers:
Lessons from Ethiopia. CAB International.